]>
Commit | Line | Data |
---|---|---|
fd79ecee DJ |
1 | /* Helper routines for parsing XML using Expat. |
2 | ||
d01e8234 | 3 | Copyright (C) 2006-2025 Free Software Foundation, Inc. |
fd79ecee DJ |
4 | |
5 | This file is part of GDB. | |
6 | ||
7 | This program is free software; you can redistribute it and/or modify | |
8 | it under the terms of the GNU General Public License as published by | |
a9762ec7 | 9 | the Free Software Foundation; either version 3 of the License, or |
fd79ecee DJ |
10 | (at your option) any later version. |
11 | ||
12 | This program is distributed in the hope that it will be useful, | |
13 | but WITHOUT ANY WARRANTY; without even the implied warranty of | |
14 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
15 | GNU General Public License for more details. | |
16 | ||
17 | You should have received a copy of the GNU General Public License | |
a9762ec7 | 18 | along with this program. If not, see <http://www.gnu.org/licenses/>. */ |
fd79ecee | 19 | |
5b9707eb | 20 | #include "cli/cli-cmds.h" |
fec4e896 | 21 | #include "xml-builtin.h" |
05a4558a | 22 | #include "xml-support.h" |
268a13a5 | 23 | #include "gdbsupport/filestuff.h" |
e0f4b3ec | 24 | #include "gdbsupport/gdb-safe-ctype.h" |
a7fc9b61 | 25 | #include <vector> |
bd8a901f | 26 | #include <string> |
e776119f DJ |
27 | |
28 | /* Debugging flag. */ | |
491144b5 | 29 | static bool debug_xml; |
fd79ecee DJ |
30 | |
31 | /* The contents of this file are only useful if XML support is | |
32 | available. */ | |
33 | #ifdef HAVE_LIBEXPAT | |
34 | ||
dbc981de | 35 | #include "gdb_expat.h" |
e776119f | 36 | |
108546a0 DJ |
37 | /* The maximum depth of <xi:include> nesting. No need to be miserly, |
38 | we just want to avoid running out of stack on loops. */ | |
39 | #define MAX_XINCLUDE_DEPTH 30 | |
40 | ||
123dc839 DJ |
41 | /* Simplified XML parser infrastructure. */ |
42 | ||
e776119f DJ |
43 | /* A parsing level -- used to keep track of the current element |
44 | nesting. */ | |
45 | struct scope_level | |
46 | { | |
a7fc9b61 PA |
47 | explicit scope_level (const gdb_xml_element *elements_ = NULL) |
48 | : elements (elements_), | |
49 | element (NULL), | |
bd8a901f | 50 | seen (0) |
a7fc9b61 PA |
51 | {} |
52 | ||
e776119f DJ |
53 | /* Elements we allow at this level. */ |
54 | const struct gdb_xml_element *elements; | |
55 | ||
56 | /* The element which we are within. */ | |
57 | const struct gdb_xml_element *element; | |
58 | ||
59 | /* Mask of which elements we've seen at this level (used for | |
60 | optional and repeatable checking). */ | |
61 | unsigned int seen; | |
62 | ||
bd8a901f PA |
63 | /* Body text accumulation. */ |
64 | std::string body; | |
e776119f | 65 | }; |
e776119f DJ |
66 | |
67 | /* The parser itself, and our additional state. */ | |
68 | struct gdb_xml_parser | |
69 | { | |
010151c9 PA |
70 | gdb_xml_parser (const char *name, |
71 | const gdb_xml_element *elements, | |
72 | void *user_data); | |
73 | ~gdb_xml_parser(); | |
74 | ||
4895cde2 PA |
75 | /* Associate DTD_NAME, which must be the name of a compiled-in DTD, |
76 | with the parser. */ | |
77 | void use_dtd (const char *dtd_name); | |
78 | ||
79 | /* Return the name of the expected / default DTD, if specified. */ | |
80 | const char *dtd_name () | |
81 | { return m_dtd_name; } | |
82 | ||
83 | /* Invoke the parser on BUFFER. BUFFER is the data to parse, which | |
84 | should be NUL-terminated. | |
85 | ||
86 | The return value is 0 for success or -1 for error. It may throw, | |
87 | but only if something unexpected goes wrong during parsing; parse | |
88 | errors will be caught, warned about, and reported as failure. */ | |
89 | int parse (const char *buffer); | |
90 | ||
91 | /* Issue a debugging message. */ | |
92 | void vdebug (const char *format, va_list ap) | |
93 | ATTRIBUTE_PRINTF (2, 0); | |
94 | ||
95 | /* Issue an error message, and stop parsing. */ | |
d9deb60b SM |
96 | [[noreturn]] void verror (const char *format, va_list ap) |
97 | ATTRIBUTE_PRINTF (2, 0); | |
4895cde2 PA |
98 | |
99 | void body_text (const XML_Char *text, int length); | |
100 | void start_element (const XML_Char *name, const XML_Char **attrs); | |
101 | void end_element (const XML_Char *name); | |
102 | ||
103 | /* Return the name of this parser. */ | |
104 | const char *name () | |
105 | { return m_name; } | |
106 | ||
107 | /* Return the user's callback data, for handlers. */ | |
108 | void *user_data () | |
109 | { return m_user_data; }; | |
110 | ||
111 | /* Are we the special <xi:include> parser? */ | |
112 | void set_is_xinclude (bool is_xinclude) | |
113 | { m_is_xinclude = is_xinclude; } | |
114 | ||
115 | /* A thrown error, if any. */ | |
94aeb44b | 116 | void set_error (gdb_exception &&error) |
4895cde2 | 117 | { |
94aeb44b | 118 | m_error = std::move (error); |
4895cde2 PA |
119 | #ifdef HAVE_XML_STOPPARSER |
120 | XML_StopParser (m_expat_parser, XML_FALSE); | |
121 | #endif | |
122 | } | |
123 | ||
124 | /* Return the underlying expat parser. */ | |
125 | XML_Parser expat_parser () | |
126 | { return m_expat_parser; } | |
127 | ||
128 | private: | |
129 | /* The underlying expat parser. */ | |
130 | XML_Parser m_expat_parser; | |
131 | ||
132 | /* Name of this parser. */ | |
133 | const char *m_name; | |
e776119f | 134 | |
4895cde2 PA |
135 | /* The user's callback data, for handlers. */ |
136 | void *m_user_data; | |
e776119f | 137 | |
a7fc9b61 | 138 | /* Scoping stack. */ |
4895cde2 | 139 | std::vector<scope_level> m_scopes; |
e776119f | 140 | |
4895cde2 PA |
141 | /* A thrown error, if any. */ |
142 | struct gdb_exception m_error; | |
108546a0 | 143 | |
4895cde2 PA |
144 | /* The line of the thrown error, or 0. */ |
145 | int m_last_line; | |
146 | ||
147 | /* The name of the expected / default DTD, if specified. */ | |
148 | const char *m_dtd_name; | |
149 | ||
150 | /* Are we the special <xi:include> parser? */ | |
151 | bool m_is_xinclude; | |
e776119f DJ |
152 | }; |
153 | ||
154 | /* Process some body text. We accumulate the text for later use; it's | |
155 | wrong to do anything with it immediately, because a single block of | |
156 | text might be broken up into multiple calls to this function. */ | |
157 | ||
4895cde2 PA |
158 | void |
159 | gdb_xml_parser::body_text (const XML_Char *text, int length) | |
160 | { | |
161 | if (m_error.reason < 0) | |
162 | return; | |
163 | ||
164 | scope_level &scope = m_scopes.back (); | |
165 | scope.body.append (text, length); | |
166 | } | |
167 | ||
e776119f DJ |
168 | static void |
169 | gdb_xml_body_text (void *data, const XML_Char *text, int length) | |
170 | { | |
19ba03f4 | 171 | struct gdb_xml_parser *parser = (struct gdb_xml_parser *) data; |
e776119f | 172 | |
4895cde2 | 173 | parser->body_text (text, length); |
e776119f | 174 | } |
fd79ecee | 175 | |
e776119f | 176 | /* Issue a debugging message from one of PARSER's handlers. */ |
fd79ecee | 177 | |
e776119f | 178 | void |
4895cde2 | 179 | gdb_xml_parser::vdebug (const char *format, va_list ap) |
e776119f | 180 | { |
4895cde2 | 181 | int line = XML_GetCurrentLineNumber (m_expat_parser); |
e776119f | 182 | |
467dc1e2 | 183 | std::string message = string_vprintf (format, ap); |
e776119f | 184 | if (line) |
6cb06a8c TT |
185 | gdb_printf (gdb_stderr, "%s (line %d): %s\n", |
186 | m_name, line, message.c_str ()); | |
e776119f | 187 | else |
6cb06a8c TT |
188 | gdb_printf (gdb_stderr, "%s: %s\n", |
189 | m_name, message.c_str ()); | |
e776119f DJ |
190 | } |
191 | ||
4895cde2 PA |
192 | void |
193 | gdb_xml_debug (struct gdb_xml_parser *parser, const char *format, ...) | |
194 | { | |
195 | if (!debug_xml) | |
196 | return; | |
197 | ||
198 | va_list ap; | |
199 | va_start (ap, format); | |
200 | parser->vdebug (format, ap); | |
201 | va_end (ap); | |
202 | } | |
203 | ||
e776119f DJ |
204 | /* Issue an error message from one of PARSER's handlers, and stop |
205 | parsing. */ | |
206 | ||
4895cde2 PA |
207 | void |
208 | gdb_xml_parser::verror (const char *format, va_list ap) | |
209 | { | |
210 | int line = XML_GetCurrentLineNumber (m_expat_parser); | |
211 | ||
212 | m_last_line = line; | |
213 | throw_verror (XML_PARSE_ERROR, format, ap); | |
214 | } | |
215 | ||
e776119f DJ |
216 | void |
217 | gdb_xml_error (struct gdb_xml_parser *parser, const char *format, ...) | |
218 | { | |
e776119f | 219 | va_list ap; |
e776119f | 220 | va_start (ap, format); |
4895cde2 PA |
221 | parser->verror (format, ap); |
222 | va_end (ap); | |
e776119f DJ |
223 | } |
224 | ||
3d2c1d41 PA |
225 | /* Find the attribute named NAME in the set of parsed attributes |
226 | ATTRIBUTES. Returns NULL if not found. */ | |
227 | ||
228 | struct gdb_xml_value * | |
4d0fdd9b SM |
229 | xml_find_attribute (std::vector<gdb_xml_value> &attributes, |
230 | const char *name) | |
3d2c1d41 | 231 | { |
4d0fdd9b SM |
232 | for (gdb_xml_value &value : attributes) |
233 | if (strcmp (value.name, name) == 0) | |
234 | return &value; | |
3d2c1d41 PA |
235 | |
236 | return NULL; | |
237 | } | |
238 | ||
4895cde2 PA |
239 | /* Handle the start of an element. NAME is the element, and ATTRS are |
240 | the names and values of this element's attributes. */ | |
e776119f | 241 | |
4895cde2 PA |
242 | void |
243 | gdb_xml_parser::start_element (const XML_Char *name, | |
244 | const XML_Char **attrs) | |
fd79ecee | 245 | { |
4895cde2 PA |
246 | if (m_error.reason < 0) |
247 | return; | |
248 | ||
e776119f DJ |
249 | const struct gdb_xml_element *element; |
250 | const struct gdb_xml_attribute *attribute; | |
e776119f | 251 | unsigned int seen; |
e776119f | 252 | |
e776119f DJ |
253 | /* Push an error scope. If we return or throw an exception before |
254 | filling this in, it will tell us to ignore children of this | |
a7fc9b61 PA |
255 | element. Note we don't take a reference to the element yet |
256 | because further below we'll process the element which may recurse | |
257 | back here and push more elements to the vector. When the | |
258 | recursion unrolls all such elements will have been popped back | |
259 | already, but if one of those pushes reallocates the vector, | |
260 | previous element references will be invalidated. */ | |
4895cde2 | 261 | m_scopes.emplace_back (); |
a7fc9b61 PA |
262 | |
263 | /* Get a reference to the current scope. */ | |
4895cde2 | 264 | scope_level &scope = m_scopes[m_scopes.size () - 2]; |
e776119f | 265 | |
4895cde2 | 266 | gdb_xml_debug (this, _("Entering element <%s>"), name); |
e776119f DJ |
267 | |
268 | /* Find this element in the list of the current scope's allowed | |
269 | children. Record that we've seen it. */ | |
270 | ||
271 | seen = 1; | |
a7fc9b61 | 272 | for (element = scope.elements; element && element->name; |
e776119f DJ |
273 | element++, seen <<= 1) |
274 | if (strcmp (element->name, name) == 0) | |
275 | break; | |
276 | ||
277 | if (element == NULL || element->name == NULL) | |
278 | { | |
108546a0 DJ |
279 | /* If we're working on XInclude, <xi:include> can be the child |
280 | of absolutely anything. Copy the previous scope's element | |
281 | list into the new scope even if there was no match. */ | |
4895cde2 | 282 | if (m_is_xinclude) |
108546a0 | 283 | { |
4895cde2 | 284 | XML_DefaultCurrent (m_expat_parser); |
108546a0 | 285 | |
4895cde2 | 286 | scope_level &unknown_scope = m_scopes.back (); |
a7fc9b61 | 287 | unknown_scope.elements = scope.elements; |
108546a0 DJ |
288 | return; |
289 | } | |
290 | ||
4895cde2 | 291 | gdb_xml_debug (this, _("Element <%s> unknown"), name); |
e776119f DJ |
292 | return; |
293 | } | |
294 | ||
a7fc9b61 | 295 | if (!(element->flags & GDB_XML_EF_REPEATABLE) && (seen & scope.seen)) |
4895cde2 | 296 | gdb_xml_error (this, _("Element <%s> only expected once"), name); |
e776119f | 297 | |
a7fc9b61 | 298 | scope.seen |= seen; |
e776119f | 299 | |
4d0fdd9b | 300 | std::vector<gdb_xml_value> attributes; |
108546a0 | 301 | |
e776119f DJ |
302 | for (attribute = element->attributes; |
303 | attribute != NULL && attribute->name != NULL; | |
304 | attribute++) | |
305 | { | |
306 | const char *val = NULL; | |
307 | const XML_Char **p; | |
308 | void *parsed_value; | |
e776119f DJ |
309 | |
310 | for (p = attrs; *p != NULL; p += 2) | |
311 | if (!strcmp (attribute->name, p[0])) | |
312 | { | |
313 | val = p[1]; | |
314 | break; | |
315 | } | |
316 | ||
317 | if (*p != NULL && val == NULL) | |
318 | { | |
4895cde2 | 319 | gdb_xml_debug (this, _("Attribute \"%s\" missing a value"), |
e776119f DJ |
320 | attribute->name); |
321 | continue; | |
322 | } | |
323 | ||
324 | if (*p == NULL && !(attribute->flags & GDB_XML_AF_OPTIONAL)) | |
325 | { | |
4895cde2 | 326 | gdb_xml_error (this, _("Required attribute \"%s\" of " |
e776119f DJ |
327 | "<%s> not specified"), |
328 | attribute->name, element->name); | |
329 | continue; | |
330 | } | |
331 | ||
332 | if (*p == NULL) | |
333 | continue; | |
334 | ||
4895cde2 | 335 | gdb_xml_debug (this, _("Parsing attribute %s=\"%s\""), |
e776119f DJ |
336 | attribute->name, val); |
337 | ||
338 | if (attribute->handler) | |
4895cde2 | 339 | parsed_value = attribute->handler (this, attribute, val); |
e776119f DJ |
340 | else |
341 | parsed_value = xstrdup (val); | |
342 | ||
4d0fdd9b | 343 | attributes.emplace_back (attribute->name, parsed_value); |
e776119f DJ |
344 | } |
345 | ||
346 | /* Check for unrecognized attributes. */ | |
347 | if (debug_xml) | |
fd79ecee | 348 | { |
e776119f DJ |
349 | const XML_Char **p; |
350 | ||
351 | for (p = attrs; *p != NULL; p += 2) | |
352 | { | |
353 | for (attribute = element->attributes; | |
354 | attribute != NULL && attribute->name != NULL; | |
355 | attribute++) | |
356 | if (strcmp (attribute->name, *p) == 0) | |
357 | break; | |
358 | ||
359 | if (attribute == NULL || attribute->name == NULL) | |
4895cde2 | 360 | gdb_xml_debug (this, _("Ignoring unknown attribute %s"), *p); |
e776119f DJ |
361 | } |
362 | } | |
363 | ||
364 | /* Call the element handler if there is one. */ | |
365 | if (element->start_handler) | |
4895cde2 | 366 | element->start_handler (this, element, m_user_data, attributes); |
e776119f | 367 | |
a7fc9b61 PA |
368 | /* Fill in a new scope level. Note that we must delay getting a |
369 | back reference till here because above we might have recursed, | |
370 | which may have reallocated the vector which invalidates | |
371 | iterators/pointers/references. */ | |
4895cde2 | 372 | scope_level &new_scope = m_scopes.back (); |
a7fc9b61 PA |
373 | new_scope.element = element; |
374 | new_scope.elements = element->children; | |
e776119f DJ |
375 | } |
376 | ||
377 | /* Wrapper for gdb_xml_start_element, to prevent throwing exceptions | |
378 | through expat. */ | |
379 | ||
380 | static void | |
381 | gdb_xml_start_element_wrapper (void *data, const XML_Char *name, | |
382 | const XML_Char **attrs) | |
383 | { | |
19ba03f4 | 384 | struct gdb_xml_parser *parser = (struct gdb_xml_parser *) data; |
e776119f | 385 | |
a70b8144 | 386 | try |
e776119f | 387 | { |
4895cde2 | 388 | parser->start_element (name, attrs); |
e776119f | 389 | } |
94aeb44b | 390 | catch (gdb_exception &ex) |
e776119f | 391 | { |
94aeb44b | 392 | parser->set_error (std::move (ex)); |
fd79ecee | 393 | } |
e776119f DJ |
394 | } |
395 | ||
4895cde2 | 396 | /* Handle the end of an element. NAME is the current element. */ |
e776119f | 397 | |
4895cde2 PA |
398 | void |
399 | gdb_xml_parser::end_element (const XML_Char *name) | |
e776119f | 400 | { |
4895cde2 PA |
401 | if (m_error.reason < 0) |
402 | return; | |
403 | ||
404 | struct scope_level *scope = &m_scopes.back (); | |
e776119f DJ |
405 | const struct gdb_xml_element *element; |
406 | unsigned int seen; | |
e776119f | 407 | |
4895cde2 | 408 | gdb_xml_debug (this, _("Leaving element <%s>"), name); |
e776119f DJ |
409 | |
410 | for (element = scope->elements, seen = 1; | |
411 | element != NULL && element->name != NULL; | |
412 | element++, seen <<= 1) | |
413 | if ((scope->seen & seen) == 0 | |
414 | && (element->flags & GDB_XML_EF_OPTIONAL) == 0) | |
4895cde2 | 415 | gdb_xml_error (this, _("Required element <%s> is missing"), |
e776119f DJ |
416 | element->name); |
417 | ||
581e13c1 | 418 | /* Call the element processor. */ |
108546a0 | 419 | if (scope->element != NULL && scope->element->end_handler) |
e776119f | 420 | { |
bd8a901f | 421 | const char *body; |
e776119f | 422 | |
bd8a901f PA |
423 | if (scope->body.empty ()) |
424 | body = ""; | |
108546a0 DJ |
425 | else |
426 | { | |
427 | int length; | |
e776119f | 428 | |
bd8a901f PA |
429 | length = scope->body.size (); |
430 | body = scope->body.c_str (); | |
e776119f | 431 | |
108546a0 | 432 | /* Strip leading and trailing whitespace. */ |
bd8a901f PA |
433 | while (length > 0 && ISSPACE (body[length - 1])) |
434 | length--; | |
435 | scope->body.erase (length); | |
108546a0 DJ |
436 | while (*body && ISSPACE (*body)) |
437 | body++; | |
438 | } | |
439 | ||
4895cde2 PA |
440 | scope->element->end_handler (this, scope->element, |
441 | m_user_data, body); | |
108546a0 DJ |
442 | } |
443 | else if (scope->element == NULL) | |
4895cde2 | 444 | XML_DefaultCurrent (m_expat_parser); |
e776119f DJ |
445 | |
446 | /* Pop the scope level. */ | |
4895cde2 | 447 | m_scopes.pop_back (); |
e776119f DJ |
448 | } |
449 | ||
450 | /* Wrapper for gdb_xml_end_element, to prevent throwing exceptions | |
451 | through expat. */ | |
452 | ||
453 | static void | |
454 | gdb_xml_end_element_wrapper (void *data, const XML_Char *name) | |
455 | { | |
19ba03f4 | 456 | struct gdb_xml_parser *parser = (struct gdb_xml_parser *) data; |
e776119f | 457 | |
a70b8144 | 458 | try |
e776119f | 459 | { |
4895cde2 | 460 | parser->end_element (name); |
e776119f | 461 | } |
94aeb44b | 462 | catch (gdb_exception &ex) |
e776119f | 463 | { |
94aeb44b | 464 | parser->set_error (std::move (ex)); |
e776119f DJ |
465 | } |
466 | } | |
467 | ||
468 | /* Free a parser and all its associated state. */ | |
469 | ||
010151c9 | 470 | gdb_xml_parser::~gdb_xml_parser () |
e776119f | 471 | { |
4895cde2 | 472 | XML_ParserFree (m_expat_parser); |
e776119f DJ |
473 | } |
474 | ||
010151c9 PA |
475 | /* Initialize a parser. */ |
476 | ||
4895cde2 | 477 | gdb_xml_parser::gdb_xml_parser (const char *name, |
010151c9 | 478 | const gdb_xml_element *elements, |
4895cde2 PA |
479 | void *user_data) |
480 | : m_name (name), | |
481 | m_user_data (user_data), | |
4895cde2 PA |
482 | m_last_line (0), |
483 | m_dtd_name (NULL), | |
484 | m_is_xinclude (false) | |
485 | { | |
486 | m_expat_parser = XML_ParserCreateNS (NULL, '!'); | |
487 | if (m_expat_parser == NULL) | |
010151c9 | 488 | malloc_failure (0); |
e776119f | 489 | |
4895cde2 | 490 | XML_SetUserData (m_expat_parser, this); |
e776119f DJ |
491 | |
492 | /* Set the callbacks. */ | |
4895cde2 | 493 | XML_SetElementHandler (m_expat_parser, gdb_xml_start_element_wrapper, |
e776119f | 494 | gdb_xml_end_element_wrapper); |
4895cde2 | 495 | XML_SetCharacterDataHandler (m_expat_parser, gdb_xml_body_text); |
e776119f DJ |
496 | |
497 | /* Initialize the outer scope. */ | |
4895cde2 | 498 | m_scopes.emplace_back (elements); |
efc0eabd PA |
499 | } |
500 | ||
108546a0 DJ |
501 | /* External entity handler. The only external entities we support |
502 | are those compiled into GDB (we do not fetch entities from the | |
503 | target). */ | |
504 | ||
505 | static int XMLCALL | |
506 | gdb_xml_fetch_external_entity (XML_Parser expat_parser, | |
507 | const XML_Char *context, | |
508 | const XML_Char *base, | |
509 | const XML_Char *systemId, | |
510 | const XML_Char *publicId) | |
511 | { | |
108546a0 DJ |
512 | XML_Parser entity_parser; |
513 | const char *text; | |
514 | enum XML_Status status; | |
515 | ||
516 | if (systemId == NULL) | |
517 | { | |
4895cde2 PA |
518 | gdb_xml_parser *parser |
519 | = (gdb_xml_parser *) XML_GetUserData (expat_parser); | |
520 | ||
521 | text = fetch_xml_builtin (parser->dtd_name ()); | |
108546a0 | 522 | if (text == NULL) |
f34652de | 523 | internal_error (_("could not locate built-in DTD %s"), |
4895cde2 | 524 | parser->dtd_name ()); |
108546a0 DJ |
525 | } |
526 | else | |
527 | { | |
528 | text = fetch_xml_builtin (systemId); | |
529 | if (text == NULL) | |
530 | return XML_STATUS_ERROR; | |
531 | } | |
532 | ||
4895cde2 PA |
533 | entity_parser = XML_ExternalEntityParserCreate (expat_parser, |
534 | context, NULL); | |
108546a0 DJ |
535 | |
536 | /* Don't use our handlers for the contents of the DTD. Just let expat | |
537 | process it. */ | |
538 | XML_SetElementHandler (entity_parser, NULL, NULL); | |
539 | XML_SetDoctypeDeclHandler (entity_parser, NULL, NULL); | |
540 | XML_SetXmlDeclHandler (entity_parser, NULL); | |
541 | XML_SetDefaultHandler (entity_parser, NULL); | |
542 | XML_SetUserData (entity_parser, NULL); | |
543 | ||
544 | status = XML_Parse (entity_parser, text, strlen (text), 1); | |
545 | ||
546 | XML_ParserFree (entity_parser); | |
547 | return status; | |
548 | } | |
549 | ||
108546a0 | 550 | void |
4895cde2 | 551 | gdb_xml_parser::use_dtd (const char *dtd_name) |
108546a0 DJ |
552 | { |
553 | enum XML_Error err; | |
554 | ||
4895cde2 | 555 | m_dtd_name = dtd_name; |
108546a0 | 556 | |
4895cde2 | 557 | XML_SetParamEntityParsing (m_expat_parser, |
108546a0 | 558 | XML_PARAM_ENTITY_PARSING_UNLESS_STANDALONE); |
4895cde2 | 559 | XML_SetExternalEntityRefHandler (m_expat_parser, |
108546a0 DJ |
560 | gdb_xml_fetch_external_entity); |
561 | ||
562 | /* Even if no DTD is provided, use the built-in DTD anyway. */ | |
4895cde2 | 563 | err = XML_UseForeignDTD (m_expat_parser, XML_TRUE); |
108546a0 | 564 | if (err != XML_ERROR_NONE) |
f34652de | 565 | internal_error (_("XML_UseForeignDTD failed: %s"), |
9b20d036 | 566 | XML_ErrorString (err)); |
108546a0 DJ |
567 | } |
568 | ||
e776119f DJ |
569 | /* Invoke PARSER on BUFFER. BUFFER is the data to parse, which |
570 | should be NUL-terminated. | |
571 | ||
572 | The return value is 0 for success or -1 for error. It may throw, | |
573 | but only if something unexpected goes wrong during parsing; parse | |
574 | errors will be caught, warned about, and reported as failure. */ | |
575 | ||
576 | int | |
4895cde2 | 577 | gdb_xml_parser::parse (const char *buffer) |
e776119f DJ |
578 | { |
579 | enum XML_Status status; | |
580 | const char *error_string; | |
581 | ||
4895cde2 | 582 | gdb_xml_debug (this, _("Starting:\n%s"), buffer); |
de584861 | 583 | |
4895cde2 | 584 | status = XML_Parse (m_expat_parser, buffer, strlen (buffer), 1); |
e776119f | 585 | |
4895cde2 | 586 | if (status == XML_STATUS_OK && m_error.reason == 0) |
e776119f DJ |
587 | return 0; |
588 | ||
4895cde2 PA |
589 | if (m_error.reason == RETURN_ERROR |
590 | && m_error.error == XML_PARSE_ERROR) | |
e776119f | 591 | { |
4895cde2 | 592 | gdb_assert (m_error.message != NULL); |
3d6e9d23 | 593 | error_string = m_error.what (); |
e776119f DJ |
594 | } |
595 | else if (status == XML_STATUS_ERROR) | |
596 | { | |
4895cde2 | 597 | enum XML_Error err = XML_GetErrorCode (m_expat_parser); |
a109c7c1 | 598 | |
e776119f DJ |
599 | error_string = XML_ErrorString (err); |
600 | } | |
601 | else | |
602 | { | |
4895cde2 | 603 | gdb_assert (m_error.reason < 0); |
94aeb44b | 604 | throw_exception (std::move (m_error)); |
e776119f DJ |
605 | } |
606 | ||
4895cde2 PA |
607 | if (m_last_line != 0) |
608 | warning (_("while parsing %s (at line %d): %s"), m_name, | |
609 | m_last_line, error_string); | |
e776119f | 610 | else |
4895cde2 | 611 | warning (_("while parsing %s: %s"), m_name, error_string); |
e776119f DJ |
612 | |
613 | return -1; | |
fd79ecee DJ |
614 | } |
615 | ||
efc0eabd PA |
616 | int |
617 | gdb_xml_parse_quick (const char *name, const char *dtd_name, | |
618 | const struct gdb_xml_element *elements, | |
619 | const char *document, void *user_data) | |
620 | { | |
010151c9 | 621 | gdb_xml_parser parser (name, elements, user_data); |
efc0eabd | 622 | if (dtd_name != NULL) |
4895cde2 PA |
623 | parser.use_dtd (dtd_name); |
624 | return parser.parse (document); | |
efc0eabd PA |
625 | } |
626 | ||
fd79ecee DJ |
627 | /* Parse a field VALSTR that we expect to contain an integer value. |
628 | The integer is returned in *VALP. The string is parsed with an | |
629 | equivalent to strtoul. | |
630 | ||
631 | Returns 0 for success, -1 for error. */ | |
632 | ||
633 | static int | |
634 | xml_parse_unsigned_integer (const char *valstr, ULONGEST *valp) | |
635 | { | |
636 | const char *endptr; | |
637 | ULONGEST result; | |
638 | ||
639 | if (*valstr == '\0') | |
640 | return -1; | |
641 | ||
642 | result = strtoulst (valstr, &endptr, 0); | |
643 | if (*endptr != '\0') | |
644 | return -1; | |
645 | ||
646 | *valp = result; | |
647 | return 0; | |
648 | } | |
649 | ||
e776119f DJ |
650 | /* Parse an integer string into a ULONGEST and return it, or call |
651 | gdb_xml_error if it could not be parsed. */ | |
fd79ecee DJ |
652 | |
653 | ULONGEST | |
e776119f | 654 | gdb_xml_parse_ulongest (struct gdb_xml_parser *parser, const char *value) |
fd79ecee DJ |
655 | { |
656 | ULONGEST result; | |
fd79ecee DJ |
657 | |
658 | if (xml_parse_unsigned_integer (value, &result) != 0) | |
e776119f DJ |
659 | gdb_xml_error (parser, _("Can't convert \"%s\" to an integer"), value); |
660 | ||
fd79ecee DJ |
661 | return result; |
662 | } | |
663 | ||
e776119f DJ |
664 | /* Parse an integer attribute into a ULONGEST. */ |
665 | ||
666 | void * | |
667 | gdb_xml_parse_attr_ulongest (struct gdb_xml_parser *parser, | |
668 | const struct gdb_xml_attribute *attribute, | |
669 | const char *value) | |
670 | { | |
671 | ULONGEST result; | |
672 | void *ret; | |
fd79ecee | 673 | |
e776119f DJ |
674 | if (xml_parse_unsigned_integer (value, &result) != 0) |
675 | gdb_xml_error (parser, _("Can't convert %s=\"%s\" to an integer"), | |
676 | attribute->name, value); | |
fd79ecee | 677 | |
8d749320 | 678 | ret = XNEW (ULONGEST); |
e776119f DJ |
679 | memcpy (ret, &result, sizeof (result)); |
680 | return ret; | |
681 | } | |
682 | ||
123dc839 DJ |
683 | /* A handler_data for yes/no boolean values. */ |
684 | ||
685 | const struct gdb_xml_enum gdb_xml_enums_boolean[] = { | |
686 | { "yes", 1 }, | |
687 | { "no", 0 }, | |
688 | { NULL, 0 } | |
689 | }; | |
690 | ||
e776119f DJ |
691 | /* Map NAME to VALUE. A struct gdb_xml_enum * should be saved as the |
692 | value of handler_data when using gdb_xml_parse_attr_enum to parse a | |
693 | fixed list of possible strings. The list is terminated by an entry | |
694 | with NAME == NULL. */ | |
695 | ||
696 | void * | |
697 | gdb_xml_parse_attr_enum (struct gdb_xml_parser *parser, | |
698 | const struct gdb_xml_attribute *attribute, | |
699 | const char *value) | |
fd79ecee | 700 | { |
19ba03f4 SM |
701 | const struct gdb_xml_enum *enums |
702 | = (const struct gdb_xml_enum *) attribute->handler_data; | |
e776119f | 703 | void *ret; |
fd79ecee | 704 | |
19ba03f4 SM |
705 | for (enums = (const struct gdb_xml_enum *) attribute->handler_data; |
706 | enums->name != NULL; enums++) | |
123dc839 | 707 | if (strcasecmp (enums->name, value) == 0) |
e776119f DJ |
708 | break; |
709 | ||
710 | if (enums->name == NULL) | |
711 | gdb_xml_error (parser, _("Unknown attribute value %s=\"%s\""), | |
712 | attribute->name, value); | |
713 | ||
714 | ret = xmalloc (sizeof (enums->value)); | |
715 | memcpy (ret, &enums->value, sizeof (enums->value)); | |
716 | return ret; | |
fd79ecee | 717 | } |
108546a0 DJ |
718 | \f |
719 | ||
720 | /* XInclude processing. This is done as a separate step from actually | |
721 | parsing the document, so that we can produce a single combined XML | |
722 | document - e.g. to hand to a front end or to simplify comparing two | |
723 | documents. We make extensive use of XML_DefaultCurrent, to pass | |
724 | input text directly into the output without reformatting or | |
725 | requoting it. | |
726 | ||
727 | We output the DOCTYPE declaration for the first document unchanged, | |
728 | if present, and discard DOCTYPEs from included documents. Only the | |
729 | one we pass through here is used when we feed the result back to | |
730 | expat. The XInclude standard explicitly does not discuss | |
731 | validation of the result; we choose to apply the same DTD applied | |
732 | to the outermost document. | |
733 | ||
734 | We can not simply include the external DTD subset in the document | |
735 | as an internal subset, because <!IGNORE> and <!INCLUDE> are valid | |
736 | only in external subsets. But if we do not pass the DTD into the | |
737 | output at all, default values will not be filled in. | |
738 | ||
739 | We don't pass through any <?xml> declaration because we generate | |
740 | UTF-8, not whatever the input encoding was. */ | |
741 | ||
742 | struct xinclude_parsing_data | |
743 | { | |
bd8a901f | 744 | xinclude_parsing_data (std::string &output_, |
8400a90d | 745 | xml_fetch_another fetcher_, |
010151c9 | 746 | int include_depth_) |
bd8a901f PA |
747 | : output (output_), |
748 | skip_depth (0), | |
010151c9 | 749 | include_depth (include_depth_), |
8400a90d | 750 | fetcher (fetcher_) |
bd8a901f | 751 | {} |
010151c9 | 752 | |
bd8a901f PA |
753 | /* Where the output goes. */ |
754 | std::string &output; | |
108546a0 DJ |
755 | |
756 | /* A count indicating whether we are in an element whose | |
757 | children should not be copied to the output, and if so, | |
758 | how deep we are nested. This is used for anything inside | |
759 | an xi:include, and for the DTD. */ | |
760 | int skip_depth; | |
761 | ||
762 | /* The number of <xi:include> elements currently being processed, | |
763 | to detect loops. */ | |
764 | int include_depth; | |
765 | ||
766 | /* A function to call to obtain additional features, and its | |
767 | baton. */ | |
768 | xml_fetch_another fetcher; | |
108546a0 DJ |
769 | }; |
770 | ||
771 | static void | |
772 | xinclude_start_include (struct gdb_xml_parser *parser, | |
773 | const struct gdb_xml_element *element, | |
4d0fdd9b SM |
774 | void *user_data, |
775 | std::vector<gdb_xml_value> &attributes) | |
108546a0 | 776 | { |
19ba03f4 SM |
777 | struct xinclude_parsing_data *data |
778 | = (struct xinclude_parsing_data *) user_data; | |
4d0fdd9b | 779 | char *href = (char *) xml_find_attribute (attributes, "href")->value.get (); |
108546a0 DJ |
780 | |
781 | gdb_xml_debug (parser, _("Processing XInclude of \"%s\""), href); | |
782 | ||
783 | if (data->include_depth > MAX_XINCLUDE_DEPTH) | |
784 | gdb_xml_error (parser, _("Maximum XInclude depth (%d) exceeded"), | |
785 | MAX_XINCLUDE_DEPTH); | |
786 | ||
6b09f134 | 787 | std::optional<gdb::char_vector> text = data->fetcher (href); |
9018be22 | 788 | if (!text) |
108546a0 | 789 | gdb_xml_error (parser, _("Could not load XML document \"%s\""), href); |
108546a0 | 790 | |
4895cde2 | 791 | if (!xml_process_xincludes (data->output, parser->name (), |
9018be22 | 792 | text->data (), data->fetcher, |
bd8a901f | 793 | data->include_depth + 1)) |
108546a0 DJ |
794 | gdb_xml_error (parser, _("Parsing \"%s\" failed"), href); |
795 | ||
108546a0 DJ |
796 | data->skip_depth++; |
797 | } | |
798 | ||
799 | static void | |
800 | xinclude_end_include (struct gdb_xml_parser *parser, | |
801 | const struct gdb_xml_element *element, | |
802 | void *user_data, const char *body_text) | |
803 | { | |
19ba03f4 SM |
804 | struct xinclude_parsing_data *data |
805 | = (struct xinclude_parsing_data *) user_data; | |
108546a0 DJ |
806 | |
807 | data->skip_depth--; | |
808 | } | |
809 | ||
810 | static void XMLCALL | |
811 | xml_xinclude_default (void *data_, const XML_Char *s, int len) | |
812 | { | |
19ba03f4 | 813 | struct gdb_xml_parser *parser = (struct gdb_xml_parser *) data_; |
4895cde2 | 814 | xinclude_parsing_data *data = (xinclude_parsing_data *) parser->user_data (); |
108546a0 DJ |
815 | |
816 | /* If we are inside of e.g. xi:include or the DTD, don't save this | |
817 | string. */ | |
818 | if (data->skip_depth) | |
819 | return; | |
820 | ||
821 | /* Otherwise just add it to the end of the document we're building | |
822 | up. */ | |
bd8a901f | 823 | data->output.append (s, len); |
108546a0 DJ |
824 | } |
825 | ||
826 | static void XMLCALL | |
827 | xml_xinclude_start_doctype (void *data_, const XML_Char *doctypeName, | |
828 | const XML_Char *sysid, const XML_Char *pubid, | |
829 | int has_internal_subset) | |
830 | { | |
19ba03f4 | 831 | struct gdb_xml_parser *parser = (struct gdb_xml_parser *) data_; |
4895cde2 | 832 | xinclude_parsing_data *data = (xinclude_parsing_data *) parser->user_data (); |
108546a0 DJ |
833 | |
834 | /* Don't print out the doctype, or the contents of the DTD internal | |
835 | subset, if any. */ | |
836 | data->skip_depth++; | |
837 | } | |
838 | ||
839 | static void XMLCALL | |
840 | xml_xinclude_end_doctype (void *data_) | |
841 | { | |
19ba03f4 | 842 | struct gdb_xml_parser *parser = (struct gdb_xml_parser *) data_; |
4895cde2 | 843 | xinclude_parsing_data *data = (xinclude_parsing_data *) parser->user_data (); |
108546a0 DJ |
844 | |
845 | data->skip_depth--; | |
846 | } | |
847 | ||
848 | static void XMLCALL | |
849 | xml_xinclude_xml_decl (void *data_, const XML_Char *version, | |
850 | const XML_Char *encoding, int standalone) | |
851 | { | |
852 | /* Do nothing - this function prevents the default handler from | |
853 | being called, thus suppressing the XML declaration from the | |
854 | output. */ | |
855 | } | |
856 | ||
108546a0 DJ |
857 | const struct gdb_xml_attribute xinclude_attributes[] = { |
858 | { "href", GDB_XML_AF_NONE, NULL, NULL }, | |
859 | { NULL, GDB_XML_AF_NONE, NULL, NULL } | |
860 | }; | |
861 | ||
862 | const struct gdb_xml_element xinclude_elements[] = { | |
863 | { "http://www.w3.org/2001/XInclude!include", xinclude_attributes, NULL, | |
864 | GDB_XML_EF_OPTIONAL | GDB_XML_EF_REPEATABLE, | |
865 | xinclude_start_include, xinclude_end_include }, | |
866 | { NULL, NULL, NULL, GDB_XML_EF_NONE, NULL, NULL } | |
867 | }; | |
868 | ||
869 | /* The main entry point for <xi:include> processing. */ | |
870 | ||
bd8a901f PA |
871 | bool |
872 | xml_process_xincludes (std::string &result, | |
873 | const char *name, const char *text, | |
8400a90d | 874 | xml_fetch_another fetcher, int depth) |
108546a0 | 875 | { |
8400a90d | 876 | xinclude_parsing_data data (result, fetcher, depth); |
108546a0 | 877 | |
010151c9 | 878 | gdb_xml_parser parser (name, xinclude_elements, &data); |
4895cde2 | 879 | parser.set_is_xinclude (true); |
108546a0 | 880 | |
4895cde2 PA |
881 | XML_SetCharacterDataHandler (parser.expat_parser (), NULL); |
882 | XML_SetDefaultHandler (parser.expat_parser (), xml_xinclude_default); | |
108546a0 DJ |
883 | |
884 | /* Always discard the XML version declarations; the only important | |
885 | thing this provides is encoding, and our result will have been | |
886 | converted to UTF-8. */ | |
4895cde2 | 887 | XML_SetXmlDeclHandler (parser.expat_parser (), xml_xinclude_xml_decl); |
108546a0 DJ |
888 | |
889 | if (depth > 0) | |
890 | /* Discard the doctype for included documents. */ | |
4895cde2 | 891 | XML_SetDoctypeDeclHandler (parser.expat_parser (), |
108546a0 DJ |
892 | xml_xinclude_start_doctype, |
893 | xml_xinclude_end_doctype); | |
894 | ||
4895cde2 | 895 | parser.use_dtd ("xinclude.dtd"); |
108546a0 | 896 | |
4895cde2 | 897 | if (parser.parse (text) == 0) |
108546a0 | 898 | { |
108546a0 | 899 | if (depth == 0) |
010151c9 | 900 | gdb_xml_debug (&parser, _("XInclude processing succeeded.")); |
bd8a901f | 901 | return true; |
108546a0 | 902 | } |
108546a0 | 903 | |
bd8a901f | 904 | return false; |
108546a0 | 905 | } |
05a4558a | 906 | #endif /* HAVE_LIBEXPAT */ |
108546a0 DJ |
907 | \f |
908 | ||
909 | /* Return an XML document which was compiled into GDB, from | |
910 | the given FILENAME, or NULL if the file was not compiled in. */ | |
911 | ||
912 | const char * | |
913 | fetch_xml_builtin (const char *filename) | |
914 | { | |
fec4e896 | 915 | const char *const (*p)[2]; |
108546a0 DJ |
916 | |
917 | for (p = xml_builtin; (*p)[0]; p++) | |
918 | if (strcmp ((*p)[0], filename) == 0) | |
919 | return (*p)[1]; | |
920 | ||
921 | return NULL; | |
922 | } | |
fd79ecee | 923 | |
05a4558a DJ |
924 | /* A to_xfer_partial helper function which reads XML files which were |
925 | compiled into GDB. The target may call this function from its own | |
926 | to_xfer_partial handler, after converting object and annex to the | |
927 | appropriate filename. */ | |
928 | ||
929 | LONGEST | |
930 | xml_builtin_xfer_partial (const char *filename, | |
931 | gdb_byte *readbuf, const gdb_byte *writebuf, | |
932 | ULONGEST offset, LONGEST len) | |
933 | { | |
934 | const char *buf; | |
935 | LONGEST len_avail; | |
936 | ||
937 | gdb_assert (readbuf != NULL && writebuf == NULL); | |
938 | gdb_assert (filename != NULL); | |
939 | ||
940 | buf = fetch_xml_builtin (filename); | |
941 | if (buf == NULL) | |
942 | return -1; | |
943 | ||
944 | len_avail = strlen (buf); | |
945 | if (offset >= len_avail) | |
946 | return 0; | |
947 | ||
948 | if (len > len_avail - offset) | |
949 | len = len_avail - offset; | |
950 | memcpy (readbuf, buf + offset, len); | |
951 | return len; | |
952 | } | |
953 | \f | |
fd79ecee DJ |
954 | |
955 | static void | |
e776119f DJ |
956 | show_debug_xml (struct ui_file *file, int from_tty, |
957 | struct cmd_list_element *c, const char *value) | |
fd79ecee | 958 | { |
6cb06a8c | 959 | gdb_printf (file, _("XML debugging is %s.\n"), value); |
fd79ecee DJ |
960 | } |
961 | ||
6b09f134 | 962 | std::optional<gdb::char_vector> |
8400a90d | 963 | xml_fetch_content_from_file (const char *filename, const char *dirname) |
a96d9b2e | 964 | { |
d419f42d | 965 | gdb_file_up file; |
a96d9b2e | 966 | |
8400a90d | 967 | if (dirname != nullptr && *dirname != '\0') |
a96d9b2e | 968 | { |
33f4dd48 SM |
969 | gdb::unique_xmalloc_ptr<char> fullname |
970 | (concat (dirname, "/", filename, (char *) NULL)); | |
a109c7c1 | 971 | |
33f4dd48 | 972 | file = gdb_fopen_cloexec (fullname.get (), FOPEN_RB); |
a96d9b2e SDJ |
973 | } |
974 | else | |
f69656d0 | 975 | file = gdb_fopen_cloexec (filename, FOPEN_RB); |
a96d9b2e SDJ |
976 | |
977 | if (file == NULL) | |
9018be22 | 978 | return {}; |
a96d9b2e | 979 | |
2edf834e | 980 | /* Read in the whole file. */ |
a96d9b2e | 981 | |
2edf834e | 982 | size_t len; |
a96d9b2e | 983 | |
2edf834e PA |
984 | if (fseek (file.get (), 0, SEEK_END) == -1) |
985 | perror_with_name (_("seek to end of file")); | |
986 | len = ftell (file.get ()); | |
987 | rewind (file.get ()); | |
a96d9b2e | 988 | |
9018be22 | 989 | gdb::char_vector text (len + 1); |
a96d9b2e | 990 | |
9018be22 | 991 | if (fread (text.data (), 1, len, file.get ()) != len |
63929e84 | 992 | || ferror (file.get ())) |
2edf834e PA |
993 | { |
994 | warning (_("Read error from \"%s\""), filename); | |
9018be22 | 995 | return {}; |
a96d9b2e SDJ |
996 | } |
997 | ||
9018be22 | 998 | text.back () = '\0'; |
a96d9b2e SDJ |
999 | return text; |
1000 | } | |
1001 | ||
5fe70629 | 1002 | INIT_GDB_FILE (xml_support) |
fd79ecee | 1003 | { |
e776119f DJ |
1004 | add_setshow_boolean_cmd ("xml", class_maintenance, &debug_xml, |
1005 | _("Set XML parser debugging."), | |
1006 | _("Show XML parser debugging."), | |
1007 | _("When set, debugging messages for XML parsers " | |
1008 | "are displayed."), | |
1009 | NULL, show_debug_xml, | |
1010 | &setdebuglist, &showdebuglist); | |
fd79ecee | 1011 | } |