Annotation of docbook2mdoc/parse.c, Revision 1.24
1.24 ! schwarze 1: /* $Id: parse.c,v 1.23 2019/04/08 14:37:31 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>
1.24 ! schwarze 20: #include <errno.h>
! 21: #include <fcntl.h>
! 22: #include <libgen.h>
1.6 schwarze 23: #include <stdarg.h>
1.1 schwarze 24: #include <stdio.h>
1.5 schwarze 25: #include <stdlib.h>
1.1 schwarze 26: #include <string.h>
27: #include <unistd.h>
28:
29: #include "node.h"
30: #include "parse.h"
31:
32: /*
33: * The implementation of the DocBook parser.
34: */
35:
1.14 schwarze 36: enum pstate {
37: PARSE_ELEM,
38: PARSE_TAG,
39: PARSE_ARG,
40: PARSE_SQ,
41: PARSE_DQ
42: };
43:
1.1 schwarze 44: /*
45: * Global parse state.
46: * Keep this as simple and small as possible.
47: */
48: struct parse {
49: const char *fname; /* Name of the input file. */
50: struct ptree *tree; /* Complete parse result. */
1.23 schwarze 51: struct pnode *doctype;
1.1 schwarze 52: struct pnode *cur; /* Current node in the tree. */
1.5 schwarze 53: enum nodeid ncur; /* Type of the current node. */
54: int line; /* Line number in the input file. */
55: int col; /* Column number in the input file. */
56: int nline; /* Line number of next token. */
57: int ncol; /* Column number of next token. */
1.4 schwarze 58: int del; /* Levels of nested nodes being deleted. */
1.23 schwarze 59: int flags;
60: #define PFLAG_WARN (1 << 0) /* Print warning messages. */
61: #define PFLAG_SPC (1 << 1) /* Whitespace before the next element. */
62: #define PFLAG_ATTR (1 << 2) /* The most recent attribute is valid. */
63: #define PFLAG_EEND (1 << 3) /* This element is self-closing. */
1.1 schwarze 64: };
65:
66: struct element {
67: const char *name; /* DocBook element name. */
68: enum nodeid node; /* Node type to generate. */
69: };
70:
71: static const struct element elements[] = {
1.3 schwarze 72: { "acronym", NODE_IGNORE },
1.1 schwarze 73: { "affiliation", NODE_AFFILIATION },
1.4 schwarze 74: { "anchor", NODE_DELETE },
1.22 schwarze 75: { "appendix", NODE_APPENDIX },
1.1 schwarze 76: { "application", NODE_APPLICATION },
77: { "arg", NODE_ARG },
1.22 schwarze 78: { "article", NODE_SECTION },
1.1 schwarze 79: { "author", NODE_AUTHOR },
80: { "authorgroup", NODE_AUTHORGROUP },
81: { "blockquote", NODE_BLOCKQUOTE },
1.22 schwarze 82: { "book", NODE_SECTION },
1.1 schwarze 83: { "bookinfo", NODE_BOOKINFO },
84: { "caution", NODE_CAUTION },
85: { "chapter", NODE_SECTION },
86: { "citerefentry", NODE_CITEREFENTRY },
87: { "citetitle", NODE_CITETITLE },
88: { "cmdsynopsis", NODE_CMDSYNOPSIS },
1.13 schwarze 89: { "code", NODE_LITERAL },
1.1 schwarze 90: { "colspec", NODE_COLSPEC },
91: { "command", NODE_COMMAND },
92: { "constant", NODE_CONSTANT },
1.7 schwarze 93: { "contrib", NODE_CONTRIB },
1.1 schwarze 94: { "copyright", NODE_COPYRIGHT },
95: { "date", NODE_DATE },
1.23 schwarze 96: { "!doctype", NODE_DOCTYPE },
97: { "!DOCTYPE", NODE_DOCTYPE },
1.1 schwarze 98: { "editor", NODE_EDITOR },
99: { "email", NODE_EMAIL },
100: { "emphasis", NODE_EMPHASIS },
1.23 schwarze 101: { "!ENTITY", NODE_ENTITY },
1.1 schwarze 102: { "entry", NODE_ENTRY },
103: { "envar", NODE_ENVAR },
1.13 schwarze 104: { "errorname", NODE_ERRORNAME },
1.1 schwarze 105: { "fieldsynopsis", NODE_FIELDSYNOPSIS },
106: { "filename", NODE_FILENAME },
1.7 schwarze 107: { "firstname", NODE_PERSONNAME },
1.1 schwarze 108: { "firstterm", NODE_FIRSTTERM },
109: { "footnote", NODE_FOOTNOTE },
110: { "funcdef", NODE_FUNCDEF },
111: { "funcprototype", NODE_FUNCPROTOTYPE },
112: { "funcsynopsis", NODE_FUNCSYNOPSIS },
113: { "funcsynopsisinfo", NODE_FUNCSYNOPSISINFO },
114: { "function", NODE_FUNCTION },
1.21 schwarze 115: { "glossary", NODE_VARIABLELIST },
116: { "glossdef", NODE_IGNORE },
117: { "glossdiv", NODE_IGNORE },
118: { "glossentry", NODE_VARLISTENTRY },
119: { "glosslist", NODE_VARIABLELIST },
1.1 schwarze 120: { "glossterm", NODE_GLOSSTERM },
121: { "group", NODE_GROUP },
122: { "holder", NODE_HOLDER },
123: { "index", NODE_INDEX },
1.4 schwarze 124: { "indexterm", NODE_DELETE },
1.1 schwarze 125: { "info", NODE_INFO },
126: { "informalequation", NODE_INFORMALEQUATION },
1.11 schwarze 127: { "informaltable", NODE_TABLE },
1.1 schwarze 128: { "inlineequation", NODE_INLINEEQUATION },
129: { "itemizedlist", NODE_ITEMIZEDLIST },
130: { "keysym", NODE_KEYSYM },
131: { "legalnotice", NODE_LEGALNOTICE },
132: { "link", NODE_LINK },
133: { "listitem", NODE_LISTITEM },
134: { "literal", NODE_LITERAL },
135: { "literallayout", NODE_LITERALLAYOUT },
136: { "manvolnum", NODE_MANVOLNUM },
137: { "member", NODE_MEMBER },
138: { "mml:math", NODE_MML_MATH },
139: { "mml:mfenced", NODE_MML_MFENCED },
140: { "mml:mfrac", NODE_MML_MFRAC },
141: { "mml:mi", NODE_MML_MI },
142: { "mml:mn", NODE_MML_MN },
143: { "mml:mo", NODE_MML_MO },
144: { "mml:mrow", NODE_MML_MROW },
145: { "mml:msub", NODE_MML_MSUB },
146: { "mml:msup", NODE_MML_MSUP },
147: { "modifier", NODE_MODIFIER },
148: { "note", NODE_NOTE },
149: { "option", NODE_OPTION },
150: { "orderedlist", NODE_ORDEREDLIST },
151: { "orgname", NODE_ORGNAME },
1.7 schwarze 152: { "othername", NODE_PERSONNAME },
1.1 schwarze 153: { "para", NODE_PARA },
154: { "paramdef", NODE_PARAMDEF },
155: { "parameter", NODE_PARAMETER },
156: { "part", NODE_SECTION },
157: { "personname", NODE_PERSONNAME },
1.3 schwarze 158: { "phrase", NODE_IGNORE },
1.1 schwarze 159: { "preface", NODE_PREFACE },
1.4 schwarze 160: { "primary", NODE_DELETE },
1.1 schwarze 161: { "programlisting", NODE_PROGRAMLISTING },
162: { "prompt", NODE_PROMPT },
163: { "quote", NODE_QUOTE },
164: { "refclass", NODE_REFCLASS },
165: { "refdescriptor", NODE_REFDESCRIPTOR },
166: { "refentry", NODE_REFENTRY },
167: { "refentryinfo", NODE_REFENTRYINFO },
168: { "refentrytitle", NODE_REFENTRYTITLE },
169: { "refmeta", NODE_REFMETA },
170: { "refmetainfo", NODE_REFMETAINFO },
171: { "refmiscinfo", NODE_REFMISCINFO },
172: { "refname", NODE_REFNAME },
173: { "refnamediv", NODE_REFNAMEDIV },
174: { "refpurpose", NODE_REFPURPOSE },
175: { "refsect1", NODE_SECTION },
176: { "refsect2", NODE_SECTION },
177: { "refsect3", NODE_SECTION },
178: { "refsection", NODE_SECTION },
179: { "refsynopsisdiv", NODE_REFSYNOPSISDIV },
180: { "releaseinfo", NODE_RELEASEINFO },
181: { "replaceable", NODE_REPLACEABLE },
182: { "row", NODE_ROW },
183: { "sbr", NODE_SBR },
184: { "screen", NODE_SCREEN },
1.4 schwarze 185: { "secondary", NODE_DELETE },
1.1 schwarze 186: { "sect1", NODE_SECTION },
187: { "sect2", NODE_SECTION },
188: { "section", NODE_SECTION },
189: { "sgmltag", NODE_SGMLTAG },
1.15 schwarze 190: { "simpara", NODE_PARA },
1.1 schwarze 191: { "simplelist", NODE_SIMPLELIST },
192: { "spanspec", NODE_SPANSPEC },
1.13 schwarze 193: { "structfield", NODE_PARAMETER },
194: { "structname", NODE_TYPE },
1.1 schwarze 195: { "subtitle", NODE_SUBTITLE },
1.7 schwarze 196: { "surname", NODE_PERSONNAME },
1.12 schwarze 197: { "symbol", NODE_CONSTANT },
1.1 schwarze 198: { "synopsis", NODE_SYNOPSIS },
199: { "table", NODE_TABLE },
200: { "tbody", NODE_TBODY },
201: { "term", NODE_TERM },
202: { "tfoot", NODE_TFOOT },
203: { "tgroup", NODE_TGROUP },
204: { "thead", NODE_THEAD },
205: { "tip", NODE_TIP },
206: { "title", NODE_TITLE },
1.3 schwarze 207: { "trademark", NODE_IGNORE },
1.1 schwarze 208: { "type", NODE_TYPE },
1.18 schwarze 209: { "ulink", NODE_LINK },
1.13 schwarze 210: { "userinput", NODE_LITERAL },
1.1 schwarze 211: { "variablelist", NODE_VARIABLELIST },
212: { "varlistentry", NODE_VARLISTENTRY },
213: { "varname", NODE_VARNAME },
214: { "warning", NODE_WARNING },
215: { "wordasword", NODE_WORDASWORD },
1.4 schwarze 216: { "xi:include", NODE_DELETE_WARN },
1.1 schwarze 217: { "year", NODE_YEAR },
1.5 schwarze 218: { NULL, NODE_IGNORE }
1.1 schwarze 219: };
220:
1.9 schwarze 221: struct entity {
222: const char *name;
223: const char *roff;
224: };
225:
226: /*
227: * XML character entity references found in the wild.
228: * Those that don't have an exact mandoc_char(7) representation
229: * are approximated, and the desired codepoint is given as a comment.
230: * Encoding them as \\[u...] would leave -Tascii out in the cold.
231: */
232: static const struct entity entities[] = {
233: { "alpha", "\\(*a" },
234: { "amp", "&" },
235: { "apos", "'" },
236: { "auml", "\\(:a" },
237: { "beta", "\\(*b" },
238: { "circ", "^" }, /* U+02C6 */
239: { "copy", "\\(co" },
240: { "dagger", "\\(dg" },
241: { "Delta", "\\(*D" },
242: { "eacute", "\\('e" },
243: { "emsp", "\\ " }, /* U+2003 */
244: { "gt", ">" },
245: { "hairsp", "\\^" },
246: { "kappa", "\\(*k" },
247: { "larr", "\\(<-" },
248: { "ldquo", "\\(lq" },
249: { "le", "\\(<=" },
250: { "lowbar", "_" },
251: { "lsqb", "[" },
252: { "lt", "<" },
253: { "mdash", "\\(em" },
254: { "minus", "\\-" },
255: { "ndash", "\\(en" },
256: { "nbsp", "\\ " },
257: { "num", "#" },
258: { "oslash", "\\(/o" },
259: { "ouml", "\\(:o" },
260: { "percnt", "%" },
261: { "quot", "\\(dq" },
262: { "rarr", "\\(->" },
263: { "rArr", "\\(rA" },
264: { "rdquo", "\\(rq" },
265: { "reg", "\\(rg" },
266: { "rho", "\\(*r" },
267: { "rsqb", "]" },
268: { "sigma", "\\(*s" },
269: { "shy", "\\&" }, /* U+00AD */
270: { "tau", "\\(*t" },
271: { "tilde", "\\[u02DC]" },
272: { "times", "\\[tmu]" },
273: { "uuml", "\\(:u" },
274: { NULL, NULL }
275: };
276:
1.23 schwarze 277: static size_t parse_string(struct parse *, char *, size_t,
278: enum pstate *, int);
1.24 ! schwarze 279: static void parse_fd(struct parse *, int);
1.23 schwarze 280:
281:
1.6 schwarze 282: static void
283: error_msg(struct parse *p, const char *fmt, ...)
284: {
285: va_list ap;
286:
287: fprintf(stderr, "%s:%d:%d: ", p->fname, p->line, p->col);
288: va_start(ap, fmt);
289: vfprintf(stderr, fmt, ap);
290: va_end(ap);
291: fputc('\n', stderr);
292: p->tree->flags |= TREE_FAIL;
293: }
294:
295: static void
296: warn_msg(struct parse *p, const char *fmt, ...)
297: {
298: va_list ap;
299:
1.23 schwarze 300: if ((p->flags & PFLAG_WARN) == 0)
1.6 schwarze 301: return;
302:
303: fprintf(stderr, "%s:%d:%d: warning: ", p->fname, p->line, p->col);
304: va_start(ap, fmt);
305: vfprintf(stderr, fmt, ap);
306: va_end(ap);
307: fputc('\n', stderr);
308: }
309:
1.1 schwarze 310: /*
311: * Process a string of characters.
312: * If a text node is already open, append to it.
313: * Otherwise, create a new one as a child of the current node.
314: */
315: static void
1.5 schwarze 316: xml_char(struct parse *ps, const char *p, int sz)
1.1 schwarze 317: {
318: struct pnode *dat;
1.16 schwarze 319: size_t newsz;
1.1 schwarze 320:
1.5 schwarze 321: if (ps->del > 0)
1.1 schwarze 322: return;
323:
1.5 schwarze 324: if (ps->cur == NULL) {
1.6 schwarze 325: error_msg(ps, "discarding text before document: %.*s", sz, p);
1.5 schwarze 326: return;
327: }
328:
1.1 schwarze 329: if (ps->cur->node != NODE_TEXT) {
330: if ((dat = calloc(1, sizeof(*dat))) == NULL) {
331: perror(NULL);
332: exit(1);
333: }
334: dat->node = NODE_TEXT;
1.23 schwarze 335: dat->spc = (ps->flags & PFLAG_SPC) != 0;
1.1 schwarze 336: dat->parent = ps->cur;
337: TAILQ_INIT(&dat->childq);
338: TAILQ_INIT(&dat->attrq);
339: TAILQ_INSERT_TAIL(&ps->cur->childq, dat, child);
340: ps->cur = dat;
341: }
342:
1.5 schwarze 343: if (ps->tree->flags & TREE_CLOSED &&
1.6 schwarze 344: ps->cur->parent == ps->tree->root)
345: warn_msg(ps, "text after end of document: %.*s", sz, p);
1.5 schwarze 346:
1.1 schwarze 347: /* Append to the current text node. */
348:
349: assert(sz >= 0);
1.23 schwarze 350: newsz = ps->cur->bsz + (ps->cur->bsz && (ps->flags & PFLAG_SPC)) + sz;
1.16 schwarze 351: ps->cur->b = realloc(ps->cur->b, newsz + 1);
1.1 schwarze 352: if (ps->cur->b == NULL) {
353: perror(NULL);
354: exit(1);
355: }
1.23 schwarze 356: if (ps->cur->bsz && (ps->flags & PFLAG_SPC))
1.16 schwarze 357: ps->cur->b[ps->cur->bsz++] = ' ';
1.1 schwarze 358: memcpy(ps->cur->b + ps->cur->bsz, p, sz);
1.16 schwarze 359: ps->cur->b[ps->cur->bsz = newsz] = '\0';
1.1 schwarze 360: ps->cur->real = ps->cur->b;
1.23 schwarze 361: ps->flags &= ~PFLAG_SPC;
1.1 schwarze 362: }
363:
1.16 schwarze 364: /*
365: * Close out the text node and strip trailing whitespace, if one is open.
366: */
1.1 schwarze 367: static void
1.16 schwarze 368: pnode_closetext(struct parse *p)
1.1 schwarze 369: {
1.16 schwarze 370: struct pnode *n;
371:
372: if ((n = p->cur) == NULL || n->node != NODE_TEXT)
373: return;
374: p->cur = n->parent;
375: while (n->bsz > 0 && isspace((unsigned char)n->b[n->bsz - 1])) {
376: n->b[--n->bsz] = '\0';
1.23 schwarze 377: p->flags |= PFLAG_SPC;
1.16 schwarze 378: }
1.1 schwarze 379: }
380:
1.9 schwarze 381: static void
382: xml_entity(struct parse *p, const char *name)
383: {
384: const struct entity *entity;
385: struct pnode *dat;
1.23 schwarze 386: const char *ccp;
387: char *cp;
388: enum pstate pstate;
1.9 schwarze 389:
390: if (p->del > 0)
391: return;
392:
393: if (p->cur == NULL) {
394: error_msg(p, "discarding entity before document: &%s;", name);
395: return;
396: }
397:
1.16 schwarze 398: pnode_closetext(p);
1.9 schwarze 399:
400: if (p->tree->flags & TREE_CLOSED && p->cur == p->tree->root)
401: warn_msg(p, "entity after end of document: &%s;", name);
402:
403: for (entity = entities; entity->name != NULL; entity++)
404: if (strcmp(name, entity->name) == 0)
405: break;
406:
407: if (entity->roff == NULL) {
1.23 schwarze 408: if (p->doctype != NULL) {
409: TAILQ_FOREACH(dat, &p->doctype->childq, child) {
410: if ((ccp = pnode_getattr_raw(dat,
411: ATTRKEY_NAME, NULL)) == NULL ||
412: strcmp(ccp, name) != 0 ||
413: (ccp = pnode_getattr_raw(dat,
414: ATTRKEY_DEFINITION, NULL)) == NULL)
415: continue;
416: if ((cp = strdup(ccp)) == NULL) {
417: perror(NULL);
418: exit(1);
419: }
420: pstate = PARSE_ELEM;
421: parse_string(p, cp, strlen(cp), &pstate, 0);
422: p->flags &= ~PFLAG_SPC;
423: free(cp);
424: return;
425: }
426: }
1.9 schwarze 427: error_msg(p, "unknown entity &%s;", name);
428: return;
429: }
430:
431: /* Create, append, and close out an entity node. */
432: if ((dat = calloc(1, sizeof(*dat))) == NULL ||
433: (dat->b = dat->real = strdup(entity->roff)) == NULL) {
434: perror(NULL);
435: exit(1);
436: }
437: dat->node = NODE_ESCAPE;
438: dat->bsz = strlen(dat->b);
1.23 schwarze 439: dat->spc = (p->flags & PFLAG_SPC) != 0;
1.9 schwarze 440: dat->parent = p->cur;
441: TAILQ_INIT(&dat->childq);
442: TAILQ_INIT(&dat->attrq);
443: TAILQ_INSERT_TAIL(&p->cur->childq, dat, child);
1.23 schwarze 444: p->flags &= ~PFLAG_SPC;
1.9 schwarze 445: }
446:
1.1 schwarze 447: /*
448: * Begin an element.
449: */
450: static void
1.5 schwarze 451: xml_elem_start(struct parse *ps, const char *name)
1.1 schwarze 452: {
1.5 schwarze 453: const struct element *elem;
454: struct pnode *dat;
1.1 schwarze 455:
1.4 schwarze 456: /*
457: * An ancestor is excluded from the tree;
458: * keep track of the number of levels excluded.
459: */
460: if (ps->del > 0) {
1.23 schwarze 461: if (*name != '!' && *name != '?')
462: ps->del++;
1.4 schwarze 463: return;
464: }
465:
1.16 schwarze 466: pnode_closetext(ps);
1.1 schwarze 467:
468: for (elem = elements; elem->name != NULL; elem++)
469: if (strcmp(elem->name, name) == 0)
470: break;
471:
1.23 schwarze 472: if (elem->name == NULL) {
473: if (*name == '!' || *name == '?')
474: return;
1.6 schwarze 475: error_msg(ps, "unknown element <%s>", name);
1.23 schwarze 476: }
1.6 schwarze 477:
1.5 schwarze 478: ps->ncur = elem->node;
1.1 schwarze 479:
1.5 schwarze 480: switch (ps->ncur) {
1.4 schwarze 481: case NODE_DELETE_WARN:
1.6 schwarze 482: warn_msg(ps, "skipping element <%s>", name);
1.2 schwarze 483: /* FALLTHROUGH */
1.4 schwarze 484: case NODE_DELETE:
485: ps->del = 1;
486: /* FALLTHROUGH */
1.2 schwarze 487: case NODE_IGNORE:
488: return;
489: case NODE_INLINEEQUATION:
1.1 schwarze 490: ps->tree->flags |= TREE_EQN;
1.2 schwarze 491: break;
492: default:
493: break;
494: }
1.1 schwarze 495:
1.6 schwarze 496: if (ps->tree->flags & TREE_CLOSED && ps->cur->parent == NULL)
497: warn_msg(ps, "element after end of document: <%s>", name);
1.5 schwarze 498:
1.1 schwarze 499: if ((dat = calloc(1, sizeof(*dat))) == NULL) {
500: perror(NULL);
501: exit(1);
502: }
1.17 schwarze 503:
504: /*
505: * Nodes that begin a new macro or request line or start by
506: * printing text always want whitespace before themselves.
507: */
508:
509: switch (dat->node = elem->node) {
1.23 schwarze 510: case NODE_DOCTYPE:
511: case NODE_ENTITY:
512: case NODE_SBR:
513: ps->flags |= PFLAG_EEND;
514: /* FALLTHROUGH */
1.22 schwarze 515: case NODE_APPENDIX:
1.17 schwarze 516: case NODE_AUTHORGROUP:
1.20 schwarze 517: case NODE_BLOCKQUOTE:
1.17 schwarze 518: case NODE_BOOKINFO:
519: case NODE_CAUTION:
520: case NODE_EDITOR:
521: case NODE_ENTRY:
522: case NODE_FUNCDEF:
523: case NODE_FUNCPROTOTYPE:
524: case NODE_INFORMALEQUATION:
525: case NODE_INLINEEQUATION:
526: case NODE_ITEMIZEDLIST:
527: case NODE_LEGALNOTICE:
528: case NODE_LITERALLAYOUT:
529: case NODE_NOTE:
530: case NODE_ORDEREDLIST:
531: case NODE_PARA:
532: case NODE_PREFACE:
533: case NODE_PROGRAMLISTING:
534: case NODE_REFMETA:
535: case NODE_REFNAMEDIV:
536: case NODE_REFSYNOPSISDIV:
537: case NODE_ROW:
538: case NODE_SCREEN:
539: case NODE_SECTION:
540: case NODE_SYNOPSIS:
541: case NODE_TGROUP:
542: case NODE_TIP:
543: case NODE_TITLE:
544: case NODE_VARIABLELIST:
545: case NODE_VARLISTENTRY:
546: case NODE_WARNING:
547: dat->spc = 1;
548: break;
549: default:
1.23 schwarze 550: dat->spc = (ps->flags & PFLAG_SPC) != 0;
1.17 schwarze 551: break;
552: }
1.1 schwarze 553: dat->parent = ps->cur;
554: TAILQ_INIT(&dat->childq);
555: TAILQ_INIT(&dat->attrq);
556:
557: if (ps->cur != NULL)
558: TAILQ_INSERT_TAIL(&ps->cur->childq, dat, child);
559:
560: ps->cur = dat;
1.23 schwarze 561: if (dat->node == NODE_DOCTYPE) {
562: if (ps->doctype == NULL)
563: ps->doctype = dat;
564: else
565: error_msg(ps, "duplicate doctype");
566: } else if (dat->parent == NULL && ps->tree->root == NULL)
1.1 schwarze 567: ps->tree->root = dat;
1.5 schwarze 568: }
569:
570: static void
571: xml_attrkey(struct parse *ps, const char *name)
572: {
573: struct pattr *attr;
1.23 schwarze 574: const char *value;
1.5 schwarze 575: enum attrkey key;
1.1 schwarze 576:
1.19 schwarze 577: if (ps->del > 0 || ps->ncur == NODE_IGNORE || *name == '\0')
1.5 schwarze 578: return;
1.23 schwarze 579:
580: if ((ps->ncur == NODE_DOCTYPE || ps->ncur == NODE_ENTITY) &&
581: TAILQ_FIRST(&ps->cur->attrq) == NULL) {
582: value = name;
583: name = "NAME";
584: } else
585: value = NULL;
586:
1.5 schwarze 587: if ((key = attrkey_parse(name)) == ATTRKEY__MAX) {
1.23 schwarze 588: ps->flags &= ~PFLAG_ATTR;
1.5 schwarze 589: return;
590: }
591: if ((attr = calloc(1, sizeof(*attr))) == NULL) {
592: perror(NULL);
593: exit(1);
594: }
595: attr->key = key;
596: attr->val = ATTRVAL__MAX;
1.23 schwarze 597: if (value == NULL) {
598: attr->rawval = NULL;
599: ps->flags |= PFLAG_ATTR;
600: } else {
601: if ((attr->rawval = strdup(value)) == NULL) {
602: perror(NULL);
603: exit(1);
604: }
605: ps->flags &= ~PFLAG_ATTR;
606: }
1.5 schwarze 607: TAILQ_INSERT_TAIL(&ps->cur->attrq, attr, child);
1.23 schwarze 608: if (ps->ncur == NODE_ENTITY && key == ATTRKEY_NAME)
609: xml_attrkey(ps, "DEFINITION");
1.5 schwarze 610: }
611:
612: static void
613: xml_attrval(struct parse *ps, const char *name)
614: {
615: struct pattr *attr;
616:
1.23 schwarze 617: if (ps->del > 0 || ps->ncur == NODE_IGNORE ||
618: (ps->flags & PFLAG_ATTR) == 0)
1.5 schwarze 619: return;
620: if ((attr = TAILQ_LAST(&ps->cur->attrq, pattrq)) == NULL)
621: return;
622: if ((attr->val = attrval_parse(name)) == ATTRVAL__MAX &&
623: (attr->rawval = strdup(name)) == NULL) {
624: perror(NULL);
625: exit(1);
1.1 schwarze 626: }
627: }
628:
629: /*
630: * Roll up the parse tree.
631: * If we're at a text node, roll that one up first.
632: */
633: static void
1.5 schwarze 634: xml_elem_end(struct parse *ps, const char *name)
1.1 schwarze 635: {
1.5 schwarze 636: const struct element *elem;
637: enum nodeid node;
1.1 schwarze 638:
1.4 schwarze 639: /*
640: * An ancestor is excluded from the tree;
641: * keep track of the number of levels excluded.
642: */
643: if (ps->del > 1) {
644: ps->del--;
645: return;
646: }
647:
1.16 schwarze 648: if (ps->del == 0)
649: pnode_closetext(ps);
1.2 schwarze 650:
1.5 schwarze 651: if (name != NULL) {
652: for (elem = elements; elem->name != NULL; elem++)
653: if (strcmp(elem->name, name) == 0)
654: break;
655: node = elem->node;
656: } else
657: node = ps->ncur;
1.2 schwarze 658:
1.5 schwarze 659: switch (node) {
1.4 schwarze 660: case NODE_DELETE_WARN:
661: case NODE_DELETE:
1.5 schwarze 662: if (ps->del > 0)
663: ps->del--;
1.4 schwarze 664: break;
1.2 schwarze 665: case NODE_IGNORE:
666: break;
1.23 schwarze 667: case NODE_DOCTYPE:
668: ps->flags &= ~PFLAG_EEND;
669: /* FALLTHROUGH */
1.2 schwarze 670: default:
1.5 schwarze 671: if (ps->cur == NULL || node != ps->cur->node) {
1.6 schwarze 672: warn_msg(ps, "element not open: </%s>", name);
1.5 schwarze 673: break;
674: }
675:
676: /*
677: * Refrain from actually closing the document element.
678: * If no more content follows, no harm is done, but if
679: * some content still follows, simply processing it is
680: * obviously better than discarding it or crashing.
681: */
682:
1.23 schwarze 683: if (ps->cur->parent != NULL || node == NODE_DOCTYPE) {
684: ps->cur = ps->cur->parent;
685: if (ps->cur != NULL)
686: ps->ncur = ps->cur->node;
687: } else
1.5 schwarze 688: ps->tree->flags |= TREE_CLOSED;
1.23 schwarze 689: ps->flags &= ~PFLAG_SPC;
1.4 schwarze 690: break;
1.2 schwarze 691: }
1.4 schwarze 692: assert(ps->del == 0);
1.1 schwarze 693: }
694:
695: struct parse *
696: parse_alloc(int warn)
697: {
698: struct parse *p;
699:
700: if ((p = calloc(1, sizeof(*p))) == NULL)
701: return NULL;
702:
703: if ((p->tree = calloc(1, sizeof(*p->tree))) == NULL) {
704: free(p);
705: return NULL;
706: }
1.23 schwarze 707: if (warn)
708: p->flags |= PFLAG_WARN;
709: else
710: p->flags &= ~PFLAG_WARN;
1.1 schwarze 711: return p;
712: }
713:
714: void
715: parse_free(struct parse *p)
716: {
717: if (p == NULL)
718: return;
719: if (p->tree != NULL) {
720: pnode_unlink(p->tree->root);
721: free(p->tree);
722: }
723: free(p);
724: }
725:
1.14 schwarze 726: static void
727: increment(struct parse *p, char *b, size_t *pend, int refill)
728: {
729: if (refill) {
730: if (b[*pend] == '\n') {
731: p->nline++;
732: p->ncol = 1;
733: } else
734: p->ncol++;
735: }
736: ++*pend;
737: }
738:
1.5 schwarze 739: /*
740: * Advance the pend pointer to the next character in the charset.
741: * If the charset starts with a space, it stands for any whitespace.
742: * Update the new input file position, used for messages.
743: * Do not overrun the buffer b of length rlen.
744: * When reaching the end, NUL-terminate the buffer and return 1;
745: * otherwise, return 0.
746: */
747: static int
748: advance(struct parse *p, char *b, size_t rlen, size_t *pend,
1.14 schwarze 749: const char *charset, int refill)
1.5 schwarze 750: {
751: int space;
752:
753: if (*charset == ' ') {
754: space = 1;
755: charset++;
756: } else
757: space = 0;
758:
1.14 schwarze 759: if (refill) {
760: p->nline = p->line;
761: p->ncol = p->col;
762: }
1.5 schwarze 763: while (*pend < rlen) {
764: if (space && isspace((unsigned char)b[*pend]))
765: break;
766: if (strchr(charset, b[*pend]) != NULL)
767: break;
1.14 schwarze 768: increment(p, b, pend, refill);
1.5 schwarze 769: }
770: if (*pend == rlen) {
771: b[rlen] = '\0';
1.14 schwarze 772: return refill;
1.5 schwarze 773: } else
774: return 0;
775: }
776:
1.14 schwarze 777: size_t
778: parse_string(struct parse *p, char *b, size_t rlen,
779: enum pstate *pstate, int refill)
780: {
781: char *cp;
782: size_t poff; /* Parse offset in b[]. */
783: size_t pend; /* Offset of the end of the current word. */
784: int elem_end;
785:
786: pend = 0;
787: for (;;) {
788:
789: /* Proceed to the next token, skipping whitespace. */
790:
791: if (refill) {
792: p->line = p->nline;
793: p->col = p->ncol;
794: }
795: if ((poff = pend) == rlen)
796: break;
797: if (isspace((unsigned char)b[pend])) {
1.23 schwarze 798: p->flags |= PFLAG_SPC;
1.14 schwarze 799: increment(p, b, &pend, refill);
800: continue;
801: }
802:
803: /*
804: * The following four cases (ARG, TAG, and starting an
805: * entity or a tag) all parse a word or quoted string.
806: * If that extends beyond the read buffer and the last
807: * read(2) still got data, they all break out of the
808: * token loop to request more data from the read loop.
809: *
810: * Also, three of them detect self-closing tags, those
811: * ending with "/>", setting the flag elem_end and
812: * calling xml_elem_end() at the very end, after
813: * handling the attribute value, attribute name, or
814: * tag name, respectively.
815: */
816:
817: /* Parse an attribute value. */
818:
819: if (*pstate >= PARSE_ARG) {
820: if (*pstate == PARSE_ARG &&
821: (b[pend] == '\'' || b[pend] == '"')) {
822: *pstate = b[pend] == '"' ?
823: PARSE_DQ : PARSE_SQ;
824: increment(p, b, &pend, refill);
825: continue;
826: }
827: if (advance(p, b, rlen, &pend,
828: *pstate == PARSE_DQ ? "\"" :
829: *pstate == PARSE_SQ ? "'" : " >", refill))
830: break;
831: *pstate = PARSE_TAG;
832: elem_end = 0;
833: if (b[pend] == '>') {
834: *pstate = PARSE_ELEM;
835: if (pend > 0 && b[pend - 1] == '/') {
836: b[pend - 1] = '\0';
837: elem_end = 1;
838: }
1.23 schwarze 839: if (p->flags & PFLAG_EEND)
840: elem_end = 1;
1.14 schwarze 841: }
842: b[pend] = '\0';
843: if (pend < rlen)
844: increment(p, b, &pend, refill);
845: xml_attrval(p, b + poff);
846: if (elem_end)
847: xml_elem_end(p, NULL);
848:
849: /* Look for an attribute name. */
850:
851: } else if (*pstate == PARSE_TAG) {
1.23 schwarze 852: switch (p->ncur) {
853: case NODE_DOCTYPE:
854: if (b[pend] == '[') {
855: *pstate = PARSE_ELEM;
856: increment(p, b, &pend, refill);
857: continue;
858: }
859: /* FALLTHROUGH */
860: case NODE_ENTITY:
861: if (b[pend] == '"' || b[pend] == '\'') {
862: *pstate = PARSE_ARG;
863: continue;
864: }
865: break;
866: default:
867: break;
868: }
1.14 schwarze 869: if (advance(p, b, rlen, &pend, " =>", refill))
870: break;
871: elem_end = 0;
872: switch (b[pend]) {
873: case '>':
874: *pstate = PARSE_ELEM;
875: if (pend > 0 && b[pend - 1] == '/') {
876: b[pend - 1] = '\0';
877: elem_end = 1;
878: }
1.23 schwarze 879: if (p->flags & PFLAG_EEND)
880: elem_end = 1;
1.14 schwarze 881: break;
882: case '=':
883: *pstate = PARSE_ARG;
884: break;
885: default:
886: break;
887: }
888: b[pend] = '\0';
889: if (pend < rlen)
890: increment(p, b, &pend, refill);
891: xml_attrkey(p, b + poff);
892: if (elem_end)
893: xml_elem_end(p, NULL);
894:
895: /* Begin an opening or closing tag. */
896:
897: } else if (b[poff] == '<') {
898: if (advance(p, b, rlen, &pend, " >", refill))
899: break;
900: if (pend > poff + 3 &&
901: strncmp(b + poff, "<!--", 4) == 0) {
902:
903: /* Skip a comment. */
904:
905: cp = strstr(b + pend - 2, "-->");
906: if (cp == NULL) {
907: if (refill)
908: break;
909: cp = b + rlen;
910: } else
911: cp += 3;
912: while (b + pend < cp)
913: increment(p, b, &pend, refill);
914: continue;
915: }
916: elem_end = 0;
917: if (b[pend] != '>')
918: *pstate = PARSE_TAG;
919: else if (pend > 0 && b[pend - 1] == '/') {
920: b[pend - 1] = '\0';
921: elem_end = 1;
922: }
923: b[pend] = '\0';
924: if (pend < rlen)
925: increment(p, b, &pend, refill);
926: if (b[++poff] == '/') {
927: elem_end = 1;
928: poff++;
1.23 schwarze 929: } else {
1.14 schwarze 930: xml_elem_start(p, b + poff);
1.23 schwarze 931: if (*pstate == PARSE_ELEM &&
932: p->flags & PFLAG_EEND)
933: elem_end = 1;
934: }
1.14 schwarze 935: if (elem_end)
936: xml_elem_end(p, b + poff);
937:
1.23 schwarze 938: /* Close a doctype. */
939:
940: } else if (p->ncur == NODE_DOCTYPE && b[poff] == ']') {
941: *pstate = PARSE_TAG;
942: increment(p, b, &pend, refill);
943:
1.14 schwarze 944: /* Process an entity. */
945:
946: } else if (b[poff] == '&') {
947: if (advance(p, b, rlen, &pend, ";", refill))
948: break;
949: b[pend] = '\0';
950: if (pend < rlen)
951: increment(p, b, &pend, refill);
952: xml_entity(p, b + poff + 1);
953:
954: /* Process text up to the next tag, entity, or EOL. */
955:
956: } else {
957: advance(p, b, rlen, &pend, "<&", refill);
958: xml_char(p, b + poff, pend - poff);
959: }
960: }
961: return poff;
962: }
963:
1.24 ! schwarze 964:
! 965: /*
! 966: * The read loop.
! 967: * If the previous token was incomplete and asked for more input,
! 968: * we have to enter the read loop once more even on EOF.
! 969: * Once rsz is 0, incomplete tokens will no longer ask for more input
! 970: * but instead use whatever there is, and then exit the read loop.
! 971: * The minus one on the size limit for read(2) is needed such that
! 972: * advance() can set b[rlen] to NUL when needed.
! 973: */
! 974: static void
! 975: parse_fd(struct parse *p, int fd)
1.1 schwarze 976: {
977: char b[4096];
1.5 schwarze 978: ssize_t rsz; /* Return value from read(2). */
1.14 schwarze 979: size_t rlen; /* Number of bytes in b[]. */
1.5 schwarze 980: size_t poff; /* Parse offset in b[]. */
1.14 schwarze 981: enum pstate pstate;
1.1 schwarze 982:
1.24 ! schwarze 983: rlen = 0;
1.14 schwarze 984: pstate = PARSE_ELEM;
985: while ((rsz = read(fd, b + rlen, sizeof(b) - rlen - 1)) >= 0 &&
986: (rlen += rsz) > 0) {
987: poff = parse_string(p, b, rlen, &pstate, rsz > 0);
1.5 schwarze 988: /* Buffer exhausted; shift left and re-fill. */
989: assert(poff > 0);
990: rlen -= poff;
1.14 schwarze 991: memmove(b, b + poff, rlen);
1.5 schwarze 992: }
1.24 ! schwarze 993: if (rsz < 0)
! 994: error_msg(p, "read: %s", strerror(errno));
! 995: }
! 996:
! 997: /*
! 998: * Open and parse a file.
! 999: */
! 1000: struct ptree *
! 1001: parse_file(struct parse *p, int fd, const char *fname)
! 1002: {
! 1003: const char *save_fname;
! 1004: int save_line, save_col;
! 1005:
! 1006: /* Save and initialize reporting data. */
! 1007:
! 1008: save_fname = p->fname;
! 1009: save_line = p->nline;
! 1010: save_col = p->ncol;
! 1011: p->fname = fname;
! 1012: p->line = 0;
! 1013: p->col = 0;
! 1014:
! 1015: /* Open the file, unless it is already open. */
! 1016:
! 1017: if (fd == -1 && (fd = open(fname, O_RDONLY, 0)) == -1) {
! 1018: error_msg(p, "open: %s", strerror(errno));
! 1019: p->fname = save_fname;
! 1020: return p->tree;
1.5 schwarze 1021: }
1.24 ! schwarze 1022:
! 1023: /*
! 1024: * After opening the starting file, change to the directory it
! 1025: * is located in, in case it wants to include any further files,
! 1026: * which are typically given with relative paths in DocBook.
! 1027: * Do this on a best-effort basis; don't complain about failure.
! 1028: */
! 1029:
! 1030: if (save_fname == NULL && (fname = dirname(fname)) != NULL &&
! 1031: strcmp(fname, ".") != 0)
! 1032: (void)chdir(fname);
! 1033:
! 1034: /* Run the read loop. */
! 1035:
! 1036: p->nline = 1;
! 1037: p->ncol = 1;
! 1038: parse_fd(p, fd);
! 1039:
! 1040: /* On the top level, finalize the parse tree. */
! 1041:
! 1042: if (save_fname == NULL) {
! 1043: pnode_closetext(p);
! 1044: if (p->tree->root == NULL)
! 1045: error_msg(p, "empty document");
! 1046: else if ((p->tree->flags & TREE_CLOSED) == 0)
! 1047: warn_msg(p, "document not closed");
! 1048: pnode_unlink(p->doctype);
! 1049: }
! 1050:
! 1051: /* Clean up. */
! 1052:
! 1053: if (fd != STDIN_FILENO)
! 1054: close(fd);
! 1055: p->fname = save_fname;
! 1056: p->nline = save_line;
! 1057: p->ncol = save_col;
1.1 schwarze 1058: return p->tree;
1059: }
CVSweb