1 /* gmarkup.c - Simple XML-like parser
3 * Copyright 2000, 2003 Red Hat, Inc.
4 * Copyright 2007, 2008 Ryan Lortie <desrt@desrt.ca>
6 * GLib is free software; you can redistribute it and/or modify it
7 * under the terms of the GNU Lesser General Public License as
8 * published by the Free Software Foundation; either version 2 of the
9 * License, or (at your option) any later version.
11 * GLib is distributed in the hope that it will be useful,
12 * but WITHOUT ANY WARRANTY; without even the implied warranty of
13 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
14 * Lesser General Public License for more details.
16 * You should have received a copy of the GNU Lesser General Public
17 * License along with GLib; see the file COPYING.LIB. If not,
18 * see <http://www.gnu.org/licenses/>.
34 #include "gstrfuncs.h"
36 #include "gtestutils.h"
42 * @Title: Simple XML Subset Parser
43 * @Short_description: parses a subset of XML
44 * @See_also: [XML Specification](http://www.w3.org/TR/REC-xml/)
46 * The "GMarkup" parser is intended to parse a simple markup format
47 * that's a subset of XML. This is a small, efficient, easy-to-use
48 * parser. It should not be used if you expect to interoperate with
49 * other applications generating full-scale XML. However, it's very
50 * useful for application data files, config files, etc. where you
51 * know your application will be the only one writing the file.
52 * Full-scale XML parsers should be able to parse the subset used by
53 * GMarkup, so you can easily migrate to full-scale XML at a later
54 * time if the need arises.
56 * GMarkup is not guaranteed to signal an error on all invalid XML;
57 * the parser may accept documents that an XML parser would not.
58 * However, XML documents which are not well-formed (which is a
59 * weaker condition than being valid. See the
60 * [XML specification](http://www.w3.org/TR/REC-xml/)
61 * for definitions of these terms.) are not considered valid GMarkup
64 * Simplifications to XML include:
66 * - Only UTF-8 encoding is allowed
68 * - No user-defined entities
70 * - Processing instructions, comments and the doctype declaration
71 * are "passed through" but are not interpreted in any way
73 * - No DTD or validation
75 * The markup format does support:
81 * - 5 standard entities: & < > " '
83 * - Character references
85 * - Sections marked as CDATA
88 G_DEFINE_QUARK (g-markup-error-quark, g_markup_error)
93 STATE_AFTER_OPEN_ANGLE,
94 STATE_AFTER_CLOSE_ANGLE,
95 STATE_AFTER_ELISION_SLASH, /* the slash that obviates need for end element */
96 STATE_INSIDE_OPEN_TAG_NAME,
97 STATE_INSIDE_ATTRIBUTE_NAME,
98 STATE_AFTER_ATTRIBUTE_NAME,
99 STATE_BETWEEN_ATTRIBUTES,
100 STATE_AFTER_ATTRIBUTE_EQUALS_SIGN,
101 STATE_INSIDE_ATTRIBUTE_VALUE_SQ,
102 STATE_INSIDE_ATTRIBUTE_VALUE_DQ,
104 STATE_AFTER_CLOSE_TAG_SLASH,
105 STATE_INSIDE_CLOSE_TAG_NAME,
106 STATE_AFTER_CLOSE_TAG_NAME,
107 STATE_INSIDE_PASSTHROUGH,
113 const char *prev_element;
114 const GMarkupParser *prev_parser;
115 gpointer prev_user_data;
116 } GMarkupRecursionTracker;
118 struct _GMarkupParseContext
120 const GMarkupParser *parser;
122 volatile gint ref_count;
124 GMarkupParseFlags flags;
129 GMarkupParseState state;
132 GDestroyNotify dnotify;
134 /* A piece of character data or an element that
135 * hasn't "ended" yet so we haven't yet called
136 * the callback for it.
138 GString *partial_chunk;
139 GSList *spare_chunks;
142 GSList *tag_stack_gstr;
143 GSList *spare_list_nodes;
145 GString **attr_names;
146 GString **attr_values;
150 const gchar *current_text;
151 gssize current_text_len;
152 const gchar *current_text_end;
154 /* used to save the start of the last interesting thingy */
159 guint document_empty : 1;
161 guint awaiting_pop : 1;
164 /* subparser support */
165 GSList *subparser_stack; /* (GMarkupRecursionTracker *) */
166 const char *subparser_element;
167 gpointer held_user_data;
171 * Helpers to reduce our allocation overhead, we have
172 * a well defined allocation lifecycle.
175 get_list_node (GMarkupParseContext *context, gpointer data)
178 if (context->spare_list_nodes != NULL)
180 node = context->spare_list_nodes;
181 context->spare_list_nodes = g_slist_remove_link (context->spare_list_nodes, node);
184 node = g_slist_alloc();
190 free_list_node (GMarkupParseContext *context, GSList *node)
193 context->spare_list_nodes = g_slist_concat (node, context->spare_list_nodes);
197 string_blank (GString *string)
199 string->str[0] = '\0';
204 * g_markup_parse_context_new:
205 * @parser: a #GMarkupParser
206 * @flags: one or more #GMarkupParseFlags
207 * @user_data: user data to pass to #GMarkupParser functions
208 * @user_data_dnotify: user data destroy notifier called when
209 * the parse context is freed
211 * Creates a new parse context. A parse context is used to parse
212 * marked-up documents. You can feed any number of documents into
213 * a context, as long as no errors occur; once an error occurs,
214 * the parse context can't continue to parse text (you have to
215 * free it and create a new parse context).
217 * Returns: a new #GMarkupParseContext
219 GMarkupParseContext *
220 g_markup_parse_context_new (const GMarkupParser *parser,
221 GMarkupParseFlags flags,
223 GDestroyNotify user_data_dnotify)
225 GMarkupParseContext *context;
227 g_return_val_if_fail (parser != NULL, NULL);
229 context = g_new (GMarkupParseContext, 1);
231 context->ref_count = 1;
232 context->parser = parser;
233 context->flags = flags;
234 context->user_data = user_data;
235 context->dnotify = user_data_dnotify;
237 context->line_number = 1;
238 context->char_number = 1;
240 context->partial_chunk = NULL;
241 context->spare_chunks = NULL;
242 context->spare_list_nodes = NULL;
244 context->state = STATE_START;
245 context->tag_stack = NULL;
246 context->tag_stack_gstr = NULL;
247 context->attr_names = NULL;
248 context->attr_values = NULL;
249 context->cur_attr = -1;
250 context->alloc_attrs = 0;
252 context->current_text = NULL;
253 context->current_text_len = -1;
254 context->current_text_end = NULL;
256 context->start = NULL;
257 context->iter = NULL;
259 context->document_empty = TRUE;
260 context->parsing = FALSE;
262 context->awaiting_pop = FALSE;
263 context->subparser_stack = NULL;
264 context->subparser_element = NULL;
266 /* this is only looked at if awaiting_pop = TRUE. initialise anyway. */
267 context->held_user_data = NULL;
269 context->balance = 0;
275 * g_markup_parse_context_ref:
276 * @context: a #GMarkupParseContext
278 * Increases the reference count of @context.
280 * Returns: the same @context
284 GMarkupParseContext *
285 g_markup_parse_context_ref (GMarkupParseContext *context)
287 g_return_val_if_fail (context != NULL, NULL);
288 g_return_val_if_fail (context->ref_count > 0, NULL);
290 g_atomic_int_inc (&context->ref_count);
296 * g_markup_parse_context_unref:
297 * @context: a #GMarkupParseContext
299 * Decreases the reference count of @context. When its reference count
300 * drops to 0, it is freed.
305 g_markup_parse_context_unref (GMarkupParseContext *context)
307 g_return_if_fail (context != NULL);
308 g_return_if_fail (context->ref_count > 0);
310 if (g_atomic_int_dec_and_test (&context->ref_count))
311 g_markup_parse_context_free (context);
315 string_full_free (gpointer ptr)
317 g_string_free (ptr, TRUE);
320 static void clear_attributes (GMarkupParseContext *context);
323 * g_markup_parse_context_free:
324 * @context: a #GMarkupParseContext
326 * Frees a #GMarkupParseContext.
328 * This function can't be called from inside one of the
329 * #GMarkupParser functions or while a subparser is pushed.
332 g_markup_parse_context_free (GMarkupParseContext *context)
334 g_return_if_fail (context != NULL);
335 g_return_if_fail (!context->parsing);
336 g_return_if_fail (!context->subparser_stack);
337 g_return_if_fail (!context->awaiting_pop);
339 if (context->dnotify)
340 (* context->dnotify) (context->user_data);
342 clear_attributes (context);
343 g_free (context->attr_names);
344 g_free (context->attr_values);
346 g_slist_free_full (context->tag_stack_gstr, string_full_free);
347 g_slist_free (context->tag_stack);
349 g_slist_free_full (context->spare_chunks, string_full_free);
350 g_slist_free (context->spare_list_nodes);
352 if (context->partial_chunk)
353 g_string_free (context->partial_chunk, TRUE);
358 static void pop_subparser_stack (GMarkupParseContext *context);
361 mark_error (GMarkupParseContext *context,
364 context->state = STATE_ERROR;
366 if (context->parser->error)
367 (*context->parser->error) (context, error, context->user_data);
369 /* report the error all the way up to free all the user-data */
370 while (context->subparser_stack)
372 pop_subparser_stack (context);
373 context->awaiting_pop = FALSE; /* already been freed */
375 if (context->parser->error)
376 (*context->parser->error) (context, error, context->user_data);
381 set_error (GMarkupParseContext *context,
385 ...) G_GNUC_PRINTF (4, 5);
388 set_error_literal (GMarkupParseContext *context,
391 const gchar *message)
395 tmp_error = g_error_new_literal (G_MARKUP_ERROR, code, message);
397 g_prefix_error (&tmp_error,
398 _("Error on line %d char %d: "),
399 context->line_number,
400 context->char_number);
402 mark_error (context, tmp_error);
404 g_propagate_error (error, tmp_error);
409 set_error (GMarkupParseContext *context,
419 va_start (args, format);
420 s = g_strdup_vprintf (format, args);
423 /* Make sure that the GError message is valid UTF-8
424 * even if it is complaining about invalid UTF-8 in the markup
426 s_valid = _g_utf8_make_valid (s);
427 set_error_literal (context, error, code, s);
434 propagate_error (GMarkupParseContext *context,
438 if (context->flags & G_MARKUP_PREFIX_ERROR_POSITION)
439 g_prefix_error (&src,
440 _("Error on line %d char %d: "),
441 context->line_number,
442 context->char_number);
444 mark_error (context, src);
446 g_propagate_error (dest, src);
449 #define IS_COMMON_NAME_END_CHAR(c) \
450 ((c) == '=' || (c) == '/' || (c) == '>' || (c) == ' ')
453 slow_name_validate (GMarkupParseContext *context,
457 const gchar *p = name;
459 if (!g_utf8_validate (name, strlen (name), NULL))
461 set_error (context, error, G_MARKUP_ERROR_BAD_UTF8,
462 _("Invalid UTF-8 encoded text in name - not valid '%s'"), name);
466 if (!(g_ascii_isalpha (*p) ||
467 (!IS_COMMON_NAME_END_CHAR (*p) &&
470 g_unichar_isalpha (g_utf8_get_char (p))))))
472 set_error (context, error, G_MARKUP_ERROR_PARSE,
473 _("'%s' is not a valid name"), name);
477 for (p = g_utf8_next_char (name); *p != '\0'; p = g_utf8_next_char (p))
480 if (!(g_ascii_isalnum (*p) ||
481 (!IS_COMMON_NAME_END_CHAR (*p) &&
486 g_unichar_isalpha (g_utf8_get_char (p))))))
488 set_error (context, error, G_MARKUP_ERROR_PARSE,
489 _("'%s' is not a valid name: '%c'"), name, *p);
497 * Use me for elements, attributes etc.
500 name_validate (GMarkupParseContext *context,
507 /* name start char */
509 if (G_UNLIKELY (IS_COMMON_NAME_END_CHAR (*p) ||
510 !(g_ascii_isalpha (*p) || *p == '_' || *p == ':')))
513 for (mask = *p++; *p != '\0'; p++)
518 if (G_UNLIKELY (!(g_ascii_isalnum (*p) ||
519 (!IS_COMMON_NAME_END_CHAR (*p) &&
527 if (mask & 0x80) /* un-common / non-ascii */
533 return slow_name_validate (context, name, error);
537 text_validate (GMarkupParseContext *context,
542 if (!g_utf8_validate (p, len, NULL))
544 set_error (context, error, G_MARKUP_ERROR_BAD_UTF8,
545 _("Invalid UTF-8 encoded text in name - not valid '%s'"), p);
553 char_str (gunichar c,
557 g_unichar_to_utf8 (c, buf);
562 utf8_str (const gchar *utf8,
565 char_str (g_utf8_get_char (utf8), buf);
571 set_unescape_error (GMarkupParseContext *context,
573 const gchar *remaining_text,
581 gint remaining_newlines;
584 remaining_newlines = 0;
589 ++remaining_newlines;
593 va_start (args, format);
594 s = g_strdup_vprintf (format, args);
597 tmp_error = g_error_new (G_MARKUP_ERROR,
599 _("Error on line %d: %s"),
600 context->line_number - remaining_newlines,
605 mark_error (context, tmp_error);
607 g_propagate_error (error, tmp_error);
611 * re-write the GString in-place, unescaping anything that escaped.
612 * most XML does not contain entities, or escaping.
615 unescape_gstring_inplace (GMarkupParseContext *context,
623 gboolean normalize_attribute;
627 /* are we unescaping an attribute or not ? */
628 if (context->state == STATE_INSIDE_ATTRIBUTE_VALUE_SQ ||
629 context->state == STATE_INSIDE_ATTRIBUTE_VALUE_DQ)
630 normalize_attribute = TRUE;
632 normalize_attribute = FALSE;
635 * Meeks' theorem: unescaping can only shrink text.
636 * for < etc. this is obvious, for  more
637 * thought is required, but this is patently so.
640 for (from = to = string->str; *from != '\0'; from++, to++)
647 if (normalize_attribute && (*to == '\t' || *to == '\n'))
651 *to = normalize_attribute ? ' ' : '\n';
660 gboolean is_hex = FALSE;
672 /* digit is between start and p */
675 l = strtoul (from, &end, 16);
677 l = strtoul (from, &end, 10);
679 if (end == from || errno != 0)
681 set_unescape_error (context, error,
682 from, G_MARKUP_ERROR_PARSE,
683 _("Failed to parse '%-.*s', which "
684 "should have been a digit "
685 "inside a character reference "
686 "(ê for example) - perhaps "
687 "the digit is too large"),
688 (int)(end - from), from);
691 else if (*end != ';')
693 set_unescape_error (context, error,
694 from, G_MARKUP_ERROR_PARSE,
695 _("Character reference did not end with a "
697 "most likely you used an ampersand "
698 "character without intending to start "
699 "an entity - escape ampersand as &"));
704 /* characters XML 1.1 permits */
705 if ((0 < l && l <= 0xD7FF) ||
706 (0xE000 <= l && l <= 0xFFFD) ||
707 (0x10000 <= l && l <= 0x10FFFF))
712 to += strlen (buf) - 1;
714 if (l >= 0x80) /* not ascii */
719 set_unescape_error (context, error,
720 from, G_MARKUP_ERROR_PARSE,
721 _("Character reference '%-.*s' does not "
722 "encode a permitted character"),
723 (int)(end - from), from);
729 else if (strncmp (from, "lt;", 3) == 0)
734 else if (strncmp (from, "gt;", 3) == 0)
739 else if (strncmp (from, "amp;", 4) == 0)
744 else if (strncmp (from, "quot;", 5) == 0)
749 else if (strncmp (from, "apos;", 5) == 0)
757 set_unescape_error (context, error,
758 from, G_MARKUP_ERROR_PARSE,
759 _("Empty entity '&;' seen; valid "
760 "entities are: & " < > '"));
763 const char *end = strchr (from, ';');
765 set_unescape_error (context, error,
766 from, G_MARKUP_ERROR_PARSE,
767 _("Entity name '%-.*s' is not known"),
768 (int)(end - from), from);
770 set_unescape_error (context, error,
771 from, G_MARKUP_ERROR_PARSE,
772 _("Entity did not end with a semicolon; "
773 "most likely you used an ampersand "
774 "character without intending to start "
775 "an entity - escape ampersand as &"));
782 g_assert (to - string->str <= string->len);
783 if (to - string->str != string->len)
784 g_string_truncate (string, to - string->str);
786 *is_ascii = !(mask & 0x80);
791 static inline gboolean
792 advance_char (GMarkupParseContext *context)
795 context->char_number++;
797 if (G_UNLIKELY (context->iter == context->current_text_end))
800 else if (G_UNLIKELY (*context->iter == '\n'))
802 context->line_number++;
803 context->char_number = 1;
809 static inline gboolean
812 return c == ' ' || c == '\t' || c == '\n' || c == '\r';
816 skip_spaces (GMarkupParseContext *context)
820 if (!xml_isspace (*context->iter))
823 while (advance_char (context));
827 advance_to_name_end (GMarkupParseContext *context)
831 if (IS_COMMON_NAME_END_CHAR (*(context->iter)))
833 if (xml_isspace (*(context->iter)))
836 while (advance_char (context));
840 release_chunk (GMarkupParseContext *context, GString *str)
845 if (str->allocated_len > 256)
846 { /* large strings are unusual and worth freeing */
847 g_string_free (str, TRUE);
851 node = get_list_node (context, str);
852 context->spare_chunks = g_slist_concat (node, context->spare_chunks);
856 add_to_partial (GMarkupParseContext *context,
857 const gchar *text_start,
858 const gchar *text_end)
860 if (context->partial_chunk == NULL)
861 { /* allocate a new chunk to parse into */
863 if (context->spare_chunks != NULL)
865 GSList *node = context->spare_chunks;
866 context->spare_chunks = g_slist_remove_link (context->spare_chunks, node);
867 context->partial_chunk = node->data;
868 free_list_node (context, node);
871 context->partial_chunk = g_string_sized_new (MAX (28, text_end - text_start));
874 if (text_start != text_end)
875 g_string_insert_len (context->partial_chunk, -1,
876 text_start, text_end - text_start);
880 truncate_partial (GMarkupParseContext *context)
882 if (context->partial_chunk != NULL)
883 string_blank (context->partial_chunk);
886 static inline const gchar*
887 current_element (GMarkupParseContext *context)
889 return context->tag_stack->data;
893 pop_subparser_stack (GMarkupParseContext *context)
895 GMarkupRecursionTracker *tracker;
897 g_assert (context->subparser_stack);
899 tracker = context->subparser_stack->data;
901 context->awaiting_pop = TRUE;
902 context->held_user_data = context->user_data;
904 context->user_data = tracker->prev_user_data;
905 context->parser = tracker->prev_parser;
906 context->subparser_element = tracker->prev_element;
907 g_slice_free (GMarkupRecursionTracker, tracker);
909 context->subparser_stack = g_slist_delete_link (context->subparser_stack,
910 context->subparser_stack);
914 push_partial_as_tag (GMarkupParseContext *context)
916 GString *str = context->partial_chunk;
917 /* sadly, this is exported by gmarkup_get_element_stack as-is */
918 context->tag_stack = g_slist_concat (get_list_node (context, str->str), context->tag_stack);
919 context->tag_stack_gstr = g_slist_concat (get_list_node (context, str), context->tag_stack_gstr);
920 context->partial_chunk = NULL;
924 pop_tag (GMarkupParseContext *context)
926 GSList *nodea, *nodeb;
928 nodea = context->tag_stack;
929 nodeb = context->tag_stack_gstr;
930 release_chunk (context, nodeb->data);
931 context->tag_stack = g_slist_remove_link (context->tag_stack, nodea);
932 context->tag_stack_gstr = g_slist_remove_link (context->tag_stack_gstr, nodeb);
933 free_list_node (context, nodea);
934 free_list_node (context, nodeb);
938 possibly_finish_subparser (GMarkupParseContext *context)
940 if (current_element (context) == context->subparser_element)
941 pop_subparser_stack (context);
945 ensure_no_outstanding_subparser (GMarkupParseContext *context)
947 if (context->awaiting_pop)
948 g_critical ("During the first end_element call after invoking a "
949 "subparser you must pop the subparser stack and handle "
950 "the freeing of the subparser user_data. This can be "
951 "done by calling the end function of the subparser. "
952 "Very probably, your program just leaked memory.");
954 /* let valgrind watch the pointer disappear... */
955 context->held_user_data = NULL;
956 context->awaiting_pop = FALSE;
960 current_attribute (GMarkupParseContext *context)
962 g_assert (context->cur_attr >= 0);
963 return context->attr_names[context->cur_attr]->str;
967 add_attribute (GMarkupParseContext *context, GString *str)
969 if (context->cur_attr + 2 >= context->alloc_attrs)
971 context->alloc_attrs += 5; /* silly magic number */
972 context->attr_names = g_realloc (context->attr_names, sizeof(GString*)*context->alloc_attrs);
973 context->attr_values = g_realloc (context->attr_values, sizeof(GString*)*context->alloc_attrs);
976 context->attr_names[context->cur_attr] = str;
977 context->attr_values[context->cur_attr] = NULL;
978 context->attr_names[context->cur_attr+1] = NULL;
979 context->attr_values[context->cur_attr+1] = NULL;
983 clear_attributes (GMarkupParseContext *context)
985 /* Go ahead and free the attributes. */
986 for (; context->cur_attr >= 0; context->cur_attr--)
988 int pos = context->cur_attr;
989 release_chunk (context, context->attr_names[pos]);
990 release_chunk (context, context->attr_values[pos]);
991 context->attr_names[pos] = context->attr_values[pos] = NULL;
993 g_assert (context->cur_attr == -1);
994 g_assert (context->attr_names == NULL ||
995 context->attr_names[0] == NULL);
996 g_assert (context->attr_values == NULL ||
997 context->attr_values[0] == NULL);
1000 /* This has to be a separate function to ensure the alloca's
1001 * are unwound on exit - otherwise we grow & blow the stack
1002 * with large documents
1005 emit_start_element (GMarkupParseContext *context,
1009 const gchar *start_name;
1010 const gchar **attr_names;
1011 const gchar **attr_values;
1014 /* In case we want to ignore qualified tags and we see that we have
1015 * one here, we push a subparser. This will ignore all tags inside of
1016 * the qualified tag.
1018 * We deal with the end of the subparser from emit_end_element.
1020 if ((context->flags & G_MARKUP_IGNORE_QUALIFIED) && strchr (current_element (context), ':'))
1022 static const GMarkupParser ignore_parser;
1023 g_markup_parse_context_push (context, &ignore_parser, NULL);
1024 clear_attributes (context);
1028 attr_names = g_newa (const gchar *, context->cur_attr + 2);
1029 attr_values = g_newa (const gchar *, context->cur_attr + 2);
1030 for (i = 0; i < context->cur_attr + 1; i++)
1032 /* Possibly omit qualified attribute names from the list */
1033 if ((context->flags & G_MARKUP_IGNORE_QUALIFIED) && strchr (context->attr_names[i]->str, ':'))
1036 attr_names[j] = context->attr_names[i]->str;
1037 attr_values[j] = context->attr_values[i]->str;
1040 attr_names[j] = NULL;
1041 attr_values[j] = NULL;
1043 /* Call user callback for element start */
1045 start_name = current_element (context);
1047 if (context->parser->start_element &&
1048 name_validate (context, start_name, error))
1049 (* context->parser->start_element) (context,
1051 (const gchar **)attr_names,
1052 (const gchar **)attr_values,
1055 clear_attributes (context);
1057 if (tmp_error != NULL)
1058 propagate_error (context, error, tmp_error);
1062 emit_end_element (GMarkupParseContext *context,
1065 /* We need to pop the tag stack and call the end_element
1066 * function, since this is the close tag
1068 GError *tmp_error = NULL;
1070 g_assert (context->tag_stack != NULL);
1072 possibly_finish_subparser (context);
1074 /* We might have just returned from our ignore subparser */
1075 if ((context->flags & G_MARKUP_IGNORE_QUALIFIED) && strchr (current_element (context), ':'))
1077 g_markup_parse_context_pop (context);
1083 if (context->parser->end_element)
1084 (* context->parser->end_element) (context,
1085 current_element (context),
1089 ensure_no_outstanding_subparser (context);
1093 mark_error (context, tmp_error);
1094 g_propagate_error (error, tmp_error);
1101 * g_markup_parse_context_parse:
1102 * @context: a #GMarkupParseContext
1103 * @text: chunk of text to parse
1104 * @text_len: length of @text in bytes
1105 * @error: return location for a #GError
1107 * Feed some data to the #GMarkupParseContext.
1109 * The data need not be valid UTF-8; an error will be signaled if
1110 * it's invalid. The data need not be an entire document; you can
1111 * feed a document into the parser incrementally, via multiple calls
1112 * to this function. Typically, as you receive data from a network
1113 * connection or file, you feed each received chunk of data into this
1114 * function, aborting the process if an error occurs. Once an error
1115 * is reported, no further data may be fed to the #GMarkupParseContext;
1116 * all errors are fatal.
1118 * Returns: %FALSE if an error occurred, %TRUE on success
1121 g_markup_parse_context_parse (GMarkupParseContext *context,
1126 g_return_val_if_fail (context != NULL, FALSE);
1127 g_return_val_if_fail (text != NULL, FALSE);
1128 g_return_val_if_fail (context->state != STATE_ERROR, FALSE);
1129 g_return_val_if_fail (!context->parsing, FALSE);
1132 text_len = strlen (text);
1137 context->parsing = TRUE;
1140 context->current_text = text;
1141 context->current_text_len = text_len;
1142 context->current_text_end = context->current_text + text_len;
1143 context->iter = context->current_text;
1144 context->start = context->iter;
1146 while (context->iter != context->current_text_end)
1148 switch (context->state)
1151 /* Possible next state: AFTER_OPEN_ANGLE */
1153 g_assert (context->tag_stack == NULL);
1155 /* whitespace is ignored outside of any elements */
1156 skip_spaces (context);
1158 if (context->iter != context->current_text_end)
1160 if (*context->iter == '<')
1162 /* Move after the open angle */
1163 advance_char (context);
1165 context->state = STATE_AFTER_OPEN_ANGLE;
1167 /* this could start a passthrough */
1168 context->start = context->iter;
1170 /* document is now non-empty */
1171 context->document_empty = FALSE;
1175 set_error_literal (context,
1177 G_MARKUP_ERROR_PARSE,
1178 _("Document must begin with an element (e.g. <book>)"));
1183 case STATE_AFTER_OPEN_ANGLE:
1184 /* Possible next states: INSIDE_OPEN_TAG_NAME,
1185 * AFTER_CLOSE_TAG_SLASH, INSIDE_PASSTHROUGH
1187 if (*context->iter == '?' ||
1188 *context->iter == '!')
1190 /* include < in the passthrough */
1191 const gchar *openangle = "<";
1192 add_to_partial (context, openangle, openangle + 1);
1193 context->start = context->iter;
1194 context->balance = 1;
1195 context->state = STATE_INSIDE_PASSTHROUGH;
1197 else if (*context->iter == '/')
1200 advance_char (context);
1202 context->state = STATE_AFTER_CLOSE_TAG_SLASH;
1204 else if (!IS_COMMON_NAME_END_CHAR (*(context->iter)))
1206 context->state = STATE_INSIDE_OPEN_TAG_NAME;
1208 /* start of tag name */
1209 context->start = context->iter;
1217 G_MARKUP_ERROR_PARSE,
1218 _("'%s' is not a valid character following "
1219 "a '<' character; it may not begin an "
1221 utf8_str (context->iter, buf));
1225 /* The AFTER_CLOSE_ANGLE state is actually sort of
1226 * broken, because it doesn't correspond to a range
1227 * of characters in the input stream as the others do,
1228 * and thus makes things harder to conceptualize
1230 case STATE_AFTER_CLOSE_ANGLE:
1231 /* Possible next states: INSIDE_TEXT, STATE_START */
1232 if (context->tag_stack == NULL)
1234 context->start = NULL;
1235 context->state = STATE_START;
1239 context->start = context->iter;
1240 context->state = STATE_INSIDE_TEXT;
1244 case STATE_AFTER_ELISION_SLASH:
1245 /* Possible next state: AFTER_CLOSE_ANGLE */
1246 if (*context->iter == '>')
1248 /* move after the close angle */
1249 advance_char (context);
1250 context->state = STATE_AFTER_CLOSE_ANGLE;
1251 emit_end_element (context, error);
1259 G_MARKUP_ERROR_PARSE,
1260 _("Odd character '%s', expected a '>' character "
1261 "to end the empty-element tag '%s'"),
1262 utf8_str (context->iter, buf),
1263 current_element (context));
1267 case STATE_INSIDE_OPEN_TAG_NAME:
1268 /* Possible next states: BETWEEN_ATTRIBUTES */
1270 /* if there's a partial chunk then it's the first part of the
1271 * tag name. If there's a context->start then it's the start
1272 * of the tag name in current_text, the partial chunk goes
1273 * before that start though.
1275 advance_to_name_end (context);
1277 if (context->iter == context->current_text_end)
1279 /* The name hasn't necessarily ended. Merge with
1280 * partial chunk, leave state unchanged.
1282 add_to_partial (context, context->start, context->iter);
1286 /* The name has ended. Combine it with the partial chunk
1287 * if any; push it on the stack; enter next state.
1289 add_to_partial (context, context->start, context->iter);
1290 push_partial_as_tag (context);
1292 context->state = STATE_BETWEEN_ATTRIBUTES;
1293 context->start = NULL;
1297 case STATE_INSIDE_ATTRIBUTE_NAME:
1298 /* Possible next states: AFTER_ATTRIBUTE_NAME */
1300 advance_to_name_end (context);
1301 add_to_partial (context, context->start, context->iter);
1303 /* read the full name, if we enter the equals sign state
1304 * then add the attribute to the list (without the value),
1305 * otherwise store a partial chunk to be prepended later.
1307 if (context->iter != context->current_text_end)
1308 context->state = STATE_AFTER_ATTRIBUTE_NAME;
1311 case STATE_AFTER_ATTRIBUTE_NAME:
1312 /* Possible next states: AFTER_ATTRIBUTE_EQUALS_SIGN */
1314 skip_spaces (context);
1316 if (context->iter != context->current_text_end)
1318 /* The name has ended. Combine it with the partial chunk
1319 * if any; push it on the stack; enter next state.
1321 if (!name_validate (context, context->partial_chunk->str, error))
1324 add_attribute (context, context->partial_chunk);
1326 context->partial_chunk = NULL;
1327 context->start = NULL;
1329 if (*context->iter == '=')
1331 advance_char (context);
1332 context->state = STATE_AFTER_ATTRIBUTE_EQUALS_SIGN;
1340 G_MARKUP_ERROR_PARSE,
1341 _("Odd character '%s', expected a '=' after "
1342 "attribute name '%s' of element '%s'"),
1343 utf8_str (context->iter, buf),
1344 current_attribute (context),
1345 current_element (context));
1351 case STATE_BETWEEN_ATTRIBUTES:
1352 /* Possible next states: AFTER_CLOSE_ANGLE,
1353 * AFTER_ELISION_SLASH, INSIDE_ATTRIBUTE_NAME
1355 skip_spaces (context);
1357 if (context->iter != context->current_text_end)
1359 if (*context->iter == '/')
1361 advance_char (context);
1362 context->state = STATE_AFTER_ELISION_SLASH;
1364 else if (*context->iter == '>')
1366 advance_char (context);
1367 context->state = STATE_AFTER_CLOSE_ANGLE;
1369 else if (!IS_COMMON_NAME_END_CHAR (*(context->iter)))
1371 context->state = STATE_INSIDE_ATTRIBUTE_NAME;
1372 /* start of attribute name */
1373 context->start = context->iter;
1381 G_MARKUP_ERROR_PARSE,
1382 _("Odd character '%s', expected a '>' or '/' "
1383 "character to end the start tag of "
1384 "element '%s', or optionally an attribute; "
1385 "perhaps you used an invalid character in "
1386 "an attribute name"),
1387 utf8_str (context->iter, buf),
1388 current_element (context));
1391 /* If we're done with attributes, invoke
1392 * the start_element callback
1394 if (context->state == STATE_AFTER_ELISION_SLASH ||
1395 context->state == STATE_AFTER_CLOSE_ANGLE)
1396 emit_start_element (context, error);
1400 case STATE_AFTER_ATTRIBUTE_EQUALS_SIGN:
1401 /* Possible next state: INSIDE_ATTRIBUTE_VALUE_[SQ/DQ] */
1403 skip_spaces (context);
1405 if (context->iter != context->current_text_end)
1407 if (*context->iter == '"')
1409 advance_char (context);
1410 context->state = STATE_INSIDE_ATTRIBUTE_VALUE_DQ;
1411 context->start = context->iter;
1413 else if (*context->iter == '\'')
1415 advance_char (context);
1416 context->state = STATE_INSIDE_ATTRIBUTE_VALUE_SQ;
1417 context->start = context->iter;
1425 G_MARKUP_ERROR_PARSE,
1426 _("Odd character '%s', expected an open quote mark "
1427 "after the equals sign when giving value for "
1428 "attribute '%s' of element '%s'"),
1429 utf8_str (context->iter, buf),
1430 current_attribute (context),
1431 current_element (context));
1436 case STATE_INSIDE_ATTRIBUTE_VALUE_SQ:
1437 case STATE_INSIDE_ATTRIBUTE_VALUE_DQ:
1438 /* Possible next states: BETWEEN_ATTRIBUTES */
1442 if (context->state == STATE_INSIDE_ATTRIBUTE_VALUE_SQ)
1453 if (*context->iter == delim)
1456 while (advance_char (context));
1458 if (context->iter == context->current_text_end)
1460 /* The value hasn't necessarily ended. Merge with
1461 * partial chunk, leave state unchanged.
1463 add_to_partial (context, context->start, context->iter);
1468 /* The value has ended at the quote mark. Combine it
1469 * with the partial chunk if any; set it for the current
1472 add_to_partial (context, context->start, context->iter);
1474 g_assert (context->cur_attr >= 0);
1476 if (unescape_gstring_inplace (context, context->partial_chunk, &is_ascii, error) &&
1477 (is_ascii || text_validate (context, context->partial_chunk->str,
1478 context->partial_chunk->len, error)))
1480 /* success, advance past quote and set state. */
1481 context->attr_values[context->cur_attr] = context->partial_chunk;
1482 context->partial_chunk = NULL;
1483 advance_char (context);
1484 context->state = STATE_BETWEEN_ATTRIBUTES;
1485 context->start = NULL;
1488 truncate_partial (context);
1492 case STATE_INSIDE_TEXT:
1493 /* Possible next states: AFTER_OPEN_ANGLE */
1496 if (*context->iter == '<')
1499 while (advance_char (context));
1501 /* The text hasn't necessarily ended. Merge with
1502 * partial chunk, leave state unchanged.
1505 add_to_partial (context, context->start, context->iter);
1507 if (context->iter != context->current_text_end)
1511 /* The text has ended at the open angle. Call the text
1514 if (unescape_gstring_inplace (context, context->partial_chunk, &is_ascii, error) &&
1515 (is_ascii || text_validate (context, context->partial_chunk->str,
1516 context->partial_chunk->len, error)))
1518 GError *tmp_error = NULL;
1520 if (context->parser->text)
1521 (*context->parser->text) (context,
1522 context->partial_chunk->str,
1523 context->partial_chunk->len,
1527 if (tmp_error == NULL)
1529 /* advance past open angle and set state. */
1530 advance_char (context);
1531 context->state = STATE_AFTER_OPEN_ANGLE;
1532 /* could begin a passthrough */
1533 context->start = context->iter;
1536 propagate_error (context, error, tmp_error);
1539 truncate_partial (context);
1543 case STATE_AFTER_CLOSE_TAG_SLASH:
1544 /* Possible next state: INSIDE_CLOSE_TAG_NAME */
1545 if (!IS_COMMON_NAME_END_CHAR (*(context->iter)))
1547 context->state = STATE_INSIDE_CLOSE_TAG_NAME;
1549 /* start of tag name */
1550 context->start = context->iter;
1558 G_MARKUP_ERROR_PARSE,
1559 _("'%s' is not a valid character following "
1560 "the characters '</'; '%s' may not begin an "
1562 utf8_str (context->iter, buf),
1563 utf8_str (context->iter, buf));
1567 case STATE_INSIDE_CLOSE_TAG_NAME:
1568 /* Possible next state: AFTER_CLOSE_TAG_NAME */
1569 advance_to_name_end (context);
1570 add_to_partial (context, context->start, context->iter);
1572 if (context->iter != context->current_text_end)
1573 context->state = STATE_AFTER_CLOSE_TAG_NAME;
1576 case STATE_AFTER_CLOSE_TAG_NAME:
1577 /* Possible next state: AFTER_CLOSE_TAG_SLASH */
1579 skip_spaces (context);
1581 if (context->iter != context->current_text_end)
1583 GString *close_name;
1585 close_name = context->partial_chunk;
1586 context->partial_chunk = NULL;
1588 if (*context->iter != '>')
1594 G_MARKUP_ERROR_PARSE,
1595 _("'%s' is not a valid character following "
1596 "the close element name '%s'; the allowed "
1597 "character is '>'"),
1598 utf8_str (context->iter, buf),
1601 else if (context->tag_stack == NULL)
1605 G_MARKUP_ERROR_PARSE,
1606 _("Element '%s' was closed, no element "
1607 "is currently open"),
1610 else if (strcmp (close_name->str, current_element (context)) != 0)
1614 G_MARKUP_ERROR_PARSE,
1615 _("Element '%s' was closed, but the currently "
1616 "open element is '%s'"),
1618 current_element (context));
1622 advance_char (context);
1623 context->state = STATE_AFTER_CLOSE_ANGLE;
1624 context->start = NULL;
1626 emit_end_element (context, error);
1628 context->partial_chunk = close_name;
1629 truncate_partial (context);
1633 case STATE_INSIDE_PASSTHROUGH:
1634 /* Possible next state: AFTER_CLOSE_ANGLE */
1637 if (*context->iter == '<')
1639 if (*context->iter == '>')
1645 add_to_partial (context, context->start, context->iter);
1646 context->start = context->iter;
1648 str = context->partial_chunk->str;
1649 len = context->partial_chunk->len;
1651 if (str[1] == '?' && str[len - 1] == '?')
1653 if (strncmp (str, "<!--", 4) == 0 &&
1654 strcmp (str + len - 2, "--") == 0)
1656 if (strncmp (str, "<![CDATA[", 9) == 0 &&
1657 strcmp (str + len - 2, "]]") == 0)
1659 if (strncmp (str, "<!DOCTYPE", 9) == 0 &&
1660 context->balance == 0)
1664 while (advance_char (context));
1666 if (context->iter == context->current_text_end)
1668 /* The passthrough hasn't necessarily ended. Merge with
1669 * partial chunk, leave state unchanged.
1671 add_to_partial (context, context->start, context->iter);
1675 /* The passthrough has ended at the close angle. Combine
1676 * it with the partial chunk if any. Call the passthrough
1677 * callback. Note that the open/close angles are
1678 * included in the text of the passthrough.
1680 GError *tmp_error = NULL;
1682 advance_char (context); /* advance past close angle */
1683 add_to_partial (context, context->start, context->iter);
1685 if (context->flags & G_MARKUP_TREAT_CDATA_AS_TEXT &&
1686 strncmp (context->partial_chunk->str, "<![CDATA[", 9) == 0)
1688 if (context->parser->text &&
1689 text_validate (context,
1690 context->partial_chunk->str + 9,
1691 context->partial_chunk->len - 12,
1693 (*context->parser->text) (context,
1694 context->partial_chunk->str + 9,
1695 context->partial_chunk->len - 12,
1699 else if (context->parser->passthrough &&
1700 text_validate (context,
1701 context->partial_chunk->str,
1702 context->partial_chunk->len,
1704 (*context->parser->passthrough) (context,
1705 context->partial_chunk->str,
1706 context->partial_chunk->len,
1710 truncate_partial (context);
1712 if (tmp_error == NULL)
1714 context->state = STATE_AFTER_CLOSE_ANGLE;
1715 context->start = context->iter; /* could begin text */
1718 propagate_error (context, error, tmp_error);
1727 g_assert_not_reached ();
1733 context->parsing = FALSE;
1735 return context->state != STATE_ERROR;
1739 * g_markup_parse_context_end_parse:
1740 * @context: a #GMarkupParseContext
1741 * @error: return location for a #GError
1743 * Signals to the #GMarkupParseContext that all data has been
1744 * fed into the parse context with g_markup_parse_context_parse().
1746 * This function reports an error if the document isn't complete,
1747 * for example if elements are still open.
1749 * Returns: %TRUE on success, %FALSE if an error was set
1752 g_markup_parse_context_end_parse (GMarkupParseContext *context,
1755 g_return_val_if_fail (context != NULL, FALSE);
1756 g_return_val_if_fail (!context->parsing, FALSE);
1757 g_return_val_if_fail (context->state != STATE_ERROR, FALSE);
1759 if (context->partial_chunk != NULL)
1761 g_string_free (context->partial_chunk, TRUE);
1762 context->partial_chunk = NULL;
1765 if (context->document_empty)
1767 set_error_literal (context, error, G_MARKUP_ERROR_EMPTY,
1768 _("Document was empty or contained only whitespace"));
1772 context->parsing = TRUE;
1774 switch (context->state)
1780 case STATE_AFTER_OPEN_ANGLE:
1781 set_error_literal (context, error, G_MARKUP_ERROR_PARSE,
1782 _("Document ended unexpectedly just after an open angle bracket '<'"));
1785 case STATE_AFTER_CLOSE_ANGLE:
1786 if (context->tag_stack != NULL)
1788 /* Error message the same as for INSIDE_TEXT */
1789 set_error (context, error, G_MARKUP_ERROR_PARSE,
1790 _("Document ended unexpectedly with elements still open - "
1791 "'%s' was the last element opened"),
1792 current_element (context));
1796 case STATE_AFTER_ELISION_SLASH:
1797 set_error (context, error, G_MARKUP_ERROR_PARSE,
1798 _("Document ended unexpectedly, expected to see a close angle "
1799 "bracket ending the tag <%s/>"), current_element (context));
1802 case STATE_INSIDE_OPEN_TAG_NAME:
1803 set_error_literal (context, error, G_MARKUP_ERROR_PARSE,
1804 _("Document ended unexpectedly inside an element name"));
1807 case STATE_INSIDE_ATTRIBUTE_NAME:
1808 case STATE_AFTER_ATTRIBUTE_NAME:
1809 set_error_literal (context, error, G_MARKUP_ERROR_PARSE,
1810 _("Document ended unexpectedly inside an attribute name"));
1813 case STATE_BETWEEN_ATTRIBUTES:
1814 set_error_literal (context, error, G_MARKUP_ERROR_PARSE,
1815 _("Document ended unexpectedly inside an element-opening "
1819 case STATE_AFTER_ATTRIBUTE_EQUALS_SIGN:
1820 set_error_literal (context, error, G_MARKUP_ERROR_PARSE,
1821 _("Document ended unexpectedly after the equals sign "
1822 "following an attribute name; no attribute value"));
1825 case STATE_INSIDE_ATTRIBUTE_VALUE_SQ:
1826 case STATE_INSIDE_ATTRIBUTE_VALUE_DQ:
1827 set_error_literal (context, error, G_MARKUP_ERROR_PARSE,
1828 _("Document ended unexpectedly while inside an attribute "
1832 case STATE_INSIDE_TEXT:
1833 g_assert (context->tag_stack != NULL);
1834 set_error (context, error, G_MARKUP_ERROR_PARSE,
1835 _("Document ended unexpectedly with elements still open - "
1836 "'%s' was the last element opened"),
1837 current_element (context));
1840 case STATE_AFTER_CLOSE_TAG_SLASH:
1841 case STATE_INSIDE_CLOSE_TAG_NAME:
1842 case STATE_AFTER_CLOSE_TAG_NAME:
1843 set_error (context, error, G_MARKUP_ERROR_PARSE,
1844 _("Document ended unexpectedly inside the close tag for "
1845 "element '%s'"), current_element (context));
1848 case STATE_INSIDE_PASSTHROUGH:
1849 set_error_literal (context, error, G_MARKUP_ERROR_PARSE,
1850 _("Document ended unexpectedly inside a comment or "
1851 "processing instruction"));
1856 g_assert_not_reached ();
1860 context->parsing = FALSE;
1862 return context->state != STATE_ERROR;
1866 * g_markup_parse_context_get_element:
1867 * @context: a #GMarkupParseContext
1869 * Retrieves the name of the currently open element.
1871 * If called from the start_element or end_element handlers this will
1872 * give the element_name as passed to those functions. For the parent
1873 * elements, see g_markup_parse_context_get_element_stack().
1875 * Returns: the name of the currently open element, or %NULL
1880 g_markup_parse_context_get_element (GMarkupParseContext *context)
1882 g_return_val_if_fail (context != NULL, NULL);
1884 if (context->tag_stack == NULL)
1887 return current_element (context);
1891 * g_markup_parse_context_get_element_stack:
1892 * @context: a #GMarkupParseContext
1894 * Retrieves the element stack from the internal state of the parser.
1896 * The returned #GSList is a list of strings where the first item is
1897 * the currently open tag (as would be returned by
1898 * g_markup_parse_context_get_element()) and the next item is its
1901 * This function is intended to be used in the start_element and
1902 * end_element handlers where g_markup_parse_context_get_element()
1903 * would merely return the name of the element that is being
1906 * Returns: the element stack, which must not be modified
1911 g_markup_parse_context_get_element_stack (GMarkupParseContext *context)
1913 g_return_val_if_fail (context != NULL, NULL);
1914 return context->tag_stack;
1918 * g_markup_parse_context_get_position:
1919 * @context: a #GMarkupParseContext
1920 * @line_number: (allow-none): return location for a line number, or %NULL
1921 * @char_number: (allow-none): return location for a char-on-line number, or %NULL
1923 * Retrieves the current line number and the number of the character on
1924 * that line. Intended for use in error messages; there are no strict
1925 * semantics for what constitutes the "current" line number other than
1926 * "the best number we could come up with for error messages."
1929 g_markup_parse_context_get_position (GMarkupParseContext *context,
1933 g_return_if_fail (context != NULL);
1936 *line_number = context->line_number;
1939 *char_number = context->char_number;
1943 * g_markup_parse_context_get_user_data:
1944 * @context: a #GMarkupParseContext
1946 * Returns the user_data associated with @context.
1948 * This will either be the user_data that was provided to
1949 * g_markup_parse_context_new() or to the most recent call
1950 * of g_markup_parse_context_push().
1952 * Returns: the provided user_data. The returned data belongs to
1953 * the markup context and will be freed when
1954 * g_markup_parse_context_free() is called.
1959 g_markup_parse_context_get_user_data (GMarkupParseContext *context)
1961 return context->user_data;
1965 * g_markup_parse_context_push:
1966 * @context: a #GMarkupParseContext
1967 * @parser: a #GMarkupParser
1968 * @user_data: user data to pass to #GMarkupParser functions
1970 * Temporarily redirects markup data to a sub-parser.
1972 * This function may only be called from the start_element handler of
1973 * a #GMarkupParser. It must be matched with a corresponding call to
1974 * g_markup_parse_context_pop() in the matching end_element handler
1975 * (except in the case that the parser aborts due to an error).
1977 * All tags, text and other data between the matching tags is
1978 * redirected to the subparser given by @parser. @user_data is used
1979 * as the user_data for that parser. @user_data is also passed to the
1980 * error callback in the event that an error occurs. This includes
1981 * errors that occur in subparsers of the subparser.
1983 * The end tag matching the start tag for which this call was made is
1984 * handled by the previous parser (which is given its own user_data)
1985 * which is why g_markup_parse_context_pop() is provided to allow "one
1986 * last access" to the @user_data provided to this function. In the
1987 * case of error, the @user_data provided here is passed directly to
1988 * the error callback of the subparser and g_markup_parse_context_pop()
1989 * should not be called. In either case, if @user_data was allocated
1990 * then it ought to be freed from both of these locations.
1992 * This function is not intended to be directly called by users
1993 * interested in invoking subparsers. Instead, it is intended to be
1994 * used by the subparsers themselves to implement a higher-level
1997 * As an example, see the following implementation of a simple
1998 * parser that counts the number of tags encountered.
2000 * |[<!-- language="C" -->
2007 * counter_start_element (GMarkupParseContext *context,
2008 * const gchar *element_name,
2009 * const gchar **attribute_names,
2010 * const gchar **attribute_values,
2011 * gpointer user_data,
2014 * CounterData *data = user_data;
2016 * data->tag_count++;
2020 * counter_error (GMarkupParseContext *context,
2022 * gpointer user_data)
2024 * CounterData *data = user_data;
2026 * g_slice_free (CounterData, data);
2029 * static GMarkupParser counter_subparser =
2031 * counter_start_element,
2039 * In order to allow this parser to be easily used as a subparser, the
2040 * following interface is provided:
2042 * |[<!-- language="C" -->
2044 * start_counting (GMarkupParseContext *context)
2046 * CounterData *data = g_slice_new (CounterData);
2048 * data->tag_count = 0;
2049 * g_markup_parse_context_push (context, &counter_subparser, data);
2053 * end_counting (GMarkupParseContext *context)
2055 * CounterData *data = g_markup_parse_context_pop (context);
2058 * result = data->tag_count;
2059 * g_slice_free (CounterData, data);
2065 * The subparser would then be used as follows:
2067 * |[<!-- language="C" -->
2068 * static void start_element (context, element_name, ...)
2070 * if (strcmp (element_name, "count-these") == 0)
2071 * start_counting (context);
2073 * // else, handle other tags...
2076 * static void end_element (context, element_name, ...)
2078 * if (strcmp (element_name, "count-these") == 0)
2079 * g_print ("Counted %d tags\n", end_counting (context));
2081 * // else, handle other tags...
2088 g_markup_parse_context_push (GMarkupParseContext *context,
2089 const GMarkupParser *parser,
2092 GMarkupRecursionTracker *tracker;
2094 tracker = g_slice_new (GMarkupRecursionTracker);
2095 tracker->prev_element = context->subparser_element;
2096 tracker->prev_parser = context->parser;
2097 tracker->prev_user_data = context->user_data;
2099 context->subparser_element = current_element (context);
2100 context->parser = parser;
2101 context->user_data = user_data;
2103 context->subparser_stack = g_slist_prepend (context->subparser_stack,
2108 * g_markup_parse_context_pop:
2109 * @context: a #GMarkupParseContext
2111 * Completes the process of a temporary sub-parser redirection.
2113 * This function exists to collect the user_data allocated by a
2114 * matching call to g_markup_parse_context_push(). It must be called
2115 * in the end_element handler corresponding to the start_element
2116 * handler during which g_markup_parse_context_push() was called.
2117 * You must not call this function from the error callback -- the
2118 * @user_data is provided directly to the callback in that case.
2120 * This function is not intended to be directly called by users
2121 * interested in invoking subparsers. Instead, it is intended to
2122 * be used by the subparsers themselves to implement a higher-level
2125 * Returns: the user data passed to g_markup_parse_context_push()
2130 g_markup_parse_context_pop (GMarkupParseContext *context)
2134 if (!context->awaiting_pop)
2135 possibly_finish_subparser (context);
2137 g_assert (context->awaiting_pop);
2139 context->awaiting_pop = FALSE;
2141 /* valgrind friendliness */
2142 user_data = context->held_user_data;
2143 context->held_user_data = NULL;
2149 append_escaped_text (GString *str,
2158 end = text + length;
2163 next = g_utf8_next_char (p);
2168 g_string_append (str, "&");
2172 g_string_append (str, "<");
2176 g_string_append (str, ">");
2180 g_string_append (str, "'");
2184 g_string_append (str, """);
2188 c = g_utf8_get_char (p);
2189 if ((0x1 <= c && c <= 0x8) ||
2190 (0xb <= c && c <= 0xc) ||
2191 (0xe <= c && c <= 0x1f) ||
2192 (0x7f <= c && c <= 0x84) ||
2193 (0x86 <= c && c <= 0x9f))
2194 g_string_append_printf (str, "&#x%x;", c);
2196 g_string_append_len (str, p, next - p);
2205 * g_markup_escape_text:
2206 * @text: some valid UTF-8 text
2207 * @length: length of @text in bytes, or -1 if the text is nul-terminated
2209 * Escapes text so that the markup parser will parse it verbatim.
2210 * Less than, greater than, ampersand, etc. are replaced with the
2211 * corresponding entities. This function would typically be used
2212 * when writing out a file to be parsed with the markup parser.
2214 * Note that this function doesn't protect whitespace and line endings
2215 * from being processed according to the XML rules for normalization
2216 * of line endings and attribute values.
2218 * Note also that this function will produce character references in
2219 * the range of  ...  for all control sequences
2220 * except for tabstop, newline and carriage return. The character
2221 * references in this range are not valid XML 1.0, but they are
2222 * valid XML 1.1 and will be accepted by the GMarkup parser.
2224 * Returns: a newly allocated string with the escaped text
2227 g_markup_escape_text (const gchar *text,
2232 g_return_val_if_fail (text != NULL, NULL);
2235 length = strlen (text);
2237 /* prealloc at least as long as original text */
2238 str = g_string_sized_new (length);
2239 append_escaped_text (str, text, length);
2241 return g_string_free (str, FALSE);
2246 * @format: a printf-style format string
2247 * @after: location to store a pointer to the character after
2248 * the returned conversion. On a %NULL return, returns the
2249 * pointer to the trailing NUL in the string
2251 * Find the next conversion in a printf-style format string.
2252 * Partially based on code from printf-parser.c,
2253 * Copyright (C) 1999-2000, 2002-2003 Free Software Foundation, Inc.
2255 * Returns: pointer to the next conversion in @format,
2256 * or %NULL, if none.
2259 find_conversion (const char *format,
2262 const char *start = format;
2265 while (*start != '\0' && *start != '%')
2282 /* Test for positional argument. */
2283 if (*cp >= '0' && *cp <= '9')
2287 for (np = cp; *np >= '0' && *np <= '9'; np++)
2293 /* Skip the flags. */
2307 /* Skip the field width. */
2312 /* Test for positional argument. */
2313 if (*cp >= '0' && *cp <= '9')
2317 for (np = cp; *np >= '0' && *np <= '9'; np++)
2325 for (; *cp >= '0' && *cp <= '9'; cp++)
2329 /* Skip the precision. */
2335 /* Test for positional argument. */
2336 if (*cp >= '0' && *cp <= '9')
2340 for (np = cp; *np >= '0' && *np <= '9'; np++)
2348 for (; *cp >= '0' && *cp <= '9'; cp++)
2353 /* Skip argument type/size specifiers. */
2354 while (*cp == 'h' ||
2363 /* Skip the conversion character. */
2371 * g_markup_vprintf_escaped:
2372 * @format: printf() style format string
2373 * @args: variable argument list, similar to vprintf()
2375 * Formats the data in @args according to @format, escaping
2376 * all string and character arguments in the fashion
2377 * of g_markup_escape_text(). See g_markup_printf_escaped().
2379 * Returns: newly allocated result from formatting
2380 * operation. Free with g_free().
2384 #pragma GCC diagnostic push
2385 #pragma GCC diagnostic ignored "-Wformat-nonliteral"
2388 g_markup_vprintf_escaped (const gchar *format,
2393 GString *result = NULL;
2394 gchar *output1 = NULL;
2395 gchar *output2 = NULL;
2396 const char *p, *op1, *op2;
2399 /* The technique here, is that we make two format strings that
2400 * have the identical conversions in the identical order to the
2401 * original strings, but differ in the text in-between. We
2402 * then use the normal g_strdup_vprintf() to format the arguments
2403 * with the two new format strings. By comparing the results,
2404 * we can figure out what segments of the output come from
2405 * the original format string, and what from the arguments,
2406 * and thus know what portions of the string to escape.
2408 * For instance, for:
2410 * g_markup_printf_escaped ("%s ate %d apples", "Susan & Fred", 5);
2412 * We form the two format strings "%sX%dX" and %sY%sY". The results
2413 * of formatting with those two strings are
2415 * "%sX%dX" => "Susan & FredX5X"
2416 * "%sY%dY" => "Susan & FredY5Y"
2418 * To find the span of the first argument, we find the first position
2419 * where the two arguments differ, which tells us that the first
2420 * argument formatted to "Susan & Fred". We then escape that
2421 * to "Susan & Fred" and join up with the intermediate portions
2422 * of the format string and the second argument to get
2423 * "Susan & Fred ate 5 apples".
2426 /* Create the two modified format strings
2428 format1 = g_string_new (NULL);
2429 format2 = g_string_new (NULL);
2434 const char *conv = find_conversion (p, &after);
2438 g_string_append_len (format1, conv, after - conv);
2439 g_string_append_c (format1, 'X');
2440 g_string_append_len (format2, conv, after - conv);
2441 g_string_append_c (format2, 'Y');
2446 /* Use them to format the arguments
2448 G_VA_COPY (args2, args);
2450 output1 = g_strdup_vprintf (format1->str, args);
2458 output2 = g_strdup_vprintf (format2->str, args2);
2462 result = g_string_new (NULL);
2464 /* Iterate through the original format string again,
2465 * copying the non-conversion portions and the escaped
2466 * converted arguments to the output string.
2474 const char *output_start;
2475 const char *conv = find_conversion (p, &after);
2478 if (!conv) /* The end, after points to the trailing \0 */
2480 g_string_append_len (result, p, after - p);
2484 g_string_append_len (result, p, conv - p);
2486 while (*op1 == *op2)
2492 escaped = g_markup_escape_text (output_start, op1 - output_start);
2493 g_string_append (result, escaped);
2502 g_string_free (format1, TRUE);
2503 g_string_free (format2, TRUE);
2508 return g_string_free (result, FALSE);
2513 #pragma GCC diagnostic pop
2516 * g_markup_printf_escaped:
2517 * @format: printf() style format string
2518 * @...: the arguments to insert in the format string
2520 * Formats arguments according to @format, escaping
2521 * all string and character arguments in the fashion
2522 * of g_markup_escape_text(). This is useful when you
2523 * want to insert literal strings into XML-style markup
2524 * output, without having to worry that the strings
2525 * might themselves contain markup.
2527 * |[<!-- language="C" -->
2528 * const char *store = "Fortnum & Mason";
2529 * const char *item = "Tea";
2532 * output = g_markup_printf_escaped ("<purchase>"
2533 * "<store>%s</store>"
2539 * Returns: newly allocated result from formatting
2540 * operation. Free with g_free().
2545 g_markup_printf_escaped (const gchar *format, ...)
2550 va_start (args, format);
2551 result = g_markup_vprintf_escaped (format, args);
2558 g_markup_parse_boolean (const char *string,
2561 char const * const falses[] = { "false", "f", "no", "n", "0" };
2562 char const * const trues[] = { "true", "t", "yes", "y", "1" };
2565 for (i = 0; i < G_N_ELEMENTS (falses); i++)
2567 if (g_ascii_strcasecmp (string, falses[i]) == 0)
2576 for (i = 0; i < G_N_ELEMENTS (trues); i++)
2578 if (g_ascii_strcasecmp (string, trues[i]) == 0)
2591 * GMarkupCollectType:
2592 * @G_MARKUP_COLLECT_INVALID: used to terminate the list of attributes
2594 * @G_MARKUP_COLLECT_STRING: collect the string pointer directly from
2595 * the attribute_values[] array. Expects a parameter of type (const
2596 * char **). If %G_MARKUP_COLLECT_OPTIONAL is specified and the
2597 * attribute isn't present then the pointer will be set to %NULL
2598 * @G_MARKUP_COLLECT_STRDUP: as with %G_MARKUP_COLLECT_STRING, but
2599 * expects a parameter of type (char **) and g_strdup()s the
2600 * returned pointer. The pointer must be freed with g_free()
2601 * @G_MARKUP_COLLECT_BOOLEAN: expects a parameter of type (gboolean *)
2602 * and parses the attribute value as a boolean. Sets %FALSE if the
2603 * attribute isn't present. Valid boolean values consist of
2604 * (case-insensitive) "false", "f", "no", "n", "0" and "true", "t",
2606 * @G_MARKUP_COLLECT_TRISTATE: as with %G_MARKUP_COLLECT_BOOLEAN, but
2607 * in the case of a missing attribute a value is set that compares
2608 * equal to neither %FALSE nor %TRUE G_MARKUP_COLLECT_OPTIONAL is
2610 * @G_MARKUP_COLLECT_OPTIONAL: can be bitwise ORed with the other fields.
2611 * If present, allows the attribute not to appear. A default value
2612 * is set depending on what value type is used
2614 * A mixed enumerated type and flags field. You must specify one type
2615 * (string, strdup, boolean, tristate). Additionally, you may optionally
2616 * bitwise OR the type with the flag %G_MARKUP_COLLECT_OPTIONAL.
2618 * It is likely that this enum will be extended in the future to
2619 * support other types.
2623 * g_markup_collect_attributes:
2624 * @element_name: the current tag name
2625 * @attribute_names: the attribute names
2626 * @attribute_values: the attribute values
2627 * @error: a pointer to a #GError or %NULL
2628 * @first_type: the #GMarkupCollectType of the first attribute
2629 * @first_attr: the name of the first attribute
2630 * @...: a pointer to the storage location of the first attribute
2631 * (or %NULL), followed by more types names and pointers, ending
2632 * with %G_MARKUP_COLLECT_INVALID
2634 * Collects the attributes of the element from the data passed to the
2635 * #GMarkupParser start_element function, dealing with common error
2636 * conditions and supporting boolean values.
2638 * This utility function is not required to write a parser but can save
2641 * The @element_name, @attribute_names, @attribute_values and @error
2642 * parameters passed to the start_element callback should be passed
2643 * unmodified to this function.
2645 * Following these arguments is a list of "supported" attributes to collect.
2646 * It is an error to specify multiple attributes with the same name. If any
2647 * attribute not in the list appears in the @attribute_names array then an
2648 * unknown attribute error will result.
2650 * The #GMarkupCollectType field allows specifying the type of collection
2651 * to perform and if a given attribute must appear or is optional.
2653 * The attribute name is simply the name of the attribute to collect.
2655 * The pointer should be of the appropriate type (see the descriptions
2656 * under #GMarkupCollectType) and may be %NULL in case a particular
2657 * attribute is to be allowed but ignored.
2659 * This function deals with issuing errors for missing attributes
2660 * (of type %G_MARKUP_ERROR_MISSING_ATTRIBUTE), unknown attributes
2661 * (of type %G_MARKUP_ERROR_UNKNOWN_ATTRIBUTE) and duplicate
2662 * attributes (of type %G_MARKUP_ERROR_INVALID_CONTENT) as well
2663 * as parse errors for boolean-valued attributes (again of type
2664 * %G_MARKUP_ERROR_INVALID_CONTENT). In all of these cases %FALSE
2665 * will be returned and @error will be set as appropriate.
2667 * Returns: %TRUE if successful
2672 g_markup_collect_attributes (const gchar *element_name,
2673 const gchar **attribute_names,
2674 const gchar **attribute_values,
2676 GMarkupCollectType first_type,
2677 const gchar *first_attr,
2680 GMarkupCollectType type;
2692 va_start (ap, first_attr);
2693 while (type != G_MARKUP_COLLECT_INVALID)
2698 mandatory = !(type & G_MARKUP_COLLECT_OPTIONAL);
2699 type &= (G_MARKUP_COLLECT_OPTIONAL - 1);
2701 /* tristate records a value != TRUE and != FALSE
2702 * for the case where the attribute is missing
2704 if (type == G_MARKUP_COLLECT_TRISTATE)
2707 for (i = 0; attribute_names[i]; i++)
2708 if (i >= 40 || !(collected & (G_GUINT64_CONSTANT(1) << i)))
2709 if (!strcmp (attribute_names[i], attr))
2712 /* ISO C99 only promises that the user can pass up to 127 arguments.
2713 * Subtracting the first 4 arguments plus the final NULL and dividing
2714 * by 3 arguments per collected attribute, we are left with a maximum
2715 * number of supported attributes of (127 - 5) / 3 = 40.
2717 * In reality, nobody is ever going to call us with anywhere close to
2718 * 40 attributes to collect, so it is safe to assume that if i > 40
2719 * then the user has given some invalid or repeated arguments. These
2720 * problems will be caught and reported at the end of the function.
2722 * We know at this point that we have an error, but we don't know
2723 * what error it is, so just continue...
2726 collected |= (G_GUINT64_CONSTANT(1) << i);
2728 value = attribute_values[i];
2730 if (value == NULL && mandatory)
2732 g_set_error (error, G_MARKUP_ERROR,
2733 G_MARKUP_ERROR_MISSING_ATTRIBUTE,
2734 "element '%s' requires attribute '%s'",
2735 element_name, attr);
2743 case G_MARKUP_COLLECT_STRING:
2745 const char **str_ptr;
2747 str_ptr = va_arg (ap, const char **);
2749 if (str_ptr != NULL)
2754 case G_MARKUP_COLLECT_STRDUP:
2758 str_ptr = va_arg (ap, char **);
2760 if (str_ptr != NULL)
2761 *str_ptr = g_strdup (value);
2765 case G_MARKUP_COLLECT_BOOLEAN:
2766 case G_MARKUP_COLLECT_TRISTATE:
2771 bool_ptr = va_arg (ap, gboolean *);
2773 if (bool_ptr != NULL)
2775 if (type == G_MARKUP_COLLECT_TRISTATE)
2776 /* constructivists rejoice!
2777 * neither false nor true...
2781 else /* G_MARKUP_COLLECT_BOOLEAN */
2787 if (!g_markup_parse_boolean (value, va_arg (ap, gboolean *)))
2789 g_set_error (error, G_MARKUP_ERROR,
2790 G_MARKUP_ERROR_INVALID_CONTENT,
2791 "element '%s', attribute '%s', value '%s' "
2792 "cannot be parsed as a boolean value",
2793 element_name, attr, value);
2803 g_assert_not_reached ();
2806 type = va_arg (ap, GMarkupCollectType);
2807 attr = va_arg (ap, const char *);
2812 /* ensure we collected all the arguments */
2813 for (i = 0; attribute_names[i]; i++)
2814 if ((collected & (G_GUINT64_CONSTANT(1) << i)) == 0)
2816 /* attribute not collected: could be caused by two things.
2818 * 1) it doesn't exist in our list of attributes
2819 * 2) it existed but was matched by a duplicate attribute earlier
2825 for (j = 0; j < i; j++)
2826 if (strcmp (attribute_names[i], attribute_names[j]) == 0)
2830 /* j is now the first occurrence of attribute_names[i] */
2832 g_set_error (error, G_MARKUP_ERROR,
2833 G_MARKUP_ERROR_UNKNOWN_ATTRIBUTE,
2834 "attribute '%s' invalid for element '%s'",
2835 attribute_names[i], element_name);
2837 g_set_error (error, G_MARKUP_ERROR,
2838 G_MARKUP_ERROR_INVALID_CONTENT,
2839 "attribute '%s' given multiple times for element '%s'",
2840 attribute_names[i], element_name);
2848 /* replay the above to free allocations */
2852 va_start (ap, first_attr);
2853 while (type != G_MARKUP_COLLECT_INVALID)
2857 ptr = va_arg (ap, gpointer);
2861 switch (type & (G_MARKUP_COLLECT_OPTIONAL - 1))
2863 case G_MARKUP_COLLECT_STRDUP:
2865 g_free (*(char **) ptr);
2867 case G_MARKUP_COLLECT_STRING:
2868 *(char **) ptr = NULL;
2871 case G_MARKUP_COLLECT_BOOLEAN:
2872 *(gboolean *) ptr = FALSE;
2875 case G_MARKUP_COLLECT_TRISTATE:
2876 *(gboolean *) ptr = -1;
2881 type = va_arg (ap, GMarkupCollectType);
2882 attr = va_arg (ap, const char *);