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 1.1 permits */
638 if ((0 < l && l <= 0xD7FF) ||
639 (0xE000 <= l && l <= 0xFFFD) ||
640 (0x10000 <= l && l <= 0x10FFFF))
643 g_string_append (ucontext->str, char_str (l, buf));
647 set_unescape_error (ucontext->context, error,
648 start, ucontext->text_end,
649 G_MARKUP_ERROR_PARSE,
650 _("Character reference '%-.*s' does not "
651 "encode a permitted character"),
656 /* Move to next state */
657 p = g_utf8_next_char (p); /* past semicolon */
658 ucontext->state = USTATE_INSIDE_TEXT;
662 set_unescape_error (ucontext->context, error,
663 start, ucontext->text_end,
664 G_MARKUP_ERROR_PARSE,
665 _("Empty character reference; "
666 "should include a digit such as "
672 set_unescape_error (ucontext->context, error,
673 start, ucontext->text_end,
674 G_MARKUP_ERROR_PARSE,
675 _("Character reference did not end with a "
677 "most likely you used an ampersand "
678 "character without intending to start "
679 "an entity - escape ampersand as &"));
686 unescape_text (GMarkupParseContext *context,
688 const gchar *text_end,
692 UnescapeContext ucontext;
695 ucontext.context = context;
696 ucontext.text = text;
697 ucontext.text_end = text_end;
698 ucontext.entity_start = NULL;
700 ucontext.str = g_string_sized_new (text_end - text);
702 ucontext.state = USTATE_INSIDE_TEXT;
705 while (p != text_end && context->state != STATE_ERROR)
707 g_assert (p < text_end);
709 switch (ucontext.state)
711 case USTATE_INSIDE_TEXT:
713 p = unescape_text_state_inside_text (&ucontext,
719 case USTATE_AFTER_AMPERSAND:
721 p = unescape_text_state_after_ampersand (&ucontext,
728 case USTATE_INSIDE_ENTITY_NAME:
730 p = unescape_text_state_inside_entity_name (&ucontext,
736 case USTATE_AFTER_CHARREF_HASH:
738 p = unescape_text_state_after_charref_hash (&ucontext,
745 g_assert_not_reached ();
750 if (context->state != STATE_ERROR)
752 switch (ucontext.state)
754 case USTATE_INSIDE_TEXT:
756 case USTATE_AFTER_AMPERSAND:
757 case USTATE_INSIDE_ENTITY_NAME:
758 set_unescape_error (context, error,
760 G_MARKUP_ERROR_PARSE,
761 _("Unfinished entity reference"));
763 case USTATE_AFTER_CHARREF_HASH:
764 set_unescape_error (context, error,
766 G_MARKUP_ERROR_PARSE,
767 _("Unfinished character reference"));
772 if (context->state == STATE_ERROR)
774 g_string_free (ucontext.str, TRUE);
780 *unescaped = ucontext.str;
785 static inline gboolean
786 advance_char (GMarkupParseContext *context)
788 context->iter = g_utf8_next_char (context->iter);
789 context->char_number += 1;
791 if (context->iter == context->current_text_end)
795 else if (*context->iter == '\n')
797 context->line_number += 1;
798 context->char_number = 1;
804 static inline gboolean
807 return c == ' ' || c == '\t' || c == '\n' || c == '\r';
811 skip_spaces (GMarkupParseContext *context)
815 if (!xml_isspace (*context->iter))
818 while (advance_char (context));
822 advance_to_name_end (GMarkupParseContext *context)
826 if (!is_name_char (context->iter))
829 while (advance_char (context));
833 add_to_partial (GMarkupParseContext *context,
834 const gchar *text_start,
835 const gchar *text_end)
837 if (context->partial_chunk == NULL)
838 context->partial_chunk = g_string_sized_new (text_end - text_start);
840 if (text_start != text_end)
841 g_string_append_len (context->partial_chunk, text_start,
842 text_end - text_start);
844 /* Invariant here that partial_chunk exists */
848 truncate_partial (GMarkupParseContext *context)
850 if (context->partial_chunk != NULL)
852 context->partial_chunk = g_string_truncate (context->partial_chunk, 0);
857 current_element (GMarkupParseContext *context)
859 return context->tag_stack->data;
863 pop_subparser_stack (GMarkupParseContext *context)
865 GMarkupRecursionTracker *tracker;
867 g_assert (context->subparser_stack);
869 tracker = context->subparser_stack->data;
871 context->awaiting_pop = TRUE;
872 context->held_user_data = context->user_data;
874 context->user_data = tracker->prev_user_data;
875 context->parser = tracker->prev_parser;
876 context->subparser_element = tracker->prev_element;
877 g_slice_free (GMarkupRecursionTracker, tracker);
879 context->subparser_stack = g_slist_delete_link (context->subparser_stack,
880 context->subparser_stack);
884 possibly_finish_subparser (GMarkupParseContext *context)
886 if (current_element (context) == context->subparser_element)
887 pop_subparser_stack (context);
891 ensure_no_outstanding_subparser (GMarkupParseContext *context)
893 if (context->awaiting_pop)
894 g_critical ("During the first end_element call after invoking a "
895 "subparser you must pop the subparser stack and handle "
896 "the freeing of the subparser user_data. This can be "
897 "done by calling the end function of the subparser. "
898 "Very probably, your program just leaked memory.");
900 /* let valgrind watch the pointer disappear... */
901 context->held_user_data = NULL;
902 context->awaiting_pop = FALSE;
906 current_attribute (GMarkupParseContext *context)
908 g_assert (context->cur_attr >= 0);
909 return context->attr_names[context->cur_attr];
913 find_current_text_end (GMarkupParseContext *context)
915 /* This function must be safe (non-segfaulting) on invalid UTF8.
916 * It assumes the string starts with a character start
918 const gchar *end = context->current_text + context->current_text_len;
922 g_assert (context->current_text_len > 0);
924 p = g_utf8_find_prev_char (context->current_text, end);
926 g_assert (p != NULL); /* since current_text was a char start */
928 /* p is now the start of the last character or character portion. */
930 next = g_utf8_next_char (p); /* this only touches *p, nothing beyond */
934 /* whole character */
935 context->current_text_end = end;
940 context->leftover_char_portion = g_string_new_len (p, end - p);
941 context->current_text_len -= (end - p);
942 context->current_text_end = p;
948 add_attribute (GMarkupParseContext *context, char *name)
950 if (context->cur_attr + 2 >= context->alloc_attrs)
952 context->alloc_attrs += 5; /* silly magic number */
953 context->attr_names = g_realloc (context->attr_names, sizeof(char*)*context->alloc_attrs);
954 context->attr_values = g_realloc (context->attr_values, sizeof(char*)*context->alloc_attrs);
957 context->attr_names[context->cur_attr] = name;
958 context->attr_values[context->cur_attr] = NULL;
959 context->attr_names[context->cur_attr+1] = NULL;
960 context->attr_values[context->cur_attr+1] = NULL;
964 * g_markup_parse_context_parse:
965 * @context: a #GMarkupParseContext
966 * @text: chunk of text to parse
967 * @text_len: length of @text in bytes
968 * @error: return location for a #GError
970 * Feed some data to the #GMarkupParseContext. The data need not
971 * be valid UTF-8; an error will be signaled if it's invalid.
972 * The data need not be an entire document; you can feed a document
973 * into the parser incrementally, via multiple calls to this function.
974 * Typically, as you receive data from a network connection or file,
975 * you feed each received chunk of data into this function, aborting
976 * the process if an error occurs. Once an error is reported, no further
977 * data may be fed to the #GMarkupParseContext; all errors are fatal.
979 * Return value: %FALSE if an error occurred, %TRUE on success
982 g_markup_parse_context_parse (GMarkupParseContext *context,
987 const gchar *first_invalid;
989 g_return_val_if_fail (context != NULL, FALSE);
990 g_return_val_if_fail (text != NULL, FALSE);
991 g_return_val_if_fail (context->state != STATE_ERROR, FALSE);
992 g_return_val_if_fail (!context->parsing, FALSE);
995 text_len = strlen (text);
1000 context->parsing = TRUE;
1002 if (context->leftover_char_portion)
1004 const gchar *first_char;
1006 if ((*text & 0xc0) != 0x80)
1009 first_char = g_utf8_find_next_char (text, text + text_len);
1013 /* leftover_char_portion was completed. Parse it. */
1014 GString *portion = context->leftover_char_portion;
1016 g_string_append_len (context->leftover_char_portion,
1017 text, first_char - text);
1019 /* hacks to allow recursion */
1020 context->parsing = FALSE;
1021 context->leftover_char_portion = NULL;
1023 if (!g_markup_parse_context_parse (context,
1024 portion->str, portion->len,
1027 g_assert (context->state == STATE_ERROR);
1030 g_string_free (portion, TRUE);
1031 context->parsing = TRUE;
1033 /* Skip the fraction of char that was in this text */
1034 text_len -= (first_char - text);
1039 /* another little chunk of the leftover char; geez
1040 * someone is inefficient.
1042 g_string_append_len (context->leftover_char_portion,
1045 if (context->leftover_char_portion->len > 7)
1047 /* The leftover char portion is too big to be
1052 G_MARKUP_ERROR_BAD_UTF8,
1053 _("Invalid UTF-8 encoded text - overlong sequence"));
1060 context->current_text = text;
1061 context->current_text_len = text_len;
1062 context->iter = context->current_text;
1063 context->start = context->iter;
1065 /* Nothing left after finishing the leftover char, or nothing
1066 * passed in to begin with.
1068 if (context->current_text_len == 0)
1071 /* find_current_text_end () assumes the string starts at
1072 * a character start, so we need to validate at least
1073 * that much. It doesn't assume any following bytes
1076 if ((*context->current_text & 0xc0) == 0x80) /* not a char start */
1080 G_MARKUP_ERROR_BAD_UTF8,
1081 _("Invalid UTF-8 encoded text - not a start char"));
1085 /* Initialize context->current_text_end, possibly adjusting
1086 * current_text_len, and add any leftover char portion
1088 find_current_text_end (context);
1090 /* Validate UTF8 (must be done after we find the end, since
1091 * we could have a trailing incomplete char)
1093 if (!g_utf8_validate (context->current_text,
1094 context->current_text_len,
1099 q = p = context->current_text;
1100 while (p != first_invalid)
1106 context->char_number = 1;
1111 context->line_number += newlines;
1112 context->char_number += g_utf8_strlen (q, first_invalid - q);
1116 G_MARKUP_ERROR_BAD_UTF8,
1117 _("Invalid UTF-8 encoded text - not valid '%s'"),
1118 g_strndup (context->current_text,
1119 context->current_text_len));
1123 while (context->iter != context->current_text_end)
1125 switch (context->state)
1128 /* Possible next state: AFTER_OPEN_ANGLE */
1130 g_assert (context->tag_stack == NULL);
1132 /* whitespace is ignored outside of any elements */
1133 skip_spaces (context);
1135 if (context->iter != context->current_text_end)
1137 if (*context->iter == '<')
1139 /* Move after the open angle */
1140 advance_char (context);
1142 context->state = STATE_AFTER_OPEN_ANGLE;
1144 /* this could start a passthrough */
1145 context->start = context->iter;
1147 /* document is now non-empty */
1148 context->document_empty = FALSE;
1154 G_MARKUP_ERROR_PARSE,
1155 _("Document must begin with an element (e.g. <book>)"));
1160 case STATE_AFTER_OPEN_ANGLE:
1161 /* Possible next states: INSIDE_OPEN_TAG_NAME,
1162 * AFTER_CLOSE_TAG_SLASH, INSIDE_PASSTHROUGH
1164 if (*context->iter == '?' ||
1165 *context->iter == '!')
1167 /* include < in the passthrough */
1168 const gchar *openangle = "<";
1169 add_to_partial (context, openangle, openangle + 1);
1170 context->start = context->iter;
1171 context->balance = 1;
1172 context->state = STATE_INSIDE_PASSTHROUGH;
1174 else if (*context->iter == '/')
1177 advance_char (context);
1179 context->state = STATE_AFTER_CLOSE_TAG_SLASH;
1181 else if (is_name_start_char (context->iter))
1183 context->state = STATE_INSIDE_OPEN_TAG_NAME;
1185 /* start of tag name */
1186 context->start = context->iter;
1194 G_MARKUP_ERROR_PARSE,
1195 _("'%s' is not a valid character following "
1196 "a '<' character; it may not begin an "
1198 utf8_str (context->iter, buf));
1202 /* The AFTER_CLOSE_ANGLE state is actually sort of
1203 * broken, because it doesn't correspond to a range
1204 * of characters in the input stream as the others do,
1205 * and thus makes things harder to conceptualize
1207 case STATE_AFTER_CLOSE_ANGLE:
1208 /* Possible next states: INSIDE_TEXT, STATE_START */
1209 if (context->tag_stack == NULL)
1211 context->start = NULL;
1212 context->state = STATE_START;
1216 context->start = context->iter;
1217 context->state = STATE_INSIDE_TEXT;
1221 case STATE_AFTER_ELISION_SLASH:
1222 /* Possible next state: AFTER_CLOSE_ANGLE */
1225 /* We need to pop the tag stack and call the end_element
1226 * function, since this is the close tag
1228 GError *tmp_error = NULL;
1230 g_assert (context->tag_stack != NULL);
1232 possibly_finish_subparser (context);
1235 if (context->parser->end_element)
1236 (* context->parser->end_element) (context,
1237 context->tag_stack->data,
1241 ensure_no_outstanding_subparser (context);
1245 mark_error (context, tmp_error);
1246 g_propagate_error (error, tmp_error);
1250 if (*context->iter == '>')
1252 /* move after the close angle */
1253 advance_char (context);
1254 context->state = STATE_AFTER_CLOSE_ANGLE;
1262 G_MARKUP_ERROR_PARSE,
1263 _("Odd character '%s', expected a '>' character "
1264 "to end the empty-element tag '%s'"),
1265 utf8_str (context->iter, buf),
1266 current_element (context));
1270 g_free (context->tag_stack->data);
1271 context->tag_stack = g_slist_delete_link (context->tag_stack,
1272 context->tag_stack);
1276 case STATE_INSIDE_OPEN_TAG_NAME:
1277 /* Possible next states: BETWEEN_ATTRIBUTES */
1279 /* if there's a partial chunk then it's the first part of the
1280 * tag name. If there's a context->start then it's the start
1281 * of the tag name in current_text, the partial chunk goes
1282 * before that start though.
1284 advance_to_name_end (context);
1286 if (context->iter == context->current_text_end)
1288 /* The name hasn't necessarily ended. Merge with
1289 * partial chunk, leave state unchanged.
1291 add_to_partial (context, context->start, context->iter);
1295 /* The name has ended. Combine it with the partial chunk
1296 * if any; push it on the stack; enter next state.
1298 add_to_partial (context, context->start, context->iter);
1299 context->tag_stack =
1300 g_slist_prepend (context->tag_stack,
1301 g_string_free (context->partial_chunk,
1304 context->partial_chunk = NULL;
1306 context->state = STATE_BETWEEN_ATTRIBUTES;
1307 context->start = NULL;
1311 case STATE_INSIDE_ATTRIBUTE_NAME:
1312 /* Possible next states: AFTER_ATTRIBUTE_NAME */
1314 advance_to_name_end (context);
1315 add_to_partial (context, context->start, context->iter);
1317 /* read the full name, if we enter the equals sign state
1318 * then add the attribute to the list (without the value),
1319 * otherwise store a partial chunk to be prepended later.
1321 if (context->iter != context->current_text_end)
1322 context->state = STATE_AFTER_ATTRIBUTE_NAME;
1325 case STATE_AFTER_ATTRIBUTE_NAME:
1326 /* Possible next states: AFTER_ATTRIBUTE_EQUALS_SIGN */
1328 skip_spaces (context);
1330 if (context->iter != context->current_text_end)
1332 /* The name has ended. Combine it with the partial chunk
1333 * if any; push it on the stack; enter next state.
1335 add_attribute (context, g_string_free (context->partial_chunk, FALSE));
1337 context->partial_chunk = NULL;
1338 context->start = NULL;
1340 if (*context->iter == '=')
1342 advance_char (context);
1343 context->state = STATE_AFTER_ATTRIBUTE_EQUALS_SIGN;
1351 G_MARKUP_ERROR_PARSE,
1352 _("Odd character '%s', expected a '=' after "
1353 "attribute name '%s' of element '%s'"),
1354 utf8_str (context->iter, buf),
1355 current_attribute (context),
1356 current_element (context));
1362 case STATE_BETWEEN_ATTRIBUTES:
1363 /* Possible next states: AFTER_CLOSE_ANGLE,
1364 * AFTER_ELISION_SLASH, INSIDE_ATTRIBUTE_NAME
1366 skip_spaces (context);
1368 if (context->iter != context->current_text_end)
1370 if (*context->iter == '/')
1372 advance_char (context);
1373 context->state = STATE_AFTER_ELISION_SLASH;
1375 else if (*context->iter == '>')
1378 advance_char (context);
1379 context->state = STATE_AFTER_CLOSE_ANGLE;
1381 else if (is_name_start_char (context->iter))
1383 context->state = STATE_INSIDE_ATTRIBUTE_NAME;
1384 /* start of attribute name */
1385 context->start = context->iter;
1393 G_MARKUP_ERROR_PARSE,
1394 _("Odd character '%s', expected a '>' or '/' "
1395 "character to end the start tag of "
1396 "element '%s', or optionally an attribute; "
1397 "perhaps you used an invalid character in "
1398 "an attribute name"),
1399 utf8_str (context->iter, buf),
1400 current_element (context));
1403 /* If we're done with attributes, invoke
1404 * the start_element callback
1406 if (context->state == STATE_AFTER_ELISION_SLASH ||
1407 context->state == STATE_AFTER_CLOSE_ANGLE)
1409 const gchar *start_name;
1410 /* Ugly, but the current code expects an empty array instead of NULL */
1411 const gchar *empty = NULL;
1412 const gchar **attr_names = ∅
1413 const gchar **attr_values = ∅
1416 /* Call user callback for element start */
1417 start_name = current_element (context);
1419 if (context->cur_attr >= 0)
1421 attr_names = (const gchar**)context->attr_names;
1422 attr_values = (const gchar**)context->attr_values;
1426 if (context->parser->start_element)
1427 (* context->parser->start_element) (context,
1429 (const gchar **)attr_names,
1430 (const gchar **)attr_values,
1434 /* Go ahead and free the attributes. */
1435 for (; context->cur_attr >= 0; context->cur_attr--)
1437 int pos = context->cur_attr;
1438 g_free (context->attr_names[pos]);
1439 g_free (context->attr_values[pos]);
1440 context->attr_names[pos] = context->attr_values[pos] = NULL;
1442 g_assert (context->cur_attr == -1);
1443 g_assert (context->attr_names == NULL ||
1444 context->attr_names[0] == NULL);
1445 g_assert (context->attr_values == NULL ||
1446 context->attr_values[0] == NULL);
1448 if (tmp_error != NULL)
1449 propagate_error (context, error, tmp_error);
1454 case STATE_AFTER_ATTRIBUTE_EQUALS_SIGN:
1455 /* Possible next state: INSIDE_ATTRIBUTE_VALUE_[SQ/DQ] */
1457 skip_spaces (context);
1459 if (context->iter != context->current_text_end)
1461 if (*context->iter == '"')
1463 advance_char (context);
1464 context->state = STATE_INSIDE_ATTRIBUTE_VALUE_DQ;
1465 context->start = context->iter;
1467 else if (*context->iter == '\'')
1469 advance_char (context);
1470 context->state = STATE_INSIDE_ATTRIBUTE_VALUE_SQ;
1471 context->start = context->iter;
1479 G_MARKUP_ERROR_PARSE,
1480 _("Odd character '%s', expected an open quote mark "
1481 "after the equals sign when giving value for "
1482 "attribute '%s' of element '%s'"),
1483 utf8_str (context->iter, buf),
1484 current_attribute (context),
1485 current_element (context));
1490 case STATE_INSIDE_ATTRIBUTE_VALUE_SQ:
1491 case STATE_INSIDE_ATTRIBUTE_VALUE_DQ:
1492 /* Possible next states: BETWEEN_ATTRIBUTES */
1496 if (context->state == STATE_INSIDE_ATTRIBUTE_VALUE_SQ)
1507 if (*context->iter == delim)
1510 while (advance_char (context));
1512 if (context->iter == context->current_text_end)
1514 /* The value hasn't necessarily ended. Merge with
1515 * partial chunk, leave state unchanged.
1517 add_to_partial (context, context->start, context->iter);
1521 /* The value has ended at the quote mark. Combine it
1522 * with the partial chunk if any; set it for the current
1527 add_to_partial (context, context->start, context->iter);
1529 g_assert (context->cur_attr >= 0);
1531 if (unescape_text (context,
1532 context->partial_chunk->str,
1533 context->partial_chunk->str +
1534 context->partial_chunk->len,
1538 /* success, advance past quote and set state. */
1539 context->attr_values[context->cur_attr] = g_string_free (unescaped, FALSE);
1540 advance_char (context);
1541 context->state = STATE_BETWEEN_ATTRIBUTES;
1542 context->start = NULL;
1545 truncate_partial (context);
1549 case STATE_INSIDE_TEXT:
1550 /* Possible next states: AFTER_OPEN_ANGLE */
1553 if (*context->iter == '<')
1556 while (advance_char (context));
1558 /* The text hasn't necessarily ended. Merge with
1559 * partial chunk, leave state unchanged.
1562 add_to_partial (context, context->start, context->iter);
1564 if (context->iter != context->current_text_end)
1566 GString *unescaped = NULL;
1568 /* The text has ended at the open angle. Call the text
1572 if (unescape_text (context,
1573 context->partial_chunk->str,
1574 context->partial_chunk->str +
1575 context->partial_chunk->len,
1579 GError *tmp_error = NULL;
1581 if (context->parser->text)
1582 (*context->parser->text) (context,
1588 g_string_free (unescaped, TRUE);
1590 if (tmp_error == NULL)
1592 /* advance past open angle and set state. */
1593 advance_char (context);
1594 context->state = STATE_AFTER_OPEN_ANGLE;
1595 /* could begin a passthrough */
1596 context->start = context->iter;
1599 propagate_error (context, error, tmp_error);
1602 truncate_partial (context);
1606 case STATE_AFTER_CLOSE_TAG_SLASH:
1607 /* Possible next state: INSIDE_CLOSE_TAG_NAME */
1608 if (is_name_start_char (context->iter))
1610 context->state = STATE_INSIDE_CLOSE_TAG_NAME;
1612 /* start of tag name */
1613 context->start = context->iter;
1621 G_MARKUP_ERROR_PARSE,
1622 _("'%s' is not a valid character following "
1623 "the characters '</'; '%s' may not begin an "
1625 utf8_str (context->iter, buf),
1626 utf8_str (context->iter, buf));
1630 case STATE_INSIDE_CLOSE_TAG_NAME:
1631 /* Possible next state: AFTER_CLOSE_TAG_NAME */
1632 advance_to_name_end (context);
1633 add_to_partial (context, context->start, context->iter);
1635 if (context->iter != context->current_text_end)
1636 context->state = STATE_AFTER_CLOSE_TAG_NAME;
1639 case STATE_AFTER_CLOSE_TAG_NAME:
1640 /* Possible next state: AFTER_CLOSE_TAG_SLASH */
1642 skip_spaces (context);
1644 if (context->iter != context->current_text_end)
1648 /* The name has ended. Combine it with the partial chunk
1649 * if any; check that it matches stack top and pop
1650 * stack; invoke proper callback; enter next state.
1652 close_name = g_string_free (context->partial_chunk, FALSE);
1653 context->partial_chunk = NULL;
1655 if (*context->iter != '>')
1661 G_MARKUP_ERROR_PARSE,
1662 _("'%s' is not a valid character following "
1663 "the close element name '%s'; the allowed "
1664 "character is '>'"),
1665 utf8_str (context->iter, buf),
1668 else if (context->tag_stack == NULL)
1672 G_MARKUP_ERROR_PARSE,
1673 _("Element '%s' was closed, no element "
1674 "is currently open"),
1677 else if (strcmp (close_name, current_element (context)) != 0)
1681 G_MARKUP_ERROR_PARSE,
1682 _("Element '%s' was closed, but the currently "
1683 "open element is '%s'"),
1685 current_element (context));
1690 advance_char (context);
1691 context->state = STATE_AFTER_CLOSE_ANGLE;
1692 context->start = NULL;
1694 possibly_finish_subparser (context);
1696 /* call the end_element callback */
1698 if (context->parser->end_element)
1699 (* context->parser->end_element) (context,
1704 ensure_no_outstanding_subparser (context);
1706 /* Pop the tag stack */
1707 g_free (context->tag_stack->data);
1708 context->tag_stack = g_slist_delete_link (context->tag_stack,
1709 context->tag_stack);
1712 propagate_error (context, error, tmp_error);
1715 g_free (close_name);
1719 case STATE_INSIDE_PASSTHROUGH:
1720 /* Possible next state: AFTER_CLOSE_ANGLE */
1723 if (*context->iter == '<')
1725 if (*context->iter == '>')
1731 add_to_partial (context, context->start, context->iter);
1732 context->start = context->iter;
1734 str = context->partial_chunk->str;
1735 len = context->partial_chunk->len;
1737 if (str[1] == '?' && str[len - 1] == '?')
1739 if (strncmp (str, "<!--", 4) == 0 &&
1740 strcmp (str + len - 2, "--") == 0)
1742 if (strncmp (str, "<![CDATA[", 9) == 0 &&
1743 strcmp (str + len - 2, "]]") == 0)
1745 if (strncmp (str, "<!DOCTYPE", 9) == 0 &&
1746 context->balance == 0)
1750 while (advance_char (context));
1752 if (context->iter == context->current_text_end)
1754 /* The passthrough hasn't necessarily ended. Merge with
1755 * partial chunk, leave state unchanged.
1757 add_to_partial (context, context->start, context->iter);
1761 /* The passthrough has ended at the close angle. Combine
1762 * it with the partial chunk if any. Call the passthrough
1763 * callback. Note that the open/close angles are
1764 * included in the text of the passthrough.
1766 GError *tmp_error = NULL;
1768 advance_char (context); /* advance past close angle */
1769 add_to_partial (context, context->start, context->iter);
1771 if (context->flags & G_MARKUP_TREAT_CDATA_AS_TEXT &&
1772 strncmp (context->partial_chunk->str, "<![CDATA[", 9) == 0)
1774 if (context->parser->text)
1775 (*context->parser->text) (context,
1776 context->partial_chunk->str + 9,
1777 context->partial_chunk->len - 12,
1781 else if (context->parser->passthrough)
1782 (*context->parser->passthrough) (context,
1783 context->partial_chunk->str,
1784 context->partial_chunk->len,
1788 truncate_partial (context);
1790 if (tmp_error == NULL)
1792 context->state = STATE_AFTER_CLOSE_ANGLE;
1793 context->start = context->iter; /* could begin text */
1796 propagate_error (context, error, tmp_error);
1805 g_assert_not_reached ();
1811 context->parsing = FALSE;
1813 return context->state != STATE_ERROR;
1817 * g_markup_parse_context_end_parse:
1818 * @context: a #GMarkupParseContext
1819 * @error: return location for a #GError
1821 * Signals to the #GMarkupParseContext that all data has been
1822 * fed into the parse context with g_markup_parse_context_parse().
1823 * This function reports an error if the document isn't complete,
1824 * for example if elements are still open.
1826 * Return value: %TRUE on success, %FALSE if an error was set
1829 g_markup_parse_context_end_parse (GMarkupParseContext *context,
1832 g_return_val_if_fail (context != NULL, FALSE);
1833 g_return_val_if_fail (!context->parsing, FALSE);
1834 g_return_val_if_fail (context->state != STATE_ERROR, FALSE);
1836 if (context->partial_chunk != NULL)
1838 g_string_free (context->partial_chunk, TRUE);
1839 context->partial_chunk = NULL;
1842 if (context->document_empty)
1844 set_error (context, error, G_MARKUP_ERROR_EMPTY,
1845 _("Document was empty or contained only whitespace"));
1849 context->parsing = TRUE;
1851 switch (context->state)
1857 case STATE_AFTER_OPEN_ANGLE:
1858 set_error (context, error, G_MARKUP_ERROR_PARSE,
1859 _("Document ended unexpectedly just after an open angle bracket '<'"));
1862 case STATE_AFTER_CLOSE_ANGLE:
1863 if (context->tag_stack != NULL)
1865 /* Error message the same as for INSIDE_TEXT */
1866 set_error (context, error, G_MARKUP_ERROR_PARSE,
1867 _("Document ended unexpectedly with elements still open - "
1868 "'%s' was the last element opened"),
1869 current_element (context));
1873 case STATE_AFTER_ELISION_SLASH:
1874 set_error (context, error, G_MARKUP_ERROR_PARSE,
1875 _("Document ended unexpectedly, expected to see a close angle "
1876 "bracket ending the tag <%s/>"), current_element (context));
1879 case STATE_INSIDE_OPEN_TAG_NAME:
1880 set_error (context, error, G_MARKUP_ERROR_PARSE,
1881 _("Document ended unexpectedly inside an element name"));
1884 case STATE_INSIDE_ATTRIBUTE_NAME:
1885 case STATE_AFTER_ATTRIBUTE_NAME:
1886 set_error (context, error, G_MARKUP_ERROR_PARSE,
1887 _("Document ended unexpectedly inside an attribute name"));
1890 case STATE_BETWEEN_ATTRIBUTES:
1891 set_error (context, error, G_MARKUP_ERROR_PARSE,
1892 _("Document ended unexpectedly inside an element-opening "
1896 case STATE_AFTER_ATTRIBUTE_EQUALS_SIGN:
1897 set_error (context, error, G_MARKUP_ERROR_PARSE,
1898 _("Document ended unexpectedly after the equals sign "
1899 "following an attribute name; no attribute value"));
1902 case STATE_INSIDE_ATTRIBUTE_VALUE_SQ:
1903 case STATE_INSIDE_ATTRIBUTE_VALUE_DQ:
1904 set_error (context, error, G_MARKUP_ERROR_PARSE,
1905 _("Document ended unexpectedly while inside an attribute "
1909 case STATE_INSIDE_TEXT:
1910 g_assert (context->tag_stack != NULL);
1911 set_error (context, error, G_MARKUP_ERROR_PARSE,
1912 _("Document ended unexpectedly with elements still open - "
1913 "'%s' was the last element opened"),
1914 current_element (context));
1917 case STATE_AFTER_CLOSE_TAG_SLASH:
1918 case STATE_INSIDE_CLOSE_TAG_NAME:
1919 case STATE_AFTER_CLOSE_TAG_NAME:
1920 set_error (context, error, G_MARKUP_ERROR_PARSE,
1921 _("Document ended unexpectedly inside the close tag for "
1922 "element '%s'"), current_element (context));
1925 case STATE_INSIDE_PASSTHROUGH:
1926 set_error (context, error, G_MARKUP_ERROR_PARSE,
1927 _("Document ended unexpectedly inside a comment or "
1928 "processing instruction"));
1933 g_assert_not_reached ();
1937 context->parsing = FALSE;
1939 return context->state != STATE_ERROR;
1943 * g_markup_parse_context_get_element:
1944 * @context: a #GMarkupParseContext
1945 * @returns: the name of the currently open element, or %NULL
1947 * Retrieves the name of the currently open element.
1949 * If called from the start_element or end_element handlers this will
1950 * give the element_name as passed to those functions. For the parent
1951 * elements, see g_markup_parse_context_get_element_stack().
1955 G_CONST_RETURN gchar *
1956 g_markup_parse_context_get_element (GMarkupParseContext *context)
1958 g_return_val_if_fail (context != NULL, NULL);
1960 if (context->tag_stack == NULL)
1963 return current_element (context);
1967 * g_markup_parse_context_get_element_stack:
1968 * @context: a #GMarkupParseContext
1970 * Retrieves the element stack from the internal state of the parser.
1971 * The returned #GSList is a list of strings where the first item is
1972 * the currently open tag (as would be returned by
1973 * g_markup_parse_context_get_element()) and the next item is its
1976 * This function is intended to be used in the start_element and
1977 * end_element handlers where g_markup_parse_context_get_element()
1978 * would merely return the name of the element that is being
1981 * Returns: the element stack, which must not be modified
1985 G_CONST_RETURN GSList *
1986 g_markup_parse_context_get_element_stack (GMarkupParseContext *context)
1988 g_return_val_if_fail (context != NULL, NULL);
1990 return context->tag_stack;
1994 * g_markup_parse_context_get_position:
1995 * @context: a #GMarkupParseContext
1996 * @line_number: return location for a line number, or %NULL
1997 * @char_number: return location for a char-on-line number, or %NULL
1999 * Retrieves the current line number and the number of the character on
2000 * that line. Intended for use in error messages; there are no strict
2001 * semantics for what constitutes the "current" line number other than
2002 * "the best number we could come up with for error messages."
2006 g_markup_parse_context_get_position (GMarkupParseContext *context,
2010 g_return_if_fail (context != NULL);
2013 *line_number = context->line_number;
2016 *char_number = context->char_number;
2020 * g_markup_parse_context_get_user_data:
2021 * @context: a #GMarkupParseContext
2023 * Returns the user_data associated with @context. This will either
2024 * be the user_data that was provided to g_markup_parse_context_new()
2025 * or to the most recent call of g_markup_parse_context_push().
2027 * Returns: the provided user_data. The returned data belongs to
2028 * the markup context and will be freed when g_markup_context_free()
2034 g_markup_parse_context_get_user_data (GMarkupParseContext *context)
2036 return context->user_data;
2040 * g_markup_parse_context_push:
2041 * @context: a #GMarkupParseContext
2042 * @parser: a #GMarkupParser
2043 * @user_data: user data to pass to #GMarkupParser functions
2045 * Temporarily redirects markup data to a sub-parser.
2047 * This function may only be called from the start_element handler of
2048 * a #GMarkupParser. It must be matched with a corresponding call to
2049 * g_markup_parse_context_pop() in the matching end_element handler
2050 * (except in the case that the parser aborts due to an error).
2052 * All tags, text and other data between the matching tags is
2053 * redirected to the subparser given by @parser. @user_data is used
2054 * as the user_data for that parser. @user_data is also passed to the
2055 * error callback in the event that an error occurs. This includes
2056 * errors that occur in subparsers of the subparser.
2058 * The end tag matching the start tag for which this call was made is
2059 * handled by the previous parser (which is given its own user_data)
2060 * which is why g_markup_parse_context_pop() is provided to allow "one
2061 * last access" to the @user_data provided to this function. In the
2062 * case of error, the @user_data provided here is passed directly to
2063 * the error callback of the subparser and g_markup_parse_context()
2064 * should not be called. In either case, if @user_data was allocated
2065 * then it ought to be freed from both of these locations.
2067 * This function is not intended to be directly called by users
2068 * interested in invoking subparsers. Instead, it is intended to be
2069 * used by the subparsers themselves to implement a higher-level
2072 * As an example, see the following implementation of a simple
2073 * parser that counts the number of tags encountered.
2082 * counter_start_element (GMarkupParseContext *context,
2083 * const gchar *element_name,
2084 * const gchar **attribute_names,
2085 * const gchar **attribute_values,
2086 * gpointer user_data,
2089 * CounterData *data = user_data;
2091 * data->tag_count++;
2095 * counter_error (GMarkupParseContext *context,
2097 * gpointer user_data)
2099 * CounterData *data = user_data;
2101 * g_slice_free (CounterData, data);
2104 * static GMarkupParser counter_subparser =
2106 * counter_start_element,
2114 * In order to allow this parser to be easily used as a subparser, the
2115 * following interface is provided:
2119 * start_counting (GMarkupParseContext *context)
2121 * CounterData *data = g_slice_new (CounterData);
2123 * data->tag_count = 0;
2124 * g_markup_parse_context_push (context, &counter_subparser, data);
2128 * end_counting (GMarkupParseContext *context)
2130 * CounterData *data = g_markup_parse_context_pop (context);
2133 * result = data->tag_count;
2134 * g_slice_free (CounterData, data);
2140 * The subparser would then be used as follows:
2143 * static void start_element (context, element_name, ...)
2145 * if (strcmp (element_name, "count-these") == 0)
2146 * start_counting (context);
2148 * /* else, handle other tags... */
2151 * static void end_element (context, element_name, ...)
2153 * if (strcmp (element_name, "count-these") == 0)
2154 * g_print ("Counted %d tags\n", end_counting (context));
2156 * /* else, handle other tags... */
2163 g_markup_parse_context_push (GMarkupParseContext *context,
2164 GMarkupParser *parser,
2167 GMarkupRecursionTracker *tracker;
2169 tracker = g_slice_new (GMarkupRecursionTracker);
2170 tracker->prev_element = context->subparser_element;
2171 tracker->prev_parser = context->parser;
2172 tracker->prev_user_data = context->user_data;
2174 context->subparser_element = current_element (context);
2175 context->parser = parser;
2176 context->user_data = user_data;
2178 context->subparser_stack = g_slist_prepend (context->subparser_stack,
2183 * g_markup_parse_context_pop:
2184 * @context: a #GMarkupParseContext
2186 * Completes the process of a temporary sub-parser redirection.
2188 * This function exists to collect the user_data allocated by a
2189 * matching call to g_markup_parse_context_push(). It must be called
2190 * in the end_element handler corresponding to the start_element
2191 * handler during which g_markup_parse_context_push() was called. You
2192 * must not call this function from the error callback -- the
2193 * @user_data is provided directly to the callback in that case.
2195 * This function is not intended to be directly called by users
2196 * interested in invoking subparsers. Instead, it is intended to be
2197 * used by the subparsers themselves to implement a higher-level
2200 * Returns: the user_data passed to g_markup_parse_context_push().
2205 g_markup_parse_context_pop (GMarkupParseContext *context)
2209 if (!context->awaiting_pop)
2210 possibly_finish_subparser (context);
2212 g_assert (context->awaiting_pop);
2214 context->awaiting_pop = FALSE;
2216 /* valgrind friendliness */
2217 user_data = context->held_user_data;
2218 context->held_user_data = NULL;
2224 append_escaped_text (GString *str,
2233 end = text + length;
2238 next = g_utf8_next_char (p);
2243 g_string_append (str, "&");
2247 g_string_append (str, "<");
2251 g_string_append (str, ">");
2255 g_string_append (str, "'");
2259 g_string_append (str, """);
2263 c = g_utf8_get_char (p);
2264 if ((0x1 <= c && c <= 0x8) ||
2265 (0xb <= c && c <= 0xc) ||
2266 (0xe <= c && c <= 0x1f) ||
2267 (0x7f <= c && c <= 0x84) ||
2268 (0x86 <= c && c <= 0x9f))
2269 g_string_append_printf (str, "&#x%x;", c);
2271 g_string_append_len (str, p, next - p);
2280 * g_markup_escape_text:
2281 * @text: some valid UTF-8 text
2282 * @length: length of @text in bytes, or -1 if the text is nul-terminated
2284 * Escapes text so that the markup parser will parse it verbatim.
2285 * Less than, greater than, ampersand, etc. are replaced with the
2286 * corresponding entities. This function would typically be used
2287 * when writing out a file to be parsed with the markup parser.
2289 * Note that this function doesn't protect whitespace and line endings
2290 * from being processed according to the XML rules for normalization
2291 * of line endings and attribute values.
2293 * Return value: a newly allocated string with the escaped text
2296 g_markup_escape_text (const gchar *text,
2301 g_return_val_if_fail (text != NULL, NULL);
2304 length = strlen (text);
2306 /* prealloc at least as long as original text */
2307 str = g_string_sized_new (length);
2308 append_escaped_text (str, text, length);
2310 return g_string_free (str, FALSE);
2315 * @format: a printf-style format string
2316 * @after: location to store a pointer to the character after
2317 * the returned conversion. On a %NULL return, returns the
2318 * pointer to the trailing NUL in the string
2320 * Find the next conversion in a printf-style format string.
2321 * Partially based on code from printf-parser.c,
2322 * Copyright (C) 1999-2000, 2002-2003 Free Software Foundation, Inc.
2324 * Return value: pointer to the next conversion in @format,
2325 * or %NULL, if none.
2328 find_conversion (const char *format,
2331 const char *start = format;
2334 while (*start != '\0' && *start != '%')
2351 /* Test for positional argument. */
2352 if (*cp >= '0' && *cp <= '9')
2356 for (np = cp; *np >= '0' && *np <= '9'; np++)
2362 /* Skip the flags. */
2376 /* Skip the field width. */
2381 /* Test for positional argument. */
2382 if (*cp >= '0' && *cp <= '9')
2386 for (np = cp; *np >= '0' && *np <= '9'; np++)
2394 for (; *cp >= '0' && *cp <= '9'; cp++)
2398 /* Skip the precision. */
2404 /* Test for positional argument. */
2405 if (*cp >= '0' && *cp <= '9')
2409 for (np = cp; *np >= '0' && *np <= '9'; np++)
2417 for (; *cp >= '0' && *cp <= '9'; cp++)
2422 /* Skip argument type/size specifiers. */
2423 while (*cp == 'h' ||
2432 /* Skip the conversion character. */
2440 * g_markup_vprintf_escaped:
2441 * @format: printf() style format string
2442 * @args: variable argument list, similar to vprintf()
2444 * Formats the data in @args according to @format, escaping
2445 * all string and character arguments in the fashion
2446 * of g_markup_escape_text(). See g_markup_printf_escaped().
2448 * Return value: newly allocated result from formatting
2449 * operation. Free with g_free().
2454 g_markup_vprintf_escaped (const char *format,
2459 GString *result = NULL;
2460 gchar *output1 = NULL;
2461 gchar *output2 = NULL;
2462 const char *p, *op1, *op2;
2465 /* The technique here, is that we make two format strings that
2466 * have the identical conversions in the identical order to the
2467 * original strings, but differ in the text in-between. We
2468 * then use the normal g_strdup_vprintf() to format the arguments
2469 * with the two new format strings. By comparing the results,
2470 * we can figure out what segments of the output come from
2471 * the the original format string, and what from the arguments,
2472 * and thus know what portions of the string to escape.
2474 * For instance, for:
2476 * g_markup_printf_escaped ("%s ate %d apples", "Susan & Fred", 5);
2478 * We form the two format strings "%sX%dX" and %sY%sY". The results
2479 * of formatting with those two strings are
2481 * "%sX%dX" => "Susan & FredX5X"
2482 * "%sY%dY" => "Susan & FredY5Y"
2484 * To find the span of the first argument, we find the first position
2485 * where the two arguments differ, which tells us that the first
2486 * argument formatted to "Susan & Fred". We then escape that
2487 * to "Susan & Fred" and join up with the intermediate portions
2488 * of the format string and the second argument to get
2489 * "Susan & Fred ate 5 apples".
2492 /* Create the two modified format strings
2494 format1 = g_string_new (NULL);
2495 format2 = g_string_new (NULL);
2500 const char *conv = find_conversion (p, &after);
2504 g_string_append_len (format1, conv, after - conv);
2505 g_string_append_c (format1, 'X');
2506 g_string_append_len (format2, conv, after - conv);
2507 g_string_append_c (format2, 'Y');
2512 /* Use them to format the arguments
2514 G_VA_COPY (args2, args);
2516 output1 = g_strdup_vprintf (format1->str, args);
2523 output2 = g_strdup_vprintf (format2->str, args2);
2528 result = g_string_new (NULL);
2530 /* Iterate through the original format string again,
2531 * copying the non-conversion portions and the escaped
2532 * converted arguments to the output string.
2540 const char *output_start;
2541 const char *conv = find_conversion (p, &after);
2544 if (!conv) /* The end, after points to the trailing \0 */
2546 g_string_append_len (result, p, after - p);
2550 g_string_append_len (result, p, conv - p);
2552 while (*op1 == *op2)
2558 escaped = g_markup_escape_text (output_start, op1 - output_start);
2559 g_string_append (result, escaped);
2568 g_string_free (format1, TRUE);
2569 g_string_free (format2, TRUE);
2574 return g_string_free (result, FALSE);
2580 * g_markup_printf_escaped:
2581 * @format: printf() style format string
2582 * @Varargs: the arguments to insert in the format string
2584 * Formats arguments according to @format, escaping
2585 * all string and character arguments in the fashion
2586 * of g_markup_escape_text(). This is useful when you
2587 * want to insert literal strings into XML-style markup
2588 * output, without having to worry that the strings
2589 * might themselves contain markup.
2592 * const char *store = "Fortnum & Mason";
2593 * const char *item = "Tea";
2596 * output = g_markup_printf_escaped ("<purchase>"
2597 * "<store>%s</store>"
2598 * "<item>%s</item>"
2599 * "</purchase>",
2603 * Return value: newly allocated result from formatting
2604 * operation. Free with g_free().
2609 g_markup_printf_escaped (const char *format, ...)
2614 va_start (args, format);
2615 result = g_markup_vprintf_escaped (format, args);
2622 g_markup_parse_boolean (const char *string,
2625 char const * const falses[] = { "false", "f", "no", "n", "0" };
2626 char const * const trues[] = { "true", "t", "yes", "y", "1" };
2629 for (i = 0; i < G_N_ELEMENTS (falses); i++)
2631 if (g_ascii_strcasecmp (string, falses[i]) == 0)
2640 for (i = 0; i < G_N_ELEMENTS (trues); i++)
2642 if (g_ascii_strcasecmp (string, trues[i]) == 0)
2655 * GMarkupCollectType:
2656 * @G_MARKUP_COLLECT_INVALID: used to terminate the list of attributes
2658 * @G_MARKUP_COLLECT_STRING: collect the string pointer directly from
2659 * the attribute_values[] array. Expects a
2660 * parameter of type (const char **). If
2661 * %G_MARKUP_COLLECT_OPTIONAL is specified
2662 * and the attribute isn't present then the
2663 * pointer will be set to %NULL.
2664 * @G_MARKUP_COLLECT_STRDUP: as with %G_MARKUP_COLLECT_STRING, but
2665 * expects a paramter of type (char **) and
2666 * g_strdup()s the returned pointer. The
2667 * pointer must be freed with g_free().
2668 * @G_MARKUP_COLLECT_BOOLEAN: expects a parameter of type (gboolean *)
2669 * and parses the attribute value as a
2670 * boolean. Sets %FALSE if the attribute
2671 * isn't present. Valid boolean values
2672 * consist of (case insensitive) "false",
2673 * "f", "no", "n", "0" and "true", "t",
2675 * @G_MARKUP_COLLECT_TRISTATE: as with %G_MARKUP_COLLECT_BOOLEAN, but
2676 * in the case of a missing attribute a
2677 * value is set that compares equal to
2678 * neither %FALSE nor %TRUE.
2679 * G_MARKUP_COLLECT_OPTIONAL is implied.
2680 * @G_MARKUP_COLLECT_OPTIONAL: can be bitwise ORed with the other
2681 * fields. If present, allows the
2682 * attribute not to appear. A default
2683 * value is set depending on what value
2686 * A mixed enumerated type and flags field. You must specify one type
2687 * (string, strdup, boolean, tristate). Additionally, you may
2688 * optionally bitwise OR the type with the flag
2689 * %G_MARKUP_COLLECT_OPTIONAL.
2691 * It is likely that this enum will be extended in the future to
2692 * support other types.
2696 * g_markup_collect_attributes:
2697 * @element_name: the current tag name
2698 * @attribute_names: the attribute names
2699 * @attribute_values: the attribute values
2700 * @error: a pointer to a #GError or %NULL
2701 * @first_type: the #GMarkupCollectType of the
2703 * @first_attr: the name of the first attribute
2704 * @...: a pointer to the storage location of the
2705 * first attribute (or %NULL), followed by
2706 * more types names and pointers, ending
2707 * with %G_MARKUP_COLLECT_INVALID.
2709 * Collects the attributes of the element from the
2710 * data passed to the #GMarkupParser start_element
2711 * function, dealing with common error conditions
2712 * and supporting boolean values.
2714 * This utility function is not required to write
2715 * a parser but can save a lot of typing.
2717 * The @element_name, @attribute_names,
2718 * @attribute_values and @error parameters passed
2719 * to the start_element callback should be passed
2720 * unmodified to this function.
2722 * Following these arguments is a list of
2723 * "supported" attributes to collect. It is an
2724 * error to specify multiple attributes with the
2725 * same name. If any attribute not in the list
2726 * appears in the @attribute_names array then an
2727 * unknown attribute error will result.
2729 * The #GMarkupCollectType field allows specifying
2730 * the type of collection to perform and if a
2731 * given attribute must appear or is optional.
2733 * The attribute name is simply the name of the
2734 * attribute to collect.
2736 * The pointer should be of the appropriate type
2737 * (see the descriptions under
2738 * #GMarkupCollectType) and may be %NULL in case a
2739 * particular attribute is to be allowed but
2742 * This function deals with issuing errors for missing attributes
2743 * (of type %G_MARKUP_ERROR_MISSING_ATTRIBUTE), unknown attributes
2744 * (of type %G_MARKUP_ERROR_UNKNOWN_ATTRIBUTE) and duplicate
2745 * attributes (of type %G_MARKUP_ERROR_INVALID_CONTENT) as well
2746 * as parse errors for boolean-valued attributes (again of type
2747 * %G_MARKUP_ERROR_INVALID_CONTENT). In all of these cases %FALSE
2748 * will be returned and @error will be set as appropriate.
2750 * Return value: %TRUE if successful
2755 g_markup_collect_attributes (const gchar *element_name,
2756 const gchar **attribute_names,
2757 const gchar **attribute_values,
2759 GMarkupCollectType first_type,
2760 const gchar *first_attr,
2763 GMarkupCollectType type;
2775 va_start (ap, first_attr);
2776 while (type != G_MARKUP_COLLECT_INVALID)
2781 mandatory = !(type & G_MARKUP_COLLECT_OPTIONAL);
2782 type &= (G_MARKUP_COLLECT_OPTIONAL - 1);
2784 /* tristate records a value != TRUE and != FALSE
2785 * for the case where the attribute is missing
2787 if (type == G_MARKUP_COLLECT_TRISTATE)
2790 for (i = 0; attribute_names[i]; i++)
2791 if (i >= 40 || !(collected & (G_GUINT64_CONSTANT(1) << i)))
2792 if (!strcmp (attribute_names[i], attr))
2795 /* ISO C99 only promises that the user can pass up to 127 arguments.
2796 * Subtracting the first 4 arguments plus the final NULL and dividing
2797 * by 3 arguments per collected attribute, we are left with a maximum
2798 * number of supported attributes of (127 - 5) / 3 = 40.
2800 * In reality, nobody is ever going to call us with anywhere close to
2801 * 40 attributes to collect, so it is safe to assume that if i > 40
2802 * then the user has given some invalid or repeated arguments. These
2803 * problems will be caught and reported at the end of the function.
2805 * We know at this point that we have an error, but we don't know
2806 * what error it is, so just continue...
2809 collected |= (G_GUINT64_CONSTANT(1) << i);
2811 value = attribute_values[i];
2813 if (value == NULL && mandatory)
2815 g_set_error (error, G_MARKUP_ERROR,
2816 G_MARKUP_ERROR_MISSING_ATTRIBUTE,
2817 "element '%s' requires attribute '%s'",
2818 element_name, attr);
2826 case G_MARKUP_COLLECT_STRING:
2828 const char **str_ptr;
2830 str_ptr = va_arg (ap, const char **);
2832 if (str_ptr != NULL)
2837 case G_MARKUP_COLLECT_STRDUP:
2841 str_ptr = va_arg (ap, char **);
2843 if (str_ptr != NULL)
2844 *str_ptr = g_strdup (value);
2848 case G_MARKUP_COLLECT_BOOLEAN:
2849 case G_MARKUP_COLLECT_TRISTATE:
2854 bool_ptr = va_arg (ap, gboolean *);
2856 if (bool_ptr != NULL)
2858 if (type == G_MARKUP_COLLECT_TRISTATE)
2859 /* constructivists rejoice!
2860 * neither false nor true...
2864 else /* G_MARKUP_COLLECT_BOOLEAN */
2870 if (!g_markup_parse_boolean (value, va_arg (ap, gboolean *)))
2872 g_set_error (error, G_MARKUP_ERROR,
2873 G_MARKUP_ERROR_INVALID_CONTENT,
2874 "element '%s', attribute '%s', value '%s' "
2875 "cannot be parsed as a boolean value",
2876 element_name, attr, value);
2886 g_assert_not_reached ();
2889 type = va_arg (ap, GMarkupCollectType);
2890 attr = va_arg (ap, const char *);
2895 /* ensure we collected all the arguments */
2896 for (i = 0; attribute_names[i]; i++)
2897 if ((collected & (G_GUINT64_CONSTANT(1) << i)) == 0)
2899 /* attribute not collected: could be caused by two things.
2901 * 1) it doesn't exist in our list of attributes
2902 * 2) it existed but was matched by a duplicate attribute earlier
2908 for (j = 0; j < i; j++)
2909 if (strcmp (attribute_names[i], attribute_names[j]) == 0)
2913 /* j is now the first occurance of attribute_names[i] */
2915 g_set_error (error, G_MARKUP_ERROR,
2916 G_MARKUP_ERROR_UNKNOWN_ATTRIBUTE,
2917 "attribute '%s' invalid for element '%s'",
2918 attribute_names[i], element_name);
2920 g_set_error (error, G_MARKUP_ERROR,
2921 G_MARKUP_ERROR_INVALID_CONTENT,
2922 "attribute '%s' given multiple times for element '%s'",
2923 attribute_names[i], element_name);
2931 /* replay the above to free allocations */
2935 va_start (ap, first_attr);
2936 while (type != G_MARKUP_COLLECT_INVALID)
2940 ptr = va_arg (ap, gpointer);
2945 switch (type & (G_MARKUP_COLLECT_OPTIONAL - 1))
2947 case G_MARKUP_COLLECT_STRDUP:
2949 g_free (*(char **) ptr);
2951 case G_MARKUP_COLLECT_STRING:
2952 *(char **) ptr = NULL;
2955 case G_MARKUP_COLLECT_BOOLEAN:
2956 *(gboolean *) ptr = FALSE;
2959 case G_MARKUP_COLLECT_TRISTATE:
2960 *(gboolean *) ptr = -1;
2964 type = va_arg (ap, GMarkupCollectType);
2965 attr = va_arg (ap, const char *);
2975 #define __G_MARKUP_C__
2976 #include "galiasdef.c"