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 * write to the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
19 * Boston, MA 02111-1307, USA.
35 g_markup_error_quark (void)
37 return g_quark_from_static_string ("g-markup-error-quark");
43 STATE_AFTER_OPEN_ANGLE,
44 STATE_AFTER_CLOSE_ANGLE,
45 STATE_AFTER_ELISION_SLASH, /* the slash that obviates need for end element */
46 STATE_INSIDE_OPEN_TAG_NAME,
47 STATE_INSIDE_ATTRIBUTE_NAME,
48 STATE_AFTER_ATTRIBUTE_NAME,
49 STATE_BETWEEN_ATTRIBUTES,
50 STATE_AFTER_ATTRIBUTE_EQUALS_SIGN,
51 STATE_INSIDE_ATTRIBUTE_VALUE_SQ,
52 STATE_INSIDE_ATTRIBUTE_VALUE_DQ,
54 STATE_AFTER_CLOSE_TAG_SLASH,
55 STATE_INSIDE_CLOSE_TAG_NAME,
56 STATE_AFTER_CLOSE_TAG_NAME,
57 STATE_INSIDE_PASSTHROUGH,
63 const char *prev_element;
64 const GMarkupParser *prev_parser;
65 gpointer prev_user_data;
66 } GMarkupRecursionTracker;
68 struct _GMarkupParseContext
70 const GMarkupParser *parser;
72 GMarkupParseFlags flags;
78 GDestroyNotify dnotify;
80 /* A piece of character data or an element that
81 * hasn't "ended" yet so we haven't yet called
82 * the callback for it.
84 GString *partial_chunk;
86 GMarkupParseState state;
93 const gchar *current_text;
94 gssize current_text_len;
95 const gchar *current_text_end;
97 GString *leftover_char_portion;
99 /* used to save the start of the last interesting thingy */
104 guint document_empty : 1;
106 guint awaiting_pop : 1;
109 /* subparser support */
110 GSList *subparser_stack; /* (GMarkupRecursionTracker *) */
111 const char *subparser_element;
112 gpointer held_user_data;
116 * g_markup_parse_context_new:
117 * @parser: a #GMarkupParser
118 * @flags: one or more #GMarkupParseFlags
119 * @user_data: user data to pass to #GMarkupParser functions
120 * @user_data_dnotify: user data destroy notifier called when the parse context is freed
122 * Creates a new parse context. A parse context is used to parse
123 * marked-up documents. You can feed any number of documents into
124 * a context, as long as no errors occur; once an error occurs,
125 * the parse context can't continue to parse text (you have to free it
126 * and create a new parse context).
128 * Return value: a new #GMarkupParseContext
130 GMarkupParseContext *
131 g_markup_parse_context_new (const GMarkupParser *parser,
132 GMarkupParseFlags flags,
134 GDestroyNotify user_data_dnotify)
136 GMarkupParseContext *context;
138 g_return_val_if_fail (parser != NULL, NULL);
140 context = g_new (GMarkupParseContext, 1);
142 context->parser = parser;
143 context->flags = flags;
144 context->user_data = user_data;
145 context->dnotify = user_data_dnotify;
147 context->line_number = 1;
148 context->char_number = 1;
150 context->partial_chunk = NULL;
152 context->state = STATE_START;
153 context->tag_stack = NULL;
154 context->attr_names = NULL;
155 context->attr_values = NULL;
156 context->cur_attr = -1;
157 context->alloc_attrs = 0;
159 context->current_text = NULL;
160 context->current_text_len = -1;
161 context->current_text_end = NULL;
162 context->leftover_char_portion = NULL;
164 context->start = NULL;
165 context->iter = NULL;
167 context->document_empty = TRUE;
168 context->parsing = FALSE;
170 context->awaiting_pop = FALSE;
171 context->subparser_stack = NULL;
172 context->subparser_element = NULL;
174 /* this is only looked at if awaiting_pop = TRUE. initialise anyway. */
175 context->held_user_data = NULL;
177 context->balance = 0;
183 * g_markup_parse_context_free:
184 * @context: a #GMarkupParseContext
186 * Frees a #GMarkupParseContext. Can't be called from inside
187 * one of the #GMarkupParser functions. Can't be called while
188 * a subparser is pushed.
191 g_markup_parse_context_free (GMarkupParseContext *context)
193 g_return_if_fail (context != NULL);
194 g_return_if_fail (!context->parsing);
195 g_return_if_fail (!context->subparser_stack);
196 g_return_if_fail (!context->awaiting_pop);
198 if (context->dnotify)
199 (* context->dnotify) (context->user_data);
201 g_strfreev (context->attr_names);
202 g_strfreev (context->attr_values);
204 g_slist_foreach (context->tag_stack, (GFunc)g_free, NULL);
205 g_slist_free (context->tag_stack);
207 if (context->partial_chunk)
208 g_string_free (context->partial_chunk, TRUE);
210 if (context->leftover_char_portion)
211 g_string_free (context->leftover_char_portion, TRUE);
216 static void pop_subparser_stack (GMarkupParseContext *context);
219 mark_error (GMarkupParseContext *context,
222 context->state = STATE_ERROR;
224 if (context->parser->error)
225 (*context->parser->error) (context, error, context->user_data);
227 /* report the error all the way up to free all the user-data */
228 while (context->subparser_stack)
230 pop_subparser_stack (context);
231 context->awaiting_pop = FALSE; /* already been freed */
233 if (context->parser->error)
234 (*context->parser->error) (context, error, context->user_data);
238 static void set_error (GMarkupParseContext *context,
242 ...) G_GNUC_PRINTF (4, 5);
245 set_error (GMarkupParseContext *context,
256 va_start (args, format);
257 s = g_strdup_vprintf (format, args);
260 /* Make sure that the GError message is valid UTF-8 even if it is
261 * complaining about invalid UTF-8 in the markup: */
262 s_valid = _g_utf8_make_valid (s);
263 tmp_error = g_error_new_literal (G_MARKUP_ERROR, code, s_valid);
268 g_prefix_error (&tmp_error,
269 _("Error on line %d char %d: "),
270 context->line_number,
271 context->char_number);
273 mark_error (context, tmp_error);
275 g_propagate_error (error, tmp_error);
279 propagate_error (GMarkupParseContext *context,
283 if (context->flags & G_MARKUP_PREFIX_ERROR_POSITION)
284 g_prefix_error (&src,
285 _("Error on line %d char %d: "),
286 context->line_number,
287 context->char_number);
289 mark_error (context, src);
291 g_propagate_error (dest, src);
294 /* To make these faster, we first use the ascii-only tests, then check
295 * for the usual non-alnum name-end chars, and only then call the
296 * expensive unicode stuff. Nobody uses non-ascii in XML tag/attribute
297 * names, so this is a reasonable hack that virtually always avoids
300 #define IS_COMMON_NAME_END_CHAR(c) \
301 ((c) == '=' || (c) == '/' || (c) == '>' || (c) == ' ')
304 is_name_start_char (const gchar *p)
306 if (g_ascii_isalpha (*p) ||
307 (!IS_COMMON_NAME_END_CHAR (*p) &&
310 g_unichar_isalpha (g_utf8_get_char (p)))))
317 is_name_char (const gchar *p)
319 if (g_ascii_isalnum (*p) ||
320 (!IS_COMMON_NAME_END_CHAR (*p) &&
325 g_unichar_isalpha (g_utf8_get_char (p)))))
333 char_str (gunichar c,
337 g_unichar_to_utf8 (c, buf);
342 utf8_str (const gchar *utf8,
345 char_str (g_utf8_get_char (utf8), buf);
350 set_unescape_error (GMarkupParseContext *context,
352 const gchar *remaining_text,
353 const gchar *remaining_text_end,
361 gint remaining_newlines;
364 remaining_newlines = 0;
366 while (p != remaining_text_end)
369 ++remaining_newlines;
373 va_start (args, format);
374 s = g_strdup_vprintf (format, args);
377 tmp_error = g_error_new (G_MARKUP_ERROR,
379 _("Error on line %d: %s"),
380 context->line_number - remaining_newlines,
385 mark_error (context, tmp_error);
387 g_propagate_error (error, tmp_error);
393 USTATE_AFTER_AMPERSAND,
394 USTATE_INSIDE_ENTITY_NAME,
395 USTATE_AFTER_CHARREF_HASH
400 GMarkupParseContext *context;
404 const gchar *text_end;
405 const gchar *entity_start;
409 unescape_text_state_inside_text (UnescapeContext *ucontext,
414 gboolean normalize_attribute;
416 if (ucontext->context->state == STATE_INSIDE_ATTRIBUTE_VALUE_SQ ||
417 ucontext->context->state == STATE_INSIDE_ATTRIBUTE_VALUE_DQ)
418 normalize_attribute = TRUE;
420 normalize_attribute = FALSE;
424 while (p != ucontext->text_end)
430 else if (normalize_attribute && (*p == '\t' || *p == '\n'))
432 g_string_append_len (ucontext->str, start, p - start);
433 g_string_append_c (ucontext->str, ' ');
434 p = g_utf8_next_char (p);
439 g_string_append_len (ucontext->str, start, p - start);
440 g_string_append_c (ucontext->str, normalize_attribute ? ' ' : '\n');
441 p = g_utf8_next_char (p);
442 if (p != ucontext->text_end && *p == '\n')
443 p = g_utf8_next_char (p);
447 p = g_utf8_next_char (p);
451 g_string_append_len (ucontext->str, start, p - start);
453 if (p != ucontext->text_end && *p == '&')
455 p = g_utf8_next_char (p);
456 ucontext->state = USTATE_AFTER_AMPERSAND;
463 unescape_text_state_after_ampersand (UnescapeContext *ucontext,
467 ucontext->entity_start = NULL;
471 p = g_utf8_next_char (p);
473 ucontext->entity_start = p;
474 ucontext->state = USTATE_AFTER_CHARREF_HASH;
476 else if (!is_name_start_char (p))
480 set_unescape_error (ucontext->context, error,
481 p, ucontext->text_end,
482 G_MARKUP_ERROR_PARSE,
483 _("Empty entity '&;' seen; valid "
484 "entities are: & " < > '"));
490 set_unescape_error (ucontext->context, error,
491 p, ucontext->text_end,
492 G_MARKUP_ERROR_PARSE,
493 _("Character '%s' is not valid at "
494 "the start of an entity name; "
495 "the & character begins an entity; "
496 "if this ampersand isn't supposed "
497 "to be an entity, escape it as "
504 ucontext->entity_start = p;
505 ucontext->state = USTATE_INSIDE_ENTITY_NAME;
512 unescape_text_state_inside_entity_name (UnescapeContext *ucontext,
516 while (p != ucontext->text_end)
520 else if (!is_name_char (p))
524 set_unescape_error (ucontext->context, error,
525 p, ucontext->text_end,
526 G_MARKUP_ERROR_PARSE,
527 _("Character '%s' is not valid "
528 "inside an entity name"),
533 p = g_utf8_next_char (p);
536 if (ucontext->context->state != STATE_ERROR)
538 if (p != ucontext->text_end)
540 gint len = p - ucontext->entity_start;
542 /* move to after semicolon */
543 p = g_utf8_next_char (p);
544 ucontext->state = USTATE_INSIDE_TEXT;
546 if (strncmp (ucontext->entity_start, "lt", len) == 0)
547 g_string_append_c (ucontext->str, '<');
548 else if (strncmp (ucontext->entity_start, "gt", len) == 0)
549 g_string_append_c (ucontext->str, '>');
550 else if (strncmp (ucontext->entity_start, "amp", len) == 0)
551 g_string_append_c (ucontext->str, '&');
552 else if (strncmp (ucontext->entity_start, "quot", len) == 0)
553 g_string_append_c (ucontext->str, '"');
554 else if (strncmp (ucontext->entity_start, "apos", len) == 0)
555 g_string_append_c (ucontext->str, '\'');
560 name = g_strndup (ucontext->entity_start, len);
561 set_unescape_error (ucontext->context, error,
562 p, ucontext->text_end,
563 G_MARKUP_ERROR_PARSE,
564 _("Entity name '%s' is not known"),
571 set_unescape_error (ucontext->context, error,
572 /* give line number of the & */
573 ucontext->entity_start, ucontext->text_end,
574 G_MARKUP_ERROR_PARSE,
575 _("Entity did not end with a semicolon; "
576 "most likely you used an ampersand "
577 "character without intending to start "
578 "an entity - escape ampersand as &"));
587 unescape_text_state_after_charref_hash (UnescapeContext *ucontext,
591 gboolean is_hex = FALSE;
594 start = ucontext->entity_start;
599 p = g_utf8_next_char (p);
603 while (p != ucontext->text_end && *p != ';')
604 p = g_utf8_next_char (p);
606 if (p != ucontext->text_end)
608 g_assert (*p == ';');
610 /* digit is between start and p */
619 l = strtoul (start, &end, 16);
621 l = strtoul (start, &end, 10);
623 if (end != p || errno != 0)
625 set_unescape_error (ucontext->context, error,
626 start, ucontext->text_end,
627 G_MARKUP_ERROR_PARSE,
628 _("Failed to parse '%-.*s', which "
629 "should have been a digit "
630 "inside a character reference "
631 "(ê for example) - perhaps "
632 "the digit is too large"),
637 /* characters XML permits */
641 (l >= 0x20 && l <= 0xD7FF) ||
642 (l >= 0xE000 && l <= 0xFFFD) ||
643 (l >= 0x10000 && l <= 0x10FFFF))
646 g_string_append (ucontext->str, char_str (l, buf));
650 set_unescape_error (ucontext->context, error,
651 start, ucontext->text_end,
652 G_MARKUP_ERROR_PARSE,
653 _("Character reference '%-.*s' does not "
654 "encode a permitted character"),
659 /* Move to next state */
660 p = g_utf8_next_char (p); /* past semicolon */
661 ucontext->state = USTATE_INSIDE_TEXT;
665 set_unescape_error (ucontext->context, error,
666 start, ucontext->text_end,
667 G_MARKUP_ERROR_PARSE,
668 _("Empty character reference; "
669 "should include a digit such as "
675 set_unescape_error (ucontext->context, error,
676 start, ucontext->text_end,
677 G_MARKUP_ERROR_PARSE,
678 _("Character reference did not end with a "
680 "most likely you used an ampersand "
681 "character without intending to start "
682 "an entity - escape ampersand as &"));
689 unescape_text (GMarkupParseContext *context,
691 const gchar *text_end,
695 UnescapeContext ucontext;
698 ucontext.context = context;
699 ucontext.text = text;
700 ucontext.text_end = text_end;
701 ucontext.entity_start = NULL;
703 ucontext.str = g_string_sized_new (text_end - text);
705 ucontext.state = USTATE_INSIDE_TEXT;
708 while (p != text_end && context->state != STATE_ERROR)
710 g_assert (p < text_end);
712 switch (ucontext.state)
714 case USTATE_INSIDE_TEXT:
716 p = unescape_text_state_inside_text (&ucontext,
722 case USTATE_AFTER_AMPERSAND:
724 p = unescape_text_state_after_ampersand (&ucontext,
731 case USTATE_INSIDE_ENTITY_NAME:
733 p = unescape_text_state_inside_entity_name (&ucontext,
739 case USTATE_AFTER_CHARREF_HASH:
741 p = unescape_text_state_after_charref_hash (&ucontext,
748 g_assert_not_reached ();
753 if (context->state != STATE_ERROR)
755 switch (ucontext.state)
757 case USTATE_INSIDE_TEXT:
759 case USTATE_AFTER_AMPERSAND:
760 case USTATE_INSIDE_ENTITY_NAME:
761 set_unescape_error (context, error,
763 G_MARKUP_ERROR_PARSE,
764 _("Unfinished entity reference"));
766 case USTATE_AFTER_CHARREF_HASH:
767 set_unescape_error (context, error,
769 G_MARKUP_ERROR_PARSE,
770 _("Unfinished character reference"));
775 if (context->state == STATE_ERROR)
777 g_string_free (ucontext.str, TRUE);
783 *unescaped = ucontext.str;
788 static inline gboolean
789 advance_char (GMarkupParseContext *context)
791 context->iter = g_utf8_next_char (context->iter);
792 context->char_number += 1;
794 if (context->iter == context->current_text_end)
798 else if (*context->iter == '\n')
800 context->line_number += 1;
801 context->char_number = 1;
807 static inline gboolean
810 return c == ' ' || c == '\t' || c == '\n' || c == '\r';
814 skip_spaces (GMarkupParseContext *context)
818 if (!xml_isspace (*context->iter))
821 while (advance_char (context));
825 advance_to_name_end (GMarkupParseContext *context)
829 if (!is_name_char (context->iter))
832 while (advance_char (context));
836 add_to_partial (GMarkupParseContext *context,
837 const gchar *text_start,
838 const gchar *text_end)
840 if (context->partial_chunk == NULL)
841 context->partial_chunk = g_string_sized_new (text_end - text_start);
843 if (text_start != text_end)
844 g_string_append_len (context->partial_chunk, text_start,
845 text_end - text_start);
847 /* Invariant here that partial_chunk exists */
851 truncate_partial (GMarkupParseContext *context)
853 if (context->partial_chunk != NULL)
855 context->partial_chunk = g_string_truncate (context->partial_chunk, 0);
860 current_element (GMarkupParseContext *context)
862 return context->tag_stack->data;
866 pop_subparser_stack (GMarkupParseContext *context)
868 GMarkupRecursionTracker *tracker;
870 g_assert (context->subparser_stack);
872 tracker = context->subparser_stack->data;
874 context->awaiting_pop = TRUE;
875 context->held_user_data = context->user_data;
877 context->user_data = tracker->prev_user_data;
878 context->parser = tracker->prev_parser;
879 context->subparser_element = tracker->prev_element;
880 g_slice_free (GMarkupRecursionTracker, tracker);
882 context->subparser_stack = g_slist_delete_link (context->subparser_stack,
883 context->subparser_stack);
887 possibly_finish_subparser (GMarkupParseContext *context)
889 if (current_element (context) == context->subparser_element)
890 pop_subparser_stack (context);
894 ensure_no_outstanding_subparser (GMarkupParseContext *context)
896 if (context->awaiting_pop)
897 g_critical ("During the first end_element call after invoking a "
898 "subparser you must pop the subparser stack and handle "
899 "the freeing of the subparser user_data. This can be "
900 "done by calling the end function of the subparser. "
901 "Very probably, your program just leaked memory.");
903 /* let valgrind watch the pointer disappear... */
904 context->held_user_data = NULL;
905 context->awaiting_pop = FALSE;
909 current_attribute (GMarkupParseContext *context)
911 g_assert (context->cur_attr >= 0);
912 return context->attr_names[context->cur_attr];
916 find_current_text_end (GMarkupParseContext *context)
918 /* This function must be safe (non-segfaulting) on invalid UTF8.
919 * It assumes the string starts with a character start
921 const gchar *end = context->current_text + context->current_text_len;
925 g_assert (context->current_text_len > 0);
927 p = g_utf8_find_prev_char (context->current_text, end);
929 g_assert (p != NULL); /* since current_text was a char start */
931 /* p is now the start of the last character or character portion. */
933 next = g_utf8_next_char (p); /* this only touches *p, nothing beyond */
937 /* whole character */
938 context->current_text_end = end;
943 context->leftover_char_portion = g_string_new_len (p, end - p);
944 context->current_text_len -= (end - p);
945 context->current_text_end = p;
951 add_attribute (GMarkupParseContext *context, char *name)
953 if (context->cur_attr + 2 >= context->alloc_attrs)
955 context->alloc_attrs += 5; /* silly magic number */
956 context->attr_names = g_realloc (context->attr_names, sizeof(char*)*context->alloc_attrs);
957 context->attr_values = g_realloc (context->attr_values, sizeof(char*)*context->alloc_attrs);
960 context->attr_names[context->cur_attr] = name;
961 context->attr_values[context->cur_attr] = NULL;
962 context->attr_names[context->cur_attr+1] = NULL;
963 context->attr_values[context->cur_attr+1] = NULL;
967 * g_markup_parse_context_parse:
968 * @context: a #GMarkupParseContext
969 * @text: chunk of text to parse
970 * @text_len: length of @text in bytes
971 * @error: return location for a #GError
973 * Feed some data to the #GMarkupParseContext. The data need not
974 * be valid UTF-8; an error will be signaled if it's invalid.
975 * The data need not be an entire document; you can feed a document
976 * into the parser incrementally, via multiple calls to this function.
977 * Typically, as you receive data from a network connection or file,
978 * you feed each received chunk of data into this function, aborting
979 * the process if an error occurs. Once an error is reported, no further
980 * data may be fed to the #GMarkupParseContext; all errors are fatal.
982 * Return value: %FALSE if an error occurred, %TRUE on success
985 g_markup_parse_context_parse (GMarkupParseContext *context,
990 const gchar *first_invalid;
992 g_return_val_if_fail (context != NULL, FALSE);
993 g_return_val_if_fail (text != NULL, FALSE);
994 g_return_val_if_fail (context->state != STATE_ERROR, FALSE);
995 g_return_val_if_fail (!context->parsing, FALSE);
998 text_len = strlen (text);
1003 context->parsing = TRUE;
1005 if (context->leftover_char_portion)
1007 const gchar *first_char;
1009 if ((*text & 0xc0) != 0x80)
1012 first_char = g_utf8_find_next_char (text, text + text_len);
1016 /* leftover_char_portion was completed. Parse it. */
1017 GString *portion = context->leftover_char_portion;
1019 g_string_append_len (context->leftover_char_portion,
1020 text, first_char - text);
1022 /* hacks to allow recursion */
1023 context->parsing = FALSE;
1024 context->leftover_char_portion = NULL;
1026 if (!g_markup_parse_context_parse (context,
1027 portion->str, portion->len,
1030 g_assert (context->state == STATE_ERROR);
1033 g_string_free (portion, TRUE);
1034 context->parsing = TRUE;
1036 /* Skip the fraction of char that was in this text */
1037 text_len -= (first_char - text);
1042 /* another little chunk of the leftover char; geez
1043 * someone is inefficient.
1045 g_string_append_len (context->leftover_char_portion,
1048 if (context->leftover_char_portion->len > 7)
1050 /* The leftover char portion is too big to be
1055 G_MARKUP_ERROR_BAD_UTF8,
1056 _("Invalid UTF-8 encoded text - overlong sequence"));
1063 context->current_text = text;
1064 context->current_text_len = text_len;
1065 context->iter = context->current_text;
1066 context->start = context->iter;
1068 /* Nothing left after finishing the leftover char, or nothing
1069 * passed in to begin with.
1071 if (context->current_text_len == 0)
1074 /* find_current_text_end () assumes the string starts at
1075 * a character start, so we need to validate at least
1076 * that much. It doesn't assume any following bytes
1079 if ((*context->current_text & 0xc0) == 0x80) /* not a char start */
1083 G_MARKUP_ERROR_BAD_UTF8,
1084 _("Invalid UTF-8 encoded text - not a start char"));
1088 /* Initialize context->current_text_end, possibly adjusting
1089 * current_text_len, and add any leftover char portion
1091 find_current_text_end (context);
1093 /* Validate UTF8 (must be done after we find the end, since
1094 * we could have a trailing incomplete char)
1096 if (!g_utf8_validate (context->current_text,
1097 context->current_text_len,
1102 q = p = context->current_text;
1103 while (p != first_invalid)
1109 context->char_number = 1;
1114 context->line_number += newlines;
1115 context->char_number += g_utf8_strlen (q, first_invalid - q);
1119 G_MARKUP_ERROR_BAD_UTF8,
1120 _("Invalid UTF-8 encoded text - not valid '%s'"),
1121 g_strndup (context->current_text,
1122 context->current_text_len));
1126 while (context->iter != context->current_text_end)
1128 switch (context->state)
1131 /* Possible next state: AFTER_OPEN_ANGLE */
1133 g_assert (context->tag_stack == NULL);
1135 /* whitespace is ignored outside of any elements */
1136 skip_spaces (context);
1138 if (context->iter != context->current_text_end)
1140 if (*context->iter == '<')
1142 /* Move after the open angle */
1143 advance_char (context);
1145 context->state = STATE_AFTER_OPEN_ANGLE;
1147 /* this could start a passthrough */
1148 context->start = context->iter;
1150 /* document is now non-empty */
1151 context->document_empty = FALSE;
1157 G_MARKUP_ERROR_PARSE,
1158 _("Document must begin with an element (e.g. <book>)"));
1163 case STATE_AFTER_OPEN_ANGLE:
1164 /* Possible next states: INSIDE_OPEN_TAG_NAME,
1165 * AFTER_CLOSE_TAG_SLASH, INSIDE_PASSTHROUGH
1167 if (*context->iter == '?' ||
1168 *context->iter == '!')
1170 /* include < in the passthrough */
1171 const gchar *openangle = "<";
1172 add_to_partial (context, openangle, openangle + 1);
1173 context->start = context->iter;
1174 context->balance = 1;
1175 context->state = STATE_INSIDE_PASSTHROUGH;
1177 else if (*context->iter == '/')
1180 advance_char (context);
1182 context->state = STATE_AFTER_CLOSE_TAG_SLASH;
1184 else if (is_name_start_char (context->iter))
1186 context->state = STATE_INSIDE_OPEN_TAG_NAME;
1188 /* start of tag name */
1189 context->start = context->iter;
1197 G_MARKUP_ERROR_PARSE,
1198 _("'%s' is not a valid character following "
1199 "a '<' character; it may not begin an "
1201 utf8_str (context->iter, buf));
1205 /* The AFTER_CLOSE_ANGLE state is actually sort of
1206 * broken, because it doesn't correspond to a range
1207 * of characters in the input stream as the others do,
1208 * and thus makes things harder to conceptualize
1210 case STATE_AFTER_CLOSE_ANGLE:
1211 /* Possible next states: INSIDE_TEXT, STATE_START */
1212 if (context->tag_stack == NULL)
1214 context->start = NULL;
1215 context->state = STATE_START;
1219 context->start = context->iter;
1220 context->state = STATE_INSIDE_TEXT;
1224 case STATE_AFTER_ELISION_SLASH:
1225 /* Possible next state: AFTER_CLOSE_ANGLE */
1228 /* We need to pop the tag stack and call the end_element
1229 * function, since this is the close tag
1231 GError *tmp_error = NULL;
1233 g_assert (context->tag_stack != NULL);
1235 possibly_finish_subparser (context);
1238 if (context->parser->end_element)
1239 (* context->parser->end_element) (context,
1240 context->tag_stack->data,
1244 ensure_no_outstanding_subparser (context);
1248 mark_error (context, tmp_error);
1249 g_propagate_error (error, tmp_error);
1253 if (*context->iter == '>')
1255 /* move after the close angle */
1256 advance_char (context);
1257 context->state = STATE_AFTER_CLOSE_ANGLE;
1265 G_MARKUP_ERROR_PARSE,
1266 _("Odd character '%s', expected a '>' character "
1267 "to end the empty-element tag '%s'"),
1268 utf8_str (context->iter, buf),
1269 current_element (context));
1273 g_free (context->tag_stack->data);
1274 context->tag_stack = g_slist_delete_link (context->tag_stack,
1275 context->tag_stack);
1279 case STATE_INSIDE_OPEN_TAG_NAME:
1280 /* Possible next states: BETWEEN_ATTRIBUTES */
1282 /* if there's a partial chunk then it's the first part of the
1283 * tag name. If there's a context->start then it's the start
1284 * of the tag name in current_text, the partial chunk goes
1285 * before that start though.
1287 advance_to_name_end (context);
1289 if (context->iter == context->current_text_end)
1291 /* The name hasn't necessarily ended. Merge with
1292 * partial chunk, leave state unchanged.
1294 add_to_partial (context, context->start, context->iter);
1298 /* The name has ended. Combine it with the partial chunk
1299 * if any; push it on the stack; enter next state.
1301 add_to_partial (context, context->start, context->iter);
1302 context->tag_stack =
1303 g_slist_prepend (context->tag_stack,
1304 g_string_free (context->partial_chunk,
1307 context->partial_chunk = NULL;
1309 context->state = STATE_BETWEEN_ATTRIBUTES;
1310 context->start = NULL;
1314 case STATE_INSIDE_ATTRIBUTE_NAME:
1315 /* Possible next states: AFTER_ATTRIBUTE_NAME */
1317 advance_to_name_end (context);
1318 add_to_partial (context, context->start, context->iter);
1320 /* read the full name, if we enter the equals sign state
1321 * then add the attribute to the list (without the value),
1322 * otherwise store a partial chunk to be prepended later.
1324 if (context->iter != context->current_text_end)
1325 context->state = STATE_AFTER_ATTRIBUTE_NAME;
1328 case STATE_AFTER_ATTRIBUTE_NAME:
1329 /* Possible next states: AFTER_ATTRIBUTE_EQUALS_SIGN */
1331 skip_spaces (context);
1333 if (context->iter != context->current_text_end)
1335 /* The name has ended. Combine it with the partial chunk
1336 * if any; push it on the stack; enter next state.
1338 add_attribute (context, g_string_free (context->partial_chunk, FALSE));
1340 context->partial_chunk = NULL;
1341 context->start = NULL;
1343 if (*context->iter == '=')
1345 advance_char (context);
1346 context->state = STATE_AFTER_ATTRIBUTE_EQUALS_SIGN;
1354 G_MARKUP_ERROR_PARSE,
1355 _("Odd character '%s', expected a '=' after "
1356 "attribute name '%s' of element '%s'"),
1357 utf8_str (context->iter, buf),
1358 current_attribute (context),
1359 current_element (context));
1365 case STATE_BETWEEN_ATTRIBUTES:
1366 /* Possible next states: AFTER_CLOSE_ANGLE,
1367 * AFTER_ELISION_SLASH, INSIDE_ATTRIBUTE_NAME
1369 skip_spaces (context);
1371 if (context->iter != context->current_text_end)
1373 if (*context->iter == '/')
1375 advance_char (context);
1376 context->state = STATE_AFTER_ELISION_SLASH;
1378 else if (*context->iter == '>')
1381 advance_char (context);
1382 context->state = STATE_AFTER_CLOSE_ANGLE;
1384 else if (is_name_start_char (context->iter))
1386 context->state = STATE_INSIDE_ATTRIBUTE_NAME;
1387 /* start of attribute name */
1388 context->start = context->iter;
1396 G_MARKUP_ERROR_PARSE,
1397 _("Odd character '%s', expected a '>' or '/' "
1398 "character to end the start tag of "
1399 "element '%s', or optionally an attribute; "
1400 "perhaps you used an invalid character in "
1401 "an attribute name"),
1402 utf8_str (context->iter, buf),
1403 current_element (context));
1406 /* If we're done with attributes, invoke
1407 * the start_element callback
1409 if (context->state == STATE_AFTER_ELISION_SLASH ||
1410 context->state == STATE_AFTER_CLOSE_ANGLE)
1412 const gchar *start_name;
1413 /* Ugly, but the current code expects an empty array instead of NULL */
1414 const gchar *empty = NULL;
1415 const gchar **attr_names = ∅
1416 const gchar **attr_values = ∅
1419 /* Call user callback for element start */
1420 start_name = current_element (context);
1422 if (context->cur_attr >= 0)
1424 attr_names = (const gchar**)context->attr_names;
1425 attr_values = (const gchar**)context->attr_values;
1429 if (context->parser->start_element)
1430 (* context->parser->start_element) (context,
1432 (const gchar **)attr_names,
1433 (const gchar **)attr_values,
1437 /* Go ahead and free the attributes. */
1438 for (; context->cur_attr >= 0; context->cur_attr--)
1440 int pos = context->cur_attr;
1441 g_free (context->attr_names[pos]);
1442 g_free (context->attr_values[pos]);
1443 context->attr_names[pos] = context->attr_values[pos] = NULL;
1445 g_assert (context->cur_attr == -1);
1446 g_assert (context->attr_names == NULL ||
1447 context->attr_names[0] == NULL);
1448 g_assert (context->attr_values == NULL ||
1449 context->attr_values[0] == NULL);
1451 if (tmp_error != NULL)
1452 propagate_error (context, error, tmp_error);
1457 case STATE_AFTER_ATTRIBUTE_EQUALS_SIGN:
1458 /* Possible next state: INSIDE_ATTRIBUTE_VALUE_[SQ/DQ] */
1460 skip_spaces (context);
1462 if (context->iter != context->current_text_end)
1464 if (*context->iter == '"')
1466 advance_char (context);
1467 context->state = STATE_INSIDE_ATTRIBUTE_VALUE_DQ;
1468 context->start = context->iter;
1470 else if (*context->iter == '\'')
1472 advance_char (context);
1473 context->state = STATE_INSIDE_ATTRIBUTE_VALUE_SQ;
1474 context->start = context->iter;
1482 G_MARKUP_ERROR_PARSE,
1483 _("Odd character '%s', expected an open quote mark "
1484 "after the equals sign when giving value for "
1485 "attribute '%s' of element '%s'"),
1486 utf8_str (context->iter, buf),
1487 current_attribute (context),
1488 current_element (context));
1493 case STATE_INSIDE_ATTRIBUTE_VALUE_SQ:
1494 case STATE_INSIDE_ATTRIBUTE_VALUE_DQ:
1495 /* Possible next states: BETWEEN_ATTRIBUTES */
1499 if (context->state == STATE_INSIDE_ATTRIBUTE_VALUE_SQ)
1510 if (*context->iter == delim)
1513 while (advance_char (context));
1515 if (context->iter == context->current_text_end)
1517 /* The value hasn't necessarily ended. Merge with
1518 * partial chunk, leave state unchanged.
1520 add_to_partial (context, context->start, context->iter);
1524 /* The value has ended at the quote mark. Combine it
1525 * with the partial chunk if any; set it for the current
1530 add_to_partial (context, context->start, context->iter);
1532 g_assert (context->cur_attr >= 0);
1534 if (unescape_text (context,
1535 context->partial_chunk->str,
1536 context->partial_chunk->str +
1537 context->partial_chunk->len,
1541 /* success, advance past quote and set state. */
1542 context->attr_values[context->cur_attr] = g_string_free (unescaped, FALSE);
1543 advance_char (context);
1544 context->state = STATE_BETWEEN_ATTRIBUTES;
1545 context->start = NULL;
1548 truncate_partial (context);
1552 case STATE_INSIDE_TEXT:
1553 /* Possible next states: AFTER_OPEN_ANGLE */
1556 if (*context->iter == '<')
1559 while (advance_char (context));
1561 /* The text hasn't necessarily ended. Merge with
1562 * partial chunk, leave state unchanged.
1565 add_to_partial (context, context->start, context->iter);
1567 if (context->iter != context->current_text_end)
1569 GString *unescaped = NULL;
1571 /* The text has ended at the open angle. Call the text
1575 if (unescape_text (context,
1576 context->partial_chunk->str,
1577 context->partial_chunk->str +
1578 context->partial_chunk->len,
1582 GError *tmp_error = NULL;
1584 if (context->parser->text)
1585 (*context->parser->text) (context,
1591 g_string_free (unescaped, TRUE);
1593 if (tmp_error == NULL)
1595 /* advance past open angle and set state. */
1596 advance_char (context);
1597 context->state = STATE_AFTER_OPEN_ANGLE;
1598 /* could begin a passthrough */
1599 context->start = context->iter;
1602 propagate_error (context, error, tmp_error);
1605 truncate_partial (context);
1609 case STATE_AFTER_CLOSE_TAG_SLASH:
1610 /* Possible next state: INSIDE_CLOSE_TAG_NAME */
1611 if (is_name_start_char (context->iter))
1613 context->state = STATE_INSIDE_CLOSE_TAG_NAME;
1615 /* start of tag name */
1616 context->start = context->iter;
1624 G_MARKUP_ERROR_PARSE,
1625 _("'%s' is not a valid character following "
1626 "the characters '</'; '%s' may not begin an "
1628 utf8_str (context->iter, buf),
1629 utf8_str (context->iter, buf));
1633 case STATE_INSIDE_CLOSE_TAG_NAME:
1634 /* Possible next state: AFTER_CLOSE_TAG_NAME */
1635 advance_to_name_end (context);
1636 add_to_partial (context, context->start, context->iter);
1638 if (context->iter != context->current_text_end)
1639 context->state = STATE_AFTER_CLOSE_TAG_NAME;
1642 case STATE_AFTER_CLOSE_TAG_NAME:
1643 /* Possible next state: AFTER_CLOSE_TAG_SLASH */
1645 skip_spaces (context);
1647 if (context->iter != context->current_text_end)
1651 /* The name has ended. Combine it with the partial chunk
1652 * if any; check that it matches stack top and pop
1653 * stack; invoke proper callback; enter next state.
1655 close_name = g_string_free (context->partial_chunk, FALSE);
1656 context->partial_chunk = NULL;
1658 if (*context->iter != '>')
1664 G_MARKUP_ERROR_PARSE,
1665 _("'%s' is not a valid character following "
1666 "the close element name '%s'; the allowed "
1667 "character is '>'"),
1668 utf8_str (context->iter, buf),
1671 else if (context->tag_stack == NULL)
1675 G_MARKUP_ERROR_PARSE,
1676 _("Element '%s' was closed, no element "
1677 "is currently open"),
1680 else if (strcmp (close_name, current_element (context)) != 0)
1684 G_MARKUP_ERROR_PARSE,
1685 _("Element '%s' was closed, but the currently "
1686 "open element is '%s'"),
1688 current_element (context));
1693 advance_char (context);
1694 context->state = STATE_AFTER_CLOSE_ANGLE;
1695 context->start = NULL;
1697 possibly_finish_subparser (context);
1699 /* call the end_element callback */
1701 if (context->parser->end_element)
1702 (* context->parser->end_element) (context,
1707 ensure_no_outstanding_subparser (context);
1709 /* Pop the tag stack */
1710 g_free (context->tag_stack->data);
1711 context->tag_stack = g_slist_delete_link (context->tag_stack,
1712 context->tag_stack);
1715 propagate_error (context, error, tmp_error);
1718 g_free (close_name);
1722 case STATE_INSIDE_PASSTHROUGH:
1723 /* Possible next state: AFTER_CLOSE_ANGLE */
1726 if (*context->iter == '<')
1728 if (*context->iter == '>')
1734 add_to_partial (context, context->start, context->iter);
1735 context->start = context->iter;
1737 str = context->partial_chunk->str;
1738 len = context->partial_chunk->len;
1740 if (str[1] == '?' && str[len - 1] == '?')
1742 if (strncmp (str, "<!--", 4) == 0 &&
1743 strcmp (str + len - 2, "--") == 0)
1745 if (strncmp (str, "<![CDATA[", 9) == 0 &&
1746 strcmp (str + len - 2, "]]") == 0)
1748 if (strncmp (str, "<!DOCTYPE", 9) == 0 &&
1749 context->balance == 0)
1753 while (advance_char (context));
1755 if (context->iter == context->current_text_end)
1757 /* The passthrough hasn't necessarily ended. Merge with
1758 * partial chunk, leave state unchanged.
1760 add_to_partial (context, context->start, context->iter);
1764 /* The passthrough has ended at the close angle. Combine
1765 * it with the partial chunk if any. Call the passthrough
1766 * callback. Note that the open/close angles are
1767 * included in the text of the passthrough.
1769 GError *tmp_error = NULL;
1771 advance_char (context); /* advance past close angle */
1772 add_to_partial (context, context->start, context->iter);
1774 if (context->flags & G_MARKUP_TREAT_CDATA_AS_TEXT &&
1775 strncmp (context->partial_chunk->str, "<![CDATA[", 9) == 0)
1777 if (context->parser->text)
1778 (*context->parser->text) (context,
1779 context->partial_chunk->str + 9,
1780 context->partial_chunk->len - 12,
1784 else if (context->parser->passthrough)
1785 (*context->parser->passthrough) (context,
1786 context->partial_chunk->str,
1787 context->partial_chunk->len,
1791 truncate_partial (context);
1793 if (tmp_error == NULL)
1795 context->state = STATE_AFTER_CLOSE_ANGLE;
1796 context->start = context->iter; /* could begin text */
1799 propagate_error (context, error, tmp_error);
1808 g_assert_not_reached ();
1814 context->parsing = FALSE;
1816 return context->state != STATE_ERROR;
1820 * g_markup_parse_context_end_parse:
1821 * @context: a #GMarkupParseContext
1822 * @error: return location for a #GError
1824 * Signals to the #GMarkupParseContext that all data has been
1825 * fed into the parse context with g_markup_parse_context_parse().
1826 * This function reports an error if the document isn't complete,
1827 * for example if elements are still open.
1829 * Return value: %TRUE on success, %FALSE if an error was set
1832 g_markup_parse_context_end_parse (GMarkupParseContext *context,
1835 g_return_val_if_fail (context != NULL, FALSE);
1836 g_return_val_if_fail (!context->parsing, FALSE);
1837 g_return_val_if_fail (context->state != STATE_ERROR, FALSE);
1839 if (context->partial_chunk != NULL)
1841 g_string_free (context->partial_chunk, TRUE);
1842 context->partial_chunk = NULL;
1845 if (context->document_empty)
1847 set_error (context, error, G_MARKUP_ERROR_EMPTY,
1848 _("Document was empty or contained only whitespace"));
1852 context->parsing = TRUE;
1854 switch (context->state)
1860 case STATE_AFTER_OPEN_ANGLE:
1861 set_error (context, error, G_MARKUP_ERROR_PARSE,
1862 _("Document ended unexpectedly just after an open angle bracket '<'"));
1865 case STATE_AFTER_CLOSE_ANGLE:
1866 if (context->tag_stack != NULL)
1868 /* Error message the same as for INSIDE_TEXT */
1869 set_error (context, error, G_MARKUP_ERROR_PARSE,
1870 _("Document ended unexpectedly with elements still open - "
1871 "'%s' was the last element opened"),
1872 current_element (context));
1876 case STATE_AFTER_ELISION_SLASH:
1877 set_error (context, error, G_MARKUP_ERROR_PARSE,
1878 _("Document ended unexpectedly, expected to see a close angle "
1879 "bracket ending the tag <%s/>"), current_element (context));
1882 case STATE_INSIDE_OPEN_TAG_NAME:
1883 set_error (context, error, G_MARKUP_ERROR_PARSE,
1884 _("Document ended unexpectedly inside an element name"));
1887 case STATE_INSIDE_ATTRIBUTE_NAME:
1888 case STATE_AFTER_ATTRIBUTE_NAME:
1889 set_error (context, error, G_MARKUP_ERROR_PARSE,
1890 _("Document ended unexpectedly inside an attribute name"));
1893 case STATE_BETWEEN_ATTRIBUTES:
1894 set_error (context, error, G_MARKUP_ERROR_PARSE,
1895 _("Document ended unexpectedly inside an element-opening "
1899 case STATE_AFTER_ATTRIBUTE_EQUALS_SIGN:
1900 set_error (context, error, G_MARKUP_ERROR_PARSE,
1901 _("Document ended unexpectedly after the equals sign "
1902 "following an attribute name; no attribute value"));
1905 case STATE_INSIDE_ATTRIBUTE_VALUE_SQ:
1906 case STATE_INSIDE_ATTRIBUTE_VALUE_DQ:
1907 set_error (context, error, G_MARKUP_ERROR_PARSE,
1908 _("Document ended unexpectedly while inside an attribute "
1912 case STATE_INSIDE_TEXT:
1913 g_assert (context->tag_stack != NULL);
1914 set_error (context, error, G_MARKUP_ERROR_PARSE,
1915 _("Document ended unexpectedly with elements still open - "
1916 "'%s' was the last element opened"),
1917 current_element (context));
1920 case STATE_AFTER_CLOSE_TAG_SLASH:
1921 case STATE_INSIDE_CLOSE_TAG_NAME:
1922 case STATE_AFTER_CLOSE_TAG_NAME:
1923 set_error (context, error, G_MARKUP_ERROR_PARSE,
1924 _("Document ended unexpectedly inside the close tag for "
1925 "element '%s'"), current_element (context));
1928 case STATE_INSIDE_PASSTHROUGH:
1929 set_error (context, error, G_MARKUP_ERROR_PARSE,
1930 _("Document ended unexpectedly inside a comment or "
1931 "processing instruction"));
1936 g_assert_not_reached ();
1940 context->parsing = FALSE;
1942 return context->state != STATE_ERROR;
1946 * g_markup_parse_context_get_element:
1947 * @context: a #GMarkupParseContext
1948 * @returns: the name of the currently open element, or %NULL
1950 * Retrieves the name of the currently open element.
1952 * If called from the start_element or end_element handlers this will
1953 * give the element_name as passed to those functions. For the parent
1954 * elements, see g_markup_parse_context_get_element_stack().
1958 G_CONST_RETURN gchar *
1959 g_markup_parse_context_get_element (GMarkupParseContext *context)
1961 g_return_val_if_fail (context != NULL, NULL);
1963 if (context->tag_stack == NULL)
1966 return current_element (context);
1970 * g_markup_parse_context_get_element_stack:
1971 * @context: a #GMarkupParseContext
1973 * Retrieves the element stack from the internal state of the parser.
1974 * The returned #GSList is a list of strings where the first item is
1975 * the currently open tag (as would be returned by
1976 * g_markup_parse_context_get_element()) and the next item is its
1979 * This function is intended to be used in the start_element and
1980 * end_element handlers where g_markup_parse_context_get_element()
1981 * would merely return the name of the element that is being
1984 * Returns: the element stack, which must not be modified
1988 G_CONST_RETURN GSList *
1989 g_markup_parse_context_get_element_stack (GMarkupParseContext *context)
1991 g_return_val_if_fail (context != NULL, NULL);
1993 return context->tag_stack;
1997 * g_markup_parse_context_get_position:
1998 * @context: a #GMarkupParseContext
1999 * @line_number: return location for a line number, or %NULL
2000 * @char_number: return location for a char-on-line number, or %NULL
2002 * Retrieves the current line number and the number of the character on
2003 * that line. Intended for use in error messages; there are no strict
2004 * semantics for what constitutes the "current" line number other than
2005 * "the best number we could come up with for error messages."
2009 g_markup_parse_context_get_position (GMarkupParseContext *context,
2013 g_return_if_fail (context != NULL);
2016 *line_number = context->line_number;
2019 *char_number = context->char_number;
2023 * g_markup_parse_context_push:
2024 * @context: a #GMarkupParseContext
2025 * @parser: a #GMarkupParser
2026 * @user_data: user data to pass to #GMarkupParser functions
2028 * Temporarily redirects markup data to a sub-parser.
2030 * This function may only be called from the start_element handler of
2031 * a #GMarkupParser. It must be matched with a corresponding call to
2032 * g_markup_parse_context_pop() in the matching end_element handler
2033 * (except in the case that the parser aborts due to an error).
2035 * All tags, text and other data between the matching tags is
2036 * redirected to the subparser given by @parser. @user_data is used
2037 * as the user_data for that parser. @user_data is also passed to the
2038 * error callback in the event that an error occurs. This includes
2039 * errors that occur in subparsers of the subparser.
2041 * The end tag matching the start tag for which this call was made is
2042 * handled by the previous parser (which is given its own user_data)
2043 * which is why g_markup_parse_context_pop() is provided to allow "one
2044 * last access" to the @user_data provided to this function. In the
2045 * case of error, the @user_data provided here is passed directly to
2046 * the error callback of the subparser and g_markup_parse_context()
2047 * should not be called. In either case, if @user_data was allocated
2048 * then it ought to be freed from both of these locations.
2050 * This function is not intended to be directly called by users
2051 * interested in invoking subparsers. Instead, it is intended to be
2052 * used by the subparsers themselves to implement a higher-level
2055 * As an example, see the following implementation of a simple
2056 * parser that counts the number of tags encountered.
2065 * counter_start_element (GMarkupParseContext *context,
2066 * const gchar *element_name,
2067 * const gchar **attribute_names,
2068 * const gchar **attribute_values,
2069 * gpointer user_data,
2072 * CounterData *data = user_data;
2074 * data->tag_count++;
2078 * counter_error (GMarkupParseContext *context,
2080 * gpointer user_data)
2082 * CounterData *data = user_data;
2084 * g_slice_free (CounterData, data);
2087 * static GMarkupParser counter_subparser =
2089 * counter_start_element,
2097 * In order to allow this parser to be easily used as a subparser, the
2098 * following interface is provided:
2102 * start_counting (GMarkupParseContext *context)
2104 * CounterData *data = g_slice_new (CounterData);
2106 * data->tag_count = 0;
2107 * g_markup_parse_context_push (context, &counter_subparser, data);
2111 * end_counting (GMarkupParseContext *context)
2113 * CounterData *data = g_markup_parse_context_pop (context);
2116 * result = data->tag_count;
2117 * g_slice_free (CounterData, data);
2123 * The subparser would then be used as follows:
2126 * static void start_element (context, element_name, ...)
2128 * if (strcmp (element_name, "count-these") == 0)
2129 * start_counting (context);
2131 * /* else, handle other tags... */
2134 * static void end_element (context, element_name, ...)
2136 * if (strcmp (element_name, "count-these") == 0)
2137 * g_print ("Counted %d tags\n", end_counting (context));
2139 * /* else, handle other tags... */
2146 g_markup_parse_context_push (GMarkupParseContext *context,
2147 GMarkupParser *parser,
2150 GMarkupRecursionTracker *tracker;
2152 tracker = g_slice_new (GMarkupRecursionTracker);
2153 tracker->prev_element = context->subparser_element;
2154 tracker->prev_parser = context->parser;
2155 tracker->prev_user_data = context->user_data;
2157 context->subparser_element = current_element (context);
2158 context->parser = parser;
2159 context->user_data = user_data;
2161 context->subparser_stack = g_slist_prepend (context->subparser_stack,
2166 * g_markup_parse_context_pop:
2167 * @context: a #GMarkupParseContext
2169 * Completes the process of a temporary sub-parser redirection.
2171 * This function exists to collect the user_data allocated by a
2172 * matching call to g_markup_parse_context_push(). It must be called
2173 * in the end_element handler corresponding to the start_element
2174 * handler during which g_markup_parse_context_push() was called. You
2175 * must not call this function from the error callback -- the
2176 * @user_data is provided directly to the callback in that case.
2178 * This function is not intended to be directly called by users
2179 * interested in invoking subparsers. Instead, it is intended to be
2180 * used by the subparsers themselves to implement a higher-level
2183 * Returns: the user_data passed to g_markup_parse_context_push().
2188 g_markup_parse_context_pop (GMarkupParseContext *context)
2192 if (!context->awaiting_pop)
2193 possibly_finish_subparser (context);
2195 g_assert (context->awaiting_pop);
2197 context->awaiting_pop = FALSE;
2199 /* valgrind friendliness */
2200 user_data = context->held_user_data;
2201 context->held_user_data = NULL;
2207 append_escaped_text (GString *str,
2216 end = text + length;
2221 next = g_utf8_next_char (p);
2226 g_string_append (str, "&");
2230 g_string_append (str, "<");
2234 g_string_append (str, ">");
2238 g_string_append (str, "'");
2242 g_string_append (str, """);
2246 c = g_utf8_get_char (p);
2247 if ((0x1 <= c && c <= 0x8) ||
2248 (0xb <= c && c <= 0xc) ||
2249 (0xe <= c && c <= 0x1f) ||
2250 (0x7f <= c && c <= 0x84) ||
2251 (0x86 <= c && c <= 0x9f))
2252 g_string_append_printf (str, "&#x%x;", c);
2254 g_string_append_len (str, p, next - p);
2263 * g_markup_escape_text:
2264 * @text: some valid UTF-8 text
2265 * @length: length of @text in bytes, or -1 if the text is nul-terminated
2267 * Escapes text so that the markup parser will parse it verbatim.
2268 * Less than, greater than, ampersand, etc. are replaced with the
2269 * corresponding entities. This function would typically be used
2270 * when writing out a file to be parsed with the markup parser.
2272 * Note that this function doesn't protect whitespace and line endings
2273 * from being processed according to the XML rules for normalization
2274 * of line endings and attribute values.
2276 * Return value: a newly allocated string with the escaped text
2279 g_markup_escape_text (const gchar *text,
2284 g_return_val_if_fail (text != NULL, NULL);
2287 length = strlen (text);
2289 /* prealloc at least as long as original text */
2290 str = g_string_sized_new (length);
2291 append_escaped_text (str, text, length);
2293 return g_string_free (str, FALSE);
2298 * @format: a printf-style format string
2299 * @after: location to store a pointer to the character after
2300 * the returned conversion. On a %NULL return, returns the
2301 * pointer to the trailing NUL in the string
2303 * Find the next conversion in a printf-style format string.
2304 * Partially based on code from printf-parser.c,
2305 * Copyright (C) 1999-2000, 2002-2003 Free Software Foundation, Inc.
2307 * Return value: pointer to the next conversion in @format,
2308 * or %NULL, if none.
2311 find_conversion (const char *format,
2314 const char *start = format;
2317 while (*start != '\0' && *start != '%')
2334 /* Test for positional argument. */
2335 if (*cp >= '0' && *cp <= '9')
2339 for (np = cp; *np >= '0' && *np <= '9'; np++)
2345 /* Skip the flags. */
2359 /* Skip the field width. */
2364 /* Test for positional argument. */
2365 if (*cp >= '0' && *cp <= '9')
2369 for (np = cp; *np >= '0' && *np <= '9'; np++)
2377 for (; *cp >= '0' && *cp <= '9'; cp++)
2381 /* Skip the precision. */
2387 /* Test for positional argument. */
2388 if (*cp >= '0' && *cp <= '9')
2392 for (np = cp; *np >= '0' && *np <= '9'; np++)
2400 for (; *cp >= '0' && *cp <= '9'; cp++)
2405 /* Skip argument type/size specifiers. */
2406 while (*cp == 'h' ||
2415 /* Skip the conversion character. */
2423 * g_markup_vprintf_escaped:
2424 * @format: printf() style format string
2425 * @args: variable argument list, similar to vprintf()
2427 * Formats the data in @args according to @format, escaping
2428 * all string and character arguments in the fashion
2429 * of g_markup_escape_text(). See g_markup_printf_escaped().
2431 * Return value: newly allocated result from formatting
2432 * operation. Free with g_free().
2437 g_markup_vprintf_escaped (const char *format,
2442 GString *result = NULL;
2443 gchar *output1 = NULL;
2444 gchar *output2 = NULL;
2445 const char *p, *op1, *op2;
2448 /* The technique here, is that we make two format strings that
2449 * have the identical conversions in the identical order to the
2450 * original strings, but differ in the text in-between. We
2451 * then use the normal g_strdup_vprintf() to format the arguments
2452 * with the two new format strings. By comparing the results,
2453 * we can figure out what segments of the output come from
2454 * the the original format string, and what from the arguments,
2455 * and thus know what portions of the string to escape.
2457 * For instance, for:
2459 * g_markup_printf_escaped ("%s ate %d apples", "Susan & Fred", 5);
2461 * We form the two format strings "%sX%dX" and %sY%sY". The results
2462 * of formatting with those two strings are
2464 * "%sX%dX" => "Susan & FredX5X"
2465 * "%sY%dY" => "Susan & FredY5Y"
2467 * To find the span of the first argument, we find the first position
2468 * where the two arguments differ, which tells us that the first
2469 * argument formatted to "Susan & Fred". We then escape that
2470 * to "Susan & Fred" and join up with the intermediate portions
2471 * of the format string and the second argument to get
2472 * "Susan & Fred ate 5 apples".
2475 /* Create the two modified format strings
2477 format1 = g_string_new (NULL);
2478 format2 = g_string_new (NULL);
2483 const char *conv = find_conversion (p, &after);
2487 g_string_append_len (format1, conv, after - conv);
2488 g_string_append_c (format1, 'X');
2489 g_string_append_len (format2, conv, after - conv);
2490 g_string_append_c (format2, 'Y');
2495 /* Use them to format the arguments
2497 G_VA_COPY (args2, args);
2499 output1 = g_strdup_vprintf (format1->str, args);
2506 output2 = g_strdup_vprintf (format2->str, args2);
2511 result = g_string_new (NULL);
2513 /* Iterate through the original format string again,
2514 * copying the non-conversion portions and the escaped
2515 * converted arguments to the output string.
2523 const char *output_start;
2524 const char *conv = find_conversion (p, &after);
2527 if (!conv) /* The end, after points to the trailing \0 */
2529 g_string_append_len (result, p, after - p);
2533 g_string_append_len (result, p, conv - p);
2535 while (*op1 == *op2)
2541 escaped = g_markup_escape_text (output_start, op1 - output_start);
2542 g_string_append (result, escaped);
2551 g_string_free (format1, TRUE);
2552 g_string_free (format2, TRUE);
2557 return g_string_free (result, FALSE);
2563 * g_markup_printf_escaped:
2564 * @format: printf() style format string
2565 * @Varargs: the arguments to insert in the format string
2567 * Formats arguments according to @format, escaping
2568 * all string and character arguments in the fashion
2569 * of g_markup_escape_text(). This is useful when you
2570 * want to insert literal strings into XML-style markup
2571 * output, without having to worry that the strings
2572 * might themselves contain markup.
2575 * const char *store = "Fortnum & Mason";
2576 * const char *item = "Tea";
2579 * output = g_markup_printf_escaped ("<purchase>"
2580 * "<store>%s</store>"
2581 * "<item>%s</item>"
2582 * "</purchase>",
2586 * Return value: newly allocated result from formatting
2587 * operation. Free with g_free().
2592 g_markup_printf_escaped (const char *format, ...)
2597 va_start (args, format);
2598 result = g_markup_vprintf_escaped (format, args);
2605 g_markup_parse_boolean (const char *string,
2608 char const * const falses[] = { "false", "f", "no", "n", "0" };
2609 char const * const trues[] = { "true", "t", "yes", "y", "1" };
2612 for (i = 0; i < G_N_ELEMENTS (falses); i++)
2614 if (g_ascii_strcasecmp (string, falses[i]) == 0)
2623 for (i = 0; i < G_N_ELEMENTS (trues); i++)
2625 if (g_ascii_strcasecmp (string, trues[i]) == 0)
2638 * GMarkupCollectType:
2639 * @G_MARKUP_COLLECT_INVALID: used to terminate the list of attributes
2641 * @G_MARKUP_COLLECT_STRING: collect the string pointer directly from
2642 * the attribute_values[] array. Expects a
2643 * parameter of type (const char **). If
2644 * %G_MARKUP_COLLECT_OPTIONAL is specified
2645 * and the attribute isn't present then the
2646 * pointer will be set to %NULL.
2647 * @G_MARKUP_COLLECT_STRDUP: as with %G_MARKUP_COLLECT_STRING, but
2648 * expects a paramter of type (char **) and
2649 * g_strdup()s the returned pointer. The
2650 * pointer must be freed with g_free().
2651 * @G_MARKUP_COLLECT_BOOLEAN: expects a parameter of type (gboolean *)
2652 * and parses the attribute value as a
2653 * boolean. Sets %FALSE if the attribute
2654 * isn't present. Valid boolean values
2655 * consist of (case insensitive) "false",
2656 * "f", "no", "n", "0" and "true", "t",
2658 * @G_MARKUP_COLLECT_TRISTATE: as with %G_MARKUP_COLLECT_BOOLEAN, but
2659 * in the case of a missing attribute a
2660 * value is set that compares equal to
2661 * neither %FALSE nor %TRUE.
2662 * G_MARKUP_COLLECT_OPTIONAL is implied.
2663 * @G_MARKUP_COLLECT_OPTIONAL: can be bitwise ORed with the other
2664 * fields. If present, allows the
2665 * attribute not to appear. A default
2666 * value is set depending on what value
2669 * A mixed enumerated type and flags field. You must specify one type
2670 * (string, strdup, boolean, tristate). Additionally, you may
2671 * optionally bitwise OR the type with the flag
2672 * %G_MARKUP_COLLECT_OPTIONAL.
2674 * It is likely that this enum will be extended in the future to
2675 * support other types.
2679 * g_markup_collect_attributes:
2680 * @element_name: the current tag name
2681 * @attribute_names: the attribute names
2682 * @attribute_values: the attribute values
2683 * @error: a pointer to a #GError or %NULL
2684 * @first_type: the #GMarkupCollectType of the
2686 * @first_attr: the name of the first attribute
2687 * @...: a pointer to the storage location of the
2688 * first attribute (or %NULL), followed by
2689 * more types names and pointers, ending
2690 * with %G_MARKUP_COLLECT_INVALID.
2692 * Collects the attributes of the element from the
2693 * data passed to the #GMarkupParser start_element
2694 * function, dealing with common error conditions
2695 * and supporting boolean values.
2697 * This utility function is not required to write
2698 * a parser but can save a lot of typing.
2700 * The @element_name, @attribute_names,
2701 * @attribute_values and @error parameters passed
2702 * to the start_element callback should be passed
2703 * unmodified to this function.
2705 * Following these arguments is a list of
2706 * "supported" attributes to collect. It is an
2707 * error to specify multiple attributes with the
2708 * same name. If any attribute not in the list
2709 * appears in the @attribute_names array then an
2710 * unknown attribute error will result.
2712 * The #GMarkupCollectType field allows specifying
2713 * the type of collection to perform and if a
2714 * given attribute must appear or is optional.
2716 * The attribute name is simply the name of the
2717 * attribute to collect.
2719 * The pointer should be of the appropriate type
2720 * (see the descriptions under
2721 * #GMarkupCollectType) and may be %NULL in case a
2722 * particular attribute is to be allowed but
2725 * This function deals with issuing errors for missing attributes
2726 * (of type %G_MARKUP_ERROR_MISSING_ATTRIBUTE), unknown attributes
2727 * (of type %G_MARKUP_ERROR_UNKNOWN_ATTRIBUTE) and duplicate
2728 * attributes (of type %G_MARKUP_ERROR_INVALID_CONTENT) as well
2729 * as parse errors for boolean-valued attributes (again of type
2730 * %G_MARKUP_ERROR_INVALID_CONTENT). In all of these cases %FALSE
2731 * will be returned and @error will be set as appropriate.
2733 * Return value: %TRUE if successful
2738 g_markup_collect_attributes (const gchar *element_name,
2739 const gchar **attribute_names,
2740 const gchar **attribute_values,
2742 GMarkupCollectType first_type,
2743 const gchar *first_attr,
2746 GMarkupCollectType type;
2758 va_start (ap, first_attr);
2759 while (type != G_MARKUP_COLLECT_INVALID)
2764 mandatory = !(type & G_MARKUP_COLLECT_OPTIONAL);
2765 type &= (G_MARKUP_COLLECT_OPTIONAL - 1);
2767 /* tristate records a value != TRUE and != FALSE
2768 * for the case where the attribute is missing
2770 if (type == G_MARKUP_COLLECT_TRISTATE)
2773 for (i = 0; attribute_names[i]; i++)
2774 if (i >= 40 || !(collected & (G_GUINT64_CONSTANT(1) << i)))
2775 if (!strcmp (attribute_names[i], attr))
2778 /* ISO C99 only promises that the user can pass up to 127 arguments.
2779 * Subtracting the first 4 arguments plus the final NULL and dividing
2780 * by 3 arguments per collected attribute, we are left with a maximum
2781 * number of supported attributes of (127 - 5) / 3 = 40.
2783 * In reality, nobody is ever going to call us with anywhere close to
2784 * 40 attributes to collect, so it is safe to assume that if i > 40
2785 * then the user has given some invalid or repeated arguments. These
2786 * problems will be caught and reported at the end of the function.
2788 * We know at this point that we have an error, but we don't know
2789 * what error it is, so just continue...
2792 collected |= (G_GUINT64_CONSTANT(1) << i);
2794 value = attribute_values[i];
2796 if (value == NULL && mandatory)
2798 g_set_error (error, G_MARKUP_ERROR,
2799 G_MARKUP_ERROR_MISSING_ATTRIBUTE,
2800 "element '%s' requires attribute '%s'",
2801 element_name, attr);
2809 case G_MARKUP_COLLECT_STRING:
2811 const char **str_ptr;
2813 str_ptr = va_arg (ap, const char **);
2815 if (str_ptr != NULL)
2820 case G_MARKUP_COLLECT_STRDUP:
2824 str_ptr = va_arg (ap, char **);
2826 if (str_ptr != NULL)
2827 *str_ptr = g_strdup (value);
2831 case G_MARKUP_COLLECT_BOOLEAN:
2832 case G_MARKUP_COLLECT_TRISTATE:
2837 bool_ptr = va_arg (ap, gboolean *);
2839 if (bool_ptr != NULL)
2841 if (type == G_MARKUP_COLLECT_TRISTATE)
2842 /* constructivists rejoice!
2843 * neither false nor true...
2847 else /* G_MARKUP_COLLECT_BOOLEAN */
2853 if (!g_markup_parse_boolean (value, va_arg (ap, gboolean *)))
2855 g_set_error (error, G_MARKUP_ERROR,
2856 G_MARKUP_ERROR_INVALID_CONTENT,
2857 "element '%s', attribute '%s', value '%s' "
2858 "cannot be parsed as a boolean value",
2859 element_name, attr, value);
2869 g_assert_not_reached ();
2872 type = va_arg (ap, GMarkupCollectType);
2873 attr = va_arg (ap, const char *);
2878 /* ensure we collected all the arguments */
2879 for (i = 0; attribute_names[i]; i++)
2880 if ((collected & (G_GUINT64_CONSTANT(1) << i)) == 0)
2882 /* attribute not collected: could be caused by two things.
2884 * 1) it doesn't exist in our list of attributes
2885 * 2) it existed but was matched by a duplicate attribute earlier
2891 for (j = 0; j < i; j++)
2892 if (strcmp (attribute_names[i], attribute_names[j]) == 0)
2896 /* j is now the first occurance of attribute_names[i] */
2898 g_set_error (error, G_MARKUP_ERROR,
2899 G_MARKUP_ERROR_UNKNOWN_ATTRIBUTE,
2900 "attribute '%s' invalid for element '%s'",
2901 attribute_names[i], element_name);
2903 g_set_error (error, G_MARKUP_ERROR,
2904 G_MARKUP_ERROR_INVALID_CONTENT,
2905 "attribute '%s' given multiple times for element '%s'",
2906 attribute_names[i], element_name);
2914 /* replay the above to free allocations */
2918 va_start (ap, first_attr);
2919 while (type != G_MARKUP_COLLECT_INVALID)
2923 ptr = va_arg (ap, gpointer);
2928 switch (type & (G_MARKUP_COLLECT_OPTIONAL - 1))
2930 case G_MARKUP_COLLECT_STRDUP:
2932 g_free (*(char **) ptr);
2934 case G_MARKUP_COLLECT_STRING:
2935 *(char **) ptr = NULL;
2938 case G_MARKUP_COLLECT_BOOLEAN:
2939 *(gboolean *) ptr = FALSE;
2942 case G_MARKUP_COLLECT_TRISTATE:
2943 *(gboolean *) ptr = -1;
2947 type = va_arg (ap, GMarkupCollectType);
2948 attr = va_arg (ap, const char *);
2958 #define __G_MARKUP_C__
2959 #include "galiasdef.c"