Annotation of src/usr.bin/mandoc/roff.c, Revision 1.245
1.245 ! schwarze 1: /* $OpenBSD: roff.c,v 1.244 2020/04/03 11:34:19 schwarze Exp $ */
1.1 schwarze 2: /*
1.244 schwarze 3: * Copyright (c) 2010-2015, 2017-2020 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);
193: static int roff_cond_text(ROFF_ARGS);
194: static int roff_cond_sub(ROFF_ARGS);
195: static int roff_ds(ROFF_ARGS);
196: static int roff_ec(ROFF_ARGS);
197: static int roff_eo(ROFF_ARGS);
198: static int roff_eqndelim(struct roff *, struct buf *, int);
1.244 schwarze 199: static int roff_evalcond(struct roff *, int, char *, int *);
1.106 schwarze 200: static int roff_evalnum(struct roff *, int,
201: const char *, int *, int *, int);
202: static int roff_evalpar(struct roff *, int,
1.133 schwarze 203: const char *, int *, int *, int);
1.71 schwarze 204: static int roff_evalstrcond(const char *, int *);
1.227 schwarze 205: static int roff_expand(struct roff *, struct buf *,
206: int, int, char);
1.42 schwarze 207: static void roff_free1(struct roff *);
1.52 schwarze 208: static void roff_freereg(struct roffreg *);
1.42 schwarze 209: static void roff_freestr(struct roffkv *);
1.85 schwarze 210: static size_t roff_getname(struct roff *, char **, int, int);
1.133 schwarze 211: static int roff_getnum(const char *, int *, int *, int);
1.56 schwarze 212: static int roff_getop(const char *, int *, char *);
1.199 schwarze 213: static int roff_getregn(struct roff *,
214: const char *, size_t, char);
1.145 schwarze 215: static int roff_getregro(const struct roff *,
216: const char *name);
1.197 schwarze 217: static const char *roff_getstrn(struct roff *,
1.187 schwarze 218: const char *, size_t, int *);
1.143 schwarze 219: static int roff_hasregn(const struct roff *,
220: const char *, size_t);
1.212 schwarze 221: static int roff_insec(ROFF_ARGS);
222: static int roff_it(ROFF_ARGS);
223: static int roff_line_ignore(ROFF_ARGS);
1.137 schwarze 224: static void roff_man_alloc1(struct roff_man *);
225: static void roff_man_free1(struct roff_man *);
1.212 schwarze 226: static int roff_manyarg(ROFF_ARGS);
1.230 schwarze 227: static int roff_noarg(ROFF_ARGS);
1.212 schwarze 228: static int roff_nop(ROFF_ARGS);
229: static int roff_nr(ROFF_ARGS);
230: static int roff_onearg(ROFF_ARGS);
1.166 schwarze 231: static enum roff_tok roff_parse(struct roff *, char *, int *,
1.87 schwarze 232: int, int);
1.212 schwarze 233: static int roff_parsetext(struct roff *, struct buf *,
1.177 schwarze 234: int, int *);
1.212 schwarze 235: static int roff_renamed(ROFF_ARGS);
236: static int roff_return(ROFF_ARGS);
237: static int roff_rm(ROFF_ARGS);
238: static int roff_rn(ROFF_ARGS);
239: static int roff_rr(ROFF_ARGS);
1.198 schwarze 240: static void roff_setregn(struct roff *, const char *,
1.199 schwarze 241: size_t, int, char, int);
1.8 schwarze 242: static void roff_setstr(struct roff *,
1.16 schwarze 243: const char *, const char *, int);
1.80 schwarze 244: static void roff_setstrn(struct roffkv **, const char *,
1.42 schwarze 245: size_t, const char *, size_t, int);
1.212 schwarze 246: static int roff_shift(ROFF_ARGS);
247: static int roff_so(ROFF_ARGS);
248: static int roff_tr(ROFF_ARGS);
249: static int roff_Dd(ROFF_ARGS);
250: static int roff_TE(ROFF_ARGS);
251: static int roff_TS(ROFF_ARGS);
252: static int roff_EQ(ROFF_ARGS);
253: static int roff_EN(ROFF_ARGS);
254: static int roff_T_(ROFF_ARGS);
255: static int roff_unsupp(ROFF_ARGS);
256: static int roff_userdef(ROFF_ARGS);
1.1 schwarze 257:
1.138 schwarze 258: /* --- constant data ------------------------------------------------------ */
259:
1.133 schwarze 260: #define ROFFNUM_SCALE (1 << 0) /* Honour scaling in roff_getnum(). */
261: #define ROFFNUM_WHITE (1 << 1) /* Skip whitespace in roff_evalnum(). */
262:
1.166 schwarze 263: const char *__roff_name[MAN_MAX + 1] = {
1.230 schwarze 264: "br", "ce", "fi", "ft",
265: "ll", "mc", "nf",
266: "po", "rj", "sp",
1.184 schwarze 267: "ta", "ti", NULL,
1.166 schwarze 268: "ab", "ad", "af", "aln",
269: "als", "am", "am1", "ami",
270: "ami1", "as", "as1", "asciify",
271: "backtrace", "bd", "bleedat", "blm",
272: "box", "boxa", "bp", "BP",
273: "break", "breakchar", "brnl", "brp",
1.177 schwarze 274: "brpnl", "c2", "cc",
1.166 schwarze 275: "cf", "cflags", "ch", "char",
276: "chop", "class", "close", "CL",
277: "color", "composite", "continue", "cp",
278: "cropat", "cs", "cu", "da",
279: "dch", "Dd", "de", "de1",
280: "defcolor", "dei", "dei1", "device",
281: "devicem", "di", "do", "ds",
282: "ds1", "dwh", "dt", "ec",
283: "ecr", "ecs", "el", "em",
284: "EN", "eo", "EP", "EQ",
285: "errprint", "ev", "evc", "ex",
286: "fallback", "fam", "fc", "fchar",
287: "fcolor", "fdeferlig", "feature", "fkern",
288: "fl", "flig", "fp", "fps",
289: "fschar", "fspacewidth", "fspecial", "ftr",
290: "fzoom", "gcolor", "hc", "hcode",
291: "hidechar", "hla", "hlm", "hpf",
292: "hpfa", "hpfcode", "hw", "hy",
293: "hylang", "hylen", "hym", "hypp",
294: "hys", "ie", "if", "ig",
295: "index", "it", "itc", "IX",
296: "kern", "kernafter", "kernbefore", "kernpair",
297: "lc", "lc_ctype", "lds", "length",
298: "letadj", "lf", "lg", "lhang",
299: "linetabs", "lnr", "lnrf", "lpfx",
1.176 schwarze 300: "ls", "lsm", "lt",
1.166 schwarze 301: "mediasize", "minss", "mk", "mso",
302: "na", "ne", "nh", "nhychar",
303: "nm", "nn", "nop", "nr",
304: "nrf", "nroff", "ns", "nx",
305: "open", "opena", "os", "output",
306: "padj", "papersize", "pc", "pev",
307: "pi", "PI", "pl", "pm",
1.184 schwarze 308: "pn", "pnr", "ps",
1.166 schwarze 309: "psbb", "pshape", "pso", "ptr",
310: "pvs", "rchar", "rd", "recursionlimit",
1.181 schwarze 311: "return", "rfschar", "rhang",
1.166 schwarze 312: "rm", "rn", "rnn", "rr",
313: "rs", "rt", "schar", "sentchar",
314: "shc", "shift", "sizes", "so",
315: "spacewidth", "special", "spreadwarn", "ss",
316: "sty", "substring", "sv", "sy",
1.172 schwarze 317: "T&", "tc", "TE",
1.173 schwarze 318: "TH", "tkf", "tl",
1.166 schwarze 319: "tm", "tm1", "tmc", "tr",
320: "track", "transchar", "trf", "trimat",
321: "trin", "trnt", "troff", "TS",
322: "uf", "ul", "unformat", "unwatch",
323: "unwatchn", "vpt", "vs", "warn",
324: "warnscale", "watch", "watchlength", "watchn",
325: "wh", "while", "write", "writec",
326: "writem", "xflag", ".", NULL,
1.178 schwarze 327: NULL, "text",
1.166 schwarze 328: "Dd", "Dt", "Os", "Sh",
329: "Ss", "Pp", "D1", "Dl",
330: "Bd", "Ed", "Bl", "El",
331: "It", "Ad", "An", "Ap",
332: "Ar", "Cd", "Cm", "Dv",
333: "Er", "Ev", "Ex", "Fa",
334: "Fd", "Fl", "Fn", "Ft",
335: "Ic", "In", "Li", "Nd",
336: "Nm", "Op", "Ot", "Pa",
337: "Rv", "St", "Va", "Vt",
338: "Xr", "%A", "%B", "%D",
339: "%I", "%J", "%N", "%O",
340: "%P", "%R", "%T", "%V",
341: "Ac", "Ao", "Aq", "At",
342: "Bc", "Bf", "Bo", "Bq",
343: "Bsx", "Bx", "Db", "Dc",
344: "Do", "Dq", "Ec", "Ef",
345: "Em", "Eo", "Fx", "Ms",
346: "No", "Ns", "Nx", "Ox",
347: "Pc", "Pf", "Po", "Pq",
348: "Qc", "Ql", "Qo", "Qq",
349: "Re", "Rs", "Sc", "So",
350: "Sq", "Sm", "Sx", "Sy",
351: "Tn", "Ux", "Xc", "Xo",
352: "Fo", "Fc", "Oo", "Oc",
353: "Bk", "Ek", "Bt", "Hf",
354: "Fr", "Ud", "Lb", "Lp",
355: "Lk", "Mt", "Brq", "Bro",
356: "Brc", "%C", "Es", "En",
1.171 schwarze 357: "Dx", "%Q", "%U", "Ta",
1.241 schwarze 358: "Tg", NULL,
1.166 schwarze 359: "TH", "SH", "SS", "TP",
1.204 schwarze 360: "TQ",
1.166 schwarze 361: "LP", "PP", "P", "IP",
362: "HP", "SM", "SB", "BI",
363: "IB", "BR", "RB", "R",
364: "B", "I", "IR", "RI",
365: "RE", "RS", "DT", "UC",
1.169 schwarze 366: "PD", "AT", "in",
1.205 schwarze 367: "SY", "YS", "OP",
368: "EX", "EE", "UR",
1.189 bentley 369: "UE", "MT", "ME", NULL
1.166 schwarze 370: };
371: const char *const *roff_name = __roff_name;
372:
373: static struct roffmac roffs[TOKEN_NONE] = {
1.230 schwarze 374: { roff_noarg, NULL, NULL, 0 }, /* br */
1.177 schwarze 375: { roff_onearg, NULL, NULL, 0 }, /* ce */
1.230 schwarze 376: { roff_noarg, NULL, NULL, 0 }, /* fi */
1.169 schwarze 377: { roff_onearg, NULL, NULL, 0 }, /* ft */
1.170 schwarze 378: { roff_onearg, NULL, NULL, 0 }, /* ll */
1.176 schwarze 379: { roff_onearg, NULL, NULL, 0 }, /* mc */
1.230 schwarze 380: { roff_noarg, NULL, NULL, 0 }, /* nf */
1.184 schwarze 381: { roff_onearg, NULL, NULL, 0 }, /* po */
1.181 schwarze 382: { roff_onearg, NULL, NULL, 0 }, /* rj */
1.171 schwarze 383: { roff_onearg, NULL, NULL, 0 }, /* sp */
1.172 schwarze 384: { roff_manyarg, NULL, NULL, 0 }, /* ta */
1.173 schwarze 385: { roff_onearg, NULL, NULL, 0 }, /* ti */
1.168 schwarze 386: { NULL, NULL, NULL, 0 }, /* ROFF_MAX */
1.167 schwarze 387: { roff_unsupp, NULL, NULL, 0 }, /* ab */
388: { roff_line_ignore, NULL, NULL, 0 }, /* ad */
389: { roff_line_ignore, NULL, NULL, 0 }, /* af */
390: { roff_unsupp, NULL, NULL, 0 }, /* aln */
1.183 schwarze 391: { roff_als, NULL, NULL, 0 }, /* als */
1.167 schwarze 392: { roff_block, roff_block_text, roff_block_sub, 0 }, /* am */
393: { roff_block, roff_block_text, roff_block_sub, 0 }, /* am1 */
394: { roff_block, roff_block_text, roff_block_sub, 0 }, /* ami */
395: { roff_block, roff_block_text, roff_block_sub, 0 }, /* ami1 */
396: { roff_ds, NULL, NULL, 0 }, /* as */
397: { roff_ds, NULL, NULL, 0 }, /* as1 */
398: { roff_unsupp, NULL, NULL, 0 }, /* asciify */
399: { roff_line_ignore, NULL, NULL, 0 }, /* backtrace */
400: { roff_line_ignore, NULL, NULL, 0 }, /* bd */
401: { roff_line_ignore, NULL, NULL, 0 }, /* bleedat */
402: { roff_unsupp, NULL, NULL, 0 }, /* blm */
403: { roff_unsupp, NULL, NULL, 0 }, /* box */
404: { roff_unsupp, NULL, NULL, 0 }, /* boxa */
405: { roff_line_ignore, NULL, NULL, 0 }, /* bp */
406: { roff_unsupp, NULL, NULL, 0 }, /* BP */
1.236 schwarze 407: { roff_break, NULL, NULL, 0 }, /* break */
1.167 schwarze 408: { roff_line_ignore, NULL, NULL, 0 }, /* breakchar */
409: { roff_line_ignore, NULL, NULL, 0 }, /* brnl */
1.230 schwarze 410: { roff_noarg, NULL, NULL, 0 }, /* brp */
1.167 schwarze 411: { roff_line_ignore, NULL, NULL, 0 }, /* brpnl */
412: { roff_unsupp, NULL, NULL, 0 }, /* c2 */
413: { roff_cc, NULL, NULL, 0 }, /* cc */
414: { roff_insec, NULL, NULL, 0 }, /* cf */
415: { roff_line_ignore, NULL, NULL, 0 }, /* cflags */
416: { roff_line_ignore, NULL, NULL, 0 }, /* ch */
1.213 schwarze 417: { roff_char, NULL, NULL, 0 }, /* char */
1.167 schwarze 418: { roff_unsupp, NULL, NULL, 0 }, /* chop */
419: { roff_line_ignore, NULL, NULL, 0 }, /* class */
420: { roff_insec, NULL, NULL, 0 }, /* close */
421: { roff_unsupp, NULL, NULL, 0 }, /* CL */
422: { roff_line_ignore, NULL, NULL, 0 }, /* color */
423: { roff_unsupp, NULL, NULL, 0 }, /* composite */
424: { roff_unsupp, NULL, NULL, 0 }, /* continue */
425: { roff_line_ignore, NULL, NULL, 0 }, /* cp */
426: { roff_line_ignore, NULL, NULL, 0 }, /* cropat */
427: { roff_line_ignore, NULL, NULL, 0 }, /* cs */
428: { roff_line_ignore, NULL, NULL, 0 }, /* cu */
429: { roff_unsupp, NULL, NULL, 0 }, /* da */
430: { roff_unsupp, NULL, NULL, 0 }, /* dch */
431: { roff_Dd, NULL, NULL, 0 }, /* Dd */
432: { roff_block, roff_block_text, roff_block_sub, 0 }, /* de */
433: { roff_block, roff_block_text, roff_block_sub, 0 }, /* de1 */
434: { roff_line_ignore, NULL, NULL, 0 }, /* defcolor */
435: { roff_block, roff_block_text, roff_block_sub, 0 }, /* dei */
436: { roff_block, roff_block_text, roff_block_sub, 0 }, /* dei1 */
437: { roff_unsupp, NULL, NULL, 0 }, /* device */
438: { roff_unsupp, NULL, NULL, 0 }, /* devicem */
439: { roff_unsupp, NULL, NULL, 0 }, /* di */
440: { roff_unsupp, NULL, NULL, 0 }, /* do */
441: { roff_ds, NULL, NULL, 0 }, /* ds */
442: { roff_ds, NULL, NULL, 0 }, /* ds1 */
443: { roff_unsupp, NULL, NULL, 0 }, /* dwh */
444: { roff_unsupp, NULL, NULL, 0 }, /* dt */
1.175 schwarze 445: { roff_ec, NULL, NULL, 0 }, /* ec */
1.167 schwarze 446: { roff_unsupp, NULL, NULL, 0 }, /* ecr */
447: { roff_unsupp, NULL, NULL, 0 }, /* ecs */
448: { roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT }, /* el */
449: { roff_unsupp, NULL, NULL, 0 }, /* em */
450: { roff_EN, NULL, NULL, 0 }, /* EN */
1.175 schwarze 451: { roff_eo, NULL, NULL, 0 }, /* eo */
1.167 schwarze 452: { roff_unsupp, NULL, NULL, 0 }, /* EP */
453: { roff_EQ, NULL, NULL, 0 }, /* EQ */
454: { roff_line_ignore, NULL, NULL, 0 }, /* errprint */
455: { roff_unsupp, NULL, NULL, 0 }, /* ev */
456: { roff_unsupp, NULL, NULL, 0 }, /* evc */
457: { roff_unsupp, NULL, NULL, 0 }, /* ex */
458: { roff_line_ignore, NULL, NULL, 0 }, /* fallback */
459: { roff_line_ignore, NULL, NULL, 0 }, /* fam */
460: { roff_unsupp, NULL, NULL, 0 }, /* fc */
461: { roff_unsupp, NULL, NULL, 0 }, /* fchar */
462: { roff_line_ignore, NULL, NULL, 0 }, /* fcolor */
463: { roff_line_ignore, NULL, NULL, 0 }, /* fdeferlig */
464: { roff_line_ignore, NULL, NULL, 0 }, /* feature */
465: { roff_line_ignore, NULL, NULL, 0 }, /* fkern */
466: { roff_line_ignore, NULL, NULL, 0 }, /* fl */
467: { roff_line_ignore, NULL, NULL, 0 }, /* flig */
468: { roff_line_ignore, NULL, NULL, 0 }, /* fp */
469: { roff_line_ignore, NULL, NULL, 0 }, /* fps */
470: { roff_unsupp, NULL, NULL, 0 }, /* fschar */
471: { roff_line_ignore, NULL, NULL, 0 }, /* fspacewidth */
472: { roff_line_ignore, NULL, NULL, 0 }, /* fspecial */
473: { roff_line_ignore, NULL, NULL, 0 }, /* ftr */
474: { roff_line_ignore, NULL, NULL, 0 }, /* fzoom */
475: { roff_line_ignore, NULL, NULL, 0 }, /* gcolor */
476: { roff_line_ignore, NULL, NULL, 0 }, /* hc */
477: { roff_line_ignore, NULL, NULL, 0 }, /* hcode */
478: { roff_line_ignore, NULL, NULL, 0 }, /* hidechar */
479: { roff_line_ignore, NULL, NULL, 0 }, /* hla */
480: { roff_line_ignore, NULL, NULL, 0 }, /* hlm */
481: { roff_line_ignore, NULL, NULL, 0 }, /* hpf */
482: { roff_line_ignore, NULL, NULL, 0 }, /* hpfa */
483: { roff_line_ignore, NULL, NULL, 0 }, /* hpfcode */
484: { roff_line_ignore, NULL, NULL, 0 }, /* hw */
485: { roff_line_ignore, NULL, NULL, 0 }, /* hy */
486: { roff_line_ignore, NULL, NULL, 0 }, /* hylang */
487: { roff_line_ignore, NULL, NULL, 0 }, /* hylen */
488: { roff_line_ignore, NULL, NULL, 0 }, /* hym */
489: { roff_line_ignore, NULL, NULL, 0 }, /* hypp */
490: { roff_line_ignore, NULL, NULL, 0 }, /* hys */
491: { roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT }, /* ie */
492: { roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT }, /* if */
493: { roff_block, roff_block_text, roff_block_sub, 0 }, /* ig */
494: { roff_unsupp, NULL, NULL, 0 }, /* index */
495: { roff_it, NULL, NULL, 0 }, /* it */
496: { roff_unsupp, NULL, NULL, 0 }, /* itc */
497: { roff_line_ignore, NULL, NULL, 0 }, /* IX */
498: { roff_line_ignore, NULL, NULL, 0 }, /* kern */
499: { roff_line_ignore, NULL, NULL, 0 }, /* kernafter */
500: { roff_line_ignore, NULL, NULL, 0 }, /* kernbefore */
501: { roff_line_ignore, NULL, NULL, 0 }, /* kernpair */
502: { roff_unsupp, NULL, NULL, 0 }, /* lc */
503: { roff_unsupp, NULL, NULL, 0 }, /* lc_ctype */
504: { roff_unsupp, NULL, NULL, 0 }, /* lds */
505: { roff_unsupp, NULL, NULL, 0 }, /* length */
506: { roff_line_ignore, NULL, NULL, 0 }, /* letadj */
507: { roff_insec, NULL, NULL, 0 }, /* lf */
508: { roff_line_ignore, NULL, NULL, 0 }, /* lg */
509: { roff_line_ignore, NULL, NULL, 0 }, /* lhang */
510: { roff_unsupp, NULL, NULL, 0 }, /* linetabs */
511: { roff_unsupp, NULL, NULL, 0 }, /* lnr */
512: { roff_unsupp, NULL, NULL, 0 }, /* lnrf */
513: { roff_unsupp, NULL, NULL, 0 }, /* lpfx */
514: { roff_line_ignore, NULL, NULL, 0 }, /* ls */
515: { roff_unsupp, NULL, NULL, 0 }, /* lsm */
516: { roff_line_ignore, NULL, NULL, 0 }, /* lt */
517: { roff_line_ignore, NULL, NULL, 0 }, /* mediasize */
518: { roff_line_ignore, NULL, NULL, 0 }, /* minss */
519: { roff_line_ignore, NULL, NULL, 0 }, /* mk */
520: { roff_insec, NULL, NULL, 0 }, /* mso */
521: { roff_line_ignore, NULL, NULL, 0 }, /* na */
522: { roff_line_ignore, NULL, NULL, 0 }, /* ne */
523: { roff_line_ignore, NULL, NULL, 0 }, /* nh */
524: { roff_line_ignore, NULL, NULL, 0 }, /* nhychar */
525: { roff_unsupp, NULL, NULL, 0 }, /* nm */
526: { roff_unsupp, NULL, NULL, 0 }, /* nn */
1.202 schwarze 527: { roff_nop, NULL, NULL, 0 }, /* nop */
1.167 schwarze 528: { roff_nr, NULL, NULL, 0 }, /* nr */
529: { roff_unsupp, NULL, NULL, 0 }, /* nrf */
530: { roff_line_ignore, NULL, NULL, 0 }, /* nroff */
531: { roff_line_ignore, NULL, NULL, 0 }, /* ns */
532: { roff_insec, NULL, NULL, 0 }, /* nx */
533: { roff_insec, NULL, NULL, 0 }, /* open */
534: { roff_insec, NULL, NULL, 0 }, /* opena */
535: { roff_line_ignore, NULL, NULL, 0 }, /* os */
536: { roff_unsupp, NULL, NULL, 0 }, /* output */
537: { roff_line_ignore, NULL, NULL, 0 }, /* padj */
538: { roff_line_ignore, NULL, NULL, 0 }, /* papersize */
539: { roff_line_ignore, NULL, NULL, 0 }, /* pc */
540: { roff_line_ignore, NULL, NULL, 0 }, /* pev */
541: { roff_insec, NULL, NULL, 0 }, /* pi */
542: { roff_unsupp, NULL, NULL, 0 }, /* PI */
543: { roff_line_ignore, NULL, NULL, 0 }, /* pl */
544: { roff_line_ignore, NULL, NULL, 0 }, /* pm */
545: { roff_line_ignore, NULL, NULL, 0 }, /* pn */
546: { roff_line_ignore, NULL, NULL, 0 }, /* pnr */
547: { roff_line_ignore, NULL, NULL, 0 }, /* ps */
548: { roff_unsupp, NULL, NULL, 0 }, /* psbb */
549: { roff_unsupp, NULL, NULL, 0 }, /* pshape */
550: { roff_insec, NULL, NULL, 0 }, /* pso */
551: { roff_line_ignore, NULL, NULL, 0 }, /* ptr */
552: { roff_line_ignore, NULL, NULL, 0 }, /* pvs */
553: { roff_unsupp, NULL, NULL, 0 }, /* rchar */
554: { roff_line_ignore, NULL, NULL, 0 }, /* rd */
555: { roff_line_ignore, NULL, NULL, 0 }, /* recursionlimit */
1.211 schwarze 556: { roff_return, NULL, NULL, 0 }, /* return */
1.167 schwarze 557: { roff_unsupp, NULL, NULL, 0 }, /* rfschar */
558: { roff_line_ignore, NULL, NULL, 0 }, /* rhang */
559: { roff_rm, NULL, NULL, 0 }, /* rm */
1.178 schwarze 560: { roff_rn, NULL, NULL, 0 }, /* rn */
1.167 schwarze 561: { roff_unsupp, NULL, NULL, 0 }, /* rnn */
562: { roff_rr, NULL, NULL, 0 }, /* rr */
563: { roff_line_ignore, NULL, NULL, 0 }, /* rs */
564: { roff_line_ignore, NULL, NULL, 0 }, /* rt */
565: { roff_unsupp, NULL, NULL, 0 }, /* schar */
566: { roff_line_ignore, NULL, NULL, 0 }, /* sentchar */
567: { roff_line_ignore, NULL, NULL, 0 }, /* shc */
1.211 schwarze 568: { roff_shift, NULL, NULL, 0 }, /* shift */
1.167 schwarze 569: { roff_line_ignore, NULL, NULL, 0 }, /* sizes */
570: { roff_so, NULL, NULL, 0 }, /* so */
571: { roff_line_ignore, NULL, NULL, 0 }, /* spacewidth */
572: { roff_line_ignore, NULL, NULL, 0 }, /* special */
573: { roff_line_ignore, NULL, NULL, 0 }, /* spreadwarn */
574: { roff_line_ignore, NULL, NULL, 0 }, /* ss */
575: { roff_line_ignore, NULL, NULL, 0 }, /* sty */
576: { roff_unsupp, NULL, NULL, 0 }, /* substring */
577: { roff_line_ignore, NULL, NULL, 0 }, /* sv */
578: { roff_insec, NULL, NULL, 0 }, /* sy */
579: { roff_T_, NULL, NULL, 0 }, /* T& */
580: { roff_unsupp, NULL, NULL, 0 }, /* tc */
581: { roff_TE, NULL, NULL, 0 }, /* TE */
1.187 schwarze 582: { roff_Dd, NULL, NULL, 0 }, /* TH */
1.167 schwarze 583: { roff_line_ignore, NULL, NULL, 0 }, /* tkf */
584: { roff_unsupp, NULL, NULL, 0 }, /* tl */
585: { roff_line_ignore, NULL, NULL, 0 }, /* tm */
586: { roff_line_ignore, NULL, NULL, 0 }, /* tm1 */
587: { roff_line_ignore, NULL, NULL, 0 }, /* tmc */
588: { roff_tr, NULL, NULL, 0 }, /* tr */
589: { roff_line_ignore, NULL, NULL, 0 }, /* track */
590: { roff_line_ignore, NULL, NULL, 0 }, /* transchar */
591: { roff_insec, NULL, NULL, 0 }, /* trf */
592: { roff_line_ignore, NULL, NULL, 0 }, /* trimat */
593: { roff_unsupp, NULL, NULL, 0 }, /* trin */
594: { roff_unsupp, NULL, NULL, 0 }, /* trnt */
595: { roff_line_ignore, NULL, NULL, 0 }, /* troff */
596: { roff_TS, NULL, NULL, 0 }, /* TS */
597: { roff_line_ignore, NULL, NULL, 0 }, /* uf */
598: { roff_line_ignore, NULL, NULL, 0 }, /* ul */
599: { roff_unsupp, NULL, NULL, 0 }, /* unformat */
600: { roff_line_ignore, NULL, NULL, 0 }, /* unwatch */
601: { roff_line_ignore, NULL, NULL, 0 }, /* unwatchn */
602: { roff_line_ignore, NULL, NULL, 0 }, /* vpt */
603: { roff_line_ignore, NULL, NULL, 0 }, /* vs */
604: { roff_line_ignore, NULL, NULL, 0 }, /* warn */
605: { roff_line_ignore, NULL, NULL, 0 }, /* warnscale */
606: { roff_line_ignore, NULL, NULL, 0 }, /* watch */
607: { roff_line_ignore, NULL, NULL, 0 }, /* watchlength */
608: { roff_line_ignore, NULL, NULL, 0 }, /* watchn */
609: { roff_unsupp, NULL, NULL, 0 }, /* wh */
1.212 schwarze 610: { roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT }, /*while*/
1.167 schwarze 611: { roff_insec, NULL, NULL, 0 }, /* write */
612: { roff_insec, NULL, NULL, 0 }, /* writec */
613: { roff_insec, NULL, NULL, 0 }, /* writem */
614: { roff_line_ignore, NULL, NULL, 0 }, /* xflag */
615: { roff_cblock, NULL, NULL, 0 }, /* . */
1.178 schwarze 616: { roff_renamed, NULL, NULL, 0 },
1.167 schwarze 617: { roff_userdef, NULL, NULL, 0 }
1.1 schwarze 618: };
619:
1.37 schwarze 620: /* Array of injected predefined strings. */
621: #define PREDEFS_MAX 38
622: static const struct predef predefs[PREDEFS_MAX] = {
623: #include "predefs.in"
624: };
625:
1.177 schwarze 626: static int roffce_lines; /* number of input lines to center */
627: static struct roff_node *roffce_node; /* active request */
1.51 schwarze 628: static int roffit_lines; /* number of lines to delay */
629: static char *roffit_macro; /* nil-terminated macro line */
630:
1.80 schwarze 631:
1.138 schwarze 632: /* --- request table ------------------------------------------------------ */
633:
1.167 schwarze 634: struct ohash *
635: roffhash_alloc(enum roff_tok mintok, enum roff_tok maxtok)
1.3 schwarze 636: {
1.167 schwarze 637: struct ohash *htab;
638: struct roffreq *req;
639: enum roff_tok tok;
640: size_t sz;
641: unsigned int slot;
1.3 schwarze 642:
1.167 schwarze 643: htab = mandoc_malloc(sizeof(*htab));
644: mandoc_ohash_init(htab, 8, offsetof(struct roffreq, name));
1.3 schwarze 645:
1.167 schwarze 646: for (tok = mintok; tok < maxtok; tok++) {
1.168 schwarze 647: if (roff_name[tok] == NULL)
648: continue;
1.167 schwarze 649: sz = strlen(roff_name[tok]);
650: req = mandoc_malloc(sizeof(*req) + sz + 1);
651: req->tok = tok;
652: memcpy(req->name, roff_name[tok], sz + 1);
653: slot = ohash_qlookup(htab, req->name);
654: ohash_insert(htab, slot, req);
1.3 schwarze 655: }
1.167 schwarze 656: return htab;
1.3 schwarze 657: }
658:
1.167 schwarze 659: void
660: roffhash_free(struct ohash *htab)
1.1 schwarze 661: {
1.167 schwarze 662: struct roffreq *req;
663: unsigned int slot;
1.1 schwarze 664:
1.167 schwarze 665: if (htab == NULL)
666: return;
667: for (req = ohash_first(htab, &slot); req != NULL;
668: req = ohash_next(htab, &slot))
669: free(req);
670: ohash_delete(htab);
671: free(htab);
672: }
673:
674: enum roff_tok
675: roffhash_find(struct ohash *htab, const char *name, size_t sz)
676: {
677: struct roffreq *req;
678: const char *end;
679:
680: if (sz) {
681: end = name + sz;
682: req = ohash_find(htab, ohash_qlookupi(htab, name, &end));
683: } else
684: req = ohash_find(htab, ohash_qlookup(htab, name));
685: return req == NULL ? TOKEN_NONE : req->tok;
1.1 schwarze 686: }
687:
1.138 schwarze 688: /* --- stack of request blocks -------------------------------------------- */
689:
1.1 schwarze 690: /*
691: * Pop the current node off of the stack of roff instructions currently
1.236 schwarze 692: * pending. Return 1 if it is a loop or 0 otherwise.
1.1 schwarze 693: */
1.212 schwarze 694: static int
1.1 schwarze 695: roffnode_pop(struct roff *r)
696: {
697: struct roffnode *p;
1.212 schwarze 698: int inloop;
1.1 schwarze 699:
1.80 schwarze 700: p = r->last;
1.212 schwarze 701: inloop = p->tok == ROFF_while;
702: r->last = p->parent;
1.16 schwarze 703: free(p->name);
704: free(p->end);
1.1 schwarze 705: free(p);
1.212 schwarze 706: return inloop;
1.1 schwarze 707: }
708:
709: /*
710: * Push a roff node onto the instruction stack. This must later be
711: * removed with roffnode_pop().
712: */
1.11 schwarze 713: static void
1.166 schwarze 714: roffnode_push(struct roff *r, enum roff_tok tok, const char *name,
1.16 schwarze 715: int line, int col)
1.1 schwarze 716: {
717: struct roffnode *p;
718:
1.11 schwarze 719: p = mandoc_calloc(1, sizeof(struct roffnode));
1.1 schwarze 720: p->tok = tok;
1.16 schwarze 721: if (name)
722: p->name = mandoc_strdup(name);
1.1 schwarze 723: p->parent = r->last;
724: p->line = line;
725: p->col = col;
1.71 schwarze 726: p->rule = p->parent ? p->parent->rule : 0;
1.1 schwarze 727:
728: r->last = p;
729: }
730:
1.138 schwarze 731: /* --- roff parser state data management ---------------------------------- */
732:
1.1 schwarze 733: static void
734: roff_free1(struct roff *r)
735: {
1.42 schwarze 736: int i;
1.27 schwarze 737:
1.218 schwarze 738: tbl_free(r->first_tbl);
1.27 schwarze 739: r->first_tbl = r->last_tbl = r->tbl = NULL;
1.1 schwarze 740:
1.219 schwarze 741: eqn_free(r->last_eqn);
1.191 schwarze 742: r->last_eqn = r->eqn = NULL;
1.32 schwarze 743:
1.211 schwarze 744: while (r->mstackpos >= 0)
745: roff_userret(r);
746:
1.1 schwarze 747: while (r->last)
748: roffnode_pop(r);
1.27 schwarze 749:
1.96 schwarze 750: free (r->rstack);
751: r->rstack = NULL;
752: r->rstacksz = 0;
753: r->rstackpos = -1;
754:
755: roff_freereg(r->regtab);
756: r->regtab = NULL;
757:
1.42 schwarze 758: roff_freestr(r->strtab);
1.178 schwarze 759: roff_freestr(r->rentab);
1.42 schwarze 760: roff_freestr(r->xmbtab);
1.178 schwarze 761: r->strtab = r->rentab = r->xmbtab = NULL;
1.42 schwarze 762:
763: if (r->xtab)
764: for (i = 0; i < 128; i++)
765: free(r->xtab[i].p);
766: free(r->xtab);
767: r->xtab = NULL;
1.1 schwarze 768: }
769:
770: void
771: roff_reset(struct roff *r)
772: {
773: roff_free1(r);
1.239 schwarze 774: r->options |= MPARSE_COMMENT;
1.99 schwarze 775: r->format = r->options & (MPARSE_MDOC | MPARSE_MAN);
1.175 schwarze 776: r->control = '\0';
777: r->escape = '\\';
1.179 schwarze 778: roffce_lines = 0;
779: roffce_node = NULL;
780: roffit_lines = 0;
781: roffit_macro = NULL;
1.1 schwarze 782: }
783:
784: void
785: roff_free(struct roff *r)
786: {
1.238 schwarze 787: int i;
1.211 schwarze 788:
1.1 schwarze 789: roff_free1(r);
1.211 schwarze 790: for (i = 0; i < r->mstacksz; i++)
791: free(r->mstack[i].argv);
792: free(r->mstack);
1.167 schwarze 793: roffhash_free(r->reqtab);
1.1 schwarze 794: free(r);
795: }
796:
797: struct roff *
1.223 schwarze 798: roff_alloc(int options)
1.1 schwarze 799: {
800: struct roff *r;
801:
1.11 schwarze 802: r = mandoc_calloc(1, sizeof(struct roff));
1.200 schwarze 803: r->reqtab = roffhash_alloc(0, ROFF_RENAMED);
1.239 schwarze 804: r->options = options | MPARSE_COMMENT;
1.99 schwarze 805: r->format = options & (MPARSE_MDOC | MPARSE_MAN);
1.211 schwarze 806: r->mstackpos = -1;
1.2 schwarze 807: r->rstackpos = -1;
1.175 schwarze 808: r->escape = '\\';
1.149 schwarze 809: return r;
1.137 schwarze 810: }
811:
1.138 schwarze 812: /* --- syntax tree state data management ---------------------------------- */
813:
1.137 schwarze 814: static void
815: roff_man_free1(struct roff_man *man)
816: {
1.228 schwarze 817: if (man->meta.first != NULL)
818: roff_node_delete(man, man->meta.first);
1.137 schwarze 819: free(man->meta.msec);
820: free(man->meta.vol);
821: free(man->meta.os);
822: free(man->meta.arch);
823: free(man->meta.title);
824: free(man->meta.name);
825: free(man->meta.date);
1.228 schwarze 826: free(man->meta.sodest);
1.137 schwarze 827: }
828:
1.229 schwarze 829: void
830: roff_state_reset(struct roff_man *man)
831: {
832: man->last = man->meta.first;
833: man->last_es = NULL;
834: man->flags = 0;
835: man->lastsec = man->lastnamed = SEC_NONE;
836: man->next = ROFF_NEXT_CHILD;
837: roff_setreg(man->roff, "nS", 0, '=');
838: }
839:
1.137 schwarze 840: static void
841: roff_man_alloc1(struct roff_man *man)
842: {
843: memset(&man->meta, 0, sizeof(man->meta));
1.228 schwarze 844: man->meta.first = mandoc_calloc(1, sizeof(*man->meta.first));
845: man->meta.first->type = ROFFT_ROOT;
846: man->meta.macroset = MACROSET_NONE;
1.229 schwarze 847: roff_state_reset(man);
1.137 schwarze 848: }
849:
850: void
851: roff_man_reset(struct roff_man *man)
852: {
853: roff_man_free1(man);
854: roff_man_alloc1(man);
855: }
856:
857: void
858: roff_man_free(struct roff_man *man)
859: {
860: roff_man_free1(man);
861: free(man);
862: }
863:
864: struct roff_man *
1.223 schwarze 865: roff_man_alloc(struct roff *roff, const char *os_s, int quick)
1.137 schwarze 866: {
867: struct roff_man *man;
868:
869: man = mandoc_calloc(1, sizeof(*man));
870: man->roff = roff;
1.188 schwarze 871: man->os_s = os_s;
1.137 schwarze 872: man->quick = quick;
873: roff_man_alloc1(man);
1.168 schwarze 874: roff->man = man;
1.149 schwarze 875: return man;
1.1 schwarze 876: }
877:
1.138 schwarze 878: /* --- syntax tree handling ----------------------------------------------- */
879:
880: struct roff_node *
881: roff_node_alloc(struct roff_man *man, int line, int pos,
882: enum roff_type type, int tok)
883: {
884: struct roff_node *n;
885:
886: n = mandoc_calloc(1, sizeof(*n));
887: n->line = line;
888: n->pos = pos;
889: n->tok = tok;
890: n->type = type;
891: n->sec = man->lastsec;
892:
893: if (man->flags & MDOC_SYNOPSIS)
1.157 schwarze 894: n->flags |= NODE_SYNPRETTY;
1.138 schwarze 895: else
1.157 schwarze 896: n->flags &= ~NODE_SYNPRETTY;
1.232 schwarze 897: if ((man->flags & (ROFF_NOFILL | ROFF_NONOFILL)) == ROFF_NOFILL)
1.231 schwarze 898: n->flags |= NODE_NOFILL;
899: else
900: n->flags &= ~NODE_NOFILL;
1.138 schwarze 901: if (man->flags & MDOC_NEWLINE)
1.157 schwarze 902: n->flags |= NODE_LINE;
1.138 schwarze 903: man->flags &= ~MDOC_NEWLINE;
904:
1.149 schwarze 905: return n;
1.138 schwarze 906: }
907:
908: void
909: roff_node_append(struct roff_man *man, struct roff_node *n)
910: {
911:
912: switch (man->next) {
913: case ROFF_NEXT_SIBLING:
1.153 schwarze 914: if (man->last->next != NULL) {
915: n->next = man->last->next;
916: man->last->next->prev = n;
917: } else
918: man->last->parent->last = n;
1.138 schwarze 919: man->last->next = n;
920: n->prev = man->last;
921: n->parent = man->last->parent;
922: break;
923: case ROFF_NEXT_CHILD:
1.159 schwarze 924: if (man->last->child != NULL) {
925: n->next = man->last->child;
926: man->last->child->prev = n;
927: } else
928: man->last->last = n;
1.138 schwarze 929: man->last->child = n;
930: n->parent = man->last;
931: break;
932: default:
933: abort();
934: }
1.154 schwarze 935: man->last = n;
1.138 schwarze 936:
937: switch (n->type) {
938: case ROFFT_HEAD:
939: n->parent->head = n;
940: break;
941: case ROFFT_BODY:
1.154 schwarze 942: if (n->end != ENDBODY_NOT)
943: return;
1.138 schwarze 944: n->parent->body = n;
945: break;
946: case ROFFT_TAIL:
947: n->parent->tail = n;
948: break;
949: default:
1.154 schwarze 950: return;
1.138 schwarze 951: }
1.154 schwarze 952:
953: /*
954: * Copy over the normalised-data pointer of our parent. Not
955: * everybody has one, but copying a null pointer is fine.
956: */
957:
958: n->norm = n->parent->norm;
959: assert(n->parent->type == ROFFT_BLOCK);
1.138 schwarze 960: }
961:
1.139 schwarze 962: void
963: roff_word_alloc(struct roff_man *man, int line, int pos, const char *word)
964: {
965: struct roff_node *n;
966:
967: n = roff_node_alloc(man, line, pos, ROFFT_TEXT, TOKEN_NONE);
968: n->string = roff_strdup(man->roff, word);
969: roff_node_append(man, n);
1.158 schwarze 970: n->flags |= NODE_VALID | NODE_ENDED;
1.139 schwarze 971: man->next = ROFF_NEXT_SIBLING;
972: }
973:
974: void
975: roff_word_append(struct roff_man *man, const char *word)
976: {
977: struct roff_node *n;
978: char *addstr, *newstr;
979:
980: n = man->last;
981: addstr = roff_strdup(man->roff, word);
982: mandoc_asprintf(&newstr, "%s %s", n->string, addstr);
983: free(addstr);
984: free(n->string);
985: n->string = newstr;
986: man->next = ROFF_NEXT_SIBLING;
1.140 schwarze 987: }
988:
989: void
990: roff_elem_alloc(struct roff_man *man, int line, int pos, int tok)
991: {
992: struct roff_node *n;
993:
994: n = roff_node_alloc(man, line, pos, ROFFT_ELEM, tok);
995: roff_node_append(man, n);
996: man->next = ROFF_NEXT_CHILD;
997: }
998:
999: struct roff_node *
1000: roff_block_alloc(struct roff_man *man, int line, int pos, int tok)
1001: {
1002: struct roff_node *n;
1003:
1004: n = roff_node_alloc(man, line, pos, ROFFT_BLOCK, tok);
1005: roff_node_append(man, n);
1006: man->next = ROFF_NEXT_CHILD;
1.149 schwarze 1007: return n;
1.139 schwarze 1008: }
1009:
1.138 schwarze 1010: struct roff_node *
1011: roff_head_alloc(struct roff_man *man, int line, int pos, int tok)
1012: {
1013: struct roff_node *n;
1014:
1015: n = roff_node_alloc(man, line, pos, ROFFT_HEAD, tok);
1016: roff_node_append(man, n);
1017: man->next = ROFF_NEXT_CHILD;
1.149 schwarze 1018: return n;
1.138 schwarze 1019: }
1020:
1021: struct roff_node *
1022: roff_body_alloc(struct roff_man *man, int line, int pos, int tok)
1023: {
1024: struct roff_node *n;
1025:
1026: n = roff_node_alloc(man, line, pos, ROFFT_BODY, tok);
1027: roff_node_append(man, n);
1028: man->next = ROFF_NEXT_CHILD;
1.149 schwarze 1029: return n;
1.139 schwarze 1030: }
1031:
1.193 schwarze 1032: static void
1.218 schwarze 1033: roff_addtbl(struct roff_man *man, int line, struct tbl_node *tbl)
1.139 schwarze 1034: {
1035: struct roff_node *n;
1.218 schwarze 1036: struct tbl_span *span;
1.139 schwarze 1037:
1.228 schwarze 1038: if (man->meta.macroset == MACROSET_MAN)
1.174 schwarze 1039: man_breakscope(man, ROFF_TS);
1.193 schwarze 1040: while ((span = tbl_span(tbl)) != NULL) {
1.218 schwarze 1041: n = roff_node_alloc(man, line, 0, ROFFT_TBL, TOKEN_NONE);
1.193 schwarze 1042: n->span = span;
1043: roff_node_append(man, n);
1044: n->flags |= NODE_VALID | NODE_ENDED;
1045: man->next = ROFF_NEXT_SIBLING;
1046: }
1.138 schwarze 1047: }
1048:
1049: void
1050: roff_node_unlink(struct roff_man *man, struct roff_node *n)
1051: {
1052:
1053: /* Adjust siblings. */
1054:
1055: if (n->prev)
1056: n->prev->next = n->next;
1057: if (n->next)
1058: n->next->prev = n->prev;
1059:
1060: /* Adjust parent. */
1061:
1062: if (n->parent != NULL) {
1063: if (n->parent->child == n)
1064: n->parent->child = n->next;
1065: if (n->parent->last == n)
1066: n->parent->last = n->prev;
1067: }
1068:
1069: /* Adjust parse point. */
1070:
1071: if (man == NULL)
1072: return;
1073: if (man->last == n) {
1074: if (n->prev == NULL) {
1075: man->last = n->parent;
1076: man->next = ROFF_NEXT_CHILD;
1077: } else {
1078: man->last = n->prev;
1079: man->next = ROFF_NEXT_SIBLING;
1080: }
1081: }
1.228 schwarze 1082: if (man->meta.first == n)
1083: man->meta.first = NULL;
1.216 schwarze 1084: }
1085:
1086: void
1087: roff_node_relink(struct roff_man *man, struct roff_node *n)
1088: {
1089: roff_node_unlink(man, n);
1090: n->prev = n->next = NULL;
1091: roff_node_append(man, n);
1.138 schwarze 1092: }
1093:
1094: void
1095: roff_node_free(struct roff_node *n)
1096: {
1097:
1098: if (n->args != NULL)
1099: mdoc_argv_free(n->args);
1100: if (n->type == ROFFT_BLOCK || n->type == ROFFT_ELEM)
1101: free(n->norm);
1.219 schwarze 1102: eqn_box_free(n->eqn);
1.138 schwarze 1103: free(n->string);
1104: free(n);
1105: }
1106:
1107: void
1108: roff_node_delete(struct roff_man *man, struct roff_node *n)
1109: {
1110:
1111: while (n->child != NULL)
1112: roff_node_delete(man, n->child);
1113: roff_node_unlink(man, n);
1114: roff_node_free(n);
1.242 schwarze 1115: }
1116:
1117: int
1118: roff_node_transparent(struct roff_node *n)
1119: {
1120: if (n == NULL)
1121: return 0;
1122: if (n->type == ROFFT_COMMENT || n->flags & NODE_NOPRT)
1123: return 1;
1.245 ! schwarze 1124: return roff_tok_transparent(n->tok);
! 1125: }
! 1126:
! 1127: int
! 1128: roff_tok_transparent(enum roff_tok tok)
! 1129: {
! 1130: switch (tok) {
1.242 schwarze 1131: case ROFF_ft:
1132: case ROFF_ll:
1133: case ROFF_mc:
1134: case ROFF_po:
1135: case ROFF_ta:
1136: case MDOC_Db:
1137: case MDOC_Es:
1138: case MDOC_Sm:
1139: case MDOC_Tg:
1140: case MAN_DT:
1141: case MAN_UC:
1142: case MAN_PD:
1143: case MAN_AT:
1144: return 1;
1145: default:
1146: return 0;
1147: }
1148: }
1149:
1150: struct roff_node *
1151: roff_node_child(struct roff_node *n)
1152: {
1153: for (n = n->child; roff_node_transparent(n); n = n->next)
1154: continue;
1155: return n;
1156: }
1157:
1158: struct roff_node *
1159: roff_node_prev(struct roff_node *n)
1160: {
1161: do {
1162: n = n->prev;
1163: } while (roff_node_transparent(n));
1164: return n;
1165: }
1166:
1167: struct roff_node *
1168: roff_node_next(struct roff_node *n)
1169: {
1170: do {
1171: n = n->next;
1172: } while (roff_node_transparent(n));
1173: return n;
1.141 schwarze 1174: }
1175:
1176: void
1177: deroff(char **dest, const struct roff_node *n)
1178: {
1179: char *cp;
1180: size_t sz;
1181:
1.243 schwarze 1182: if (n->string == NULL) {
1.141 schwarze 1183: for (n = n->child; n != NULL; n = n->next)
1184: deroff(dest, n);
1185: return;
1186: }
1187:
1.160 schwarze 1188: /* Skip leading whitespace. */
1.141 schwarze 1189:
1.160 schwarze 1190: for (cp = n->string; *cp != '\0'; cp++) {
1.161 schwarze 1191: if (cp[0] == '\\' && cp[1] != '\0' &&
1192: strchr(" %&0^|~", cp[1]) != NULL)
1.141 schwarze 1193: cp++;
1.160 schwarze 1194: else if ( ! isspace((unsigned char)*cp))
1.141 schwarze 1195: break;
1196: }
1197:
1.161 schwarze 1198: /* Skip trailing backslash. */
1199:
1200: sz = strlen(cp);
1.162 schwarze 1201: if (sz > 0 && cp[sz - 1] == '\\')
1.161 schwarze 1202: sz--;
1203:
1.141 schwarze 1204: /* Skip trailing whitespace. */
1205:
1.161 schwarze 1206: for (; sz; sz--)
1.141 schwarze 1207: if ( ! isspace((unsigned char)cp[sz-1]))
1208: break;
1209:
1210: /* Skip empty strings. */
1211:
1212: if (sz == 0)
1213: return;
1214:
1215: if (*dest == NULL) {
1216: *dest = mandoc_strndup(cp, sz);
1217: return;
1218: }
1219:
1220: mandoc_asprintf(&cp, "%s %*s", *dest, (int)sz, cp);
1221: free(*dest);
1222: *dest = cp;
1.138 schwarze 1223: }
1224:
1225: /* --- main functions of the roff parser ---------------------------------- */
1226:
1.8 schwarze 1227: /*
1.227 schwarze 1228: * In the current line, expand escape sequences that produce parsable
1229: * input text. Also check the syntax of the remaining escape sequences,
1230: * which typically produce output glyphs or change formatter state.
1.8 schwarze 1231: */
1.212 schwarze 1232: static int
1.227 schwarze 1233: roff_expand(struct roff *r, struct buf *buf, int ln, int pos, char newesc)
1.8 schwarze 1234: {
1.211 schwarze 1235: struct mctx *ctx; /* current macro call context */
1.81 schwarze 1236: char ubuf[24]; /* buffer to print the number */
1.200 schwarze 1237: struct roff_node *n; /* used for header comments */
1.78 schwarze 1238: const char *start; /* start of the string to process */
1.82 schwarze 1239: char *stesc; /* start of an escape sequence ('\\') */
1.224 schwarze 1240: const char *esct; /* type of esccape sequence */
1.200 schwarze 1241: char *ep; /* end of comment string */
1.23 schwarze 1242: const char *stnam; /* start of the name, after "[(*" */
1243: const char *cp; /* end of the name, e.g. before ']' */
1244: const char *res; /* the string to be substituted */
1.110 schwarze 1245: char *nbuf; /* new buffer to copy buf->buf to */
1.53 schwarze 1246: size_t maxl; /* expected length of the escape name */
1247: size_t naml; /* actual length of the escape name */
1.211 schwarze 1248: size_t asz; /* length of the replacement */
1249: size_t rsz; /* length of the rest of the string */
1.109 schwarze 1250: int inaml; /* length returned from mandoc_escape() */
1.53 schwarze 1251: int expand_count; /* to avoid infinite loops */
1.79 schwarze 1252: int npos; /* position in numeric expression */
1.91 schwarze 1253: int arg_complete; /* argument not interrupted by eol */
1.211 schwarze 1254: int quote_args; /* true for \\$@, false for \\$* */
1.175 schwarze 1255: int done; /* no more input available */
1.187 schwarze 1256: int deftype; /* type of definition to paste */
1.195 schwarze 1257: int rcsid; /* kind of RCS id seen */
1.224 schwarze 1258: enum mandocerr err; /* for escape sequence problems */
1.199 schwarze 1259: char sign; /* increment number register */
1.79 schwarze 1260: char term; /* character terminating the escape */
1.8 schwarze 1261:
1.175 schwarze 1262: /* Search forward for comments. */
1263:
1264: done = 0;
1265: start = buf->buf + pos;
1266: for (stesc = buf->buf + pos; *stesc != '\0'; stesc++) {
1.227 schwarze 1267: if (stesc[0] != newesc || stesc[1] == '\0')
1.175 schwarze 1268: continue;
1269: stesc++;
1270: if (*stesc != '"' && *stesc != '#')
1271: continue;
1.185 schwarze 1272:
1273: /* Comment found, look for RCS id. */
1274:
1.195 schwarze 1275: rcsid = 0;
1.185 schwarze 1276: if ((cp = strstr(stesc, "$" "OpenBSD")) != NULL) {
1.195 schwarze 1277: rcsid = 1 << MANDOC_OS_OPENBSD;
1.185 schwarze 1278: cp += 8;
1279: } else if ((cp = strstr(stesc, "$" "NetBSD")) != NULL) {
1.195 schwarze 1280: rcsid = 1 << MANDOC_OS_NETBSD;
1.185 schwarze 1281: cp += 7;
1282: }
1283: if (cp != NULL &&
1284: isalnum((unsigned char)*cp) == 0 &&
1.186 schwarze 1285: strchr(cp, '$') != NULL) {
1.195 schwarze 1286: if (r->man->meta.rcsids & rcsid)
1.222 schwarze 1287: mandoc_msg(MANDOCERR_RCS_REP, ln,
1288: (int)(stesc - buf->buf) + 1,
1289: "%s", stesc + 1);
1.195 schwarze 1290: r->man->meta.rcsids |= rcsid;
1.186 schwarze 1291: }
1.185 schwarze 1292:
1293: /* Handle trailing whitespace. */
1294:
1.200 schwarze 1295: ep = strchr(stesc--, '\0') - 1;
1296: if (*ep == '\n') {
1.175 schwarze 1297: done = 1;
1.200 schwarze 1298: ep--;
1.175 schwarze 1299: }
1.200 schwarze 1300: if (*ep == ' ' || *ep == '\t')
1.222 schwarze 1301: mandoc_msg(MANDOCERR_SPACE_EOL,
1302: ln, (int)(ep - buf->buf), NULL);
1.200 schwarze 1303:
1304: /*
1305: * Save comments preceding the title macro
1306: * in the syntax tree.
1307: */
1308:
1.239 schwarze 1309: if (newesc != ASCII_ESC && r->options & MPARSE_COMMENT) {
1.200 schwarze 1310: while (*ep == ' ' || *ep == '\t')
1311: ep--;
1312: ep[1] = '\0';
1313: n = roff_node_alloc(r->man,
1314: ln, stesc + 1 - buf->buf,
1315: ROFFT_COMMENT, TOKEN_NONE);
1316: n->string = mandoc_strdup(stesc + 2);
1317: roff_node_append(r->man, n);
1318: n->flags |= NODE_VALID | NODE_ENDED;
1319: r->man->next = ROFF_NEXT_SIBLING;
1320: }
1321:
1.206 schwarze 1322: /* Line continuation with comment. */
1323:
1324: if (stesc[1] == '#') {
1325: *stesc = '\0';
1.212 schwarze 1326: return ROFF_IGN | ROFF_APPEND;
1.206 schwarze 1327: }
1328:
1329: /* Discard normal comments. */
1.200 schwarze 1330:
1.207 schwarze 1331: while (stesc > start && stesc[-1] == ' ' &&
1332: (stesc == start + 1 || stesc[-2] != '\\'))
1.175 schwarze 1333: stesc--;
1334: *stesc = '\0';
1335: break;
1336: }
1337: if (stesc == start)
1338: return ROFF_CONT;
1339: stesc--;
1340:
1341: /* Notice the end of the input. */
1342:
1343: if (*stesc == '\n') {
1344: *stesc-- = '\0';
1345: done = 1;
1346: }
1347:
1.43 schwarze 1348: expand_count = 0;
1.175 schwarze 1349: while (stesc >= start) {
1.227 schwarze 1350: if (*stesc != newesc) {
1.43 schwarze 1351:
1.227 schwarze 1352: /*
1353: * If we have a non-standard escape character,
1354: * escape literal backslashes because all
1355: * processing in subsequent functions uses
1356: * the standard escaping rules.
1357: */
1.78 schwarze 1358:
1.227 schwarze 1359: if (newesc != ASCII_ESC && *stesc == '\\') {
1.175 schwarze 1360: *stesc = '\0';
1361: buf->sz = mandoc_asprintf(&nbuf, "%s\\e%s",
1362: buf->buf, stesc + 1) + 1;
1363: start = nbuf + pos;
1364: stesc = nbuf + (stesc - buf->buf);
1365: free(buf->buf);
1366: buf->buf = nbuf;
1367: }
1.227 schwarze 1368:
1369: /* Search backwards for the next escape. */
1370:
1.175 schwarze 1371: stesc--;
1.78 schwarze 1372: continue;
1.175 schwarze 1373: }
1.78 schwarze 1374:
1375: /* If it is escaped, skip it. */
1376:
1377: for (cp = stesc - 1; cp >= start; cp--)
1.175 schwarze 1378: if (*cp != r->escape)
1.78 schwarze 1379: break;
1380:
1.110 schwarze 1381: if ((stesc - cp) % 2 == 0) {
1.175 schwarze 1382: while (stesc > cp)
1383: *stesc-- = '\\';
1.78 schwarze 1384: continue;
1.175 schwarze 1385: } else if (stesc[1] != '\0') {
1386: *stesc = '\\';
1387: } else {
1388: *stesc-- = '\0';
1389: if (done)
1390: continue;
1391: else
1.212 schwarze 1392: return ROFF_IGN | ROFF_APPEND;
1.78 schwarze 1393: }
1.23 schwarze 1394:
1.79 schwarze 1395: /* Decide whether to expand or to check only. */
1.23 schwarze 1396:
1.79 schwarze 1397: term = '\0';
1.78 schwarze 1398: cp = stesc + 1;
1.224 schwarze 1399: if (*cp == 'E')
1400: cp++;
1401: esct = cp;
1402: switch (*esct) {
1.80 schwarze 1403: case '*':
1.211 schwarze 1404: case '$':
1.53 schwarze 1405: res = NULL;
1406: break;
1.80 schwarze 1407: case 'B':
1408: case 'w':
1.79 schwarze 1409: term = cp[1];
1410: /* FALLTHROUGH */
1.80 schwarze 1411: case 'n':
1.199 schwarze 1412: sign = cp[1];
1413: if (sign == '+' || sign == '-')
1414: cp++;
1.53 schwarze 1415: res = ubuf;
1416: break;
1417: default:
1.224 schwarze 1418: err = MANDOCERR_OK;
1419: switch(mandoc_escape(&cp, &stnam, &inaml)) {
1420: case ESCAPE_SPECIAL:
1421: if (mchars_spec2cp(stnam, inaml) >= 0)
1422: break;
1423: /* FALLTHROUGH */
1424: case ESCAPE_ERROR:
1425: err = MANDOCERR_ESC_BAD;
1426: break;
1427: case ESCAPE_UNDEF:
1428: err = MANDOCERR_ESC_UNDEF;
1429: break;
1430: case ESCAPE_UNSUPP:
1431: err = MANDOCERR_ESC_UNSUPP;
1432: break;
1433: default:
1434: break;
1435: }
1436: if (err != MANDOCERR_OK)
1437: mandoc_msg(err, ln, (int)(stesc - buf->buf),
1.92 schwarze 1438: "%.*s", (int)(cp - stesc), stesc);
1.175 schwarze 1439: stesc--;
1.78 schwarze 1440: continue;
1.42 schwarze 1441: }
1442:
1.78 schwarze 1443: if (EXPAND_LIMIT < ++expand_count) {
1.222 schwarze 1444: mandoc_msg(MANDOCERR_ROFFLOOP,
1.110 schwarze 1445: ln, (int)(stesc - buf->buf), NULL);
1.149 schwarze 1446: return ROFF_IGN;
1.78 schwarze 1447: }
1.23 schwarze 1448:
1449: /*
1450: * The third character decides the length
1.53 schwarze 1451: * of the name of the string or register.
1.23 schwarze 1452: * Save a pointer to the name.
1453: */
1454:
1.110 schwarze 1455: if (term == '\0') {
1.79 schwarze 1456: switch (*++cp) {
1.80 schwarze 1457: case '\0':
1.79 schwarze 1458: maxl = 0;
1459: break;
1.80 schwarze 1460: case '(':
1.79 schwarze 1461: cp++;
1462: maxl = 2;
1463: break;
1.80 schwarze 1464: case '[':
1.79 schwarze 1465: cp++;
1466: term = ']';
1467: maxl = 0;
1468: break;
1469: default:
1470: maxl = 1;
1471: break;
1472: }
1473: } else {
1474: cp += 2;
1.8 schwarze 1475: maxl = 0;
1476: }
1.23 schwarze 1477: stnam = cp;
1.8 schwarze 1478:
1.23 schwarze 1479: /* Advance to the end of the name. */
1.8 schwarze 1480:
1.125 schwarze 1481: naml = 0;
1.91 schwarze 1482: arg_complete = 1;
1.125 schwarze 1483: while (maxl == 0 || naml < maxl) {
1.110 schwarze 1484: if (*cp == '\0') {
1.222 schwarze 1485: mandoc_msg(MANDOCERR_ESC_BAD, ln,
1486: (int)(stesc - buf->buf), "%s", stesc);
1.91 schwarze 1487: arg_complete = 0;
1.79 schwarze 1488: break;
1.42 schwarze 1489: }
1.110 schwarze 1490: if (maxl == 0 && *cp == term) {
1.79 schwarze 1491: cp++;
1.125 schwarze 1492: break;
1493: }
1.224 schwarze 1494: if (*cp++ != '\\' || *esct != 'w') {
1.125 schwarze 1495: naml++;
1496: continue;
1497: }
1498: switch (mandoc_escape(&cp, NULL, NULL)) {
1499: case ESCAPE_SPECIAL:
1500: case ESCAPE_UNICODE:
1501: case ESCAPE_NUMBERED:
1.224 schwarze 1502: case ESCAPE_UNDEF:
1.125 schwarze 1503: case ESCAPE_OVERSTRIKE:
1504: naml++;
1505: break;
1506: default:
1.8 schwarze 1507: break;
1.79 schwarze 1508: }
1.8 schwarze 1509: }
1510:
1.23 schwarze 1511: /*
1512: * Retrieve the replacement string; if it is
1513: * undefined, resume searching for escapes.
1514: */
1515:
1.224 schwarze 1516: switch (*esct) {
1.80 schwarze 1517: case '*':
1.187 schwarze 1518: if (arg_complete) {
1519: deftype = ROFFDEF_USER | ROFFDEF_PRE;
1520: res = roff_getstrn(r, stnam, naml, &deftype);
1.203 schwarze 1521:
1522: /*
1523: * If not overriden, let \*(.T
1524: * through to the formatters.
1525: */
1526:
1527: if (res == NULL && naml == 2 &&
1528: stnam[0] == '.' && stnam[1] == 'T') {
1529: roff_setstrn(&r->strtab,
1530: ".T", 2, NULL, 0, 0);
1531: stesc--;
1532: continue;
1533: }
1.187 schwarze 1534: }
1.79 schwarze 1535: break;
1.211 schwarze 1536: case '$':
1537: if (r->mstackpos < 0) {
1.222 schwarze 1538: mandoc_msg(MANDOCERR_ARG_UNDEF, ln,
1539: (int)(stesc - buf->buf), "%.3s", stesc);
1.211 schwarze 1540: break;
1541: }
1542: ctx = r->mstack + r->mstackpos;
1.224 schwarze 1543: npos = esct[1] - '1';
1.211 schwarze 1544: if (npos >= 0 && npos <= 8) {
1545: res = npos < ctx->argc ?
1546: ctx->argv[npos] : "";
1547: break;
1548: }
1.224 schwarze 1549: if (esct[1] == '*')
1.211 schwarze 1550: quote_args = 0;
1.224 schwarze 1551: else if (esct[1] == '@')
1.211 schwarze 1552: quote_args = 1;
1553: else {
1.222 schwarze 1554: mandoc_msg(MANDOCERR_ARG_NONUM, ln,
1555: (int)(stesc - buf->buf), "%.3s", stesc);
1.211 schwarze 1556: break;
1557: }
1558: asz = 0;
1559: for (npos = 0; npos < ctx->argc; npos++) {
1560: if (npos)
1561: asz++; /* blank */
1562: if (quote_args)
1563: asz += 2; /* quotes */
1564: asz += strlen(ctx->argv[npos]);
1565: }
1566: if (asz != 3) {
1567: rsz = buf->sz - (stesc - buf->buf) - 3;
1568: if (asz < 3)
1569: memmove(stesc + asz, stesc + 3, rsz);
1570: buf->sz += asz - 3;
1571: nbuf = mandoc_realloc(buf->buf, buf->sz);
1572: start = nbuf + pos;
1573: stesc = nbuf + (stesc - buf->buf);
1574: buf->buf = nbuf;
1575: if (asz > 3)
1576: memmove(stesc + asz, stesc + 3, rsz);
1577: }
1578: for (npos = 0; npos < ctx->argc; npos++) {
1579: if (npos)
1580: *stesc++ = ' ';
1581: if (quote_args)
1582: *stesc++ = '"';
1583: cp = ctx->argv[npos];
1584: while (*cp != '\0')
1585: *stesc++ = *cp++;
1586: if (quote_args)
1587: *stesc++ = '"';
1588: }
1589: continue;
1.80 schwarze 1590: case 'B':
1.79 schwarze 1591: npos = 0;
1.91 schwarze 1592: ubuf[0] = arg_complete &&
1.133 schwarze 1593: roff_evalnum(r, ln, stnam, &npos,
1594: NULL, ROFFNUM_SCALE) &&
1.91 schwarze 1595: stnam + npos + 1 == cp ? '1' : '0';
1.79 schwarze 1596: ubuf[1] = '\0';
1597: break;
1.80 schwarze 1598: case 'n':
1.91 schwarze 1599: if (arg_complete)
1600: (void)snprintf(ubuf, sizeof(ubuf), "%d",
1.199 schwarze 1601: roff_getregn(r, stnam, naml, sign));
1.91 schwarze 1602: else
1603: ubuf[0] = '\0';
1.79 schwarze 1604: break;
1.80 schwarze 1605: case 'w':
1.91 schwarze 1606: /* use even incomplete args */
1.81 schwarze 1607: (void)snprintf(ubuf, sizeof(ubuf), "%d",
1.79 schwarze 1608: 24 * (int)naml);
1609: break;
1610: }
1.8 schwarze 1611:
1.110 schwarze 1612: if (res == NULL) {
1.224 schwarze 1613: if (*esct == '*')
1.222 schwarze 1614: mandoc_msg(MANDOCERR_STR_UNDEF,
1615: ln, (int)(stesc - buf->buf),
1.211 schwarze 1616: "%.*s", (int)naml, stnam);
1.37 schwarze 1617: res = "";
1.118 schwarze 1618: } else if (buf->sz + strlen(res) > SHRT_MAX) {
1.222 schwarze 1619: mandoc_msg(MANDOCERR_ROFFLOOP,
1.118 schwarze 1620: ln, (int)(stesc - buf->buf), NULL);
1.149 schwarze 1621: return ROFF_IGN;
1.8 schwarze 1622: }
1623:
1.23 schwarze 1624: /* Replace the escape sequence by the string. */
1625:
1.82 schwarze 1626: *stesc = '\0';
1.110 schwarze 1627: buf->sz = mandoc_asprintf(&nbuf, "%s%s%s",
1628: buf->buf, res, cp) + 1;
1.8 schwarze 1629:
1.78 schwarze 1630: /* Prepare for the next replacement. */
1.8 schwarze 1631:
1.78 schwarze 1632: start = nbuf + pos;
1.110 schwarze 1633: stesc = nbuf + (stesc - buf->buf) + strlen(res);
1634: free(buf->buf);
1635: buf->buf = nbuf;
1.42 schwarze 1636: }
1.149 schwarze 1637: return ROFF_CONT;
1.42 schwarze 1638: }
1.225 schwarze 1639:
1640: /*
1641: * Parse a quoted or unquoted roff-style request or macro argument.
1642: * Return a pointer to the parsed argument, which is either the original
1643: * pointer or advanced by one byte in case the argument is quoted.
1644: * NUL-terminate the argument in place.
1645: * Collapse pairs of quotes inside quoted arguments.
1646: * Advance the argument pointer to the next argument,
1647: * or to the NUL byte terminating the argument line.
1648: */
1649: char *
1.227 schwarze 1650: roff_getarg(struct roff *r, char **cpp, int ln, int *pos)
1.225 schwarze 1651: {
1.227 schwarze 1652: struct buf buf;
1.238 schwarze 1653: char *cp, *start;
1.227 schwarze 1654: int newesc, pairs, quoted, white;
1.225 schwarze 1655:
1656: /* Quoting can only start with a new word. */
1657: start = *cpp;
1658: quoted = 0;
1659: if ('"' == *start) {
1660: quoted = 1;
1661: start++;
1662: }
1663:
1.227 schwarze 1664: newesc = pairs = white = 0;
1.225 schwarze 1665: for (cp = start; '\0' != *cp; cp++) {
1666:
1667: /*
1668: * Move the following text left
1669: * after quoted quotes and after "\\" and "\t".
1670: */
1671: if (pairs)
1672: cp[-pairs] = cp[0];
1673:
1674: if ('\\' == cp[0]) {
1675: /*
1676: * In copy mode, translate double to single
1677: * backslashes and backslash-t to literal tabs.
1678: */
1679: switch (cp[1]) {
1680: case 'a':
1681: case 't':
1.226 schwarze 1682: cp[-pairs] = '\t';
1.227 schwarze 1683: pairs++;
1684: cp++;
1685: break;
1.225 schwarze 1686: case '\\':
1.227 schwarze 1687: newesc = 1;
1688: cp[-pairs] = ASCII_ESC;
1.225 schwarze 1689: pairs++;
1690: cp++;
1691: break;
1692: case ' ':
1693: /* Skip escaped blanks. */
1694: if (0 == quoted)
1695: cp++;
1696: break;
1697: default:
1698: break;
1699: }
1700: } else if (0 == quoted) {
1701: if (' ' == cp[0]) {
1702: /* Unescaped blanks end unquoted args. */
1703: white = 1;
1704: break;
1705: }
1706: } else if ('"' == cp[0]) {
1707: if ('"' == cp[1]) {
1708: /* Quoted quotes collapse. */
1709: pairs++;
1710: cp++;
1711: } else {
1712: /* Unquoted quotes end quoted args. */
1713: quoted = 2;
1714: break;
1715: }
1716: }
1717: }
1718:
1719: /* Quoted argument without a closing quote. */
1720: if (1 == quoted)
1721: mandoc_msg(MANDOCERR_ARG_QUOTE, ln, *pos, NULL);
1722:
1723: /* NUL-terminate this argument and move to the next one. */
1724: if (pairs)
1725: cp[-pairs] = '\0';
1726: if ('\0' != *cp) {
1727: *cp++ = '\0';
1728: while (' ' == *cp)
1729: cp++;
1730: }
1731: *pos += (int)(cp - start) + (quoted ? 1 : 0);
1732: *cpp = cp;
1733:
1734: if ('\0' == *cp && (white || ' ' == cp[-1]))
1735: mandoc_msg(MANDOCERR_SPACE_EOL, ln, *pos, NULL);
1736:
1.227 schwarze 1737: start = mandoc_strdup(start);
1738: if (newesc == 0)
1739: return start;
1740:
1741: buf.buf = start;
1742: buf.sz = strlen(start) + 1;
1743: buf.next = NULL;
1744: if (roff_expand(r, &buf, ln, 0, ASCII_ESC) & ROFF_IGN) {
1745: free(buf.buf);
1746: buf.buf = mandoc_strdup("");
1747: }
1748: return buf.buf;
1.225 schwarze 1749: }
1750:
1.42 schwarze 1751:
1752: /*
1.147 schwarze 1753: * Process text streams.
1.42 schwarze 1754: */
1.212 schwarze 1755: static int
1.177 schwarze 1756: roff_parsetext(struct roff *r, struct buf *buf, int pos, int *offs)
1.42 schwarze 1757: {
1758: size_t sz;
1759: const char *start;
1.51 schwarze 1760: char *p;
1761: int isz;
1.42 schwarze 1762: enum mandoc_esc esc;
1763:
1.147 schwarze 1764: /* Spring the input line trap. */
1765:
1766: if (roffit_lines == 1) {
1767: isz = mandoc_asprintf(&p, "%s\n.%s", buf->buf, roffit_macro);
1768: free(buf->buf);
1769: buf->buf = p;
1770: buf->sz = isz + 1;
1771: *offs = 0;
1772: free(roffit_macro);
1773: roffit_lines = 0;
1.149 schwarze 1774: return ROFF_REPARSE;
1.147 schwarze 1775: } else if (roffit_lines > 1)
1776: --roffit_lines;
1777:
1.177 schwarze 1778: if (roffce_node != NULL && buf->buf[pos] != '\0') {
1779: if (roffce_lines < 1) {
1780: r->man->last = roffce_node;
1781: r->man->next = ROFF_NEXT_SIBLING;
1782: roffce_lines = 0;
1783: roffce_node = NULL;
1784: } else
1785: roffce_lines--;
1786: }
1787:
1.147 schwarze 1788: /* Convert all breakable hyphens into ASCII_HYPH. */
1789:
1.110 schwarze 1790: start = p = buf->buf + pos;
1.42 schwarze 1791:
1.110 schwarze 1792: while (*p != '\0') {
1.42 schwarze 1793: sz = strcspn(p, "-\\");
1794: p += sz;
1795:
1.110 schwarze 1796: if (*p == '\0')
1.42 schwarze 1797: break;
1798:
1.110 schwarze 1799: if (*p == '\\') {
1.42 schwarze 1800: /* Skip over escapes. */
1801: p++;
1.62 schwarze 1802: esc = mandoc_escape((const char **)&p, NULL, NULL);
1.110 schwarze 1803: if (esc == ESCAPE_ERROR)
1.42 schwarze 1804: break;
1.136 schwarze 1805: while (*p == '-')
1806: p++;
1.42 schwarze 1807: continue;
1808: } else if (p == start) {
1809: p++;
1810: continue;
1811: }
1812:
1.44 schwarze 1813: if (isalpha((unsigned char)p[-1]) &&
1814: isalpha((unsigned char)p[1]))
1.42 schwarze 1815: *p = ASCII_HYPH;
1816: p++;
1.8 schwarze 1817: }
1.149 schwarze 1818: return ROFF_CONT;
1.8 schwarze 1819: }
1820:
1.212 schwarze 1821: int
1.110 schwarze 1822: roff_parseln(struct roff *r, int ln, struct buf *buf, int *offs)
1.1 schwarze 1823: {
1.166 schwarze 1824: enum roff_tok t;
1.212 schwarze 1825: int e;
1.110 schwarze 1826: int pos; /* parse point */
1.115 schwarze 1827: int spos; /* saved parse point for messages */
1.110 schwarze 1828: int ppos; /* original offset in buf->buf */
1829: int ctl; /* macro line (boolean) */
1830:
1831: ppos = pos = *offs;
1.1 schwarze 1832:
1.102 schwarze 1833: /* Handle in-line equation delimiters. */
1834:
1.108 schwarze 1835: if (r->tbl == NULL &&
1836: r->last_eqn != NULL && r->last_eqn->delim &&
1.102 schwarze 1837: (r->eqn == NULL || r->eqn_inline)) {
1.110 schwarze 1838: e = roff_eqndelim(r, buf, pos);
1.102 schwarze 1839: if (e == ROFF_REPARSE)
1.149 schwarze 1840: return e;
1.102 schwarze 1841: assert(e == ROFF_CONT);
1842: }
1843:
1844: /* Expand some escape sequences. */
1.8 schwarze 1845:
1.227 schwarze 1846: e = roff_expand(r, buf, ln, pos, r->escape);
1.212 schwarze 1847: if ((e & ROFF_MASK) == ROFF_IGN)
1.149 schwarze 1848: return e;
1.110 schwarze 1849: assert(e == ROFF_CONT);
1.8 schwarze 1850:
1.110 schwarze 1851: ctl = roff_getcontrol(r, buf->buf, &pos);
1.35 schwarze 1852:
1.8 schwarze 1853: /*
1.2 schwarze 1854: * First, if a scope is open and we're not a macro, pass the
1.124 schwarze 1855: * text through the macro's filter.
1856: * Equations process all content themselves.
1857: * Tables process almost all content themselves, but we want
1858: * to warn about macros before passing it there.
1.2 schwarze 1859: */
1860:
1.124 schwarze 1861: if (r->last != NULL && ! ctl) {
1.2 schwarze 1862: t = r->last->tok;
1.110 schwarze 1863: e = (*roffs[t].text)(r, t, buf, ln, pos, pos, offs);
1.212 schwarze 1864: if ((e & ROFF_MASK) == ROFF_IGN)
1.149 schwarze 1865: return e;
1.212 schwarze 1866: e &= ~ROFF_MASK;
1867: } else
1868: e = ROFF_IGN;
1.191 schwarze 1869: if (r->eqn != NULL && strncmp(buf->buf + ppos, ".EN", 3)) {
1870: eqn_read(r->eqn, buf->buf + ppos);
1.212 schwarze 1871: return e;
1.191 schwarze 1872: }
1.193 schwarze 1873: if (r->tbl != NULL && (ctl == 0 || buf->buf[pos] == '\0')) {
1874: tbl_read(r->tbl, ln, buf->buf, ppos);
1.218 schwarze 1875: roff_addtbl(r->man, ln, r->tbl);
1.212 schwarze 1876: return e;
1.193 schwarze 1877: }
1.239 schwarze 1878: if ( ! ctl) {
1879: r->options &= ~MPARSE_COMMENT;
1.212 schwarze 1880: return roff_parsetext(r, buf, pos, offs) | e;
1.239 schwarze 1881: }
1.100 schwarze 1882:
1883: /* Skip empty request lines. */
1884:
1.110 schwarze 1885: if (buf->buf[pos] == '"') {
1.222 schwarze 1886: mandoc_msg(MANDOCERR_COMMENT_BAD, ln, pos, NULL);
1.149 schwarze 1887: return ROFF_IGN;
1.110 schwarze 1888: } else if (buf->buf[pos] == '\0')
1.149 schwarze 1889: return ROFF_IGN;
1.2 schwarze 1890:
1891: /*
1892: * If a scope is open, go to the child handler for that macro,
1893: * as it may want to preprocess before doing anything with it.
1.32 schwarze 1894: * Don't do so if an equation is open.
1.2 schwarze 1895: */
1896:
1897: if (r->last) {
1.1 schwarze 1898: t = r->last->tok;
1.149 schwarze 1899: return (*roffs[t].sub)(r, t, buf, ln, ppos, pos, offs);
1.2 schwarze 1900: }
1901:
1.115 schwarze 1902: /* No scope is open. This is a new request or macro. */
1903:
1.239 schwarze 1904: r->options &= ~MPARSE_COMMENT;
1.115 schwarze 1905: spos = pos;
1906: t = roff_parse(r, buf->buf, &pos, ln, ppos);
1907:
1908: /* Tables ignore most macros. */
1909:
1.180 schwarze 1910: if (r->tbl != NULL && (t == TOKEN_NONE || t == ROFF_TS ||
1.181 schwarze 1911: t == ROFF_br || t == ROFF_ce || t == ROFF_rj || t == ROFF_sp)) {
1.222 schwarze 1912: mandoc_msg(MANDOCERR_TBLMACRO,
1913: ln, pos, "%s", buf->buf + spos);
1.180 schwarze 1914: if (t != TOKEN_NONE)
1.149 schwarze 1915: return ROFF_IGN;
1.129 schwarze 1916: while (buf->buf[pos] != '\0' && buf->buf[pos] != ' ')
1917: pos++;
1.163 schwarze 1918: while (buf->buf[pos] == ' ')
1.129 schwarze 1919: pos++;
1.193 schwarze 1920: tbl_read(r->tbl, ln, buf->buf, pos);
1.218 schwarze 1921: roff_addtbl(r->man, ln, r->tbl);
1.193 schwarze 1922: return ROFF_IGN;
1.115 schwarze 1923: }
1924:
1.177 schwarze 1925: /* For now, let high level macros abort .ce mode. */
1926:
1927: if (ctl && roffce_node != NULL &&
1.196 schwarze 1928: (t == TOKEN_NONE || t == ROFF_Dd || t == ROFF_EQ ||
1929: t == ROFF_TH || t == ROFF_TS)) {
1.177 schwarze 1930: r->man->last = roffce_node;
1931: r->man->next = ROFF_NEXT_SIBLING;
1932: roffce_lines = 0;
1933: roffce_node = NULL;
1934: }
1935:
1.2 schwarze 1936: /*
1.115 schwarze 1937: * This is neither a roff request nor a user-defined macro.
1938: * Let the standard macro set parsers handle it.
1.2 schwarze 1939: */
1940:
1.166 schwarze 1941: if (t == TOKEN_NONE)
1.149 schwarze 1942: return ROFF_CONT;
1.115 schwarze 1943:
1944: /* Execute a roff request or a user defined macro. */
1.1 schwarze 1945:
1.168 schwarze 1946: return (*roffs[t].proc)(r, t, buf, ln, spos, pos, offs);
1.2 schwarze 1947: }
1948:
1.211 schwarze 1949: /*
1950: * Internal interface function to tell the roff parser that execution
1951: * of the current macro ended. This is required because macro
1952: * definitions usually do not end with a .return request.
1953: */
1954: void
1955: roff_userret(struct roff *r)
1956: {
1957: struct mctx *ctx;
1958: int i;
1959:
1960: assert(r->mstackpos >= 0);
1961: ctx = r->mstack + r->mstackpos;
1962: for (i = 0; i < ctx->argc; i++)
1963: free(ctx->argv[i]);
1964: ctx->argc = 0;
1965: r->mstackpos--;
1966: }
1967:
1.27 schwarze 1968: void
1.2 schwarze 1969: roff_endparse(struct roff *r)
1970: {
1.193 schwarze 1971: if (r->last != NULL)
1.222 schwarze 1972: mandoc_msg(MANDOCERR_BLK_NOEND, r->last->line,
1973: r->last->col, "%s", roff_name[r->last->tok]);
1.27 schwarze 1974:
1.193 schwarze 1975: if (r->eqn != NULL) {
1.222 schwarze 1976: mandoc_msg(MANDOCERR_BLK_NOEND,
1.191 schwarze 1977: r->eqn->node->line, r->eqn->node->pos, "EQ");
1978: eqn_parse(r->eqn);
1979: r->eqn = NULL;
1.32 schwarze 1980: }
1981:
1.193 schwarze 1982: if (r->tbl != NULL) {
1.218 schwarze 1983: tbl_end(r->tbl, 1);
1.193 schwarze 1984: r->tbl = NULL;
1.27 schwarze 1985: }
1.1 schwarze 1986: }
1987:
1988: /*
1989: * Parse a roff node's type from the input buffer. This must be in the
1990: * form of ".foo xxx" in the usual way.
1991: */
1.166 schwarze 1992: static enum roff_tok
1.87 schwarze 1993: roff_parse(struct roff *r, char *buf, int *pos, int ln, int ppos)
1.1 schwarze 1994: {
1.87 schwarze 1995: char *cp;
1.16 schwarze 1996: const char *mac;
1997: size_t maclen;
1.187 schwarze 1998: int deftype;
1.166 schwarze 1999: enum roff_tok t;
1.1 schwarze 2000:
1.87 schwarze 2001: cp = buf + *pos;
2002:
2003: if ('\0' == *cp || '"' == *cp || '\t' == *cp || ' ' == *cp)
1.166 schwarze 2004: return TOKEN_NONE;
1.1 schwarze 2005:
1.87 schwarze 2006: mac = cp;
2007: maclen = roff_getname(r, &cp, ln, ppos);
1.1 schwarze 2008:
1.187 schwarze 2009: deftype = ROFFDEF_USER | ROFFDEF_REN;
2010: r->current_string = roff_getstrn(r, mac, maclen, &deftype);
2011: switch (deftype) {
2012: case ROFFDEF_USER:
2013: t = ROFF_USERDEF;
2014: break;
2015: case ROFFDEF_REN:
2016: t = ROFF_RENAMED;
2017: break;
2018: default:
2019: t = roffhash_find(r->reqtab, mac, maclen);
2020: break;
2021: }
1.166 schwarze 2022: if (t != TOKEN_NONE)
1.87 schwarze 2023: *pos = cp - buf;
1.197 schwarze 2024: else if (deftype == ROFFDEF_UNDEF) {
2025: /* Using an undefined macro defines it to be empty. */
2026: roff_setstrn(&r->strtab, mac, maclen, "", 0, 0);
2027: roff_setstrn(&r->rentab, mac, maclen, NULL, 0, 0);
2028: }
1.149 schwarze 2029: return t;
1.1 schwarze 2030: }
2031:
1.138 schwarze 2032: /* --- handling of request blocks ----------------------------------------- */
2033:
1.212 schwarze 2034: static int
1.2 schwarze 2035: roff_cblock(ROFF_ARGS)
1.1 schwarze 2036: {
2037:
1.2 schwarze 2038: /*
2039: * A block-close `..' should only be invoked as a child of an
2040: * ignore macro, otherwise raise a warning and just ignore it.
2041: */
2042:
1.110 schwarze 2043: if (r->last == NULL) {
1.222 schwarze 2044: mandoc_msg(MANDOCERR_BLK_NOTOPEN, ln, ppos, "..");
1.149 schwarze 2045: return ROFF_IGN;
1.2 schwarze 2046: }
1.1 schwarze 2047:
1.2 schwarze 2048: switch (r->last->tok) {
1.80 schwarze 2049: case ROFF_am:
1.93 schwarze 2050: /* ROFF_am1 is remapped to ROFF_am in roff_block(). */
1.80 schwarze 2051: case ROFF_ami:
2052: case ROFF_de:
1.23 schwarze 2053: /* ROFF_de1 is remapped to ROFF_de in roff_block(). */
1.80 schwarze 2054: case ROFF_dei:
2055: case ROFF_ig:
1.2 schwarze 2056: break;
2057: default:
1.222 schwarze 2058: mandoc_msg(MANDOCERR_BLK_NOTOPEN, ln, ppos, "..");
1.149 schwarze 2059: return ROFF_IGN;
1.2 schwarze 2060: }
2061:
1.110 schwarze 2062: if (buf->buf[pos] != '\0')
1.222 schwarze 2063: mandoc_msg(MANDOCERR_ARG_SKIP, ln, pos,
1.110 schwarze 2064: ".. %s", buf->buf + pos);
1.2 schwarze 2065:
2066: roffnode_pop(r);
2067: roffnode_cleanscope(r);
1.149 schwarze 2068: return ROFF_IGN;
1.2 schwarze 2069:
2070: }
1.1 schwarze 2071:
1.236 schwarze 2072: /*
2073: * Pop all nodes ending at the end of the current input line.
2074: * Return the number of loops ended.
2075: */
1.212 schwarze 2076: static int
1.2 schwarze 2077: roffnode_cleanscope(struct roff *r)
2078: {
1.212 schwarze 2079: int inloop;
1.1 schwarze 2080:
1.212 schwarze 2081: inloop = 0;
2082: while (r->last != NULL) {
1.46 schwarze 2083: if (--r->last->endspan != 0)
1.2 schwarze 2084: break;
1.212 schwarze 2085: inloop += roffnode_pop(r);
1.2 schwarze 2086: }
1.212 schwarze 2087: return inloop;
1.2 schwarze 2088: }
1.1 schwarze 2089:
1.236 schwarze 2090: /*
2091: * Handle the closing \} of a conditional block.
2092: * Apart from generating warnings, this only pops nodes.
2093: * Return the number of loops ended.
2094: */
1.212 schwarze 2095: static int
1.68 schwarze 2096: roff_ccond(struct roff *r, int ln, int ppos)
1.2 schwarze 2097: {
2098: if (NULL == r->last) {
1.222 schwarze 2099: mandoc_msg(MANDOCERR_BLK_NOTOPEN, ln, ppos, "\\}");
1.212 schwarze 2100: return 0;
1.2 schwarze 2101: }
1.1 schwarze 2102:
1.2 schwarze 2103: switch (r->last->tok) {
1.80 schwarze 2104: case ROFF_el:
2105: case ROFF_ie:
2106: case ROFF_if:
1.212 schwarze 2107: case ROFF_while:
1.2 schwarze 2108: break;
2109: default:
1.222 schwarze 2110: mandoc_msg(MANDOCERR_BLK_NOTOPEN, ln, ppos, "\\}");
1.212 schwarze 2111: return 0;
1.2 schwarze 2112: }
1.1 schwarze 2113:
1.2 schwarze 2114: if (r->last->endspan > -1) {
1.222 schwarze 2115: mandoc_msg(MANDOCERR_BLK_NOTOPEN, ln, ppos, "\\}");
1.212 schwarze 2116: return 0;
1.2 schwarze 2117: }
2118:
1.212 schwarze 2119: return roffnode_pop(r) + roffnode_cleanscope(r);
1.1 schwarze 2120: }
2121:
1.212 schwarze 2122: static int
1.2 schwarze 2123: roff_block(ROFF_ARGS)
1.1 schwarze 2124: {
1.187 schwarze 2125: const char *name, *value;
2126: char *call, *cp, *iname, *rname;
2127: size_t csz, namesz, rsz;
2128: int deftype;
1.16 schwarze 2129:
1.93 schwarze 2130: /* Ignore groff compatibility mode for now. */
1.2 schwarze 2131:
1.110 schwarze 2132: if (tok == ROFF_de1)
1.93 schwarze 2133: tok = ROFF_de;
1.123 schwarze 2134: else if (tok == ROFF_dei1)
2135: tok = ROFF_dei;
1.110 schwarze 2136: else if (tok == ROFF_am1)
1.93 schwarze 2137: tok = ROFF_am;
1.123 schwarze 2138: else if (tok == ROFF_ami1)
2139: tok = ROFF_ami;
1.93 schwarze 2140:
2141: /* Parse the macro name argument. */
2142:
1.110 schwarze 2143: cp = buf->buf + pos;
2144: if (tok == ROFF_ig) {
1.93 schwarze 2145: iname = NULL;
2146: namesz = 0;
2147: } else {
2148: iname = cp;
2149: namesz = roff_getname(r, &cp, ln, ppos);
2150: iname[namesz] = '\0';
2151: }
1.22 schwarze 2152:
1.93 schwarze 2153: /* Resolve the macro name argument if it is indirect. */
1.22 schwarze 2154:
1.110 schwarze 2155: if (namesz && (tok == ROFF_dei || tok == ROFF_ami)) {
1.187 schwarze 2156: deftype = ROFFDEF_USER;
2157: name = roff_getstrn(r, iname, namesz, &deftype);
2158: if (name == NULL) {
1.222 schwarze 2159: mandoc_msg(MANDOCERR_STR_UNDEF,
2160: ln, (int)(iname - buf->buf),
1.93 schwarze 2161: "%.*s", (int)namesz, iname);
2162: namesz = 0;
2163: } else
2164: namesz = strlen(name);
2165: } else
2166: name = iname;
1.22 schwarze 2167:
1.110 schwarze 2168: if (namesz == 0 && tok != ROFF_ig) {
1.222 schwarze 2169: mandoc_msg(MANDOCERR_REQ_EMPTY,
2170: ln, ppos, "%s", roff_name[tok]);
1.149 schwarze 2171: return ROFF_IGN;
1.93 schwarze 2172: }
1.2 schwarze 2173:
1.16 schwarze 2174: roffnode_push(r, tok, name, ln, ppos);
2175:
2176: /*
2177: * At the beginning of a `de' macro, clear the existing string
2178: * with the same name, if there is one. New content will be
1.66 schwarze 2179: * appended from roff_block_text() in multiline mode.
1.16 schwarze 2180: */
1.22 schwarze 2181:
1.183 schwarze 2182: if (tok == ROFF_de || tok == ROFF_dei) {
1.86 schwarze 2183: roff_setstrn(&r->strtab, name, namesz, "", 0, 0);
1.183 schwarze 2184: roff_setstrn(&r->rentab, name, namesz, NULL, 0, 0);
1.187 schwarze 2185: } else if (tok == ROFF_am || tok == ROFF_ami) {
2186: deftype = ROFFDEF_ANY;
2187: value = roff_getstrn(r, iname, namesz, &deftype);
2188: switch (deftype) { /* Before appending, ... */
2189: case ROFFDEF_PRE: /* copy predefined to user-defined. */
2190: roff_setstrn(&r->strtab, name, namesz,
2191: value, strlen(value), 0);
2192: break;
2193: case ROFFDEF_REN: /* call original standard macro. */
2194: csz = mandoc_asprintf(&call, ".%.*s \\$* \\\"\n",
2195: (int)strlen(value), value);
2196: roff_setstrn(&r->strtab, name, namesz, call, csz, 0);
2197: roff_setstrn(&r->rentab, name, namesz, NULL, 0, 0);
2198: free(call);
2199: break;
2200: case ROFFDEF_STD: /* rename and call standard macro. */
2201: rsz = mandoc_asprintf(&rname, "__%s_renamed", name);
2202: roff_setstrn(&r->rentab, rname, rsz, name, namesz, 0);
2203: csz = mandoc_asprintf(&call, ".%.*s \\$* \\\"\n",
2204: (int)rsz, rname);
2205: roff_setstrn(&r->strtab, name, namesz, call, csz, 0);
2206: free(call);
2207: free(rname);
2208: break;
2209: default:
2210: break;
2211: }
1.183 schwarze 2212: }
1.2 schwarze 2213:
1.110 schwarze 2214: if (*cp == '\0')
1.149 schwarze 2215: return ROFF_IGN;
1.1 schwarze 2216:
1.93 schwarze 2217: /* Get the custom end marker. */
1.22 schwarze 2218:
1.93 schwarze 2219: iname = cp;
1.86 schwarze 2220: namesz = roff_getname(r, &cp, ln, ppos);
1.93 schwarze 2221:
2222: /* Resolve the end marker if it is indirect. */
2223:
1.110 schwarze 2224: if (namesz && (tok == ROFF_dei || tok == ROFF_ami)) {
1.187 schwarze 2225: deftype = ROFFDEF_USER;
2226: name = roff_getstrn(r, iname, namesz, &deftype);
2227: if (name == NULL) {
1.222 schwarze 2228: mandoc_msg(MANDOCERR_STR_UNDEF,
2229: ln, (int)(iname - buf->buf),
1.93 schwarze 2230: "%.*s", (int)namesz, iname);
2231: namesz = 0;
2232: } else
2233: namesz = strlen(name);
2234: } else
2235: name = iname;
2236:
1.86 schwarze 2237: if (namesz)
2238: r->last->end = mandoc_strndup(name, namesz);
1.2 schwarze 2239:
1.110 schwarze 2240: if (*cp != '\0')
1.222 schwarze 2241: mandoc_msg(MANDOCERR_ARG_EXCESS,
1.167 schwarze 2242: ln, pos, ".%s ... %s", roff_name[tok], cp);
1.1 schwarze 2243:
1.149 schwarze 2244: return ROFF_IGN;
1.1 schwarze 2245: }
2246:
1.212 schwarze 2247: static int
1.2 schwarze 2248: roff_block_sub(ROFF_ARGS)
1.1 schwarze 2249: {
1.166 schwarze 2250: enum roff_tok t;
1.2 schwarze 2251: int i, j;
2252:
2253: /*
2254: * First check whether a custom macro exists at this level. If
2255: * it does, then check against it. This is some of groff's
2256: * stranger behaviours. If we encountered a custom end-scope
2257: * tag and that tag also happens to be a "real" macro, then we
2258: * need to try interpreting it again as a real macro. If it's
2259: * not, then return ignore. Else continue.
2260: */
2261:
2262: if (r->last->end) {
1.35 schwarze 2263: for (i = pos, j = 0; r->last->end[j]; j++, i++)
1.110 schwarze 2264: if (buf->buf[i] != r->last->end[j])
1.2 schwarze 2265: break;
1.1 schwarze 2266:
1.110 schwarze 2267: if (r->last->end[j] == '\0' &&
2268: (buf->buf[i] == '\0' ||
2269: buf->buf[i] == ' ' ||
2270: buf->buf[i] == '\t')) {
1.2 schwarze 2271: roffnode_pop(r);
2272: roffnode_cleanscope(r);
1.1 schwarze 2273:
1.110 schwarze 2274: while (buf->buf[i] == ' ' || buf->buf[i] == '\t')
1.35 schwarze 2275: i++;
2276:
2277: pos = i;
1.110 schwarze 2278: if (roff_parse(r, buf->buf, &pos, ln, ppos) !=
1.166 schwarze 2279: TOKEN_NONE)
1.149 schwarze 2280: return ROFF_RERUN;
2281: return ROFF_IGN;
1.2 schwarze 2282: }
1.1 schwarze 2283: }
2284:
1.2 schwarze 2285: /*
2286: * If we have no custom end-query or lookup failed, then try
2287: * pulling it out of the hashtable.
2288: */
1.1 schwarze 2289:
1.110 schwarze 2290: t = roff_parse(r, buf->buf, &pos, ln, ppos);
1.1 schwarze 2291:
1.110 schwarze 2292: if (t != ROFF_cblock) {
2293: if (tok != ROFF_ig)
2294: roff_setstr(r, r->last->name, buf->buf + ppos, 2);
1.149 schwarze 2295: return ROFF_IGN;
1.16 schwarze 2296: }
1.1 schwarze 2297:
1.149 schwarze 2298: return (*roffs[t].proc)(r, t, buf, ln, ppos, pos, offs);
1.2 schwarze 2299: }
2300:
1.212 schwarze 2301: static int
1.2 schwarze 2302: roff_block_text(ROFF_ARGS)
2303: {
2304:
1.110 schwarze 2305: if (tok != ROFF_ig)
2306: roff_setstr(r, r->last->name, buf->buf + pos, 2);
1.16 schwarze 2307:
1.149 schwarze 2308: return ROFF_IGN;
1.2 schwarze 2309: }
2310:
1.212 schwarze 2311: static int
1.2 schwarze 2312: roff_cond_sub(ROFF_ARGS)
2313: {
1.236 schwarze 2314: struct roffnode *bl;
1.212 schwarze 2315: char *ep;
2316: int endloop, irc, rr;
1.166 schwarze 2317: enum roff_tok t;
1.2 schwarze 2318:
1.212 schwarze 2319: irc = ROFF_IGN;
1.2 schwarze 2320: rr = r->last->rule;
1.212 schwarze 2321: endloop = tok != ROFF_while ? ROFF_IGN :
2322: rr ? ROFF_LOOPCONT : ROFF_LOOPEXIT;
2323: if (roffnode_cleanscope(r))
2324: irc |= endloop;
1.39 schwarze 2325:
1.69 schwarze 2326: /*
2327: * If `\}' occurs on a macro line without a preceding macro,
2328: * drop the line completely.
2329: */
2330:
1.110 schwarze 2331: ep = buf->buf + pos;
2332: if (ep[0] == '\\' && ep[1] == '}')
1.71 schwarze 2333: rr = 0;
1.69 schwarze 2334:
1.215 schwarze 2335: /*
2336: * The closing delimiter `\}' rewinds the conditional scope
2337: * but is otherwise ignored when interpreting the line.
2338: */
1.39 schwarze 2339:
1.110 schwarze 2340: while ((ep = strchr(ep, '\\')) != NULL) {
1.190 schwarze 2341: switch (ep[1]) {
2342: case '}':
2343: memmove(ep, ep + 2, strlen(ep + 2) + 1);
1.212 schwarze 2344: if (roff_ccond(r, ln, ep - buf->buf))
2345: irc |= endloop;
1.190 schwarze 2346: break;
2347: case '\0':
2348: ++ep;
2349: break;
2350: default:
2351: ep += 2;
2352: break;
1.70 schwarze 2353: }
1.50 schwarze 2354: }
1.190 schwarze 2355:
1.240 schwarze 2356: t = roff_parse(r, buf->buf, &pos, ln, ppos);
2357:
2358: /* For now, let high level macros abort .ce mode. */
2359:
2360: if (roffce_node != NULL &&
2361: (t == TOKEN_NONE || t == ROFF_Dd || t == ROFF_EQ ||
2362: t == ROFF_TH || t == ROFF_TS)) {
2363: r->man->last = roffce_node;
2364: r->man->next = ROFF_NEXT_SIBLING;
2365: roffce_lines = 0;
2366: roffce_node = NULL;
2367: }
2368:
1.190 schwarze 2369: /*
2370: * Fully handle known macros when they are structurally
2371: * required or when the conditional evaluated to true.
2372: */
2373:
1.236 schwarze 2374: if (t == ROFF_break) {
2375: if (irc & ROFF_LOOPMASK)
2376: irc = ROFF_IGN | ROFF_LOOPEXIT;
2377: else if (rr) {
2378: for (bl = r->last; bl != NULL; bl = bl->parent) {
2379: bl->rule = 0;
2380: if (bl->tok == ROFF_while)
2381: break;
2382: }
2383: }
2384: } else if (t != TOKEN_NONE &&
2385: (rr || roffs[t].flags & ROFFMAC_STRUCT))
2386: irc |= (*roffs[t].proc)(r, t, buf, ln, ppos, pos, offs);
2387: else
2388: irc |= rr ? ROFF_CONT : ROFF_IGN;
1.212 schwarze 2389: return irc;
1.2 schwarze 2390: }
2391:
1.212 schwarze 2392: static int
1.2 schwarze 2393: roff_cond_text(ROFF_ARGS)
2394: {
1.37 schwarze 2395: char *ep;
1.212 schwarze 2396: int endloop, irc, rr;
1.2 schwarze 2397:
1.212 schwarze 2398: irc = ROFF_IGN;
1.2 schwarze 2399: rr = r->last->rule;
1.212 schwarze 2400: endloop = tok != ROFF_while ? ROFF_IGN :
2401: rr ? ROFF_LOOPCONT : ROFF_LOOPEXIT;
2402: if (roffnode_cleanscope(r))
2403: irc |= endloop;
1.1 schwarze 2404:
1.215 schwarze 2405: /*
2406: * If `\}' occurs on a text line with neither preceding
2407: * nor following characters, drop the line completely.
2408: */
2409:
1.110 schwarze 2410: ep = buf->buf + pos;
1.215 schwarze 2411: if (strcmp(ep, "\\}") == 0)
2412: rr = 0;
2413:
2414: /*
2415: * The closing delimiter `\}' rewinds the conditional scope
2416: * but is otherwise ignored when interpreting the line.
2417: */
2418:
1.110 schwarze 2419: while ((ep = strchr(ep, '\\')) != NULL) {
1.215 schwarze 2420: switch (ep[1]) {
2421: case '}':
2422: memmove(ep, ep + 2, strlen(ep + 2) + 1);
2423: if (roff_ccond(r, ln, ep - buf->buf))
1.212 schwarze 2424: irc |= endloop;
1.215 schwarze 2425: break;
2426: case '\0':
2427: ++ep;
2428: break;
2429: default:
2430: ep += 2;
2431: break;
1.70 schwarze 2432: }
1.2 schwarze 2433: }
1.212 schwarze 2434: if (rr)
2435: irc |= ROFF_CONT;
2436: return irc;
1.2 schwarze 2437: }
2438:
1.138 schwarze 2439: /* --- handling of numeric and conditional expressions -------------------- */
2440:
1.77 schwarze 2441: /*
2442: * Parse a single signed integer number. Stop at the first non-digit.
2443: * If there is at least one digit, return success and advance the
2444: * parse point, else return failure and let the parse point unchanged.
2445: * Ignore overflows, treat them just like the C language.
2446: */
1.56 schwarze 2447: static int
1.133 schwarze 2448: roff_getnum(const char *v, int *pos, int *res, int flags)
1.56 schwarze 2449: {
1.133 schwarze 2450: int myres, scaled, n, p;
1.79 schwarze 2451:
2452: if (NULL == res)
2453: res = &myres;
1.56 schwarze 2454:
2455: p = *pos;
2456: n = v[p] == '-';
1.133 schwarze 2457: if (n || v[p] == '+')
1.56 schwarze 2458: p++;
2459:
1.133 schwarze 2460: if (flags & ROFFNUM_WHITE)
2461: while (isspace((unsigned char)v[p]))
2462: p++;
2463:
1.56 schwarze 2464: for (*res = 0; isdigit((unsigned char)v[p]); p++)
1.77 schwarze 2465: *res = 10 * *res + v[p] - '0';
1.56 schwarze 2466: if (p == *pos + n)
2467: return 0;
2468:
2469: if (n)
2470: *res = -*res;
2471:
1.126 schwarze 2472: /* Each number may be followed by one optional scaling unit. */
2473:
2474: switch (v[p]) {
2475: case 'f':
1.133 schwarze 2476: scaled = *res * 65536;
1.126 schwarze 2477: break;
2478: case 'i':
1.133 schwarze 2479: scaled = *res * 240;
1.126 schwarze 2480: break;
2481: case 'c':
1.133 schwarze 2482: scaled = *res * 240 / 2.54;
1.126 schwarze 2483: break;
2484: case 'v':
2485: case 'P':
1.133 schwarze 2486: scaled = *res * 40;
1.126 schwarze 2487: break;
2488: case 'm':
2489: case 'n':
1.133 schwarze 2490: scaled = *res * 24;
1.126 schwarze 2491: break;
2492: case 'p':
1.133 schwarze 2493: scaled = *res * 10 / 3;
1.126 schwarze 2494: break;
2495: case 'u':
1.133 schwarze 2496: scaled = *res;
1.126 schwarze 2497: break;
2498: case 'M':
1.133 schwarze 2499: scaled = *res * 6 / 25;
1.126 schwarze 2500: break;
2501: default:
1.133 schwarze 2502: scaled = *res;
1.126 schwarze 2503: p--;
2504: break;
2505: }
1.133 schwarze 2506: if (flags & ROFFNUM_SCALE)
2507: *res = scaled;
1.126 schwarze 2508:
2509: *pos = p + 1;
1.149 schwarze 2510: return 1;
1.56 schwarze 2511: }
2512:
1.71 schwarze 2513: /*
2514: * Evaluate a string comparison condition.
2515: * The first character is the delimiter.
2516: * Succeed if the string up to its second occurrence
2517: * matches the string up to its third occurence.
2518: * Advance the cursor after the third occurrence
2519: * or lacking that, to the end of the line.
2520: */
2521: static int
2522: roff_evalstrcond(const char *v, int *pos)
2523: {
2524: const char *s1, *s2, *s3;
2525: int match;
2526:
2527: match = 0;
2528: s1 = v + *pos; /* initial delimiter */
2529: s2 = s1 + 1; /* for scanning the first string */
2530: s3 = strchr(s2, *s1); /* for scanning the second string */
2531:
2532: if (NULL == s3) /* found no middle delimiter */
2533: goto out;
2534:
2535: while ('\0' != *++s3) {
2536: if (*s2 != *s3) { /* mismatch */
2537: s3 = strchr(s3, *s1);
2538: break;
2539: }
2540: if (*s3 == *s1) { /* found the final delimiter */
2541: match = 1;
2542: break;
2543: }
2544: s2++;
2545: }
2546:
2547: out:
2548: if (NULL == s3)
2549: s3 = strchr(s2, '\0');
1.114 schwarze 2550: else if (*s3 != '\0')
1.71 schwarze 2551: s3++;
2552: *pos = s3 - v;
1.149 schwarze 2553: return match;
1.71 schwarze 2554: }
2555:
1.77 schwarze 2556: /*
2557: * Evaluate an optionally negated single character, numerical,
2558: * or string condition.
2559: */
1.71 schwarze 2560: static int
1.143 schwarze 2561: roff_evalcond(struct roff *r, int ln, char *v, int *pos)
1.5 schwarze 2562: {
1.208 schwarze 2563: const char *start, *end;
2564: char *cp, *name;
2565: size_t sz;
2566: int deftype, len, number, savepos, istrue, wanttrue;
1.5 schwarze 2567:
1.71 schwarze 2568: if ('!' == v[*pos]) {
2569: wanttrue = 0;
2570: (*pos)++;
2571: } else
2572: wanttrue = 1;
2573:
1.5 schwarze 2574: switch (v[*pos]) {
1.112 schwarze 2575: case '\0':
1.149 schwarze 2576: return 0;
1.80 schwarze 2577: case 'n':
2578: case 'o':
1.5 schwarze 2579: (*pos)++;
1.149 schwarze 2580: return wanttrue;
1.80 schwarze 2581: case 'e':
2582: case 't':
1.111 schwarze 2583: case 'v':
1.5 schwarze 2584: (*pos)++;
1.149 schwarze 2585: return !wanttrue;
1.208 schwarze 2586: case 'c':
2587: do {
2588: (*pos)++;
2589: } while (v[*pos] == ' ');
2590:
2591: /*
2592: * Quirk for groff compatibility:
2593: * The horizontal tab is neither available nor unavailable.
2594: */
2595:
2596: if (v[*pos] == '\t') {
2597: (*pos)++;
2598: return 0;
2599: }
2600:
2601: /* Printable ASCII characters are available. */
2602:
2603: if (v[*pos] != '\\') {
2604: (*pos)++;
2605: return wanttrue;
2606: }
2607:
2608: end = v + ++*pos;
2609: switch (mandoc_escape(&end, &start, &len)) {
2610: case ESCAPE_SPECIAL:
2611: istrue = mchars_spec2cp(start, len) != -1;
2612: break;
2613: case ESCAPE_UNICODE:
2614: istrue = 1;
2615: break;
2616: case ESCAPE_NUMBERED:
2617: istrue = mchars_num2char(start, len) != -1;
2618: break;
2619: default:
2620: istrue = !wanttrue;
2621: break;
2622: }
2623: *pos = end - v;
2624: return istrue == wanttrue;
1.182 schwarze 2625: case 'd':
1.143 schwarze 2626: case 'r':
1.182 schwarze 2627: cp = v + *pos + 1;
2628: while (*cp == ' ')
2629: cp++;
2630: name = cp;
2631: sz = roff_getname(r, &cp, ln, cp - v);
1.187 schwarze 2632: if (sz == 0)
2633: istrue = 0;
2634: else if (v[*pos] == 'r')
2635: istrue = roff_hasregn(r, name, sz);
2636: else {
2637: deftype = ROFFDEF_ANY;
2638: roff_getstrn(r, name, sz, &deftype);
2639: istrue = !!deftype;
2640: }
1.235 schwarze 2641: *pos = (name + sz) - v;
1.182 schwarze 2642: return istrue == wanttrue;
1.5 schwarze 2643: default:
2644: break;
2645: }
2646:
1.113 schwarze 2647: savepos = *pos;
1.133 schwarze 2648: if (roff_evalnum(r, ln, v, pos, &number, ROFFNUM_SCALE))
1.149 schwarze 2649: return (number > 0) == wanttrue;
1.113 schwarze 2650: else if (*pos == savepos)
1.149 schwarze 2651: return roff_evalstrcond(v, pos) == wanttrue;
1.77 schwarze 2652: else
1.149 schwarze 2653: return 0;
1.5 schwarze 2654: }
2655:
1.212 schwarze 2656: static int
1.21 schwarze 2657: roff_line_ignore(ROFF_ARGS)
1.6 schwarze 2658: {
1.30 schwarze 2659:
1.149 schwarze 2660: return ROFF_IGN;
1.21 schwarze 2661: }
2662:
1.212 schwarze 2663: static int
1.123 schwarze 2664: roff_insec(ROFF_ARGS)
2665: {
2666:
1.222 schwarze 2667: mandoc_msg(MANDOCERR_REQ_INSEC, ln, ppos, "%s", roff_name[tok]);
1.149 schwarze 2668: return ROFF_IGN;
1.123 schwarze 2669: }
2670:
1.212 schwarze 2671: static int
1.123 schwarze 2672: roff_unsupp(ROFF_ARGS)
2673: {
2674:
1.222 schwarze 2675: mandoc_msg(MANDOCERR_REQ_UNSUPP, ln, ppos, "%s", roff_name[tok]);
1.149 schwarze 2676: return ROFF_IGN;
1.123 schwarze 2677: }
2678:
1.212 schwarze 2679: static int
1.2 schwarze 2680: roff_cond(ROFF_ARGS)
2681: {
1.212 schwarze 2682: int irc;
1.46 schwarze 2683:
2684: roffnode_push(r, tok, NULL, ln, ppos);
1.2 schwarze 2685:
1.80 schwarze 2686: /*
1.35 schwarze 2687: * An `.el' has no conditional body: it will consume the value
2688: * of the current rstack entry set in prior `ie' calls or
1.80 schwarze 2689: * defaults to DENY.
1.35 schwarze 2690: *
2691: * If we're not an `el', however, then evaluate the conditional.
2692: */
1.1 schwarze 2693:
1.110 schwarze 2694: r->last->rule = tok == ROFF_el ?
1.80 schwarze 2695: (r->rstackpos < 0 ? 0 : r->rstack[r->rstackpos--]) :
1.110 schwarze 2696: roff_evalcond(r, ln, buf->buf, &pos);
1.2 schwarze 2697:
1.35 schwarze 2698: /*
2699: * An if-else will put the NEGATION of the current evaluated
2700: * conditional into the stack of rules.
2701: */
2702:
1.110 schwarze 2703: if (tok == ROFF_ie) {
1.96 schwarze 2704: if (r->rstackpos + 1 == r->rstacksz) {
2705: r->rstacksz += 16;
2706: r->rstack = mandoc_reallocarray(r->rstack,
2707: r->rstacksz, sizeof(int));
1.35 schwarze 2708: }
1.71 schwarze 2709: r->rstack[++r->rstackpos] = !r->last->rule;
1.2 schwarze 2710: }
1.5 schwarze 2711:
2712: /* If the parent has false as its rule, then so do we. */
2713:
1.71 schwarze 2714: if (r->last->parent && !r->last->parent->rule)
2715: r->last->rule = 0;
1.5 schwarze 2716:
2717: /*
1.46 schwarze 2718: * Determine scope.
2719: * If there is nothing on the line after the conditional,
2720: * not even whitespace, use next-line scope.
1.212 schwarze 2721: * Except that .while does not support next-line scope.
1.5 schwarze 2722: */
1.2 schwarze 2723:
1.212 schwarze 2724: if (buf->buf[pos] == '\0' && tok != ROFF_while) {
1.46 schwarze 2725: r->last->endspan = 2;
2726: goto out;
2727: }
2728:
1.110 schwarze 2729: while (buf->buf[pos] == ' ')
1.46 schwarze 2730: pos++;
2731:
2732: /* An opening brace requests multiline scope. */
1.2 schwarze 2733:
1.110 schwarze 2734: if (buf->buf[pos] == '\\' && buf->buf[pos + 1] == '{') {
1.2 schwarze 2735: r->last->endspan = -1;
2736: pos += 2;
1.144 schwarze 2737: while (buf->buf[pos] == ' ')
2738: pos++;
1.46 schwarze 2739: goto out;
1.80 schwarze 2740: }
1.2 schwarze 2741:
2742: /*
1.46 schwarze 2743: * Anything else following the conditional causes
2744: * single-line scope. Warn if the scope contains
2745: * nothing but trailing whitespace.
1.2 schwarze 2746: */
2747:
1.110 schwarze 2748: if (buf->buf[pos] == '\0')
1.222 schwarze 2749: mandoc_msg(MANDOCERR_COND_EMPTY,
2750: ln, ppos, "%s", roff_name[tok]);
1.2 schwarze 2751:
1.46 schwarze 2752: r->last->endspan = 1;
1.1 schwarze 2753:
1.46 schwarze 2754: out:
1.2 schwarze 2755: *offs = pos;
1.212 schwarze 2756: irc = ROFF_RERUN;
2757: if (tok == ROFF_while)
2758: irc |= ROFF_WHILE;
2759: return irc;
1.1 schwarze 2760: }
2761:
1.212 schwarze 2762: static int
1.7 schwarze 2763: roff_ds(ROFF_ARGS)
2764: {
1.85 schwarze 2765: char *string;
2766: const char *name;
2767: size_t namesz;
1.10 schwarze 2768:
1.123 schwarze 2769: /* Ignore groff compatibility mode for now. */
2770:
2771: if (tok == ROFF_ds1)
2772: tok = ROFF_ds;
2773: else if (tok == ROFF_as1)
2774: tok = ROFF_as;
2775:
1.10 schwarze 2776: /*
1.85 schwarze 2777: * The first word is the name of the string.
2778: * If it is empty or terminated by an escape sequence,
2779: * abort the `ds' request without defining anything.
1.10 schwarze 2780: */
1.7 schwarze 2781:
1.110 schwarze 2782: name = string = buf->buf + pos;
2783: if (*name == '\0')
1.149 schwarze 2784: return ROFF_IGN;
1.7 schwarze 2785:
1.85 schwarze 2786: namesz = roff_getname(r, &string, ln, pos);
1.235 schwarze 2787: switch (name[namesz]) {
2788: case '\\':
1.149 schwarze 2789: return ROFF_IGN;
1.235 schwarze 2790: case '\t':
2791: string = buf->buf + pos + namesz;
2792: break;
2793: default:
2794: break;
2795: }
1.85 schwarze 2796:
2797: /* Read past the initial double-quote, if any. */
1.110 schwarze 2798: if (*string == '"')
1.7 schwarze 2799: string++;
2800:
1.10 schwarze 2801: /* The rest is the value. */
1.85 schwarze 2802: roff_setstrn(&r->strtab, name, namesz, string, strlen(string),
2803: ROFF_as == tok);
1.183 schwarze 2804: roff_setstrn(&r->rentab, name, namesz, NULL, 0, 0);
1.149 schwarze 2805: return ROFF_IGN;
1.7 schwarze 2806: }
2807:
1.77 schwarze 2808: /*
2809: * Parse a single operator, one or two characters long.
2810: * If the operator is recognized, return success and advance the
2811: * parse point, else return failure and let the parse point unchanged.
2812: */
2813: static int
2814: roff_getop(const char *v, int *pos, char *res)
2815: {
2816:
2817: *res = v[*pos];
2818:
2819: switch (*res) {
1.80 schwarze 2820: case '+':
2821: case '-':
2822: case '*':
2823: case '/':
2824: case '%':
2825: case '&':
2826: case ':':
1.77 schwarze 2827: break;
2828: case '<':
2829: switch (v[*pos + 1]) {
1.80 schwarze 2830: case '=':
1.77 schwarze 2831: *res = 'l';
2832: (*pos)++;
2833: break;
1.80 schwarze 2834: case '>':
1.77 schwarze 2835: *res = '!';
2836: (*pos)++;
2837: break;
1.80 schwarze 2838: case '?':
1.77 schwarze 2839: *res = 'i';
2840: (*pos)++;
2841: break;
2842: default:
2843: break;
2844: }
2845: break;
2846: case '>':
2847: switch (v[*pos + 1]) {
1.80 schwarze 2848: case '=':
1.77 schwarze 2849: *res = 'g';
2850: (*pos)++;
2851: break;
1.80 schwarze 2852: case '?':
1.77 schwarze 2853: *res = 'a';
2854: (*pos)++;
2855: break;
2856: default:
2857: break;
2858: }
2859: break;
2860: case '=':
2861: if ('=' == v[*pos + 1])
2862: (*pos)++;
2863: break;
2864: default:
1.149 schwarze 2865: return 0;
1.77 schwarze 2866: }
2867: (*pos)++;
2868:
1.149 schwarze 2869: return *res;
1.77 schwarze 2870: }
2871:
2872: /*
2873: * Evaluate either a parenthesized numeric expression
2874: * or a single signed integer number.
2875: */
2876: static int
1.106 schwarze 2877: roff_evalpar(struct roff *r, int ln,
1.133 schwarze 2878: const char *v, int *pos, int *res, int flags)
1.77 schwarze 2879: {
2880:
2881: if ('(' != v[*pos])
1.149 schwarze 2882: return roff_getnum(v, pos, res, flags);
1.77 schwarze 2883:
2884: (*pos)++;
1.133 schwarze 2885: if ( ! roff_evalnum(r, ln, v, pos, res, flags | ROFFNUM_WHITE))
1.149 schwarze 2886: return 0;
1.77 schwarze 2887:
1.79 schwarze 2888: /*
2889: * Omission of the closing parenthesis
2890: * is an error in validation mode,
2891: * but ignored in evaluation mode.
2892: */
2893:
1.77 schwarze 2894: if (')' == v[*pos])
2895: (*pos)++;
1.79 schwarze 2896: else if (NULL == res)
1.149 schwarze 2897: return 0;
1.77 schwarze 2898:
1.149 schwarze 2899: return 1;
1.77 schwarze 2900: }
2901:
2902: /*
2903: * Evaluate a complete numeric expression.
2904: * Proceed left to right, there is no concept of precedence.
2905: */
2906: static int
1.106 schwarze 2907: roff_evalnum(struct roff *r, int ln, const char *v,
1.133 schwarze 2908: int *pos, int *res, int flags)
1.77 schwarze 2909: {
2910: int mypos, operand2;
2911: char operator;
2912:
2913: if (NULL == pos) {
2914: mypos = 0;
2915: pos = &mypos;
2916: }
2917:
1.133 schwarze 2918: if (flags & ROFFNUM_WHITE)
1.77 schwarze 2919: while (isspace((unsigned char)v[*pos]))
2920: (*pos)++;
2921:
1.133 schwarze 2922: if ( ! roff_evalpar(r, ln, v, pos, res, flags))
1.149 schwarze 2923: return 0;
1.77 schwarze 2924:
2925: while (1) {
1.133 schwarze 2926: if (flags & ROFFNUM_WHITE)
1.77 schwarze 2927: while (isspace((unsigned char)v[*pos]))
2928: (*pos)++;
2929:
2930: if ( ! roff_getop(v, pos, &operator))
2931: break;
2932:
1.133 schwarze 2933: if (flags & ROFFNUM_WHITE)
1.77 schwarze 2934: while (isspace((unsigned char)v[*pos]))
2935: (*pos)++;
2936:
1.133 schwarze 2937: if ( ! roff_evalpar(r, ln, v, pos, &operand2, flags))
1.149 schwarze 2938: return 0;
1.77 schwarze 2939:
1.133 schwarze 2940: if (flags & ROFFNUM_WHITE)
1.77 schwarze 2941: while (isspace((unsigned char)v[*pos]))
2942: (*pos)++;
1.79 schwarze 2943:
2944: if (NULL == res)
2945: continue;
1.77 schwarze 2946:
2947: switch (operator) {
1.80 schwarze 2948: case '+':
1.77 schwarze 2949: *res += operand2;
2950: break;
1.80 schwarze 2951: case '-':
1.77 schwarze 2952: *res -= operand2;
2953: break;
1.80 schwarze 2954: case '*':
1.77 schwarze 2955: *res *= operand2;
2956: break;
1.80 schwarze 2957: case '/':
1.116 schwarze 2958: if (operand2 == 0) {
1.106 schwarze 2959: mandoc_msg(MANDOCERR_DIVZERO,
1.222 schwarze 2960: ln, *pos, "%s", v);
1.106 schwarze 2961: *res = 0;
2962: break;
2963: }
1.77 schwarze 2964: *res /= operand2;
2965: break;
1.80 schwarze 2966: case '%':
1.116 schwarze 2967: if (operand2 == 0) {
2968: mandoc_msg(MANDOCERR_DIVZERO,
1.222 schwarze 2969: ln, *pos, "%s", v);
1.116 schwarze 2970: *res = 0;
2971: break;
2972: }
1.77 schwarze 2973: *res %= operand2;
2974: break;
1.80 schwarze 2975: case '<':
1.77 schwarze 2976: *res = *res < operand2;
2977: break;
1.80 schwarze 2978: case '>':
1.77 schwarze 2979: *res = *res > operand2;
2980: break;
1.80 schwarze 2981: case 'l':
1.77 schwarze 2982: *res = *res <= operand2;
2983: break;
1.80 schwarze 2984: case 'g':
1.77 schwarze 2985: *res = *res >= operand2;
2986: break;
1.80 schwarze 2987: case '=':
1.77 schwarze 2988: *res = *res == operand2;
2989: break;
1.80 schwarze 2990: case '!':
1.77 schwarze 2991: *res = *res != operand2;
2992: break;
1.80 schwarze 2993: case '&':
1.77 schwarze 2994: *res = *res && operand2;
2995: break;
1.80 schwarze 2996: case ':':
1.77 schwarze 2997: *res = *res || operand2;
2998: break;
1.80 schwarze 2999: case 'i':
1.77 schwarze 3000: if (operand2 < *res)
3001: *res = operand2;
3002: break;
1.80 schwarze 3003: case 'a':
1.77 schwarze 3004: if (operand2 > *res)
3005: *res = operand2;
3006: break;
3007: default:
3008: abort();
3009: }
3010: }
1.149 schwarze 3011: return 1;
1.77 schwarze 3012: }
3013:
1.138 schwarze 3014: /* --- register management ------------------------------------------------ */
3015:
1.52 schwarze 3016: void
1.60 schwarze 3017: roff_setreg(struct roff *r, const char *name, int val, char sign)
1.41 schwarze 3018: {
1.199 schwarze 3019: roff_setregn(r, name, strlen(name), val, sign, INT_MIN);
1.198 schwarze 3020: }
3021:
3022: static void
3023: roff_setregn(struct roff *r, const char *name, size_t len,
1.199 schwarze 3024: int val, char sign, int step)
1.198 schwarze 3025: {
1.52 schwarze 3026: struct roffreg *reg;
3027:
3028: /* Search for an existing register with the same name. */
3029: reg = r->regtab;
3030:
1.198 schwarze 3031: while (reg != NULL && (reg->key.sz != len ||
3032: strncmp(reg->key.p, name, len) != 0))
1.52 schwarze 3033: reg = reg->next;
1.41 schwarze 3034:
1.52 schwarze 3035: if (NULL == reg) {
3036: /* Create a new register. */
3037: reg = mandoc_malloc(sizeof(struct roffreg));
1.198 schwarze 3038: reg->key.p = mandoc_strndup(name, len);
3039: reg->key.sz = len;
1.60 schwarze 3040: reg->val = 0;
1.199 schwarze 3041: reg->step = 0;
1.52 schwarze 3042: reg->next = r->regtab;
3043: r->regtab = reg;
3044: }
3045:
1.60 schwarze 3046: if ('+' == sign)
3047: reg->val += val;
3048: else if ('-' == sign)
3049: reg->val -= val;
3050: else
3051: reg->val = val;
1.199 schwarze 3052: if (step != INT_MIN)
3053: reg->step = step;
1.41 schwarze 3054: }
3055:
1.65 schwarze 3056: /*
3057: * Handle some predefined read-only number registers.
3058: * For now, return -1 if the requested register is not predefined;
3059: * in case a predefined read-only register having the value -1
3060: * were to turn up, another special value would have to be chosen.
3061: */
3062: static int
1.145 schwarze 3063: roff_getregro(const struct roff *r, const char *name)
1.65 schwarze 3064: {
3065:
3066: switch (*name) {
1.145 schwarze 3067: case '$': /* Number of arguments of the last macro evaluated. */
1.211 schwarze 3068: return r->mstackpos < 0 ? 0 : r->mstack[r->mstackpos].argc;
1.80 schwarze 3069: case 'A': /* ASCII approximation mode is always off. */
1.149 schwarze 3070: return 0;
1.80 schwarze 3071: case 'g': /* Groff compatibility mode is always on. */
1.149 schwarze 3072: return 1;
1.80 schwarze 3073: case 'H': /* Fixed horizontal resolution. */
1.149 schwarze 3074: return 24;
1.80 schwarze 3075: case 'j': /* Always adjust left margin only. */
1.149 schwarze 3076: return 0;
1.80 schwarze 3077: case 'T': /* Some output device is always defined. */
1.149 schwarze 3078: return 1;
1.80 schwarze 3079: case 'V': /* Fixed vertical resolution. */
1.149 schwarze 3080: return 40;
1.65 schwarze 3081: default:
1.149 schwarze 3082: return -1;
1.65 schwarze 3083: }
3084: }
3085:
1.53 schwarze 3086: int
1.198 schwarze 3087: roff_getreg(struct roff *r, const char *name)
1.41 schwarze 3088: {
1.199 schwarze 3089: return roff_getregn(r, name, strlen(name), '\0');
1.53 schwarze 3090: }
3091:
3092: static int
1.199 schwarze 3093: roff_getregn(struct roff *r, const char *name, size_t len, char sign)
1.53 schwarze 3094: {
3095: struct roffreg *reg;
1.65 schwarze 3096: int val;
3097:
3098: if ('.' == name[0] && 2 == len) {
1.145 schwarze 3099: val = roff_getregro(r, name + 1);
1.65 schwarze 3100: if (-1 != val)
1.149 schwarze 3101: return val;
1.65 schwarze 3102: }
1.53 schwarze 3103:
1.199 schwarze 3104: for (reg = r->regtab; reg; reg = reg->next) {
1.53 schwarze 3105: if (len == reg->key.sz &&
1.199 schwarze 3106: 0 == strncmp(name, reg->key.p, len)) {
3107: switch (sign) {
3108: case '+':
3109: reg->val += reg->step;
3110: break;
3111: case '-':
3112: reg->val -= reg->step;
3113: break;
3114: default:
3115: break;
3116: }
1.149 schwarze 3117: return reg->val;
1.199 schwarze 3118: }
3119: }
1.143 schwarze 3120:
1.199 schwarze 3121: roff_setregn(r, name, len, 0, '\0', INT_MIN);
1.149 schwarze 3122: return 0;
1.143 schwarze 3123: }
3124:
3125: static int
3126: roff_hasregn(const struct roff *r, const char *name, size_t len)
3127: {
3128: struct roffreg *reg;
3129: int val;
3130:
3131: if ('.' == name[0] && 2 == len) {
1.145 schwarze 3132: val = roff_getregro(r, name + 1);
1.143 schwarze 3133: if (-1 != val)
1.149 schwarze 3134: return 1;
1.143 schwarze 3135: }
3136:
3137: for (reg = r->regtab; reg; reg = reg->next)
3138: if (len == reg->key.sz &&
3139: 0 == strncmp(name, reg->key.p, len))
1.149 schwarze 3140: return 1;
1.41 schwarze 3141:
1.149 schwarze 3142: return 0;
1.41 schwarze 3143: }
3144:
1.52 schwarze 3145: static void
3146: roff_freereg(struct roffreg *reg)
1.41 schwarze 3147: {
1.52 schwarze 3148: struct roffreg *old_reg;
1.41 schwarze 3149:
1.52 schwarze 3150: while (NULL != reg) {
3151: free(reg->key.p);
3152: old_reg = reg;
3153: reg = reg->next;
3154: free(old_reg);
3155: }
1.41 schwarze 3156: }
1.7 schwarze 3157:
1.212 schwarze 3158: static int
1.6 schwarze 3159: roff_nr(ROFF_ARGS)
1.1 schwarze 3160: {
1.199 schwarze 3161: char *key, *val, *step;
1.85 schwarze 3162: size_t keysz;
1.199 schwarze 3163: int iv, is, len;
1.60 schwarze 3164: char sign;
1.6 schwarze 3165:
1.110 schwarze 3166: key = val = buf->buf + pos;
3167: if (*key == '\0')
1.149 schwarze 3168: return ROFF_IGN;
1.85 schwarze 3169:
3170: keysz = roff_getname(r, &val, ln, pos);
1.235 schwarze 3171: if (key[keysz] == '\\' || key[keysz] == '\t')
1.149 schwarze 3172: return ROFF_IGN;
1.6 schwarze 3173:
1.60 schwarze 3174: sign = *val;
1.110 schwarze 3175: if (sign == '+' || sign == '-')
1.60 schwarze 3176: val++;
3177:
1.199 schwarze 3178: len = 0;
3179: if (roff_evalnum(r, ln, val, &len, &iv, ROFFNUM_SCALE) == 0)
3180: return ROFF_IGN;
3181:
3182: step = val + len;
3183: while (isspace((unsigned char)*step))
3184: step++;
3185: if (roff_evalnum(r, ln, step, NULL, &is, 0) == 0)
3186: is = INT_MIN;
1.1 schwarze 3187:
1.199 schwarze 3188: roff_setregn(r, key, keysz, iv, sign, is);
1.149 schwarze 3189: return ROFF_IGN;
1.76 schwarze 3190: }
3191:
1.212 schwarze 3192: static int
1.76 schwarze 3193: roff_rr(ROFF_ARGS)
3194: {
3195: struct roffreg *reg, **prev;
1.85 schwarze 3196: char *name, *cp;
3197: size_t namesz;
1.76 schwarze 3198:
1.110 schwarze 3199: name = cp = buf->buf + pos;
3200: if (*name == '\0')
1.149 schwarze 3201: return ROFF_IGN;
1.85 schwarze 3202: namesz = roff_getname(r, &cp, ln, pos);
3203: name[namesz] = '\0';
1.76 schwarze 3204:
3205: prev = &r->regtab;
3206: while (1) {
3207: reg = *prev;
1.110 schwarze 3208: if (reg == NULL || !strcmp(name, reg->key.p))
1.76 schwarze 3209: break;
3210: prev = ®->next;
3211: }
1.110 schwarze 3212: if (reg != NULL) {
1.76 schwarze 3213: *prev = reg->next;
3214: free(reg->key.p);
3215: free(reg);
3216: }
1.149 schwarze 3217: return ROFF_IGN;
1.29 schwarze 3218: }
3219:
1.138 schwarze 3220: /* --- handler functions for roff requests -------------------------------- */
3221:
1.212 schwarze 3222: static int
1.29 schwarze 3223: roff_rm(ROFF_ARGS)
3224: {
3225: const char *name;
3226: char *cp;
1.85 schwarze 3227: size_t namesz;
1.29 schwarze 3228:
1.110 schwarze 3229: cp = buf->buf + pos;
3230: while (*cp != '\0') {
1.85 schwarze 3231: name = cp;
1.110 schwarze 3232: namesz = roff_getname(r, &cp, ln, (int)(cp - buf->buf));
1.85 schwarze 3233: roff_setstrn(&r->strtab, name, namesz, NULL, 0, 0);
1.183 schwarze 3234: roff_setstrn(&r->rentab, name, namesz, NULL, 0, 0);
1.235 schwarze 3235: if (name[namesz] == '\\' || name[namesz] == '\t')
1.85 schwarze 3236: break;
1.29 schwarze 3237: }
1.149 schwarze 3238: return ROFF_IGN;
1.51 schwarze 3239: }
3240:
1.212 schwarze 3241: static int
1.51 schwarze 3242: roff_it(ROFF_ARGS)
3243: {
3244: int iv;
3245:
3246: /* Parse the number of lines. */
1.133 schwarze 3247:
3248: if ( ! roff_evalnum(r, ln, buf->buf, &pos, &iv, 0)) {
1.222 schwarze 3249: mandoc_msg(MANDOCERR_IT_NONUM,
3250: ln, ppos, "%s", buf->buf + 1);
1.149 schwarze 3251: return ROFF_IGN;
1.51 schwarze 3252: }
3253:
1.134 schwarze 3254: while (isspace((unsigned char)buf->buf[pos]))
3255: pos++;
3256:
3257: /*
3258: * Arm the input line trap.
3259: * Special-casing "an-trap" is an ugly workaround to cope
3260: * with DocBook stupidly fiddling with man(7) internals.
3261: */
1.133 schwarze 3262:
1.51 schwarze 3263: roffit_lines = iv;
1.134 schwarze 3264: roffit_macro = mandoc_strdup(iv != 1 ||
3265: strcmp(buf->buf + pos, "an-trap") ?
3266: buf->buf + pos : "br");
1.149 schwarze 3267: return ROFF_IGN;
1.47 schwarze 3268: }
3269:
1.212 schwarze 3270: static int
1.47 schwarze 3271: roff_Dd(ROFF_ARGS)
3272: {
1.187 schwarze 3273: int mask;
3274: enum roff_tok t, te;
1.99 schwarze 3275:
1.187 schwarze 3276: switch (tok) {
3277: case ROFF_Dd:
3278: tok = MDOC_Dd;
3279: te = MDOC_MAX;
3280: if (r->format == 0)
3281: r->format = MPARSE_MDOC;
3282: mask = MPARSE_MDOC | MPARSE_QUICK;
3283: break;
3284: case ROFF_TH:
3285: tok = MAN_TH;
3286: te = MAN_MAX;
3287: if (r->format == 0)
3288: r->format = MPARSE_MAN;
3289: mask = MPARSE_QUICK;
3290: break;
3291: default:
3292: abort();
3293: }
3294: if ((r->options & mask) == 0)
3295: for (t = tok; t < te; t++)
3296: roff_setstr(r, roff_name[t], NULL, 0);
1.149 schwarze 3297: return ROFF_CONT;
1.14 schwarze 3298: }
3299:
1.212 schwarze 3300: static int
1.27 schwarze 3301: roff_TE(ROFF_ARGS)
3302: {
1.233 schwarze 3303: r->man->flags &= ~ROFF_NONOFILL;
1.193 schwarze 3304: if (r->tbl == NULL) {
1.222 schwarze 3305: mandoc_msg(MANDOCERR_BLK_NOTOPEN, ln, ppos, "TE");
1.193 schwarze 3306: return ROFF_IGN;
3307: }
1.218 schwarze 3308: if (tbl_end(r->tbl, 0) == 0) {
1.193 schwarze 3309: r->tbl = NULL;
1.130 schwarze 3310: free(buf->buf);
3311: buf->buf = mandoc_strdup(".sp");
3312: buf->sz = 4;
1.201 schwarze 3313: *offs = 0;
1.149 schwarze 3314: return ROFF_REPARSE;
1.130 schwarze 3315: }
1.193 schwarze 3316: r->tbl = NULL;
1.149 schwarze 3317: return ROFF_IGN;
1.27 schwarze 3318: }
3319:
1.212 schwarze 3320: static int
1.27 schwarze 3321: roff_T_(ROFF_ARGS)
3322: {
3323:
3324: if (NULL == r->tbl)
1.222 schwarze 3325: mandoc_msg(MANDOCERR_BLK_NOTOPEN, ln, ppos, "T&");
1.27 schwarze 3326: else
1.168 schwarze 3327: tbl_restart(ln, ppos, r->tbl);
1.27 schwarze 3328:
1.149 schwarze 3329: return ROFF_IGN;
1.27 schwarze 3330: }
3331:
1.102 schwarze 3332: /*
3333: * Handle in-line equation delimiters.
3334: */
1.212 schwarze 3335: static int
1.110 schwarze 3336: roff_eqndelim(struct roff *r, struct buf *buf, int pos)
1.41 schwarze 3337: {
1.105 schwarze 3338: char *cp1, *cp2;
3339: const char *bef_pr, *bef_nl, *mac, *aft_nl, *aft_pr;
1.41 schwarze 3340:
1.102 schwarze 3341: /*
3342: * Outside equations, look for an opening delimiter.
3343: * If we are inside an equation, we already know it is
3344: * in-line, or this function wouldn't have been called;
3345: * so look for a closing delimiter.
3346: */
3347:
1.110 schwarze 3348: cp1 = buf->buf + pos;
1.102 schwarze 3349: cp2 = strchr(cp1, r->eqn == NULL ?
3350: r->last_eqn->odelim : r->last_eqn->cdelim);
3351: if (cp2 == NULL)
1.149 schwarze 3352: return ROFF_CONT;
1.102 schwarze 3353:
1.105 schwarze 3354: *cp2++ = '\0';
3355: bef_pr = bef_nl = aft_nl = aft_pr = "";
3356:
3357: /* Handle preceding text, protecting whitespace. */
3358:
1.110 schwarze 3359: if (*buf->buf != '\0') {
1.105 schwarze 3360: if (r->eqn == NULL)
3361: bef_pr = "\\&";
3362: bef_nl = "\n";
3363: }
3364:
3365: /*
3366: * Prepare replacing the delimiter with an equation macro
3367: * and drop leading white space from the equation.
3368: */
1.102 schwarze 3369:
1.105 schwarze 3370: if (r->eqn == NULL) {
3371: while (*cp2 == ' ')
3372: cp2++;
3373: mac = ".EQ";
3374: } else
3375: mac = ".EN";
3376:
3377: /* Handle following text, protecting whitespace. */
3378:
3379: if (*cp2 != '\0') {
3380: aft_nl = "\n";
3381: if (r->eqn != NULL)
3382: aft_pr = "\\&";
3383: }
3384:
3385: /* Do the actual replacement. */
3386:
1.110 schwarze 3387: buf->sz = mandoc_asprintf(&cp1, "%s%s%s%s%s%s%s", buf->buf,
1.105 schwarze 3388: bef_pr, bef_nl, mac, aft_nl, aft_pr, cp2) + 1;
1.110 schwarze 3389: free(buf->buf);
3390: buf->buf = cp1;
1.102 schwarze 3391:
3392: /* Toggle the in-line state of the eqn subsystem. */
3393:
3394: r->eqn_inline = r->eqn == NULL;
1.149 schwarze 3395: return ROFF_REPARSE;
1.41 schwarze 3396: }
3397:
1.212 schwarze 3398: static int
1.107 schwarze 3399: roff_EQ(ROFF_ARGS)
1.32 schwarze 3400: {
1.191 schwarze 3401: struct roff_node *n;
3402:
1.228 schwarze 3403: if (r->man->meta.macroset == MACROSET_MAN)
1.194 schwarze 3404: man_breakscope(r->man, ROFF_EQ);
1.191 schwarze 3405: n = roff_node_alloc(r->man, ln, ppos, ROFFT_EQN, TOKEN_NONE);
3406: if (ln > r->man->last->line)
3407: n->flags |= NODE_LINE;
1.220 schwarze 3408: n->eqn = eqn_box_new();
1.191 schwarze 3409: roff_node_append(r->man, n);
3410: r->man->next = ROFF_NEXT_SIBLING;
1.32 schwarze 3411:
1.110 schwarze 3412: assert(r->eqn == NULL);
1.191 schwarze 3413: if (r->last_eqn == NULL)
1.223 schwarze 3414: r->last_eqn = eqn_alloc();
1.191 schwarze 3415: else
3416: eqn_reset(r->last_eqn);
3417: r->eqn = r->last_eqn;
3418: r->eqn->node = n;
1.41 schwarze 3419:
1.110 schwarze 3420: if (buf->buf[pos] != '\0')
1.222 schwarze 3421: mandoc_msg(MANDOCERR_ARG_SKIP, ln, pos,
1.110 schwarze 3422: ".EQ %s", buf->buf + pos);
1.41 schwarze 3423:
1.149 schwarze 3424: return ROFF_IGN;
1.32 schwarze 3425: }
3426:
1.212 schwarze 3427: static int
1.32 schwarze 3428: roff_EN(ROFF_ARGS)
3429: {
1.191 schwarze 3430: if (r->eqn != NULL) {
3431: eqn_parse(r->eqn);
3432: r->eqn = NULL;
3433: } else
1.222 schwarze 3434: mandoc_msg(MANDOCERR_BLK_NOTOPEN, ln, ppos, "EN");
1.191 schwarze 3435: if (buf->buf[pos] != '\0')
1.222 schwarze 3436: mandoc_msg(MANDOCERR_ARG_SKIP, ln, pos,
1.191 schwarze 3437: "EN %s", buf->buf + pos);
1.149 schwarze 3438: return ROFF_IGN;
1.32 schwarze 3439: }
3440:
1.212 schwarze 3441: static int
1.27 schwarze 3442: roff_TS(ROFF_ARGS)
3443: {
1.193 schwarze 3444: if (r->tbl != NULL) {
1.222 schwarze 3445: mandoc_msg(MANDOCERR_BLK_BROKEN, ln, ppos, "TS breaks TS");
1.218 schwarze 3446: tbl_end(r->tbl, 0);
1.27 schwarze 3447: }
1.233 schwarze 3448: r->man->flags |= ROFF_NONOFILL;
1.223 schwarze 3449: r->tbl = tbl_alloc(ppos, ln, r->last_tbl);
1.218 schwarze 3450: if (r->last_tbl == NULL)
1.193 schwarze 3451: r->first_tbl = r->tbl;
3452: r->last_tbl = r->tbl;
1.169 schwarze 3453: return ROFF_IGN;
3454: }
3455:
1.212 schwarze 3456: static int
1.230 schwarze 3457: roff_noarg(ROFF_ARGS)
3458: {
3459: if (r->man->flags & (MAN_BLINE | MAN_ELINE))
3460: man_breakscope(r->man, tok);
3461: if (tok == ROFF_brp)
3462: tok = ROFF_br;
3463: roff_elem_alloc(r->man, ln, ppos, tok);
3464: if (buf->buf[pos] != '\0')
3465: mandoc_msg(MANDOCERR_ARG_SKIP, ln, pos,
3466: "%s %s", roff_name[tok], buf->buf + pos);
3467: if (tok == ROFF_nf)
3468: r->man->flags |= ROFF_NOFILL;
3469: else if (tok == ROFF_fi)
3470: r->man->flags &= ~ROFF_NOFILL;
3471: r->man->last->flags |= NODE_LINE | NODE_VALID | NODE_ENDED;
3472: r->man->next = ROFF_NEXT_SIBLING;
3473: return ROFF_IGN;
3474: }
3475:
3476: static int
1.169 schwarze 3477: roff_onearg(ROFF_ARGS)
3478: {
3479: struct roff_node *n;
3480: char *cp;
1.177 schwarze 3481: int npos;
1.169 schwarze 3482:
1.174 schwarze 3483: if (r->man->flags & (MAN_BLINE | MAN_ELINE) &&
1.192 schwarze 3484: (tok == ROFF_ce || tok == ROFF_rj || tok == ROFF_sp ||
3485: tok == ROFF_ti))
1.174 schwarze 3486: man_breakscope(r->man, tok);
3487:
1.181 schwarze 3488: if (roffce_node != NULL && (tok == ROFF_ce || tok == ROFF_rj)) {
1.177 schwarze 3489: r->man->last = roffce_node;
3490: r->man->next = ROFF_NEXT_SIBLING;
3491: }
3492:
1.169 schwarze 3493: roff_elem_alloc(r->man, ln, ppos, tok);
3494: n = r->man->last;
3495:
3496: cp = buf->buf + pos;
3497: if (*cp != '\0') {
3498: while (*cp != '\0' && *cp != ' ')
3499: cp++;
3500: while (*cp == ' ')
3501: *cp++ = '\0';
3502: if (*cp != '\0')
1.222 schwarze 3503: mandoc_msg(MANDOCERR_ARG_EXCESS,
3504: ln, (int)(cp - buf->buf),
1.169 schwarze 3505: "%s ... %s", roff_name[tok], cp);
3506: roff_word_alloc(r->man, ln, pos, buf->buf + pos);
1.172 schwarze 3507: }
3508:
1.181 schwarze 3509: if (tok == ROFF_ce || tok == ROFF_rj) {
3510: if (r->man->last->type == ROFFT_ELEM) {
1.177 schwarze 3511: roff_word_alloc(r->man, ln, pos, "1");
3512: r->man->last->flags |= NODE_NOSRC;
3513: }
3514: npos = 0;
3515: if (roff_evalnum(r, ln, r->man->last->string, &npos,
3516: &roffce_lines, 0) == 0) {
1.222 schwarze 3517: mandoc_msg(MANDOCERR_CE_NONUM,
3518: ln, pos, "ce %s", buf->buf + pos);
1.177 schwarze 3519: roffce_lines = 1;
3520: }
3521: if (roffce_lines < 1) {
3522: r->man->last = r->man->last->parent;
3523: roffce_node = NULL;
3524: roffce_lines = 0;
3525: } else
3526: roffce_node = r->man->last->parent;
3527: } else {
3528: n->flags |= NODE_VALID | NODE_ENDED;
3529: r->man->last = n;
3530: }
3531: n->flags |= NODE_LINE;
1.172 schwarze 3532: r->man->next = ROFF_NEXT_SIBLING;
3533: return ROFF_IGN;
3534: }
3535:
1.212 schwarze 3536: static int
1.172 schwarze 3537: roff_manyarg(ROFF_ARGS)
3538: {
3539: struct roff_node *n;
3540: char *sp, *ep;
3541:
3542: roff_elem_alloc(r->man, ln, ppos, tok);
3543: n = r->man->last;
3544:
3545: for (sp = ep = buf->buf + pos; *sp != '\0'; sp = ep) {
3546: while (*ep != '\0' && *ep != ' ')
3547: ep++;
3548: while (*ep == ' ')
3549: *ep++ = '\0';
3550: roff_word_alloc(r->man, ln, sp - buf->buf, sp);
1.169 schwarze 3551: }
3552:
3553: n->flags |= NODE_LINE | NODE_VALID | NODE_ENDED;
3554: r->man->last = n;
3555: r->man->next = ROFF_NEXT_SIBLING;
1.149 schwarze 3556: return ROFF_IGN;
1.123 schwarze 3557: }
3558:
1.212 schwarze 3559: static int
1.183 schwarze 3560: roff_als(ROFF_ARGS)
3561: {
3562: char *oldn, *newn, *end, *value;
3563: size_t oldsz, newsz, valsz;
3564:
3565: newn = oldn = buf->buf + pos;
3566: if (*newn == '\0')
3567: return ROFF_IGN;
3568:
3569: newsz = roff_getname(r, &oldn, ln, pos);
1.235 schwarze 3570: if (newn[newsz] == '\\' || newn[newsz] == '\t' || *oldn == '\0')
1.183 schwarze 3571: return ROFF_IGN;
3572:
3573: end = oldn;
3574: oldsz = roff_getname(r, &end, ln, oldn - buf->buf);
3575: if (oldsz == 0)
3576: return ROFF_IGN;
3577:
1.210 schwarze 3578: valsz = mandoc_asprintf(&value, ".%.*s \\$@\\\"\n",
1.187 schwarze 3579: (int)oldsz, oldn);
1.183 schwarze 3580: roff_setstrn(&r->strtab, newn, newsz, value, valsz, 0);
3581: roff_setstrn(&r->rentab, newn, newsz, NULL, 0, 0);
3582: free(value);
1.236 schwarze 3583: return ROFF_IGN;
3584: }
3585:
3586: /*
3587: * The .break request only makes sense inside conditionals,
3588: * and that case is already handled in roff_cond_sub().
3589: */
3590: static int
3591: roff_break(ROFF_ARGS)
3592: {
3593: mandoc_msg(MANDOCERR_BLK_NOTOPEN, ln, pos, "break");
1.168 schwarze 3594: return ROFF_IGN;
1.27 schwarze 3595: }
3596:
1.212 schwarze 3597: static int
1.48 schwarze 3598: roff_cc(ROFF_ARGS)
3599: {
3600: const char *p;
3601:
1.110 schwarze 3602: p = buf->buf + pos;
1.48 schwarze 3603:
1.110 schwarze 3604: if (*p == '\0' || (r->control = *p++) == '.')
1.175 schwarze 3605: r->control = '\0';
1.48 schwarze 3606:
1.110 schwarze 3607: if (*p != '\0')
1.222 schwarze 3608: mandoc_msg(MANDOCERR_ARG_EXCESS,
1.132 schwarze 3609: ln, p - buf->buf, "cc ... %s", p);
1.48 schwarze 3610:
1.213 schwarze 3611: return ROFF_IGN;
3612: }
3613:
3614: static int
3615: roff_char(ROFF_ARGS)
3616: {
3617: const char *p, *kp, *vp;
3618: size_t ksz, vsz;
3619: int font;
3620:
3621: /* Parse the character to be replaced. */
3622:
3623: kp = buf->buf + pos;
3624: p = kp + 1;
3625: if (*kp == '\0' || (*kp == '\\' &&
3626: mandoc_escape(&p, NULL, NULL) != ESCAPE_SPECIAL) ||
3627: (*p != ' ' && *p != '\0')) {
1.222 schwarze 3628: mandoc_msg(MANDOCERR_CHAR_ARG, ln, pos, "char %s", kp);
1.213 schwarze 3629: return ROFF_IGN;
3630: }
3631: ksz = p - kp;
3632: while (*p == ' ')
3633: p++;
3634:
3635: /*
3636: * If the replacement string contains a font escape sequence,
3637: * we have to restore the font at the end.
3638: */
3639:
3640: vp = p;
3641: vsz = strlen(p);
3642: font = 0;
3643: while (*p != '\0') {
3644: if (*p++ != '\\')
3645: continue;
3646: switch (mandoc_escape(&p, NULL, NULL)) {
3647: case ESCAPE_FONT:
3648: case ESCAPE_FONTROMAN:
3649: case ESCAPE_FONTITALIC:
3650: case ESCAPE_FONTBOLD:
3651: case ESCAPE_FONTBI:
1.214 schwarze 3652: case ESCAPE_FONTCW:
1.213 schwarze 3653: case ESCAPE_FONTPREV:
3654: font++;
3655: break;
3656: default:
3657: break;
3658: }
3659: }
3660: if (font > 1)
1.222 schwarze 3661: mandoc_msg(MANDOCERR_CHAR_FONT,
3662: ln, (int)(vp - buf->buf), "%s", vp);
1.213 schwarze 3663:
3664: /*
3665: * Approximate the effect of .char using the .tr tables.
3666: * XXX In groff, .char and .tr interact differently.
3667: */
3668:
3669: if (ksz == 1) {
3670: if (r->xtab == NULL)
3671: r->xtab = mandoc_calloc(128, sizeof(*r->xtab));
3672: assert((unsigned int)*kp < 128);
3673: free(r->xtab[(int)*kp].p);
3674: r->xtab[(int)*kp].sz = mandoc_asprintf(&r->xtab[(int)*kp].p,
3675: "%s%s", vp, font ? "\fP" : "");
3676: } else {
3677: roff_setstrn(&r->xmbtab, kp, ksz, vp, vsz, 0);
3678: if (font)
3679: roff_setstrn(&r->xmbtab, kp, ksz, "\\fP", 3, 1);
3680: }
1.149 schwarze 3681: return ROFF_IGN;
1.48 schwarze 3682: }
3683:
1.212 schwarze 3684: static int
1.175 schwarze 3685: roff_ec(ROFF_ARGS)
3686: {
3687: const char *p;
3688:
3689: p = buf->buf + pos;
3690: if (*p == '\0')
3691: r->escape = '\\';
3692: else {
3693: r->escape = *p;
3694: if (*++p != '\0')
1.222 schwarze 3695: mandoc_msg(MANDOCERR_ARG_EXCESS, ln,
3696: (int)(p - buf->buf), "ec ... %s", p);
1.175 schwarze 3697: }
3698: return ROFF_IGN;
3699: }
3700:
1.212 schwarze 3701: static int
1.175 schwarze 3702: roff_eo(ROFF_ARGS)
3703: {
3704: r->escape = '\0';
3705: if (buf->buf[pos] != '\0')
1.222 schwarze 3706: mandoc_msg(MANDOCERR_ARG_SKIP,
1.175 schwarze 3707: ln, pos, "eo %s", buf->buf + pos);
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;
3852: int levels, i;
3853:
3854: levels = 1;
3855: if (buf->buf[pos] != '\0' &&
3856: roff_evalnum(r, ln, buf->buf, &pos, &levels, 0) == 0) {
1.222 schwarze 3857: mandoc_msg(MANDOCERR_CE_NONUM,
1.211 schwarze 3858: ln, pos, "shift %s", buf->buf + pos);
3859: levels = 1;
3860: }
3861: if (r->mstackpos < 0) {
1.222 schwarze 3862: mandoc_msg(MANDOCERR_REQ_NOMAC, ln, ppos, "shift");
1.211 schwarze 3863: return ROFF_IGN;
3864: }
3865: ctx = r->mstack + r->mstackpos;
3866: if (levels > ctx->argc) {
1.222 schwarze 3867: mandoc_msg(MANDOCERR_SHIFT,
1.211 schwarze 3868: ln, pos, "%d, but max is %d", levels, ctx->argc);
3869: levels = ctx->argc;
3870: }
3871: if (levels == 0)
3872: return ROFF_IGN;
3873: for (i = 0; i < levels; i++)
3874: free(ctx->argv[i]);
3875: ctx->argc -= levels;
3876: for (i = 0; i < ctx->argc; i++)
3877: ctx->argv[i] = ctx->argv[i + levels];
3878: return ROFF_IGN;
3879: }
3880:
1.212 schwarze 3881: static int
1.14 schwarze 3882: roff_so(ROFF_ARGS)
3883: {
1.121 schwarze 3884: char *name, *cp;
1.15 schwarze 3885:
1.110 schwarze 3886: name = buf->buf + pos;
1.222 schwarze 3887: mandoc_msg(MANDOCERR_SO, ln, ppos, "so %s", name);
1.14 schwarze 3888:
1.22 schwarze 3889: /*
3890: * Handle `so'. Be EXTREMELY careful, as we shouldn't be
3891: * opening anything that's not in our cwd or anything beneath
3892: * it. Thus, explicitly disallow traversing up the file-system
3893: * or using absolute paths.
3894: */
3895:
1.110 schwarze 3896: if (*name == '/' || strstr(name, "../") || strstr(name, "/..")) {
1.222 schwarze 3897: mandoc_msg(MANDOCERR_SO_PATH, ln, ppos, ".so %s", name);
1.121 schwarze 3898: buf->sz = mandoc_asprintf(&cp,
3899: ".sp\nSee the file %s.\n.sp", name) + 1;
3900: free(buf->buf);
3901: buf->buf = cp;
3902: *offs = 0;
1.149 schwarze 3903: return ROFF_REPARSE;
1.14 schwarze 3904: }
3905:
3906: *offs = pos;
1.149 schwarze 3907: return ROFF_SO;
1.7 schwarze 3908: }
3909:
1.138 schwarze 3910: /* --- user defined strings and macros ------------------------------------ */
3911:
1.212 schwarze 3912: static int
1.16 schwarze 3913: roff_userdef(ROFF_ARGS)
1.12 schwarze 3914: {
1.211 schwarze 3915: struct mctx *ctx;
3916: char *arg, *ap, *dst, *src;
3917: size_t sz;
3918:
1.237 schwarze 3919: /* If the macro is empty, ignore it altogether. */
3920:
3921: if (*r->current_string == '\0')
3922: return ROFF_IGN;
3923:
1.211 schwarze 3924: /* Initialize a new macro stack context. */
3925:
3926: if (++r->mstackpos == r->mstacksz) {
3927: r->mstack = mandoc_recallocarray(r->mstack,
3928: r->mstacksz, r->mstacksz + 8, sizeof(*r->mstack));
3929: r->mstacksz += 8;
3930: }
3931: ctx = r->mstack + r->mstackpos;
3932: ctx->argsz = 0;
3933: ctx->argc = 0;
3934: ctx->argv = NULL;
3935:
3936: /*
3937: * Collect pointers to macro argument strings,
3938: * NUL-terminating them and escaping quotes.
3939: */
3940:
3941: src = buf->buf + pos;
3942: while (*src != '\0') {
3943: if (ctx->argc == ctx->argsz) {
3944: ctx->argsz += 8;
3945: ctx->argv = mandoc_reallocarray(ctx->argv,
3946: ctx->argsz, sizeof(*ctx->argv));
3947: }
1.227 schwarze 3948: arg = roff_getarg(r, &src, ln, &pos);
1.211 schwarze 3949: sz = 1; /* For the terminating NUL. */
3950: for (ap = arg; *ap != '\0'; ap++)
3951: sz += *ap == '"' ? 4 : 1;
3952: ctx->argv[ctx->argc++] = dst = mandoc_malloc(sz);
3953: for (ap = arg; *ap != '\0'; ap++) {
3954: if (*ap == '"') {
3955: memcpy(dst, "\\(dq", 4);
3956: dst += 4;
3957: } else
3958: *dst++ = *ap;
1.16 schwarze 3959: }
1.211 schwarze 3960: *dst = '\0';
1.227 schwarze 3961: free(arg);
1.209 schwarze 3962: }
3963:
1.211 schwarze 3964: /* Replace the macro invocation by the macro definition. */
1.135 schwarze 3965:
1.110 schwarze 3966: free(buf->buf);
1.211 schwarze 3967: buf->buf = mandoc_strdup(r->current_string);
3968: buf->sz = strlen(buf->buf) + 1;
1.120 schwarze 3969: *offs = 0;
1.16 schwarze 3970:
1.237 schwarze 3971: return buf->buf[buf->sz - 2] == '\n' ?
1.212 schwarze 3972: ROFF_REPARSE | ROFF_USERCALL : ROFF_IGN | ROFF_APPEND;
1.12 schwarze 3973: }
1.28 schwarze 3974:
1.178 schwarze 3975: /*
3976: * Calling a high-level macro that was renamed with .rn.
3977: * r->current_string has already been set up by roff_parse().
3978: */
1.212 schwarze 3979: static int
1.178 schwarze 3980: roff_renamed(ROFF_ARGS)
3981: {
3982: char *nbuf;
3983:
1.187 schwarze 3984: buf->sz = mandoc_asprintf(&nbuf, ".%s%s%s", r->current_string,
3985: buf->buf[pos] == '\0' ? "" : " ", buf->buf + pos) + 1;
1.178 schwarze 3986: free(buf->buf);
3987: buf->buf = nbuf;
1.201 schwarze 3988: *offs = 0;
1.178 schwarze 3989: return ROFF_CONT;
3990: }
3991:
1.234 schwarze 3992: /*
3993: * Measure the length in bytes of the roff identifier at *cpp
3994: * and advance the pointer to the next word.
3995: */
1.85 schwarze 3996: static size_t
1.28 schwarze 3997: roff_getname(struct roff *r, char **cpp, int ln, int pos)
3998: {
3999: char *name, *cp;
1.85 schwarze 4000: size_t namesz;
1.28 schwarze 4001:
4002: name = *cpp;
1.234 schwarze 4003: if (*name == '\0')
1.149 schwarze 4004: return 0;
1.28 schwarze 4005:
1.234 schwarze 4006: /* Advance cp to the byte after the end of the name. */
4007:
1.85 schwarze 4008: for (cp = name; 1; cp++) {
1.234 schwarze 4009: namesz = cp - name;
1.235 schwarze 4010: if (*cp == '\0')
1.85 schwarze 4011: break;
1.235 schwarze 4012: if (*cp == ' ' || *cp == '\t') {
4013: cp++;
4014: break;
4015: }
1.234 schwarze 4016: if (*cp != '\\')
1.28 schwarze 4017: continue;
1.234 schwarze 4018: if (cp[1] == '{' || cp[1] == '}')
1.88 schwarze 4019: break;
1.234 schwarze 4020: if (*++cp == '\\')
1.28 schwarze 4021: continue;
1.222 schwarze 4022: mandoc_msg(MANDOCERR_NAMESC, ln, pos,
1.97 schwarze 4023: "%.*s", (int)(cp - name + 1), name);
1.85 schwarze 4024: mandoc_escape((const char **)&cp, NULL, NULL);
4025: break;
1.28 schwarze 4026: }
4027:
4028: /* Read past spaces. */
1.234 schwarze 4029:
4030: while (*cp == ' ')
1.28 schwarze 4031: cp++;
4032:
4033: *cpp = cp;
1.149 schwarze 4034: return namesz;
1.28 schwarze 4035: }
4036:
1.16 schwarze 4037: /*
4038: * Store *string into the user-defined string called *name.
4039: * To clear an existing entry, call with (*r, *name, NULL, 0).
1.66 schwarze 4040: * append == 0: replace mode
4041: * append == 1: single-line append mode
4042: * append == 2: multiline append mode, append '\n' after each call
1.16 schwarze 4043: */
1.8 schwarze 4044: static void
1.16 schwarze 4045: roff_setstr(struct roff *r, const char *name, const char *string,
1.66 schwarze 4046: int append)
1.7 schwarze 4047: {
1.187 schwarze 4048: size_t namesz;
1.42 schwarze 4049:
1.187 schwarze 4050: namesz = strlen(name);
4051: roff_setstrn(&r->strtab, name, namesz, string,
1.80 schwarze 4052: string ? strlen(string) : 0, append);
1.187 schwarze 4053: roff_setstrn(&r->rentab, name, namesz, NULL, 0, 0);
1.42 schwarze 4054: }
4055:
4056: static void
4057: roff_setstrn(struct roffkv **r, const char *name, size_t namesz,
1.66 schwarze 4058: const char *string, size_t stringsz, int append)
1.42 schwarze 4059: {
4060: struct roffkv *n;
4061: char *c;
4062: int i;
4063: size_t oldch, newch;
1.7 schwarze 4064:
1.16 schwarze 4065: /* Search for an existing string with the same name. */
1.42 schwarze 4066: n = *r;
4067:
1.84 schwarze 4068: while (n && (namesz != n->key.sz ||
4069: strncmp(n->key.p, name, namesz)))
1.7 schwarze 4070: n = n->next;
1.8 schwarze 4071:
4072: if (NULL == n) {
1.16 schwarze 4073: /* Create a new string table entry. */
1.42 schwarze 4074: n = mandoc_malloc(sizeof(struct roffkv));
4075: n->key.p = mandoc_strndup(name, namesz);
4076: n->key.sz = namesz;
4077: n->val.p = NULL;
4078: n->val.sz = 0;
4079: n->next = *r;
4080: *r = n;
1.66 schwarze 4081: } else if (0 == append) {
1.42 schwarze 4082: free(n->val.p);
4083: n->val.p = NULL;
4084: n->val.sz = 0;
1.16 schwarze 4085: }
4086:
4087: if (NULL == string)
4088: return;
4089:
4090: /*
4091: * One additional byte for the '\n' in multiline mode,
4092: * and one for the terminating '\0'.
4093: */
1.66 schwarze 4094: newch = stringsz + (1 < append ? 2u : 1u);
1.42 schwarze 4095:
4096: if (NULL == n->val.p) {
4097: n->val.p = mandoc_malloc(newch);
4098: *n->val.p = '\0';
1.16 schwarze 4099: oldch = 0;
4100: } else {
1.42 schwarze 4101: oldch = n->val.sz;
4102: n->val.p = mandoc_realloc(n->val.p, oldch + newch);
1.16 schwarze 4103: }
4104:
4105: /* Skip existing content in the destination buffer. */
1.42 schwarze 4106: c = n->val.p + (int)oldch;
1.16 schwarze 4107:
4108: /* Append new content to the destination buffer. */
1.42 schwarze 4109: i = 0;
4110: while (i < (int)stringsz) {
1.16 schwarze 4111: /*
4112: * Rudimentary roff copy mode:
4113: * Handle escaped backslashes.
4114: */
1.42 schwarze 4115: if ('\\' == string[i] && '\\' == string[i + 1])
4116: i++;
4117: *c++ = string[i++];
1.16 schwarze 4118: }
1.8 schwarze 4119:
1.16 schwarze 4120: /* Append terminating bytes. */
1.66 schwarze 4121: if (1 < append)
1.16 schwarze 4122: *c++ = '\n';
1.42 schwarze 4123:
1.16 schwarze 4124: *c = '\0';
1.42 schwarze 4125: n->val.sz = (int)(c - n->val.p);
1.7 schwarze 4126: }
4127:
1.8 schwarze 4128: static const char *
1.197 schwarze 4129: roff_getstrn(struct roff *r, const char *name, size_t len,
1.187 schwarze 4130: int *deftype)
1.7 schwarze 4131: {
1.187 schwarze 4132: const struct roffkv *n;
1.197 schwarze 4133: int found, i;
1.187 schwarze 4134: enum roff_tok tok;
4135:
1.197 schwarze 4136: found = 0;
4137: for (n = r->strtab; n != NULL; n = n->next) {
4138: if (strncmp(name, n->key.p, len) != 0 ||
4139: n->key.p[len] != '\0' || n->val.p == NULL)
4140: continue;
4141: if (*deftype & ROFFDEF_USER) {
4142: *deftype = ROFFDEF_USER;
4143: return n->val.p;
4144: } else {
4145: found = 1;
4146: break;
4147: }
4148: }
4149: for (n = r->rentab; n != NULL; n = n->next) {
4150: if (strncmp(name, n->key.p, len) != 0 ||
4151: n->key.p[len] != '\0' || n->val.p == NULL)
4152: continue;
4153: if (*deftype & ROFFDEF_REN) {
4154: *deftype = ROFFDEF_REN;
4155: return n->val.p;
4156: } else {
4157: found = 1;
4158: break;
1.187 schwarze 4159: }
4160: }
1.197 schwarze 4161: for (i = 0; i < PREDEFS_MAX; i++) {
4162: if (strncmp(name, predefs[i].name, len) != 0 ||
4163: predefs[i].name[len] != '\0')
4164: continue;
4165: if (*deftype & ROFFDEF_PRE) {
4166: *deftype = ROFFDEF_PRE;
4167: return predefs[i].str;
4168: } else {
4169: found = 1;
4170: break;
1.187 schwarze 4171: }
4172: }
1.228 schwarze 4173: if (r->man->meta.macroset != MACROSET_MAN) {
1.197 schwarze 4174: for (tok = MDOC_Dd; tok < MDOC_MAX; tok++) {
4175: if (strncmp(name, roff_name[tok], len) != 0 ||
4176: roff_name[tok][len] != '\0')
4177: continue;
4178: if (*deftype & ROFFDEF_STD) {
4179: *deftype = ROFFDEF_STD;
4180: return NULL;
4181: } else {
4182: found = 1;
4183: break;
1.187 schwarze 4184: }
4185: }
4186: }
1.228 schwarze 4187: if (r->man->meta.macroset != MACROSET_MDOC) {
1.197 schwarze 4188: for (tok = MAN_TH; tok < MAN_MAX; tok++) {
4189: if (strncmp(name, roff_name[tok], len) != 0 ||
4190: roff_name[tok][len] != '\0')
4191: continue;
4192: if (*deftype & ROFFDEF_STD) {
4193: *deftype = ROFFDEF_STD;
4194: return NULL;
4195: } else {
4196: found = 1;
4197: break;
1.187 schwarze 4198: }
4199: }
1.197 schwarze 4200: }
4201:
4202: if (found == 0 && *deftype != ROFFDEF_ANY) {
4203: if (*deftype & ROFFDEF_REN) {
4204: /*
4205: * This might still be a request,
4206: * so do not treat it as undefined yet.
4207: */
4208: *deftype = ROFFDEF_UNDEF;
4209: return NULL;
1.187 schwarze 4210: }
1.197 schwarze 4211:
4212: /* Using an undefined string defines it to be empty. */
4213:
4214: roff_setstrn(&r->strtab, name, len, "", 0, 0);
4215: roff_setstrn(&r->rentab, name, len, NULL, 0, 0);
1.187 schwarze 4216: }
1.197 schwarze 4217:
1.187 schwarze 4218: *deftype = 0;
1.149 schwarze 4219: return NULL;
1.7 schwarze 4220: }
4221:
1.8 schwarze 4222: static void
1.42 schwarze 4223: roff_freestr(struct roffkv *r)
1.7 schwarze 4224: {
1.42 schwarze 4225: struct roffkv *n, *nn;
1.7 schwarze 4226:
1.42 schwarze 4227: for (n = r; n; n = nn) {
4228: free(n->key.p);
4229: free(n->val.p);
1.7 schwarze 4230: nn = n->next;
4231: free(n);
4232: }
1.27 schwarze 4233: }
1.138 schwarze 4234:
4235: /* --- accessors and utility functions ------------------------------------ */
1.42 schwarze 4236:
4237: /*
4238: * Duplicate an input string, making the appropriate character
4239: * conversations (as stipulated by `tr') along the way.
4240: * Returns a heap-allocated string with all the replacements made.
4241: */
4242: char *
4243: roff_strdup(const struct roff *r, const char *p)
4244: {
4245: const struct roffkv *cp;
4246: char *res;
4247: const char *pp;
4248: size_t ssz, sz;
4249: enum mandoc_esc esc;
4250:
4251: if (NULL == r->xmbtab && NULL == r->xtab)
1.149 schwarze 4252: return mandoc_strdup(p);
1.42 schwarze 4253: else if ('\0' == *p)
1.149 schwarze 4254: return mandoc_strdup("");
1.42 schwarze 4255:
4256: /*
4257: * Step through each character looking for term matches
4258: * (remember that a `tr' can be invoked with an escape, which is
4259: * a glyph but the escape is multi-character).
4260: * We only do this if the character hash has been initialised
4261: * and the string is >0 length.
4262: */
4263:
4264: res = NULL;
4265: ssz = 0;
4266:
4267: while ('\0' != *p) {
1.161 schwarze 4268: assert((unsigned int)*p < 128);
4269: if ('\\' != *p && r->xtab && r->xtab[(unsigned int)*p].p) {
1.42 schwarze 4270: sz = r->xtab[(int)*p].sz;
4271: res = mandoc_realloc(res, ssz + sz + 1);
4272: memcpy(res + ssz, r->xtab[(int)*p].p, sz);
4273: ssz += sz;
4274: p++;
4275: continue;
4276: } else if ('\\' != *p) {
4277: res = mandoc_realloc(res, ssz + 2);
4278: res[ssz++] = *p++;
4279: continue;
4280: }
4281:
4282: /* Search for term matches. */
4283: for (cp = r->xmbtab; cp; cp = cp->next)
4284: if (0 == strncmp(p, cp->key.p, cp->key.sz))
4285: break;
4286:
4287: if (NULL != cp) {
4288: /*
4289: * A match has been found.
4290: * Append the match to the array and move
4291: * forward by its keysize.
4292: */
1.80 schwarze 4293: res = mandoc_realloc(res,
4294: ssz + cp->val.sz + 1);
1.42 schwarze 4295: memcpy(res + ssz, cp->val.p, cp->val.sz);
4296: ssz += cp->val.sz;
4297: p += (int)cp->key.sz;
4298: continue;
4299: }
4300:
4301: /*
4302: * Handle escapes carefully: we need to copy
4303: * over just the escape itself, or else we might
4304: * do replacements within the escape itself.
4305: * Make sure to pass along the bogus string.
4306: */
4307: pp = p++;
4308: esc = mandoc_escape(&p, NULL, NULL);
4309: if (ESCAPE_ERROR == esc) {
4310: sz = strlen(pp);
4311: res = mandoc_realloc(res, ssz + sz + 1);
4312: memcpy(res + ssz, pp, sz);
4313: break;
4314: }
1.80 schwarze 4315: /*
4316: * We bail out on bad escapes.
1.42 schwarze 4317: * No need to warn: we already did so when
1.227 schwarze 4318: * roff_expand() was called.
1.42 schwarze 4319: */
4320: sz = (int)(p - pp);
4321: res = mandoc_realloc(res, ssz + sz + 1);
4322: memcpy(res + ssz, pp, sz);
4323: ssz += sz;
4324: }
4325:
4326: res[(int)ssz] = '\0';
1.149 schwarze 4327: return res;
1.99 schwarze 4328: }
4329:
4330: int
4331: roff_getformat(const struct roff *r)
4332: {
4333:
1.149 schwarze 4334: return r->format;
1.48 schwarze 4335: }
4336:
4337: /*
1.80 schwarze 4338: * Find out whether a line is a macro line or not.
1.48 schwarze 4339: * If it is, adjust the current position and return one; if it isn't,
4340: * return zero and don't change the current position.
4341: * If the control character has been set with `.cc', then let that grain
4342: * precedence.
4343: * This is slighly contrary to groff, where using the non-breaking
4344: * control character when `cc' has been invoked will cause the
4345: * non-breaking macro contents to be printed verbatim.
4346: */
4347: int
4348: roff_getcontrol(const struct roff *r, const char *cp, int *ppos)
4349: {
4350: int pos;
4351:
4352: pos = *ppos;
4353:
1.175 schwarze 4354: if (r->control != '\0' && cp[pos] == r->control)
1.48 schwarze 4355: pos++;
1.175 schwarze 4356: else if (r->control != '\0')
1.149 schwarze 4357: return 0;
1.48 schwarze 4358: else if ('\\' == cp[pos] && '.' == cp[pos + 1])
4359: pos += 2;
4360: else if ('.' == cp[pos] || '\'' == cp[pos])
4361: pos++;
4362: else
1.149 schwarze 4363: return 0;
1.48 schwarze 4364:
4365: while (' ' == cp[pos] || '\t' == cp[pos])
4366: pos++;
4367:
4368: *ppos = pos;
1.149 schwarze 4369: return 1;
1.1 schwarze 4370: }