Commit | Line | Data |
---|---|---|
fd79ecee DJ |
1 | /* Helper routines for parsing XML using Expat. |
2 | ||
b811d2c2 | 3 | Copyright (C) 2006-2020 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 DJ |
19 | |
20 | #include "defs.h" | |
e776119f | 21 | #include "gdbcmd.h" |
fec4e896 | 22 | #include "xml-builtin.h" |
05a4558a | 23 | #include "xml-support.h" |
268a13a5 | 24 | #include "gdbsupport/filestuff.h" |
05a4558a | 25 | #include "safe-ctype.h" |
a7fc9b61 | 26 | #include <vector> |
bd8a901f | 27 | #include <string> |
e776119f DJ |
28 | |
29 | /* Debugging flag. */ | |
491144b5 | 30 | static bool debug_xml; |
fd79ecee DJ |
31 | |
32 | /* The contents of this file are only useful if XML support is | |
33 | available. */ | |
34 | #ifdef HAVE_LIBEXPAT | |
35 | ||
dbc981de | 36 | #include "gdb_expat.h" |
e776119f | 37 | |
108546a0 DJ |
38 | /* The maximum depth of <xi:include> nesting. No need to be miserly, |
39 | we just want to avoid running out of stack on loops. */ | |
40 | #define MAX_XINCLUDE_DEPTH 30 | |
41 | ||
123dc839 DJ |
42 | /* Simplified XML parser infrastructure. */ |
43 | ||
e776119f DJ |
44 | /* A parsing level -- used to keep track of the current element |
45 | nesting. */ | |
46 | struct scope_level | |
47 | { | |
a7fc9b61 PA |
48 | explicit scope_level (const gdb_xml_element *elements_ = NULL) |
49 | : elements (elements_), | |
50 | element (NULL), | |
bd8a901f | 51 | seen (0) |
a7fc9b61 PA |
52 | {} |
53 | ||
e776119f DJ |
54 | /* Elements we allow at this level. */ |
55 | const struct gdb_xml_element *elements; | |
56 | ||
57 | /* The element which we are within. */ | |
58 | const struct gdb_xml_element *element; | |
59 | ||
60 | /* Mask of which elements we've seen at this level (used for | |
61 | optional and repeatable checking). */ | |
62 | unsigned int seen; | |
63 | ||
bd8a901f PA |
64 | /* Body text accumulation. */ |
65 | std::string body; | |
e776119f | 66 | }; |
e776119f DJ |
67 | |
68 | /* The parser itself, and our additional state. */ | |
69 | struct gdb_xml_parser | |
70 | { | |
010151c9 PA |
71 | gdb_xml_parser (const char *name, |
72 | const gdb_xml_element *elements, | |
73 | void *user_data); | |
74 | ~gdb_xml_parser(); | |
75 | ||
4895cde2 PA |
76 | /* Associate DTD_NAME, which must be the name of a compiled-in DTD, |
77 | with the parser. */ | |
78 | void use_dtd (const char *dtd_name); | |
79 | ||
80 | /* Return the name of the expected / default DTD, if specified. */ | |
81 | const char *dtd_name () | |
82 | { return m_dtd_name; } | |
83 | ||
84 | /* Invoke the parser on BUFFER. BUFFER is the data to parse, which | |
85 | should be NUL-terminated. | |
86 | ||
87 | The return value is 0 for success or -1 for error. It may throw, | |
88 | but only if something unexpected goes wrong during parsing; parse | |
89 | errors will be caught, warned about, and reported as failure. */ | |
90 | int parse (const char *buffer); | |
91 | ||
92 | /* Issue a debugging message. */ | |
93 | void vdebug (const char *format, va_list ap) | |
94 | ATTRIBUTE_PRINTF (2, 0); | |
95 | ||
96 | /* Issue an error message, and stop parsing. */ | |
97 | void verror (const char *format, va_list ap) | |
98 | ATTRIBUTE_NORETURN ATTRIBUTE_PRINTF (2, 0); | |
99 | ||
100 | void body_text (const XML_Char *text, int length); | |
101 | void start_element (const XML_Char *name, const XML_Char **attrs); | |
102 | void end_element (const XML_Char *name); | |
103 | ||
104 | /* Return the name of this parser. */ | |
105 | const char *name () | |
106 | { return m_name; } | |
107 | ||
108 | /* Return the user's callback data, for handlers. */ | |
109 | void *user_data () | |
110 | { return m_user_data; }; | |
111 | ||
112 | /* Are we the special <xi:include> parser? */ | |
113 | void set_is_xinclude (bool is_xinclude) | |
114 | { m_is_xinclude = is_xinclude; } | |
115 | ||
116 | /* A thrown error, if any. */ | |
94aeb44b | 117 | void set_error (gdb_exception &&error) |
4895cde2 | 118 | { |
94aeb44b | 119 | m_error = std::move (error); |
4895cde2 PA |
120 | #ifdef HAVE_XML_STOPPARSER |
121 | XML_StopParser (m_expat_parser, XML_FALSE); | |
122 | #endif | |
123 | } | |
124 | ||
125 | /* Return the underlying expat parser. */ | |
126 | XML_Parser expat_parser () | |
127 | { return m_expat_parser; } | |
128 | ||
129 | private: | |
130 | /* The underlying expat parser. */ | |
131 | XML_Parser m_expat_parser; | |
132 | ||
133 | /* Name of this parser. */ | |
134 | const char *m_name; | |
e776119f | 135 | |
4895cde2 PA |
136 | /* The user's callback data, for handlers. */ |
137 | void *m_user_data; | |
e776119f | 138 | |
a7fc9b61 | 139 | /* Scoping stack. */ |
4895cde2 | 140 | std::vector<scope_level> m_scopes; |
e776119f | 141 | |
4895cde2 PA |
142 | /* A thrown error, if any. */ |
143 | struct gdb_exception m_error; | |
108546a0 | 144 | |
4895cde2 PA |
145 | /* The line of the thrown error, or 0. */ |
146 | int m_last_line; | |
147 | ||
148 | /* The name of the expected / default DTD, if specified. */ | |
149 | const char *m_dtd_name; | |
150 | ||
151 | /* Are we the special <xi:include> parser? */ | |
152 | bool m_is_xinclude; | |
e776119f DJ |
153 | }; |
154 | ||
155 | /* Process some body text. We accumulate the text for later use; it's | |
156 | wrong to do anything with it immediately, because a single block of | |
157 | text might be broken up into multiple calls to this function. */ | |
158 | ||
4895cde2 PA |
159 | void |
160 | gdb_xml_parser::body_text (const XML_Char *text, int length) | |
161 | { | |
162 | if (m_error.reason < 0) | |
163 | return; | |
164 | ||
165 | scope_level &scope = m_scopes.back (); | |
166 | scope.body.append (text, length); | |
167 | } | |
168 | ||
e776119f DJ |
169 | static void |
170 | gdb_xml_body_text (void *data, const XML_Char *text, int length) | |
171 | { | |
19ba03f4 | 172 | struct gdb_xml_parser *parser = (struct gdb_xml_parser *) data; |
e776119f | 173 | |
4895cde2 | 174 | parser->body_text (text, length); |
e776119f | 175 | } |
fd79ecee | 176 | |
e776119f | 177 | /* Issue a debugging message from one of PARSER's handlers. */ |
fd79ecee | 178 | |
e776119f | 179 | void |
4895cde2 | 180 | gdb_xml_parser::vdebug (const char *format, va_list ap) |
e776119f | 181 | { |
4895cde2 | 182 | int line = XML_GetCurrentLineNumber (m_expat_parser); |
e776119f | 183 | |
467dc1e2 | 184 | std::string message = string_vprintf (format, ap); |
e776119f DJ |
185 | if (line) |
186 | fprintf_unfiltered (gdb_stderr, "%s (line %d): %s\n", | |
467dc1e2 | 187 | m_name, line, message.c_str ()); |
e776119f DJ |
188 | else |
189 | fprintf_unfiltered (gdb_stderr, "%s: %s\n", | |
467dc1e2 | 190 | m_name, message.c_str ()); |
e776119f DJ |
191 | } |
192 | ||
4895cde2 PA |
193 | void |
194 | gdb_xml_debug (struct gdb_xml_parser *parser, const char *format, ...) | |
195 | { | |
196 | if (!debug_xml) | |
197 | return; | |
198 | ||
199 | va_list ap; | |
200 | va_start (ap, format); | |
201 | parser->vdebug (format, ap); | |
202 | va_end (ap); | |
203 | } | |
204 | ||
e776119f DJ |
205 | /* Issue an error message from one of PARSER's handlers, and stop |
206 | parsing. */ | |
207 | ||
4895cde2 PA |
208 | void |
209 | gdb_xml_parser::verror (const char *format, va_list ap) | |
210 | { | |
211 | int line = XML_GetCurrentLineNumber (m_expat_parser); | |
212 | ||
213 | m_last_line = line; | |
214 | throw_verror (XML_PARSE_ERROR, format, ap); | |
215 | } | |
216 | ||
e776119f DJ |
217 | void |
218 | gdb_xml_error (struct gdb_xml_parser *parser, const char *format, ...) | |
219 | { | |
e776119f | 220 | va_list ap; |
e776119f | 221 | va_start (ap, format); |
4895cde2 PA |
222 | parser->verror (format, ap); |
223 | va_end (ap); | |
e776119f DJ |
224 | } |
225 | ||
3d2c1d41 PA |
226 | /* Find the attribute named NAME in the set of parsed attributes |
227 | ATTRIBUTES. Returns NULL if not found. */ | |
228 | ||
229 | struct gdb_xml_value * | |
4d0fdd9b SM |
230 | xml_find_attribute (std::vector<gdb_xml_value> &attributes, |
231 | const char *name) | |
3d2c1d41 | 232 | { |
4d0fdd9b SM |
233 | for (gdb_xml_value &value : attributes) |
234 | if (strcmp (value.name, name) == 0) | |
235 | return &value; | |
3d2c1d41 PA |
236 | |
237 | return NULL; | |
238 | } | |
239 | ||
4895cde2 PA |
240 | /* Handle the start of an element. NAME is the element, and ATTRS are |
241 | the names and values of this element's attributes. */ | |
e776119f | 242 | |
4895cde2 PA |
243 | void |
244 | gdb_xml_parser::start_element (const XML_Char *name, | |
245 | const XML_Char **attrs) | |
fd79ecee | 246 | { |
4895cde2 PA |
247 | if (m_error.reason < 0) |
248 | return; | |
249 | ||
e776119f DJ |
250 | const struct gdb_xml_element *element; |
251 | const struct gdb_xml_attribute *attribute; | |
e776119f | 252 | unsigned int seen; |
e776119f | 253 | |
e776119f DJ |
254 | /* Push an error scope. If we return or throw an exception before |
255 | filling this in, it will tell us to ignore children of this | |
a7fc9b61 PA |
256 | element. Note we don't take a reference to the element yet |
257 | because further below we'll process the element which may recurse | |
258 | back here and push more elements to the vector. When the | |
259 | recursion unrolls all such elements will have been popped back | |
260 | already, but if one of those pushes reallocates the vector, | |
261 | previous element references will be invalidated. */ | |
4895cde2 | 262 | m_scopes.emplace_back (); |
a7fc9b61 PA |
263 | |
264 | /* Get a reference to the current scope. */ | |
4895cde2 | 265 | scope_level &scope = m_scopes[m_scopes.size () - 2]; |
e776119f | 266 | |
4895cde2 | 267 | gdb_xml_debug (this, _("Entering element <%s>"), name); |
e776119f DJ |
268 | |
269 | /* Find this element in the list of the current scope's allowed | |
270 | children. Record that we've seen it. */ | |
271 | ||
272 | seen = 1; | |
a7fc9b61 | 273 | for (element = scope.elements; element && element->name; |
e776119f DJ |
274 | element++, seen <<= 1) |
275 | if (strcmp (element->name, name) == 0) | |
276 | break; | |
277 | ||
278 | if (element == NULL || element->name == NULL) | |
279 | { | |
108546a0 DJ |
280 | /* If we're working on XInclude, <xi:include> can be the child |
281 | of absolutely anything. Copy the previous scope's element | |
282 | list into the new scope even if there was no match. */ | |
4895cde2 | 283 | if (m_is_xinclude) |
108546a0 | 284 | { |
4895cde2 | 285 | XML_DefaultCurrent (m_expat_parser); |
108546a0 | 286 | |
4895cde2 | 287 | scope_level &unknown_scope = m_scopes.back (); |
a7fc9b61 | 288 | unknown_scope.elements = scope.elements; |
108546a0 DJ |
289 | return; |
290 | } | |
291 | ||
4895cde2 | 292 | gdb_xml_debug (this, _("Element <%s> unknown"), name); |
e776119f DJ |
293 | return; |
294 | } | |
295 | ||
a7fc9b61 | 296 | if (!(element->flags & GDB_XML_EF_REPEATABLE) && (seen & scope.seen)) |
4895cde2 | 297 | gdb_xml_error (this, _("Element <%s> only expected once"), name); |
e776119f | 298 | |
a7fc9b61 | 299 | scope.seen |= seen; |
e776119f | 300 | |
4d0fdd9b | 301 | std::vector<gdb_xml_value> attributes; |
108546a0 | 302 | |
e776119f DJ |
303 | for (attribute = element->attributes; |
304 | attribute != NULL && attribute->name != NULL; | |
305 | attribute++) | |
306 | { | |
307 | const char *val = NULL; | |
308 | const XML_Char **p; | |
309 | void *parsed_value; | |
e776119f DJ |
310 | |
311 | for (p = attrs; *p != NULL; p += 2) | |
312 | if (!strcmp (attribute->name, p[0])) | |
313 | { | |
314 | val = p[1]; | |
315 | break; | |
316 | } | |
317 | ||
318 | if (*p != NULL && val == NULL) | |
319 | { | |
4895cde2 | 320 | gdb_xml_debug (this, _("Attribute \"%s\" missing a value"), |
e776119f DJ |
321 | attribute->name); |
322 | continue; | |
323 | } | |
324 | ||
325 | if (*p == NULL && !(attribute->flags & GDB_XML_AF_OPTIONAL)) | |
326 | { | |
4895cde2 | 327 | gdb_xml_error (this, _("Required attribute \"%s\" of " |
e776119f DJ |
328 | "<%s> not specified"), |
329 | attribute->name, element->name); | |
330 | continue; | |
331 | } | |
332 | ||
333 | if (*p == NULL) | |
334 | continue; | |
335 | ||
4895cde2 | 336 | gdb_xml_debug (this, _("Parsing attribute %s=\"%s\""), |
e776119f DJ |
337 | attribute->name, val); |
338 | ||
339 | if (attribute->handler) | |
4895cde2 | 340 | parsed_value = attribute->handler (this, attribute, val); |
e776119f DJ |
341 | else |
342 | parsed_value = xstrdup (val); | |
343 | ||
4d0fdd9b | 344 | attributes.emplace_back (attribute->name, parsed_value); |
e776119f DJ |
345 | } |
346 | ||
347 | /* Check for unrecognized attributes. */ | |
348 | if (debug_xml) | |
fd79ecee | 349 | { |
e776119f DJ |
350 | const XML_Char **p; |
351 | ||
352 | for (p = attrs; *p != NULL; p += 2) | |
353 | { | |
354 | for (attribute = element->attributes; | |
355 | attribute != NULL && attribute->name != NULL; | |
356 | attribute++) | |
357 | if (strcmp (attribute->name, *p) == 0) | |
358 | break; | |
359 | ||
360 | if (attribute == NULL || attribute->name == NULL) | |
4895cde2 | 361 | gdb_xml_debug (this, _("Ignoring unknown attribute %s"), *p); |
e776119f DJ |
362 | } |
363 | } | |
364 | ||
365 | /* Call the element handler if there is one. */ | |
366 | if (element->start_handler) | |
4895cde2 | 367 | element->start_handler (this, element, m_user_data, attributes); |
e776119f | 368 | |
a7fc9b61 PA |
369 | /* Fill in a new scope level. Note that we must delay getting a |
370 | back reference till here because above we might have recursed, | |
371 | which may have reallocated the vector which invalidates | |
372 | iterators/pointers/references. */ | |
4895cde2 | 373 | scope_level &new_scope = m_scopes.back (); |
a7fc9b61 PA |
374 | new_scope.element = element; |
375 | new_scope.elements = element->children; | |
e776119f DJ |
376 | } |
377 | ||
378 | /* Wrapper for gdb_xml_start_element, to prevent throwing exceptions | |
379 | through expat. */ | |
380 | ||
381 | static void | |
382 | gdb_xml_start_element_wrapper (void *data, const XML_Char *name, | |
383 | const XML_Char **attrs) | |
384 | { | |
19ba03f4 | 385 | struct gdb_xml_parser *parser = (struct gdb_xml_parser *) data; |
e776119f | 386 | |
a70b8144 | 387 | try |
e776119f | 388 | { |
4895cde2 | 389 | parser->start_element (name, attrs); |
e776119f | 390 | } |
94aeb44b | 391 | catch (gdb_exception &ex) |
e776119f | 392 | { |
94aeb44b | 393 | parser->set_error (std::move (ex)); |
fd79ecee | 394 | } |
e776119f DJ |
395 | } |
396 | ||
4895cde2 | 397 | /* Handle the end of an element. NAME is the current element. */ |
e776119f | 398 | |
4895cde2 PA |
399 | void |
400 | gdb_xml_parser::end_element (const XML_Char *name) | |
e776119f | 401 | { |
4895cde2 PA |
402 | if (m_error.reason < 0) |
403 | return; | |
404 | ||
405 | struct scope_level *scope = &m_scopes.back (); | |
e776119f DJ |
406 | const struct gdb_xml_element *element; |
407 | unsigned int seen; | |
e776119f | 408 | |
4895cde2 | 409 | gdb_xml_debug (this, _("Leaving element <%s>"), name); |
e776119f DJ |
410 | |
411 | for (element = scope->elements, seen = 1; | |
412 | element != NULL && element->name != NULL; | |
413 | element++, seen <<= 1) | |
414 | if ((scope->seen & seen) == 0 | |
415 | && (element->flags & GDB_XML_EF_OPTIONAL) == 0) | |
4895cde2 | 416 | gdb_xml_error (this, _("Required element <%s> is missing"), |
e776119f DJ |
417 | element->name); |
418 | ||
581e13c1 | 419 | /* Call the element processor. */ |
108546a0 | 420 | if (scope->element != NULL && scope->element->end_handler) |
e776119f | 421 | { |
bd8a901f | 422 | const char *body; |
e776119f | 423 | |
bd8a901f PA |
424 | if (scope->body.empty ()) |
425 | body = ""; | |
108546a0 DJ |
426 | else |
427 | { | |
428 | int length; | |
e776119f | 429 | |
bd8a901f PA |
430 | length = scope->body.size (); |
431 | body = scope->body.c_str (); | |
e776119f | 432 | |
108546a0 | 433 | /* Strip leading and trailing whitespace. */ |
bd8a901f PA |
434 | while (length > 0 && ISSPACE (body[length - 1])) |
435 | length--; | |
436 | scope->body.erase (length); | |
108546a0 DJ |
437 | while (*body && ISSPACE (*body)) |
438 | body++; | |
439 | } | |
440 | ||
4895cde2 PA |
441 | scope->element->end_handler (this, scope->element, |
442 | m_user_data, body); | |
108546a0 DJ |
443 | } |
444 | else if (scope->element == NULL) | |
4895cde2 | 445 | XML_DefaultCurrent (m_expat_parser); |
e776119f DJ |
446 | |
447 | /* Pop the scope level. */ | |
4895cde2 | 448 | m_scopes.pop_back (); |
e776119f DJ |
449 | } |
450 | ||
451 | /* Wrapper for gdb_xml_end_element, to prevent throwing exceptions | |
452 | through expat. */ | |
453 | ||
454 | static void | |
455 | gdb_xml_end_element_wrapper (void *data, const XML_Char *name) | |
456 | { | |
19ba03f4 | 457 | struct gdb_xml_parser *parser = (struct gdb_xml_parser *) data; |
e776119f | 458 | |
a70b8144 | 459 | try |
e776119f | 460 | { |
4895cde2 | 461 | parser->end_element (name); |
e776119f | 462 | } |
94aeb44b | 463 | catch (gdb_exception &ex) |
e776119f | 464 | { |
94aeb44b | 465 | parser->set_error (std::move (ex)); |
e776119f DJ |
466 | } |
467 | } | |
468 | ||
469 | /* Free a parser and all its associated state. */ | |
470 | ||
010151c9 | 471 | gdb_xml_parser::~gdb_xml_parser () |
e776119f | 472 | { |
4895cde2 | 473 | XML_ParserFree (m_expat_parser); |
e776119f DJ |
474 | } |
475 | ||
010151c9 PA |
476 | /* Initialize a parser. */ |
477 | ||
4895cde2 | 478 | gdb_xml_parser::gdb_xml_parser (const char *name, |
010151c9 | 479 | const gdb_xml_element *elements, |
4895cde2 PA |
480 | void *user_data) |
481 | : m_name (name), | |
482 | m_user_data (user_data), | |
4895cde2 PA |
483 | m_last_line (0), |
484 | m_dtd_name (NULL), | |
485 | m_is_xinclude (false) | |
486 | { | |
487 | m_expat_parser = XML_ParserCreateNS (NULL, '!'); | |
488 | if (m_expat_parser == NULL) | |
010151c9 | 489 | malloc_failure (0); |
e776119f | 490 | |
4895cde2 | 491 | XML_SetUserData (m_expat_parser, this); |
e776119f DJ |
492 | |
493 | /* Set the callbacks. */ | |
4895cde2 | 494 | XML_SetElementHandler (m_expat_parser, gdb_xml_start_element_wrapper, |
e776119f | 495 | gdb_xml_end_element_wrapper); |
4895cde2 | 496 | XML_SetCharacterDataHandler (m_expat_parser, gdb_xml_body_text); |
e776119f DJ |
497 | |
498 | /* Initialize the outer scope. */ | |
4895cde2 | 499 | m_scopes.emplace_back (elements); |
efc0eabd PA |
500 | } |
501 | ||
108546a0 DJ |
502 | /* External entity handler. The only external entities we support |
503 | are those compiled into GDB (we do not fetch entities from the | |
504 | target). */ | |
505 | ||
506 | static int XMLCALL | |
507 | gdb_xml_fetch_external_entity (XML_Parser expat_parser, | |
508 | const XML_Char *context, | |
509 | const XML_Char *base, | |
510 | const XML_Char *systemId, | |
511 | const XML_Char *publicId) | |
512 | { | |
108546a0 DJ |
513 | XML_Parser entity_parser; |
514 | const char *text; | |
515 | enum XML_Status status; | |
516 | ||
517 | if (systemId == NULL) | |
518 | { | |
4895cde2 PA |
519 | gdb_xml_parser *parser |
520 | = (gdb_xml_parser *) XML_GetUserData (expat_parser); | |
521 | ||
522 | text = fetch_xml_builtin (parser->dtd_name ()); | |
108546a0 | 523 | if (text == NULL) |
9b20d036 MS |
524 | internal_error (__FILE__, __LINE__, |
525 | _("could not locate built-in DTD %s"), | |
4895cde2 | 526 | parser->dtd_name ()); |
108546a0 DJ |
527 | } |
528 | else | |
529 | { | |
530 | text = fetch_xml_builtin (systemId); | |
531 | if (text == NULL) | |
532 | return XML_STATUS_ERROR; | |
533 | } | |
534 | ||
4895cde2 PA |
535 | entity_parser = XML_ExternalEntityParserCreate (expat_parser, |
536 | context, NULL); | |
108546a0 DJ |
537 | |
538 | /* Don't use our handlers for the contents of the DTD. Just let expat | |
539 | process it. */ | |
540 | XML_SetElementHandler (entity_parser, NULL, NULL); | |
541 | XML_SetDoctypeDeclHandler (entity_parser, NULL, NULL); | |
542 | XML_SetXmlDeclHandler (entity_parser, NULL); | |
543 | XML_SetDefaultHandler (entity_parser, NULL); | |
544 | XML_SetUserData (entity_parser, NULL); | |
545 | ||
546 | status = XML_Parse (entity_parser, text, strlen (text), 1); | |
547 | ||
548 | XML_ParserFree (entity_parser); | |
549 | return status; | |
550 | } | |
551 | ||
108546a0 | 552 | void |
4895cde2 | 553 | gdb_xml_parser::use_dtd (const char *dtd_name) |
108546a0 DJ |
554 | { |
555 | enum XML_Error err; | |
556 | ||
4895cde2 | 557 | m_dtd_name = dtd_name; |
108546a0 | 558 | |
4895cde2 | 559 | XML_SetParamEntityParsing (m_expat_parser, |
108546a0 | 560 | XML_PARAM_ENTITY_PARSING_UNLESS_STANDALONE); |
4895cde2 | 561 | XML_SetExternalEntityRefHandler (m_expat_parser, |
108546a0 DJ |
562 | gdb_xml_fetch_external_entity); |
563 | ||
564 | /* Even if no DTD is provided, use the built-in DTD anyway. */ | |
4895cde2 | 565 | err = XML_UseForeignDTD (m_expat_parser, XML_TRUE); |
108546a0 DJ |
566 | if (err != XML_ERROR_NONE) |
567 | internal_error (__FILE__, __LINE__, | |
9b20d036 MS |
568 | _("XML_UseForeignDTD failed: %s"), |
569 | XML_ErrorString (err)); | |
108546a0 DJ |
570 | } |
571 | ||
e776119f DJ |
572 | /* Invoke PARSER on BUFFER. BUFFER is the data to parse, which |
573 | should be NUL-terminated. | |
574 | ||
575 | The return value is 0 for success or -1 for error. It may throw, | |
576 | but only if something unexpected goes wrong during parsing; parse | |
577 | errors will be caught, warned about, and reported as failure. */ | |
578 | ||
579 | int | |
4895cde2 | 580 | gdb_xml_parser::parse (const char *buffer) |
e776119f DJ |
581 | { |
582 | enum XML_Status status; | |
583 | const char *error_string; | |
584 | ||
4895cde2 | 585 | gdb_xml_debug (this, _("Starting:\n%s"), buffer); |
de584861 | 586 | |
4895cde2 | 587 | status = XML_Parse (m_expat_parser, buffer, strlen (buffer), 1); |
e776119f | 588 | |
4895cde2 | 589 | if (status == XML_STATUS_OK && m_error.reason == 0) |
e776119f DJ |
590 | return 0; |
591 | ||
4895cde2 PA |
592 | if (m_error.reason == RETURN_ERROR |
593 | && m_error.error == XML_PARSE_ERROR) | |
e776119f | 594 | { |
4895cde2 | 595 | gdb_assert (m_error.message != NULL); |
3d6e9d23 | 596 | error_string = m_error.what (); |
e776119f DJ |
597 | } |
598 | else if (status == XML_STATUS_ERROR) | |
599 | { | |
4895cde2 | 600 | enum XML_Error err = XML_GetErrorCode (m_expat_parser); |
a109c7c1 | 601 | |
e776119f DJ |
602 | error_string = XML_ErrorString (err); |
603 | } | |
604 | else | |
605 | { | |
4895cde2 | 606 | gdb_assert (m_error.reason < 0); |
94aeb44b | 607 | throw_exception (std::move (m_error)); |
e776119f DJ |
608 | } |
609 | ||
4895cde2 PA |
610 | if (m_last_line != 0) |
611 | warning (_("while parsing %s (at line %d): %s"), m_name, | |
612 | m_last_line, error_string); | |
e776119f | 613 | else |
4895cde2 | 614 | warning (_("while parsing %s: %s"), m_name, error_string); |
e776119f DJ |
615 | |
616 | return -1; | |
fd79ecee DJ |
617 | } |
618 | ||
efc0eabd PA |
619 | int |
620 | gdb_xml_parse_quick (const char *name, const char *dtd_name, | |
621 | const struct gdb_xml_element *elements, | |
622 | const char *document, void *user_data) | |
623 | { | |
010151c9 | 624 | gdb_xml_parser parser (name, elements, user_data); |
efc0eabd | 625 | if (dtd_name != NULL) |
4895cde2 PA |
626 | parser.use_dtd (dtd_name); |
627 | return parser.parse (document); | |
efc0eabd PA |
628 | } |
629 | ||
fd79ecee DJ |
630 | /* Parse a field VALSTR that we expect to contain an integer value. |
631 | The integer is returned in *VALP. The string is parsed with an | |
632 | equivalent to strtoul. | |
633 | ||
634 | Returns 0 for success, -1 for error. */ | |
635 | ||
636 | static int | |
637 | xml_parse_unsigned_integer (const char *valstr, ULONGEST *valp) | |
638 | { | |
639 | const char *endptr; | |
640 | ULONGEST result; | |
641 | ||
642 | if (*valstr == '\0') | |
643 | return -1; | |
644 | ||
645 | result = strtoulst (valstr, &endptr, 0); | |
646 | if (*endptr != '\0') | |
647 | return -1; | |
648 | ||
649 | *valp = result; | |
650 | return 0; | |
651 | } | |
652 | ||
e776119f DJ |
653 | /* Parse an integer string into a ULONGEST and return it, or call |
654 | gdb_xml_error if it could not be parsed. */ | |
fd79ecee DJ |
655 | |
656 | ULONGEST | |
e776119f | 657 | gdb_xml_parse_ulongest (struct gdb_xml_parser *parser, const char *value) |
fd79ecee DJ |
658 | { |
659 | ULONGEST result; | |
fd79ecee DJ |
660 | |
661 | if (xml_parse_unsigned_integer (value, &result) != 0) | |
e776119f DJ |
662 | gdb_xml_error (parser, _("Can't convert \"%s\" to an integer"), value); |
663 | ||
fd79ecee DJ |
664 | return result; |
665 | } | |
666 | ||
e776119f DJ |
667 | /* Parse an integer attribute into a ULONGEST. */ |
668 | ||
669 | void * | |
670 | gdb_xml_parse_attr_ulongest (struct gdb_xml_parser *parser, | |
671 | const struct gdb_xml_attribute *attribute, | |
672 | const char *value) | |
673 | { | |
674 | ULONGEST result; | |
675 | void *ret; | |
fd79ecee | 676 | |
e776119f DJ |
677 | if (xml_parse_unsigned_integer (value, &result) != 0) |
678 | gdb_xml_error (parser, _("Can't convert %s=\"%s\" to an integer"), | |
679 | attribute->name, value); | |
fd79ecee | 680 | |
8d749320 | 681 | ret = XNEW (ULONGEST); |
e776119f DJ |
682 | memcpy (ret, &result, sizeof (result)); |
683 | return ret; | |
684 | } | |
685 | ||
123dc839 DJ |
686 | /* A handler_data for yes/no boolean values. */ |
687 | ||
688 | const struct gdb_xml_enum gdb_xml_enums_boolean[] = { | |
689 | { "yes", 1 }, | |
690 | { "no", 0 }, | |
691 | { NULL, 0 } | |
692 | }; | |
693 | ||
e776119f DJ |
694 | /* Map NAME to VALUE. A struct gdb_xml_enum * should be saved as the |
695 | value of handler_data when using gdb_xml_parse_attr_enum to parse a | |
696 | fixed list of possible strings. The list is terminated by an entry | |
697 | with NAME == NULL. */ | |
698 | ||
699 | void * | |
700 | gdb_xml_parse_attr_enum (struct gdb_xml_parser *parser, | |
701 | const struct gdb_xml_attribute *attribute, | |
702 | const char *value) | |
fd79ecee | 703 | { |
19ba03f4 SM |
704 | const struct gdb_xml_enum *enums |
705 | = (const struct gdb_xml_enum *) attribute->handler_data; | |
e776119f | 706 | void *ret; |
fd79ecee | 707 | |
19ba03f4 SM |
708 | for (enums = (const struct gdb_xml_enum *) attribute->handler_data; |
709 | enums->name != NULL; enums++) | |
123dc839 | 710 | if (strcasecmp (enums->name, value) == 0) |
e776119f DJ |
711 | break; |
712 | ||
713 | if (enums->name == NULL) | |
714 | gdb_xml_error (parser, _("Unknown attribute value %s=\"%s\""), | |
715 | attribute->name, value); | |
716 | ||
717 | ret = xmalloc (sizeof (enums->value)); | |
718 | memcpy (ret, &enums->value, sizeof (enums->value)); | |
719 | return ret; | |
fd79ecee | 720 | } |
108546a0 DJ |
721 | \f |
722 | ||
723 | /* XInclude processing. This is done as a separate step from actually | |
724 | parsing the document, so that we can produce a single combined XML | |
725 | document - e.g. to hand to a front end or to simplify comparing two | |
726 | documents. We make extensive use of XML_DefaultCurrent, to pass | |
727 | input text directly into the output without reformatting or | |
728 | requoting it. | |
729 | ||
730 | We output the DOCTYPE declaration for the first document unchanged, | |
731 | if present, and discard DOCTYPEs from included documents. Only the | |
732 | one we pass through here is used when we feed the result back to | |
733 | expat. The XInclude standard explicitly does not discuss | |
734 | validation of the result; we choose to apply the same DTD applied | |
735 | to the outermost document. | |
736 | ||
737 | We can not simply include the external DTD subset in the document | |
738 | as an internal subset, because <!IGNORE> and <!INCLUDE> are valid | |
739 | only in external subsets. But if we do not pass the DTD into the | |
740 | output at all, default values will not be filled in. | |
741 | ||
742 | We don't pass through any <?xml> declaration because we generate | |
743 | UTF-8, not whatever the input encoding was. */ | |
744 | ||
745 | struct xinclude_parsing_data | |
746 | { | |
bd8a901f PA |
747 | xinclude_parsing_data (std::string &output_, |
748 | xml_fetch_another fetcher_, void *fetcher_baton_, | |
010151c9 | 749 | int include_depth_) |
bd8a901f PA |
750 | : output (output_), |
751 | skip_depth (0), | |
010151c9 PA |
752 | include_depth (include_depth_), |
753 | fetcher (fetcher_), | |
754 | fetcher_baton (fetcher_baton_) | |
bd8a901f | 755 | {} |
010151c9 | 756 | |
bd8a901f PA |
757 | /* Where the output goes. */ |
758 | std::string &output; | |
108546a0 DJ |
759 | |
760 | /* A count indicating whether we are in an element whose | |
761 | children should not be copied to the output, and if so, | |
762 | how deep we are nested. This is used for anything inside | |
763 | an xi:include, and for the DTD. */ | |
764 | int skip_depth; | |
765 | ||
766 | /* The number of <xi:include> elements currently being processed, | |
767 | to detect loops. */ | |
768 | int include_depth; | |
769 | ||
770 | /* A function to call to obtain additional features, and its | |
771 | baton. */ | |
772 | xml_fetch_another fetcher; | |
773 | void *fetcher_baton; | |
774 | }; | |
775 | ||
776 | static void | |
777 | xinclude_start_include (struct gdb_xml_parser *parser, | |
778 | const struct gdb_xml_element *element, | |
4d0fdd9b SM |
779 | void *user_data, |
780 | std::vector<gdb_xml_value> &attributes) | |
108546a0 | 781 | { |
19ba03f4 SM |
782 | struct xinclude_parsing_data *data |
783 | = (struct xinclude_parsing_data *) user_data; | |
4d0fdd9b | 784 | char *href = (char *) xml_find_attribute (attributes, "href")->value.get (); |
108546a0 DJ |
785 | |
786 | gdb_xml_debug (parser, _("Processing XInclude of \"%s\""), href); | |
787 | ||
788 | if (data->include_depth > MAX_XINCLUDE_DEPTH) | |
789 | gdb_xml_error (parser, _("Maximum XInclude depth (%d) exceeded"), | |
790 | MAX_XINCLUDE_DEPTH); | |
791 | ||
9018be22 SM |
792 | gdb::optional<gdb::char_vector> text |
793 | = data->fetcher (href, data->fetcher_baton); | |
794 | if (!text) | |
108546a0 | 795 | gdb_xml_error (parser, _("Could not load XML document \"%s\""), href); |
108546a0 | 796 | |
4895cde2 | 797 | if (!xml_process_xincludes (data->output, parser->name (), |
9018be22 | 798 | text->data (), data->fetcher, |
bd8a901f PA |
799 | data->fetcher_baton, |
800 | data->include_depth + 1)) | |
108546a0 DJ |
801 | gdb_xml_error (parser, _("Parsing \"%s\" failed"), href); |
802 | ||
108546a0 DJ |
803 | data->skip_depth++; |
804 | } | |
805 | ||
806 | static void | |
807 | xinclude_end_include (struct gdb_xml_parser *parser, | |
808 | const struct gdb_xml_element *element, | |
809 | void *user_data, const char *body_text) | |
810 | { | |
19ba03f4 SM |
811 | struct xinclude_parsing_data *data |
812 | = (struct xinclude_parsing_data *) user_data; | |
108546a0 DJ |
813 | |
814 | data->skip_depth--; | |
815 | } | |
816 | ||
817 | static void XMLCALL | |
818 | xml_xinclude_default (void *data_, const XML_Char *s, int len) | |
819 | { | |
19ba03f4 | 820 | struct gdb_xml_parser *parser = (struct gdb_xml_parser *) data_; |
4895cde2 | 821 | xinclude_parsing_data *data = (xinclude_parsing_data *) parser->user_data (); |
108546a0 DJ |
822 | |
823 | /* If we are inside of e.g. xi:include or the DTD, don't save this | |
824 | string. */ | |
825 | if (data->skip_depth) | |
826 | return; | |
827 | ||
828 | /* Otherwise just add it to the end of the document we're building | |
829 | up. */ | |
bd8a901f | 830 | data->output.append (s, len); |
108546a0 DJ |
831 | } |
832 | ||
833 | static void XMLCALL | |
834 | xml_xinclude_start_doctype (void *data_, const XML_Char *doctypeName, | |
835 | const XML_Char *sysid, const XML_Char *pubid, | |
836 | int has_internal_subset) | |
837 | { | |
19ba03f4 | 838 | struct gdb_xml_parser *parser = (struct gdb_xml_parser *) data_; |
4895cde2 | 839 | xinclude_parsing_data *data = (xinclude_parsing_data *) parser->user_data (); |
108546a0 DJ |
840 | |
841 | /* Don't print out the doctype, or the contents of the DTD internal | |
842 | subset, if any. */ | |
843 | data->skip_depth++; | |
844 | } | |
845 | ||
846 | static void XMLCALL | |
847 | xml_xinclude_end_doctype (void *data_) | |
848 | { | |
19ba03f4 | 849 | struct gdb_xml_parser *parser = (struct gdb_xml_parser *) data_; |
4895cde2 | 850 | xinclude_parsing_data *data = (xinclude_parsing_data *) parser->user_data (); |
108546a0 DJ |
851 | |
852 | data->skip_depth--; | |
853 | } | |
854 | ||
855 | static void XMLCALL | |
856 | xml_xinclude_xml_decl (void *data_, const XML_Char *version, | |
857 | const XML_Char *encoding, int standalone) | |
858 | { | |
859 | /* Do nothing - this function prevents the default handler from | |
860 | being called, thus suppressing the XML declaration from the | |
861 | output. */ | |
862 | } | |
863 | ||
108546a0 DJ |
864 | const struct gdb_xml_attribute xinclude_attributes[] = { |
865 | { "href", GDB_XML_AF_NONE, NULL, NULL }, | |
866 | { NULL, GDB_XML_AF_NONE, NULL, NULL } | |
867 | }; | |
868 | ||
869 | const struct gdb_xml_element xinclude_elements[] = { | |
870 | { "http://www.w3.org/2001/XInclude!include", xinclude_attributes, NULL, | |
871 | GDB_XML_EF_OPTIONAL | GDB_XML_EF_REPEATABLE, | |
872 | xinclude_start_include, xinclude_end_include }, | |
873 | { NULL, NULL, NULL, GDB_XML_EF_NONE, NULL, NULL } | |
874 | }; | |
875 | ||
876 | /* The main entry point for <xi:include> processing. */ | |
877 | ||
bd8a901f PA |
878 | bool |
879 | xml_process_xincludes (std::string &result, | |
880 | const char *name, const char *text, | |
108546a0 DJ |
881 | xml_fetch_another fetcher, void *fetcher_baton, |
882 | int depth) | |
883 | { | |
bd8a901f | 884 | xinclude_parsing_data data (result, fetcher, fetcher_baton, depth); |
108546a0 | 885 | |
010151c9 | 886 | gdb_xml_parser parser (name, xinclude_elements, &data); |
4895cde2 | 887 | parser.set_is_xinclude (true); |
108546a0 | 888 | |
4895cde2 PA |
889 | XML_SetCharacterDataHandler (parser.expat_parser (), NULL); |
890 | XML_SetDefaultHandler (parser.expat_parser (), xml_xinclude_default); | |
108546a0 DJ |
891 | |
892 | /* Always discard the XML version declarations; the only important | |
893 | thing this provides is encoding, and our result will have been | |
894 | converted to UTF-8. */ | |
4895cde2 | 895 | XML_SetXmlDeclHandler (parser.expat_parser (), xml_xinclude_xml_decl); |
108546a0 DJ |
896 | |
897 | if (depth > 0) | |
898 | /* Discard the doctype for included documents. */ | |
4895cde2 | 899 | XML_SetDoctypeDeclHandler (parser.expat_parser (), |
108546a0 DJ |
900 | xml_xinclude_start_doctype, |
901 | xml_xinclude_end_doctype); | |
902 | ||
4895cde2 | 903 | parser.use_dtd ("xinclude.dtd"); |
108546a0 | 904 | |
4895cde2 | 905 | if (parser.parse (text) == 0) |
108546a0 | 906 | { |
108546a0 | 907 | if (depth == 0) |
010151c9 | 908 | gdb_xml_debug (&parser, _("XInclude processing succeeded.")); |
bd8a901f | 909 | return true; |
108546a0 | 910 | } |
108546a0 | 911 | |
bd8a901f | 912 | return false; |
108546a0 | 913 | } |
05a4558a | 914 | #endif /* HAVE_LIBEXPAT */ |
108546a0 DJ |
915 | \f |
916 | ||
917 | /* Return an XML document which was compiled into GDB, from | |
918 | the given FILENAME, or NULL if the file was not compiled in. */ | |
919 | ||
920 | const char * | |
921 | fetch_xml_builtin (const char *filename) | |
922 | { | |
fec4e896 | 923 | const char *const (*p)[2]; |
108546a0 DJ |
924 | |
925 | for (p = xml_builtin; (*p)[0]; p++) | |
926 | if (strcmp ((*p)[0], filename) == 0) | |
927 | return (*p)[1]; | |
928 | ||
929 | return NULL; | |
930 | } | |
fd79ecee | 931 | |
05a4558a DJ |
932 | /* A to_xfer_partial helper function which reads XML files which were |
933 | compiled into GDB. The target may call this function from its own | |
934 | to_xfer_partial handler, after converting object and annex to the | |
935 | appropriate filename. */ | |
936 | ||
937 | LONGEST | |
938 | xml_builtin_xfer_partial (const char *filename, | |
939 | gdb_byte *readbuf, const gdb_byte *writebuf, | |
940 | ULONGEST offset, LONGEST len) | |
941 | { | |
942 | const char *buf; | |
943 | LONGEST len_avail; | |
944 | ||
945 | gdb_assert (readbuf != NULL && writebuf == NULL); | |
946 | gdb_assert (filename != NULL); | |
947 | ||
948 | buf = fetch_xml_builtin (filename); | |
949 | if (buf == NULL) | |
950 | return -1; | |
951 | ||
952 | len_avail = strlen (buf); | |
953 | if (offset >= len_avail) | |
954 | return 0; | |
955 | ||
956 | if (len > len_avail - offset) | |
957 | len = len_avail - offset; | |
958 | memcpy (readbuf, buf + offset, len); | |
959 | return len; | |
960 | } | |
961 | \f | |
fd79ecee DJ |
962 | |
963 | static void | |
e776119f DJ |
964 | show_debug_xml (struct ui_file *file, int from_tty, |
965 | struct cmd_list_element *c, const char *value) | |
fd79ecee | 966 | { |
e776119f | 967 | fprintf_filtered (file, _("XML debugging is %s.\n"), value); |
fd79ecee DJ |
968 | } |
969 | ||
9018be22 | 970 | gdb::optional<gdb::char_vector> |
a96d9b2e SDJ |
971 | xml_fetch_content_from_file (const char *filename, void *baton) |
972 | { | |
19ba03f4 | 973 | const char *dirname = (const char *) baton; |
d419f42d | 974 | gdb_file_up file; |
a96d9b2e SDJ |
975 | |
976 | if (dirname && *dirname) | |
977 | { | |
978 | char *fullname = concat (dirname, "/", filename, (char *) NULL); | |
a109c7c1 | 979 | |
f69656d0 | 980 | file = gdb_fopen_cloexec (fullname, FOPEN_RB); |
a96d9b2e SDJ |
981 | xfree (fullname); |
982 | } | |
983 | else | |
f69656d0 | 984 | file = gdb_fopen_cloexec (filename, FOPEN_RB); |
a96d9b2e SDJ |
985 | |
986 | if (file == NULL) | |
9018be22 | 987 | return {}; |
a96d9b2e | 988 | |
2edf834e | 989 | /* Read in the whole file. */ |
a96d9b2e | 990 | |
2edf834e | 991 | size_t len; |
a96d9b2e | 992 | |
2edf834e PA |
993 | if (fseek (file.get (), 0, SEEK_END) == -1) |
994 | perror_with_name (_("seek to end of file")); | |
995 | len = ftell (file.get ()); | |
996 | rewind (file.get ()); | |
a96d9b2e | 997 | |
9018be22 | 998 | gdb::char_vector text (len + 1); |
a96d9b2e | 999 | |
9018be22 | 1000 | if (fread (text.data (), 1, len, file.get ()) != len |
63929e84 | 1001 | || ferror (file.get ())) |
2edf834e PA |
1002 | { |
1003 | warning (_("Read error from \"%s\""), filename); | |
9018be22 | 1004 | return {}; |
a96d9b2e SDJ |
1005 | } |
1006 | ||
9018be22 | 1007 | text.back () = '\0'; |
a96d9b2e SDJ |
1008 | return text; |
1009 | } | |
1010 | ||
6c265988 SM |
1011 | void _initialize_xml_support (); |
1012 | void _initialize_xml_support (); | |
fd79ecee | 1013 | void |
6c265988 | 1014 | _initialize_xml_support () |
fd79ecee | 1015 | { |
e776119f DJ |
1016 | add_setshow_boolean_cmd ("xml", class_maintenance, &debug_xml, |
1017 | _("Set XML parser debugging."), | |
1018 | _("Show XML parser debugging."), | |
1019 | _("When set, debugging messages for XML parsers " | |
1020 | "are displayed."), | |
1021 | NULL, show_debug_xml, | |
1022 | &setdebuglist, &showdebuglist); | |
fd79ecee | 1023 | } |