Annotation of src/usr.bin/mandoc/roff.c, Revision 1.264
1.264 ! schwarze 1: /* $OpenBSD: roff.c,v 1.263 2022/05/31 20:21:40 schwarze Exp $ */
1.1 schwarze 2: /*
1.253 schwarze 3: * Copyright (c) 2010-2015, 2017-2022 Ingo Schwarze <schwarze@openbsd.org>
1.139 schwarze 4: * Copyright (c) 2008-2012, 2014 Kristaps Dzonsons <kristaps@bsd.lv>
1.1 schwarze 5: *
6: * Permission to use, copy, modify, and distribute this software for any
7: * purpose with or without fee is hereby granted, provided that the above
8: * copyright notice and this permission notice appear in all copies.
9: *
1.16 schwarze 10: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES
1.1 schwarze 11: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
1.16 schwarze 12: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR
1.1 schwarze 13: * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
14: * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
15: * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
16: * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
1.244 schwarze 17: *
18: * Implementation of the roff(7) parser for mandoc(1).
1.1 schwarze 19: */
1.99 schwarze 20: #include <sys/types.h>
21:
1.1 schwarze 22: #include <assert.h>
1.3 schwarze 23: #include <ctype.h>
1.117 schwarze 24: #include <limits.h>
1.167 schwarze 25: #include <stddef.h>
26: #include <stdint.h>
1.51 schwarze 27: #include <stdio.h>
1.1 schwarze 28: #include <stdlib.h>
29: #include <string.h>
30:
1.74 schwarze 31: #include "mandoc_aux.h"
1.167 schwarze 32: #include "mandoc_ohash.h"
1.217 schwarze 33: #include "mandoc.h"
1.137 schwarze 34: #include "roff.h"
1.221 schwarze 35: #include "mandoc_parse.h"
1.98 daniel 36: #include "libmandoc.h"
1.138 schwarze 37: #include "roff_int.h"
1.218 schwarze 38: #include "tbl_parse.h"
1.219 schwarze 39: #include "eqn_parse.h"
1.2 schwarze 40:
1.227 schwarze 41: /*
42: * ASCII_ESC is used to signal from roff_getarg() to roff_expand()
43: * that an escape sequence resulted from copy-in processing and
44: * needs to be checked or interpolated. As it is used nowhere
45: * else, it is defined here rather than in a header file.
46: */
47: #define ASCII_ESC 27
48:
1.43 schwarze 49: /* Maximum number of string expansions per line, to break infinite loops. */
50: #define EXPAND_LIMIT 1000
51:
1.187 schwarze 52: /* Types of definitions of macros and strings. */
53: #define ROFFDEF_USER (1 << 1) /* User-defined. */
54: #define ROFFDEF_PRE (1 << 2) /* Predefined. */
55: #define ROFFDEF_REN (1 << 3) /* Renamed standard macro. */
56: #define ROFFDEF_STD (1 << 4) /* mdoc(7) or man(7) macro. */
57: #define ROFFDEF_ANY (ROFFDEF_USER | ROFFDEF_PRE | \
58: ROFFDEF_REN | ROFFDEF_STD)
1.197 schwarze 59: #define ROFFDEF_UNDEF (1 << 5) /* Completely undefined. */
1.187 schwarze 60:
1.138 schwarze 61: /* --- data types --------------------------------------------------------- */
62:
1.41 schwarze 63: /*
1.42 schwarze 64: * An incredibly-simple string buffer.
65: */
1.8 schwarze 66: struct roffstr {
1.42 schwarze 67: char *p; /* nil-terminated buffer */
68: size_t sz; /* saved strlen(p) */
69: };
70:
71: /*
72: * A key-value roffstr pair as part of a singly-linked list.
73: */
74: struct roffkv {
75: struct roffstr key;
76: struct roffstr val;
77: struct roffkv *next; /* next in list */
1.8 schwarze 78: };
79:
1.52 schwarze 80: /*
81: * A single number register as part of a singly-linked list.
82: */
83: struct roffreg {
84: struct roffstr key;
1.53 schwarze 85: int val;
1.199 schwarze 86: int step;
1.52 schwarze 87: struct roffreg *next;
88: };
89:
1.167 schwarze 90: /*
91: * Association of request and macro names with token IDs.
92: */
93: struct roffreq {
94: enum roff_tok tok;
95: char name[];
96: };
97:
1.211 schwarze 98: /*
99: * A macro processing context.
100: * More than one is needed when macro calls are nested.
101: */
102: struct mctx {
103: char **argv;
104: int argc;
105: int argsz;
106: };
107:
1.1 schwarze 108: struct roff {
1.168 schwarze 109: struct roff_man *man; /* mdoc or man parser */
1.1 schwarze 110: struct roffnode *last; /* leaf of stack */
1.211 schwarze 111: struct mctx *mstack; /* stack of macro contexts */
1.96 schwarze 112: int *rstack; /* stack of inverted `ie' values */
1.167 schwarze 113: struct ohash *reqtab; /* request lookup table */
1.52 schwarze 114: struct roffreg *regtab; /* number registers */
1.42 schwarze 115: struct roffkv *strtab; /* user-defined strings & macros */
1.178 schwarze 116: struct roffkv *rentab; /* renamed strings & macros */
1.42 schwarze 117: struct roffkv *xmbtab; /* multi-byte trans table (`tr') */
118: struct roffstr *xtab; /* single-byte trans table (`tr') */
1.16 schwarze 119: const char *current_string; /* value of last called user macro */
1.27 schwarze 120: struct tbl_node *first_tbl; /* first table parsed */
121: struct tbl_node *last_tbl; /* last table parsed */
122: struct tbl_node *tbl; /* current table being parsed */
1.191 schwarze 123: struct eqn_node *last_eqn; /* equation parser */
124: struct eqn_node *eqn; /* active equation parser */
1.102 schwarze 125: int eqn_inline; /* current equation is inline */
1.96 schwarze 126: int options; /* parse options */
1.211 schwarze 127: int mstacksz; /* current size of mstack */
128: int mstackpos; /* position in mstack */
1.96 schwarze 129: int rstacksz; /* current size limit of rstack */
130: int rstackpos; /* position in rstack */
1.99 schwarze 131: int format; /* current file in mdoc or man format */
1.96 schwarze 132: char control; /* control character */
1.175 schwarze 133: char escape; /* escape character */
1.1 schwarze 134: };
135:
1.236 schwarze 136: /*
137: * A macro definition, condition, or ignored block.
138: */
1.1 schwarze 139: struct roffnode {
1.166 schwarze 140: enum roff_tok tok; /* type of node */
1.1 schwarze 141: struct roffnode *parent; /* up one in stack */
142: int line; /* parse line */
143: int col; /* parse col */
1.16 schwarze 144: char *name; /* node name, e.g. macro name */
1.236 schwarze 145: char *end; /* custom end macro of the block */
146: int endspan; /* scope to: 1=eol 2=next line -1=\} */
147: int rule; /* content is: 1=evaluated 0=skipped */
1.1 schwarze 148: };
149:
150: #define ROFF_ARGS struct roff *r, /* parse ctx */ \
1.166 schwarze 151: enum roff_tok tok, /* tok of macro */ \
1.110 schwarze 152: struct buf *buf, /* input buffer */ \
1.1 schwarze 153: int ln, /* parse line */ \
1.2 schwarze 154: int ppos, /* original pos in buffer */ \
155: int pos, /* current pos in buffer */ \
156: int *offs /* reset offset of buffer data */
1.1 schwarze 157:
1.212 schwarze 158: typedef int (*roffproc)(ROFF_ARGS);
1.1 schwarze 159:
160: struct roffmac {
1.2 schwarze 161: roffproc proc; /* process new macro */
162: roffproc text; /* process as child text of macro */
163: roffproc sub; /* process as child of macro */
164: int flags;
165: #define ROFFMAC_STRUCT (1 << 0) /* always interpret */
1.1 schwarze 166: };
167:
1.37 schwarze 168: struct predef {
169: const char *name; /* predefined input name */
170: const char *str; /* replacement symbol */
171: };
172:
173: #define PREDEF(__name, __str) \
174: { (__name), (__str) },
175:
1.138 schwarze 176: /* --- function prototypes ------------------------------------------------ */
177:
1.212 schwarze 178: static int roffnode_cleanscope(struct roff *);
179: static int roffnode_pop(struct roff *);
1.166 schwarze 180: static void roffnode_push(struct roff *, enum roff_tok,
1.42 schwarze 181: const char *, int, int);
1.218 schwarze 182: static void roff_addtbl(struct roff_man *, int, struct tbl_node *);
1.212 schwarze 183: static int roff_als(ROFF_ARGS);
184: static int roff_block(ROFF_ARGS);
185: static int roff_block_text(ROFF_ARGS);
186: static int roff_block_sub(ROFF_ARGS);
1.236 schwarze 187: static int roff_break(ROFF_ARGS);
1.212 schwarze 188: static int roff_cblock(ROFF_ARGS);
189: static int roff_cc(ROFF_ARGS);
190: static int roff_ccond(struct roff *, int, int);
1.213 schwarze 191: static int roff_char(ROFF_ARGS);
1.212 schwarze 192: static int roff_cond(ROFF_ARGS);
1.247 schwarze 193: static int roff_cond_checkend(ROFF_ARGS);
1.212 schwarze 194: static int roff_cond_text(ROFF_ARGS);
195: static int roff_cond_sub(ROFF_ARGS);
196: static int roff_ds(ROFF_ARGS);
197: static int roff_ec(ROFF_ARGS);
198: static int roff_eo(ROFF_ARGS);
199: static int roff_eqndelim(struct roff *, struct buf *, int);
1.244 schwarze 200: static int roff_evalcond(struct roff *, int, char *, int *);
1.106 schwarze 201: static int roff_evalnum(struct roff *, int,
202: const char *, int *, int *, int);
203: static int roff_evalpar(struct roff *, int,
1.133 schwarze 204: const char *, int *, int *, int);
1.71 schwarze 205: static int roff_evalstrcond(const char *, int *);
1.227 schwarze 206: static int roff_expand(struct roff *, struct buf *,
207: int, int, char);
1.260 schwarze 208: static void roff_expand_patch(struct buf *, int,
209: const char *, int);
1.42 schwarze 210: static void roff_free1(struct roff *);
1.52 schwarze 211: static void roff_freereg(struct roffreg *);
1.42 schwarze 212: static void roff_freestr(struct roffkv *);
1.85 schwarze 213: static size_t roff_getname(struct roff *, char **, int, int);
1.133 schwarze 214: static int roff_getnum(const char *, int *, int *, int);
1.56 schwarze 215: static int roff_getop(const char *, int *, char *);
1.199 schwarze 216: static int roff_getregn(struct roff *,
217: const char *, size_t, char);
1.145 schwarze 218: static int roff_getregro(const struct roff *,
219: const char *name);
1.197 schwarze 220: static const char *roff_getstrn(struct roff *,
1.187 schwarze 221: const char *, size_t, int *);
1.143 schwarze 222: static int roff_hasregn(const struct roff *,
223: const char *, size_t);
1.212 schwarze 224: static int roff_insec(ROFF_ARGS);
225: static int roff_it(ROFF_ARGS);
226: static int roff_line_ignore(ROFF_ARGS);
1.137 schwarze 227: static void roff_man_alloc1(struct roff_man *);
228: static void roff_man_free1(struct roff_man *);
1.212 schwarze 229: static int roff_manyarg(ROFF_ARGS);
1.256 schwarze 230: static int roff_mc(ROFF_ARGS);
1.230 schwarze 231: static int roff_noarg(ROFF_ARGS);
1.212 schwarze 232: static int roff_nop(ROFF_ARGS);
233: static int roff_nr(ROFF_ARGS);
234: static int roff_onearg(ROFF_ARGS);
1.166 schwarze 235: static enum roff_tok roff_parse(struct roff *, char *, int *,
1.87 schwarze 236: int, int);
1.259 schwarze 237: static int roff_parse_comment(struct roff *, struct buf *,
238: int, int, char);
1.212 schwarze 239: static int roff_parsetext(struct roff *, struct buf *,
1.177 schwarze 240: int, int *);
1.212 schwarze 241: static int roff_renamed(ROFF_ARGS);
1.258 schwarze 242: static int roff_req_or_macro(ROFF_ARGS);
1.212 schwarze 243: static int roff_return(ROFF_ARGS);
244: static int roff_rm(ROFF_ARGS);
245: static int roff_rn(ROFF_ARGS);
246: static int roff_rr(ROFF_ARGS);
1.198 schwarze 247: static void roff_setregn(struct roff *, const char *,
1.199 schwarze 248: size_t, int, char, int);
1.8 schwarze 249: static void roff_setstr(struct roff *,
1.16 schwarze 250: const char *, const char *, int);
1.80 schwarze 251: static void roff_setstrn(struct roffkv **, const char *,
1.42 schwarze 252: size_t, const char *, size_t, int);
1.212 schwarze 253: static int roff_shift(ROFF_ARGS);
254: static int roff_so(ROFF_ARGS);
255: static int roff_tr(ROFF_ARGS);
256: static int roff_Dd(ROFF_ARGS);
257: static int roff_TE(ROFF_ARGS);
258: static int roff_TS(ROFF_ARGS);
259: static int roff_EQ(ROFF_ARGS);
260: static int roff_EN(ROFF_ARGS);
261: static int roff_T_(ROFF_ARGS);
262: static int roff_unsupp(ROFF_ARGS);
263: static int roff_userdef(ROFF_ARGS);
1.1 schwarze 264:
1.138 schwarze 265: /* --- constant data ------------------------------------------------------ */
266:
1.133 schwarze 267: #define ROFFNUM_SCALE (1 << 0) /* Honour scaling in roff_getnum(). */
268: #define ROFFNUM_WHITE (1 << 1) /* Skip whitespace in roff_evalnum(). */
269:
1.166 schwarze 270: const char *__roff_name[MAN_MAX + 1] = {
1.230 schwarze 271: "br", "ce", "fi", "ft",
272: "ll", "mc", "nf",
273: "po", "rj", "sp",
1.184 schwarze 274: "ta", "ti", NULL,
1.166 schwarze 275: "ab", "ad", "af", "aln",
276: "als", "am", "am1", "ami",
277: "ami1", "as", "as1", "asciify",
278: "backtrace", "bd", "bleedat", "blm",
279: "box", "boxa", "bp", "BP",
280: "break", "breakchar", "brnl", "brp",
1.177 schwarze 281: "brpnl", "c2", "cc",
1.166 schwarze 282: "cf", "cflags", "ch", "char",
283: "chop", "class", "close", "CL",
284: "color", "composite", "continue", "cp",
285: "cropat", "cs", "cu", "da",
286: "dch", "Dd", "de", "de1",
287: "defcolor", "dei", "dei1", "device",
288: "devicem", "di", "do", "ds",
289: "ds1", "dwh", "dt", "ec",
290: "ecr", "ecs", "el", "em",
291: "EN", "eo", "EP", "EQ",
292: "errprint", "ev", "evc", "ex",
293: "fallback", "fam", "fc", "fchar",
294: "fcolor", "fdeferlig", "feature", "fkern",
295: "fl", "flig", "fp", "fps",
296: "fschar", "fspacewidth", "fspecial", "ftr",
297: "fzoom", "gcolor", "hc", "hcode",
298: "hidechar", "hla", "hlm", "hpf",
299: "hpfa", "hpfcode", "hw", "hy",
300: "hylang", "hylen", "hym", "hypp",
301: "hys", "ie", "if", "ig",
302: "index", "it", "itc", "IX",
303: "kern", "kernafter", "kernbefore", "kernpair",
304: "lc", "lc_ctype", "lds", "length",
305: "letadj", "lf", "lg", "lhang",
306: "linetabs", "lnr", "lnrf", "lpfx",
1.176 schwarze 307: "ls", "lsm", "lt",
1.166 schwarze 308: "mediasize", "minss", "mk", "mso",
309: "na", "ne", "nh", "nhychar",
310: "nm", "nn", "nop", "nr",
311: "nrf", "nroff", "ns", "nx",
312: "open", "opena", "os", "output",
313: "padj", "papersize", "pc", "pev",
314: "pi", "PI", "pl", "pm",
1.184 schwarze 315: "pn", "pnr", "ps",
1.166 schwarze 316: "psbb", "pshape", "pso", "ptr",
317: "pvs", "rchar", "rd", "recursionlimit",
1.181 schwarze 318: "return", "rfschar", "rhang",
1.166 schwarze 319: "rm", "rn", "rnn", "rr",
320: "rs", "rt", "schar", "sentchar",
321: "shc", "shift", "sizes", "so",
322: "spacewidth", "special", "spreadwarn", "ss",
323: "sty", "substring", "sv", "sy",
1.172 schwarze 324: "T&", "tc", "TE",
1.173 schwarze 325: "TH", "tkf", "tl",
1.166 schwarze 326: "tm", "tm1", "tmc", "tr",
327: "track", "transchar", "trf", "trimat",
328: "trin", "trnt", "troff", "TS",
329: "uf", "ul", "unformat", "unwatch",
330: "unwatchn", "vpt", "vs", "warn",
331: "warnscale", "watch", "watchlength", "watchn",
332: "wh", "while", "write", "writec",
333: "writem", "xflag", ".", NULL,
1.178 schwarze 334: NULL, "text",
1.166 schwarze 335: "Dd", "Dt", "Os", "Sh",
336: "Ss", "Pp", "D1", "Dl",
337: "Bd", "Ed", "Bl", "El",
338: "It", "Ad", "An", "Ap",
339: "Ar", "Cd", "Cm", "Dv",
340: "Er", "Ev", "Ex", "Fa",
341: "Fd", "Fl", "Fn", "Ft",
342: "Ic", "In", "Li", "Nd",
343: "Nm", "Op", "Ot", "Pa",
344: "Rv", "St", "Va", "Vt",
345: "Xr", "%A", "%B", "%D",
346: "%I", "%J", "%N", "%O",
347: "%P", "%R", "%T", "%V",
348: "Ac", "Ao", "Aq", "At",
349: "Bc", "Bf", "Bo", "Bq",
350: "Bsx", "Bx", "Db", "Dc",
351: "Do", "Dq", "Ec", "Ef",
352: "Em", "Eo", "Fx", "Ms",
353: "No", "Ns", "Nx", "Ox",
354: "Pc", "Pf", "Po", "Pq",
355: "Qc", "Ql", "Qo", "Qq",
356: "Re", "Rs", "Sc", "So",
357: "Sq", "Sm", "Sx", "Sy",
358: "Tn", "Ux", "Xc", "Xo",
359: "Fo", "Fc", "Oo", "Oc",
360: "Bk", "Ek", "Bt", "Hf",
361: "Fr", "Ud", "Lb", "Lp",
362: "Lk", "Mt", "Brq", "Bro",
363: "Brc", "%C", "Es", "En",
1.171 schwarze 364: "Dx", "%Q", "%U", "Ta",
1.241 schwarze 365: "Tg", NULL,
1.166 schwarze 366: "TH", "SH", "SS", "TP",
1.204 schwarze 367: "TQ",
1.166 schwarze 368: "LP", "PP", "P", "IP",
369: "HP", "SM", "SB", "BI",
370: "IB", "BR", "RB", "R",
371: "B", "I", "IR", "RI",
372: "RE", "RS", "DT", "UC",
1.169 schwarze 373: "PD", "AT", "in",
1.205 schwarze 374: "SY", "YS", "OP",
375: "EX", "EE", "UR",
1.189 bentley 376: "UE", "MT", "ME", NULL
1.166 schwarze 377: };
378: const char *const *roff_name = __roff_name;
379:
380: static struct roffmac roffs[TOKEN_NONE] = {
1.230 schwarze 381: { roff_noarg, NULL, NULL, 0 }, /* br */
1.177 schwarze 382: { roff_onearg, NULL, NULL, 0 }, /* ce */
1.230 schwarze 383: { roff_noarg, NULL, NULL, 0 }, /* fi */
1.169 schwarze 384: { roff_onearg, NULL, NULL, 0 }, /* ft */
1.170 schwarze 385: { roff_onearg, NULL, NULL, 0 }, /* ll */
1.256 schwarze 386: { roff_mc, NULL, NULL, 0 }, /* mc */
1.230 schwarze 387: { roff_noarg, NULL, NULL, 0 }, /* nf */
1.184 schwarze 388: { roff_onearg, NULL, NULL, 0 }, /* po */
1.181 schwarze 389: { roff_onearg, NULL, NULL, 0 }, /* rj */
1.171 schwarze 390: { roff_onearg, NULL, NULL, 0 }, /* sp */
1.172 schwarze 391: { roff_manyarg, NULL, NULL, 0 }, /* ta */
1.173 schwarze 392: { roff_onearg, NULL, NULL, 0 }, /* ti */
1.168 schwarze 393: { NULL, NULL, NULL, 0 }, /* ROFF_MAX */
1.167 schwarze 394: { roff_unsupp, NULL, NULL, 0 }, /* ab */
395: { roff_line_ignore, NULL, NULL, 0 }, /* ad */
396: { roff_line_ignore, NULL, NULL, 0 }, /* af */
397: { roff_unsupp, NULL, NULL, 0 }, /* aln */
1.183 schwarze 398: { roff_als, NULL, NULL, 0 }, /* als */
1.167 schwarze 399: { roff_block, roff_block_text, roff_block_sub, 0 }, /* am */
400: { roff_block, roff_block_text, roff_block_sub, 0 }, /* am1 */
401: { roff_block, roff_block_text, roff_block_sub, 0 }, /* ami */
402: { roff_block, roff_block_text, roff_block_sub, 0 }, /* ami1 */
403: { roff_ds, NULL, NULL, 0 }, /* as */
404: { roff_ds, NULL, NULL, 0 }, /* as1 */
405: { roff_unsupp, NULL, NULL, 0 }, /* asciify */
406: { roff_line_ignore, NULL, NULL, 0 }, /* backtrace */
407: { roff_line_ignore, NULL, NULL, 0 }, /* bd */
408: { roff_line_ignore, NULL, NULL, 0 }, /* bleedat */
409: { roff_unsupp, NULL, NULL, 0 }, /* blm */
410: { roff_unsupp, NULL, NULL, 0 }, /* box */
411: { roff_unsupp, NULL, NULL, 0 }, /* boxa */
412: { roff_line_ignore, NULL, NULL, 0 }, /* bp */
413: { roff_unsupp, NULL, NULL, 0 }, /* BP */
1.236 schwarze 414: { roff_break, NULL, NULL, 0 }, /* break */
1.167 schwarze 415: { roff_line_ignore, NULL, NULL, 0 }, /* breakchar */
416: { roff_line_ignore, NULL, NULL, 0 }, /* brnl */
1.230 schwarze 417: { roff_noarg, NULL, NULL, 0 }, /* brp */
1.167 schwarze 418: { roff_line_ignore, NULL, NULL, 0 }, /* brpnl */
419: { roff_unsupp, NULL, NULL, 0 }, /* c2 */
420: { roff_cc, NULL, NULL, 0 }, /* cc */
421: { roff_insec, NULL, NULL, 0 }, /* cf */
422: { roff_line_ignore, NULL, NULL, 0 }, /* cflags */
423: { roff_line_ignore, NULL, NULL, 0 }, /* ch */
1.213 schwarze 424: { roff_char, NULL, NULL, 0 }, /* char */
1.167 schwarze 425: { roff_unsupp, NULL, NULL, 0 }, /* chop */
426: { roff_line_ignore, NULL, NULL, 0 }, /* class */
427: { roff_insec, NULL, NULL, 0 }, /* close */
428: { roff_unsupp, NULL, NULL, 0 }, /* CL */
429: { roff_line_ignore, NULL, NULL, 0 }, /* color */
430: { roff_unsupp, NULL, NULL, 0 }, /* composite */
431: { roff_unsupp, NULL, NULL, 0 }, /* continue */
432: { roff_line_ignore, NULL, NULL, 0 }, /* cp */
433: { roff_line_ignore, NULL, NULL, 0 }, /* cropat */
434: { roff_line_ignore, NULL, NULL, 0 }, /* cs */
435: { roff_line_ignore, NULL, NULL, 0 }, /* cu */
436: { roff_unsupp, NULL, NULL, 0 }, /* da */
437: { roff_unsupp, NULL, NULL, 0 }, /* dch */
438: { roff_Dd, NULL, NULL, 0 }, /* Dd */
439: { roff_block, roff_block_text, roff_block_sub, 0 }, /* de */
440: { roff_block, roff_block_text, roff_block_sub, 0 }, /* de1 */
441: { roff_line_ignore, NULL, NULL, 0 }, /* defcolor */
442: { roff_block, roff_block_text, roff_block_sub, 0 }, /* dei */
443: { roff_block, roff_block_text, roff_block_sub, 0 }, /* dei1 */
444: { roff_unsupp, NULL, NULL, 0 }, /* device */
445: { roff_unsupp, NULL, NULL, 0 }, /* devicem */
446: { roff_unsupp, NULL, NULL, 0 }, /* di */
447: { roff_unsupp, NULL, NULL, 0 }, /* do */
448: { roff_ds, NULL, NULL, 0 }, /* ds */
449: { roff_ds, NULL, NULL, 0 }, /* ds1 */
450: { roff_unsupp, NULL, NULL, 0 }, /* dwh */
451: { roff_unsupp, NULL, NULL, 0 }, /* dt */
1.175 schwarze 452: { roff_ec, NULL, NULL, 0 }, /* ec */
1.167 schwarze 453: { roff_unsupp, NULL, NULL, 0 }, /* ecr */
454: { roff_unsupp, NULL, NULL, 0 }, /* ecs */
455: { roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT }, /* el */
456: { roff_unsupp, NULL, NULL, 0 }, /* em */
457: { roff_EN, NULL, NULL, 0 }, /* EN */
1.175 schwarze 458: { roff_eo, NULL, NULL, 0 }, /* eo */
1.167 schwarze 459: { roff_unsupp, NULL, NULL, 0 }, /* EP */
460: { roff_EQ, NULL, NULL, 0 }, /* EQ */
461: { roff_line_ignore, NULL, NULL, 0 }, /* errprint */
462: { roff_unsupp, NULL, NULL, 0 }, /* ev */
463: { roff_unsupp, NULL, NULL, 0 }, /* evc */
464: { roff_unsupp, NULL, NULL, 0 }, /* ex */
465: { roff_line_ignore, NULL, NULL, 0 }, /* fallback */
466: { roff_line_ignore, NULL, NULL, 0 }, /* fam */
467: { roff_unsupp, NULL, NULL, 0 }, /* fc */
468: { roff_unsupp, NULL, NULL, 0 }, /* fchar */
469: { roff_line_ignore, NULL, NULL, 0 }, /* fcolor */
470: { roff_line_ignore, NULL, NULL, 0 }, /* fdeferlig */
471: { roff_line_ignore, NULL, NULL, 0 }, /* feature */
472: { roff_line_ignore, NULL, NULL, 0 }, /* fkern */
473: { roff_line_ignore, NULL, NULL, 0 }, /* fl */
474: { roff_line_ignore, NULL, NULL, 0 }, /* flig */
475: { roff_line_ignore, NULL, NULL, 0 }, /* fp */
476: { roff_line_ignore, NULL, NULL, 0 }, /* fps */
477: { roff_unsupp, NULL, NULL, 0 }, /* fschar */
478: { roff_line_ignore, NULL, NULL, 0 }, /* fspacewidth */
479: { roff_line_ignore, NULL, NULL, 0 }, /* fspecial */
480: { roff_line_ignore, NULL, NULL, 0 }, /* ftr */
481: { roff_line_ignore, NULL, NULL, 0 }, /* fzoom */
482: { roff_line_ignore, NULL, NULL, 0 }, /* gcolor */
483: { roff_line_ignore, NULL, NULL, 0 }, /* hc */
484: { roff_line_ignore, NULL, NULL, 0 }, /* hcode */
485: { roff_line_ignore, NULL, NULL, 0 }, /* hidechar */
486: { roff_line_ignore, NULL, NULL, 0 }, /* hla */
487: { roff_line_ignore, NULL, NULL, 0 }, /* hlm */
488: { roff_line_ignore, NULL, NULL, 0 }, /* hpf */
489: { roff_line_ignore, NULL, NULL, 0 }, /* hpfa */
490: { roff_line_ignore, NULL, NULL, 0 }, /* hpfcode */
491: { roff_line_ignore, NULL, NULL, 0 }, /* hw */
492: { roff_line_ignore, NULL, NULL, 0 }, /* hy */
493: { roff_line_ignore, NULL, NULL, 0 }, /* hylang */
494: { roff_line_ignore, NULL, NULL, 0 }, /* hylen */
495: { roff_line_ignore, NULL, NULL, 0 }, /* hym */
496: { roff_line_ignore, NULL, NULL, 0 }, /* hypp */
497: { roff_line_ignore, NULL, NULL, 0 }, /* hys */
498: { roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT }, /* ie */
499: { roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT }, /* if */
500: { roff_block, roff_block_text, roff_block_sub, 0 }, /* ig */
501: { roff_unsupp, NULL, NULL, 0 }, /* index */
502: { roff_it, NULL, NULL, 0 }, /* it */
503: { roff_unsupp, NULL, NULL, 0 }, /* itc */
504: { roff_line_ignore, NULL, NULL, 0 }, /* IX */
505: { roff_line_ignore, NULL, NULL, 0 }, /* kern */
506: { roff_line_ignore, NULL, NULL, 0 }, /* kernafter */
507: { roff_line_ignore, NULL, NULL, 0 }, /* kernbefore */
508: { roff_line_ignore, NULL, NULL, 0 }, /* kernpair */
509: { roff_unsupp, NULL, NULL, 0 }, /* lc */
510: { roff_unsupp, NULL, NULL, 0 }, /* lc_ctype */
511: { roff_unsupp, NULL, NULL, 0 }, /* lds */
512: { roff_unsupp, NULL, NULL, 0 }, /* length */
513: { roff_line_ignore, NULL, NULL, 0 }, /* letadj */
514: { roff_insec, NULL, NULL, 0 }, /* lf */
515: { roff_line_ignore, NULL, NULL, 0 }, /* lg */
516: { roff_line_ignore, NULL, NULL, 0 }, /* lhang */
517: { roff_unsupp, NULL, NULL, 0 }, /* linetabs */
518: { roff_unsupp, NULL, NULL, 0 }, /* lnr */
519: { roff_unsupp, NULL, NULL, 0 }, /* lnrf */
520: { roff_unsupp, NULL, NULL, 0 }, /* lpfx */
521: { roff_line_ignore, NULL, NULL, 0 }, /* ls */
522: { roff_unsupp, NULL, NULL, 0 }, /* lsm */
523: { roff_line_ignore, NULL, NULL, 0 }, /* lt */
524: { roff_line_ignore, NULL, NULL, 0 }, /* mediasize */
525: { roff_line_ignore, NULL, NULL, 0 }, /* minss */
526: { roff_line_ignore, NULL, NULL, 0 }, /* mk */
527: { roff_insec, NULL, NULL, 0 }, /* mso */
528: { roff_line_ignore, NULL, NULL, 0 }, /* na */
529: { roff_line_ignore, NULL, NULL, 0 }, /* ne */
530: { roff_line_ignore, NULL, NULL, 0 }, /* nh */
531: { roff_line_ignore, NULL, NULL, 0 }, /* nhychar */
532: { roff_unsupp, NULL, NULL, 0 }, /* nm */
533: { roff_unsupp, NULL, NULL, 0 }, /* nn */
1.202 schwarze 534: { roff_nop, NULL, NULL, 0 }, /* nop */
1.167 schwarze 535: { roff_nr, NULL, NULL, 0 }, /* nr */
536: { roff_unsupp, NULL, NULL, 0 }, /* nrf */
537: { roff_line_ignore, NULL, NULL, 0 }, /* nroff */
538: { roff_line_ignore, NULL, NULL, 0 }, /* ns */
539: { roff_insec, NULL, NULL, 0 }, /* nx */
540: { roff_insec, NULL, NULL, 0 }, /* open */
541: { roff_insec, NULL, NULL, 0 }, /* opena */
542: { roff_line_ignore, NULL, NULL, 0 }, /* os */
543: { roff_unsupp, NULL, NULL, 0 }, /* output */
544: { roff_line_ignore, NULL, NULL, 0 }, /* padj */
545: { roff_line_ignore, NULL, NULL, 0 }, /* papersize */
546: { roff_line_ignore, NULL, NULL, 0 }, /* pc */
547: { roff_line_ignore, NULL, NULL, 0 }, /* pev */
548: { roff_insec, NULL, NULL, 0 }, /* pi */
549: { roff_unsupp, NULL, NULL, 0 }, /* PI */
550: { roff_line_ignore, NULL, NULL, 0 }, /* pl */
551: { roff_line_ignore, NULL, NULL, 0 }, /* pm */
552: { roff_line_ignore, NULL, NULL, 0 }, /* pn */
553: { roff_line_ignore, NULL, NULL, 0 }, /* pnr */
554: { roff_line_ignore, NULL, NULL, 0 }, /* ps */
555: { roff_unsupp, NULL, NULL, 0 }, /* psbb */
556: { roff_unsupp, NULL, NULL, 0 }, /* pshape */
557: { roff_insec, NULL, NULL, 0 }, /* pso */
558: { roff_line_ignore, NULL, NULL, 0 }, /* ptr */
559: { roff_line_ignore, NULL, NULL, 0 }, /* pvs */
560: { roff_unsupp, NULL, NULL, 0 }, /* rchar */
561: { roff_line_ignore, NULL, NULL, 0 }, /* rd */
562: { roff_line_ignore, NULL, NULL, 0 }, /* recursionlimit */
1.211 schwarze 563: { roff_return, NULL, NULL, 0 }, /* return */
1.167 schwarze 564: { roff_unsupp, NULL, NULL, 0 }, /* rfschar */
565: { roff_line_ignore, NULL, NULL, 0 }, /* rhang */
566: { roff_rm, NULL, NULL, 0 }, /* rm */
1.178 schwarze 567: { roff_rn, NULL, NULL, 0 }, /* rn */
1.167 schwarze 568: { roff_unsupp, NULL, NULL, 0 }, /* rnn */
569: { roff_rr, NULL, NULL, 0 }, /* rr */
570: { roff_line_ignore, NULL, NULL, 0 }, /* rs */
571: { roff_line_ignore, NULL, NULL, 0 }, /* rt */
572: { roff_unsupp, NULL, NULL, 0 }, /* schar */
573: { roff_line_ignore, NULL, NULL, 0 }, /* sentchar */
574: { roff_line_ignore, NULL, NULL, 0 }, /* shc */
1.211 schwarze 575: { roff_shift, NULL, NULL, 0 }, /* shift */
1.167 schwarze 576: { roff_line_ignore, NULL, NULL, 0 }, /* sizes */
577: { roff_so, NULL, NULL, 0 }, /* so */
578: { roff_line_ignore, NULL, NULL, 0 }, /* spacewidth */
579: { roff_line_ignore, NULL, NULL, 0 }, /* special */
580: { roff_line_ignore, NULL, NULL, 0 }, /* spreadwarn */
581: { roff_line_ignore, NULL, NULL, 0 }, /* ss */
582: { roff_line_ignore, NULL, NULL, 0 }, /* sty */
583: { roff_unsupp, NULL, NULL, 0 }, /* substring */
584: { roff_line_ignore, NULL, NULL, 0 }, /* sv */
585: { roff_insec, NULL, NULL, 0 }, /* sy */
586: { roff_T_, NULL, NULL, 0 }, /* T& */
587: { roff_unsupp, NULL, NULL, 0 }, /* tc */
588: { roff_TE, NULL, NULL, 0 }, /* TE */
1.187 schwarze 589: { roff_Dd, NULL, NULL, 0 }, /* TH */
1.167 schwarze 590: { roff_line_ignore, NULL, NULL, 0 }, /* tkf */
591: { roff_unsupp, NULL, NULL, 0 }, /* tl */
592: { roff_line_ignore, NULL, NULL, 0 }, /* tm */
593: { roff_line_ignore, NULL, NULL, 0 }, /* tm1 */
594: { roff_line_ignore, NULL, NULL, 0 }, /* tmc */
595: { roff_tr, NULL, NULL, 0 }, /* tr */
596: { roff_line_ignore, NULL, NULL, 0 }, /* track */
597: { roff_line_ignore, NULL, NULL, 0 }, /* transchar */
598: { roff_insec, NULL, NULL, 0 }, /* trf */
599: { roff_line_ignore, NULL, NULL, 0 }, /* trimat */
600: { roff_unsupp, NULL, NULL, 0 }, /* trin */
601: { roff_unsupp, NULL, NULL, 0 }, /* trnt */
602: { roff_line_ignore, NULL, NULL, 0 }, /* troff */
603: { roff_TS, NULL, NULL, 0 }, /* TS */
604: { roff_line_ignore, NULL, NULL, 0 }, /* uf */
605: { roff_line_ignore, NULL, NULL, 0 }, /* ul */
606: { roff_unsupp, NULL, NULL, 0 }, /* unformat */
607: { roff_line_ignore, NULL, NULL, 0 }, /* unwatch */
608: { roff_line_ignore, NULL, NULL, 0 }, /* unwatchn */
609: { roff_line_ignore, NULL, NULL, 0 }, /* vpt */
610: { roff_line_ignore, NULL, NULL, 0 }, /* vs */
611: { roff_line_ignore, NULL, NULL, 0 }, /* warn */
612: { roff_line_ignore, NULL, NULL, 0 }, /* warnscale */
613: { roff_line_ignore, NULL, NULL, 0 }, /* watch */
614: { roff_line_ignore, NULL, NULL, 0 }, /* watchlength */
615: { roff_line_ignore, NULL, NULL, 0 }, /* watchn */
616: { roff_unsupp, NULL, NULL, 0 }, /* wh */
1.212 schwarze 617: { roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT }, /*while*/
1.167 schwarze 618: { roff_insec, NULL, NULL, 0 }, /* write */
619: { roff_insec, NULL, NULL, 0 }, /* writec */
620: { roff_insec, NULL, NULL, 0 }, /* writem */
621: { roff_line_ignore, NULL, NULL, 0 }, /* xflag */
622: { roff_cblock, NULL, NULL, 0 }, /* . */
1.178 schwarze 623: { roff_renamed, NULL, NULL, 0 },
1.167 schwarze 624: { roff_userdef, NULL, NULL, 0 }
1.1 schwarze 625: };
626:
1.37 schwarze 627: /* Array of injected predefined strings. */
628: #define PREDEFS_MAX 38
629: static const struct predef predefs[PREDEFS_MAX] = {
630: #include "predefs.in"
631: };
632:
1.177 schwarze 633: static int roffce_lines; /* number of input lines to center */
634: static struct roff_node *roffce_node; /* active request */
1.51 schwarze 635: static int roffit_lines; /* number of lines to delay */
636: static char *roffit_macro; /* nil-terminated macro line */
637:
1.80 schwarze 638:
1.138 schwarze 639: /* --- request table ------------------------------------------------------ */
640:
1.167 schwarze 641: struct ohash *
642: roffhash_alloc(enum roff_tok mintok, enum roff_tok maxtok)
1.3 schwarze 643: {
1.167 schwarze 644: struct ohash *htab;
645: struct roffreq *req;
646: enum roff_tok tok;
647: size_t sz;
648: unsigned int slot;
1.3 schwarze 649:
1.167 schwarze 650: htab = mandoc_malloc(sizeof(*htab));
651: mandoc_ohash_init(htab, 8, offsetof(struct roffreq, name));
1.3 schwarze 652:
1.167 schwarze 653: for (tok = mintok; tok < maxtok; tok++) {
1.168 schwarze 654: if (roff_name[tok] == NULL)
655: continue;
1.167 schwarze 656: sz = strlen(roff_name[tok]);
657: req = mandoc_malloc(sizeof(*req) + sz + 1);
658: req->tok = tok;
659: memcpy(req->name, roff_name[tok], sz + 1);
660: slot = ohash_qlookup(htab, req->name);
661: ohash_insert(htab, slot, req);
1.3 schwarze 662: }
1.167 schwarze 663: return htab;
1.3 schwarze 664: }
665:
1.167 schwarze 666: void
667: roffhash_free(struct ohash *htab)
1.1 schwarze 668: {
1.167 schwarze 669: struct roffreq *req;
670: unsigned int slot;
1.1 schwarze 671:
1.167 schwarze 672: if (htab == NULL)
673: return;
674: for (req = ohash_first(htab, &slot); req != NULL;
675: req = ohash_next(htab, &slot))
676: free(req);
677: ohash_delete(htab);
678: free(htab);
679: }
680:
681: enum roff_tok
682: roffhash_find(struct ohash *htab, const char *name, size_t sz)
683: {
684: struct roffreq *req;
685: const char *end;
686:
687: if (sz) {
688: end = name + sz;
689: req = ohash_find(htab, ohash_qlookupi(htab, name, &end));
690: } else
691: req = ohash_find(htab, ohash_qlookup(htab, name));
692: return req == NULL ? TOKEN_NONE : req->tok;
1.1 schwarze 693: }
694:
1.138 schwarze 695: /* --- stack of request blocks -------------------------------------------- */
696:
1.1 schwarze 697: /*
698: * Pop the current node off of the stack of roff instructions currently
1.236 schwarze 699: * pending. Return 1 if it is a loop or 0 otherwise.
1.1 schwarze 700: */
1.212 schwarze 701: static int
1.1 schwarze 702: roffnode_pop(struct roff *r)
703: {
704: struct roffnode *p;
1.212 schwarze 705: int inloop;
1.1 schwarze 706:
1.80 schwarze 707: p = r->last;
1.212 schwarze 708: inloop = p->tok == ROFF_while;
709: r->last = p->parent;
1.16 schwarze 710: free(p->name);
711: free(p->end);
1.1 schwarze 712: free(p);
1.212 schwarze 713: return inloop;
1.1 schwarze 714: }
715:
716: /*
717: * Push a roff node onto the instruction stack. This must later be
718: * removed with roffnode_pop().
719: */
1.11 schwarze 720: static void
1.166 schwarze 721: roffnode_push(struct roff *r, enum roff_tok tok, const char *name,
1.16 schwarze 722: int line, int col)
1.1 schwarze 723: {
724: struct roffnode *p;
725:
1.11 schwarze 726: p = mandoc_calloc(1, sizeof(struct roffnode));
1.1 schwarze 727: p->tok = tok;
1.16 schwarze 728: if (name)
729: p->name = mandoc_strdup(name);
1.1 schwarze 730: p->parent = r->last;
731: p->line = line;
732: p->col = col;
1.71 schwarze 733: p->rule = p->parent ? p->parent->rule : 0;
1.1 schwarze 734:
735: r->last = p;
736: }
737:
1.138 schwarze 738: /* --- roff parser state data management ---------------------------------- */
739:
1.1 schwarze 740: static void
741: roff_free1(struct roff *r)
742: {
1.42 schwarze 743: int i;
1.27 schwarze 744:
1.218 schwarze 745: tbl_free(r->first_tbl);
1.27 schwarze 746: r->first_tbl = r->last_tbl = r->tbl = NULL;
1.1 schwarze 747:
1.219 schwarze 748: eqn_free(r->last_eqn);
1.191 schwarze 749: r->last_eqn = r->eqn = NULL;
1.32 schwarze 750:
1.211 schwarze 751: while (r->mstackpos >= 0)
752: roff_userret(r);
753:
1.1 schwarze 754: while (r->last)
755: roffnode_pop(r);
1.27 schwarze 756:
1.96 schwarze 757: free (r->rstack);
758: r->rstack = NULL;
759: r->rstacksz = 0;
760: r->rstackpos = -1;
761:
762: roff_freereg(r->regtab);
763: r->regtab = NULL;
764:
1.42 schwarze 765: roff_freestr(r->strtab);
1.178 schwarze 766: roff_freestr(r->rentab);
1.42 schwarze 767: roff_freestr(r->xmbtab);
1.178 schwarze 768: r->strtab = r->rentab = r->xmbtab = NULL;
1.42 schwarze 769:
770: if (r->xtab)
771: for (i = 0; i < 128; i++)
772: free(r->xtab[i].p);
773: free(r->xtab);
774: r->xtab = NULL;
1.1 schwarze 775: }
776:
777: void
778: roff_reset(struct roff *r)
779: {
780: roff_free1(r);
1.239 schwarze 781: r->options |= MPARSE_COMMENT;
1.99 schwarze 782: r->format = r->options & (MPARSE_MDOC | MPARSE_MAN);
1.175 schwarze 783: r->control = '\0';
784: r->escape = '\\';
1.179 schwarze 785: roffce_lines = 0;
786: roffce_node = NULL;
787: roffit_lines = 0;
788: roffit_macro = NULL;
1.1 schwarze 789: }
790:
791: void
792: roff_free(struct roff *r)
793: {
1.238 schwarze 794: int i;
1.211 schwarze 795:
1.1 schwarze 796: roff_free1(r);
1.211 schwarze 797: for (i = 0; i < r->mstacksz; i++)
798: free(r->mstack[i].argv);
799: free(r->mstack);
1.167 schwarze 800: roffhash_free(r->reqtab);
1.1 schwarze 801: free(r);
802: }
803:
804: struct roff *
1.223 schwarze 805: roff_alloc(int options)
1.1 schwarze 806: {
807: struct roff *r;
808:
1.11 schwarze 809: r = mandoc_calloc(1, sizeof(struct roff));
1.200 schwarze 810: r->reqtab = roffhash_alloc(0, ROFF_RENAMED);
1.239 schwarze 811: r->options = options | MPARSE_COMMENT;
1.99 schwarze 812: r->format = options & (MPARSE_MDOC | MPARSE_MAN);
1.211 schwarze 813: r->mstackpos = -1;
1.2 schwarze 814: r->rstackpos = -1;
1.175 schwarze 815: r->escape = '\\';
1.149 schwarze 816: return r;
1.137 schwarze 817: }
818:
1.138 schwarze 819: /* --- syntax tree state data management ---------------------------------- */
820:
1.137 schwarze 821: static void
822: roff_man_free1(struct roff_man *man)
823: {
1.228 schwarze 824: if (man->meta.first != NULL)
825: roff_node_delete(man, man->meta.first);
1.137 schwarze 826: free(man->meta.msec);
827: free(man->meta.vol);
828: free(man->meta.os);
829: free(man->meta.arch);
830: free(man->meta.title);
831: free(man->meta.name);
832: free(man->meta.date);
1.228 schwarze 833: free(man->meta.sodest);
1.137 schwarze 834: }
835:
1.229 schwarze 836: void
837: roff_state_reset(struct roff_man *man)
838: {
839: man->last = man->meta.first;
840: man->last_es = NULL;
841: man->flags = 0;
842: man->lastsec = man->lastnamed = SEC_NONE;
843: man->next = ROFF_NEXT_CHILD;
844: roff_setreg(man->roff, "nS", 0, '=');
845: }
846:
1.137 schwarze 847: static void
848: roff_man_alloc1(struct roff_man *man)
849: {
850: memset(&man->meta, 0, sizeof(man->meta));
1.228 schwarze 851: man->meta.first = mandoc_calloc(1, sizeof(*man->meta.first));
852: man->meta.first->type = ROFFT_ROOT;
853: man->meta.macroset = MACROSET_NONE;
1.229 schwarze 854: roff_state_reset(man);
1.137 schwarze 855: }
856:
857: void
858: roff_man_reset(struct roff_man *man)
859: {
860: roff_man_free1(man);
861: roff_man_alloc1(man);
862: }
863:
864: void
865: roff_man_free(struct roff_man *man)
866: {
867: roff_man_free1(man);
1.252 schwarze 868: free(man->os_r);
1.137 schwarze 869: free(man);
870: }
871:
872: struct roff_man *
1.223 schwarze 873: roff_man_alloc(struct roff *roff, const char *os_s, int quick)
1.137 schwarze 874: {
875: struct roff_man *man;
876:
877: man = mandoc_calloc(1, sizeof(*man));
878: man->roff = roff;
1.188 schwarze 879: man->os_s = os_s;
1.137 schwarze 880: man->quick = quick;
881: roff_man_alloc1(man);
1.168 schwarze 882: roff->man = man;
1.149 schwarze 883: return man;
1.1 schwarze 884: }
885:
1.138 schwarze 886: /* --- syntax tree handling ----------------------------------------------- */
887:
888: struct roff_node *
889: roff_node_alloc(struct roff_man *man, int line, int pos,
890: enum roff_type type, int tok)
891: {
892: struct roff_node *n;
893:
894: n = mandoc_calloc(1, sizeof(*n));
895: n->line = line;
896: n->pos = pos;
897: n->tok = tok;
898: n->type = type;
899: n->sec = man->lastsec;
900:
901: if (man->flags & MDOC_SYNOPSIS)
1.157 schwarze 902: n->flags |= NODE_SYNPRETTY;
1.138 schwarze 903: else
1.157 schwarze 904: n->flags &= ~NODE_SYNPRETTY;
1.232 schwarze 905: if ((man->flags & (ROFF_NOFILL | ROFF_NONOFILL)) == ROFF_NOFILL)
1.231 schwarze 906: n->flags |= NODE_NOFILL;
907: else
908: n->flags &= ~NODE_NOFILL;
1.138 schwarze 909: if (man->flags & MDOC_NEWLINE)
1.157 schwarze 910: n->flags |= NODE_LINE;
1.138 schwarze 911: man->flags &= ~MDOC_NEWLINE;
912:
1.149 schwarze 913: return n;
1.138 schwarze 914: }
915:
916: void
917: roff_node_append(struct roff_man *man, struct roff_node *n)
918: {
919:
920: switch (man->next) {
921: case ROFF_NEXT_SIBLING:
1.153 schwarze 922: if (man->last->next != NULL) {
923: n->next = man->last->next;
924: man->last->next->prev = n;
925: } else
926: man->last->parent->last = n;
1.138 schwarze 927: man->last->next = n;
928: n->prev = man->last;
929: n->parent = man->last->parent;
930: break;
931: case ROFF_NEXT_CHILD:
1.159 schwarze 932: if (man->last->child != NULL) {
933: n->next = man->last->child;
934: man->last->child->prev = n;
935: } else
936: man->last->last = n;
1.138 schwarze 937: man->last->child = n;
938: n->parent = man->last;
939: break;
940: default:
941: abort();
942: }
1.154 schwarze 943: man->last = n;
1.138 schwarze 944:
945: switch (n->type) {
946: case ROFFT_HEAD:
947: n->parent->head = n;
948: break;
949: case ROFFT_BODY:
1.154 schwarze 950: if (n->end != ENDBODY_NOT)
951: return;
1.138 schwarze 952: n->parent->body = n;
953: break;
954: case ROFFT_TAIL:
955: n->parent->tail = n;
956: break;
957: default:
1.154 schwarze 958: return;
1.138 schwarze 959: }
1.154 schwarze 960:
961: /*
962: * Copy over the normalised-data pointer of our parent. Not
963: * everybody has one, but copying a null pointer is fine.
964: */
965:
966: n->norm = n->parent->norm;
967: assert(n->parent->type == ROFFT_BLOCK);
1.138 schwarze 968: }
969:
1.139 schwarze 970: void
971: roff_word_alloc(struct roff_man *man, int line, int pos, const char *word)
972: {
973: struct roff_node *n;
974:
975: n = roff_node_alloc(man, line, pos, ROFFT_TEXT, TOKEN_NONE);
976: n->string = roff_strdup(man->roff, word);
977: roff_node_append(man, n);
1.158 schwarze 978: n->flags |= NODE_VALID | NODE_ENDED;
1.139 schwarze 979: man->next = ROFF_NEXT_SIBLING;
980: }
981:
982: void
983: roff_word_append(struct roff_man *man, const char *word)
984: {
985: struct roff_node *n;
986: char *addstr, *newstr;
987:
988: n = man->last;
989: addstr = roff_strdup(man->roff, word);
990: mandoc_asprintf(&newstr, "%s %s", n->string, addstr);
991: free(addstr);
992: free(n->string);
993: n->string = newstr;
994: man->next = ROFF_NEXT_SIBLING;
1.140 schwarze 995: }
996:
997: void
998: roff_elem_alloc(struct roff_man *man, int line, int pos, int tok)
999: {
1000: struct roff_node *n;
1001:
1002: n = roff_node_alloc(man, line, pos, ROFFT_ELEM, tok);
1003: roff_node_append(man, n);
1004: man->next = ROFF_NEXT_CHILD;
1005: }
1006:
1007: struct roff_node *
1008: roff_block_alloc(struct roff_man *man, int line, int pos, int tok)
1009: {
1010: struct roff_node *n;
1011:
1012: n = roff_node_alloc(man, line, pos, ROFFT_BLOCK, tok);
1013: roff_node_append(man, n);
1014: man->next = ROFF_NEXT_CHILD;
1.149 schwarze 1015: return n;
1.139 schwarze 1016: }
1017:
1.138 schwarze 1018: struct roff_node *
1019: roff_head_alloc(struct roff_man *man, int line, int pos, int tok)
1020: {
1021: struct roff_node *n;
1022:
1023: n = roff_node_alloc(man, line, pos, ROFFT_HEAD, tok);
1024: roff_node_append(man, n);
1025: man->next = ROFF_NEXT_CHILD;
1.149 schwarze 1026: return n;
1.138 schwarze 1027: }
1028:
1029: struct roff_node *
1030: roff_body_alloc(struct roff_man *man, int line, int pos, int tok)
1031: {
1032: struct roff_node *n;
1033:
1034: n = roff_node_alloc(man, line, pos, ROFFT_BODY, tok);
1035: roff_node_append(man, n);
1036: man->next = ROFF_NEXT_CHILD;
1.149 schwarze 1037: return n;
1.139 schwarze 1038: }
1039:
1.193 schwarze 1040: static void
1.218 schwarze 1041: roff_addtbl(struct roff_man *man, int line, struct tbl_node *tbl)
1.139 schwarze 1042: {
1043: struct roff_node *n;
1.218 schwarze 1044: struct tbl_span *span;
1.139 schwarze 1045:
1.228 schwarze 1046: if (man->meta.macroset == MACROSET_MAN)
1.174 schwarze 1047: man_breakscope(man, ROFF_TS);
1.193 schwarze 1048: while ((span = tbl_span(tbl)) != NULL) {
1.218 schwarze 1049: n = roff_node_alloc(man, line, 0, ROFFT_TBL, TOKEN_NONE);
1.193 schwarze 1050: n->span = span;
1051: roff_node_append(man, n);
1052: n->flags |= NODE_VALID | NODE_ENDED;
1053: man->next = ROFF_NEXT_SIBLING;
1054: }
1.138 schwarze 1055: }
1056:
1057: void
1058: roff_node_unlink(struct roff_man *man, struct roff_node *n)
1059: {
1060:
1061: /* Adjust siblings. */
1062:
1063: if (n->prev)
1064: n->prev->next = n->next;
1065: if (n->next)
1066: n->next->prev = n->prev;
1067:
1068: /* Adjust parent. */
1069:
1070: if (n->parent != NULL) {
1071: if (n->parent->child == n)
1072: n->parent->child = n->next;
1073: if (n->parent->last == n)
1074: n->parent->last = n->prev;
1075: }
1076:
1077: /* Adjust parse point. */
1078:
1079: if (man == NULL)
1080: return;
1081: if (man->last == n) {
1082: if (n->prev == NULL) {
1083: man->last = n->parent;
1084: man->next = ROFF_NEXT_CHILD;
1085: } else {
1086: man->last = n->prev;
1087: man->next = ROFF_NEXT_SIBLING;
1088: }
1089: }
1.228 schwarze 1090: if (man->meta.first == n)
1091: man->meta.first = NULL;
1.216 schwarze 1092: }
1093:
1094: void
1095: roff_node_relink(struct roff_man *man, struct roff_node *n)
1096: {
1097: roff_node_unlink(man, n);
1098: n->prev = n->next = NULL;
1099: roff_node_append(man, n);
1.138 schwarze 1100: }
1101:
1102: void
1103: roff_node_free(struct roff_node *n)
1104: {
1105:
1106: if (n->args != NULL)
1107: mdoc_argv_free(n->args);
1108: if (n->type == ROFFT_BLOCK || n->type == ROFFT_ELEM)
1109: free(n->norm);
1.219 schwarze 1110: eqn_box_free(n->eqn);
1.138 schwarze 1111: free(n->string);
1.246 schwarze 1112: free(n->tag);
1.138 schwarze 1113: free(n);
1114: }
1115:
1116: void
1117: roff_node_delete(struct roff_man *man, struct roff_node *n)
1118: {
1119:
1120: while (n->child != NULL)
1121: roff_node_delete(man, n->child);
1122: roff_node_unlink(man, n);
1123: roff_node_free(n);
1.242 schwarze 1124: }
1125:
1126: int
1127: roff_node_transparent(struct roff_node *n)
1128: {
1129: if (n == NULL)
1130: return 0;
1131: if (n->type == ROFFT_COMMENT || n->flags & NODE_NOPRT)
1132: return 1;
1.245 schwarze 1133: return roff_tok_transparent(n->tok);
1134: }
1135:
1136: int
1137: roff_tok_transparent(enum roff_tok tok)
1138: {
1139: switch (tok) {
1.242 schwarze 1140: case ROFF_ft:
1141: case ROFF_ll:
1142: case ROFF_mc:
1143: case ROFF_po:
1144: case ROFF_ta:
1145: case MDOC_Db:
1146: case MDOC_Es:
1147: case MDOC_Sm:
1148: case MDOC_Tg:
1149: case MAN_DT:
1150: case MAN_UC:
1151: case MAN_PD:
1152: case MAN_AT:
1153: return 1;
1154: default:
1155: return 0;
1156: }
1157: }
1158:
1159: struct roff_node *
1160: roff_node_child(struct roff_node *n)
1161: {
1162: for (n = n->child; roff_node_transparent(n); n = n->next)
1163: continue;
1164: return n;
1165: }
1166:
1167: struct roff_node *
1168: roff_node_prev(struct roff_node *n)
1169: {
1170: do {
1171: n = n->prev;
1172: } while (roff_node_transparent(n));
1173: return n;
1174: }
1175:
1176: struct roff_node *
1177: roff_node_next(struct roff_node *n)
1178: {
1179: do {
1180: n = n->next;
1181: } while (roff_node_transparent(n));
1182: return n;
1.141 schwarze 1183: }
1184:
1185: void
1186: deroff(char **dest, const struct roff_node *n)
1187: {
1188: char *cp;
1189: size_t sz;
1190:
1.243 schwarze 1191: if (n->string == NULL) {
1.141 schwarze 1192: for (n = n->child; n != NULL; n = n->next)
1193: deroff(dest, n);
1194: return;
1195: }
1196:
1.160 schwarze 1197: /* Skip leading whitespace. */
1.141 schwarze 1198:
1.160 schwarze 1199: for (cp = n->string; *cp != '\0'; cp++) {
1.161 schwarze 1200: if (cp[0] == '\\' && cp[1] != '\0' &&
1201: strchr(" %&0^|~", cp[1]) != NULL)
1.141 schwarze 1202: cp++;
1.160 schwarze 1203: else if ( ! isspace((unsigned char)*cp))
1.141 schwarze 1204: break;
1205: }
1206:
1.161 schwarze 1207: /* Skip trailing backslash. */
1208:
1209: sz = strlen(cp);
1.162 schwarze 1210: if (sz > 0 && cp[sz - 1] == '\\')
1.161 schwarze 1211: sz--;
1212:
1.141 schwarze 1213: /* Skip trailing whitespace. */
1214:
1.161 schwarze 1215: for (; sz; sz--)
1.141 schwarze 1216: if ( ! isspace((unsigned char)cp[sz-1]))
1217: break;
1218:
1219: /* Skip empty strings. */
1220:
1221: if (sz == 0)
1222: return;
1223:
1224: if (*dest == NULL) {
1225: *dest = mandoc_strndup(cp, sz);
1226: return;
1227: }
1228:
1229: mandoc_asprintf(&cp, "%s %*s", *dest, (int)sz, cp);
1230: free(*dest);
1231: *dest = cp;
1.138 schwarze 1232: }
1233:
1234: /* --- main functions of the roff parser ---------------------------------- */
1235:
1.260 schwarze 1236: /*
1237: * Save comments preceding the title macro, for example in order to
1238: * preserve Copyright and license headers in HTML output,
1239: * provide diagnostics about RCS ids and trailing whitespace in comments,
1240: * then discard comments including preceding whitespace.
1241: * This function also handles input line continuation.
1242: */
1.259 schwarze 1243: static int
1.260 schwarze 1244: roff_parse_comment(struct roff *r, struct buf *buf, int ln, int pos, char ec)
1.259 schwarze 1245: {
1246: struct roff_node *n; /* used for header comments */
1247: const char *start; /* start of the string to process */
1248: const char *cp; /* for RCS id parsing */
1249: char *stesc; /* start of an escape sequence ('\\') */
1250: char *ep; /* end of comment string */
1251: int rcsid; /* kind of RCS id seen */
1252:
1253: for (start = stesc = buf->buf + pos;; stesc++) {
1.260 schwarze 1254: /*
1255: * XXX Ugly hack: Remove the newline character that
1256: * mparse_buf_r() appended to mark the end of input
1257: * if it is not preceded by an escape character.
1258: */
1259: if (stesc[0] == '\n') {
1260: assert(stesc[1] == '\0');
1261: stesc[0] = '\0';
1262: }
1263:
1.259 schwarze 1264: /* The line ends without continuation or comment. */
1265: if (stesc[0] == '\0')
1266: return ROFF_CONT;
1267:
1268: /* Unescaped byte: skip it. */
1.260 schwarze 1269: if (stesc[0] != ec)
1.259 schwarze 1270: continue;
1271:
1.260 schwarze 1272: /*
1273: * XXX Ugly hack: Do not attempt to append another line
1274: * if the function mparse_buf_r() appended a newline
1275: * character to indicate the end of input.
1276: */
1277: if (stesc[1] == '\n') {
1278: assert(stesc[2] == '\0');
1279: stesc[0] = '\0';
1280: return ROFF_CONT;
1281: }
1282:
1283: /*
1284: * An escape character at the end of an input line
1285: * requests line continuation.
1286: */
1.259 schwarze 1287: if (stesc[1] == '\0') {
1288: stesc[0] = '\0';
1289: return ROFF_IGN | ROFF_APPEND;
1290: }
1291:
1292: /* Found a comment: process it. */
1293: if (stesc[1] == '"' || stesc[1] == '#')
1294: break;
1295:
1296: /* Escaped escape character: skip them both. */
1.260 schwarze 1297: if (stesc[1] == ec)
1.259 schwarze 1298: stesc++;
1299: }
1300:
1301: /* Look for an RCS id in the comment. */
1302:
1303: rcsid = 0;
1304: if ((cp = strstr(stesc + 2, "$" "OpenBSD")) != NULL) {
1305: rcsid = 1 << MANDOC_OS_OPENBSD;
1306: cp += 8;
1307: } else if ((cp = strstr(stesc + 2, "$" "NetBSD")) != NULL) {
1308: rcsid = 1 << MANDOC_OS_NETBSD;
1309: cp += 7;
1310: }
1311: if (cp != NULL && isalnum((unsigned char)*cp) == 0 &&
1312: strchr(cp, '$') != NULL) {
1313: if (r->man->meta.rcsids & rcsid)
1314: mandoc_msg(MANDOCERR_RCS_REP, ln,
1315: (int)(stesc - buf->buf) + 2, "%s", stesc + 1);
1316: r->man->meta.rcsids |= rcsid;
1317: }
1318:
1319: /* Warn about trailing whitespace at the end of the comment. */
1320:
1321: ep = strchr(stesc + 2, '\0') - 1;
1322: if (*ep == '\n')
1323: *ep-- = '\0';
1324: if (*ep == ' ' || *ep == '\t')
1325: mandoc_msg(MANDOCERR_SPACE_EOL,
1326: ln, (int)(ep - buf->buf), NULL);
1327:
1328: /* Save comments preceding the title macro in the syntax tree. */
1329:
1330: if (r->options & MPARSE_COMMENT) {
1331: while (*ep == ' ' || *ep == '\t')
1332: ep--;
1333: ep[1] = '\0';
1334: n = roff_node_alloc(r->man, ln, stesc + 1 - buf->buf,
1335: ROFFT_COMMENT, TOKEN_NONE);
1336: n->string = mandoc_strdup(stesc + 2);
1337: roff_node_append(r->man, n);
1338: n->flags |= NODE_VALID | NODE_ENDED;
1339: r->man->next = ROFF_NEXT_SIBLING;
1340: }
1341:
1342: /* The comment requests line continuation. */
1343:
1344: if (stesc[1] == '#') {
1345: *stesc = '\0';
1346: return ROFF_IGN | ROFF_APPEND;
1347: }
1348:
1349: /* Discard the comment including preceding whitespace. */
1350:
1351: while (stesc > start && stesc[-1] == ' ' &&
1352: (stesc == start + 1 || stesc[-2] != '\\'))
1353: stesc--;
1354: *stesc = '\0';
1355: return ROFF_CONT;
1356: }
1357:
1.8 schwarze 1358: /*
1.227 schwarze 1359: * In the current line, expand escape sequences that produce parsable
1360: * input text. Also check the syntax of the remaining escape sequences,
1361: * which typically produce output glyphs or change formatter state.
1.8 schwarze 1362: */
1.212 schwarze 1363: static int
1.260 schwarze 1364: roff_expand(struct roff *r, struct buf *buf, int ln, int pos, char ec)
1.8 schwarze 1365: {
1.260 schwarze 1366: char ubuf[24]; /* buffer to print a number */
1367: struct mctx *ctx; /* current macro call context */
1368: const char *res; /* the string to be pasted */
1369: const char *src; /* source for copying */
1370: char *dst; /* destination for copying */
1371: int iesc; /* index of leading escape char */
1372: int inam; /* index of the escape name */
1373: int iarg; /* index beginning the argument */
1374: int iendarg; /* index right after the argument */
1375: int iend; /* index right after the sequence */
1376: int deftype; /* type of definition to paste */
1377: int argi; /* macro argument index */
1378: int quote_args; /* true for \\$@, false for \\$* */
1379: int asz; /* length of the replacement */
1380: int rsz; /* length of the rest of the string */
1381: int npos; /* position in numeric expression */
1.53 schwarze 1382: int expand_count; /* to avoid infinite loops */
1.175 schwarze 1383:
1.43 schwarze 1384: expand_count = 0;
1.260 schwarze 1385: while (buf->buf[pos] != '\0') {
1.43 schwarze 1386:
1.260 schwarze 1387: /*
1388: * Skip plain ASCII characters.
1389: * If we have a non-standard escape character,
1390: * escape literal backslashes because all processing in
1391: * subsequent functions uses the standard escaping rules.
1392: */
1.78 schwarze 1393:
1.260 schwarze 1394: if (buf->buf[pos] != ec) {
1395: if (ec != ASCII_ESC && buf->buf[pos] == '\\') {
1396: roff_expand_patch(buf, pos, "\\e", pos + 1);
1397: pos++;
1.175 schwarze 1398: }
1.260 schwarze 1399: pos++;
1.78 schwarze 1400: continue;
1.175 schwarze 1401: }
1.78 schwarze 1402:
1.260 schwarze 1403: /*
1404: * Parse escape sequences,
1405: * issue diagnostic messages when appropriate,
1406: * and skip sequences that do not need expansion.
1407: * If we have a non-standard escape character, translate
1408: * it to backslashes and translate backslashes to \e.
1409: */
1.78 schwarze 1410:
1.264 ! schwarze 1411: if (roff_escape(buf->buf, ln, pos, &iesc, &inam,
! 1412: &iarg, &iendarg, &iend) != ESCAPE_EXPAND) {
1.260 schwarze 1413: while (pos < iend) {
1414: if (buf->buf[pos] == ec) {
1415: buf->buf[pos] = '\\';
1416: if (pos + 1 < iend)
1417: pos++;
1418: } else if (buf->buf[pos] == '\\') {
1419: roff_expand_patch(buf,
1420: pos, "\\e", pos + 1);
1421: pos++;
1422: iend++;
1423: }
1424: pos++;
1.224 schwarze 1425: }
1.78 schwarze 1426: continue;
1.42 schwarze 1427: }
1.260 schwarze 1428:
1429: /* Handle expansion. */
1430:
1431: res = NULL;
1432: switch (buf->buf[inam]) {
1433: case '*':
1434: if (iendarg == iarg)
1.79 schwarze 1435: break;
1.260 schwarze 1436: deftype = ROFFDEF_USER | ROFFDEF_PRE;
1437: if ((res = roff_getstrn(r, buf->buf + iarg,
1438: iendarg - iarg, &deftype)) != NULL)
1.79 schwarze 1439: break;
1.8 schwarze 1440:
1.260 schwarze 1441: /*
1442: * If not overriden,
1443: * let \*(.T through to the formatters.
1444: */
1.8 schwarze 1445:
1.260 schwarze 1446: if (iendarg - iarg == 2 &&
1447: buf->buf[iarg] == '.' &&
1448: buf->buf[iarg + 1] == 'T') {
1449: roff_setstrn(&r->strtab, ".T", 2, NULL, 0, 0);
1450: pos = iend;
1.125 schwarze 1451: continue;
1452: }
1.8 schwarze 1453:
1.260 schwarze 1454: mandoc_msg(MANDOCERR_STR_UNDEF, ln, iesc,
1455: "%.*s", iendarg - iarg, buf->buf + iarg);
1456: break;
1.23 schwarze 1457:
1.211 schwarze 1458: case '$':
1459: if (r->mstackpos < 0) {
1.260 schwarze 1460: mandoc_msg(MANDOCERR_ARG_UNDEF, ln, iesc,
1461: "%.*s", iend - iesc, buf->buf + iesc);
1.211 schwarze 1462: break;
1463: }
1464: ctx = r->mstack + r->mstackpos;
1.260 schwarze 1465: argi = buf->buf[iarg] - '1';
1466: if (argi >= 0 && argi <= 8) {
1467: if (argi < ctx->argc)
1468: res = ctx->argv[argi];
1.211 schwarze 1469: break;
1470: }
1.260 schwarze 1471: if (buf->buf[iarg] == '*')
1.211 schwarze 1472: quote_args = 0;
1.260 schwarze 1473: else if (buf->buf[iarg] == '@')
1.211 schwarze 1474: quote_args = 1;
1475: else {
1.260 schwarze 1476: mandoc_msg(MANDOCERR_ARG_NONUM, ln, iesc,
1477: "%.*s", iend - iesc, buf->buf + iesc);
1.211 schwarze 1478: break;
1479: }
1480: asz = 0;
1.260 schwarze 1481: for (argi = 0; argi < ctx->argc; argi++) {
1482: if (argi)
1.211 schwarze 1483: asz++; /* blank */
1484: if (quote_args)
1485: asz += 2; /* quotes */
1.260 schwarze 1486: asz += strlen(ctx->argv[argi]);
1.211 schwarze 1487: }
1.260 schwarze 1488: if (asz != iend - iesc) {
1489: rsz = buf->sz - iend;
1490: if (asz < iend - iesc)
1491: memmove(buf->buf + iesc + asz,
1492: buf->buf + iend, rsz);
1493: buf->sz = iesc + asz + rsz;
1494: buf->buf = mandoc_realloc(buf->buf, buf->sz);
1495: if (asz > iend - iesc)
1496: memmove(buf->buf + iesc + asz,
1497: buf->buf + iend, rsz);
1.211 schwarze 1498: }
1.260 schwarze 1499: dst = buf->buf + iesc;
1500: for (argi = 0; argi < ctx->argc; argi++) {
1501: if (argi)
1502: *dst++ = ' ';
1.211 schwarze 1503: if (quote_args)
1.260 schwarze 1504: *dst++ = '"';
1505: src = ctx->argv[argi];
1506: while (*src != '\0')
1507: *dst++ = *src++;
1.211 schwarze 1508: if (quote_args)
1.260 schwarze 1509: *dst++ = '"';
1.211 schwarze 1510: }
1511: continue;
1.263 schwarze 1512: case 'A':
1513: ubuf[0] = iendarg > iarg ? '1' : '0';
1514: ubuf[1] = '\0';
1515: res = ubuf;
1516: break;
1.80 schwarze 1517: case 'B':
1.79 schwarze 1518: npos = 0;
1.260 schwarze 1519: ubuf[0] = iendarg > iarg && iend > iendarg &&
1520: roff_evalnum(r, ln, buf->buf + iarg, &npos,
1521: NULL, ROFFNUM_SCALE) &&
1522: npos == iendarg - iarg ? '1' : '0';
1.79 schwarze 1523: ubuf[1] = '\0';
1.260 schwarze 1524: res = ubuf;
1.79 schwarze 1525: break;
1.261 schwarze 1526: case 'V':
1527: mandoc_msg(MANDOCERR_UNSUPP, ln, iesc,
1528: "%.*s", iend - iesc, buf->buf + iesc);
1529: roff_expand_patch(buf, iendarg, "}", iend);
1530: roff_expand_patch(buf, iesc, "${", iarg);
1531: continue;
1.262 schwarze 1532: case 'g':
1533: break;
1.80 schwarze 1534: case 'n':
1.260 schwarze 1535: if (iendarg > iarg)
1.91 schwarze 1536: (void)snprintf(ubuf, sizeof(ubuf), "%d",
1.260 schwarze 1537: roff_getregn(r, buf->buf + iarg,
1538: iendarg - iarg, buf->buf[inam + 1]));
1.91 schwarze 1539: else
1540: ubuf[0] = '\0';
1.260 schwarze 1541: res = ubuf;
1.79 schwarze 1542: break;
1.80 schwarze 1543: case 'w':
1.260 schwarze 1544: (void)snprintf(ubuf, sizeof(ubuf),
1545: "%d", (iendarg - iarg) * 24);
1546: res = ubuf;
1547: break;
1548: default:
1.79 schwarze 1549: break;
1550: }
1.260 schwarze 1551: if (res == NULL)
1.37 schwarze 1552: res = "";
1.260 schwarze 1553: if (++expand_count > EXPAND_LIMIT ||
1554: buf->sz + strlen(res) > SHRT_MAX) {
1555: mandoc_msg(MANDOCERR_ROFFLOOP, ln, iesc, NULL);
1.149 schwarze 1556: return ROFF_IGN;
1.8 schwarze 1557: }
1.260 schwarze 1558: roff_expand_patch(buf, iesc, res, iend);
1559: }
1560: return ROFF_CONT;
1561: }
1.8 schwarze 1562:
1.260 schwarze 1563: /*
1564: * Replace the substring from the start position (inclusive)
1565: * to end position (exclusive) with the repl(acement) string.
1566: */
1567: static void
1568: roff_expand_patch(struct buf *buf, int start, const char *repl, int end)
1569: {
1570: char *nbuf;
1.23 schwarze 1571:
1.261 schwarze 1572: buf->sz = mandoc_asprintf(&nbuf, "%.*s%s%s", start, buf->buf,
1573: repl, buf->buf + end) + 1;
1.260 schwarze 1574: free(buf->buf);
1575: buf->buf = nbuf;
1.42 schwarze 1576: }
1.225 schwarze 1577:
1578: /*
1579: * Parse a quoted or unquoted roff-style request or macro argument.
1580: * Return a pointer to the parsed argument, which is either the original
1581: * pointer or advanced by one byte in case the argument is quoted.
1582: * NUL-terminate the argument in place.
1583: * Collapse pairs of quotes inside quoted arguments.
1584: * Advance the argument pointer to the next argument,
1585: * or to the NUL byte terminating the argument line.
1586: */
1587: char *
1.227 schwarze 1588: roff_getarg(struct roff *r, char **cpp, int ln, int *pos)
1.225 schwarze 1589: {
1.227 schwarze 1590: struct buf buf;
1.238 schwarze 1591: char *cp, *start;
1.227 schwarze 1592: int newesc, pairs, quoted, white;
1.225 schwarze 1593:
1594: /* Quoting can only start with a new word. */
1595: start = *cpp;
1596: quoted = 0;
1597: if ('"' == *start) {
1598: quoted = 1;
1599: start++;
1600: }
1601:
1.227 schwarze 1602: newesc = pairs = white = 0;
1.225 schwarze 1603: for (cp = start; '\0' != *cp; cp++) {
1604:
1605: /*
1606: * Move the following text left
1607: * after quoted quotes and after "\\" and "\t".
1608: */
1609: if (pairs)
1610: cp[-pairs] = cp[0];
1611:
1612: if ('\\' == cp[0]) {
1613: /*
1614: * In copy mode, translate double to single
1615: * backslashes and backslash-t to literal tabs.
1616: */
1617: switch (cp[1]) {
1618: case 'a':
1619: case 't':
1.226 schwarze 1620: cp[-pairs] = '\t';
1.227 schwarze 1621: pairs++;
1622: cp++;
1623: break;
1.225 schwarze 1624: case '\\':
1.227 schwarze 1625: newesc = 1;
1626: cp[-pairs] = ASCII_ESC;
1.225 schwarze 1627: pairs++;
1628: cp++;
1629: break;
1630: case ' ':
1631: /* Skip escaped blanks. */
1632: if (0 == quoted)
1633: cp++;
1634: break;
1635: default:
1636: break;
1637: }
1638: } else if (0 == quoted) {
1639: if (' ' == cp[0]) {
1640: /* Unescaped blanks end unquoted args. */
1641: white = 1;
1642: break;
1643: }
1644: } else if ('"' == cp[0]) {
1645: if ('"' == cp[1]) {
1646: /* Quoted quotes collapse. */
1647: pairs++;
1648: cp++;
1649: } else {
1650: /* Unquoted quotes end quoted args. */
1651: quoted = 2;
1652: break;
1653: }
1654: }
1655: }
1656:
1657: /* Quoted argument without a closing quote. */
1658: if (1 == quoted)
1659: mandoc_msg(MANDOCERR_ARG_QUOTE, ln, *pos, NULL);
1660:
1661: /* NUL-terminate this argument and move to the next one. */
1662: if (pairs)
1663: cp[-pairs] = '\0';
1664: if ('\0' != *cp) {
1665: *cp++ = '\0';
1666: while (' ' == *cp)
1667: cp++;
1668: }
1669: *pos += (int)(cp - start) + (quoted ? 1 : 0);
1670: *cpp = cp;
1671:
1672: if ('\0' == *cp && (white || ' ' == cp[-1]))
1673: mandoc_msg(MANDOCERR_SPACE_EOL, ln, *pos, NULL);
1674:
1.227 schwarze 1675: start = mandoc_strdup(start);
1676: if (newesc == 0)
1677: return start;
1678:
1679: buf.buf = start;
1680: buf.sz = strlen(start) + 1;
1681: buf.next = NULL;
1682: if (roff_expand(r, &buf, ln, 0, ASCII_ESC) & ROFF_IGN) {
1683: free(buf.buf);
1684: buf.buf = mandoc_strdup("");
1685: }
1686: return buf.buf;
1.225 schwarze 1687: }
1688:
1.42 schwarze 1689:
1690: /*
1.147 schwarze 1691: * Process text streams.
1.42 schwarze 1692: */
1.212 schwarze 1693: static int
1.177 schwarze 1694: roff_parsetext(struct roff *r, struct buf *buf, int pos, int *offs)
1.42 schwarze 1695: {
1696: size_t sz;
1697: const char *start;
1.51 schwarze 1698: char *p;
1699: int isz;
1.42 schwarze 1700: enum mandoc_esc esc;
1701:
1.147 schwarze 1702: /* Spring the input line trap. */
1703:
1704: if (roffit_lines == 1) {
1705: isz = mandoc_asprintf(&p, "%s\n.%s", buf->buf, roffit_macro);
1706: free(buf->buf);
1707: buf->buf = p;
1708: buf->sz = isz + 1;
1709: *offs = 0;
1710: free(roffit_macro);
1711: roffit_lines = 0;
1.149 schwarze 1712: return ROFF_REPARSE;
1.147 schwarze 1713: } else if (roffit_lines > 1)
1714: --roffit_lines;
1715:
1.177 schwarze 1716: if (roffce_node != NULL && buf->buf[pos] != '\0') {
1717: if (roffce_lines < 1) {
1718: r->man->last = roffce_node;
1719: r->man->next = ROFF_NEXT_SIBLING;
1720: roffce_lines = 0;
1721: roffce_node = NULL;
1722: } else
1723: roffce_lines--;
1724: }
1725:
1.147 schwarze 1726: /* Convert all breakable hyphens into ASCII_HYPH. */
1727:
1.110 schwarze 1728: start = p = buf->buf + pos;
1.42 schwarze 1729:
1.110 schwarze 1730: while (*p != '\0') {
1.42 schwarze 1731: sz = strcspn(p, "-\\");
1732: p += sz;
1733:
1.110 schwarze 1734: if (*p == '\0')
1.42 schwarze 1735: break;
1736:
1.110 schwarze 1737: if (*p == '\\') {
1.42 schwarze 1738: /* Skip over escapes. */
1739: p++;
1.62 schwarze 1740: esc = mandoc_escape((const char **)&p, NULL, NULL);
1.110 schwarze 1741: if (esc == ESCAPE_ERROR)
1.42 schwarze 1742: break;
1.136 schwarze 1743: while (*p == '-')
1744: p++;
1.42 schwarze 1745: continue;
1746: } else if (p == start) {
1747: p++;
1748: continue;
1749: }
1750:
1.44 schwarze 1751: if (isalpha((unsigned char)p[-1]) &&
1752: isalpha((unsigned char)p[1]))
1.42 schwarze 1753: *p = ASCII_HYPH;
1754: p++;
1.8 schwarze 1755: }
1.149 schwarze 1756: return ROFF_CONT;
1.8 schwarze 1757: }
1758:
1.212 schwarze 1759: int
1.249 schwarze 1760: roff_parseln(struct roff *r, int ln, struct buf *buf, int *offs, size_t len)
1.1 schwarze 1761: {
1.166 schwarze 1762: enum roff_tok t;
1.212 schwarze 1763: int e;
1.110 schwarze 1764: int pos; /* parse point */
1.115 schwarze 1765: int spos; /* saved parse point for messages */
1.110 schwarze 1766: int ppos; /* original offset in buf->buf */
1767: int ctl; /* macro line (boolean) */
1768:
1769: ppos = pos = *offs;
1.249 schwarze 1770:
1771: if (len > 80 && r->tbl == NULL && r->eqn == NULL &&
1772: (r->man->flags & ROFF_NOFILL) == 0 &&
1773: strchr(" .\\", buf->buf[pos]) == NULL &&
1774: buf->buf[pos] != r->control &&
1775: strcspn(buf->buf, " ") < 80)
1776: mandoc_msg(MANDOCERR_TEXT_LONG, ln, (int)len - 1,
1777: "%.20s...", buf->buf + pos);
1.1 schwarze 1778:
1.102 schwarze 1779: /* Handle in-line equation delimiters. */
1780:
1.108 schwarze 1781: if (r->tbl == NULL &&
1782: r->last_eqn != NULL && r->last_eqn->delim &&
1.102 schwarze 1783: (r->eqn == NULL || r->eqn_inline)) {
1.110 schwarze 1784: e = roff_eqndelim(r, buf, pos);
1.102 schwarze 1785: if (e == ROFF_REPARSE)
1.149 schwarze 1786: return e;
1.102 schwarze 1787: assert(e == ROFF_CONT);
1788: }
1789:
1.259 schwarze 1790: /* Handle comments and escape sequences. */
1791:
1792: e = roff_parse_comment(r, buf, ln, pos, r->escape);
1793: if ((e & ROFF_MASK) == ROFF_IGN)
1794: return e;
1795: assert(e == ROFF_CONT);
1.8 schwarze 1796:
1.227 schwarze 1797: e = roff_expand(r, buf, ln, pos, r->escape);
1.212 schwarze 1798: if ((e & ROFF_MASK) == ROFF_IGN)
1.149 schwarze 1799: return e;
1.110 schwarze 1800: assert(e == ROFF_CONT);
1.8 schwarze 1801:
1.110 schwarze 1802: ctl = roff_getcontrol(r, buf->buf, &pos);
1.35 schwarze 1803:
1.8 schwarze 1804: /*
1.2 schwarze 1805: * First, if a scope is open and we're not a macro, pass the
1.124 schwarze 1806: * text through the macro's filter.
1807: * Equations process all content themselves.
1808: * Tables process almost all content themselves, but we want
1809: * to warn about macros before passing it there.
1.2 schwarze 1810: */
1811:
1.124 schwarze 1812: if (r->last != NULL && ! ctl) {
1.2 schwarze 1813: t = r->last->tok;
1.110 schwarze 1814: e = (*roffs[t].text)(r, t, buf, ln, pos, pos, offs);
1.212 schwarze 1815: if ((e & ROFF_MASK) == ROFF_IGN)
1.149 schwarze 1816: return e;
1.212 schwarze 1817: e &= ~ROFF_MASK;
1818: } else
1819: e = ROFF_IGN;
1.191 schwarze 1820: if (r->eqn != NULL && strncmp(buf->buf + ppos, ".EN", 3)) {
1821: eqn_read(r->eqn, buf->buf + ppos);
1.212 schwarze 1822: return e;
1.191 schwarze 1823: }
1.193 schwarze 1824: if (r->tbl != NULL && (ctl == 0 || buf->buf[pos] == '\0')) {
1825: tbl_read(r->tbl, ln, buf->buf, ppos);
1.218 schwarze 1826: roff_addtbl(r->man, ln, r->tbl);
1.212 schwarze 1827: return e;
1.193 schwarze 1828: }
1.239 schwarze 1829: if ( ! ctl) {
1830: r->options &= ~MPARSE_COMMENT;
1.212 schwarze 1831: return roff_parsetext(r, buf, pos, offs) | e;
1.239 schwarze 1832: }
1.100 schwarze 1833:
1834: /* Skip empty request lines. */
1835:
1.110 schwarze 1836: if (buf->buf[pos] == '"') {
1.222 schwarze 1837: mandoc_msg(MANDOCERR_COMMENT_BAD, ln, pos, NULL);
1.149 schwarze 1838: return ROFF_IGN;
1.110 schwarze 1839: } else if (buf->buf[pos] == '\0')
1.149 schwarze 1840: return ROFF_IGN;
1.2 schwarze 1841:
1842: /*
1843: * If a scope is open, go to the child handler for that macro,
1844: * as it may want to preprocess before doing anything with it.
1845: */
1846:
1847: if (r->last) {
1.1 schwarze 1848: t = r->last->tok;
1.149 schwarze 1849: return (*roffs[t].sub)(r, t, buf, ln, ppos, pos, offs);
1.2 schwarze 1850: }
1851:
1.239 schwarze 1852: r->options &= ~MPARSE_COMMENT;
1.115 schwarze 1853: spos = pos;
1854: t = roff_parse(r, buf->buf, &pos, ln, ppos);
1.258 schwarze 1855: return roff_req_or_macro(r, t, buf, ln, spos, pos, offs);
1856: }
1857:
1858: /*
1859: * Handle a new request or macro.
1860: * May be called outside any scope or from inside a conditional scope.
1861: */
1862: static int
1863: roff_req_or_macro(ROFF_ARGS) {
1.115 schwarze 1864:
1.258 schwarze 1865: /* For now, tables ignore most macros and some request. */
1.115 schwarze 1866:
1.258 schwarze 1867: if (r->tbl != NULL && (tok == TOKEN_NONE || tok == ROFF_TS ||
1868: tok == ROFF_br || tok == ROFF_ce || tok == ROFF_rj ||
1869: tok == ROFF_sp)) {
1.222 schwarze 1870: mandoc_msg(MANDOCERR_TBLMACRO,
1.258 schwarze 1871: ln, ppos, "%s", buf->buf + ppos);
1872: if (tok != TOKEN_NONE)
1.149 schwarze 1873: return ROFF_IGN;
1.129 schwarze 1874: while (buf->buf[pos] != '\0' && buf->buf[pos] != ' ')
1875: pos++;
1.163 schwarze 1876: while (buf->buf[pos] == ' ')
1.129 schwarze 1877: pos++;
1.193 schwarze 1878: tbl_read(r->tbl, ln, buf->buf, pos);
1.218 schwarze 1879: roff_addtbl(r->man, ln, r->tbl);
1.193 schwarze 1880: return ROFF_IGN;
1.115 schwarze 1881: }
1882:
1.177 schwarze 1883: /* For now, let high level macros abort .ce mode. */
1884:
1.258 schwarze 1885: if (roffce_node != NULL &&
1886: (tok == TOKEN_NONE || tok == ROFF_Dd || tok == ROFF_EQ ||
1887: tok == ROFF_TH || tok == ROFF_TS)) {
1.177 schwarze 1888: r->man->last = roffce_node;
1889: r->man->next = ROFF_NEXT_SIBLING;
1890: roffce_lines = 0;
1891: roffce_node = NULL;
1892: }
1893:
1.2 schwarze 1894: /*
1.115 schwarze 1895: * This is neither a roff request nor a user-defined macro.
1896: * Let the standard macro set parsers handle it.
1.2 schwarze 1897: */
1898:
1.258 schwarze 1899: if (tok == TOKEN_NONE)
1.149 schwarze 1900: return ROFF_CONT;
1.115 schwarze 1901:
1.258 schwarze 1902: /* Execute a roff request or a user-defined macro. */
1.1 schwarze 1903:
1.258 schwarze 1904: return (*roffs[tok].proc)(r, tok, buf, ln, ppos, pos, offs);
1.2 schwarze 1905: }
1906:
1.211 schwarze 1907: /*
1908: * Internal interface function to tell the roff parser that execution
1909: * of the current macro ended. This is required because macro
1910: * definitions usually do not end with a .return request.
1911: */
1912: void
1913: roff_userret(struct roff *r)
1914: {
1915: struct mctx *ctx;
1916: int i;
1917:
1918: assert(r->mstackpos >= 0);
1919: ctx = r->mstack + r->mstackpos;
1920: for (i = 0; i < ctx->argc; i++)
1921: free(ctx->argv[i]);
1922: ctx->argc = 0;
1923: r->mstackpos--;
1924: }
1925:
1.27 schwarze 1926: void
1.2 schwarze 1927: roff_endparse(struct roff *r)
1928: {
1.193 schwarze 1929: if (r->last != NULL)
1.222 schwarze 1930: mandoc_msg(MANDOCERR_BLK_NOEND, r->last->line,
1931: r->last->col, "%s", roff_name[r->last->tok]);
1.27 schwarze 1932:
1.193 schwarze 1933: if (r->eqn != NULL) {
1.222 schwarze 1934: mandoc_msg(MANDOCERR_BLK_NOEND,
1.191 schwarze 1935: r->eqn->node->line, r->eqn->node->pos, "EQ");
1936: eqn_parse(r->eqn);
1937: r->eqn = NULL;
1.32 schwarze 1938: }
1939:
1.193 schwarze 1940: if (r->tbl != NULL) {
1.218 schwarze 1941: tbl_end(r->tbl, 1);
1.193 schwarze 1942: r->tbl = NULL;
1.27 schwarze 1943: }
1.1 schwarze 1944: }
1945:
1946: /*
1.258 schwarze 1947: * Parse the request or macro name at buf[*pos].
1948: * Return ROFF_RENAMED, ROFF_USERDEF, or a ROFF_* token value.
1949: * For empty, undefined, mdoc(7), and man(7) macros, return TOKEN_NONE.
1950: * As a side effect, set r->current_string to the definition or to NULL.
1.1 schwarze 1951: */
1.166 schwarze 1952: static enum roff_tok
1.87 schwarze 1953: roff_parse(struct roff *r, char *buf, int *pos, int ln, int ppos)
1.1 schwarze 1954: {
1.87 schwarze 1955: char *cp;
1.16 schwarze 1956: const char *mac;
1957: size_t maclen;
1.187 schwarze 1958: int deftype;
1.166 schwarze 1959: enum roff_tok t;
1.1 schwarze 1960:
1.87 schwarze 1961: cp = buf + *pos;
1962:
1963: if ('\0' == *cp || '"' == *cp || '\t' == *cp || ' ' == *cp)
1.166 schwarze 1964: return TOKEN_NONE;
1.1 schwarze 1965:
1.87 schwarze 1966: mac = cp;
1967: maclen = roff_getname(r, &cp, ln, ppos);
1.1 schwarze 1968:
1.187 schwarze 1969: deftype = ROFFDEF_USER | ROFFDEF_REN;
1970: r->current_string = roff_getstrn(r, mac, maclen, &deftype);
1971: switch (deftype) {
1972: case ROFFDEF_USER:
1973: t = ROFF_USERDEF;
1974: break;
1975: case ROFFDEF_REN:
1976: t = ROFF_RENAMED;
1977: break;
1978: default:
1979: t = roffhash_find(r->reqtab, mac, maclen);
1980: break;
1981: }
1.166 schwarze 1982: if (t != TOKEN_NONE)
1.87 schwarze 1983: *pos = cp - buf;
1.197 schwarze 1984: else if (deftype == ROFFDEF_UNDEF) {
1985: /* Using an undefined macro defines it to be empty. */
1986: roff_setstrn(&r->strtab, mac, maclen, "", 0, 0);
1987: roff_setstrn(&r->rentab, mac, maclen, NULL, 0, 0);
1988: }
1.149 schwarze 1989: return t;
1.1 schwarze 1990: }
1991:
1.138 schwarze 1992: /* --- handling of request blocks ----------------------------------------- */
1993:
1.247 schwarze 1994: /*
1995: * Close a macro definition block or an "ignore" block.
1996: */
1.212 schwarze 1997: static int
1.2 schwarze 1998: roff_cblock(ROFF_ARGS)
1.1 schwarze 1999: {
1.247 schwarze 2000: int rr;
1.2 schwarze 2001:
1.110 schwarze 2002: if (r->last == NULL) {
1.222 schwarze 2003: mandoc_msg(MANDOCERR_BLK_NOTOPEN, ln, ppos, "..");
1.149 schwarze 2004: return ROFF_IGN;
1.2 schwarze 2005: }
1.1 schwarze 2006:
1.2 schwarze 2007: switch (r->last->tok) {
1.80 schwarze 2008: case ROFF_am:
2009: case ROFF_ami:
2010: case ROFF_de:
2011: case ROFF_dei:
2012: case ROFF_ig:
1.2 schwarze 2013: break;
1.247 schwarze 2014: case ROFF_am1:
2015: case ROFF_de1:
2016: /* Remapped in roff_block(). */
2017: abort();
1.2 schwarze 2018: default:
1.222 schwarze 2019: mandoc_msg(MANDOCERR_BLK_NOTOPEN, ln, ppos, "..");
1.149 schwarze 2020: return ROFF_IGN;
1.2 schwarze 2021: }
2022:
1.247 schwarze 2023: roffnode_pop(r);
2024: roffnode_cleanscope(r);
2025:
2026: /*
2027: * If a conditional block with braces is still open,
2028: * check for "\}" block end markers.
2029: */
2030:
2031: if (r->last != NULL && r->last->endspan < 0) {
2032: rr = 1; /* If arguments follow "\}", warn about them. */
2033: roff_cond_checkend(r, tok, buf, ln, ppos, pos, &rr);
2034: }
2035:
1.110 schwarze 2036: if (buf->buf[pos] != '\0')
1.222 schwarze 2037: mandoc_msg(MANDOCERR_ARG_SKIP, ln, pos,
1.110 schwarze 2038: ".. %s", buf->buf + pos);
1.2 schwarze 2039:
1.149 schwarze 2040: return ROFF_IGN;
1.2 schwarze 2041: }
1.1 schwarze 2042:
1.236 schwarze 2043: /*
2044: * Pop all nodes ending at the end of the current input line.
2045: * Return the number of loops ended.
2046: */
1.212 schwarze 2047: static int
1.2 schwarze 2048: roffnode_cleanscope(struct roff *r)
2049: {
1.212 schwarze 2050: int inloop;
1.1 schwarze 2051:
1.212 schwarze 2052: inloop = 0;
1.247 schwarze 2053: while (r->last != NULL && r->last->endspan > 0) {
1.46 schwarze 2054: if (--r->last->endspan != 0)
1.2 schwarze 2055: break;
1.212 schwarze 2056: inloop += roffnode_pop(r);
1.2 schwarze 2057: }
1.212 schwarze 2058: return inloop;
1.2 schwarze 2059: }
1.1 schwarze 2060:
1.236 schwarze 2061: /*
1.247 schwarze 2062: * Handle the closing "\}" of a conditional block.
1.236 schwarze 2063: * Apart from generating warnings, this only pops nodes.
2064: * Return the number of loops ended.
2065: */
1.212 schwarze 2066: static int
1.68 schwarze 2067: roff_ccond(struct roff *r, int ln, int ppos)
1.2 schwarze 2068: {
2069: if (NULL == r->last) {
1.222 schwarze 2070: mandoc_msg(MANDOCERR_BLK_NOTOPEN, ln, ppos, "\\}");
1.212 schwarze 2071: return 0;
1.2 schwarze 2072: }
1.1 schwarze 2073:
1.2 schwarze 2074: switch (r->last->tok) {
1.80 schwarze 2075: case ROFF_el:
2076: case ROFF_ie:
2077: case ROFF_if:
1.212 schwarze 2078: case ROFF_while:
1.2 schwarze 2079: break;
2080: default:
1.222 schwarze 2081: mandoc_msg(MANDOCERR_BLK_NOTOPEN, ln, ppos, "\\}");
1.212 schwarze 2082: return 0;
1.2 schwarze 2083: }
1.1 schwarze 2084:
1.2 schwarze 2085: if (r->last->endspan > -1) {
1.222 schwarze 2086: mandoc_msg(MANDOCERR_BLK_NOTOPEN, ln, ppos, "\\}");
1.212 schwarze 2087: return 0;
1.2 schwarze 2088: }
2089:
1.212 schwarze 2090: return roffnode_pop(r) + roffnode_cleanscope(r);
1.1 schwarze 2091: }
2092:
1.212 schwarze 2093: static int
1.2 schwarze 2094: roff_block(ROFF_ARGS)
1.1 schwarze 2095: {
1.187 schwarze 2096: const char *name, *value;
2097: char *call, *cp, *iname, *rname;
2098: size_t csz, namesz, rsz;
2099: int deftype;
1.16 schwarze 2100:
1.93 schwarze 2101: /* Ignore groff compatibility mode for now. */
1.2 schwarze 2102:
1.110 schwarze 2103: if (tok == ROFF_de1)
1.93 schwarze 2104: tok = ROFF_de;
1.123 schwarze 2105: else if (tok == ROFF_dei1)
2106: tok = ROFF_dei;
1.110 schwarze 2107: else if (tok == ROFF_am1)
1.93 schwarze 2108: tok = ROFF_am;
1.123 schwarze 2109: else if (tok == ROFF_ami1)
2110: tok = ROFF_ami;
1.93 schwarze 2111:
2112: /* Parse the macro name argument. */
2113:
1.110 schwarze 2114: cp = buf->buf + pos;
2115: if (tok == ROFF_ig) {
1.93 schwarze 2116: iname = NULL;
2117: namesz = 0;
2118: } else {
2119: iname = cp;
2120: namesz = roff_getname(r, &cp, ln, ppos);
2121: iname[namesz] = '\0';
2122: }
1.22 schwarze 2123:
1.93 schwarze 2124: /* Resolve the macro name argument if it is indirect. */
1.22 schwarze 2125:
1.110 schwarze 2126: if (namesz && (tok == ROFF_dei || tok == ROFF_ami)) {
1.187 schwarze 2127: deftype = ROFFDEF_USER;
2128: name = roff_getstrn(r, iname, namesz, &deftype);
2129: if (name == NULL) {
1.222 schwarze 2130: mandoc_msg(MANDOCERR_STR_UNDEF,
2131: ln, (int)(iname - buf->buf),
1.93 schwarze 2132: "%.*s", (int)namesz, iname);
2133: namesz = 0;
2134: } else
2135: namesz = strlen(name);
2136: } else
2137: name = iname;
1.22 schwarze 2138:
1.110 schwarze 2139: if (namesz == 0 && tok != ROFF_ig) {
1.222 schwarze 2140: mandoc_msg(MANDOCERR_REQ_EMPTY,
2141: ln, ppos, "%s", roff_name[tok]);
1.149 schwarze 2142: return ROFF_IGN;
1.93 schwarze 2143: }
1.2 schwarze 2144:
1.16 schwarze 2145: roffnode_push(r, tok, name, ln, ppos);
2146:
2147: /*
2148: * At the beginning of a `de' macro, clear the existing string
2149: * with the same name, if there is one. New content will be
1.66 schwarze 2150: * appended from roff_block_text() in multiline mode.
1.16 schwarze 2151: */
1.22 schwarze 2152:
1.183 schwarze 2153: if (tok == ROFF_de || tok == ROFF_dei) {
1.86 schwarze 2154: roff_setstrn(&r->strtab, name, namesz, "", 0, 0);
1.183 schwarze 2155: roff_setstrn(&r->rentab, name, namesz, NULL, 0, 0);
1.187 schwarze 2156: } else if (tok == ROFF_am || tok == ROFF_ami) {
2157: deftype = ROFFDEF_ANY;
2158: value = roff_getstrn(r, iname, namesz, &deftype);
2159: switch (deftype) { /* Before appending, ... */
2160: case ROFFDEF_PRE: /* copy predefined to user-defined. */
2161: roff_setstrn(&r->strtab, name, namesz,
2162: value, strlen(value), 0);
2163: break;
2164: case ROFFDEF_REN: /* call original standard macro. */
2165: csz = mandoc_asprintf(&call, ".%.*s \\$* \\\"\n",
2166: (int)strlen(value), value);
2167: roff_setstrn(&r->strtab, name, namesz, call, csz, 0);
2168: roff_setstrn(&r->rentab, name, namesz, NULL, 0, 0);
2169: free(call);
2170: break;
2171: case ROFFDEF_STD: /* rename and call standard macro. */
2172: rsz = mandoc_asprintf(&rname, "__%s_renamed", name);
2173: roff_setstrn(&r->rentab, rname, rsz, name, namesz, 0);
2174: csz = mandoc_asprintf(&call, ".%.*s \\$* \\\"\n",
2175: (int)rsz, rname);
2176: roff_setstrn(&r->strtab, name, namesz, call, csz, 0);
2177: free(call);
2178: free(rname);
2179: break;
2180: default:
2181: break;
2182: }
1.183 schwarze 2183: }
1.2 schwarze 2184:
1.110 schwarze 2185: if (*cp == '\0')
1.149 schwarze 2186: return ROFF_IGN;
1.1 schwarze 2187:
1.93 schwarze 2188: /* Get the custom end marker. */
1.22 schwarze 2189:
1.93 schwarze 2190: iname = cp;
1.86 schwarze 2191: namesz = roff_getname(r, &cp, ln, ppos);
1.93 schwarze 2192:
2193: /* Resolve the end marker if it is indirect. */
2194:
1.110 schwarze 2195: if (namesz && (tok == ROFF_dei || tok == ROFF_ami)) {
1.187 schwarze 2196: deftype = ROFFDEF_USER;
2197: name = roff_getstrn(r, iname, namesz, &deftype);
2198: if (name == NULL) {
1.222 schwarze 2199: mandoc_msg(MANDOCERR_STR_UNDEF,
2200: ln, (int)(iname - buf->buf),
1.93 schwarze 2201: "%.*s", (int)namesz, iname);
2202: namesz = 0;
2203: } else
2204: namesz = strlen(name);
2205: } else
2206: name = iname;
2207:
1.86 schwarze 2208: if (namesz)
2209: r->last->end = mandoc_strndup(name, namesz);
1.2 schwarze 2210:
1.110 schwarze 2211: if (*cp != '\0')
1.222 schwarze 2212: mandoc_msg(MANDOCERR_ARG_EXCESS,
1.167 schwarze 2213: ln, pos, ".%s ... %s", roff_name[tok], cp);
1.1 schwarze 2214:
1.149 schwarze 2215: return ROFF_IGN;
1.1 schwarze 2216: }
2217:
1.212 schwarze 2218: static int
1.2 schwarze 2219: roff_block_sub(ROFF_ARGS)
1.1 schwarze 2220: {
1.166 schwarze 2221: enum roff_tok t;
1.2 schwarze 2222: int i, j;
2223:
2224: /*
1.257 schwarze 2225: * If a custom end marker is a user-defined or predefined macro
2226: * or a request, interpret it.
1.2 schwarze 2227: */
2228:
2229: if (r->last->end) {
1.35 schwarze 2230: for (i = pos, j = 0; r->last->end[j]; j++, i++)
1.110 schwarze 2231: if (buf->buf[i] != r->last->end[j])
1.2 schwarze 2232: break;
1.1 schwarze 2233:
1.110 schwarze 2234: if (r->last->end[j] == '\0' &&
2235: (buf->buf[i] == '\0' ||
2236: buf->buf[i] == ' ' ||
2237: buf->buf[i] == '\t')) {
1.2 schwarze 2238: roffnode_pop(r);
2239: roffnode_cleanscope(r);
1.1 schwarze 2240:
1.110 schwarze 2241: while (buf->buf[i] == ' ' || buf->buf[i] == '\t')
1.35 schwarze 2242: i++;
2243:
2244: pos = i;
1.110 schwarze 2245: if (roff_parse(r, buf->buf, &pos, ln, ppos) !=
1.166 schwarze 2246: TOKEN_NONE)
1.149 schwarze 2247: return ROFF_RERUN;
2248: return ROFF_IGN;
1.2 schwarze 2249: }
1.1 schwarze 2250: }
2251:
1.257 schwarze 2252: /* Handle the standard end marker. */
1.1 schwarze 2253:
1.110 schwarze 2254: t = roff_parse(r, buf->buf, &pos, ln, ppos);
1.257 schwarze 2255: if (t == ROFF_cblock)
2256: return roff_cblock(r, t, buf, ln, ppos, pos, offs);
1.1 schwarze 2257:
1.257 schwarze 2258: /* Not an end marker, so append the line to the block. */
1.1 schwarze 2259:
1.257 schwarze 2260: if (tok != ROFF_ig)
2261: roff_setstr(r, r->last->name, buf->buf + ppos, 2);
2262: return ROFF_IGN;
1.2 schwarze 2263: }
2264:
1.212 schwarze 2265: static int
1.2 schwarze 2266: roff_block_text(ROFF_ARGS)
2267: {
2268:
1.110 schwarze 2269: if (tok != ROFF_ig)
2270: roff_setstr(r, r->last->name, buf->buf + pos, 2);
1.16 schwarze 2271:
1.149 schwarze 2272: return ROFF_IGN;
1.2 schwarze 2273: }
2274:
1.247 schwarze 2275: /*
2276: * Check for a closing "\}" and handle it.
2277: * In this function, the final "int *offs" argument is used for
2278: * different purposes than elsewhere:
2279: * Input: *offs == 0: caller wants to discard arguments following \}
2280: * *offs == 1: caller wants to preserve text following \}
2281: * Output: *offs = 0: tell caller to discard input line
2282: * *offs = 1: tell caller to use input line
2283: */
1.212 schwarze 2284: static int
1.247 schwarze 2285: roff_cond_checkend(ROFF_ARGS)
1.2 schwarze 2286: {
1.212 schwarze 2287: char *ep;
2288: int endloop, irc, rr;
1.2 schwarze 2289:
1.212 schwarze 2290: irc = ROFF_IGN;
1.2 schwarze 2291: rr = r->last->rule;
1.212 schwarze 2292: endloop = tok != ROFF_while ? ROFF_IGN :
2293: rr ? ROFF_LOOPCONT : ROFF_LOOPEXIT;
2294: if (roffnode_cleanscope(r))
2295: irc |= endloop;
1.39 schwarze 2296:
1.69 schwarze 2297: /*
1.247 schwarze 2298: * If "\}" occurs on a macro line without a preceding macro or
2299: * a text line contains nothing else, drop the line completely.
1.69 schwarze 2300: */
2301:
1.110 schwarze 2302: ep = buf->buf + pos;
1.247 schwarze 2303: if (ep[0] == '\\' && ep[1] == '}' && (ep[2] == '\0' || *offs == 0))
1.71 schwarze 2304: rr = 0;
1.69 schwarze 2305:
1.215 schwarze 2306: /*
1.247 schwarze 2307: * The closing delimiter "\}" rewinds the conditional scope
1.215 schwarze 2308: * but is otherwise ignored when interpreting the line.
2309: */
1.39 schwarze 2310:
1.110 schwarze 2311: while ((ep = strchr(ep, '\\')) != NULL) {
1.190 schwarze 2312: switch (ep[1]) {
2313: case '}':
1.248 schwarze 2314: if (ep[2] == '\0')
2315: ep[0] = '\0';
2316: else if (rr)
1.247 schwarze 2317: ep[1] = '&';
2318: else
2319: memmove(ep, ep + 2, strlen(ep + 2) + 1);
1.212 schwarze 2320: if (roff_ccond(r, ln, ep - buf->buf))
2321: irc |= endloop;
1.190 schwarze 2322: break;
2323: case '\0':
2324: ++ep;
2325: break;
2326: default:
2327: ep += 2;
2328: break;
1.70 schwarze 2329: }
1.50 schwarze 2330: }
1.247 schwarze 2331: *offs = rr;
2332: return irc;
2333: }
1.190 schwarze 2334:
1.247 schwarze 2335: /*
2336: * Parse and process a request or macro line in conditional scope.
2337: */
2338: static int
2339: roff_cond_sub(ROFF_ARGS)
2340: {
2341: struct roffnode *bl;
1.258 schwarze 2342: int irc, rr, spos;
1.247 schwarze 2343: enum roff_tok t;
2344:
2345: rr = 0; /* If arguments follow "\}", skip them. */
2346: irc = roff_cond_checkend(r, tok, buf, ln, ppos, pos, &rr);
1.258 schwarze 2347: spos = pos;
1.240 schwarze 2348: t = roff_parse(r, buf->buf, &pos, ln, ppos);
2349:
1.190 schwarze 2350: /*
1.258 schwarze 2351: * Handle requests and macros if the conditional evaluated
2352: * to true or if they are structurally required.
2353: * The .break request is always handled specially.
1.190 schwarze 2354: */
2355:
1.236 schwarze 2356: if (t == ROFF_break) {
2357: if (irc & ROFF_LOOPMASK)
2358: irc = ROFF_IGN | ROFF_LOOPEXIT;
2359: else if (rr) {
2360: for (bl = r->last; bl != NULL; bl = bl->parent) {
2361: bl->rule = 0;
2362: if (bl->tok == ROFF_while)
2363: break;
2364: }
2365: }
1.258 schwarze 2366: } else if (rr || (t < TOKEN_NONE && roffs[t].flags & ROFFMAC_STRUCT)) {
2367: irc |= roff_req_or_macro(r, t, buf, ln, spos, pos, offs);
1.255 schwarze 2368: if (irc & ROFF_WHILE)
2369: irc &= ~(ROFF_LOOPCONT | ROFF_LOOPEXIT);
1.258 schwarze 2370: }
1.212 schwarze 2371: return irc;
1.2 schwarze 2372: }
2373:
1.247 schwarze 2374: /*
2375: * Parse and process a text line in conditional scope.
2376: */
1.212 schwarze 2377: static int
1.2 schwarze 2378: roff_cond_text(ROFF_ARGS)
2379: {
1.247 schwarze 2380: int irc, rr;
1.215 schwarze 2381:
1.247 schwarze 2382: rr = 1; /* If arguments follow "\}", preserve them. */
2383: irc = roff_cond_checkend(r, tok, buf, ln, ppos, pos, &rr);
1.212 schwarze 2384: if (rr)
2385: irc |= ROFF_CONT;
2386: return irc;
1.2 schwarze 2387: }
2388:
1.138 schwarze 2389: /* --- handling of numeric and conditional expressions -------------------- */
2390:
1.77 schwarze 2391: /*
2392: * Parse a single signed integer number. Stop at the first non-digit.
2393: * If there is at least one digit, return success and advance the
2394: * parse point, else return failure and let the parse point unchanged.
2395: * Ignore overflows, treat them just like the C language.
2396: */
1.56 schwarze 2397: static int
1.133 schwarze 2398: roff_getnum(const char *v, int *pos, int *res, int flags)
1.56 schwarze 2399: {
1.133 schwarze 2400: int myres, scaled, n, p;
1.79 schwarze 2401:
2402: if (NULL == res)
2403: res = &myres;
1.56 schwarze 2404:
2405: p = *pos;
2406: n = v[p] == '-';
1.133 schwarze 2407: if (n || v[p] == '+')
1.56 schwarze 2408: p++;
2409:
1.133 schwarze 2410: if (flags & ROFFNUM_WHITE)
2411: while (isspace((unsigned char)v[p]))
2412: p++;
2413:
1.56 schwarze 2414: for (*res = 0; isdigit((unsigned char)v[p]); p++)
1.77 schwarze 2415: *res = 10 * *res + v[p] - '0';
1.56 schwarze 2416: if (p == *pos + n)
2417: return 0;
2418:
2419: if (n)
2420: *res = -*res;
2421:
1.126 schwarze 2422: /* Each number may be followed by one optional scaling unit. */
2423:
2424: switch (v[p]) {
2425: case 'f':
1.133 schwarze 2426: scaled = *res * 65536;
1.126 schwarze 2427: break;
2428: case 'i':
1.133 schwarze 2429: scaled = *res * 240;
1.126 schwarze 2430: break;
2431: case 'c':
1.133 schwarze 2432: scaled = *res * 240 / 2.54;
1.126 schwarze 2433: break;
2434: case 'v':
2435: case 'P':
1.133 schwarze 2436: scaled = *res * 40;
1.126 schwarze 2437: break;
2438: case 'm':
2439: case 'n':
1.133 schwarze 2440: scaled = *res * 24;
1.126 schwarze 2441: break;
2442: case 'p':
1.133 schwarze 2443: scaled = *res * 10 / 3;
1.126 schwarze 2444: break;
2445: case 'u':
1.133 schwarze 2446: scaled = *res;
1.126 schwarze 2447: break;
2448: case 'M':
1.133 schwarze 2449: scaled = *res * 6 / 25;
1.126 schwarze 2450: break;
2451: default:
1.133 schwarze 2452: scaled = *res;
1.126 schwarze 2453: p--;
2454: break;
2455: }
1.133 schwarze 2456: if (flags & ROFFNUM_SCALE)
2457: *res = scaled;
1.126 schwarze 2458:
2459: *pos = p + 1;
1.149 schwarze 2460: return 1;
1.56 schwarze 2461: }
2462:
1.71 schwarze 2463: /*
2464: * Evaluate a string comparison condition.
2465: * The first character is the delimiter.
2466: * Succeed if the string up to its second occurrence
2467: * matches the string up to its third occurence.
2468: * Advance the cursor after the third occurrence
2469: * or lacking that, to the end of the line.
2470: */
2471: static int
2472: roff_evalstrcond(const char *v, int *pos)
2473: {
2474: const char *s1, *s2, *s3;
2475: int match;
2476:
2477: match = 0;
2478: s1 = v + *pos; /* initial delimiter */
2479: s2 = s1 + 1; /* for scanning the first string */
2480: s3 = strchr(s2, *s1); /* for scanning the second string */
2481:
2482: if (NULL == s3) /* found no middle delimiter */
2483: goto out;
2484:
2485: while ('\0' != *++s3) {
2486: if (*s2 != *s3) { /* mismatch */
2487: s3 = strchr(s3, *s1);
2488: break;
2489: }
2490: if (*s3 == *s1) { /* found the final delimiter */
2491: match = 1;
2492: break;
2493: }
2494: s2++;
2495: }
2496:
2497: out:
2498: if (NULL == s3)
2499: s3 = strchr(s2, '\0');
1.114 schwarze 2500: else if (*s3 != '\0')
1.71 schwarze 2501: s3++;
2502: *pos = s3 - v;
1.149 schwarze 2503: return match;
1.71 schwarze 2504: }
2505:
1.77 schwarze 2506: /*
2507: * Evaluate an optionally negated single character, numerical,
2508: * or string condition.
2509: */
1.71 schwarze 2510: static int
1.143 schwarze 2511: roff_evalcond(struct roff *r, int ln, char *v, int *pos)
1.5 schwarze 2512: {
1.208 schwarze 2513: const char *start, *end;
2514: char *cp, *name;
2515: size_t sz;
2516: int deftype, len, number, savepos, istrue, wanttrue;
1.5 schwarze 2517:
1.71 schwarze 2518: if ('!' == v[*pos]) {
2519: wanttrue = 0;
2520: (*pos)++;
2521: } else
2522: wanttrue = 1;
2523:
1.5 schwarze 2524: switch (v[*pos]) {
1.112 schwarze 2525: case '\0':
1.149 schwarze 2526: return 0;
1.80 schwarze 2527: case 'n':
2528: case 'o':
1.5 schwarze 2529: (*pos)++;
1.149 schwarze 2530: return wanttrue;
1.80 schwarze 2531: case 'e':
2532: case 't':
1.111 schwarze 2533: case 'v':
1.5 schwarze 2534: (*pos)++;
1.149 schwarze 2535: return !wanttrue;
1.208 schwarze 2536: case 'c':
2537: do {
2538: (*pos)++;
2539: } while (v[*pos] == ' ');
2540:
2541: /*
2542: * Quirk for groff compatibility:
2543: * The horizontal tab is neither available nor unavailable.
2544: */
2545:
2546: if (v[*pos] == '\t') {
2547: (*pos)++;
2548: return 0;
2549: }
2550:
2551: /* Printable ASCII characters are available. */
2552:
2553: if (v[*pos] != '\\') {
2554: (*pos)++;
2555: return wanttrue;
2556: }
2557:
2558: end = v + ++*pos;
2559: switch (mandoc_escape(&end, &start, &len)) {
2560: case ESCAPE_SPECIAL:
2561: istrue = mchars_spec2cp(start, len) != -1;
2562: break;
2563: case ESCAPE_UNICODE:
2564: istrue = 1;
2565: break;
2566: case ESCAPE_NUMBERED:
2567: istrue = mchars_num2char(start, len) != -1;
2568: break;
2569: default:
2570: istrue = !wanttrue;
2571: break;
2572: }
2573: *pos = end - v;
2574: return istrue == wanttrue;
1.182 schwarze 2575: case 'd':
1.143 schwarze 2576: case 'r':
1.182 schwarze 2577: cp = v + *pos + 1;
2578: while (*cp == ' ')
2579: cp++;
2580: name = cp;
2581: sz = roff_getname(r, &cp, ln, cp - v);
1.187 schwarze 2582: if (sz == 0)
2583: istrue = 0;
2584: else if (v[*pos] == 'r')
2585: istrue = roff_hasregn(r, name, sz);
2586: else {
2587: deftype = ROFFDEF_ANY;
2588: roff_getstrn(r, name, sz, &deftype);
2589: istrue = !!deftype;
2590: }
1.235 schwarze 2591: *pos = (name + sz) - v;
1.182 schwarze 2592: return istrue == wanttrue;
1.5 schwarze 2593: default:
2594: break;
2595: }
2596:
1.113 schwarze 2597: savepos = *pos;
1.133 schwarze 2598: if (roff_evalnum(r, ln, v, pos, &number, ROFFNUM_SCALE))
1.149 schwarze 2599: return (number > 0) == wanttrue;
1.113 schwarze 2600: else if (*pos == savepos)
1.149 schwarze 2601: return roff_evalstrcond(v, pos) == wanttrue;
1.77 schwarze 2602: else
1.149 schwarze 2603: return 0;
1.5 schwarze 2604: }
2605:
1.212 schwarze 2606: static int
1.21 schwarze 2607: roff_line_ignore(ROFF_ARGS)
1.6 schwarze 2608: {
1.30 schwarze 2609:
1.149 schwarze 2610: return ROFF_IGN;
1.21 schwarze 2611: }
2612:
1.212 schwarze 2613: static int
1.123 schwarze 2614: roff_insec(ROFF_ARGS)
2615: {
2616:
1.222 schwarze 2617: mandoc_msg(MANDOCERR_REQ_INSEC, ln, ppos, "%s", roff_name[tok]);
1.149 schwarze 2618: return ROFF_IGN;
1.123 schwarze 2619: }
2620:
1.212 schwarze 2621: static int
1.123 schwarze 2622: roff_unsupp(ROFF_ARGS)
2623: {
2624:
1.222 schwarze 2625: mandoc_msg(MANDOCERR_REQ_UNSUPP, ln, ppos, "%s", roff_name[tok]);
1.149 schwarze 2626: return ROFF_IGN;
1.123 schwarze 2627: }
2628:
1.212 schwarze 2629: static int
1.2 schwarze 2630: roff_cond(ROFF_ARGS)
2631: {
1.212 schwarze 2632: int irc;
1.46 schwarze 2633:
2634: roffnode_push(r, tok, NULL, ln, ppos);
1.2 schwarze 2635:
1.80 schwarze 2636: /*
1.35 schwarze 2637: * An `.el' has no conditional body: it will consume the value
2638: * of the current rstack entry set in prior `ie' calls or
1.80 schwarze 2639: * defaults to DENY.
1.35 schwarze 2640: *
2641: * If we're not an `el', however, then evaluate the conditional.
2642: */
1.1 schwarze 2643:
1.110 schwarze 2644: r->last->rule = tok == ROFF_el ?
1.80 schwarze 2645: (r->rstackpos < 0 ? 0 : r->rstack[r->rstackpos--]) :
1.110 schwarze 2646: roff_evalcond(r, ln, buf->buf, &pos);
1.2 schwarze 2647:
1.35 schwarze 2648: /*
2649: * An if-else will put the NEGATION of the current evaluated
2650: * conditional into the stack of rules.
2651: */
2652:
1.110 schwarze 2653: if (tok == ROFF_ie) {
1.96 schwarze 2654: if (r->rstackpos + 1 == r->rstacksz) {
2655: r->rstacksz += 16;
2656: r->rstack = mandoc_reallocarray(r->rstack,
2657: r->rstacksz, sizeof(int));
1.35 schwarze 2658: }
1.71 schwarze 2659: r->rstack[++r->rstackpos] = !r->last->rule;
1.2 schwarze 2660: }
1.5 schwarze 2661:
2662: /* If the parent has false as its rule, then so do we. */
2663:
1.71 schwarze 2664: if (r->last->parent && !r->last->parent->rule)
2665: r->last->rule = 0;
1.5 schwarze 2666:
2667: /*
1.46 schwarze 2668: * Determine scope.
2669: * If there is nothing on the line after the conditional,
2670: * not even whitespace, use next-line scope.
1.212 schwarze 2671: * Except that .while does not support next-line scope.
1.5 schwarze 2672: */
1.2 schwarze 2673:
1.212 schwarze 2674: if (buf->buf[pos] == '\0' && tok != ROFF_while) {
1.46 schwarze 2675: r->last->endspan = 2;
2676: goto out;
2677: }
2678:
1.110 schwarze 2679: while (buf->buf[pos] == ' ')
1.46 schwarze 2680: pos++;
2681:
2682: /* An opening brace requests multiline scope. */
1.2 schwarze 2683:
1.110 schwarze 2684: if (buf->buf[pos] == '\\' && buf->buf[pos + 1] == '{') {
1.2 schwarze 2685: r->last->endspan = -1;
2686: pos += 2;
1.144 schwarze 2687: while (buf->buf[pos] == ' ')
2688: pos++;
1.46 schwarze 2689: goto out;
1.80 schwarze 2690: }
1.2 schwarze 2691:
2692: /*
1.46 schwarze 2693: * Anything else following the conditional causes
2694: * single-line scope. Warn if the scope contains
2695: * nothing but trailing whitespace.
1.2 schwarze 2696: */
2697:
1.110 schwarze 2698: if (buf->buf[pos] == '\0')
1.222 schwarze 2699: mandoc_msg(MANDOCERR_COND_EMPTY,
2700: ln, ppos, "%s", roff_name[tok]);
1.2 schwarze 2701:
1.46 schwarze 2702: r->last->endspan = 1;
1.1 schwarze 2703:
1.46 schwarze 2704: out:
1.2 schwarze 2705: *offs = pos;
1.212 schwarze 2706: irc = ROFF_RERUN;
2707: if (tok == ROFF_while)
2708: irc |= ROFF_WHILE;
2709: return irc;
1.1 schwarze 2710: }
2711:
1.212 schwarze 2712: static int
1.7 schwarze 2713: roff_ds(ROFF_ARGS)
2714: {
1.85 schwarze 2715: char *string;
2716: const char *name;
2717: size_t namesz;
1.10 schwarze 2718:
1.123 schwarze 2719: /* Ignore groff compatibility mode for now. */
2720:
2721: if (tok == ROFF_ds1)
2722: tok = ROFF_ds;
2723: else if (tok == ROFF_as1)
2724: tok = ROFF_as;
2725:
1.10 schwarze 2726: /*
1.85 schwarze 2727: * The first word is the name of the string.
2728: * If it is empty or terminated by an escape sequence,
2729: * abort the `ds' request without defining anything.
1.10 schwarze 2730: */
1.7 schwarze 2731:
1.110 schwarze 2732: name = string = buf->buf + pos;
2733: if (*name == '\0')
1.149 schwarze 2734: return ROFF_IGN;
1.7 schwarze 2735:
1.85 schwarze 2736: namesz = roff_getname(r, &string, ln, pos);
1.235 schwarze 2737: switch (name[namesz]) {
2738: case '\\':
1.149 schwarze 2739: return ROFF_IGN;
1.235 schwarze 2740: case '\t':
2741: string = buf->buf + pos + namesz;
2742: break;
2743: default:
2744: break;
2745: }
1.85 schwarze 2746:
2747: /* Read past the initial double-quote, if any. */
1.110 schwarze 2748: if (*string == '"')
1.7 schwarze 2749: string++;
2750:
1.10 schwarze 2751: /* The rest is the value. */
1.85 schwarze 2752: roff_setstrn(&r->strtab, name, namesz, string, strlen(string),
2753: ROFF_as == tok);
1.183 schwarze 2754: roff_setstrn(&r->rentab, name, namesz, NULL, 0, 0);
1.149 schwarze 2755: return ROFF_IGN;
1.7 schwarze 2756: }
2757:
1.77 schwarze 2758: /*
2759: * Parse a single operator, one or two characters long.
2760: * If the operator is recognized, return success and advance the
2761: * parse point, else return failure and let the parse point unchanged.
2762: */
2763: static int
2764: roff_getop(const char *v, int *pos, char *res)
2765: {
2766:
2767: *res = v[*pos];
2768:
2769: switch (*res) {
1.80 schwarze 2770: case '+':
2771: case '-':
2772: case '*':
2773: case '/':
2774: case '%':
2775: case '&':
2776: case ':':
1.77 schwarze 2777: break;
2778: case '<':
2779: switch (v[*pos + 1]) {
1.80 schwarze 2780: case '=':
1.77 schwarze 2781: *res = 'l';
2782: (*pos)++;
2783: break;
1.80 schwarze 2784: case '>':
1.77 schwarze 2785: *res = '!';
2786: (*pos)++;
2787: break;
1.80 schwarze 2788: case '?':
1.77 schwarze 2789: *res = 'i';
2790: (*pos)++;
2791: break;
2792: default:
2793: break;
2794: }
2795: break;
2796: case '>':
2797: switch (v[*pos + 1]) {
1.80 schwarze 2798: case '=':
1.77 schwarze 2799: *res = 'g';
2800: (*pos)++;
2801: break;
1.80 schwarze 2802: case '?':
1.77 schwarze 2803: *res = 'a';
2804: (*pos)++;
2805: break;
2806: default:
2807: break;
2808: }
2809: break;
2810: case '=':
2811: if ('=' == v[*pos + 1])
2812: (*pos)++;
2813: break;
2814: default:
1.149 schwarze 2815: return 0;
1.77 schwarze 2816: }
2817: (*pos)++;
2818:
1.149 schwarze 2819: return *res;
1.77 schwarze 2820: }
2821:
2822: /*
2823: * Evaluate either a parenthesized numeric expression
2824: * or a single signed integer number.
2825: */
2826: static int
1.106 schwarze 2827: roff_evalpar(struct roff *r, int ln,
1.133 schwarze 2828: const char *v, int *pos, int *res, int flags)
1.77 schwarze 2829: {
2830:
2831: if ('(' != v[*pos])
1.149 schwarze 2832: return roff_getnum(v, pos, res, flags);
1.77 schwarze 2833:
2834: (*pos)++;
1.133 schwarze 2835: if ( ! roff_evalnum(r, ln, v, pos, res, flags | ROFFNUM_WHITE))
1.149 schwarze 2836: return 0;
1.77 schwarze 2837:
1.79 schwarze 2838: /*
2839: * Omission of the closing parenthesis
2840: * is an error in validation mode,
2841: * but ignored in evaluation mode.
2842: */
2843:
1.77 schwarze 2844: if (')' == v[*pos])
2845: (*pos)++;
1.79 schwarze 2846: else if (NULL == res)
1.149 schwarze 2847: return 0;
1.77 schwarze 2848:
1.149 schwarze 2849: return 1;
1.77 schwarze 2850: }
2851:
2852: /*
2853: * Evaluate a complete numeric expression.
2854: * Proceed left to right, there is no concept of precedence.
2855: */
2856: static int
1.106 schwarze 2857: roff_evalnum(struct roff *r, int ln, const char *v,
1.133 schwarze 2858: int *pos, int *res, int flags)
1.77 schwarze 2859: {
2860: int mypos, operand2;
2861: char operator;
2862:
2863: if (NULL == pos) {
2864: mypos = 0;
2865: pos = &mypos;
2866: }
2867:
1.133 schwarze 2868: if (flags & ROFFNUM_WHITE)
1.77 schwarze 2869: while (isspace((unsigned char)v[*pos]))
2870: (*pos)++;
2871:
1.133 schwarze 2872: if ( ! roff_evalpar(r, ln, v, pos, res, flags))
1.149 schwarze 2873: return 0;
1.77 schwarze 2874:
2875: while (1) {
1.133 schwarze 2876: if (flags & ROFFNUM_WHITE)
1.77 schwarze 2877: while (isspace((unsigned char)v[*pos]))
2878: (*pos)++;
2879:
2880: if ( ! roff_getop(v, pos, &operator))
2881: break;
2882:
1.133 schwarze 2883: if (flags & ROFFNUM_WHITE)
1.77 schwarze 2884: while (isspace((unsigned char)v[*pos]))
2885: (*pos)++;
2886:
1.133 schwarze 2887: if ( ! roff_evalpar(r, ln, v, pos, &operand2, flags))
1.149 schwarze 2888: return 0;
1.77 schwarze 2889:
1.133 schwarze 2890: if (flags & ROFFNUM_WHITE)
1.77 schwarze 2891: while (isspace((unsigned char)v[*pos]))
2892: (*pos)++;
1.79 schwarze 2893:
2894: if (NULL == res)
2895: continue;
1.77 schwarze 2896:
2897: switch (operator) {
1.80 schwarze 2898: case '+':
1.77 schwarze 2899: *res += operand2;
2900: break;
1.80 schwarze 2901: case '-':
1.77 schwarze 2902: *res -= operand2;
2903: break;
1.80 schwarze 2904: case '*':
1.77 schwarze 2905: *res *= operand2;
2906: break;
1.80 schwarze 2907: case '/':
1.116 schwarze 2908: if (operand2 == 0) {
1.106 schwarze 2909: mandoc_msg(MANDOCERR_DIVZERO,
1.222 schwarze 2910: ln, *pos, "%s", v);
1.106 schwarze 2911: *res = 0;
2912: break;
2913: }
1.77 schwarze 2914: *res /= operand2;
2915: break;
1.80 schwarze 2916: case '%':
1.116 schwarze 2917: if (operand2 == 0) {
2918: mandoc_msg(MANDOCERR_DIVZERO,
1.222 schwarze 2919: ln, *pos, "%s", v);
1.116 schwarze 2920: *res = 0;
2921: break;
2922: }
1.77 schwarze 2923: *res %= operand2;
2924: break;
1.80 schwarze 2925: case '<':
1.77 schwarze 2926: *res = *res < operand2;
2927: break;
1.80 schwarze 2928: case '>':
1.77 schwarze 2929: *res = *res > operand2;
2930: break;
1.80 schwarze 2931: case 'l':
1.77 schwarze 2932: *res = *res <= operand2;
2933: break;
1.80 schwarze 2934: case 'g':
1.77 schwarze 2935: *res = *res >= operand2;
2936: break;
1.80 schwarze 2937: case '=':
1.77 schwarze 2938: *res = *res == operand2;
2939: break;
1.80 schwarze 2940: case '!':
1.77 schwarze 2941: *res = *res != operand2;
2942: break;
1.80 schwarze 2943: case '&':
1.77 schwarze 2944: *res = *res && operand2;
2945: break;
1.80 schwarze 2946: case ':':
1.77 schwarze 2947: *res = *res || operand2;
2948: break;
1.80 schwarze 2949: case 'i':
1.77 schwarze 2950: if (operand2 < *res)
2951: *res = operand2;
2952: break;
1.80 schwarze 2953: case 'a':
1.77 schwarze 2954: if (operand2 > *res)
2955: *res = operand2;
2956: break;
2957: default:
2958: abort();
2959: }
2960: }
1.149 schwarze 2961: return 1;
1.77 schwarze 2962: }
2963:
1.138 schwarze 2964: /* --- register management ------------------------------------------------ */
2965:
1.52 schwarze 2966: void
1.60 schwarze 2967: roff_setreg(struct roff *r, const char *name, int val, char sign)
1.41 schwarze 2968: {
1.199 schwarze 2969: roff_setregn(r, name, strlen(name), val, sign, INT_MIN);
1.198 schwarze 2970: }
2971:
2972: static void
2973: roff_setregn(struct roff *r, const char *name, size_t len,
1.199 schwarze 2974: int val, char sign, int step)
1.198 schwarze 2975: {
1.52 schwarze 2976: struct roffreg *reg;
2977:
2978: /* Search for an existing register with the same name. */
2979: reg = r->regtab;
2980:
1.198 schwarze 2981: while (reg != NULL && (reg->key.sz != len ||
2982: strncmp(reg->key.p, name, len) != 0))
1.52 schwarze 2983: reg = reg->next;
1.41 schwarze 2984:
1.52 schwarze 2985: if (NULL == reg) {
2986: /* Create a new register. */
2987: reg = mandoc_malloc(sizeof(struct roffreg));
1.198 schwarze 2988: reg->key.p = mandoc_strndup(name, len);
2989: reg->key.sz = len;
1.60 schwarze 2990: reg->val = 0;
1.199 schwarze 2991: reg->step = 0;
1.52 schwarze 2992: reg->next = r->regtab;
2993: r->regtab = reg;
2994: }
2995:
1.60 schwarze 2996: if ('+' == sign)
2997: reg->val += val;
2998: else if ('-' == sign)
2999: reg->val -= val;
3000: else
3001: reg->val = val;
1.199 schwarze 3002: if (step != INT_MIN)
3003: reg->step = step;
1.41 schwarze 3004: }
3005:
1.65 schwarze 3006: /*
3007: * Handle some predefined read-only number registers.
3008: * For now, return -1 if the requested register is not predefined;
3009: * in case a predefined read-only register having the value -1
3010: * were to turn up, another special value would have to be chosen.
3011: */
3012: static int
1.145 schwarze 3013: roff_getregro(const struct roff *r, const char *name)
1.65 schwarze 3014: {
3015:
3016: switch (*name) {
1.145 schwarze 3017: case '$': /* Number of arguments of the last macro evaluated. */
1.211 schwarze 3018: return r->mstackpos < 0 ? 0 : r->mstack[r->mstackpos].argc;
1.80 schwarze 3019: case 'A': /* ASCII approximation mode is always off. */
1.149 schwarze 3020: return 0;
1.80 schwarze 3021: case 'g': /* Groff compatibility mode is always on. */
1.149 schwarze 3022: return 1;
1.80 schwarze 3023: case 'H': /* Fixed horizontal resolution. */
1.149 schwarze 3024: return 24;
1.80 schwarze 3025: case 'j': /* Always adjust left margin only. */
1.149 schwarze 3026: return 0;
1.80 schwarze 3027: case 'T': /* Some output device is always defined. */
1.149 schwarze 3028: return 1;
1.80 schwarze 3029: case 'V': /* Fixed vertical resolution. */
1.149 schwarze 3030: return 40;
1.65 schwarze 3031: default:
1.149 schwarze 3032: return -1;
1.65 schwarze 3033: }
3034: }
3035:
1.53 schwarze 3036: int
1.198 schwarze 3037: roff_getreg(struct roff *r, const char *name)
1.41 schwarze 3038: {
1.199 schwarze 3039: return roff_getregn(r, name, strlen(name), '\0');
1.53 schwarze 3040: }
3041:
3042: static int
1.199 schwarze 3043: roff_getregn(struct roff *r, const char *name, size_t len, char sign)
1.53 schwarze 3044: {
3045: struct roffreg *reg;
1.65 schwarze 3046: int val;
3047:
3048: if ('.' == name[0] && 2 == len) {
1.145 schwarze 3049: val = roff_getregro(r, name + 1);
1.65 schwarze 3050: if (-1 != val)
1.149 schwarze 3051: return val;
1.65 schwarze 3052: }
1.53 schwarze 3053:
1.199 schwarze 3054: for (reg = r->regtab; reg; reg = reg->next) {
1.53 schwarze 3055: if (len == reg->key.sz &&
1.199 schwarze 3056: 0 == strncmp(name, reg->key.p, len)) {
3057: switch (sign) {
3058: case '+':
3059: reg->val += reg->step;
3060: break;
3061: case '-':
3062: reg->val -= reg->step;
3063: break;
3064: default:
3065: break;
3066: }
1.149 schwarze 3067: return reg->val;
1.199 schwarze 3068: }
3069: }
1.143 schwarze 3070:
1.199 schwarze 3071: roff_setregn(r, name, len, 0, '\0', INT_MIN);
1.149 schwarze 3072: return 0;
1.143 schwarze 3073: }
3074:
3075: static int
3076: roff_hasregn(const struct roff *r, const char *name, size_t len)
3077: {
3078: struct roffreg *reg;
3079: int val;
3080:
3081: if ('.' == name[0] && 2 == len) {
1.145 schwarze 3082: val = roff_getregro(r, name + 1);
1.143 schwarze 3083: if (-1 != val)
1.149 schwarze 3084: return 1;
1.143 schwarze 3085: }
3086:
3087: for (reg = r->regtab; reg; reg = reg->next)
3088: if (len == reg->key.sz &&
3089: 0 == strncmp(name, reg->key.p, len))
1.149 schwarze 3090: return 1;
1.41 schwarze 3091:
1.149 schwarze 3092: return 0;
1.41 schwarze 3093: }
3094:
1.52 schwarze 3095: static void
3096: roff_freereg(struct roffreg *reg)
1.41 schwarze 3097: {
1.52 schwarze 3098: struct roffreg *old_reg;
1.41 schwarze 3099:
1.52 schwarze 3100: while (NULL != reg) {
3101: free(reg->key.p);
3102: old_reg = reg;
3103: reg = reg->next;
3104: free(old_reg);
3105: }
1.41 schwarze 3106: }
1.7 schwarze 3107:
1.212 schwarze 3108: static int
1.6 schwarze 3109: roff_nr(ROFF_ARGS)
1.1 schwarze 3110: {
1.199 schwarze 3111: char *key, *val, *step;
1.85 schwarze 3112: size_t keysz;
1.199 schwarze 3113: int iv, is, len;
1.60 schwarze 3114: char sign;
1.6 schwarze 3115:
1.110 schwarze 3116: key = val = buf->buf + pos;
3117: if (*key == '\0')
1.149 schwarze 3118: return ROFF_IGN;
1.85 schwarze 3119:
3120: keysz = roff_getname(r, &val, ln, pos);
1.235 schwarze 3121: if (key[keysz] == '\\' || key[keysz] == '\t')
1.149 schwarze 3122: return ROFF_IGN;
1.6 schwarze 3123:
1.60 schwarze 3124: sign = *val;
1.110 schwarze 3125: if (sign == '+' || sign == '-')
1.60 schwarze 3126: val++;
3127:
1.199 schwarze 3128: len = 0;
3129: if (roff_evalnum(r, ln, val, &len, &iv, ROFFNUM_SCALE) == 0)
3130: return ROFF_IGN;
3131:
3132: step = val + len;
3133: while (isspace((unsigned char)*step))
3134: step++;
3135: if (roff_evalnum(r, ln, step, NULL, &is, 0) == 0)
3136: is = INT_MIN;
1.1 schwarze 3137:
1.199 schwarze 3138: roff_setregn(r, key, keysz, iv, sign, is);
1.149 schwarze 3139: return ROFF_IGN;
1.76 schwarze 3140: }
3141:
1.212 schwarze 3142: static int
1.76 schwarze 3143: roff_rr(ROFF_ARGS)
3144: {
3145: struct roffreg *reg, **prev;
1.85 schwarze 3146: char *name, *cp;
3147: size_t namesz;
1.76 schwarze 3148:
1.110 schwarze 3149: name = cp = buf->buf + pos;
3150: if (*name == '\0')
1.149 schwarze 3151: return ROFF_IGN;
1.85 schwarze 3152: namesz = roff_getname(r, &cp, ln, pos);
3153: name[namesz] = '\0';
1.76 schwarze 3154:
3155: prev = &r->regtab;
3156: while (1) {
3157: reg = *prev;
1.110 schwarze 3158: if (reg == NULL || !strcmp(name, reg->key.p))
1.76 schwarze 3159: break;
3160: prev = ®->next;
3161: }
1.110 schwarze 3162: if (reg != NULL) {
1.76 schwarze 3163: *prev = reg->next;
3164: free(reg->key.p);
3165: free(reg);
3166: }
1.149 schwarze 3167: return ROFF_IGN;
1.29 schwarze 3168: }
3169:
1.138 schwarze 3170: /* --- handler functions for roff requests -------------------------------- */
3171:
1.212 schwarze 3172: static int
1.29 schwarze 3173: roff_rm(ROFF_ARGS)
3174: {
3175: const char *name;
3176: char *cp;
1.85 schwarze 3177: size_t namesz;
1.29 schwarze 3178:
1.110 schwarze 3179: cp = buf->buf + pos;
3180: while (*cp != '\0') {
1.85 schwarze 3181: name = cp;
1.110 schwarze 3182: namesz = roff_getname(r, &cp, ln, (int)(cp - buf->buf));
1.85 schwarze 3183: roff_setstrn(&r->strtab, name, namesz, NULL, 0, 0);
1.183 schwarze 3184: roff_setstrn(&r->rentab, name, namesz, NULL, 0, 0);
1.235 schwarze 3185: if (name[namesz] == '\\' || name[namesz] == '\t')
1.85 schwarze 3186: break;
1.29 schwarze 3187: }
1.149 schwarze 3188: return ROFF_IGN;
1.51 schwarze 3189: }
3190:
1.212 schwarze 3191: static int
1.51 schwarze 3192: roff_it(ROFF_ARGS)
3193: {
3194: int iv;
3195:
3196: /* Parse the number of lines. */
1.133 schwarze 3197:
3198: if ( ! roff_evalnum(r, ln, buf->buf, &pos, &iv, 0)) {
1.222 schwarze 3199: mandoc_msg(MANDOCERR_IT_NONUM,
3200: ln, ppos, "%s", buf->buf + 1);
1.149 schwarze 3201: return ROFF_IGN;
1.51 schwarze 3202: }
3203:
1.134 schwarze 3204: while (isspace((unsigned char)buf->buf[pos]))
3205: pos++;
3206:
3207: /*
3208: * Arm the input line trap.
3209: * Special-casing "an-trap" is an ugly workaround to cope
3210: * with DocBook stupidly fiddling with man(7) internals.
3211: */
1.133 schwarze 3212:
1.51 schwarze 3213: roffit_lines = iv;
1.134 schwarze 3214: roffit_macro = mandoc_strdup(iv != 1 ||
3215: strcmp(buf->buf + pos, "an-trap") ?
3216: buf->buf + pos : "br");
1.149 schwarze 3217: return ROFF_IGN;
1.47 schwarze 3218: }
3219:
1.212 schwarze 3220: static int
1.47 schwarze 3221: roff_Dd(ROFF_ARGS)
3222: {
1.187 schwarze 3223: int mask;
3224: enum roff_tok t, te;
1.99 schwarze 3225:
1.187 schwarze 3226: switch (tok) {
3227: case ROFF_Dd:
3228: tok = MDOC_Dd;
3229: te = MDOC_MAX;
3230: if (r->format == 0)
3231: r->format = MPARSE_MDOC;
3232: mask = MPARSE_MDOC | MPARSE_QUICK;
3233: break;
3234: case ROFF_TH:
3235: tok = MAN_TH;
3236: te = MAN_MAX;
3237: if (r->format == 0)
3238: r->format = MPARSE_MAN;
3239: mask = MPARSE_QUICK;
3240: break;
3241: default:
3242: abort();
3243: }
3244: if ((r->options & mask) == 0)
3245: for (t = tok; t < te; t++)
3246: roff_setstr(r, roff_name[t], NULL, 0);
1.149 schwarze 3247: return ROFF_CONT;
1.14 schwarze 3248: }
3249:
1.212 schwarze 3250: static int
1.27 schwarze 3251: roff_TE(ROFF_ARGS)
3252: {
1.233 schwarze 3253: r->man->flags &= ~ROFF_NONOFILL;
1.193 schwarze 3254: if (r->tbl == NULL) {
1.222 schwarze 3255: mandoc_msg(MANDOCERR_BLK_NOTOPEN, ln, ppos, "TE");
1.193 schwarze 3256: return ROFF_IGN;
3257: }
1.218 schwarze 3258: if (tbl_end(r->tbl, 0) == 0) {
1.193 schwarze 3259: r->tbl = NULL;
1.130 schwarze 3260: free(buf->buf);
3261: buf->buf = mandoc_strdup(".sp");
3262: buf->sz = 4;
1.201 schwarze 3263: *offs = 0;
1.149 schwarze 3264: return ROFF_REPARSE;
1.130 schwarze 3265: }
1.193 schwarze 3266: r->tbl = NULL;
1.149 schwarze 3267: return ROFF_IGN;
1.27 schwarze 3268: }
3269:
1.212 schwarze 3270: static int
1.27 schwarze 3271: roff_T_(ROFF_ARGS)
3272: {
3273:
3274: if (NULL == r->tbl)
1.222 schwarze 3275: mandoc_msg(MANDOCERR_BLK_NOTOPEN, ln, ppos, "T&");
1.27 schwarze 3276: else
1.168 schwarze 3277: tbl_restart(ln, ppos, r->tbl);
1.27 schwarze 3278:
1.149 schwarze 3279: return ROFF_IGN;
1.27 schwarze 3280: }
3281:
1.102 schwarze 3282: /*
3283: * Handle in-line equation delimiters.
3284: */
1.212 schwarze 3285: static int
1.110 schwarze 3286: roff_eqndelim(struct roff *r, struct buf *buf, int pos)
1.41 schwarze 3287: {
1.105 schwarze 3288: char *cp1, *cp2;
3289: const char *bef_pr, *bef_nl, *mac, *aft_nl, *aft_pr;
1.41 schwarze 3290:
1.102 schwarze 3291: /*
3292: * Outside equations, look for an opening delimiter.
3293: * If we are inside an equation, we already know it is
3294: * in-line, or this function wouldn't have been called;
3295: * so look for a closing delimiter.
3296: */
3297:
1.110 schwarze 3298: cp1 = buf->buf + pos;
1.102 schwarze 3299: cp2 = strchr(cp1, r->eqn == NULL ?
3300: r->last_eqn->odelim : r->last_eqn->cdelim);
3301: if (cp2 == NULL)
1.149 schwarze 3302: return ROFF_CONT;
1.102 schwarze 3303:
1.105 schwarze 3304: *cp2++ = '\0';
3305: bef_pr = bef_nl = aft_nl = aft_pr = "";
3306:
3307: /* Handle preceding text, protecting whitespace. */
3308:
1.110 schwarze 3309: if (*buf->buf != '\0') {
1.105 schwarze 3310: if (r->eqn == NULL)
3311: bef_pr = "\\&";
3312: bef_nl = "\n";
3313: }
3314:
3315: /*
3316: * Prepare replacing the delimiter with an equation macro
3317: * and drop leading white space from the equation.
3318: */
1.102 schwarze 3319:
1.105 schwarze 3320: if (r->eqn == NULL) {
3321: while (*cp2 == ' ')
3322: cp2++;
3323: mac = ".EQ";
3324: } else
3325: mac = ".EN";
3326:
3327: /* Handle following text, protecting whitespace. */
3328:
3329: if (*cp2 != '\0') {
3330: aft_nl = "\n";
3331: if (r->eqn != NULL)
3332: aft_pr = "\\&";
3333: }
3334:
3335: /* Do the actual replacement. */
3336:
1.110 schwarze 3337: buf->sz = mandoc_asprintf(&cp1, "%s%s%s%s%s%s%s", buf->buf,
1.105 schwarze 3338: bef_pr, bef_nl, mac, aft_nl, aft_pr, cp2) + 1;
1.110 schwarze 3339: free(buf->buf);
3340: buf->buf = cp1;
1.102 schwarze 3341:
3342: /* Toggle the in-line state of the eqn subsystem. */
3343:
3344: r->eqn_inline = r->eqn == NULL;
1.149 schwarze 3345: return ROFF_REPARSE;
1.41 schwarze 3346: }
3347:
1.212 schwarze 3348: static int
1.107 schwarze 3349: roff_EQ(ROFF_ARGS)
1.32 schwarze 3350: {
1.191 schwarze 3351: struct roff_node *n;
3352:
1.228 schwarze 3353: if (r->man->meta.macroset == MACROSET_MAN)
1.194 schwarze 3354: man_breakscope(r->man, ROFF_EQ);
1.191 schwarze 3355: n = roff_node_alloc(r->man, ln, ppos, ROFFT_EQN, TOKEN_NONE);
3356: if (ln > r->man->last->line)
3357: n->flags |= NODE_LINE;
1.220 schwarze 3358: n->eqn = eqn_box_new();
1.191 schwarze 3359: roff_node_append(r->man, n);
3360: r->man->next = ROFF_NEXT_SIBLING;
1.32 schwarze 3361:
1.110 schwarze 3362: assert(r->eqn == NULL);
1.191 schwarze 3363: if (r->last_eqn == NULL)
1.223 schwarze 3364: r->last_eqn = eqn_alloc();
1.191 schwarze 3365: else
3366: eqn_reset(r->last_eqn);
3367: r->eqn = r->last_eqn;
3368: r->eqn->node = n;
1.41 schwarze 3369:
1.110 schwarze 3370: if (buf->buf[pos] != '\0')
1.222 schwarze 3371: mandoc_msg(MANDOCERR_ARG_SKIP, ln, pos,
1.110 schwarze 3372: ".EQ %s", buf->buf + pos);
1.41 schwarze 3373:
1.149 schwarze 3374: return ROFF_IGN;
1.32 schwarze 3375: }
3376:
1.212 schwarze 3377: static int
1.32 schwarze 3378: roff_EN(ROFF_ARGS)
3379: {
1.191 schwarze 3380: if (r->eqn != NULL) {
3381: eqn_parse(r->eqn);
3382: r->eqn = NULL;
3383: } else
1.222 schwarze 3384: mandoc_msg(MANDOCERR_BLK_NOTOPEN, ln, ppos, "EN");
1.191 schwarze 3385: if (buf->buf[pos] != '\0')
1.222 schwarze 3386: mandoc_msg(MANDOCERR_ARG_SKIP, ln, pos,
1.191 schwarze 3387: "EN %s", buf->buf + pos);
1.149 schwarze 3388: return ROFF_IGN;
1.32 schwarze 3389: }
3390:
1.212 schwarze 3391: static int
1.27 schwarze 3392: roff_TS(ROFF_ARGS)
3393: {
1.193 schwarze 3394: if (r->tbl != NULL) {
1.222 schwarze 3395: mandoc_msg(MANDOCERR_BLK_BROKEN, ln, ppos, "TS breaks TS");
1.218 schwarze 3396: tbl_end(r->tbl, 0);
1.27 schwarze 3397: }
1.233 schwarze 3398: r->man->flags |= ROFF_NONOFILL;
1.223 schwarze 3399: r->tbl = tbl_alloc(ppos, ln, r->last_tbl);
1.218 schwarze 3400: if (r->last_tbl == NULL)
1.193 schwarze 3401: r->first_tbl = r->tbl;
3402: r->last_tbl = r->tbl;
1.169 schwarze 3403: return ROFF_IGN;
3404: }
3405:
1.212 schwarze 3406: static int
1.230 schwarze 3407: roff_noarg(ROFF_ARGS)
3408: {
3409: if (r->man->flags & (MAN_BLINE | MAN_ELINE))
3410: man_breakscope(r->man, tok);
3411: if (tok == ROFF_brp)
3412: tok = ROFF_br;
3413: roff_elem_alloc(r->man, ln, ppos, tok);
3414: if (buf->buf[pos] != '\0')
3415: mandoc_msg(MANDOCERR_ARG_SKIP, ln, pos,
3416: "%s %s", roff_name[tok], buf->buf + pos);
3417: if (tok == ROFF_nf)
3418: r->man->flags |= ROFF_NOFILL;
3419: else if (tok == ROFF_fi)
3420: r->man->flags &= ~ROFF_NOFILL;
3421: r->man->last->flags |= NODE_LINE | NODE_VALID | NODE_ENDED;
3422: r->man->next = ROFF_NEXT_SIBLING;
3423: return ROFF_IGN;
3424: }
3425:
3426: static int
1.169 schwarze 3427: roff_onearg(ROFF_ARGS)
3428: {
3429: struct roff_node *n;
3430: char *cp;
1.177 schwarze 3431: int npos;
1.169 schwarze 3432:
1.174 schwarze 3433: if (r->man->flags & (MAN_BLINE | MAN_ELINE) &&
1.192 schwarze 3434: (tok == ROFF_ce || tok == ROFF_rj || tok == ROFF_sp ||
3435: tok == ROFF_ti))
1.174 schwarze 3436: man_breakscope(r->man, tok);
3437:
1.181 schwarze 3438: if (roffce_node != NULL && (tok == ROFF_ce || tok == ROFF_rj)) {
1.177 schwarze 3439: r->man->last = roffce_node;
3440: r->man->next = ROFF_NEXT_SIBLING;
3441: }
3442:
1.169 schwarze 3443: roff_elem_alloc(r->man, ln, ppos, tok);
3444: n = r->man->last;
3445:
3446: cp = buf->buf + pos;
3447: if (*cp != '\0') {
3448: while (*cp != '\0' && *cp != ' ')
3449: cp++;
3450: while (*cp == ' ')
3451: *cp++ = '\0';
3452: if (*cp != '\0')
1.222 schwarze 3453: mandoc_msg(MANDOCERR_ARG_EXCESS,
3454: ln, (int)(cp - buf->buf),
1.169 schwarze 3455: "%s ... %s", roff_name[tok], cp);
3456: roff_word_alloc(r->man, ln, pos, buf->buf + pos);
1.172 schwarze 3457: }
3458:
1.181 schwarze 3459: if (tok == ROFF_ce || tok == ROFF_rj) {
3460: if (r->man->last->type == ROFFT_ELEM) {
1.177 schwarze 3461: roff_word_alloc(r->man, ln, pos, "1");
3462: r->man->last->flags |= NODE_NOSRC;
3463: }
3464: npos = 0;
3465: if (roff_evalnum(r, ln, r->man->last->string, &npos,
3466: &roffce_lines, 0) == 0) {
1.222 schwarze 3467: mandoc_msg(MANDOCERR_CE_NONUM,
3468: ln, pos, "ce %s", buf->buf + pos);
1.177 schwarze 3469: roffce_lines = 1;
3470: }
3471: if (roffce_lines < 1) {
3472: r->man->last = r->man->last->parent;
3473: roffce_node = NULL;
3474: roffce_lines = 0;
3475: } else
3476: roffce_node = r->man->last->parent;
3477: } else {
3478: n->flags |= NODE_VALID | NODE_ENDED;
3479: r->man->last = n;
3480: }
3481: n->flags |= NODE_LINE;
1.172 schwarze 3482: r->man->next = ROFF_NEXT_SIBLING;
3483: return ROFF_IGN;
3484: }
3485:
1.212 schwarze 3486: static int
1.172 schwarze 3487: roff_manyarg(ROFF_ARGS)
3488: {
3489: struct roff_node *n;
3490: char *sp, *ep;
3491:
3492: roff_elem_alloc(r->man, ln, ppos, tok);
3493: n = r->man->last;
3494:
3495: for (sp = ep = buf->buf + pos; *sp != '\0'; sp = ep) {
3496: while (*ep != '\0' && *ep != ' ')
3497: ep++;
3498: while (*ep == ' ')
3499: *ep++ = '\0';
3500: roff_word_alloc(r->man, ln, sp - buf->buf, sp);
1.169 schwarze 3501: }
3502:
3503: n->flags |= NODE_LINE | NODE_VALID | NODE_ENDED;
3504: r->man->last = n;
3505: r->man->next = ROFF_NEXT_SIBLING;
1.149 schwarze 3506: return ROFF_IGN;
1.123 schwarze 3507: }
3508:
1.212 schwarze 3509: static int
1.183 schwarze 3510: roff_als(ROFF_ARGS)
3511: {
3512: char *oldn, *newn, *end, *value;
3513: size_t oldsz, newsz, valsz;
3514:
3515: newn = oldn = buf->buf + pos;
3516: if (*newn == '\0')
3517: return ROFF_IGN;
3518:
3519: newsz = roff_getname(r, &oldn, ln, pos);
1.235 schwarze 3520: if (newn[newsz] == '\\' || newn[newsz] == '\t' || *oldn == '\0')
1.183 schwarze 3521: return ROFF_IGN;
3522:
3523: end = oldn;
3524: oldsz = roff_getname(r, &end, ln, oldn - buf->buf);
3525: if (oldsz == 0)
3526: return ROFF_IGN;
3527:
1.210 schwarze 3528: valsz = mandoc_asprintf(&value, ".%.*s \\$@\\\"\n",
1.187 schwarze 3529: (int)oldsz, oldn);
1.183 schwarze 3530: roff_setstrn(&r->strtab, newn, newsz, value, valsz, 0);
3531: roff_setstrn(&r->rentab, newn, newsz, NULL, 0, 0);
3532: free(value);
1.236 schwarze 3533: return ROFF_IGN;
3534: }
3535:
3536: /*
3537: * The .break request only makes sense inside conditionals,
3538: * and that case is already handled in roff_cond_sub().
3539: */
3540: static int
3541: roff_break(ROFF_ARGS)
3542: {
3543: mandoc_msg(MANDOCERR_BLK_NOTOPEN, ln, pos, "break");
1.168 schwarze 3544: return ROFF_IGN;
1.27 schwarze 3545: }
3546:
1.212 schwarze 3547: static int
1.48 schwarze 3548: roff_cc(ROFF_ARGS)
3549: {
3550: const char *p;
3551:
1.110 schwarze 3552: p = buf->buf + pos;
1.48 schwarze 3553:
1.110 schwarze 3554: if (*p == '\0' || (r->control = *p++) == '.')
1.175 schwarze 3555: r->control = '\0';
1.48 schwarze 3556:
1.110 schwarze 3557: if (*p != '\0')
1.222 schwarze 3558: mandoc_msg(MANDOCERR_ARG_EXCESS,
1.132 schwarze 3559: ln, p - buf->buf, "cc ... %s", p);
1.48 schwarze 3560:
1.213 schwarze 3561: return ROFF_IGN;
3562: }
3563:
3564: static int
3565: roff_char(ROFF_ARGS)
3566: {
3567: const char *p, *kp, *vp;
3568: size_t ksz, vsz;
3569: int font;
3570:
3571: /* Parse the character to be replaced. */
3572:
3573: kp = buf->buf + pos;
3574: p = kp + 1;
3575: if (*kp == '\0' || (*kp == '\\' &&
3576: mandoc_escape(&p, NULL, NULL) != ESCAPE_SPECIAL) ||
3577: (*p != ' ' && *p != '\0')) {
1.222 schwarze 3578: mandoc_msg(MANDOCERR_CHAR_ARG, ln, pos, "char %s", kp);
1.213 schwarze 3579: return ROFF_IGN;
3580: }
3581: ksz = p - kp;
3582: while (*p == ' ')
3583: p++;
3584:
3585: /*
3586: * If the replacement string contains a font escape sequence,
3587: * we have to restore the font at the end.
3588: */
3589:
3590: vp = p;
3591: vsz = strlen(p);
3592: font = 0;
3593: while (*p != '\0') {
3594: if (*p++ != '\\')
3595: continue;
3596: switch (mandoc_escape(&p, NULL, NULL)) {
3597: case ESCAPE_FONT:
3598: case ESCAPE_FONTROMAN:
3599: case ESCAPE_FONTITALIC:
3600: case ESCAPE_FONTBOLD:
3601: case ESCAPE_FONTBI:
1.250 schwarze 3602: case ESCAPE_FONTCR:
3603: case ESCAPE_FONTCB:
3604: case ESCAPE_FONTCI:
1.213 schwarze 3605: case ESCAPE_FONTPREV:
3606: font++;
3607: break;
3608: default:
3609: break;
3610: }
3611: }
3612: if (font > 1)
1.222 schwarze 3613: mandoc_msg(MANDOCERR_CHAR_FONT,
3614: ln, (int)(vp - buf->buf), "%s", vp);
1.213 schwarze 3615:
3616: /*
3617: * Approximate the effect of .char using the .tr tables.
3618: * XXX In groff, .char and .tr interact differently.
3619: */
3620:
3621: if (ksz == 1) {
3622: if (r->xtab == NULL)
3623: r->xtab = mandoc_calloc(128, sizeof(*r->xtab));
3624: assert((unsigned int)*kp < 128);
3625: free(r->xtab[(int)*kp].p);
3626: r->xtab[(int)*kp].sz = mandoc_asprintf(&r->xtab[(int)*kp].p,
3627: "%s%s", vp, font ? "\fP" : "");
3628: } else {
3629: roff_setstrn(&r->xmbtab, kp, ksz, vp, vsz, 0);
3630: if (font)
3631: roff_setstrn(&r->xmbtab, kp, ksz, "\\fP", 3, 1);
3632: }
1.149 schwarze 3633: return ROFF_IGN;
1.48 schwarze 3634: }
3635:
1.212 schwarze 3636: static int
1.175 schwarze 3637: roff_ec(ROFF_ARGS)
3638: {
3639: const char *p;
3640:
3641: p = buf->buf + pos;
3642: if (*p == '\0')
3643: r->escape = '\\';
3644: else {
3645: r->escape = *p;
3646: if (*++p != '\0')
1.222 schwarze 3647: mandoc_msg(MANDOCERR_ARG_EXCESS, ln,
3648: (int)(p - buf->buf), "ec ... %s", p);
1.175 schwarze 3649: }
3650: return ROFF_IGN;
3651: }
3652:
1.212 schwarze 3653: static int
1.175 schwarze 3654: roff_eo(ROFF_ARGS)
3655: {
3656: r->escape = '\0';
3657: if (buf->buf[pos] != '\0')
1.222 schwarze 3658: mandoc_msg(MANDOCERR_ARG_SKIP,
1.175 schwarze 3659: ln, pos, "eo %s", buf->buf + pos);
1.256 schwarze 3660: return ROFF_IGN;
3661: }
3662:
3663: static int
3664: roff_mc(ROFF_ARGS)
3665: {
3666: struct roff_node *n;
3667: char *cp;
3668:
3669: /* Parse the first argument. */
3670:
3671: cp = buf->buf + pos;
3672: if (*cp != '\0')
3673: cp++;
3674: if (buf->buf[pos] == '\\') {
3675: switch (mandoc_escape((const char **)&cp, NULL, NULL)) {
3676: case ESCAPE_SPECIAL:
3677: case ESCAPE_UNICODE:
3678: case ESCAPE_NUMBERED:
3679: break;
3680: default:
3681: *cp = '\0';
3682: mandoc_msg(MANDOCERR_MC_ESC, ln, pos,
3683: "mc %s", buf->buf + pos);
3684: buf->buf[pos] = '\0';
3685: break;
3686: }
3687: }
3688:
3689: /* Ignore additional arguments. */
3690:
3691: while (*cp == ' ')
3692: *cp++ = '\0';
3693: if (*cp != '\0') {
3694: mandoc_msg(MANDOCERR_MC_DIST, ln, (int)(cp - buf->buf),
3695: "mc ... %s", cp);
3696: *cp = '\0';
3697: }
3698:
3699: /* Create the .mc node. */
3700:
3701: roff_elem_alloc(r->man, ln, ppos, tok);
3702: n = r->man->last;
3703: if (buf->buf[pos] != '\0')
3704: roff_word_alloc(r->man, ln, pos, buf->buf + pos);
3705: n->flags |= NODE_LINE | NODE_VALID | NODE_ENDED;
3706: r->man->last = n;
3707: r->man->next = ROFF_NEXT_SIBLING;
1.175 schwarze 3708: return ROFF_IGN;
1.202 schwarze 3709: }
3710:
1.212 schwarze 3711: static int
1.202 schwarze 3712: roff_nop(ROFF_ARGS)
3713: {
3714: while (buf->buf[pos] == ' ')
3715: pos++;
3716: *offs = pos;
3717: return ROFF_RERUN;
1.175 schwarze 3718: }
3719:
1.212 schwarze 3720: static int
1.42 schwarze 3721: roff_tr(ROFF_ARGS)
3722: {
3723: const char *p, *first, *second;
3724: size_t fsz, ssz;
3725: enum mandoc_esc esc;
3726:
1.110 schwarze 3727: p = buf->buf + pos;
1.42 schwarze 3728:
1.110 schwarze 3729: if (*p == '\0') {
1.222 schwarze 3730: mandoc_msg(MANDOCERR_REQ_EMPTY, ln, ppos, "tr");
1.149 schwarze 3731: return ROFF_IGN;
1.42 schwarze 3732: }
3733:
1.110 schwarze 3734: while (*p != '\0') {
1.42 schwarze 3735: fsz = ssz = 1;
3736:
3737: first = p++;
1.110 schwarze 3738: if (*first == '\\') {
1.42 schwarze 3739: esc = mandoc_escape(&p, NULL, NULL);
1.110 schwarze 3740: if (esc == ESCAPE_ERROR) {
1.222 schwarze 3741: mandoc_msg(MANDOCERR_ESC_BAD, ln,
3742: (int)(p - buf->buf), "%s", first);
1.149 schwarze 3743: return ROFF_IGN;
1.42 schwarze 3744: }
3745: fsz = (size_t)(p - first);
3746: }
3747:
3748: second = p++;
1.110 schwarze 3749: if (*second == '\\') {
1.42 schwarze 3750: esc = mandoc_escape(&p, NULL, NULL);
1.110 schwarze 3751: if (esc == ESCAPE_ERROR) {
1.222 schwarze 3752: mandoc_msg(MANDOCERR_ESC_BAD, ln,
3753: (int)(p - buf->buf), "%s", second);
1.149 schwarze 3754: return ROFF_IGN;
1.42 schwarze 3755: }
3756: ssz = (size_t)(p - second);
1.110 schwarze 3757: } else if (*second == '\0') {
1.222 schwarze 3758: mandoc_msg(MANDOCERR_TR_ODD, ln,
3759: (int)(first - buf->buf), "tr %s", first);
1.42 schwarze 3760: second = " ";
3761: p--;
3762: }
3763:
3764: if (fsz > 1) {
1.80 schwarze 3765: roff_setstrn(&r->xmbtab, first, fsz,
3766: second, ssz, 0);
1.42 schwarze 3767: continue;
3768: }
3769:
1.110 schwarze 3770: if (r->xtab == NULL)
1.80 schwarze 3771: r->xtab = mandoc_calloc(128,
3772: sizeof(struct roffstr));
1.42 schwarze 3773:
3774: free(r->xtab[(int)*first].p);
3775: r->xtab[(int)*first].p = mandoc_strndup(second, ssz);
3776: r->xtab[(int)*first].sz = ssz;
3777: }
3778:
1.149 schwarze 3779: return ROFF_IGN;
1.42 schwarze 3780: }
3781:
1.211 schwarze 3782: /*
3783: * Implementation of the .return request.
3784: * There is no need to call roff_userret() from here.
3785: * The read module will call that after rewinding the reader stack
3786: * to the place from where the current macro was called.
3787: */
1.212 schwarze 3788: static int
1.211 schwarze 3789: roff_return(ROFF_ARGS)
3790: {
3791: if (r->mstackpos >= 0)
1.212 schwarze 3792: return ROFF_IGN | ROFF_USERRET;
1.211 schwarze 3793:
1.222 schwarze 3794: mandoc_msg(MANDOCERR_REQ_NOMAC, ln, ppos, "return");
1.211 schwarze 3795: return ROFF_IGN;
3796: }
3797:
1.212 schwarze 3798: static int
1.178 schwarze 3799: roff_rn(ROFF_ARGS)
3800: {
3801: const char *value;
3802: char *oldn, *newn, *end;
3803: size_t oldsz, newsz;
1.187 schwarze 3804: int deftype;
1.178 schwarze 3805:
3806: oldn = newn = buf->buf + pos;
3807: if (*oldn == '\0')
3808: return ROFF_IGN;
3809:
3810: oldsz = roff_getname(r, &newn, ln, pos);
1.235 schwarze 3811: if (oldn[oldsz] == '\\' || oldn[oldsz] == '\t' || *newn == '\0')
1.178 schwarze 3812: return ROFF_IGN;
3813:
3814: end = newn;
3815: newsz = roff_getname(r, &end, ln, newn - buf->buf);
3816: if (newsz == 0)
3817: return ROFF_IGN;
3818:
1.187 schwarze 3819: deftype = ROFFDEF_ANY;
3820: value = roff_getstrn(r, oldn, oldsz, &deftype);
3821: switch (deftype) {
3822: case ROFFDEF_USER:
1.178 schwarze 3823: roff_setstrn(&r->strtab, newn, newsz, value, strlen(value), 0);
3824: roff_setstrn(&r->strtab, oldn, oldsz, NULL, 0, 0);
3825: roff_setstrn(&r->rentab, newn, newsz, NULL, 0, 0);
1.187 schwarze 3826: break;
3827: case ROFFDEF_PRE:
3828: roff_setstrn(&r->strtab, newn, newsz, value, strlen(value), 0);
3829: roff_setstrn(&r->rentab, newn, newsz, NULL, 0, 0);
3830: break;
3831: case ROFFDEF_REN:
1.178 schwarze 3832: roff_setstrn(&r->rentab, newn, newsz, value, strlen(value), 0);
3833: roff_setstrn(&r->rentab, oldn, oldsz, NULL, 0, 0);
1.187 schwarze 3834: roff_setstrn(&r->strtab, newn, newsz, NULL, 0, 0);
3835: break;
3836: case ROFFDEF_STD:
1.178 schwarze 3837: roff_setstrn(&r->rentab, newn, newsz, oldn, oldsz, 0);
1.187 schwarze 3838: roff_setstrn(&r->strtab, newn, newsz, NULL, 0, 0);
3839: break;
3840: default:
3841: roff_setstrn(&r->strtab, newn, newsz, NULL, 0, 0);
3842: roff_setstrn(&r->rentab, newn, newsz, NULL, 0, 0);
3843: break;
3844: }
1.178 schwarze 3845: return ROFF_IGN;
3846: }
3847:
1.212 schwarze 3848: static int
1.211 schwarze 3849: roff_shift(ROFF_ARGS)
3850: {
3851: struct mctx *ctx;
1.254 schwarze 3852: int argpos, levels, i;
1.211 schwarze 3853:
1.254 schwarze 3854: argpos = pos;
1.211 schwarze 3855: levels = 1;
3856: if (buf->buf[pos] != '\0' &&
3857: roff_evalnum(r, ln, buf->buf, &pos, &levels, 0) == 0) {
1.222 schwarze 3858: mandoc_msg(MANDOCERR_CE_NONUM,
1.211 schwarze 3859: ln, pos, "shift %s", buf->buf + pos);
3860: levels = 1;
3861: }
3862: if (r->mstackpos < 0) {
1.222 schwarze 3863: mandoc_msg(MANDOCERR_REQ_NOMAC, ln, ppos, "shift");
1.211 schwarze 3864: return ROFF_IGN;
3865: }
3866: ctx = r->mstack + r->mstackpos;
3867: if (levels > ctx->argc) {
1.222 schwarze 3868: mandoc_msg(MANDOCERR_SHIFT,
1.254 schwarze 3869: ln, argpos, "%d, but max is %d", levels, ctx->argc);
1.211 schwarze 3870: levels = ctx->argc;
1.254 schwarze 3871: }
3872: if (levels < 0) {
3873: mandoc_msg(MANDOCERR_ARG_NEG, ln, argpos, "shift %d", levels);
3874: levels = 0;
1.211 schwarze 3875: }
3876: if (levels == 0)
3877: return ROFF_IGN;
3878: for (i = 0; i < levels; i++)
3879: free(ctx->argv[i]);
3880: ctx->argc -= levels;
3881: for (i = 0; i < ctx->argc; i++)
3882: ctx->argv[i] = ctx->argv[i + levels];
3883: return ROFF_IGN;
3884: }
3885:
1.212 schwarze 3886: static int
1.14 schwarze 3887: roff_so(ROFF_ARGS)
3888: {
1.121 schwarze 3889: char *name, *cp;
1.15 schwarze 3890:
1.110 schwarze 3891: name = buf->buf + pos;
1.222 schwarze 3892: mandoc_msg(MANDOCERR_SO, ln, ppos, "so %s", name);
1.14 schwarze 3893:
1.22 schwarze 3894: /*
3895: * Handle `so'. Be EXTREMELY careful, as we shouldn't be
3896: * opening anything that's not in our cwd or anything beneath
3897: * it. Thus, explicitly disallow traversing up the file-system
3898: * or using absolute paths.
3899: */
3900:
1.110 schwarze 3901: if (*name == '/' || strstr(name, "../") || strstr(name, "/..")) {
1.222 schwarze 3902: mandoc_msg(MANDOCERR_SO_PATH, ln, ppos, ".so %s", name);
1.121 schwarze 3903: buf->sz = mandoc_asprintf(&cp,
3904: ".sp\nSee the file %s.\n.sp", name) + 1;
3905: free(buf->buf);
3906: buf->buf = cp;
3907: *offs = 0;
1.149 schwarze 3908: return ROFF_REPARSE;
1.14 schwarze 3909: }
3910:
3911: *offs = pos;
1.149 schwarze 3912: return ROFF_SO;
1.7 schwarze 3913: }
3914:
1.138 schwarze 3915: /* --- user defined strings and macros ------------------------------------ */
3916:
1.212 schwarze 3917: static int
1.16 schwarze 3918: roff_userdef(ROFF_ARGS)
1.12 schwarze 3919: {
1.211 schwarze 3920: struct mctx *ctx;
3921: char *arg, *ap, *dst, *src;
3922: size_t sz;
3923:
1.237 schwarze 3924: /* If the macro is empty, ignore it altogether. */
3925:
3926: if (*r->current_string == '\0')
3927: return ROFF_IGN;
3928:
1.211 schwarze 3929: /* Initialize a new macro stack context. */
3930:
3931: if (++r->mstackpos == r->mstacksz) {
3932: r->mstack = mandoc_recallocarray(r->mstack,
3933: r->mstacksz, r->mstacksz + 8, sizeof(*r->mstack));
3934: r->mstacksz += 8;
3935: }
3936: ctx = r->mstack + r->mstackpos;
3937: ctx->argc = 0;
3938:
3939: /*
3940: * Collect pointers to macro argument strings,
3941: * NUL-terminating them and escaping quotes.
3942: */
3943:
3944: src = buf->buf + pos;
3945: while (*src != '\0') {
3946: if (ctx->argc == ctx->argsz) {
3947: ctx->argsz += 8;
3948: ctx->argv = mandoc_reallocarray(ctx->argv,
3949: ctx->argsz, sizeof(*ctx->argv));
3950: }
1.227 schwarze 3951: arg = roff_getarg(r, &src, ln, &pos);
1.211 schwarze 3952: sz = 1; /* For the terminating NUL. */
3953: for (ap = arg; *ap != '\0'; ap++)
3954: sz += *ap == '"' ? 4 : 1;
3955: ctx->argv[ctx->argc++] = dst = mandoc_malloc(sz);
3956: for (ap = arg; *ap != '\0'; ap++) {
3957: if (*ap == '"') {
3958: memcpy(dst, "\\(dq", 4);
3959: dst += 4;
3960: } else
3961: *dst++ = *ap;
1.16 schwarze 3962: }
1.211 schwarze 3963: *dst = '\0';
1.227 schwarze 3964: free(arg);
1.209 schwarze 3965: }
3966:
1.211 schwarze 3967: /* Replace the macro invocation by the macro definition. */
1.135 schwarze 3968:
1.110 schwarze 3969: free(buf->buf);
1.211 schwarze 3970: buf->buf = mandoc_strdup(r->current_string);
3971: buf->sz = strlen(buf->buf) + 1;
1.120 schwarze 3972: *offs = 0;
1.16 schwarze 3973:
1.237 schwarze 3974: return buf->buf[buf->sz - 2] == '\n' ?
1.212 schwarze 3975: ROFF_REPARSE | ROFF_USERCALL : ROFF_IGN | ROFF_APPEND;
1.12 schwarze 3976: }
1.28 schwarze 3977:
1.178 schwarze 3978: /*
3979: * Calling a high-level macro that was renamed with .rn.
3980: * r->current_string has already been set up by roff_parse().
3981: */
1.212 schwarze 3982: static int
1.178 schwarze 3983: roff_renamed(ROFF_ARGS)
3984: {
3985: char *nbuf;
3986:
1.187 schwarze 3987: buf->sz = mandoc_asprintf(&nbuf, ".%s%s%s", r->current_string,
3988: buf->buf[pos] == '\0' ? "" : " ", buf->buf + pos) + 1;
1.178 schwarze 3989: free(buf->buf);
3990: buf->buf = nbuf;
1.201 schwarze 3991: *offs = 0;
1.178 schwarze 3992: return ROFF_CONT;
3993: }
3994:
1.234 schwarze 3995: /*
3996: * Measure the length in bytes of the roff identifier at *cpp
3997: * and advance the pointer to the next word.
3998: */
1.85 schwarze 3999: static size_t
1.28 schwarze 4000: roff_getname(struct roff *r, char **cpp, int ln, int pos)
4001: {
4002: char *name, *cp;
1.85 schwarze 4003: size_t namesz;
1.28 schwarze 4004:
4005: name = *cpp;
1.234 schwarze 4006: if (*name == '\0')
1.149 schwarze 4007: return 0;
1.28 schwarze 4008:
1.234 schwarze 4009: /* Advance cp to the byte after the end of the name. */
4010:
1.85 schwarze 4011: for (cp = name; 1; cp++) {
1.234 schwarze 4012: namesz = cp - name;
1.235 schwarze 4013: if (*cp == '\0')
1.85 schwarze 4014: break;
1.235 schwarze 4015: if (*cp == ' ' || *cp == '\t') {
4016: cp++;
4017: break;
4018: }
1.234 schwarze 4019: if (*cp != '\\')
1.28 schwarze 4020: continue;
1.234 schwarze 4021: if (cp[1] == '{' || cp[1] == '}')
1.88 schwarze 4022: break;
1.234 schwarze 4023: if (*++cp == '\\')
1.28 schwarze 4024: continue;
1.222 schwarze 4025: mandoc_msg(MANDOCERR_NAMESC, ln, pos,
1.97 schwarze 4026: "%.*s", (int)(cp - name + 1), name);
1.85 schwarze 4027: mandoc_escape((const char **)&cp, NULL, NULL);
4028: break;
1.28 schwarze 4029: }
4030:
4031: /* Read past spaces. */
1.234 schwarze 4032:
4033: while (*cp == ' ')
1.28 schwarze 4034: cp++;
4035:
4036: *cpp = cp;
1.149 schwarze 4037: return namesz;
1.28 schwarze 4038: }
4039:
1.16 schwarze 4040: /*
4041: * Store *string into the user-defined string called *name.
4042: * To clear an existing entry, call with (*r, *name, NULL, 0).
1.66 schwarze 4043: * append == 0: replace mode
4044: * append == 1: single-line append mode
4045: * append == 2: multiline append mode, append '\n' after each call
1.16 schwarze 4046: */
1.8 schwarze 4047: static void
1.16 schwarze 4048: roff_setstr(struct roff *r, const char *name, const char *string,
1.66 schwarze 4049: int append)
1.7 schwarze 4050: {
1.187 schwarze 4051: size_t namesz;
1.42 schwarze 4052:
1.187 schwarze 4053: namesz = strlen(name);
4054: roff_setstrn(&r->strtab, name, namesz, string,
1.80 schwarze 4055: string ? strlen(string) : 0, append);
1.187 schwarze 4056: roff_setstrn(&r->rentab, name, namesz, NULL, 0, 0);
1.42 schwarze 4057: }
4058:
4059: static void
4060: roff_setstrn(struct roffkv **r, const char *name, size_t namesz,
1.66 schwarze 4061: const char *string, size_t stringsz, int append)
1.42 schwarze 4062: {
4063: struct roffkv *n;
4064: char *c;
4065: int i;
4066: size_t oldch, newch;
1.7 schwarze 4067:
1.16 schwarze 4068: /* Search for an existing string with the same name. */
1.42 schwarze 4069: n = *r;
4070:
1.84 schwarze 4071: while (n && (namesz != n->key.sz ||
4072: strncmp(n->key.p, name, namesz)))
1.7 schwarze 4073: n = n->next;
1.8 schwarze 4074:
4075: if (NULL == n) {
1.16 schwarze 4076: /* Create a new string table entry. */
1.42 schwarze 4077: n = mandoc_malloc(sizeof(struct roffkv));
4078: n->key.p = mandoc_strndup(name, namesz);
4079: n->key.sz = namesz;
4080: n->val.p = NULL;
4081: n->val.sz = 0;
4082: n->next = *r;
4083: *r = n;
1.66 schwarze 4084: } else if (0 == append) {
1.42 schwarze 4085: free(n->val.p);
4086: n->val.p = NULL;
4087: n->val.sz = 0;
1.16 schwarze 4088: }
4089:
4090: if (NULL == string)
4091: return;
4092:
4093: /*
4094: * One additional byte for the '\n' in multiline mode,
4095: * and one for the terminating '\0'.
4096: */
1.66 schwarze 4097: newch = stringsz + (1 < append ? 2u : 1u);
1.42 schwarze 4098:
4099: if (NULL == n->val.p) {
4100: n->val.p = mandoc_malloc(newch);
4101: *n->val.p = '\0';
1.16 schwarze 4102: oldch = 0;
4103: } else {
1.42 schwarze 4104: oldch = n->val.sz;
4105: n->val.p = mandoc_realloc(n->val.p, oldch + newch);
1.16 schwarze 4106: }
4107:
4108: /* Skip existing content in the destination buffer. */
1.42 schwarze 4109: c = n->val.p + (int)oldch;
1.16 schwarze 4110:
4111: /* Append new content to the destination buffer. */
1.42 schwarze 4112: i = 0;
4113: while (i < (int)stringsz) {
1.16 schwarze 4114: /*
4115: * Rudimentary roff copy mode:
4116: * Handle escaped backslashes.
4117: */
1.42 schwarze 4118: if ('\\' == string[i] && '\\' == string[i + 1])
4119: i++;
4120: *c++ = string[i++];
1.16 schwarze 4121: }
1.8 schwarze 4122:
1.16 schwarze 4123: /* Append terminating bytes. */
1.66 schwarze 4124: if (1 < append)
1.16 schwarze 4125: *c++ = '\n';
1.42 schwarze 4126:
1.16 schwarze 4127: *c = '\0';
1.42 schwarze 4128: n->val.sz = (int)(c - n->val.p);
1.7 schwarze 4129: }
4130:
1.8 schwarze 4131: static const char *
1.197 schwarze 4132: roff_getstrn(struct roff *r, const char *name, size_t len,
1.187 schwarze 4133: int *deftype)
1.7 schwarze 4134: {
1.187 schwarze 4135: const struct roffkv *n;
1.197 schwarze 4136: int found, i;
1.187 schwarze 4137: enum roff_tok tok;
4138:
1.197 schwarze 4139: found = 0;
4140: for (n = r->strtab; n != NULL; n = n->next) {
4141: if (strncmp(name, n->key.p, len) != 0 ||
4142: n->key.p[len] != '\0' || n->val.p == NULL)
4143: continue;
4144: if (*deftype & ROFFDEF_USER) {
4145: *deftype = ROFFDEF_USER;
4146: return n->val.p;
4147: } else {
4148: found = 1;
4149: break;
4150: }
4151: }
4152: for (n = r->rentab; n != NULL; n = n->next) {
4153: if (strncmp(name, n->key.p, len) != 0 ||
4154: n->key.p[len] != '\0' || n->val.p == NULL)
4155: continue;
4156: if (*deftype & ROFFDEF_REN) {
4157: *deftype = ROFFDEF_REN;
4158: return n->val.p;
4159: } else {
4160: found = 1;
4161: break;
1.187 schwarze 4162: }
4163: }
1.197 schwarze 4164: for (i = 0; i < PREDEFS_MAX; i++) {
4165: if (strncmp(name, predefs[i].name, len) != 0 ||
4166: predefs[i].name[len] != '\0')
4167: continue;
4168: if (*deftype & ROFFDEF_PRE) {
4169: *deftype = ROFFDEF_PRE;
4170: return predefs[i].str;
4171: } else {
4172: found = 1;
4173: break;
1.187 schwarze 4174: }
4175: }
1.228 schwarze 4176: if (r->man->meta.macroset != MACROSET_MAN) {
1.197 schwarze 4177: for (tok = MDOC_Dd; tok < MDOC_MAX; tok++) {
4178: if (strncmp(name, roff_name[tok], len) != 0 ||
4179: roff_name[tok][len] != '\0')
4180: continue;
4181: if (*deftype & ROFFDEF_STD) {
4182: *deftype = ROFFDEF_STD;
4183: return NULL;
4184: } else {
4185: found = 1;
4186: break;
1.187 schwarze 4187: }
4188: }
4189: }
1.228 schwarze 4190: if (r->man->meta.macroset != MACROSET_MDOC) {
1.197 schwarze 4191: for (tok = MAN_TH; tok < MAN_MAX; tok++) {
4192: if (strncmp(name, roff_name[tok], len) != 0 ||
4193: roff_name[tok][len] != '\0')
4194: continue;
4195: if (*deftype & ROFFDEF_STD) {
4196: *deftype = ROFFDEF_STD;
4197: return NULL;
4198: } else {
4199: found = 1;
4200: break;
1.187 schwarze 4201: }
4202: }
1.197 schwarze 4203: }
4204:
4205: if (found == 0 && *deftype != ROFFDEF_ANY) {
4206: if (*deftype & ROFFDEF_REN) {
4207: /*
4208: * This might still be a request,
4209: * so do not treat it as undefined yet.
4210: */
4211: *deftype = ROFFDEF_UNDEF;
4212: return NULL;
1.187 schwarze 4213: }
1.197 schwarze 4214:
4215: /* Using an undefined string defines it to be empty. */
4216:
4217: roff_setstrn(&r->strtab, name, len, "", 0, 0);
4218: roff_setstrn(&r->rentab, name, len, NULL, 0, 0);
1.187 schwarze 4219: }
1.197 schwarze 4220:
1.187 schwarze 4221: *deftype = 0;
1.149 schwarze 4222: return NULL;
1.7 schwarze 4223: }
4224:
1.8 schwarze 4225: static void
1.42 schwarze 4226: roff_freestr(struct roffkv *r)
1.7 schwarze 4227: {
1.42 schwarze 4228: struct roffkv *n, *nn;
1.7 schwarze 4229:
1.42 schwarze 4230: for (n = r; n; n = nn) {
4231: free(n->key.p);
4232: free(n->val.p);
1.7 schwarze 4233: nn = n->next;
4234: free(n);
4235: }
1.27 schwarze 4236: }
1.138 schwarze 4237:
4238: /* --- accessors and utility functions ------------------------------------ */
1.42 schwarze 4239:
4240: /*
4241: * Duplicate an input string, making the appropriate character
4242: * conversations (as stipulated by `tr') along the way.
4243: * Returns a heap-allocated string with all the replacements made.
4244: */
4245: char *
4246: roff_strdup(const struct roff *r, const char *p)
4247: {
4248: const struct roffkv *cp;
4249: char *res;
4250: const char *pp;
4251: size_t ssz, sz;
4252: enum mandoc_esc esc;
4253:
4254: if (NULL == r->xmbtab && NULL == r->xtab)
1.149 schwarze 4255: return mandoc_strdup(p);
1.42 schwarze 4256: else if ('\0' == *p)
1.149 schwarze 4257: return mandoc_strdup("");
1.42 schwarze 4258:
4259: /*
4260: * Step through each character looking for term matches
4261: * (remember that a `tr' can be invoked with an escape, which is
4262: * a glyph but the escape is multi-character).
4263: * We only do this if the character hash has been initialised
4264: * and the string is >0 length.
4265: */
4266:
4267: res = NULL;
4268: ssz = 0;
4269:
4270: while ('\0' != *p) {
1.161 schwarze 4271: assert((unsigned int)*p < 128);
4272: if ('\\' != *p && r->xtab && r->xtab[(unsigned int)*p].p) {
1.42 schwarze 4273: sz = r->xtab[(int)*p].sz;
4274: res = mandoc_realloc(res, ssz + sz + 1);
4275: memcpy(res + ssz, r->xtab[(int)*p].p, sz);
4276: ssz += sz;
4277: p++;
4278: continue;
4279: } else if ('\\' != *p) {
4280: res = mandoc_realloc(res, ssz + 2);
4281: res[ssz++] = *p++;
4282: continue;
4283: }
4284:
4285: /* Search for term matches. */
4286: for (cp = r->xmbtab; cp; cp = cp->next)
4287: if (0 == strncmp(p, cp->key.p, cp->key.sz))
4288: break;
4289:
4290: if (NULL != cp) {
4291: /*
4292: * A match has been found.
4293: * Append the match to the array and move
4294: * forward by its keysize.
4295: */
1.80 schwarze 4296: res = mandoc_realloc(res,
4297: ssz + cp->val.sz + 1);
1.42 schwarze 4298: memcpy(res + ssz, cp->val.p, cp->val.sz);
4299: ssz += cp->val.sz;
4300: p += (int)cp->key.sz;
4301: continue;
4302: }
4303:
4304: /*
4305: * Handle escapes carefully: we need to copy
4306: * over just the escape itself, or else we might
4307: * do replacements within the escape itself.
4308: * Make sure to pass along the bogus string.
4309: */
4310: pp = p++;
4311: esc = mandoc_escape(&p, NULL, NULL);
4312: if (ESCAPE_ERROR == esc) {
4313: sz = strlen(pp);
4314: res = mandoc_realloc(res, ssz + sz + 1);
4315: memcpy(res + ssz, pp, sz);
4316: break;
4317: }
1.80 schwarze 4318: /*
4319: * We bail out on bad escapes.
1.42 schwarze 4320: * No need to warn: we already did so when
1.227 schwarze 4321: * roff_expand() was called.
1.42 schwarze 4322: */
4323: sz = (int)(p - pp);
4324: res = mandoc_realloc(res, ssz + sz + 1);
4325: memcpy(res + ssz, pp, sz);
4326: ssz += sz;
4327: }
4328:
4329: res[(int)ssz] = '\0';
1.149 schwarze 4330: return res;
1.99 schwarze 4331: }
4332:
4333: int
4334: roff_getformat(const struct roff *r)
4335: {
4336:
1.149 schwarze 4337: return r->format;
1.48 schwarze 4338: }
4339:
4340: /*
1.80 schwarze 4341: * Find out whether a line is a macro line or not.
1.48 schwarze 4342: * If it is, adjust the current position and return one; if it isn't,
4343: * return zero and don't change the current position.
4344: * If the control character has been set with `.cc', then let that grain
4345: * precedence.
4346: * This is slighly contrary to groff, where using the non-breaking
4347: * control character when `cc' has been invoked will cause the
4348: * non-breaking macro contents to be printed verbatim.
4349: */
4350: int
4351: roff_getcontrol(const struct roff *r, const char *cp, int *ppos)
4352: {
4353: int pos;
4354:
4355: pos = *ppos;
4356:
1.175 schwarze 4357: if (r->control != '\0' && cp[pos] == r->control)
1.48 schwarze 4358: pos++;
1.175 schwarze 4359: else if (r->control != '\0')
1.149 schwarze 4360: return 0;
1.48 schwarze 4361: else if ('\\' == cp[pos] && '.' == cp[pos + 1])
4362: pos += 2;
4363: else if ('.' == cp[pos] || '\'' == cp[pos])
4364: pos++;
4365: else
1.149 schwarze 4366: return 0;
1.48 schwarze 4367:
4368: while (' ' == cp[pos] || '\t' == cp[pos])
4369: pos++;
4370:
4371: *ppos = pos;
1.149 schwarze 4372: return 1;
1.1 schwarze 4373: }