Annotation of src/usr.bin/make/lowparse.c, Revision 1.31
1.31 ! espie 1: /* $OpenBSD: lowparse.c,v 1.30 2012/10/02 10:29:31 espie Exp $ */
1.1 espie 2:
3: /* low-level parsing functions. */
4:
5: /*
6: * Copyright (c) 1999,2000 Marc Espie.
7: *
8: * Extensive code changes for the OpenBSD project.
9: *
10: * Redistribution and use in source and binary forms, with or without
11: * modification, are permitted provided that the following conditions
12: * are met:
13: * 1. Redistributions of source code must retain the above copyright
14: * notice, this list of conditions and the following disclaimer.
15: * 2. Redistributions in binary form must reproduce the above copyright
16: * notice, this list of conditions and the following disclaimer in the
17: * documentation and/or other materials provided with the distribution.
18: *
19: * THIS SOFTWARE IS PROVIDED BY THE OPENBSD PROJECT AND CONTRIBUTORS
20: * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
21: * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
22: * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE OPENBSD
23: * PROJECT OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
24: * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
25: * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
26: * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
27: * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
28: * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
29: * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
30: */
31:
1.9 espie 32: #include <assert.h>
33: #include <stddef.h>
1.1 espie 34: #include <stdio.h>
1.11 espie 35: #include <stdlib.h>
1.10 espie 36: #include <string.h>
1.9 espie 37: #include <unistd.h>
38: #include "config.h"
39: #include "defines.h"
1.1 espie 40: #include "buf.h"
41: #include "lowparse.h"
1.9 espie 42: #include "error.h"
43: #include "lst.h"
44: #include "memory.h"
1.30 espie 45: #include "pathnames.h"
1.27 espie 46: #ifndef LOCATION_TYPE
1.26 espie 47: #include "location.h"
1.27 espie 48: #endif
1.26 espie 49:
1.9 espie 50:
51: /* XXX check whether we can free filenames at the end, for a proper
52: * definition of `end'. */
1.1 espie 53:
1.9 espie 54: #if 0
1.1 espie 55: static LIST fileNames; /* file names to free at end */
56: #endif
57:
1.22 espie 58: /* Input stream structure: file or string.
59: * Files have str == NULL, F != NULL.
60: * Strings have F == NULL, str != NULL.
61: */
62: struct input_stream {
1.26 espie 63: Location origin; /* Name of file and line number */
1.21 espie 64: FILE *F; /* Open stream, or NULL if pure string. */
65: char *str; /* Input string, if F == NULL. */
66:
67: /* Line buffer. */
68: char *ptr; /* Where we are. */
69: char *end; /* Don't overdo it. */
1.22 espie 70: };
1.2 espie 71:
1.22 espie 72: static struct input_stream *current; /* the input_stream being parsed. */
1.2 espie 73:
1.22 espie 74: static LIST input_stack; /* Stack of input_stream waiting to be parsed
1.9 espie 75: * (includes and loop reparses) */
1.1 espie 76:
1.28 espie 77: /* record gnode location for proper reporting at runtime */
78: static Location *post_parse = NULL;
79:
1.22 espie 80: /* input_stream ctors.
1.9 espie 81: *
1.22 espie 82: * obj = new_input_file(filename, filehandle);
83: * Create input stream from filename, filehandle. */
84: static struct input_stream *new_input_file(const char *, FILE *);
1.26 espie 85: /* obj = new_input_string(str, origin);
86: * Create input stream from str, origin. */
87: static struct input_stream *new_input_string(char *, const Location *);
1.22 espie 88: /* free_input_stream(obj);
89: * Discard consumed input stream, closing files, freeing memory. */
90: static void free_input_stream(struct input_stream *);
1.9 espie 91:
92:
1.20 espie 93: /* Handling basic character reading.
1.22 espie 94: * c = read_char();
1.9 espie 95: * New character c from current input stream, or EOF at end of stream. */
1.22 espie 96: #define read_char() \
97: current->ptr < current->end ? *current->ptr++ : grab_new_line_and_readchar()
98: /* char = grab_new_line_and_readchar();
99: * Guts for read_char. Grabs a new line off fgetln when we have
100: * consumed the current line and returns the first char, or EOF at end of
1.9 espie 101: * stream. */
1.22 espie 102: static int grab_new_line_and_readchar(void);
103: /* c = skip_to_end_of_line();
1.9 espie 104: * Skips to the end of the current line, returns either '\n' or EOF. */
1.22 espie 105: static int skip_to_end_of_line(void);
1.1 espie 106:
107:
1.9 espie 108: /* Helper functions to handle basic parsing. */
1.22 espie 109: /* read_logical_line(buffer, firstchar);
1.9 espie 110: * Grabs logical line into buffer, the first character has already been
111: * read into firstchar. */
1.22 espie 112: static void read_logical_line(Buffer, int);
1.1 espie 113:
1.9 espie 114: /* firstchar = ParseSkipEmptyLines(buffer);
115: * Scans lines, skipping empty lines. May put some characters into
116: * buffer, returns the first character useful to continue parsing
117: * (e.g., not a backslash or a space. */
1.22 espie 118: static int skip_empty_lines_and_read_char(Buffer);
1.1 espie 119:
1.29 espie 120: const char *curdir;
121: size_t curdir_len;
122:
123: void
124: Parse_setcurdir(const char *dir)
125: {
126: curdir = dir;
127: curdir_len = strlen(dir);
128: }
129:
1.30 espie 130: static bool
131: startswith(const char *f, const char *s, size_t len)
132: {
133: return strncmp(f, s, len) == 0 && f[len] == '/';
134: }
135:
1.29 espie 136: static const char *
137: simplify(const char *filename)
138: {
1.30 espie 139: if (startswith(filename, curdir, curdir_len))
1.29 espie 140: return filename + curdir_len + 1;
1.30 espie 141: else if (startswith(filename, _PATH_DEFSYSPATH,
142: sizeof(_PATH_DEFSYSPATH)-1)) {
143: size_t sz;
144: char *buf;
145: sz = strlen(filename) - sizeof(_PATH_DEFSYSPATH)+3;
146: buf = emalloc(sz);
147: snprintf(buf, sz, "<%s>", filename+sizeof(_PATH_DEFSYSPATH));
148: return buf;
149: } else
1.29 espie 150: return filename;
151: }
152:
1.22 espie 153: static struct input_stream *
154: new_input_file(const char *name, FILE *stream)
1.1 espie 155: {
1.22 espie 156: struct input_stream *istream;
1.9 espie 157: #if 0
1.21 espie 158: Lst_AtEnd(&fileNames, name);
1.1 espie 159: #endif
160:
1.22 espie 161: istream = emalloc(sizeof(*istream));
1.29 espie 162: istream->origin.fname = simplify(name);
1.22 espie 163: istream->str = NULL;
1.21 espie 164: /* Naturally enough, we start reading at line 0. */
1.26 espie 165: istream->origin.lineno = 0;
1.22 espie 166: istream->F = stream;
167: istream->ptr = istream->end = NULL;
168: return istream;
1.1 espie 169: }
170:
171: static void
1.22 espie 172: free_input_stream(struct input_stream *istream)
1.1 espie 173: {
1.22 espie 174: if (istream->F && fileno(istream->F) != STDIN_FILENO)
175: (void)fclose(istream->F);
176: free(istream->str);
1.21 espie 177: /* Note we can't free the file names yet, as they are embedded in GN
178: * for error reports. */
1.22 espie 179: free(istream);
1.1 espie 180: }
181:
1.22 espie 182: static struct input_stream *
1.26 espie 183: new_input_string(char *str, const Location *origin)
1.1 espie 184: {
1.22 espie 185: struct input_stream *istream;
1.1 espie 186:
1.22 espie 187: istream = emalloc(sizeof(*istream));
1.26 espie 188: /* No malloc, name is always taken from an already existing istream
189: * and strings are used in for loops, so we need to reset the line counter
190: * to an appropriate value. */
191: istream->origin = *origin;
1.22 espie 192: istream->F = NULL;
193: istream->ptr = istream->str = str;
194: istream->end = str + strlen(str);
195: return istream;
1.1 espie 196: }
197:
198:
199: void
1.18 espie 200: Parse_FromString(char *str, unsigned long lineno)
1.1 espie 201: {
1.26 espie 202: Location origin;
203:
204: origin.fname = current->origin.fname;
205: origin.lineno = lineno;
1.21 espie 206: if (DEBUG(FOR))
207: (void)fprintf(stderr, "%s\n----\n", str);
1.1 espie 208:
1.25 espie 209: Lst_Push(&input_stack, current);
210: assert(current != NULL);
1.26 espie 211: current = new_input_string(str, &origin);
1.1 espie 212: }
213:
214:
1.6 espie 215: void
1.18 espie 216: Parse_FromFile(const char *name, FILE *stream)
1.1 espie 217: {
1.21 espie 218: if (current != NULL)
219: Lst_Push(&input_stack, current);
1.22 espie 220: current = new_input_file(name, stream);
1.1 espie 221: }
222:
1.9 espie 223: bool
1.18 espie 224: Parse_NextFile(void)
1.1 espie 225: {
1.21 espie 226: if (current != NULL)
1.22 espie 227: free_input_stream(current);
228: current = (struct input_stream *)Lst_Pop(&input_stack);
1.21 espie 229: return current != NULL;
1.1 espie 230: }
231:
1.5 espie 232: static int
1.22 espie 233: grab_new_line_and_readchar(void)
1.1 espie 234: {
1.21 espie 235: size_t len;
1.1 espie 236:
1.21 espie 237: if (current->F) {
238: current->ptr = fgetln(current->F, &len);
239: if (current->ptr) {
240: current->end = current->ptr + len;
241: return *current->ptr++;
242: } else {
243: current->end = NULL;
244: }
1.1 espie 245: }
1.21 espie 246: return EOF;
1.1 espie 247: }
248:
1.6 espie 249: static int
1.22 espie 250: skip_to_end_of_line(void)
1.1 espie 251: {
1.21 espie 252: if (current->F) {
253: if (current->end - current->ptr > 1)
254: current->ptr = current->end - 1;
255: if (*current->ptr == '\n')
256: return *current->ptr++;
257: return EOF;
258: } else {
259: int c;
1.6 espie 260:
1.21 espie 261: do {
1.22 espie 262: c = read_char();
1.21 espie 263: } while (c != '\n' && c != EOF);
264: return c;
265: }
1.2 espie 266: }
267:
1.6 espie 268:
1.2 espie 269: char *
1.18 espie 270: Parse_ReadNextConditionalLine(Buffer linebuf)
1.2 espie 271: {
1.21 espie 272: int c;
1.6 espie 273:
274: /* If first char isn't dot, skip to end of line, handling \ */
1.22 espie 275: while ((c = read_char()) != '.') {
276: for (;c != '\n'; c = read_char()) {
1.21 espie 277: if (c == '\\') {
1.22 espie 278: c = read_char();
1.21 espie 279: if (c == '\n')
1.26 espie 280: current->origin.lineno++;
1.21 espie 281: }
1.31 ! espie 282: if (c == EOF)
! 283: /* Unclosed conditional, reported by cond.c */
1.21 espie 284: return NULL;
285: }
1.26 espie 286: current->origin.lineno++;
1.6 espie 287: }
1.2 espie 288:
1.6 espie 289: /* This is the line we need to copy */
1.21 espie 290: return Parse_ReadUnparsedLine(linebuf, "conditional");
1.6 espie 291: }
1.2 espie 292:
1.6 espie 293: static void
1.22 espie 294: read_logical_line(Buffer linebuf, int c)
1.6 espie 295: {
1.21 espie 296: for (;;) {
297: if (c == '\n') {
1.26 espie 298: current->origin.lineno++;
1.21 espie 299: break;
300: }
301: if (c == EOF)
302: break;
303: Buf_AddChar(linebuf, c);
1.22 espie 304: c = read_char();
1.21 espie 305: while (c == '\\') {
1.22 espie 306: c = read_char();
1.21 espie 307: if (c == '\n') {
308: Buf_AddSpace(linebuf);
1.26 espie 309: current->origin.lineno++;
1.21 espie 310: do {
1.22 espie 311: c = read_char();
1.21 espie 312: } while (c == ' ' || c == '\t');
313: } else {
314: Buf_AddChar(linebuf, '\\');
315: if (c == '\\') {
316: Buf_AddChar(linebuf, '\\');
1.22 espie 317: c = read_char();
1.21 espie 318: }
319: break;
320: }
1.6 espie 321: }
322: }
1.2 espie 323: }
324:
325: char *
1.18 espie 326: Parse_ReadUnparsedLine(Buffer linebuf, const char *type)
1.6 espie 327: {
1.21 espie 328: int c;
1.6 espie 329:
1.21 espie 330: Buf_Reset(linebuf);
1.22 espie 331: c = read_char();
1.21 espie 332: if (c == EOF) {
333: Parse_Error(PARSE_FATAL, "Unclosed %s", type);
334: return NULL;
335: }
1.6 espie 336:
1.21 espie 337: /* Handle '\' at beginning of line, since \\n needs special treatment */
338: while (c == '\\') {
1.22 espie 339: c = read_char();
1.21 espie 340: if (c == '\n') {
1.26 espie 341: current->origin.lineno++;
1.21 espie 342: do {
1.22 espie 343: c = read_char();
1.21 espie 344: } while (c == ' ' || c == '\t');
345: } else {
346: Buf_AddChar(linebuf, '\\');
347: if (c == '\\') {
348: Buf_AddChar(linebuf, '\\');
1.22 espie 349: c = read_char();
1.21 espie 350: }
351: break;
352: }
1.2 espie 353: }
1.22 espie 354: read_logical_line(linebuf, c);
1.6 espie 355:
1.21 espie 356: return Buf_Retrieve(linebuf);
1.6 espie 357: }
358:
1.9 espie 359: /* This is a fairly complex function, but without it, we could not skip
360: * blocks of comments without reading them. */
1.6 espie 361: static int
1.22 espie 362: skip_empty_lines_and_read_char(Buffer linebuf)
1.6 espie 363: {
1.21 espie 364: int c; /* the current character */
1.2 espie 365:
1.21 espie 366: for (;;) {
367: Buf_Reset(linebuf);
1.22 espie 368: c = read_char();
1.21 espie 369: /* Strip leading spaces, fold on '\n' */
370: if (c == ' ') {
371: do {
1.22 espie 372: c = read_char();
1.21 espie 373: } while (c == ' ' || c == '\t');
374: while (c == '\\') {
1.22 espie 375: c = read_char();
1.21 espie 376: if (c == '\n') {
1.26 espie 377: current->origin.lineno++;
1.21 espie 378: do {
1.22 espie 379: c = read_char();
1.21 espie 380: } while (c == ' ' || c == '\t');
381: } else {
382: Buf_AddChar(linebuf, '\\');
383: if (c == '\\') {
384: Buf_AddChar(linebuf, '\\');
1.22 espie 385: c = read_char();
1.21 espie 386: }
387: if (c == EOF)
388: return '\n';
389: else
390: return c;
391: }
392: }
393: assert(c != '\t');
394: }
395: if (c == '#')
1.22 espie 396: c = skip_to_end_of_line();
1.21 espie 397: /* Almost identical to spaces, except this occurs after
398: * comments have been taken care of, and we keep the tab
399: * itself. */
400: if (c == '\t') {
401: Buf_AddChar(linebuf, '\t');
402: do {
1.22 espie 403: c = read_char();
1.21 espie 404: } while (c == ' ' || c == '\t');
405: while (c == '\\') {
1.22 espie 406: c = read_char();
1.21 espie 407: if (c == '\n') {
1.26 espie 408: current->origin.lineno++;
1.21 espie 409: do {
1.22 espie 410: c = read_char();
1.21 espie 411: } while (c == ' ' || c == '\t');
412: } else {
413: Buf_AddChar(linebuf, '\\');
414: if (c == '\\') {
415: Buf_AddChar(linebuf, '\\');
1.22 espie 416: c = read_char();
1.21 espie 417: }
418: if (c == EOF)
419: return '\n';
420: else
421: return c;
422: }
423: }
1.2 espie 424: }
1.21 espie 425: if (c == '\n')
1.26 espie 426: current->origin.lineno++;
1.21 espie 427: else
1.6 espie 428: return c;
1.2 espie 429: }
1.6 espie 430: }
1.2 espie 431:
1.9 espie 432: /* Parse_ReadNormalLine removes beginning and trailing blanks (but keeps
1.18 espie 433: * the first tab), handles escaped newlines, and skips over uninteresting
1.20 espie 434: * lines.
1.6 espie 435: *
1.18 espie 436: * The line number is incremented, which implies that continuation
437: * lines are numbered with the last line number (we could do better, at a
1.9 espie 438: * price).
1.6 espie 439: *
1.9 espie 440: * Trivial comments are also removed, but we can't do more, as
441: * we don't know which lines are shell commands or not. */
1.6 espie 442: char *
1.18 espie 443: Parse_ReadNormalLine(Buffer linebuf)
1.6 espie 444: {
1.21 espie 445: int c; /* the current character */
1.6 espie 446:
1.22 espie 447: c = skip_empty_lines_and_read_char(linebuf);
1.6 espie 448:
1.21 espie 449: if (c == EOF)
450: return NULL;
451: else {
1.22 espie 452: read_logical_line(linebuf, c);
1.21 espie 453: Buf_KillTrailingSpaces(linebuf);
454: return Buf_Retrieve(linebuf);
455: }
1.1 espie 456: }
457:
458: unsigned long
1.18 espie 459: Parse_Getlineno(void)
1.1 espie 460: {
1.26 espie 461: return current ? current->origin.lineno : 0;
1.1 espie 462: }
463:
464: const char *
1.18 espie 465: Parse_Getfilename(void)
1.1 espie 466: {
1.26 espie 467: return current ? current->origin.fname : NULL;
468: }
469:
470: void
1.28 espie 471: Parse_SetLocation(Location *origin)
472: {
473: post_parse = origin;
474: }
475:
476: void
1.26 espie 477: Parse_FillLocation(Location *origin)
478: {
1.28 espie 479: if (post_parse) {
480: *origin = *post_parse;
481: } else {
482: origin->lineno = Parse_Getlineno();
483: origin->fname = Parse_Getfilename();
484: }
1.1 espie 485: }
486:
487: void
1.18 espie 488: Parse_ReportErrors(void)
1.1 espie 489: {
1.31 ! espie 490: if (fatal_errors)
1.21 espie 491: exit(1);
1.31 ! espie 492: else
1.21 espie 493: assert(current == NULL);
1.1 espie 494: }