2 * Sherlock Library -- A simple XML parser
4 * (c) 2007--2008 Pavel Charvat <pchar@ucw.cz>
6 * This software may be freely distributed and used according to the terms
7 * of the GNU Lesser General Public License.
12 #include "sherlock/sherlock.h"
13 #include "sherlock/xml/xml.h"
14 #include "sherlock/xml/dtd.h"
15 #include "sherlock/xml/internals.h"
16 #include "lib/fastbuf.h"
17 #include "lib/ff-unicode.h"
18 #include "lib/unicode.h"
19 #include "lib/chartype.h"
20 #include "lib/hashfunc.h"
24 /*** Basic parsing ***/
27 xml_fatal_expected(struct xml_context *ctx, uns c)
29 if (c >= 32 && c < 128)
30 xml_fatal(ctx, "Expected '%c'", c);
32 xml_fatal(ctx, "Expected U+%04x", c);
36 xml_fatal_expected_white(struct xml_context *ctx)
38 xml_fatal(ctx, "Expected a white space");
42 xml_fatal_expected_quot(struct xml_context *ctx)
44 xml_fatal(ctx, "Expected a quotation mark");
48 xml_parse_eq(struct xml_context *ctx)
50 /* Eq ::= S? '=' S? */
51 xml_parse_white(ctx, 0);
52 xml_parse_char(ctx, '=');
53 xml_parse_white(ctx, 0);
56 /*** Names and nmtokens ***/
59 xml_parse_string(struct xml_context *ctx, struct mempool *pool, uns first_cat, uns next_cat, char *err)
61 char *p = mp_start_noalign(pool, 1);
62 if (unlikely(!(xml_peek_cat(ctx) & first_cat)))
63 xml_fatal(ctx, "%s", err);
66 p = mp_spread(pool, p, 5);
67 p = utf8_32_put(p, xml_skip_char(ctx));
69 while (xml_peek_cat(ctx) & next_cat);
71 return mp_end(pool, p);
75 xml_skip_string(struct xml_context *ctx, uns first_cat, uns next_cat, char *err)
77 if (unlikely(!(xml_get_cat(ctx) & first_cat)))
78 xml_fatal(ctx, "%s", err);
79 while (xml_peek_cat(ctx) & next_cat)
84 xml_parse_name(struct xml_context *ctx, struct mempool *pool)
86 /* Name ::= NameStartChar (NameChar)* */
87 return xml_parse_string(ctx, pool, ctx->cat_sname, ctx->cat_name, "Expected a name");
91 xml_skip_name(struct xml_context *ctx)
93 xml_skip_string(ctx, ctx->cat_sname, ctx->cat_name, "Expected a name");
97 xml_parse_nmtoken(struct xml_context *ctx, struct mempool *pool)
99 /* Nmtoken ::= (NameChar)+ */
100 return xml_parse_string(ctx, pool, ctx->cat_name, ctx->cat_name, "Expected a nmtoken");
103 /*** Simple literals ***/
106 xml_parse_system_literal(struct xml_context *ctx, struct mempool *pool)
108 /* SystemLiteral ::= ('"' [^"]* '"') | ("'" [^']* "'") */
109 char *p = mp_start_noalign(pool, 1);
110 uns q = xml_parse_quote(ctx), c;
111 while ((c = xml_get_char(ctx)) != q)
113 p = mp_spread(pool, p, 5);
114 p = utf8_32_put(p, c);
117 return mp_end(pool, p);
121 xml_parse_pubid_literal(struct xml_context *ctx, struct mempool *pool)
123 /* PubidLiteral ::= '"' PubidChar* '"' | "'" (PubidChar - "'")* "'" */
124 char *p = mp_start_noalign(pool, 1);
125 uns q = xml_parse_quote(ctx), c;
126 while ((c = xml_get_char(ctx)) != q)
128 if (unlikely(!(xml_last_cat(ctx) & XML_CHAR_PUBID)))
129 xml_fatal(ctx, "Expected a pubid character");
130 p = mp_spread(pool, p, 2);
134 return mp_end(pool, p);
140 xml_push_comment(struct xml_context *ctx)
142 TRACE(ctx, "push_comment");
143 /* Comment ::= '<!--' ((Char - '-') | ('-' (Char - '-')))* '-->'
144 * Already parsed: '<!-' */
145 xml_parse_char(ctx, '-');
146 struct xml_node *n = xml_push_dom(ctx, NULL);
147 n->type = XML_NODE_COMMENT;
148 char *p = mp_start_noalign(ctx->pool, 6);
151 if (xml_get_char(ctx) == '-')
152 if (xml_get_char(ctx) == '-')
156 p = utf8_32_put(p, xml_last_char(ctx));
157 p = mp_spread(ctx->pool, p, 6);
159 xml_parse_char(ctx, '>');
161 n->len = p - (char *)mp_ptr(ctx->pool);
162 n->text = mp_end(ctx->pool, p + 1);
163 if ((ctx->flags & XML_REPORT_COMMENTS) && ctx->h_comment)
168 xml_pop_comment(struct xml_context *ctx)
170 xml_pop_dom(ctx, !(ctx->flags & XML_ALLOC_COMMENTS));
172 TRACE(ctx, "pop_comment");
176 xml_skip_comment(struct xml_context *ctx)
178 TRACE(ctx, "skip_comment");
179 xml_parse_char(ctx, '-');
180 while (xml_get_char(ctx) != '-' || xml_get_char(ctx) != '-');
181 xml_parse_char(ctx, '>');
185 /*** Processing instructions ***/
188 xml_push_pi(struct xml_context *ctx)
190 TRACE(ctx, "push_pi");
191 /* Parses a PI to ctx->value and ctx->name:
192 * PI ::= '<?' PITarget (S (Char* - (Char* '?>' Char*)))? '?>'
193 * PITarget ::= Name - (('X' | 'x') ('M' | 'm') ('L' | 'l'))
194 * Already parsed: '<?' */
195 struct xml_node *n = xml_push_dom(ctx, NULL);
196 n->type = XML_NODE_PI;
197 n->name = xml_parse_name(ctx, ctx->pool);
198 if (unlikely(!strcasecmp(n->name, "xml")))
199 xml_error(ctx, "Reserved PI target");
200 char *p = mp_start_noalign(ctx->pool, 5);
201 if (!xml_parse_white(ctx, 0))
202 xml_parse_seq(ctx, "?>");
206 if (xml_get_char(ctx) == '?')
207 if (xml_peek_char(ctx) == '>')
215 p = utf8_32_put(p, xml_last_char(ctx));
216 p = mp_spread(ctx->pool, p, 5);
219 n->len = p - (char *)mp_ptr(ctx->pool);
220 n->text = mp_end(ctx->pool, p + 1);
221 if ((ctx->flags & XML_REPORT_PIS) && ctx->h_pi)
226 xml_pop_pi(struct xml_context *ctx)
228 xml_pop_dom(ctx, !(ctx->flags & XML_ALLOC_PIS));
230 TRACE(ctx, "pop_pi");
234 xml_skip_pi(struct xml_context *ctx)
236 TRACE(ctx, "skip_pi");
237 if (ctx->flags & XML_VALIDATING)
239 struct mempool_state state;
240 mp_save(ctx->stack, &state);
241 if (unlikely(!strcasecmp(xml_parse_name(ctx, ctx->stack), "xml")))
242 xml_error(ctx, "Reserved PI target");
243 mp_restore(ctx->stack, &state);
244 if (!xml_parse_white(ctx, 0))
246 xml_parse_seq(ctx, "?>");
252 if (xml_get_char(ctx) == '?')
253 if (xml_peek_char(ctx) == '>')
259 /*** Character references ***/
262 xml_parse_char_ref(struct xml_context *ctx)
264 TRACE(ctx, "parse_char_ref");
265 /* CharRef ::= '&#' [0-9]+ ';' | '&#x' [0-9a-fA-F]+ ';'
266 * Already parsed: '&#' */
268 if (xml_get_char(ctx) == 'x')
270 if (!(xml_get_cat(ctx) & XML_CHAR_XDIGIT))
272 xml_error(ctx, "Expected a hexadecimal value of character reference");
277 v = (v << 4) + Cxvalue(xml_last_char(ctx));
279 while (v < 0x110000 && (xml_get_cat(ctx) & XML_CHAR_XDIGIT));
283 if (!(xml_last_cat(ctx) & XML_CHAR_DIGIT))
285 xml_error(ctx, "Expected a numeric value of character reference");
290 v = v * 10 + xml_last_char(ctx) - '0';
292 while (v < 0x110000 && (xml_get_cat(ctx) & XML_CHAR_DIGIT));
294 uns cat = xml_char_cat(v);
295 if (!(cat & ctx->cat_unrestricted))
297 xml_error(ctx, "Character reference out of range");
300 if (xml_last_char(ctx) == ';')
305 xml_error(ctx, "Expected ';'");
307 while (xml_last_char(ctx) != ';')
310 return UNI_REPLACEMENT;
313 /*** References to general entities ***/
316 xml_parse_ref(struct xml_context *ctx)
318 /* Reference ::= EntityRef | CharRef
319 * EntityRef ::= '&' Name ';'
320 * Already parsed: '&' */
321 struct fastbuf *out = &ctx->chars;
322 if (xml_peek_char(ctx) == '#')
325 bput_utf8_32(out, xml_parse_char_ref(ctx));
329 TRACE(ctx, "parse_ge_ref");
330 struct mempool_state state;
331 mp_save(ctx->stack, &state);
332 char *name = xml_parse_name(ctx, ctx->stack);
333 xml_parse_char(ctx, ';');
334 struct xml_dtd_entity *ent = xml_dtd_find_entity(ctx, name);
337 xml_error(ctx, "Unknown entity &%s;", name);
342 else if (ent->flags & XML_DTD_ENTITY_TRIVIAL)
344 TRACE(ctx, "Trivial entity &%s;", name);
345 bputs(out, ent->text);
349 TRACE(ctx, "Pushed entity &%s;", name);
350 mp_restore(ctx->stack, &state);
352 xml_push_entity(ctx, ent);
355 mp_restore(ctx->stack, &state);
360 /*** Character data ***/
363 xml_spout_chars(struct fastbuf *fb)
365 if (fb->bptr < fb->bufend)
367 struct xml_context *ctx = SKIP_BACK(struct xml_context, chars, fb);
368 struct mempool *pool = ctx->pool;
369 if (fb->bufend != fb->buffer)
371 TRACE(ctx, "growing chars");
372 uns len = fb->bufend - fb->buffer;
373 uns reported = fb->bstop - fb->buffer;
374 fb->buffer = mp_expand(pool);
375 fb->bufend = fb->buffer + mp_avail(pool);
376 fb->bptr = fb->buffer + len;
377 fb->bstop = fb->buffer + reported;
381 TRACE(ctx, "starting chars");
382 mp_save(pool, &ctx->chars_state);
383 fb->bptr = fb->buffer = fb->bstop = mp_start_noalign(pool, 2);
384 fb->bufend = fb->buffer + mp_avail(pool) - 1;
389 xml_end_chars(struct xml_context *ctx, char **out)
391 struct fastbuf *fb = &ctx->chars;
392 uns len = fb->bptr - fb->buffer;
395 TRACE(ctx, "ending chars");
397 *out = mp_end(ctx->pool, fb->bptr + 1);
398 fb->bufend = fb->bstop = fb->bptr = fb->buffer;
404 xml_report_chars(struct xml_context *ctx, char **out)
406 struct fastbuf *fb = &ctx->chars;
407 uns len = fb->bptr - fb->buffer;
412 fb->bstop = fb->bptr;
418 xml_flush_chars(struct xml_context *ctx)
421 uns len = xml_end_chars(ctx, &text), rlen;
424 if ((ctx->flags & XML_REPORT_CHARS) && ctx->h_block && (rlen = xml_report_chars(ctx, &rtext)))
425 ctx->h_block(ctx, rtext, rlen);
426 if (!(ctx->flags & XML_ALLOC_CHARS) && !(ctx->flags & XML_REPORT_CHARS) && !ctx->h_chars)
428 mp_restore(ctx->pool, &ctx->chars_state);
431 struct xml_node *n = xml_push_dom(ctx, &ctx->chars_state);
432 n->type = XML_NODE_CHARS;
435 if ((ctx->flags & XML_REPORT_CHARS) && ctx->h_chars)
442 xml_pop_chars(struct xml_context *ctx)
444 xml_pop_dom(ctx, !(ctx->flags & XML_ALLOC_CHARS));
445 TRACE(ctx, "pop_chars");
449 xml_append_chars(struct xml_context *ctx)
451 TRACE(ctx, "append_chars");
452 struct fastbuf *out = &ctx->chars;
453 while (xml_get_char(ctx) != '<')
454 if (xml_last_char(ctx) == '&')
460 bput_utf8_32(out, xml_last_char(ctx));
464 /*** CDATA sections ***/
467 xml_append_cdata(struct xml_context *ctx)
469 /* CDSect :== '<![CDATA[' (Char* - (Char* ']]>' Char*)) ']]>'
470 * Already parsed: '<![' */
471 TRACE(ctx, "append_cdata");
472 xml_parse_seq(ctx, "CDATA[");
473 struct fastbuf *out = &ctx->chars;
476 if ((ctx->flags & XML_REPORT_CHARS) && ctx->h_block && (rlen = xml_report_chars(ctx, &rtext)))
477 ctx->h_block(ctx, rtext, rlen);
480 if (xml_get_char(ctx) == ']')
482 if (xml_get_char(ctx) == ']')
483 if (xml_get_char(ctx) == '>')
489 bput_utf8_32(out, xml_last_char(ctx));
491 if ((ctx->flags & XML_REPORT_CHARS) && ctx->h_cdata && (rlen = xml_report_chars(ctx, &rtext)))
492 ctx->h_cdata(ctx, rtext, rlen);
497 xml_skip_cdata(struct xml_context *ctx)
499 TRACE(ctx, "skip_cdata");
500 xml_parse_seq(ctx, "CDATA[");
501 while (xml_get_char(ctx) != ']' || xml_get_char(ctx) != ']' || xml_get_char(ctx) != '>');
505 /*** Attribute values ***/
508 xml_parse_attr_value(struct xml_context *ctx, struct xml_dtd_attr *attr UNUSED)
510 TRACE(ctx, "parse_attr_value");
511 /* AttValue ::= '"' ([^<&"] | Reference)* '"' | "'" ([^<&'] | Reference)* "'" */
512 /* FIXME: -- check value constrains / normalize leading/trailing WS and repeated WS */
513 struct mempool_state state;
514 uns quote = xml_parse_quote(ctx);
515 mp_save(ctx->stack, &state);
516 struct fastbuf *out = &ctx->chars;
517 struct xml_source *src = ctx->src;
520 uns c = xml_get_char(ctx);
526 else if (c == quote && src == ctx->src)
529 xml_error(ctx, "Attribute value must not contain '<'");
530 else if (xml_last_cat(ctx) & XML_CHAR_WHITE)
533 bput_utf8_32(out, c);
535 mp_restore(ctx->stack, &state);
537 return xml_end_chars(ctx, &text) ? text : "";
541 xml_normalize_white(struct xml_context *ctx UNUSED, char *text)
543 char *s = text, *d = text;
552 while (*++s == 0x20);
555 if (d != text && d[-1] == 0x20)
563 struct xml_attrs_table;
566 xml_attrs_hash(struct xml_attrs_table *t UNUSED, struct xml_node *e, char *n)
568 return hash_pointer(e) ^ hash_string(n);
572 xml_attrs_eq(struct xml_attrs_table *t UNUSED, struct xml_node *e1, char *n1, struct xml_node *e2, char *n2)
574 return (e1 == e2) && !strcmp(n1, n2);
578 xml_attrs_init_key(struct xml_attrs_table *t UNUSED, struct xml_attr *a, struct xml_node *e, char *name)
584 slist_add_tail(&e->attrs, &a->n);
587 #define HASH_PREFIX(x) xml_attrs_##x
588 #define HASH_NODE struct xml_attr
589 #define HASH_KEY_COMPLEX(x) x elem, x name
590 #define HASH_KEY_DECL struct xml_node *elem, char *name
591 #define HASH_TABLE_DYNAMIC
593 #define HASH_GIVE_HASHFN
594 #define HASH_GIVE_INIT_KEY
595 #define HASH_WANT_CLEANUP
596 #define HASH_WANT_REMOVE
597 #define HASH_WANT_LOOKUP
598 #define HASH_WANT_FIND
599 #define HASH_GIVE_ALLOC
601 #include "lib/hashtable.h"
604 xml_parse_attr(struct xml_context *ctx)
606 TRACE(ctx, "parse_attr");
607 /* Attribute ::= Name Eq AttValue */
608 struct xml_node *e = ctx->node;
609 char *n = xml_parse_name(ctx, ctx->pool);
610 struct xml_attr *a = xml_attrs_lookup(ctx->tab_attrs, e, n);
612 char *v = xml_parse_attr_value(ctx, NULL);
615 xml_error(ctx, "Attribute %s is not unique in element <%s>", n, e->name);
621 else if (!(a->dtd = xml_dtd_find_attr(ctx, e->dtd, a->name)))
622 xml_error(ctx, "Undefined attribute %s in element <%s>", n, e->name);
624 xml_validate_attr(ctx, a->dtd, a->val);
628 xml_attr_find(struct xml_context *ctx, struct xml_node *node, char *name)
630 return xml_attrs_find(ctx->tab_attrs, node, name);
634 xml_attrs_table_init(struct xml_context *ctx)
636 xml_attrs_init(ctx->tab_attrs = xml_hash_new(ctx->pool, sizeof(struct xml_attrs_table)));
640 xml_attrs_table_cleanup(struct xml_context *ctx)
642 xml_attrs_cleanup(ctx->tab_attrs);
648 xml_push_element(struct xml_context *ctx)
650 TRACE(ctx, "push_element");
651 /* EmptyElemTag | STag
652 * EmptyElemTag ::= '<' Name (S Attribute)* S? '/>'
653 * STag ::= '<' Name (S Attribute)* S? '>'
654 * Already parsed: '<' */
655 struct xml_node *e = xml_push_dom(ctx, NULL);
656 clist_init(&e->sons);
657 e->type = XML_NODE_ELEM;
658 e->name = xml_parse_name(ctx, ctx->pool);
659 slist_init(&e->attrs);
663 if (ctx->doctype && strcmp(e->name, ctx->doctype))
664 xml_error(ctx, "The root element <%s> does not match the document type <%s>", e->name, ctx->doctype);
668 else if (!(e->dtd = xml_dtd_find_elem(ctx, e->name)))
669 xml_error(ctx, "Undefined element <%s>", e->name);
672 // FIXME: validate regular expressions
676 uns white = xml_parse_white(ctx, 0);
677 uns c = xml_get_char(ctx);
680 xml_parse_char(ctx, '>');
681 ctx->flags |= XML_EMPTY_ELEM_TAG;
687 xml_fatal_expected_white(ctx);
691 if ((ctx->flags & XML_REPORT_TAGS) && ctx->h_stag)
696 xml_pop_element(struct xml_context *ctx)
698 TRACE(ctx, "pop_element");
699 if ((ctx->flags & XML_REPORT_TAGS) && ctx->h_etag)
701 struct xml_node *e = ctx->node;
702 uns free = !(ctx->flags & XML_ALLOC_TAGS);
707 /* Restore hash table of attributes */
708 SLIST_FOR_EACH(struct xml_attr *, a, e->attrs)
709 xml_attrs_remove(ctx->tab_attrs, a);
711 while (n = clist_head(&e->sons))
713 if (n->type == XML_NODE_ELEM)
715 SLIST_FOR_EACH(struct xml_attr *, a, n->attrs)
716 xml_attrs_remove(ctx->tab_attrs, a);
717 clist_insert_list_after(&n->sons, &n->n);
722 xml_pop_dom(ctx, free);
727 xml_parse_etag(struct xml_context *ctx)
729 /* ETag ::= '</' Name S? '>'
730 * Already parsed: '<' */
731 struct xml_node *e = ctx->node;
737 n = utf8_32_get(n, &c);
738 if (xml_get_char(ctx) != c)
741 xml_parse_white(ctx, 0);
742 if (xml_get_char(ctx) != '>')
745 xml_error(ctx, "Invalid ETag, expected </%s>", e->name);
746 while (xml_get_char(ctx) != '>');
751 /*** Document type declaration ***/
754 xml_parse_doctype_decl(struct xml_context *ctx)
756 TRACE(ctx, "parse_doctype_decl");
757 /* doctypedecl ::= '<!DOCTYPE' S Name (S ExternalID)? S? ('[' intSubset ']' S?)? '>'
758 * Already parsed: '<!'
759 * Terminated before '[' or '>' */
761 xml_fatal(ctx, "Multiple document types not allowed");
762 xml_parse_seq(ctx, "DOCTYPE");
763 xml_parse_white(ctx, 1);
764 ctx->doctype = xml_parse_name(ctx, ctx->pool);
765 TRACE(ctx, "doctype=%s", ctx->doctype);
767 if (xml_parse_white(ctx, 0) && ((c = xml_peek_char(ctx)) == 'S' || c == 'P'))
771 xml_parse_seq(ctx, "SYSTEM");
772 xml_parse_white(ctx, 1);
773 ctx->system_id = xml_parse_system_literal(ctx, ctx->pool);
777 xml_parse_seq(ctx, "PUBLIC");
778 xml_parse_white(ctx, 1);
779 ctx->public_id = xml_parse_pubid_literal(ctx, ctx->pool);
780 xml_parse_white(ctx, 1);
781 ctx->system_id = xml_parse_system_literal(ctx, ctx->pool);
783 xml_parse_white(ctx, 0);
784 ctx->flags |= XML_HAS_EXTERNAL_SUBSET;
786 if (xml_peek_char(ctx) == '[')
787 ctx->flags |= XML_HAS_INTERNAL_SUBSET;
788 if (ctx->h_doctype_decl)
789 ctx->h_doctype_decl(ctx);
794 ///////////////////////////////////////////////////////////////////////////////////////////////////////////
796 /* DTD: Internal subset */
799 xml_parse_internal_subset(struct xml_context *ctx)
801 // FIXME: comments/pi have no parent
803 * intSubset :== (markupdecl | DeclSep)
804 * Already parsed: ']' */
807 xml_parse_white(ctx, 0);
808 uns c = xml_get_char(ctx);
811 if ((c = xml_get_char(ctx)) == '!')
812 switch (c = xml_get_char(ctx))
815 xml_push_comment(ctx);
816 xml_pop_comment(ctx);
819 xml_parse_seq(ctx, "OTATION");
820 xml_parse_notation_decl(ctx);
823 if ((c = xml_get_char(ctx)) == 'N')
825 xml_parse_seq(ctx, "TITY");
826 xml_parse_entity_decl(ctx);
830 xml_parse_seq(ctx, "EMENT");
831 xml_parse_element_decl(ctx);
837 xml_parse_seq(ctx, "TTLIST");
838 xml_parse_attr_list_decl(ctx);
851 xml_parse_pe_ref(ctx);
861 xml_fatal(ctx, "Invalid markup in the internal subset");
864 /*** The State Machine ***/
867 xml_next(struct xml_context *ctx)
869 /* A nasty state machine */
871 #define PULL(x) do { if (ctx->pull & XML_PULL_##x) return ctx->state = XML_STATE_##x; case XML_STATE_##x: ; } while (0)
872 #define PULL_STATE(x, s) do { if (ctx->pull & XML_PULL_##x) return ctx->state = XML_STATE_##s, XML_STATE_##x; case XML_STATE_##s: ; } while (0)
874 TRACE(ctx, "xml_next (state=%u)", ctx->state);
876 ctx->throw_buf = &throw_buf;
877 if (setjmp(throw_buf))
880 if (ctx->err_code == XML_ERR_EOF && ctx->h_fatal)
882 TRACE(ctx, "raised fatal error");
883 return ctx->state = XML_STATE_EOF;
888 case XML_STATE_START:
889 TRACE(ctx, "entering prolog");
890 ctx->flags |= XML_SRC_DOCUMENT | XML_SRC_EXPECTED_DECL;
891 if (ctx->h_document_start)
892 ctx->h_document_start(ctx);
896 ctx->h_xml_decl(ctx);
899 /* Misc* (doctypedecl Misc*)? */
902 xml_parse_white(ctx, 0);
903 xml_parse_char(ctx, '<');
905 if ((c = xml_get_char(ctx)) == '?')
906 /* Processing intruction */
907 if (!(ctx->flags & XML_REPORT_PIS))
912 PULL_STATE(PI, PROLOG_PI);
917 /* Found the root tag */
921 else if (xml_get_char(ctx) == '-')
922 if (!(ctx->flags & XML_REPORT_COMMENTS))
923 xml_skip_comment(ctx);
926 xml_push_comment(ctx);
927 PULL_STATE(COMMENT, PROLOG_COMMENT);
928 xml_pop_comment(ctx);
934 xml_parse_doctype_decl(ctx);
936 if (xml_peek_char(ctx) == '[')
940 if (ctx->flags & XML_PARSE_DTD)
943 if (ctx->h_dtd_start)
944 ctx->h_dtd_start(ctx);
945 // FIXME: pull iface?
946 xml_parse_internal_subset(ctx);
947 // FIXME: external subset
952 xml_skip_internal_subset(ctx);
954 xml_parse_white(ctx, 0);
955 xml_parse_char(ctx, '>');
960 case XML_STATE_CHARS:
964 if (xml_peek_char(ctx) != '<')
967 xml_append_chars(ctx);
975 if ((c = xml_get_char(ctx)) == '?')
978 if (!(ctx->flags & (XML_REPORT_PIS | XML_ALLOC_PIS)))
982 if (xml_flush_chars(ctx))
984 PULL_STATE(CHARS, CHARS_BEFORE_PI);
994 if ((c = xml_get_char(ctx)) == '-')
997 if (!(ctx->flags & (XML_REPORT_COMMENTS | XML_ALLOC_COMMENTS)))
998 xml_skip_comment(ctx);
1001 if (xml_flush_chars(ctx))
1003 PULL_STATE(CHARS, CHARS_BEFORE_COMMENT);
1006 xml_push_comment(ctx);
1008 xml_pop_comment(ctx);
1014 xml_append_cdata(ctx);
1017 xml_fatal(ctx, "Unexpected character after '<!'");
1021 /* STag | EmptyElemTag */
1022 xml_unget_char(ctx);
1023 if (xml_flush_chars(ctx))
1025 PULL_STATE(CHARS, CHARS_BEFORE_STAG);
1029 xml_push_element(ctx);
1031 if (ctx->flags & XML_EMPTY_ELEM_TAG)
1038 if (xml_flush_chars(ctx))
1040 PULL_STATE(CHARS, CHARS_BEFORE_ETAG);
1044 xml_parse_etag(ctx);
1047 xml_pop_element(ctx);
1055 TRACE(ctx, "entering epilog");
1058 /* Epilog whitespace is the only place, where a valid document can reach EOF */
1059 if (setjmp(throw_buf))
1060 if (ctx->err_code == XML_ERR_EOF)
1062 TRACE(ctx, "reached EOF");
1063 ctx->state = XML_STATE_EOF;
1064 if (ctx->h_document_end)
1065 ctx->h_document_end(ctx);
1068 ctx->err_msg = NULL;
1069 return XML_STATE_EOF;
1073 xml_parse_white(ctx, 0);
1074 if (setjmp(throw_buf))
1078 xml_parse_char(ctx, '<');
1080 if ((c = xml_get_char(ctx)) == '?')
1081 /* Processing instruction */
1082 if (!(ctx->flags & XML_REPORT_PIS))
1087 PULL_STATE(PI, EPILOG_PI);
1092 xml_parse_char(ctx, '-');
1094 if (!(ctx->flags & XML_REPORT_COMMENTS))
1095 xml_skip_comment(ctx);
1098 xml_push_comment(ctx);
1099 PULL_STATE(COMMENT, EPILOG_COMMENT);
1100 xml_pop_comment(ctx);
1104 xml_fatal(ctx, "Syntax error in the epilog");
1112 xml_parse(struct xml_context *ctx)
1114 /* This cycle should run only once unless the user overrides the value of ctx->pull in a SAX handler */
1119 while (xml_next(ctx));
1120 return ctx->err_code;