Annotation of docbook2mdoc/parse.c, Revision 1.15
1.15 ! schwarze 1: /* $Id: parse.c,v 1.14 2019/04/05 14:37:36 schwarze Exp $ */
1.1 schwarze 2: /*
3: * Copyright (c) 2014 Kristaps Dzonsons <kristaps@bsd.lv>
4: * Copyright (c) 2019 Ingo Schwarze <schwarze@openbsd.org>
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: *
10: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES
11: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
12: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR
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.
17: */
18: #include <assert.h>
19: #include <ctype.h>
1.6 schwarze 20: #include <stdarg.h>
1.1 schwarze 21: #include <stdio.h>
1.5 schwarze 22: #include <stdlib.h>
1.1 schwarze 23: #include <string.h>
24: #include <unistd.h>
25:
26: #include "node.h"
27: #include "parse.h"
28:
29: /*
30: * The implementation of the DocBook parser.
31: */
32:
1.14 schwarze 33: enum pstate {
34: PARSE_ELEM,
35: PARSE_TAG,
36: PARSE_ARG,
37: PARSE_SQ,
38: PARSE_DQ
39: };
40:
1.1 schwarze 41: /*
42: * Global parse state.
43: * Keep this as simple and small as possible.
44: */
45: struct parse {
46: const char *fname; /* Name of the input file. */
47: struct ptree *tree; /* Complete parse result. */
48: struct pnode *cur; /* Current node in the tree. */
1.5 schwarze 49: enum nodeid ncur; /* Type of the current node. */
50: int line; /* Line number in the input file. */
51: int col; /* Column number in the input file. */
52: int nline; /* Line number of next token. */
53: int ncol; /* Column number of next token. */
1.4 schwarze 54: int del; /* Levels of nested nodes being deleted. */
1.5 schwarze 55: int attr; /* The most recent attribute is valid. */
1.1 schwarze 56: int warn;
57: };
58:
59: struct element {
60: const char *name; /* DocBook element name. */
61: enum nodeid node; /* Node type to generate. */
62: };
63:
64: static const struct element elements[] = {
1.3 schwarze 65: { "acronym", NODE_IGNORE },
1.1 schwarze 66: { "affiliation", NODE_AFFILIATION },
1.4 schwarze 67: { "anchor", NODE_DELETE },
1.1 schwarze 68: { "application", NODE_APPLICATION },
69: { "arg", NODE_ARG },
70: { "author", NODE_AUTHOR },
71: { "authorgroup", NODE_AUTHORGROUP },
72: { "blockquote", NODE_BLOCKQUOTE },
73: { "book", NODE_BOOK },
74: { "bookinfo", NODE_BOOKINFO },
75: { "caution", NODE_CAUTION },
76: { "chapter", NODE_SECTION },
77: { "citerefentry", NODE_CITEREFENTRY },
78: { "citetitle", NODE_CITETITLE },
79: { "cmdsynopsis", NODE_CMDSYNOPSIS },
1.13 schwarze 80: { "code", NODE_LITERAL },
1.1 schwarze 81: { "colspec", NODE_COLSPEC },
82: { "command", NODE_COMMAND },
83: { "constant", NODE_CONSTANT },
1.7 schwarze 84: { "contrib", NODE_CONTRIB },
1.1 schwarze 85: { "copyright", NODE_COPYRIGHT },
86: { "date", NODE_DATE },
87: { "editor", NODE_EDITOR },
88: { "email", NODE_EMAIL },
89: { "emphasis", NODE_EMPHASIS },
90: { "entry", NODE_ENTRY },
91: { "envar", NODE_ENVAR },
1.13 schwarze 92: { "errorname", NODE_ERRORNAME },
1.1 schwarze 93: { "fieldsynopsis", NODE_FIELDSYNOPSIS },
94: { "filename", NODE_FILENAME },
1.7 schwarze 95: { "firstname", NODE_PERSONNAME },
1.1 schwarze 96: { "firstterm", NODE_FIRSTTERM },
97: { "footnote", NODE_FOOTNOTE },
98: { "funcdef", NODE_FUNCDEF },
99: { "funcprototype", NODE_FUNCPROTOTYPE },
100: { "funcsynopsis", NODE_FUNCSYNOPSIS },
101: { "funcsynopsisinfo", NODE_FUNCSYNOPSISINFO },
102: { "function", NODE_FUNCTION },
103: { "glossterm", NODE_GLOSSTERM },
104: { "group", NODE_GROUP },
105: { "holder", NODE_HOLDER },
106: { "index", NODE_INDEX },
1.4 schwarze 107: { "indexterm", NODE_DELETE },
1.1 schwarze 108: { "info", NODE_INFO },
109: { "informalequation", NODE_INFORMALEQUATION },
1.11 schwarze 110: { "informaltable", NODE_TABLE },
1.1 schwarze 111: { "inlineequation", NODE_INLINEEQUATION },
112: { "itemizedlist", NODE_ITEMIZEDLIST },
113: { "keysym", NODE_KEYSYM },
114: { "legalnotice", NODE_LEGALNOTICE },
115: { "link", NODE_LINK },
116: { "listitem", NODE_LISTITEM },
117: { "literal", NODE_LITERAL },
118: { "literallayout", NODE_LITERALLAYOUT },
119: { "manvolnum", NODE_MANVOLNUM },
120: { "member", NODE_MEMBER },
121: { "mml:math", NODE_MML_MATH },
122: { "mml:mfenced", NODE_MML_MFENCED },
123: { "mml:mfrac", NODE_MML_MFRAC },
124: { "mml:mi", NODE_MML_MI },
125: { "mml:mn", NODE_MML_MN },
126: { "mml:mo", NODE_MML_MO },
127: { "mml:mrow", NODE_MML_MROW },
128: { "mml:msub", NODE_MML_MSUB },
129: { "mml:msup", NODE_MML_MSUP },
130: { "modifier", NODE_MODIFIER },
131: { "note", NODE_NOTE },
132: { "option", NODE_OPTION },
133: { "orderedlist", NODE_ORDEREDLIST },
134: { "orgname", NODE_ORGNAME },
1.7 schwarze 135: { "othername", NODE_PERSONNAME },
1.1 schwarze 136: { "para", NODE_PARA },
137: { "paramdef", NODE_PARAMDEF },
138: { "parameter", NODE_PARAMETER },
139: { "part", NODE_SECTION },
140: { "personname", NODE_PERSONNAME },
1.3 schwarze 141: { "phrase", NODE_IGNORE },
1.1 schwarze 142: { "preface", NODE_PREFACE },
1.4 schwarze 143: { "primary", NODE_DELETE },
1.1 schwarze 144: { "programlisting", NODE_PROGRAMLISTING },
145: { "prompt", NODE_PROMPT },
146: { "quote", NODE_QUOTE },
147: { "refclass", NODE_REFCLASS },
148: { "refdescriptor", NODE_REFDESCRIPTOR },
149: { "refentry", NODE_REFENTRY },
150: { "refentryinfo", NODE_REFENTRYINFO },
151: { "refentrytitle", NODE_REFENTRYTITLE },
152: { "refmeta", NODE_REFMETA },
153: { "refmetainfo", NODE_REFMETAINFO },
154: { "refmiscinfo", NODE_REFMISCINFO },
155: { "refname", NODE_REFNAME },
156: { "refnamediv", NODE_REFNAMEDIV },
157: { "refpurpose", NODE_REFPURPOSE },
158: { "refsect1", NODE_SECTION },
159: { "refsect2", NODE_SECTION },
160: { "refsect3", NODE_SECTION },
161: { "refsection", NODE_SECTION },
162: { "refsynopsisdiv", NODE_REFSYNOPSISDIV },
163: { "releaseinfo", NODE_RELEASEINFO },
164: { "replaceable", NODE_REPLACEABLE },
165: { "row", NODE_ROW },
166: { "sbr", NODE_SBR },
167: { "screen", NODE_SCREEN },
1.4 schwarze 168: { "secondary", NODE_DELETE },
1.1 schwarze 169: { "sect1", NODE_SECTION },
170: { "sect2", NODE_SECTION },
171: { "section", NODE_SECTION },
172: { "sgmltag", NODE_SGMLTAG },
1.15 ! schwarze 173: { "simpara", NODE_PARA },
1.1 schwarze 174: { "simplelist", NODE_SIMPLELIST },
175: { "spanspec", NODE_SPANSPEC },
1.13 schwarze 176: { "structfield", NODE_PARAMETER },
177: { "structname", NODE_TYPE },
1.1 schwarze 178: { "subtitle", NODE_SUBTITLE },
1.7 schwarze 179: { "surname", NODE_PERSONNAME },
1.12 schwarze 180: { "symbol", NODE_CONSTANT },
1.1 schwarze 181: { "synopsis", NODE_SYNOPSIS },
182: { "table", NODE_TABLE },
183: { "tbody", NODE_TBODY },
184: { "term", NODE_TERM },
185: { "tfoot", NODE_TFOOT },
186: { "tgroup", NODE_TGROUP },
187: { "thead", NODE_THEAD },
188: { "tip", NODE_TIP },
189: { "title", NODE_TITLE },
1.3 schwarze 190: { "trademark", NODE_IGNORE },
1.1 schwarze 191: { "type", NODE_TYPE },
192: { "ulink", NODE_ULINK },
1.13 schwarze 193: { "userinput", NODE_LITERAL },
1.1 schwarze 194: { "variablelist", NODE_VARIABLELIST },
195: { "varlistentry", NODE_VARLISTENTRY },
196: { "varname", NODE_VARNAME },
197: { "warning", NODE_WARNING },
198: { "wordasword", NODE_WORDASWORD },
1.4 schwarze 199: { "xi:include", NODE_DELETE_WARN },
1.1 schwarze 200: { "year", NODE_YEAR },
1.5 schwarze 201: { NULL, NODE_IGNORE }
1.1 schwarze 202: };
203:
1.9 schwarze 204: struct entity {
205: const char *name;
206: const char *roff;
207: };
208:
209: /*
210: * XML character entity references found in the wild.
211: * Those that don't have an exact mandoc_char(7) representation
212: * are approximated, and the desired codepoint is given as a comment.
213: * Encoding them as \\[u...] would leave -Tascii out in the cold.
214: */
215: static const struct entity entities[] = {
216: { "alpha", "\\(*a" },
217: { "amp", "&" },
218: { "apos", "'" },
219: { "auml", "\\(:a" },
220: { "beta", "\\(*b" },
221: { "circ", "^" }, /* U+02C6 */
222: { "copy", "\\(co" },
223: { "dagger", "\\(dg" },
224: { "Delta", "\\(*D" },
225: { "eacute", "\\('e" },
226: { "emsp", "\\ " }, /* U+2003 */
227: { "gt", ">" },
228: { "hairsp", "\\^" },
229: { "kappa", "\\(*k" },
230: { "larr", "\\(<-" },
231: { "ldquo", "\\(lq" },
232: { "le", "\\(<=" },
233: { "lowbar", "_" },
234: { "lsqb", "[" },
235: { "lt", "<" },
236: { "mdash", "\\(em" },
237: { "minus", "\\-" },
238: { "ndash", "\\(en" },
239: { "nbsp", "\\ " },
240: { "num", "#" },
241: { "oslash", "\\(/o" },
242: { "ouml", "\\(:o" },
243: { "percnt", "%" },
244: { "quot", "\\(dq" },
245: { "rarr", "\\(->" },
246: { "rArr", "\\(rA" },
247: { "rdquo", "\\(rq" },
248: { "reg", "\\(rg" },
249: { "rho", "\\(*r" },
250: { "rsqb", "]" },
251: { "sigma", "\\(*s" },
252: { "shy", "\\&" }, /* U+00AD */
253: { "tau", "\\(*t" },
254: { "tilde", "\\[u02DC]" },
255: { "times", "\\[tmu]" },
256: { "uuml", "\\(:u" },
257: { NULL, NULL }
258: };
259:
1.6 schwarze 260: static void
261: error_msg(struct parse *p, const char *fmt, ...)
262: {
263: va_list ap;
264:
265: fprintf(stderr, "%s:%d:%d: ", p->fname, p->line, p->col);
266: va_start(ap, fmt);
267: vfprintf(stderr, fmt, ap);
268: va_end(ap);
269: fputc('\n', stderr);
270: p->tree->flags |= TREE_FAIL;
271: }
272:
273: static void
274: warn_msg(struct parse *p, const char *fmt, ...)
275: {
276: va_list ap;
277:
278: if (p->warn == 0)
279: return;
280:
281: fprintf(stderr, "%s:%d:%d: warning: ", p->fname, p->line, p->col);
282: va_start(ap, fmt);
283: vfprintf(stderr, fmt, ap);
284: va_end(ap);
285: fputc('\n', stderr);
286: }
287:
1.1 schwarze 288: /*
289: * Process a string of characters.
290: * If a text node is already open, append to it.
291: * Otherwise, create a new one as a child of the current node.
292: */
293: static void
1.5 schwarze 294: xml_char(struct parse *ps, const char *p, int sz)
1.1 schwarze 295: {
296: struct pnode *dat;
297:
1.5 schwarze 298: if (ps->del > 0)
1.1 schwarze 299: return;
300:
1.5 schwarze 301: if (ps->cur == NULL) {
1.6 schwarze 302: error_msg(ps, "discarding text before document: %.*s", sz, p);
1.5 schwarze 303: return;
304: }
305:
1.1 schwarze 306: if (ps->cur->node != NODE_TEXT) {
307: if ((dat = calloc(1, sizeof(*dat))) == NULL) {
308: perror(NULL);
309: exit(1);
310: }
311: dat->node = NODE_TEXT;
312: dat->parent = ps->cur;
313: TAILQ_INIT(&dat->childq);
314: TAILQ_INIT(&dat->attrq);
315: TAILQ_INSERT_TAIL(&ps->cur->childq, dat, child);
316: ps->cur = dat;
317: }
318:
1.5 schwarze 319: if (ps->tree->flags & TREE_CLOSED &&
1.6 schwarze 320: ps->cur->parent == ps->tree->root)
321: warn_msg(ps, "text after end of document: %.*s", sz, p);
1.5 schwarze 322:
1.1 schwarze 323: /* Append to the current text node. */
324:
325: assert(sz >= 0);
326: ps->cur->b = realloc(ps->cur->b, ps->cur->bsz + sz + 1);
327: if (ps->cur->b == NULL) {
328: perror(NULL);
329: exit(1);
330: }
331: memcpy(ps->cur->b + ps->cur->bsz, p, sz);
332: ps->cur->bsz += sz;
333: ps->cur->b[ps->cur->bsz] = '\0';
334: ps->cur->real = ps->cur->b;
335: }
336:
337: static void
338: pnode_trim(struct pnode *pn)
339: {
340: assert(pn->node == NODE_TEXT);
341: for (; pn->bsz > 0; pn->b[--pn->bsz] = '\0')
342: if (isspace((unsigned char)pn->b[pn->bsz - 1]) == 0)
343: break;
344: }
345:
1.9 schwarze 346: static void
347: xml_entity(struct parse *p, const char *name)
348: {
349: const struct entity *entity;
350: struct pnode *dat;
351:
352: if (p->del > 0)
353: return;
354:
355: if (p->cur == NULL) {
356: error_msg(p, "discarding entity before document: &%s;", name);
357: return;
358: }
359:
360: /* Close out the text node, if there is one. */
361: if (p->cur->node == NODE_TEXT) {
362: pnode_trim(p->cur);
363: p->cur = p->cur->parent;
364: }
365:
366: if (p->tree->flags & TREE_CLOSED && p->cur == p->tree->root)
367: warn_msg(p, "entity after end of document: &%s;", name);
368:
369: for (entity = entities; entity->name != NULL; entity++)
370: if (strcmp(name, entity->name) == 0)
371: break;
372:
373: if (entity->roff == NULL) {
374: error_msg(p, "unknown entity &%s;", name);
375: return;
376: }
377:
378: /* Create, append, and close out an entity node. */
379: if ((dat = calloc(1, sizeof(*dat))) == NULL ||
380: (dat->b = dat->real = strdup(entity->roff)) == NULL) {
381: perror(NULL);
382: exit(1);
383: }
384: dat->node = NODE_ESCAPE;
385: dat->bsz = strlen(dat->b);
386: dat->parent = p->cur;
387: TAILQ_INIT(&dat->childq);
388: TAILQ_INIT(&dat->attrq);
389: TAILQ_INSERT_TAIL(&p->cur->childq, dat, child);
390: }
391:
1.1 schwarze 392: /*
393: * Begin an element.
394: */
395: static void
1.5 schwarze 396: xml_elem_start(struct parse *ps, const char *name)
1.1 schwarze 397: {
1.5 schwarze 398: const struct element *elem;
399: struct pnode *dat;
1.1 schwarze 400:
1.5 schwarze 401: if (*name == '!' || *name == '?')
1.1 schwarze 402: return;
403:
1.4 schwarze 404: /*
405: * An ancestor is excluded from the tree;
406: * keep track of the number of levels excluded.
407: */
408: if (ps->del > 0) {
409: ps->del++;
410: return;
411: }
412:
1.1 schwarze 413: /* Close out the text node, if there is one. */
414: if (ps->cur != NULL && ps->cur->node == NODE_TEXT) {
415: pnode_trim(ps->cur);
416: ps->cur = ps->cur->parent;
417: }
418:
419: for (elem = elements; elem->name != NULL; elem++)
420: if (strcmp(elem->name, name) == 0)
421: break;
422:
1.6 schwarze 423: if (elem->name == NULL)
424: error_msg(ps, "unknown element <%s>", name);
425:
1.5 schwarze 426: ps->ncur = elem->node;
1.1 schwarze 427:
1.5 schwarze 428: switch (ps->ncur) {
1.4 schwarze 429: case NODE_DELETE_WARN:
1.6 schwarze 430: warn_msg(ps, "skipping element <%s>", name);
1.2 schwarze 431: /* FALLTHROUGH */
1.4 schwarze 432: case NODE_DELETE:
433: ps->del = 1;
434: /* FALLTHROUGH */
1.2 schwarze 435: case NODE_IGNORE:
436: return;
437: case NODE_INLINEEQUATION:
1.1 schwarze 438: ps->tree->flags |= TREE_EQN;
1.2 schwarze 439: break;
440: default:
441: break;
442: }
1.1 schwarze 443:
1.6 schwarze 444: if (ps->tree->flags & TREE_CLOSED && ps->cur->parent == NULL)
445: warn_msg(ps, "element after end of document: <%s>", name);
1.5 schwarze 446:
1.1 schwarze 447: if ((dat = calloc(1, sizeof(*dat))) == NULL) {
448: perror(NULL);
449: exit(1);
450: }
451: dat->node = elem->node;
452: dat->parent = ps->cur;
453: TAILQ_INIT(&dat->childq);
454: TAILQ_INIT(&dat->attrq);
455:
456: if (ps->cur != NULL)
457: TAILQ_INSERT_TAIL(&ps->cur->childq, dat, child);
458:
459: ps->cur = dat;
460: if (ps->tree->root == NULL)
461: ps->tree->root = dat;
1.5 schwarze 462: }
463:
464: static void
465: xml_attrkey(struct parse *ps, const char *name)
466: {
467: struct pattr *attr;
468: enum attrkey key;
1.1 schwarze 469:
1.5 schwarze 470: if (ps->del > 0 || *name == '\0')
471: return;
472: if ((key = attrkey_parse(name)) == ATTRKEY__MAX) {
473: ps->attr = 0;
474: return;
475: }
476: if ((attr = calloc(1, sizeof(*attr))) == NULL) {
477: perror(NULL);
478: exit(1);
479: }
480: attr->key = key;
481: attr->val = ATTRVAL__MAX;
482: attr->rawval = NULL;
483: TAILQ_INSERT_TAIL(&ps->cur->attrq, attr, child);
484: ps->attr = 1;
485: }
486:
487: static void
488: xml_attrval(struct parse *ps, const char *name)
489: {
490: struct pattr *attr;
491:
492: if (ps->del > 0 || ps->attr == 0)
493: return;
494: if ((attr = TAILQ_LAST(&ps->cur->attrq, pattrq)) == NULL)
495: return;
496: if ((attr->val = attrval_parse(name)) == ATTRVAL__MAX &&
497: (attr->rawval = strdup(name)) == NULL) {
498: perror(NULL);
499: exit(1);
1.1 schwarze 500: }
501: }
502:
503: /*
504: * Roll up the parse tree.
505: * If we're at a text node, roll that one up first.
506: */
507: static void
1.5 schwarze 508: xml_elem_end(struct parse *ps, const char *name)
1.1 schwarze 509: {
1.5 schwarze 510: const struct element *elem;
511: enum nodeid node;
1.1 schwarze 512:
1.4 schwarze 513: /*
514: * An ancestor is excluded from the tree;
515: * keep track of the number of levels excluded.
516: */
517: if (ps->del > 1) {
518: ps->del--;
519: return;
520: }
521:
1.1 schwarze 522: /* Close out the text node, if there is one. */
1.5 schwarze 523: if (ps->del == 0 && ps->cur != NULL && ps->cur->node == NODE_TEXT) {
1.1 schwarze 524: pnode_trim(ps->cur);
525: ps->cur = ps->cur->parent;
526: }
1.2 schwarze 527:
1.5 schwarze 528: if (name != NULL) {
529: for (elem = elements; elem->name != NULL; elem++)
530: if (strcmp(elem->name, name) == 0)
531: break;
532: node = elem->node;
533: } else
534: node = ps->ncur;
1.2 schwarze 535:
1.5 schwarze 536: switch (node) {
1.4 schwarze 537: case NODE_DELETE_WARN:
538: case NODE_DELETE:
1.5 schwarze 539: if (ps->del > 0)
540: ps->del--;
1.4 schwarze 541: break;
1.2 schwarze 542: case NODE_IGNORE:
543: break;
544: default:
1.5 schwarze 545: if (ps->cur == NULL || node != ps->cur->node) {
1.6 schwarze 546: warn_msg(ps, "element not open: </%s>", name);
1.5 schwarze 547: break;
548: }
549:
550: /*
551: * Refrain from actually closing the document element.
552: * If no more content follows, no harm is done, but if
553: * some content still follows, simply processing it is
554: * obviously better than discarding it or crashing.
555: */
556:
557: if (ps->cur->parent == NULL)
558: ps->tree->flags |= TREE_CLOSED;
559: else
560: ps->cur = ps->cur->parent;
1.4 schwarze 561: break;
1.2 schwarze 562: }
1.4 schwarze 563: assert(ps->del == 0);
1.1 schwarze 564: }
565:
566: struct parse *
567: parse_alloc(int warn)
568: {
569: struct parse *p;
570:
571: if ((p = calloc(1, sizeof(*p))) == NULL)
572: return NULL;
573:
574: if ((p->tree = calloc(1, sizeof(*p->tree))) == NULL) {
575: free(p);
576: return NULL;
577: }
578: p->warn = warn;
579: return p;
580: }
581:
582: void
583: parse_free(struct parse *p)
584: {
585: if (p == NULL)
586: return;
587: if (p->tree != NULL) {
588: pnode_unlink(p->tree->root);
589: free(p->tree);
590: }
591: free(p);
592: }
593:
1.14 schwarze 594: static void
595: increment(struct parse *p, char *b, size_t *pend, int refill)
596: {
597: if (refill) {
598: if (b[*pend] == '\n') {
599: p->nline++;
600: p->ncol = 1;
601: } else
602: p->ncol++;
603: }
604: ++*pend;
605: }
606:
1.5 schwarze 607: /*
608: * Advance the pend pointer to the next character in the charset.
609: * If the charset starts with a space, it stands for any whitespace.
610: * Update the new input file position, used for messages.
611: * Do not overrun the buffer b of length rlen.
612: * When reaching the end, NUL-terminate the buffer and return 1;
613: * otherwise, return 0.
614: */
615: static int
616: advance(struct parse *p, char *b, size_t rlen, size_t *pend,
1.14 schwarze 617: const char *charset, int refill)
1.5 schwarze 618: {
619: int space;
620:
621: if (*charset == ' ') {
622: space = 1;
623: charset++;
624: } else
625: space = 0;
626:
1.14 schwarze 627: if (refill) {
628: p->nline = p->line;
629: p->ncol = p->col;
630: }
1.5 schwarze 631: while (*pend < rlen) {
632: if (space && isspace((unsigned char)b[*pend]))
633: break;
634: if (strchr(charset, b[*pend]) != NULL)
635: break;
1.14 schwarze 636: increment(p, b, pend, refill);
1.5 schwarze 637: }
638: if (*pend == rlen) {
639: b[rlen] = '\0';
1.14 schwarze 640: return refill;
1.5 schwarze 641: } else
642: return 0;
643: }
644:
1.14 schwarze 645: size_t
646: parse_string(struct parse *p, char *b, size_t rlen,
647: enum pstate *pstate, int refill)
648: {
649: char *cp;
650: size_t poff; /* Parse offset in b[]. */
651: size_t pend; /* Offset of the end of the current word. */
652: int elem_end;
653:
654: pend = 0;
655: for (;;) {
656:
657: /* Proceed to the next token, skipping whitespace. */
658:
659: if (refill) {
660: p->line = p->nline;
661: p->col = p->ncol;
662: }
663: if ((poff = pend) == rlen)
664: break;
665: if (isspace((unsigned char)b[pend])) {
666: increment(p, b, &pend, refill);
667: continue;
668: }
669:
670: /*
671: * The following four cases (ARG, TAG, and starting an
672: * entity or a tag) all parse a word or quoted string.
673: * If that extends beyond the read buffer and the last
674: * read(2) still got data, they all break out of the
675: * token loop to request more data from the read loop.
676: *
677: * Also, three of them detect self-closing tags, those
678: * ending with "/>", setting the flag elem_end and
679: * calling xml_elem_end() at the very end, after
680: * handling the attribute value, attribute name, or
681: * tag name, respectively.
682: */
683:
684: /* Parse an attribute value. */
685:
686: if (*pstate >= PARSE_ARG) {
687: if (*pstate == PARSE_ARG &&
688: (b[pend] == '\'' || b[pend] == '"')) {
689: *pstate = b[pend] == '"' ?
690: PARSE_DQ : PARSE_SQ;
691: increment(p, b, &pend, refill);
692: continue;
693: }
694: if (advance(p, b, rlen, &pend,
695: *pstate == PARSE_DQ ? "\"" :
696: *pstate == PARSE_SQ ? "'" : " >", refill))
697: break;
698: *pstate = PARSE_TAG;
699: elem_end = 0;
700: if (b[pend] == '>') {
701: *pstate = PARSE_ELEM;
702: if (pend > 0 && b[pend - 1] == '/') {
703: b[pend - 1] = '\0';
704: elem_end = 1;
705: }
706: }
707: b[pend] = '\0';
708: if (pend < rlen)
709: increment(p, b, &pend, refill);
710: xml_attrval(p, b + poff);
711: if (elem_end)
712: xml_elem_end(p, NULL);
713:
714: /* Look for an attribute name. */
715:
716: } else if (*pstate == PARSE_TAG) {
717: if (advance(p, b, rlen, &pend, " =>", refill))
718: break;
719: elem_end = 0;
720: switch (b[pend]) {
721: case '>':
722: *pstate = PARSE_ELEM;
723: if (pend > 0 && b[pend - 1] == '/') {
724: b[pend - 1] = '\0';
725: elem_end = 1;
726: }
727: break;
728: case '=':
729: *pstate = PARSE_ARG;
730: break;
731: default:
732: break;
733: }
734: b[pend] = '\0';
735: if (pend < rlen)
736: increment(p, b, &pend, refill);
737: xml_attrkey(p, b + poff);
738: if (elem_end)
739: xml_elem_end(p, NULL);
740:
741: /* Begin an opening or closing tag. */
742:
743: } else if (b[poff] == '<') {
744: if (advance(p, b, rlen, &pend, " >", refill))
745: break;
746: if (pend > poff + 3 &&
747: strncmp(b + poff, "<!--", 4) == 0) {
748:
749: /* Skip a comment. */
750:
751: cp = strstr(b + pend - 2, "-->");
752: if (cp == NULL) {
753: if (refill)
754: break;
755: cp = b + rlen;
756: } else
757: cp += 3;
758: while (b + pend < cp)
759: increment(p, b, &pend, refill);
760: continue;
761: }
762: elem_end = 0;
763: if (b[pend] != '>')
764: *pstate = PARSE_TAG;
765: else if (pend > 0 && b[pend - 1] == '/') {
766: b[pend - 1] = '\0';
767: elem_end = 1;
768: }
769: b[pend] = '\0';
770: if (pend < rlen)
771: increment(p, b, &pend, refill);
772: if (b[++poff] == '/') {
773: elem_end = 1;
774: poff++;
775: } else
776: xml_elem_start(p, b + poff);
777: if (elem_end)
778: xml_elem_end(p, b + poff);
779:
780: /* Process an entity. */
781:
782: } else if (b[poff] == '&') {
783: if (advance(p, b, rlen, &pend, ";", refill))
784: break;
785: b[pend] = '\0';
786: if (pend < rlen)
787: increment(p, b, &pend, refill);
788: xml_entity(p, b + poff + 1);
789:
790: /* Process text up to the next tag, entity, or EOL. */
791:
792: } else {
793: advance(p, b, rlen, &pend, "<&", refill);
794: xml_char(p, b + poff, pend - poff);
795: }
796: }
797: return poff;
798: }
799:
1.1 schwarze 800: struct ptree *
801: parse_file(struct parse *p, int fd, const char *fname)
802: {
803: char b[4096];
1.5 schwarze 804: ssize_t rsz; /* Return value from read(2). */
1.14 schwarze 805: size_t rlen; /* Number of bytes in b[]. */
1.5 schwarze 806: size_t poff; /* Parse offset in b[]. */
1.14 schwarze 807: enum pstate pstate;
1.1 schwarze 808:
809: p->fname = fname;
1.5 schwarze 810: p->nline = 1;
811: p->ncol = 1;
1.14 schwarze 812: pstate = PARSE_ELEM;
1.5 schwarze 813: rlen = 0;
814:
815: /*
816: * Read loop.
817: *
1.14 schwarze 818: * If the previous token was incomplete and asked for more
819: * input, we have to enter the read loop once more even on EOF.
1.5 schwarze 820: * Once rsz is 0, incomplete tokens will no longer ask
821: * for more input but instead use whatever there is,
822: * and then exit the read loop.
823: * The minus one on the size limit for read(2) is needed
824: * such that advance() can set b[rlen] to NUL when needed.
825: */
826:
1.14 schwarze 827: while ((rsz = read(fd, b + rlen, sizeof(b) - rlen - 1)) >= 0 &&
828: (rlen += rsz) > 0) {
829: poff = parse_string(p, b, rlen, &pstate, rsz > 0);
1.5 schwarze 830: /* Buffer exhausted; shift left and re-fill. */
831: assert(poff > 0);
832: rlen -= poff;
1.14 schwarze 833: memmove(b, b + poff, rlen);
1.5 schwarze 834: }
835: if (rsz < 0) {
836: perror(fname);
837: p->tree->flags |= TREE_FAIL;
838: }
839: if (p->cur != NULL && p->cur->node == NODE_TEXT) {
840: pnode_trim(p->cur);
841: p->cur = p->cur->parent;
842: }
1.6 schwarze 843: if ((p->tree->flags & TREE_CLOSED) == 0)
844: warn_msg(p, "document not closed");
1.1 schwarze 845: return p->tree;
846: }
CVSweb