?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
qasf/tests: Fix a couple of spelling errors in ok() messages.
[wine/zf.git] / dlls / xmllite / reader.c
blobeddc4d8eec0a86e7cba947878d2aedbcbfe746bf
1 /*
2 * IXmlReader implementation
3 *
4 * Copyright 2010, 2012-2013, 2016-2017 Nikolay Sivov
5 *
6 * This library is free software; you can redistribute it and/or
7 * modify it under the terms of the GNU Lesser General Public
8 * License as published by the Free Software Foundation; either
9 * version 2.1 of the License, or (at your option) any later version.
10 *
11 * This library 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.
15 *
16 * You should have received a copy of the GNU Lesser General Public
17 * License along with this library; if not, write to the Free Software
18 * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301, USA
19 */
20
21 #define COBJMACROS
22
23 #include <stdio.h>
24 #include <stdarg.h>
25 #include <assert.h>
26 #include "windef.h"
27 #include "winbase.h"
28 #include "initguid.h"
29 #include "objbase.h"
30 #include "xmllite.h"
31 #include "xmllite_private.h"
32
33 #include "wine/debug.h"
34 #include "wine/list.h"
35
36 WINE_DEFAULT_DEBUG_CHANNEL(xmllite);
37
38 /* not defined in public headers */
39 DEFINE_GUID(IID_IXmlReaderInput, 0x0b3ccc9b, 0x9214, 0x428b, 0xa2, 0xae, 0xef, 0x3a, 0xa8, 0x71, 0xaf, 0xda);
40
41 typedef enum
42 {
43 XmlReadInState_Initial,
44 XmlReadInState_XmlDecl,
45 XmlReadInState_Misc_DTD,
46 XmlReadInState_DTD,
47 XmlReadInState_DTD_Misc,
48 XmlReadInState_Element,
49 XmlReadInState_Content,
50 XmlReadInState_MiscEnd, /* optional Misc at the end of a document */
51 XmlReadInState_Eof
52 } XmlReaderInternalState;
53
54 /* This state denotes where parsing was interrupted by input problem.
55 Reader resumes parsing using this information. */
56 typedef enum
57 {
58 XmlReadResumeState_Initial,
59 XmlReadResumeState_PITarget,
60 XmlReadResumeState_PIBody,
61 XmlReadResumeState_CDATA,
62 XmlReadResumeState_Comment,
63 XmlReadResumeState_STag,
64 XmlReadResumeState_CharData,
65 XmlReadResumeState_Whitespace
66 } XmlReaderResumeState;
67
68 /* saved pointer index to resume from particular input position */
69 typedef enum
70 {
71 XmlReadResume_Name, /* PITarget, name for NCName, prefix for QName */
72 XmlReadResume_Local, /* local for QName */
73 XmlReadResume_Body, /* PI body, comment text, CDATA text, CharData text */
74 XmlReadResume_Last
75 } XmlReaderResume;
76
77 typedef enum
78 {
79 StringValue_LocalName,
80 StringValue_Prefix,
81 StringValue_QualifiedName,
82 StringValue_Value,
83 StringValue_Last
84 } XmlReaderStringValue;
85
86 static const WCHAR usasciiW[] = {'U','S','-','A','S','C','I','I',0};
87 static const WCHAR utf16W[] = {'U','T','F','-','1','6',0};
88 static const WCHAR utf8W[] = {'U','T','F','-','8',0};
89
90 static const WCHAR dblquoteW[] = {'\"',0};
91 static const WCHAR quoteW[] = {'\'',0};
92 static const WCHAR ltW[] = {'<',0};
93 static const WCHAR gtW[] = {'>',0};
94 static const WCHAR commentW[] = {'<','!','-','-',0};
95 static const WCHAR piW[] = {'<','?',0};
96
97 BOOL is_namestartchar(WCHAR ch);
98
99 static const char *debugstr_nodetype(XmlNodeType nodetype)
100 {
101 static const char * const type_names[] =
102 {
103 "None",
104 "Element",
105 "Attribute",
106 "Text",
107 "CDATA",
108 "",
109 "",
110 "ProcessingInstruction",
111 "Comment",
112 "",
113 "DocumentType",
114 "",
115 "",
116 "Whitespace",
117 "",
118 "EndElement",
119 "",
120 "XmlDeclaration"
121 };
122
123 if (nodetype > _XmlNodeType_Last)
124 return wine_dbg_sprintf("unknown type=%d", nodetype);
125
126 return type_names[nodetype];
127 }
128
129 static const char *debugstr_reader_prop(XmlReaderProperty prop)
130 {
131 static const char * const prop_names[] =
132 {
133 "MultiLanguage",
134 "ConformanceLevel",
135 "RandomAccess",
136 "XmlResolver",
137 "DtdProcessing",
138 "ReadState",
139 "MaxElementDepth",
140 "MaxEntityExpansion"
141 };
142
143 if (prop > _XmlReaderProperty_Last)
144 return wine_dbg_sprintf("unknown property=%d", prop);
145
146 return prop_names[prop];
147 }
148
149 struct xml_encoding_data
150 {
151 const WCHAR *name;
152 xml_encoding enc;
153 UINT cp;
154 };
155
156 static const struct xml_encoding_data xml_encoding_map[] = {
157 { usasciiW, XmlEncoding_USASCII, 20127 },
158 { utf16W, XmlEncoding_UTF16, 1200 },
159 { utf8W, XmlEncoding_UTF8, CP_UTF8 },
160 };
161
162 const WCHAR *get_encoding_name(xml_encoding encoding)
163 {
164 return xml_encoding_map[encoding].name;
165 }
166
167 xml_encoding get_encoding_from_codepage(UINT codepage)
168 {
169 int i;
170 for (i = 0; i < ARRAY_SIZE(xml_encoding_map); i++)
171 {
172 if (xml_encoding_map[i].cp == codepage) return xml_encoding_map[i].enc;
173 }
174 return XmlEncoding_Unknown;
175 }
176
177 typedef struct
178 {
179 char *data;
180 UINT cur;
181 unsigned int allocated;
182 unsigned int written;
183 BOOL prev_cr;
184 } encoded_buffer;
185
186 typedef struct input_buffer input_buffer;
187
188 typedef struct
189 {
190 IXmlReaderInput IXmlReaderInput_iface;
191 LONG ref;
192 /* reference passed on IXmlReaderInput creation, is kept when input is created */
193 IUnknown *input;
194 IMalloc *imalloc;
195 xml_encoding encoding;
196 BOOL hint;
197 WCHAR *baseuri;
198 /* stream reference set after SetInput() call from reader,
199 stored as sequential stream, cause currently
200 optimizations possible with IStream aren't implemented */
201 ISequentialStream *stream;
202 input_buffer *buffer;
203 unsigned int pending : 1;
204 } xmlreaderinput;
205
206 static const struct IUnknownVtbl xmlreaderinputvtbl;
207
208 /* Structure to hold parsed string of specific length.
209
210 Reader stores node value as 'start' pointer, on request
211 a null-terminated version of it is allocated.
212
213 To init a strval variable use reader_init_strval(),
214 to set strval as a reader value use reader_set_strval().
215 */
216 typedef struct
217 {
218 WCHAR *str; /* allocated null-terminated string */
219 UINT len; /* length in WCHARs, altered after ReadValueChunk */
220 UINT start; /* input position where value starts */
221 } strval;
222
223 static WCHAR emptyW[] = {0};
224 static WCHAR xmlW[] = {'x','m','l',0};
225 static WCHAR xmlnsW[] = {'x','m','l','n','s',0};
226 static const strval strval_empty = { emptyW };
227 static const strval strval_xml = { xmlW, 3 };
228 static const strval strval_xmlns = { xmlnsW, 5 };
229
230 struct reader_position
231 {
232 UINT line_number;
233 UINT line_position;
234 };
235
236 enum attribute_flags
237 {
238 ATTRIBUTE_NS_DEFINITION = 0x1,
239 ATTRIBUTE_DEFAULT_NS_DEFINITION = 0x2,
240 };
241
242 struct attribute
243 {
244 struct list entry;
245 strval prefix;
246 strval localname;
247 strval qname;
248 strval value;
249 struct reader_position position;
250 unsigned int flags;
251 };
252
253 struct element
254 {
255 struct list entry;
256 strval prefix;
257 strval localname;
258 strval qname;
259 struct reader_position position;
260 };
261
262 struct ns
263 {
264 struct list entry;
265 strval prefix;
266 strval uri;
267 struct element *element;
268 };
269
270 typedef struct
271 {
272 IXmlReader IXmlReader_iface;
273 LONG ref;
274 xmlreaderinput *input;
275 IMalloc *imalloc;
276 XmlReadState state;
277 HRESULT error; /* error set on XmlReadState_Error */
278 XmlReaderInternalState instate;
279 XmlReaderResumeState resumestate;
280 XmlNodeType nodetype;
281 DtdProcessing dtdmode;
282 IXmlResolver *resolver;
283 IUnknown *mlang;
284 struct reader_position position;
285 struct list attrs; /* attributes list for current node */
286 struct attribute *attr; /* current attribute */
287 UINT attr_count;
288 struct list nsdef;
289 struct list ns;
290 struct list elements;
291 int chunk_read_off;
292 strval strvalues[StringValue_Last];
293 UINT depth;
294 UINT max_depth;
295 BOOL is_empty_element;
296 struct element empty_element; /* used for empty elements without end tag <a />,
297 and to keep <?xml reader position */
298 UINT resume[XmlReadResume_Last]; /* offsets used to resume reader */
299 } xmlreader;
300
301 struct input_buffer
302 {
303 encoded_buffer utf16;
304 encoded_buffer encoded;
305 UINT code_page;
306 xmlreaderinput *input;
307 };
308
309 static inline xmlreader *impl_from_IXmlReader(IXmlReader *iface)
310 {
311 return CONTAINING_RECORD(iface, xmlreader, IXmlReader_iface);
312 }
313
314 static inline xmlreaderinput *impl_from_IXmlReaderInput(IXmlReaderInput *iface)
315 {
316 return CONTAINING_RECORD(iface, xmlreaderinput, IXmlReaderInput_iface);
317 }
318
319 /* reader memory allocation functions */
320 static inline void *reader_alloc(xmlreader *reader, size_t len)
321 {
322 return m_alloc(reader->imalloc, len);
323 }
324
325 static inline void *reader_alloc_zero(xmlreader *reader, size_t len)
326 {
327 void *ret = reader_alloc(reader, len);
328 if (ret)
329 memset(ret, 0, len);
330 return ret;
331 }
332
333 static inline void reader_free(xmlreader *reader, void *mem)
334 {
335 m_free(reader->imalloc, mem);
336 }
337
338 /* Just return pointer from offset, no attempt to read more. */
339 static inline WCHAR *reader_get_ptr2(const xmlreader *reader, UINT offset)
340 {
341 encoded_buffer *buffer = &reader->input->buffer->utf16;
342 return (WCHAR*)buffer->data + offset;
343 }
344
345 static inline WCHAR *reader_get_strptr(const xmlreader *reader, const strval *v)
346 {
347 return v->str ? v->str : reader_get_ptr2(reader, v->start);
348 }
349
350 static HRESULT reader_strvaldup(xmlreader *reader, const strval *src, strval *dest)
351 {
352 *dest = *src;
353
354 if (src->str != strval_empty.str)
355 {
356 dest->str = reader_alloc(reader, (dest->len+1)*sizeof(WCHAR));
357 if (!dest->str) return E_OUTOFMEMORY;
358 memcpy(dest->str, reader_get_strptr(reader, src), dest->len*sizeof(WCHAR));
359 dest->str[dest->len] = 0;
360 dest->start = 0;
361 }
362
363 return S_OK;
364 }
365
366 /* reader input memory allocation functions */
367 static inline void *readerinput_alloc(xmlreaderinput *input, size_t len)
368 {
369 return m_alloc(input->imalloc, len);
370 }
371
372 static inline void *readerinput_realloc(xmlreaderinput *input, void *mem, size_t len)
373 {
374 return m_realloc(input->imalloc, mem, len);
375 }
376
377 static inline void readerinput_free(xmlreaderinput *input, void *mem)
378 {
379 m_free(input->imalloc, mem);
380 }
381
382 static inline WCHAR *readerinput_strdupW(xmlreaderinput *input, const WCHAR *str)
383 {
384 LPWSTR ret = NULL;
385
386 if(str) {
387 DWORD size;
388
389 size = (lstrlenW(str)+1)*sizeof(WCHAR);
390 ret = readerinput_alloc(input, size);
391 if (ret) memcpy(ret, str, size);
392 }
393
394 return ret;
395 }
396
397 /* This one frees stored string value if needed */
398 static void reader_free_strvalued(xmlreader *reader, strval *v)
399 {
400 if (v->str != strval_empty.str)
401 {
402 reader_free(reader, v->str);
403 *v = strval_empty;
404 }
405 }
406
407 static void reader_clear_attrs(xmlreader *reader)
408 {
409 struct attribute *attr, *attr2;
410 LIST_FOR_EACH_ENTRY_SAFE(attr, attr2, &reader->attrs, struct attribute, entry)
411 {
412 reader_free_strvalued(reader, &attr->localname);
413 reader_free_strvalued(reader, &attr->value);
414 reader_free(reader, attr);
415 }
416 list_init(&reader->attrs);
417 reader->attr_count = 0;
418 reader->attr = NULL;
419 }
420
421 /* attribute data holds pointers to buffer data, so buffer shrink is not possible
422 while we are on a node with attributes */
423 static HRESULT reader_add_attr(xmlreader *reader, strval *prefix, strval *localname, strval *qname,
424 strval *value, const struct reader_position *position, unsigned int flags)
425 {
426 struct attribute *attr;
427 HRESULT hr;
428
429 attr = reader_alloc(reader, sizeof(*attr));
430 if (!attr) return E_OUTOFMEMORY;
431
432 hr = reader_strvaldup(reader, localname, &attr->localname);
433 if (hr == S_OK)
434 {
435 hr = reader_strvaldup(reader, value, &attr->value);
436 if (hr != S_OK)
437 reader_free_strvalued(reader, &attr->value);
438 }
439 if (hr != S_OK)
440 {
441 reader_free(reader, attr);
442 return hr;
443 }
444
445 if (prefix)
446 attr->prefix = *prefix;
447 else
448 memset(&attr->prefix, 0, sizeof(attr->prefix));
449 attr->qname = qname ? *qname : *localname;
450 attr->position = *position;
451 attr->flags = flags;
452 list_add_tail(&reader->attrs, &attr->entry);
453 reader->attr_count++;
454
455 return S_OK;
456 }
457
458 /* Returns current element, doesn't check if reader is actually positioned on it. */
459 static struct element *reader_get_element(xmlreader *reader)
460 {
461 if (reader->is_empty_element)
462 return &reader->empty_element;
463
464 return LIST_ENTRY(list_head(&reader->elements), struct element, entry);
465 }
466
467 static inline void reader_init_strvalue(UINT start, UINT len, strval *v)
468 {
469 v->start = start;
470 v->len = len;
471 v->str = NULL;
472 }
473
474 static inline const char* debug_strval(const xmlreader *reader, const strval *v)
475 {
476 return debugstr_wn(reader_get_strptr(reader, v), v->len);
477 }
478
479 /* used to initialize from constant string */
480 static inline void reader_init_cstrvalue(WCHAR *str, UINT len, strval *v)
481 {
482 v->start = 0;
483 v->len = len;
484 v->str = str;
485 }
486
487 static void reader_free_strvalue(xmlreader *reader, XmlReaderStringValue type)
488 {
489 reader_free_strvalued(reader, &reader->strvalues[type]);
490 }
491
492 static void reader_free_strvalues(xmlreader *reader)
493 {
494 int type;
495 for (type = 0; type < StringValue_Last; type++)
496 reader_free_strvalue(reader, type);
497 }
498
499 /* This helper should only be used to test if strings are the same,
500 it doesn't try to sort. */
501 static inline int strval_eq(const xmlreader *reader, const strval *str1, const strval *str2)
502 {
503 if (str1->len != str2->len) return 0;
504 return !memcmp(reader_get_strptr(reader, str1), reader_get_strptr(reader, str2), str1->len*sizeof(WCHAR));
505 }
506
507 static void reader_clear_elements(xmlreader *reader)
508 {
509 struct element *elem, *elem2;
510 LIST_FOR_EACH_ENTRY_SAFE(elem, elem2, &reader->elements, struct element, entry)
511 {
512 reader_free_strvalued(reader, &elem->prefix);
513 reader_free_strvalued(reader, &elem->localname);
514 reader_free_strvalued(reader, &elem->qname);
515 reader_free(reader, elem);
516 }
517 list_init(&reader->elements);
518 reader_free_strvalued(reader, &reader->empty_element.localname);
519 reader_free_strvalued(reader, &reader->empty_element.qname);
520 reader->is_empty_element = FALSE;
521 }
522
523 static struct ns *reader_lookup_ns(xmlreader *reader, const strval *prefix)
524 {
525 struct list *nslist = prefix ? &reader->ns : &reader->nsdef;
526 struct ns *ns;
527
528 LIST_FOR_EACH_ENTRY_REV(ns, nslist, struct ns, entry) {
529 if (strval_eq(reader, prefix, &ns->prefix))
530 return ns;
531 }
532
533 return NULL;
534 }
535
536 static HRESULT reader_inc_depth(xmlreader *reader)
537 {
538 return (++reader->depth >= reader->max_depth && reader->max_depth) ? SC_E_MAXELEMENTDEPTH : S_OK;
539 }
540
541 static void reader_dec_depth(xmlreader *reader)
542 {
543 if (reader->depth)
544 reader->depth--;
545 }
546
547 static HRESULT reader_push_ns(xmlreader *reader, const strval *prefix, const strval *uri, BOOL def)
548 {
549 struct ns *ns;
550 HRESULT hr;
551
552 ns = reader_alloc(reader, sizeof(*ns));
553 if (!ns) return E_OUTOFMEMORY;
554
555 if (def)
556 memset(&ns->prefix, 0, sizeof(ns->prefix));
557 else {
558 hr = reader_strvaldup(reader, prefix, &ns->prefix);
559 if (FAILED(hr)) {
560 reader_free(reader, ns);
561 return hr;
562 }
563 }
564
565 hr = reader_strvaldup(reader, uri, &ns->uri);
566 if (FAILED(hr)) {
567 reader_free_strvalued(reader, &ns->prefix);
568 reader_free(reader, ns);
569 return hr;
570 }
571
572 ns->element = NULL;
573 list_add_head(def ? &reader->nsdef : &reader->ns, &ns->entry);
574 return hr;
575 }
576
577 static void reader_free_element(xmlreader *reader, struct element *element)
578 {
579 reader_free_strvalued(reader, &element->prefix);
580 reader_free_strvalued(reader, &element->localname);
581 reader_free_strvalued(reader, &element->qname);
582 reader_free(reader, element);
583 }
584
585 static void reader_mark_ns_nodes(xmlreader *reader, struct element *element)
586 {
587 struct ns *ns;
588
589 LIST_FOR_EACH_ENTRY(ns, &reader->ns, struct ns, entry) {
590 if (ns->element)
591 break;
592 ns->element = element;
593 }
594
595 LIST_FOR_EACH_ENTRY(ns, &reader->nsdef, struct ns, entry) {
596 if (ns->element)
597 break;
598 ns->element = element;
599 }
600 }
601
602 static HRESULT reader_push_element(xmlreader *reader, strval *prefix, strval *localname,
603 strval *qname, const struct reader_position *position)
604 {
605 struct element *element;
606 HRESULT hr;
607
608 element = reader_alloc_zero(reader, sizeof(*element));
609 if (!element)
610 return E_OUTOFMEMORY;
611
612 if ((hr = reader_strvaldup(reader, prefix, &element->prefix)) == S_OK &&
613 (hr = reader_strvaldup(reader, localname, &element->localname)) == S_OK &&
614 (hr = reader_strvaldup(reader, qname, &element->qname)) == S_OK)
615 {
616 list_add_head(&reader->elements, &element->entry);
617 reader_mark_ns_nodes(reader, element);
618 reader->is_empty_element = FALSE;
619 element->position = *position;
620 }
621 else
622 reader_free_element(reader, element);
623
624 return hr;
625 }
626
627 static void reader_pop_ns_nodes(xmlreader *reader, struct element *element)
628 {
629 struct ns *ns, *ns2;
630
631 LIST_FOR_EACH_ENTRY_SAFE_REV(ns, ns2, &reader->ns, struct ns, entry) {
632 if (ns->element != element)
633 break;
634
635 list_remove(&ns->entry);
636 reader_free_strvalued(reader, &ns->prefix);
637 reader_free_strvalued(reader, &ns->uri);
638 reader_free(reader, ns);
639 }
640
641 if (!list_empty(&reader->nsdef)) {
642 ns = LIST_ENTRY(list_head(&reader->nsdef), struct ns, entry);
643 if (ns->element == element) {
644 list_remove(&ns->entry);
645 reader_free_strvalued(reader, &ns->prefix);
646 reader_free_strvalued(reader, &ns->uri);
647 reader_free(reader, ns);
648 }
649 }
650 }
651
652 static void reader_pop_element(xmlreader *reader)
653 {
654 struct element *element;
655
656 if (list_empty(&reader->elements))
657 return;
658
659 element = LIST_ENTRY(list_head(&reader->elements), struct element, entry);
660 list_remove(&element->entry);
661
662 reader_pop_ns_nodes(reader, element);
663 reader_free_element(reader, element);
664
665 /* It was a root element, the rest is expected as Misc */
666 if (list_empty(&reader->elements))
667 reader->instate = XmlReadInState_MiscEnd;
668 }
669
670 /* Always make a copy, cause strings are supposed to be null terminated. Null pointer for 'value'
671 means node value is to be determined. */
672 static void reader_set_strvalue(xmlreader *reader, XmlReaderStringValue type, const strval *value)
673 {
674 strval *v = &reader->strvalues[type];
675
676 reader_free_strvalue(reader, type);
677 if (!value)
678 {
679 v->str = NULL;
680 v->start = 0;
681 v->len = 0;
682 return;
683 }
684
685 if (value->str == strval_empty.str)
686 *v = *value;
687 else
688 {
689 if (type == StringValue_Value)
690 {
691 /* defer allocation for value string */
692 v->str = NULL;
693 v->start = value->start;
694 v->len = value->len;
695 }
696 else
697 {
698 v->str = reader_alloc(reader, (value->len + 1)*sizeof(WCHAR));
699 memcpy(v->str, reader_get_strptr(reader, value), value->len*sizeof(WCHAR));
700 v->str[value->len] = 0;
701 v->len = value->len;
702 }
703 }
704 }
705
706 static inline int is_reader_pending(xmlreader *reader)
707 {
708 return reader->input->pending;
709 }
710
711 static HRESULT init_encoded_buffer(xmlreaderinput *input, encoded_buffer *buffer)
712 {
713 const int initial_len = 0x2000;
714 buffer->data = readerinput_alloc(input, initial_len);
715 if (!buffer->data) return E_OUTOFMEMORY;
716
717 memset(buffer->data, 0, 4);
718 buffer->cur = 0;
719 buffer->allocated = initial_len;
720 buffer->written = 0;
721 buffer->prev_cr = FALSE;
722
723 return S_OK;
724 }
725
726 static void free_encoded_buffer(xmlreaderinput *input, encoded_buffer *buffer)
727 {
728 readerinput_free(input, buffer->data);
729 }
730
731 HRESULT get_code_page(xml_encoding encoding, UINT *cp)
732 {
733 if (encoding == XmlEncoding_Unknown)
734 {
735 FIXME("unsupported encoding %d\n", encoding);
736 return E_NOTIMPL;
737 }
738
739 *cp = xml_encoding_map[encoding].cp;
740
741 return S_OK;
742 }
743
744 xml_encoding parse_encoding_name(const WCHAR *name, int len)
745 {
746 int min, max, n, c;
747
748 if (!name) return XmlEncoding_Unknown;
749
750 min = 0;
751 max = ARRAY_SIZE(xml_encoding_map) - 1;
752
753 while (min <= max)
754 {
755 n = (min+max)/2;
756
757 if (len != -1)
758 c = wcsnicmp(xml_encoding_map[n].name, name, len);
759 else
760 c = wcsicmp(xml_encoding_map[n].name, name);
761 if (!c)
762 return xml_encoding_map[n].enc;
763
764 if (c > 0)
765 max = n-1;
766 else
767 min = n+1;
768 }
769
770 return XmlEncoding_Unknown;
771 }
772
773 static HRESULT alloc_input_buffer(xmlreaderinput *input)
774 {
775 input_buffer *buffer;
776 HRESULT hr;
777
778 input->buffer = NULL;
779
780 buffer = readerinput_alloc(input, sizeof(*buffer));
781 if (!buffer) return E_OUTOFMEMORY;
782
783 buffer->input = input;
784 buffer->code_page = ~0; /* code page is unknown at this point */
785 hr = init_encoded_buffer(input, &buffer->utf16);
786 if (hr != S_OK) {
787 readerinput_free(input, buffer);
788 return hr;
789 }
790
791 hr = init_encoded_buffer(input, &buffer->encoded);
792 if (hr != S_OK) {
793 free_encoded_buffer(input, &buffer->utf16);
794 readerinput_free(input, buffer);
795 return hr;
796 }
797
798 input->buffer = buffer;
799 return S_OK;
800 }
801
802 static void free_input_buffer(input_buffer *buffer)
803 {
804 free_encoded_buffer(buffer->input, &buffer->encoded);
805 free_encoded_buffer(buffer->input, &buffer->utf16);
806 readerinput_free(buffer->input, buffer);
807 }
808
809 static void readerinput_release_stream(xmlreaderinput *readerinput)
810 {
811 if (readerinput->stream) {
812 ISequentialStream_Release(readerinput->stream);
813 readerinput->stream = NULL;
814 }
815 }
816
817 /* Queries already stored interface for IStream/ISequentialStream.
818 Interface supplied on creation will be overwritten */
819 static inline HRESULT readerinput_query_for_stream(xmlreaderinput *readerinput)
820 {
821 HRESULT hr;
822
823 readerinput_release_stream(readerinput);
824 hr = IUnknown_QueryInterface(readerinput->input, &IID_IStream, (void**)&readerinput->stream);
825 if (hr != S_OK)
826 hr = IUnknown_QueryInterface(readerinput->input, &IID_ISequentialStream, (void**)&readerinput->stream);
827
828 return hr;
829 }
830
831 /* reads a chunk to raw buffer */
832 static HRESULT readerinput_growraw(xmlreaderinput *readerinput)
833 {
834 encoded_buffer *buffer = &readerinput->buffer->encoded;
835 /* to make sure aligned length won't exceed allocated length */
836 ULONG len = buffer->allocated - buffer->written - 4;
837 ULONG read;
838 HRESULT hr;
839
840 /* always try to get aligned to 4 bytes, so the only case we can get partially read characters is
841 variable width encodings like UTF-8 */
842 len = (len + 3) & ~3;
843 /* try to use allocated space or grow */
844 if (buffer->allocated - buffer->written < len)
845 {
846 buffer->allocated *= 2;
847 buffer->data = readerinput_realloc(readerinput, buffer->data, buffer->allocated);
848 len = buffer->allocated - buffer->written;
849 }
850
851 read = 0;
852 hr = ISequentialStream_Read(readerinput->stream, buffer->data + buffer->written, len, &read);
853 TRACE("written=%d, alloc=%d, requested=%d, read=%d, ret=0x%08x\n", buffer->written, buffer->allocated, len, read, hr);
854 readerinput->pending = hr == E_PENDING;
855 if (FAILED(hr)) return hr;
856 buffer->written += read;
857
858 return hr;
859 }
860
861 /* grows UTF-16 buffer so it has at least 'length' WCHAR chars free on return */
862 static void readerinput_grow(xmlreaderinput *readerinput, int length)
863 {
864 encoded_buffer *buffer = &readerinput->buffer->utf16;
865
866 length *= sizeof(WCHAR);
867 /* grow if needed, plus 4 bytes to be sure null terminator will fit in */
868 if (buffer->allocated < buffer->written + length + 4)
869 {
870 int grown_size = max(2*buffer->allocated, buffer->allocated + length);
871 buffer->data = readerinput_realloc(readerinput, buffer->data, grown_size);
872 buffer->allocated = grown_size;
873 }
874 }
875
876 static inline BOOL readerinput_is_utf8(xmlreaderinput *readerinput)
877 {
878 static const char startA[] = {'<','?'};
879 static const char commentA[] = {'<','!'};
880 encoded_buffer *buffer = &readerinput->buffer->encoded;
881 unsigned char *ptr = (unsigned char*)buffer->data;
882
883 return !memcmp(buffer->data, startA, sizeof(startA)) ||
884 !memcmp(buffer->data, commentA, sizeof(commentA)) ||
885 /* test start byte */
886 (ptr[0] == '<' &&
887 (
888 (ptr[1] && (ptr[1] <= 0x7f)) ||
889 (buffer->data[1] >> 5) == 0x6 || /* 2 bytes */
890 (buffer->data[1] >> 4) == 0xe || /* 3 bytes */
891 (buffer->data[1] >> 3) == 0x1e) /* 4 bytes */
892 );
893 }
894
895 static HRESULT readerinput_detectencoding(xmlreaderinput *readerinput, xml_encoding *enc)
896 {
897 encoded_buffer *buffer = &readerinput->buffer->encoded;
898 static const char utf8bom[] = {0xef,0xbb,0xbf};
899 static const char utf16lebom[] = {0xff,0xfe};
900 WCHAR *ptrW;
901
902 *enc = XmlEncoding_Unknown;
903
904 if (buffer->written <= 3)
905 {
906 HRESULT hr = readerinput_growraw(readerinput);
907 if (FAILED(hr)) return hr;
908 if (buffer->written < 3) return MX_E_INPUTEND;
909 }
910
911 ptrW = (WCHAR *)buffer->data;
912 /* try start symbols if we have enough data to do that, input buffer should contain
913 first chunk already */
914 if (readerinput_is_utf8(readerinput))
915 *enc = XmlEncoding_UTF8;
916 else if (*ptrW == '<')
917 {
918 ptrW++;
919 if (*ptrW == '?' || *ptrW == '!' || is_namestartchar(*ptrW))
920 *enc = XmlEncoding_UTF16;
921 }
922 /* try with BOM now */
923 else if (!memcmp(buffer->data, utf8bom, sizeof(utf8bom)))
924 {
925 buffer->cur += sizeof(utf8bom);
926 *enc = XmlEncoding_UTF8;
927 }
928 else if (!memcmp(buffer->data, utf16lebom, sizeof(utf16lebom)))
929 {
930 buffer->cur += sizeof(utf16lebom);
931 *enc = XmlEncoding_UTF16;
932 }
933
934 return S_OK;
935 }
936
937 static int readerinput_get_utf8_convlen(xmlreaderinput *readerinput)
938 {
939 encoded_buffer *buffer = &readerinput->buffer->encoded;
940 int len = buffer->written;
941
942 /* complete single byte char */
943 if (!(buffer->data[len-1] & 0x80)) return len;
944
945 /* find start byte of multibyte char */
946 while (--len && !(buffer->data[len] & 0xc0))
947 ;
948
949 return len;
950 }
951
952 /* Returns byte length of complete char sequence for buffer code page,
953 it's relative to current buffer position which is currently used for BOM handling
954 only. */
955 static int readerinput_get_convlen(xmlreaderinput *readerinput)
956 {
957 encoded_buffer *buffer = &readerinput->buffer->encoded;
958 int len;
959
960 if (readerinput->buffer->code_page == CP_UTF8)
961 len = readerinput_get_utf8_convlen(readerinput);
962 else
963 len = buffer->written;
964
965 TRACE("%d\n", len - buffer->cur);
966 return len - buffer->cur;
967 }
968
969 /* It's possible that raw buffer has some leftovers from last conversion - some char
970 sequence that doesn't represent a full code point. Length argument should be calculated with
971 readerinput_get_convlen(), if it's -1 it will be calculated here. */
972 static void readerinput_shrinkraw(xmlreaderinput *readerinput, int len)
973 {
974 encoded_buffer *buffer = &readerinput->buffer->encoded;
975
976 if (len == -1)
977 len = readerinput_get_convlen(readerinput);
978
979 memmove(buffer->data, buffer->data + buffer->cur + (buffer->written - len), len);
980 /* everything below cur is lost too */
981 buffer->written -= len + buffer->cur;
982 /* after this point we don't need cur offset really,
983 it's used only to mark where actual data begins when first chunk is read */
984 buffer->cur = 0;
985 }
986
987 static void fixup_buffer_cr(encoded_buffer *buffer, int off)
988 {
989 BOOL prev_cr = buffer->prev_cr;
990 const WCHAR *src;
991 WCHAR *dest;
992
993 src = dest = (WCHAR*)buffer->data + off;
994 while ((const char*)src < buffer->data + buffer->written)
995 {
996 if (*src == '\r')
997 {
998 *dest++ = '\n';
999 src++;
1000 prev_cr = TRUE;
1001 continue;
1002 }
1003 if(prev_cr && *src == '\n')
1004 src++;
1005 else
1006 *dest++ = *src++;
1007 prev_cr = FALSE;
1008 }
1009
1010 buffer->written = (char*)dest - buffer->data;
1011 buffer->prev_cr = prev_cr;
1012 *dest = 0;
1013 }
1014
1015 /* note that raw buffer content is kept */
1016 static void readerinput_switchencoding(xmlreaderinput *readerinput, xml_encoding enc)
1017 {
1018 encoded_buffer *src = &readerinput->buffer->encoded;
1019 encoded_buffer *dest = &readerinput->buffer->utf16;
1020 int len, dest_len;
1021 UINT cp = ~0u;
1022 HRESULT hr;
1023 WCHAR *ptr;
1024
1025 hr = get_code_page(enc, &cp);
1026 if (FAILED(hr)) return;
1027
1028 readerinput->buffer->code_page = cp;
1029 len = readerinput_get_convlen(readerinput);
1030
1031 TRACE("switching to cp %d\n", cp);
1032
1033 /* just copy in this case */
1034 if (enc == XmlEncoding_UTF16)
1035 {
1036 readerinput_grow(readerinput, len);
1037 memcpy(dest->data, src->data + src->cur, len);
1038 dest->written += len*sizeof(WCHAR);
1039 }
1040 else
1041 {
1042 dest_len = MultiByteToWideChar(cp, 0, src->data + src->cur, len, NULL, 0);
1043 readerinput_grow(readerinput, dest_len);
1044 ptr = (WCHAR*)dest->data;
1045 MultiByteToWideChar(cp, 0, src->data + src->cur, len, ptr, dest_len);
1046 ptr[dest_len] = 0;
1047 dest->written += dest_len*sizeof(WCHAR);
1048 }
1049
1050 fixup_buffer_cr(dest, 0);
1051 }
1052
1053 /* shrinks parsed data a buffer begins with */
1054 static void reader_shrink(xmlreader *reader)
1055 {
1056 encoded_buffer *buffer = &reader->input->buffer->utf16;
1057
1058 /* avoid to move too often using threshold shrink length */
1059 if (buffer->cur*sizeof(WCHAR) > buffer->written / 2)
1060 {
1061 buffer->written -= buffer->cur*sizeof(WCHAR);
1062 memmove(buffer->data, (WCHAR*)buffer->data + buffer->cur, buffer->written);
1063 buffer->cur = 0;
1064 *(WCHAR*)&buffer->data[buffer->written] = 0;
1065 }
1066 }
1067
1068 /* This is a normal way for reader to get new data converted from raw buffer to utf16 buffer.
1069 It won't attempt to shrink but will grow destination buffer if needed */
1070 static HRESULT reader_more(xmlreader *reader)
1071 {
1072 xmlreaderinput *readerinput = reader->input;
1073 encoded_buffer *src = &readerinput->buffer->encoded;
1074 encoded_buffer *dest = &readerinput->buffer->utf16;
1075 UINT cp = readerinput->buffer->code_page;
1076 int len, dest_len, prev_len;
1077 HRESULT hr;
1078 WCHAR *ptr;
1079
1080 /* get some raw data from stream first */
1081 hr = readerinput_growraw(readerinput);
1082 len = readerinput_get_convlen(readerinput);
1083 prev_len = dest->written / sizeof(WCHAR);
1084
1085 /* just copy for UTF-16 case */
1086 if (cp == 1200)
1087 {
1088 readerinput_grow(readerinput, len);
1089 memcpy(dest->data + dest->written, src->data + src->cur, len);
1090 dest->written += len*sizeof(WCHAR);
1091 }
1092 else
1093 {
1094 dest_len = MultiByteToWideChar(cp, 0, src->data + src->cur, len, NULL, 0);
1095 readerinput_grow(readerinput, dest_len);
1096 ptr = (WCHAR*)(dest->data + dest->written);
1097 MultiByteToWideChar(cp, 0, src->data + src->cur, len, ptr, dest_len);
1098 ptr[dest_len] = 0;
1099 dest->written += dest_len*sizeof(WCHAR);
1100 /* get rid of processed data */
1101 readerinput_shrinkraw(readerinput, len);
1102 }
1103
1104 fixup_buffer_cr(dest, prev_len);
1105 return hr;
1106 }
1107
1108 static inline UINT reader_get_cur(xmlreader *reader)
1109 {
1110 return reader->input->buffer->utf16.cur;
1111 }
1112
1113 static inline WCHAR *reader_get_ptr(xmlreader *reader)
1114 {
1115 encoded_buffer *buffer = &reader->input->buffer->utf16;
1116 WCHAR *ptr = (WCHAR*)buffer->data + buffer->cur;
1117 if (!*ptr) reader_more(reader);
1118 return (WCHAR*)buffer->data + buffer->cur;
1119 }
1120
1121 static int reader_cmp(xmlreader *reader, const WCHAR *str)
1122 {
1123 int i=0;
1124 const WCHAR *ptr = reader_get_ptr(reader);
1125 while (str[i])
1126 {
1127 if (!ptr[i])
1128 {
1129 reader_more(reader);
1130 ptr = reader_get_ptr(reader);
1131 }
1132 if (str[i] != ptr[i])
1133 return ptr[i] - str[i];
1134 i++;
1135 }
1136 return 0;
1137 }
1138
1139 static void reader_update_position(xmlreader *reader, WCHAR ch)
1140 {
1141 if (ch == '\r')
1142 reader->position.line_position = 1;
1143 else if (ch == '\n')
1144 {
1145 reader->position.line_number++;
1146 reader->position.line_position = 1;
1147 }
1148 else
1149 reader->position.line_position++;
1150 }
1151
1152 /* moves cursor n WCHARs forward */
1153 static void reader_skipn(xmlreader *reader, int n)
1154 {
1155 encoded_buffer *buffer = &reader->input->buffer->utf16;
1156 const WCHAR *ptr;
1157
1158 while (*(ptr = reader_get_ptr(reader)) && n--)
1159 {
1160 reader_update_position(reader, *ptr);
1161 buffer->cur++;
1162 }
1163 }
1164
1165 static inline BOOL is_wchar_space(WCHAR ch)
1166 {
1167 return ch == ' ' || ch == '\t' || ch == '\r' || ch == '\n';
1168 }
1169
1170 /* [3] S ::= (#x20 | #x9 | #xD | #xA)+ */
1171 static int reader_skipspaces(xmlreader *reader)
1172 {
1173 const WCHAR *ptr = reader_get_ptr(reader);
1174 UINT start = reader_get_cur(reader);
1175
1176 while (is_wchar_space(*ptr))
1177 {
1178 reader_skipn(reader, 1);
1179 ptr = reader_get_ptr(reader);
1180 }
1181
1182 return reader_get_cur(reader) - start;
1183 }
1184
1185 /* [26] VersionNum ::= '1.' [0-9]+ */
1186 static HRESULT reader_parse_versionnum(xmlreader *reader, strval *val)
1187 {
1188 static const WCHAR onedotW[] = {'1','.',0};
1189 WCHAR *ptr, *ptr2;
1190 UINT start;
1191
1192 if (reader_cmp(reader, onedotW)) return WC_E_XMLDECL;
1193
1194 start = reader_get_cur(reader);
1195 /* skip "1." */
1196 reader_skipn(reader, 2);
1197
1198 ptr2 = ptr = reader_get_ptr(reader);
1199 while (*ptr >= '0' && *ptr <= '9')
1200 {
1201 reader_skipn(reader, 1);
1202 ptr = reader_get_ptr(reader);
1203 }
1204
1205 if (ptr2 == ptr) return WC_E_DIGIT;
1206 reader_init_strvalue(start, reader_get_cur(reader)-start, val);
1207 TRACE("version=%s\n", debug_strval(reader, val));
1208 return S_OK;
1209 }
1210
1211 /* [25] Eq ::= S? '=' S? */
1212 static HRESULT reader_parse_eq(xmlreader *reader)
1213 {
1214 static const WCHAR eqW[] = {'=',0};
1215 reader_skipspaces(reader);
1216 if (reader_cmp(reader, eqW)) return WC_E_EQUAL;
1217 /* skip '=' */
1218 reader_skipn(reader, 1);
1219 reader_skipspaces(reader);
1220 return S_OK;
1221 }
1222
1223 /* [24] VersionInfo ::= S 'version' Eq ("'" VersionNum "'" | '"' VersionNum '"') */
1224 static HRESULT reader_parse_versioninfo(xmlreader *reader)
1225 {
1226 static const WCHAR versionW[] = {'v','e','r','s','i','o','n',0};
1227 struct reader_position position;
1228 strval val, name;
1229 HRESULT hr;
1230
1231 if (!reader_skipspaces(reader)) return WC_E_WHITESPACE;
1232
1233 position = reader->position;
1234 if (reader_cmp(reader, versionW)) return WC_E_XMLDECL;
1235 reader_init_strvalue(reader_get_cur(reader), 7, &name);
1236 /* skip 'version' */
1237 reader_skipn(reader, 7);
1238
1239 hr = reader_parse_eq(reader);
1240 if (FAILED(hr)) return hr;
1241
1242 if (reader_cmp(reader, quoteW) && reader_cmp(reader, dblquoteW))
1243 return WC_E_QUOTE;
1244 /* skip "'"|'"' */
1245 reader_skipn(reader, 1);
1246
1247 hr = reader_parse_versionnum(reader, &val);
1248 if (FAILED(hr)) return hr;
1249
1250 if (reader_cmp(reader, quoteW) && reader_cmp(reader, dblquoteW))
1251 return WC_E_QUOTE;
1252
1253 /* skip "'"|'"' */
1254 reader_skipn(reader, 1);
1255
1256 return reader_add_attr(reader, NULL, &name, NULL, &val, &position, 0);
1257 }
1258
1259 /* ([A-Za-z0-9._] | '-') */
1260 static inline BOOL is_wchar_encname(WCHAR ch)
1261 {
1262 return ((ch >= 'A' && ch <= 'Z') ||
1263 (ch >= 'a' && ch <= 'z') ||
1264 (ch >= '0' && ch <= '9') ||
1265 (ch == '.') || (ch == '_') ||
1266 (ch == '-'));
1267 }
1268
1269 /* [81] EncName ::= [A-Za-z] ([A-Za-z0-9._] | '-')* */
1270 static HRESULT reader_parse_encname(xmlreader *reader, strval *val)
1271 {
1272 WCHAR *start = reader_get_ptr(reader), *ptr;
1273 xml_encoding enc;
1274 int len;
1275
1276 if ((*start < 'A' || *start > 'Z') && (*start < 'a' || *start > 'z'))
1277 return WC_E_ENCNAME;
1278
1279 val->start = reader_get_cur(reader);
1280
1281 ptr = start;
1282 while (is_wchar_encname(*++ptr))
1283 ;
1284
1285 len = ptr - start;
1286 enc = parse_encoding_name(start, len);
1287 TRACE("encoding name %s\n", debugstr_wn(start, len));
1288 val->str = start;
1289 val->len = len;
1290
1291 if (enc == XmlEncoding_Unknown)
1292 return WC_E_ENCNAME;
1293
1294 /* skip encoding name */
1295 reader_skipn(reader, len);
1296 return S_OK;
1297 }
1298
1299 /* [80] EncodingDecl ::= S 'encoding' Eq ('"' EncName '"' | "'" EncName "'" ) */
1300 static HRESULT reader_parse_encdecl(xmlreader *reader)
1301 {
1302 static const WCHAR encodingW[] = {'e','n','c','o','d','i','n','g',0};
1303 struct reader_position position;
1304 strval name, val;
1305 HRESULT hr;
1306
1307 if (!reader_skipspaces(reader)) return S_FALSE;
1308
1309 position = reader->position;
1310 if (reader_cmp(reader, encodingW)) return S_FALSE;
1311 name.str = reader_get_ptr(reader);
1312 name.start = reader_get_cur(reader);
1313 name.len = 8;
1314 /* skip 'encoding' */
1315 reader_skipn(reader, 8);
1316
1317 hr = reader_parse_eq(reader);
1318 if (FAILED(hr)) return hr;
1319
1320 if (reader_cmp(reader, quoteW) && reader_cmp(reader, dblquoteW))
1321 return WC_E_QUOTE;
1322 /* skip "'"|'"' */
1323 reader_skipn(reader, 1);
1324
1325 hr = reader_parse_encname(reader, &val);
1326 if (FAILED(hr)) return hr;
1327
1328 if (reader_cmp(reader, quoteW) && reader_cmp(reader, dblquoteW))
1329 return WC_E_QUOTE;
1330
1331 /* skip "'"|'"' */
1332 reader_skipn(reader, 1);
1333
1334 return reader_add_attr(reader, NULL, &name, NULL, &val, &position, 0);
1335 }
1336
1337 /* [32] SDDecl ::= S 'standalone' Eq (("'" ('yes' | 'no') "'") | ('"' ('yes' | 'no') '"')) */
1338 static HRESULT reader_parse_sddecl(xmlreader *reader)
1339 {
1340 static const WCHAR standaloneW[] = {'s','t','a','n','d','a','l','o','n','e',0};
1341 static const WCHAR yesW[] = {'y','e','s',0};
1342 static const WCHAR noW[] = {'n','o',0};
1343 struct reader_position position;
1344 strval name, val;
1345 UINT start;
1346 HRESULT hr;
1347
1348 if (!reader_skipspaces(reader)) return S_FALSE;
1349
1350 position = reader->position;
1351 if (reader_cmp(reader, standaloneW)) return S_FALSE;
1352 reader_init_strvalue(reader_get_cur(reader), 10, &name);
1353 /* skip 'standalone' */
1354 reader_skipn(reader, 10);
1355
1356 hr = reader_parse_eq(reader);
1357 if (FAILED(hr)) return hr;
1358
1359 if (reader_cmp(reader, quoteW) && reader_cmp(reader, dblquoteW))
1360 return WC_E_QUOTE;
1361 /* skip "'"|'"' */
1362 reader_skipn(reader, 1);
1363
1364 if (reader_cmp(reader, yesW) && reader_cmp(reader, noW))
1365 return WC_E_XMLDECL;
1366
1367 start = reader_get_cur(reader);
1368 /* skip 'yes'|'no' */
1369 reader_skipn(reader, reader_cmp(reader, yesW) ? 2 : 3);
1370 reader_init_strvalue(start, reader_get_cur(reader)-start, &val);
1371 TRACE("standalone=%s\n", debug_strval(reader, &val));
1372
1373 if (reader_cmp(reader, quoteW) && reader_cmp(reader, dblquoteW))
1374 return WC_E_QUOTE;
1375 /* skip "'"|'"' */
1376 reader_skipn(reader, 1);
1377
1378 return reader_add_attr(reader, NULL, &name, NULL, &val, &position, 0);
1379 }
1380
1381 /* [23] XMLDecl ::= '<?xml' VersionInfo EncodingDecl? SDDecl? S? '?>' */
1382 static HRESULT reader_parse_xmldecl(xmlreader *reader)
1383 {
1384 static const WCHAR xmldeclW[] = {'<','?','x','m','l',' ',0};
1385 static const WCHAR declcloseW[] = {'?','>',0};
1386 struct reader_position position;
1387 HRESULT hr;
1388
1389 /* check if we have "<?xml " */
1390 if (reader_cmp(reader, xmldeclW))
1391 return S_FALSE;
1392
1393 reader_skipn(reader, 2);
1394 position = reader->position;
1395 reader_skipn(reader, 3);
1396 hr = reader_parse_versioninfo(reader);
1397 if (FAILED(hr))
1398 return hr;
1399
1400 hr = reader_parse_encdecl(reader);
1401 if (FAILED(hr))
1402 return hr;
1403
1404 hr = reader_parse_sddecl(reader);
1405 if (FAILED(hr))
1406 return hr;
1407
1408 reader_skipspaces(reader);
1409 if (reader_cmp(reader, declcloseW))
1410 return WC_E_XMLDECL;
1411
1412 /* skip '?>' */
1413 reader_skipn(reader, 2);
1414
1415 reader->nodetype = XmlNodeType_XmlDeclaration;
1416 reader->empty_element.position = position;
1417 reader_set_strvalue(reader, StringValue_LocalName, &strval_xml);
1418 reader_set_strvalue(reader, StringValue_QualifiedName, &strval_xml);
1419
1420 return S_OK;
1421 }
1422
1423 /* [15] Comment ::= '<!--' ((Char - '-') | ('-' (Char - '-')))* '-->' */
1424 static HRESULT reader_parse_comment(xmlreader *reader)
1425 {
1426 WCHAR *ptr;
1427 UINT start;
1428
1429 if (reader->resumestate == XmlReadResumeState_Comment)
1430 {
1431 start = reader->resume[XmlReadResume_Body];
1432 ptr = reader_get_ptr(reader);
1433 }
1434 else
1435 {
1436 /* skip '<!--' */
1437 reader_skipn(reader, 4);
1438 reader_shrink(reader);
1439 ptr = reader_get_ptr(reader);
1440 start = reader_get_cur(reader);
1441 reader->nodetype = XmlNodeType_Comment;
1442 reader->resume[XmlReadResume_Body] = start;
1443 reader->resumestate = XmlReadResumeState_Comment;
1444 reader_set_strvalue(reader, StringValue_Value, NULL);
1445 }
1446
1447 /* will exit when there's no more data, it won't attempt to
1448 read more from stream */
1449 while (*ptr)
1450 {
1451 if (ptr[0] == '-')
1452 {
1453 if (ptr[1] == '-')
1454 {
1455 if (ptr[2] == '>')
1456 {
1457 strval value;
1458
1459 reader_init_strvalue(start, reader_get_cur(reader)-start, &value);
1460 TRACE("%s\n", debug_strval(reader, &value));
1461
1462 /* skip rest of markup '->' */
1463 reader_skipn(reader, 3);
1464
1465 reader_set_strvalue(reader, StringValue_Value, &value);
1466 reader->resume[XmlReadResume_Body] = 0;
1467 reader->resumestate = XmlReadResumeState_Initial;
1468 return S_OK;
1469 }
1470 else
1471 return WC_E_COMMENT;
1472 }
1473 }
1474
1475 reader_skipn(reader, 1);
1476 ptr++;
1477 }
1478
1479 return S_OK;
1480 }
1481
1482 /* [2] Char ::= #x9 | #xA | #xD | [#x20-#xD7FF] | [#xE000-#xFFFD] | [#x10000-#x10FFFF] */
1483 static inline BOOL is_char(WCHAR ch)
1484 {
1485 return (ch == '\t') || (ch == '\r') || (ch == '\n') ||
1486 (ch >= 0x20 && ch <= 0xd7ff) ||
1487 (ch >= 0xd800 && ch <= 0xdbff) || /* high surrogate */
1488 (ch >= 0xdc00 && ch <= 0xdfff) || /* low surrogate */
1489 (ch >= 0xe000 && ch <= 0xfffd);
1490 }
1491
1492 /* [13] PubidChar ::= #x20 | #xD | #xA | [a-zA-Z0-9] | [-'()+,./:=?;!*#@$_%] */
1493 BOOL is_pubchar(WCHAR ch)
1494 {
1495 return (ch == ' ') ||
1496 (ch >= 'a' && ch <= 'z') ||
1497 (ch >= 'A' && ch <= 'Z') ||
1498 (ch >= '0' && ch <= '9') ||
1499 (ch >= '-' && ch <= ';') || /* '()*+,-./:; */
1500 (ch == '=') || (ch == '?') ||
1501 (ch == '@') || (ch == '!') ||
1502 (ch >= '#' && ch <= '%') || /* #$% */
1503 (ch == '_') || (ch == '\r') || (ch == '\n');
1504 }
1505
1506 BOOL is_namestartchar(WCHAR ch)
1507 {
1508 return (ch == ':') || (ch >= 'A' && ch <= 'Z') ||
1509 (ch == '_') || (ch >= 'a' && ch <= 'z') ||
1510 (ch >= 0xc0 && ch <= 0xd6) ||
1511 (ch >= 0xd8 && ch <= 0xf6) ||
1512 (ch >= 0xf8 && ch <= 0x2ff) ||
1513 (ch >= 0x370 && ch <= 0x37d) ||
1514 (ch >= 0x37f && ch <= 0x1fff) ||
1515 (ch >= 0x200c && ch <= 0x200d) ||
1516 (ch >= 0x2070 && ch <= 0x218f) ||
1517 (ch >= 0x2c00 && ch <= 0x2fef) ||
1518 (ch >= 0x3001 && ch <= 0xd7ff) ||
1519 (ch >= 0xd800 && ch <= 0xdbff) || /* high surrogate */
1520 (ch >= 0xdc00 && ch <= 0xdfff) || /* low surrogate */
1521 (ch >= 0xf900 && ch <= 0xfdcf) ||
1522 (ch >= 0xfdf0 && ch <= 0xfffd);
1523 }
1524
1525 /* [4 NS] NCName ::= Name - (Char* ':' Char*) */
1526 BOOL is_ncnamechar(WCHAR ch)
1527 {
1528 return (ch >= 'A' && ch <= 'Z') ||
1529 (ch == '_') || (ch >= 'a' && ch <= 'z') ||
1530 (ch == '-') || (ch == '.') ||
1531 (ch >= '0' && ch <= '9') ||
1532 (ch == 0xb7) ||
1533 (ch >= 0xc0 && ch <= 0xd6) ||
1534 (ch >= 0xd8 && ch <= 0xf6) ||
1535 (ch >= 0xf8 && ch <= 0x2ff) ||
1536 (ch >= 0x300 && ch <= 0x36f) ||
1537 (ch >= 0x370 && ch <= 0x37d) ||
1538 (ch >= 0x37f && ch <= 0x1fff) ||
1539 (ch >= 0x200c && ch <= 0x200d) ||
1540 (ch >= 0x203f && ch <= 0x2040) ||
1541 (ch >= 0x2070 && ch <= 0x218f) ||
1542 (ch >= 0x2c00 && ch <= 0x2fef) ||
1543 (ch >= 0x3001 && ch <= 0xd7ff) ||
1544 (ch >= 0xd800 && ch <= 0xdbff) || /* high surrogate */
1545 (ch >= 0xdc00 && ch <= 0xdfff) || /* low surrogate */
1546 (ch >= 0xf900 && ch <= 0xfdcf) ||
1547 (ch >= 0xfdf0 && ch <= 0xfffd);
1548 }
1549
1550 BOOL is_namechar(WCHAR ch)
1551 {
1552 return (ch == ':') || is_ncnamechar(ch);
1553 }
1554
1555 static XmlNodeType reader_get_nodetype(const xmlreader *reader)
1556 {
1557 /* When we're on attribute always return attribute type, container node type is kept.
1558 Note that container is not necessarily an element, and attribute doesn't mean it's
1559 an attribute in XML spec terms. */
1560 return reader->attr ? XmlNodeType_Attribute : reader->nodetype;
1561 }
1562
1563 /* [4] NameStartChar ::= ":" | [A-Z] | "_" | [a-z] | [#xC0-#xD6] | [#xD8-#xF6] | [#xF8-#x2FF] | [#x370-#x37D] |
1564 [#x37F-#x1FFF] | [#x200C-#x200D] | [#x2070-#x218F] | [#x2C00-#x2FEF] | [#x3001-#xD7FF] |
1565 [#xF900-#xFDCF] | [#xFDF0-#xFFFD] | [#x10000-#xEFFFF]
1566 [4a] NameChar ::= NameStartChar | "-" | "." | [0-9] | #xB7 | [#x0300-#x036F] | [#x203F-#x2040]
1567 [5] Name ::= NameStartChar (NameChar)* */
1568 static HRESULT reader_parse_name(xmlreader *reader, strval *name)
1569 {
1570 WCHAR *ptr;
1571 UINT start;
1572
1573 if (reader->resume[XmlReadResume_Name])
1574 {
1575 start = reader->resume[XmlReadResume_Name];
1576 ptr = reader_get_ptr(reader);
1577 }
1578 else
1579 {
1580 ptr = reader_get_ptr(reader);
1581 start = reader_get_cur(reader);
1582 if (!is_namestartchar(*ptr)) return WC_E_NAMECHARACTER;
1583 }
1584
1585 while (is_namechar(*ptr))
1586 {
1587 reader_skipn(reader, 1);
1588 ptr = reader_get_ptr(reader);
1589 }
1590
1591 if (is_reader_pending(reader))
1592 {
1593 reader->resume[XmlReadResume_Name] = start;
1594 return E_PENDING;
1595 }
1596 else
1597 reader->resume[XmlReadResume_Name] = 0;
1598
1599 reader_init_strvalue(start, reader_get_cur(reader)-start, name);
1600 TRACE("name %s:%d\n", debug_strval(reader, name), name->len);
1601
1602 return S_OK;
1603 }
1604
1605 /* [17] PITarget ::= Name - (('X' | 'x') ('M' | 'm') ('L' | 'l')) */
1606 static HRESULT reader_parse_pitarget(xmlreader *reader, strval *target)
1607 {
1608 static const WCHAR xmlW[] = {'x','m','l'};
1609 static const strval xmlval = { (WCHAR*)xmlW, 3 };
1610 strval name;
1611 WCHAR *ptr;
1612 HRESULT hr;
1613 UINT i;
1614
1615 hr = reader_parse_name(reader, &name);
1616 if (FAILED(hr)) return is_reader_pending(reader) ? E_PENDING : WC_E_PI;
1617
1618 /* now that we got name check for illegal content */
1619 if (strval_eq(reader, &name, &xmlval))
1620 return WC_E_LEADINGXML;
1621
1622 /* PITarget can't be a qualified name */
1623 ptr = reader_get_strptr(reader, &name);
1624 for (i = 0; i < name.len; i++)
1625 if (ptr[i] == ':')
1626 return i ? NC_E_NAMECOLON : WC_E_PI;
1627
1628 TRACE("pitarget %s:%d\n", debug_strval(reader, &name), name.len);
1629 *target = name;
1630 return S_OK;
1631 }
1632
1633 /* [16] PI ::= '<?' PITarget (S (Char* - (Char* '?>' Char*)))? '?>' */
1634 static HRESULT reader_parse_pi(xmlreader *reader)
1635 {
1636 strval target;
1637 WCHAR *ptr;
1638 UINT start;
1639 HRESULT hr;
1640
1641 switch (reader->resumestate)
1642 {
1643 case XmlReadResumeState_Initial:
1644 /* skip '<?' */
1645 reader_skipn(reader, 2);
1646 reader_shrink(reader);
1647 reader->resumestate = XmlReadResumeState_PITarget;
1648 case XmlReadResumeState_PITarget:
1649 hr = reader_parse_pitarget(reader, &target);
1650 if (FAILED(hr)) return hr;
1651 reader_set_strvalue(reader, StringValue_LocalName, &target);
1652 reader_set_strvalue(reader, StringValue_QualifiedName, &target);
1653 reader_set_strvalue(reader, StringValue_Value, &strval_empty);
1654 reader->resumestate = XmlReadResumeState_PIBody;
1655 reader->resume[XmlReadResume_Body] = reader_get_cur(reader);
1656 default:
1657 ;
1658 }
1659
1660 start = reader->resume[XmlReadResume_Body];
1661 ptr = reader_get_ptr(reader);
1662 while (*ptr)
1663 {
1664 if (ptr[0] == '?')
1665 {
1666 if (ptr[1] == '>')
1667 {
1668 UINT cur = reader_get_cur(reader);
1669 strval value;
1670
1671 /* strip all leading whitespace chars */
1672 while (start < cur)
1673 {
1674 ptr = reader_get_ptr2(reader, start);
1675 if (!is_wchar_space(*ptr)) break;
1676 start++;
1677 }
1678
1679 reader_init_strvalue(start, cur-start, &value);
1680
1681 /* skip '?>' */
1682 reader_skipn(reader, 2);
1683 TRACE("%s\n", debug_strval(reader, &value));
1684 reader->nodetype = XmlNodeType_ProcessingInstruction;
1685 reader->resumestate = XmlReadResumeState_Initial;
1686 reader->resume[XmlReadResume_Body] = 0;
1687 reader_set_strvalue(reader, StringValue_Value, &value);
1688 return S_OK;
1689 }
1690 }
1691
1692 reader_skipn(reader, 1);
1693 ptr = reader_get_ptr(reader);
1694 }
1695
1696 return S_OK;
1697 }
1698
1699 /* This one is used to parse significant whitespace nodes, like in Misc production */
1700 static HRESULT reader_parse_whitespace(xmlreader *reader)
1701 {
1702 switch (reader->resumestate)
1703 {
1704 case XmlReadResumeState_Initial:
1705 reader_shrink(reader);
1706 reader->resumestate = XmlReadResumeState_Whitespace;
1707 reader->resume[XmlReadResume_Body] = reader_get_cur(reader);
1708 reader->nodetype = XmlNodeType_Whitespace;
1709 reader_set_strvalue(reader, StringValue_LocalName, &strval_empty);
1710 reader_set_strvalue(reader, StringValue_QualifiedName, &strval_empty);
1711 reader_set_strvalue(reader, StringValue_Value, &strval_empty);
1712 /* fallthrough */
1713 case XmlReadResumeState_Whitespace:
1714 {
1715 strval value;
1716 UINT start;
1717
1718 reader_skipspaces(reader);
1719 if (is_reader_pending(reader)) return S_OK;
1720
1721 start = reader->resume[XmlReadResume_Body];
1722 reader_init_strvalue(start, reader_get_cur(reader)-start, &value);
1723 reader_set_strvalue(reader, StringValue_Value, &value);
1724 TRACE("%s\n", debug_strval(reader, &value));
1725 reader->resumestate = XmlReadResumeState_Initial;
1726 }
1727 default:
1728 ;
1729 }
1730
1731 return S_OK;
1732 }
1733
1734 /* [27] Misc ::= Comment | PI | S */
1735 static HRESULT reader_parse_misc(xmlreader *reader)
1736 {
1737 HRESULT hr = S_FALSE;
1738
1739 if (reader->resumestate != XmlReadResumeState_Initial)
1740 {
1741 hr = reader_more(reader);
1742 if (FAILED(hr)) return hr;
1743
1744 /* finish current node */
1745 switch (reader->resumestate)
1746 {
1747 case XmlReadResumeState_PITarget:
1748 case XmlReadResumeState_PIBody:
1749 return reader_parse_pi(reader);
1750 case XmlReadResumeState_Comment:
1751 return reader_parse_comment(reader);
1752 case XmlReadResumeState_Whitespace:
1753 return reader_parse_whitespace(reader);
1754 default:
1755 ERR("unknown resume state %d\n", reader->resumestate);
1756 }
1757 }
1758
1759 while (1)
1760 {
1761 const WCHAR *cur = reader_get_ptr(reader);
1762
1763 if (is_wchar_space(*cur))
1764 hr = reader_parse_whitespace(reader);
1765 else if (!reader_cmp(reader, commentW))
1766 hr = reader_parse_comment(reader);
1767 else if (!reader_cmp(reader, piW))
1768 hr = reader_parse_pi(reader);
1769 else
1770 break;
1771
1772 if (hr != S_FALSE) return hr;
1773 }
1774
1775 return hr;
1776 }
1777
1778 /* [11] SystemLiteral ::= ('"' [^"]* '"') | ("'" [^']* "'") */
1779 static HRESULT reader_parse_sys_literal(xmlreader *reader, strval *literal)
1780 {
1781 WCHAR *cur = reader_get_ptr(reader), quote;
1782 UINT start;
1783
1784 if (*cur != '"' && *cur != '\'') return WC_E_QUOTE;
1785
1786 quote = *cur;
1787 reader_skipn(reader, 1);
1788
1789 cur = reader_get_ptr(reader);
1790 start = reader_get_cur(reader);
1791 while (is_char(*cur) && *cur != quote)
1792 {
1793 reader_skipn(reader, 1);
1794 cur = reader_get_ptr(reader);
1795 }
1796 reader_init_strvalue(start, reader_get_cur(reader)-start, literal);
1797 if (*cur == quote) reader_skipn(reader, 1);
1798
1799 TRACE("%s\n", debug_strval(reader, literal));
1800 return S_OK;
1801 }
1802
1803 /* [12] PubidLiteral ::= '"' PubidChar* '"' | "'" (PubidChar - "'")* "'"
1804 [13] PubidChar ::= #x20 | #xD | #xA | [a-zA-Z0-9] | [-'()+,./:=?;!*#@$_%] */
1805 static HRESULT reader_parse_pub_literal(xmlreader *reader, strval *literal)
1806 {
1807 WCHAR *cur = reader_get_ptr(reader), quote;
1808 UINT start;
1809
1810 if (*cur != '"' && *cur != '\'') return WC_E_QUOTE;
1811
1812 quote = *cur;
1813 reader_skipn(reader, 1);
1814
1815 start = reader_get_cur(reader);
1816 cur = reader_get_ptr(reader);
1817 while (is_pubchar(*cur) && *cur != quote)
1818 {
1819 reader_skipn(reader, 1);
1820 cur = reader_get_ptr(reader);
1821 }
1822 reader_init_strvalue(start, reader_get_cur(reader)-start, literal);
1823 if (*cur == quote) reader_skipn(reader, 1);
1824
1825 TRACE("%s\n", debug_strval(reader, literal));
1826 return S_OK;
1827 }
1828
1829 /* [75] ExternalID ::= 'SYSTEM' S SystemLiteral | 'PUBLIC' S PubidLiteral S SystemLiteral */
1830 static HRESULT reader_parse_externalid(xmlreader *reader)
1831 {
1832 static WCHAR systemW[] = {'S','Y','S','T','E','M',0};
1833 static WCHAR publicW[] = {'P','U','B','L','I','C',0};
1834 struct reader_position position = reader->position;
1835 strval name, sys;
1836 HRESULT hr;
1837 int cnt;
1838
1839 if (!reader_cmp(reader, publicW)) {
1840 strval pub;
1841
1842 /* public id */
1843 reader_skipn(reader, 6);
1844 cnt = reader_skipspaces(reader);
1845 if (!cnt) return WC_E_WHITESPACE;
1846
1847 hr = reader_parse_pub_literal(reader, &pub);
1848 if (FAILED(hr)) return hr;
1849
1850 reader_init_cstrvalue(publicW, lstrlenW(publicW), &name);
1851 hr = reader_add_attr(reader, NULL, &name, NULL, &pub, &position, 0);
1852 if (FAILED(hr)) return hr;
1853
1854 cnt = reader_skipspaces(reader);
1855 if (!cnt) return S_OK;
1856
1857 /* optional system id */
1858 hr = reader_parse_sys_literal(reader, &sys);
1859 if (FAILED(hr)) return S_OK;
1860
1861 reader_init_cstrvalue(systemW, lstrlenW(systemW), &name);
1862 hr = reader_add_attr(reader, NULL, &name, NULL, &sys, &position, 0);
1863 if (FAILED(hr)) return hr;
1864
1865 return S_OK;
1866 } else if (!reader_cmp(reader, systemW)) {
1867 /* system id */
1868 reader_skipn(reader, 6);
1869 cnt = reader_skipspaces(reader);
1870 if (!cnt) return WC_E_WHITESPACE;
1871
1872 hr = reader_parse_sys_literal(reader, &sys);
1873 if (FAILED(hr)) return hr;
1874
1875 reader_init_cstrvalue(systemW, lstrlenW(systemW), &name);
1876 return reader_add_attr(reader, NULL, &name, NULL, &sys, &position, 0);
1877 }
1878
1879 return S_FALSE;
1880 }
1881
1882 /* [28] doctypedecl ::= '<!DOCTYPE' S Name (S ExternalID)? S? ('[' intSubset ']' S?)? '>' */
1883 static HRESULT reader_parse_dtd(xmlreader *reader)
1884 {
1885 static const WCHAR doctypeW[] = {'<','!','D','O','C','T','Y','P','E',0};
1886 strval name;
1887 WCHAR *cur;
1888 HRESULT hr;
1889
1890 /* check if we have "<!DOCTYPE" */
1891 if (reader_cmp(reader, doctypeW)) return S_FALSE;
1892 reader_shrink(reader);
1893
1894 /* DTD processing is not allowed by default */
1895 if (reader->dtdmode == DtdProcessing_Prohibit) return WC_E_DTDPROHIBITED;
1896
1897 reader_skipn(reader, 9);
1898 if (!reader_skipspaces(reader)) return WC_E_WHITESPACE;
1899
1900 /* name */
1901 hr = reader_parse_name(reader, &name);
1902 if (FAILED(hr)) return WC_E_DECLDOCTYPE;
1903
1904 reader_skipspaces(reader);
1905
1906 hr = reader_parse_externalid(reader);
1907 if (FAILED(hr)) return hr;
1908
1909 reader_skipspaces(reader);
1910
1911 cur = reader_get_ptr(reader);
1912 if (*cur != '>')
1913 {
1914 FIXME("internal subset parsing not implemented\n");
1915 return E_NOTIMPL;
1916 }
1917
1918 /* skip '>' */
1919 reader_skipn(reader, 1);
1920
1921 reader->nodetype = XmlNodeType_DocumentType;
1922 reader_set_strvalue(reader, StringValue_LocalName, &name);
1923 reader_set_strvalue(reader, StringValue_QualifiedName, &name);
1924
1925 return S_OK;
1926 }
1927
1928 /* [11 NS] LocalPart ::= NCName */
1929 static HRESULT reader_parse_local(xmlreader *reader, strval *local, BOOL check_for_separator)
1930 {
1931 WCHAR *ptr;
1932 UINT start;
1933
1934 if (reader->resume[XmlReadResume_Local])
1935 {
1936 start = reader->resume[XmlReadResume_Local];
1937 ptr = reader_get_ptr(reader);
1938 }
1939 else
1940 {
1941 ptr = reader_get_ptr(reader);
1942 start = reader_get_cur(reader);
1943 }
1944
1945 while (is_ncnamechar(*ptr))
1946 {
1947 reader_skipn(reader, 1);
1948 ptr = reader_get_ptr(reader);
1949 }
1950
1951 if (check_for_separator && *ptr == ':')
1952 return NC_E_QNAMECOLON;
1953
1954 if (is_reader_pending(reader))
1955 {
1956 reader->resume[XmlReadResume_Local] = start;
1957 return E_PENDING;
1958 }
1959 else
1960 reader->resume[XmlReadResume_Local] = 0;
1961
1962 reader_init_strvalue(start, reader_get_cur(reader)-start, local);
1963
1964 return S_OK;
1965 }
1966
1967 /* [7 NS] QName ::= PrefixedName | UnprefixedName
1968 [8 NS] PrefixedName ::= Prefix ':' LocalPart
1969 [9 NS] UnprefixedName ::= LocalPart
1970 [10 NS] Prefix ::= NCName */
1971 static HRESULT reader_parse_qname(xmlreader *reader, strval *prefix, strval *local, strval *qname)
1972 {
1973 WCHAR *ptr;
1974 UINT start;
1975 HRESULT hr;
1976
1977 if (reader->resume[XmlReadResume_Name])
1978 {
1979 start = reader->resume[XmlReadResume_Name];
1980 ptr = reader_get_ptr(reader);
1981 }
1982 else
1983 {
1984 ptr = reader_get_ptr(reader);
1985 start = reader_get_cur(reader);
1986 reader->resume[XmlReadResume_Name] = start;
1987 if (!is_ncnamechar(*ptr)) return NC_E_QNAMECHARACTER;
1988 }
1989
1990 if (reader->resume[XmlReadResume_Local])
1991 {
1992 hr = reader_parse_local(reader, local, FALSE);
1993 if (FAILED(hr)) return hr;
1994
1995 reader_init_strvalue(reader->resume[XmlReadResume_Name],
1996 local->start - reader->resume[XmlReadResume_Name] - 1,
1997 prefix);
1998 }
1999 else
2000 {
2001 /* skip prefix part */
2002 while (is_ncnamechar(*ptr))
2003 {
2004 reader_skipn(reader, 1);
2005 ptr = reader_get_ptr(reader);
2006 }
2007
2008 if (is_reader_pending(reader)) return E_PENDING;
2009
2010 /* got a qualified name */
2011 if (*ptr == ':')
2012 {
2013 reader_init_strvalue(start, reader_get_cur(reader)-start, prefix);
2014
2015 /* skip ':' */
2016 reader_skipn(reader, 1);
2017 hr = reader_parse_local(reader, local, TRUE);
2018 if (FAILED(hr)) return hr;
2019 }
2020 else
2021 {
2022 reader_init_strvalue(reader->resume[XmlReadResume_Name], reader_get_cur(reader)-reader->resume[XmlReadResume_Name], local);
2023 reader_init_strvalue(0, 0, prefix);
2024 }
2025 }
2026
2027 if (prefix->len)
2028 TRACE("qname %s:%s\n", debug_strval(reader, prefix), debug_strval(reader, local));
2029 else
2030 TRACE("ncname %s\n", debug_strval(reader, local));
2031
2032 reader_init_strvalue(prefix->len ? prefix->start : local->start,
2033 /* count ':' too */
2034 (prefix->len ? prefix->len + 1 : 0) + local->len,
2035 qname);
2036
2037 reader->resume[XmlReadResume_Name] = 0;
2038 reader->resume[XmlReadResume_Local] = 0;
2039
2040 return S_OK;
2041 }
2042
2043 static WCHAR get_predefined_entity(const xmlreader *reader, const strval *name)
2044 {
2045 static const WCHAR entltW[] = {'l','t'};
2046 static const WCHAR entgtW[] = {'g','t'};
2047 static const WCHAR entampW[] = {'a','m','p'};
2048 static const WCHAR entaposW[] = {'a','p','o','s'};
2049 static const WCHAR entquotW[] = {'q','u','o','t'};
2050 static const strval lt = { (WCHAR*)entltW, 2 };
2051 static const strval gt = { (WCHAR*)entgtW, 2 };
2052 static const strval amp = { (WCHAR*)entampW, 3 };
2053 static const strval apos = { (WCHAR*)entaposW, 4 };
2054 static const strval quot = { (WCHAR*)entquotW, 4 };
2055 WCHAR *str = reader_get_strptr(reader, name);
2056
2057 switch (*str)
2058 {
2059 case 'l':
2060 if (strval_eq(reader, name, &lt)) return '<';
2061 break;
2062 case 'g':
2063 if (strval_eq(reader, name, &gt)) return '>';
2064 break;
2065 case 'a':
2066 if (strval_eq(reader, name, &amp))
2067 return '&';
2068 else if (strval_eq(reader, name, &apos))
2069 return '\'';
2070 break;
2071 case 'q':
2072 if (strval_eq(reader, name, &quot)) return '\"';
2073 break;
2074 default:
2075 ;
2076 }
2077
2078 return 0;
2079 }
2080
2081 /* [66] CharRef ::= '&#' [0-9]+ ';' | '&#x' [0-9a-fA-F]+ ';'
2082 [67] Reference ::= EntityRef | CharRef
2083 [68] EntityRef ::= '&' Name ';' */
2084 static HRESULT reader_parse_reference(xmlreader *reader)
2085 {
2086 encoded_buffer *buffer = &reader->input->buffer->utf16;
2087 WCHAR *start = reader_get_ptr(reader), *ptr;
2088 UINT cur = reader_get_cur(reader);
2089 WCHAR ch = 0;
2090 int len;
2091
2092 /* skip '&' */
2093 reader_skipn(reader, 1);
2094 ptr = reader_get_ptr(reader);
2095
2096 if (*ptr == '#')
2097 {
2098 reader_skipn(reader, 1);
2099 ptr = reader_get_ptr(reader);
2100
2101 /* hex char or decimal */
2102 if (*ptr == 'x')
2103 {
2104 reader_skipn(reader, 1);
2105 ptr = reader_get_ptr(reader);
2106
2107 while (*ptr != ';')
2108 {
2109 if ((*ptr >= '0' && *ptr <= '9'))
2110 ch = ch*16 + *ptr - '0';
2111 else if ((*ptr >= 'a' && *ptr <= 'f'))
2112 ch = ch*16 + *ptr - 'a' + 10;
2113 else if ((*ptr >= 'A' && *ptr <= 'F'))
2114 ch = ch*16 + *ptr - 'A' + 10;
2115 else
2116 return ch ? WC_E_SEMICOLON : WC_E_HEXDIGIT;
2117 reader_skipn(reader, 1);
2118 ptr = reader_get_ptr(reader);
2119 }
2120 }
2121 else
2122 {
2123 while (*ptr != ';')
2124 {
2125 if ((*ptr >= '0' && *ptr <= '9'))
2126 {
2127 ch = ch*10 + *ptr - '0';
2128 reader_skipn(reader, 1);
2129 ptr = reader_get_ptr(reader);
2130 }
2131 else
2132 return ch ? WC_E_SEMICOLON : WC_E_DIGIT;
2133 }
2134 }
2135
2136 if (!is_char(ch)) return WC_E_XMLCHARACTER;
2137
2138 /* normalize */
2139 if (is_wchar_space(ch)) ch = ' ';
2140
2141 ptr = reader_get_ptr(reader);
2142 start = reader_get_ptr2(reader, cur);
2143 len = buffer->written - ((char *)ptr - buffer->data);
2144 memmove(start + 1, ptr + 1, len);
2145
2146 buffer->written -= (reader_get_cur(reader) - cur) * sizeof(WCHAR);
2147 buffer->cur = cur + 1;
2148
2149 *start = ch;
2150 }
2151 else
2152 {
2153 strval name;
2154 HRESULT hr;
2155
2156 hr = reader_parse_name(reader, &name);
2157 if (FAILED(hr)) return hr;
2158
2159 ptr = reader_get_ptr(reader);
2160 if (*ptr != ';') return WC_E_SEMICOLON;
2161
2162 /* predefined entities resolve to a single character */
2163 ch = get_predefined_entity(reader, &name);
2164 if (ch)
2165 {
2166 len = buffer->written - ((char*)ptr - buffer->data) - sizeof(WCHAR);
2167 memmove(start+1, ptr+1, len);
2168 buffer->cur = cur + 1;
2169 buffer->written -= (ptr - start) * sizeof(WCHAR);
2170
2171 *start = ch;
2172 }
2173 else
2174 {
2175 FIXME("undeclared entity %s\n", debug_strval(reader, &name));
2176 return WC_E_UNDECLAREDENTITY;
2177 }
2178
2179 }
2180
2181 return S_OK;
2182 }
2183
2184 /* [10 NS] AttValue ::= '"' ([^<&"] | Reference)* '"' | "'" ([^<&'] | Reference)* "'" */
2185 static HRESULT reader_parse_attvalue(xmlreader *reader, strval *value)
2186 {
2187 WCHAR *ptr, quote;
2188 UINT start;
2189
2190 ptr = reader_get_ptr(reader);
2191
2192 /* skip opening quote */
2193 quote = *ptr;
2194 if (quote != '\"' && quote != '\'') return WC_E_QUOTE;
2195 reader_skipn(reader, 1);
2196
2197 ptr = reader_get_ptr(reader);
2198 start = reader_get_cur(reader);
2199 while (*ptr)
2200 {
2201 if (*ptr == '<') return WC_E_LESSTHAN;
2202
2203 if (*ptr == quote)
2204 {
2205 reader_init_strvalue(start, reader_get_cur(reader)-start, value);
2206 /* skip closing quote */
2207 reader_skipn(reader, 1);
2208 return S_OK;
2209 }
2210
2211 if (*ptr == '&')
2212 {
2213 HRESULT hr = reader_parse_reference(reader);
2214 if (FAILED(hr)) return hr;
2215 }
2216 else
2217 {
2218 /* replace all whitespace chars with ' ' */
2219 if (is_wchar_space(*ptr)) *ptr = ' ';
2220 reader_skipn(reader, 1);
2221 }
2222 ptr = reader_get_ptr(reader);
2223 }
2224
2225 return WC_E_QUOTE;
2226 }
2227
2228 /* [1 NS] NSAttName ::= PrefixedAttName | DefaultAttName
2229 [2 NS] PrefixedAttName ::= 'xmlns:' NCName
2230 [3 NS] DefaultAttName ::= 'xmlns'
2231 [15 NS] Attribute ::= NSAttName Eq AttValue | QName Eq AttValue */
2232 static HRESULT reader_parse_attribute(xmlreader *reader)
2233 {
2234 struct reader_position position = reader->position;
2235 strval prefix, local, qname, value;
2236 enum attribute_flags flags = 0;
2237 HRESULT hr;
2238
2239 hr = reader_parse_qname(reader, &prefix, &local, &qname);
2240 if (FAILED(hr)) return hr;
2241
2242 if (strval_eq(reader, &prefix, &strval_xmlns))
2243 flags |= ATTRIBUTE_NS_DEFINITION;
2244
2245 if (strval_eq(reader, &qname, &strval_xmlns))
2246 flags |= ATTRIBUTE_DEFAULT_NS_DEFINITION;
2247
2248 hr = reader_parse_eq(reader);
2249 if (FAILED(hr)) return hr;
2250
2251 hr = reader_parse_attvalue(reader, &value);
2252 if (FAILED(hr)) return hr;
2253
2254 if (flags & (ATTRIBUTE_NS_DEFINITION | ATTRIBUTE_DEFAULT_NS_DEFINITION))
2255 reader_push_ns(reader, &local, &value, !!(flags & ATTRIBUTE_DEFAULT_NS_DEFINITION));
2256
2257 TRACE("%s=%s\n", debug_strval(reader, &local), debug_strval(reader, &value));
2258 return reader_add_attr(reader, &prefix, &local, &qname, &value, &position, flags);
2259 }
2260
2261 /* [12 NS] STag ::= '<' QName (S Attribute)* S? '>'
2262 [14 NS] EmptyElemTag ::= '<' QName (S Attribute)* S? '/>' */
2263 static HRESULT reader_parse_stag(xmlreader *reader, strval *prefix, strval *local, strval *qname)
2264 {
2265 struct reader_position position = reader->position;
2266 HRESULT hr;
2267
2268 hr = reader_parse_qname(reader, prefix, local, qname);
2269 if (FAILED(hr)) return hr;
2270
2271 for (;;)
2272 {
2273 static const WCHAR endW[] = {'/','>',0};
2274
2275 reader_skipspaces(reader);
2276
2277 /* empty element */
2278 if ((reader->is_empty_element = !reader_cmp(reader, endW)))
2279 {
2280 struct element *element = &reader->empty_element;
2281
2282 /* skip '/>' */
2283 reader_skipn(reader, 2);
2284
2285 reader_free_strvalued(reader, &element->qname);
2286 reader_free_strvalued(reader, &element->localname);
2287
2288 element->prefix = *prefix;
2289 reader_strvaldup(reader, qname, &element->qname);
2290 reader_strvaldup(reader, local, &element->localname);
2291 element->position = position;
2292 reader_mark_ns_nodes(reader, element);
2293 return S_OK;
2294 }
2295
2296 /* got a start tag */
2297 if (!reader_cmp(reader, gtW))
2298 {
2299 /* skip '>' */
2300 reader_skipn(reader, 1);
2301 return reader_push_element(reader, prefix, local, qname, &position);
2302 }
2303
2304 hr = reader_parse_attribute(reader);
2305 if (FAILED(hr)) return hr;
2306 }
2307
2308 return S_OK;
2309 }
2310
2311 /* [39] element ::= EmptyElemTag | STag content ETag */
2312 static HRESULT reader_parse_element(xmlreader *reader)
2313 {
2314 HRESULT hr;
2315
2316 switch (reader->resumestate)
2317 {
2318 case XmlReadResumeState_Initial:
2319 /* check if we are really on element */
2320 if (reader_cmp(reader, ltW)) return S_FALSE;
2321
2322 /* skip '<' */
2323 reader_skipn(reader, 1);
2324
2325 reader_shrink(reader);
2326 reader->resumestate = XmlReadResumeState_STag;
2327 case XmlReadResumeState_STag:
2328 {
2329 strval qname, prefix, local;
2330
2331 /* this handles empty elements too */
2332 hr = reader_parse_stag(reader, &prefix, &local, &qname);
2333 if (FAILED(hr)) return hr;
2334
2335 /* FIXME: need to check for defined namespace to reject invalid prefix */
2336
2337 /* if we got empty element and stack is empty go straight to Misc */
2338 if (reader->is_empty_element && list_empty(&reader->elements))
2339 reader->instate = XmlReadInState_MiscEnd;
2340 else
2341 reader->instate = XmlReadInState_Content;
2342
2343 reader->nodetype = XmlNodeType_Element;
2344 reader->resumestate = XmlReadResumeState_Initial;
2345 reader_set_strvalue(reader, StringValue_Prefix, &prefix);
2346 reader_set_strvalue(reader, StringValue_QualifiedName, &qname);
2347 reader_set_strvalue(reader, StringValue_Value, &strval_empty);
2348 break;
2349 }
2350 default:
2351 hr = E_FAIL;
2352 }
2353
2354 return hr;
2355 }
2356
2357 /* [13 NS] ETag ::= '</' QName S? '>' */
2358 static HRESULT reader_parse_endtag(xmlreader *reader)
2359 {
2360 struct reader_position position;
2361 strval prefix, local, qname;
2362 struct element *element;
2363 HRESULT hr;
2364
2365 /* skip '</' */
2366 reader_skipn(reader, 2);
2367
2368 position = reader->position;
2369 hr = reader_parse_qname(reader, &prefix, &local, &qname);
2370 if (FAILED(hr)) return hr;
2371
2372 reader_skipspaces(reader);
2373
2374 if (reader_cmp(reader, gtW)) return WC_E_GREATERTHAN;
2375
2376 /* skip '>' */
2377 reader_skipn(reader, 1);
2378
2379 /* Element stack should never be empty at this point, cause we shouldn't get to
2380 content parsing if it's empty. */
2381 element = LIST_ENTRY(list_head(&reader->elements), struct element, entry);
2382 if (!strval_eq(reader, &element->qname, &qname)) return WC_E_ELEMENTMATCH;
2383
2384 /* update position stored for start tag, we won't be using it */
2385 element->position = position;
2386
2387 reader->nodetype = XmlNodeType_EndElement;
2388 reader->is_empty_element = FALSE;
2389 reader_set_strvalue(reader, StringValue_Prefix, &prefix);
2390
2391 return S_OK;
2392 }
2393
2394 /* [18] CDSect ::= CDStart CData CDEnd
2395 [19] CDStart ::= '<![CDATA['
2396 [20] CData ::= (Char* - (Char* ']]>' Char*))
2397 [21] CDEnd ::= ']]>' */
2398 static HRESULT reader_parse_cdata(xmlreader *reader)
2399 {
2400 WCHAR *ptr;
2401 UINT start;
2402
2403 if (reader->resumestate == XmlReadResumeState_CDATA)
2404 {
2405 start = reader->resume[XmlReadResume_Body];
2406 ptr = reader_get_ptr(reader);
2407 }
2408 else
2409 {
2410 /* skip markup '<![CDATA[' */
2411 reader_skipn(reader, 9);
2412 reader_shrink(reader);
2413 ptr = reader_get_ptr(reader);
2414 start = reader_get_cur(reader);
2415 reader->nodetype = XmlNodeType_CDATA;
2416 reader->resume[XmlReadResume_Body] = start;
2417 reader->resumestate = XmlReadResumeState_CDATA;
2418 reader_set_strvalue(reader, StringValue_Value, NULL);
2419 }
2420
2421 while (*ptr)
2422 {
2423 if (*ptr == ']' && *(ptr+1) == ']' && *(ptr+2) == '>')
2424 {
2425 strval value;
2426
2427 reader_init_strvalue(start, reader_get_cur(reader)-start, &value);
2428
2429 /* skip ']]>' */
2430 reader_skipn(reader, 3);
2431 TRACE("%s\n", debug_strval(reader, &value));
2432
2433 reader_set_strvalue(reader, StringValue_Value, &value);
2434 reader->resume[XmlReadResume_Body] = 0;
2435 reader->resumestate = XmlReadResumeState_Initial;
2436 return S_OK;
2437 }
2438 else
2439 {
2440 reader_skipn(reader, 1);
2441 ptr = reader_get_ptr(reader);
2442 }
2443 }
2444
2445 return S_OK;
2446 }
2447
2448 /* [14] CharData ::= [^<&]* - ([^<&]* ']]>' [^<&]*) */
2449 static HRESULT reader_parse_chardata(xmlreader *reader)
2450 {
2451 struct reader_position position;
2452 WCHAR *ptr;
2453 UINT start;
2454
2455 if (reader->resumestate == XmlReadResumeState_CharData)
2456 {
2457 start = reader->resume[XmlReadResume_Body];
2458 ptr = reader_get_ptr(reader);
2459 }
2460 else
2461 {
2462 reader_shrink(reader);
2463 ptr = reader_get_ptr(reader);
2464 start = reader_get_cur(reader);
2465 /* There's no text */
2466 if (!*ptr || *ptr == '<') return S_OK;
2467 reader->nodetype = is_wchar_space(*ptr) ? XmlNodeType_Whitespace : XmlNodeType_Text;
2468 reader->resume[XmlReadResume_Body] = start;
2469 reader->resumestate = XmlReadResumeState_CharData;
2470 reader_set_strvalue(reader, StringValue_Value, NULL);
2471 }
2472
2473 position = reader->position;
2474 while (*ptr)
2475 {
2476 static const WCHAR ampW[] = {'&',0};
2477
2478 /* CDATA closing sequence ']]>' is not allowed */
2479 if (ptr[0] == ']' && ptr[1] == ']' && ptr[2] == '>')
2480 return WC_E_CDSECTEND;
2481
2482 /* Found next markup part */
2483 if (ptr[0] == '<')
2484 {
2485 strval value;
2486
2487 reader->empty_element.position = position;
2488 reader_init_strvalue(start, reader_get_cur(reader)-start, &value);
2489 reader_set_strvalue(reader, StringValue_Value, &value);
2490 reader->resume[XmlReadResume_Body] = 0;
2491 reader->resumestate = XmlReadResumeState_Initial;
2492 return S_OK;
2493 }
2494
2495 /* this covers a case when text has leading whitespace chars */
2496 if (!is_wchar_space(*ptr)) reader->nodetype = XmlNodeType_Text;
2497
2498 if (!reader_cmp(reader, ampW))
2499 reader_parse_reference(reader);
2500 else
2501 reader_skipn(reader, 1);
2502
2503 ptr = reader_get_ptr(reader);
2504 }
2505
2506 return S_OK;
2507 }
2508
2509 /* [43] content ::= CharData? ((element | Reference | CDSect | PI | Comment) CharData?)* */
2510 static HRESULT reader_parse_content(xmlreader *reader)
2511 {
2512 static const WCHAR cdstartW[] = {'<','!','[','C','D','A','T','A','[',0};
2513 static const WCHAR etagW[] = {'<','/',0};
2514
2515 if (reader->resumestate != XmlReadResumeState_Initial)
2516 {
2517 switch (reader->resumestate)
2518 {
2519 case XmlReadResumeState_CDATA:
2520 return reader_parse_cdata(reader);
2521 case XmlReadResumeState_Comment:
2522 return reader_parse_comment(reader);
2523 case XmlReadResumeState_PIBody:
2524 case XmlReadResumeState_PITarget:
2525 return reader_parse_pi(reader);
2526 case XmlReadResumeState_CharData:
2527 return reader_parse_chardata(reader);
2528 default:
2529 ERR("unknown resume state %d\n", reader->resumestate);
2530 }
2531 }
2532
2533 reader_shrink(reader);
2534
2535 /* handle end tag here, it indicates end of content as well */
2536 if (!reader_cmp(reader, etagW))
2537 return reader_parse_endtag(reader);
2538
2539 if (!reader_cmp(reader, commentW))
2540 return reader_parse_comment(reader);
2541
2542 if (!reader_cmp(reader, piW))
2543 return reader_parse_pi(reader);
2544
2545 if (!reader_cmp(reader, cdstartW))
2546 return reader_parse_cdata(reader);
2547
2548 if (!reader_cmp(reader, ltW))
2549 return reader_parse_element(reader);
2550
2551 /* what's left must be CharData */
2552 return reader_parse_chardata(reader);
2553 }
2554
2555 static HRESULT reader_parse_nextnode(xmlreader *reader)
2556 {
2557 XmlNodeType nodetype = reader_get_nodetype(reader);
2558 HRESULT hr;
2559
2560 if (!is_reader_pending(reader))
2561 {
2562 reader->chunk_read_off = 0;
2563 reader_clear_attrs(reader);
2564 }
2565
2566 /* When moving from EndElement or empty element, pop its own namespace definitions */
2567 switch (nodetype)
2568 {
2569 case XmlNodeType_Attribute:
2570 reader_dec_depth(reader);
2571 /* fallthrough */
2572 case XmlNodeType_Element:
2573 if (reader->is_empty_element)
2574 reader_pop_ns_nodes(reader, &reader->empty_element);
2575 else if (FAILED(hr = reader_inc_depth(reader)))
2576 return hr;
2577 break;
2578 case XmlNodeType_EndElement:
2579 reader_pop_element(reader);
2580 reader_dec_depth(reader);
2581 break;
2582 default:
2583 ;
2584 }
2585
2586 for (;;)
2587 {
2588 switch (reader->instate)
2589 {
2590 /* if it's a first call for a new input we need to detect stream encoding */
2591 case XmlReadInState_Initial:
2592 {
2593 xml_encoding enc;
2594
2595 hr = readerinput_growraw(reader->input);
2596 if (FAILED(hr)) return hr;
2597
2598 reader->position.line_number = 1;
2599 reader->position.line_position = 1;
2600
2601 /* try to detect encoding by BOM or data and set input code page */
2602 hr = readerinput_detectencoding(reader->input, &enc);
2603 TRACE("detected encoding %s, 0x%08x\n", enc == XmlEncoding_Unknown ? "(unknown)" :
2604 debugstr_w(xml_encoding_map[enc].name), hr);
2605 if (FAILED(hr)) return hr;
2606
2607 /* always switch first time cause we have to put something in */
2608 readerinput_switchencoding(reader->input, enc);
2609
2610 /* parse xml declaration */
2611 hr = reader_parse_xmldecl(reader);
2612 if (FAILED(hr)) return hr;
2613
2614 readerinput_shrinkraw(reader->input, -1);
2615 reader->instate = XmlReadInState_Misc_DTD;
2616 if (hr == S_OK) return hr;
2617 }
2618 break;
2619 case XmlReadInState_Misc_DTD:
2620 hr = reader_parse_misc(reader);
2621 if (FAILED(hr)) return hr;
2622
2623 if (hr == S_FALSE)
2624 reader->instate = XmlReadInState_DTD;
2625 else
2626 return hr;
2627 break;
2628 case XmlReadInState_DTD:
2629 hr = reader_parse_dtd(reader);
2630 if (FAILED(hr)) return hr;
2631
2632 if (hr == S_OK)
2633 {
2634 reader->instate = XmlReadInState_DTD_Misc;
2635 return hr;
2636 }
2637 else
2638 reader->instate = XmlReadInState_Element;
2639 break;
2640 case XmlReadInState_DTD_Misc:
2641 hr = reader_parse_misc(reader);
2642 if (FAILED(hr)) return hr;
2643
2644 if (hr == S_FALSE)
2645 reader->instate = XmlReadInState_Element;
2646 else
2647 return hr;
2648 break;
2649 case XmlReadInState_Element:
2650 return reader_parse_element(reader);
2651 case XmlReadInState_Content:
2652 return reader_parse_content(reader);
2653 case XmlReadInState_MiscEnd:
2654 hr = reader_parse_misc(reader);
2655 if (hr != S_FALSE) return hr;
2656
2657 if (*reader_get_ptr(reader))
2658 {
2659 WARN("found garbage in the end of XML\n");
2660 return WC_E_SYNTAX;
2661 }
2662
2663 reader->instate = XmlReadInState_Eof;
2664 reader->state = XmlReadState_EndOfFile;
2665 reader->nodetype = XmlNodeType_None;
2666 return hr;
2667 case XmlReadInState_Eof:
2668 return S_FALSE;
2669 default:
2670 FIXME("internal state %d not handled\n", reader->instate);
2671 return E_NOTIMPL;
2672 }
2673 }
2674
2675 return E_NOTIMPL;
2676 }
2677
2678 static HRESULT WINAPI xmlreader_QueryInterface(IXmlReader *iface, REFIID riid, void** ppvObject)
2679 {
2680 xmlreader *This = impl_from_IXmlReader(iface);
2681
2682 TRACE("(%p)->(%s %p)\n", This, debugstr_guid(riid), ppvObject);
2683
2684 if (IsEqualGUID(riid, &IID_IUnknown) ||
2685 IsEqualGUID(riid, &IID_IXmlReader))
2686 {
2687 *ppvObject = iface;
2688 }
2689 else
2690 {
2691 FIXME("interface %s not implemented\n", debugstr_guid(riid));
2692 *ppvObject = NULL;
2693 return E_NOINTERFACE;
2694 }
2695
2696 IXmlReader_AddRef(iface);
2697
2698 return S_OK;
2699 }
2700
2701 static ULONG WINAPI xmlreader_AddRef(IXmlReader *iface)
2702 {
2703 xmlreader *This = impl_from_IXmlReader(iface);
2704 ULONG ref = InterlockedIncrement(&This->ref);
2705 TRACE("(%p)->(%d)\n", This, ref);
2706 return ref;
2707 }
2708
2709 static void reader_clear_ns(xmlreader *reader)
2710 {
2711 struct ns *ns, *ns2;
2712
2713 LIST_FOR_EACH_ENTRY_SAFE(ns, ns2, &reader->ns, struct ns, entry) {
2714 list_remove(&ns->entry);
2715 reader_free_strvalued(reader, &ns->prefix);
2716 reader_free_strvalued(reader, &ns->uri);
2717 reader_free(reader, ns);
2718 }
2719
2720 LIST_FOR_EACH_ENTRY_SAFE(ns, ns2, &reader->nsdef, struct ns, entry) {
2721 list_remove(&ns->entry);
2722 reader_free_strvalued(reader, &ns->uri);
2723 reader_free(reader, ns);
2724 }
2725 }
2726
2727 static void reader_reset_parser(xmlreader *reader)
2728 {
2729 reader->position.line_number = 0;
2730 reader->position.line_position = 0;
2731
2732 reader_clear_elements(reader);
2733 reader_clear_attrs(reader);
2734 reader_clear_ns(reader);
2735 reader_free_strvalues(reader);
2736
2737 reader->depth = 0;
2738 reader->nodetype = XmlNodeType_None;
2739 reader->resumestate = XmlReadResumeState_Initial;
2740 memset(reader->resume, 0, sizeof(reader->resume));
2741 reader->is_empty_element = FALSE;
2742 }
2743
2744 static ULONG WINAPI xmlreader_Release(IXmlReader *iface)
2745 {
2746 xmlreader *This = impl_from_IXmlReader(iface);
2747 LONG ref = InterlockedDecrement(&This->ref);
2748
2749 TRACE("(%p)->(%d)\n", This, ref);
2750
2751 if (ref == 0)
2752 {
2753 IMalloc *imalloc = This->imalloc;
2754 reader_reset_parser(This);
2755 if (This->input) IUnknown_Release(&This->input->IXmlReaderInput_iface);
2756 if (This->resolver) IXmlResolver_Release(This->resolver);
2757 if (This->mlang) IUnknown_Release(This->mlang);
2758 reader_free(This, This);
2759 if (imalloc) IMalloc_Release(imalloc);
2760 }
2761
2762 return ref;
2763 }
2764
2765 static HRESULT WINAPI xmlreader_SetInput(IXmlReader* iface, IUnknown *input)
2766 {
2767 xmlreader *This = impl_from_IXmlReader(iface);
2768 IXmlReaderInput *readerinput;
2769 HRESULT hr;
2770
2771 TRACE("(%p)->(%p)\n", This, input);
2772
2773 if (This->input)
2774 {
2775 readerinput_release_stream(This->input);
2776 IUnknown_Release(&This->input->IXmlReaderInput_iface);
2777 This->input = NULL;
2778 }
2779
2780 reader_reset_parser(This);
2781
2782 /* just reset current input */
2783 if (!input)
2784 {
2785 This->state = XmlReadState_Initial;
2786 return S_OK;
2787 }
2788
2789 /* now try IXmlReaderInput, ISequentialStream, IStream */
2790 hr = IUnknown_QueryInterface(input, &IID_IXmlReaderInput, (void**)&readerinput);
2791 if (hr == S_OK)
2792 {
2793 if (readerinput->lpVtbl == &xmlreaderinputvtbl)
2794 This->input = impl_from_IXmlReaderInput(readerinput);
2795 else
2796 {
2797 ERR("got external IXmlReaderInput implementation: %p, vtbl=%p\n",
2798 readerinput, readerinput->lpVtbl);
2799 IUnknown_Release(readerinput);
2800 return E_FAIL;
2801
2802 }
2803 }
2804
2805 if (hr != S_OK || !readerinput)
2806 {
2807 /* create IXmlReaderInput basing on supplied interface */
2808 hr = CreateXmlReaderInputWithEncodingName(input,
2809 This->imalloc, NULL, FALSE, NULL, &readerinput);
2810 if (hr != S_OK) return hr;
2811 This->input = impl_from_IXmlReaderInput(readerinput);
2812 }
2813
2814 /* set stream for supplied IXmlReaderInput */
2815 hr = readerinput_query_for_stream(This->input);
2816 if (hr == S_OK)
2817 {
2818 This->state = XmlReadState_Initial;
2819 This->instate = XmlReadInState_Initial;
2820 }
2821 return hr;
2822 }
2823
2824 static HRESULT WINAPI xmlreader_GetProperty(IXmlReader* iface, UINT property, LONG_PTR *value)
2825 {
2826 xmlreader *This = impl_from_IXmlReader(iface);
2827
2828 TRACE("(%p)->(%s %p)\n", This, debugstr_reader_prop(property), value);
2829
2830 if (!value) return E_INVALIDARG;
2831
2832 switch (property)
2833 {
2834 case XmlReaderProperty_MultiLanguage:
2835 *value = (LONG_PTR)This->mlang;
2836 if (This->mlang)
2837 IUnknown_AddRef(This->mlang);
2838 break;
2839 case XmlReaderProperty_XmlResolver:
2840 *value = (LONG_PTR)This->resolver;
2841 if (This->resolver)
2842 IXmlResolver_AddRef(This->resolver);
2843 break;
2844 case XmlReaderProperty_DtdProcessing:
2845 *value = This->dtdmode;
2846 break;
2847 case XmlReaderProperty_ReadState:
2848 *value = This->state;
2849 break;
2850 case XmlReaderProperty_MaxElementDepth:
2851 *value = This->max_depth;
2852 break;
2853 default:
2854 FIXME("Unimplemented property (%u)\n", property);
2855 return E_NOTIMPL;
2856 }
2857
2858 return S_OK;
2859 }
2860
2861 static HRESULT WINAPI xmlreader_SetProperty(IXmlReader* iface, UINT property, LONG_PTR value)
2862 {
2863 xmlreader *This = impl_from_IXmlReader(iface);
2864
2865 TRACE("(%p)->(%s 0x%lx)\n", This, debugstr_reader_prop(property), value);
2866
2867 switch (property)
2868 {
2869 case XmlReaderProperty_MultiLanguage:
2870 if (This->mlang)
2871 IUnknown_Release(This->mlang);
2872 This->mlang = (IUnknown*)value;
2873 if (This->mlang)
2874 IUnknown_AddRef(This->mlang);
2875 if (This->mlang)
2876 FIXME("Ignoring MultiLanguage %p\n", This->mlang);
2877 break;
2878 case XmlReaderProperty_XmlResolver:
2879 if (This->resolver)
2880 IXmlResolver_Release(This->resolver);
2881 This->resolver = (IXmlResolver*)value;
2882 if (This->resolver)
2883 IXmlResolver_AddRef(This->resolver);
2884 break;
2885 case XmlReaderProperty_DtdProcessing:
2886 if (value < 0 || value > _DtdProcessing_Last) return E_INVALIDARG;
2887 This->dtdmode = value;
2888 break;
2889 case XmlReaderProperty_MaxElementDepth:
2890 This->max_depth = value;
2891 break;
2892 default:
2893 FIXME("Unimplemented property (%u)\n", property);
2894 return E_NOTIMPL;
2895 }
2896
2897 return S_OK;
2898 }
2899
2900 static HRESULT WINAPI xmlreader_Read(IXmlReader* iface, XmlNodeType *nodetype)
2901 {
2902 xmlreader *This = impl_from_IXmlReader(iface);
2903 XmlNodeType oldtype = This->nodetype;
2904 XmlNodeType type;
2905 HRESULT hr;
2906
2907 TRACE("(%p)->(%p)\n", This, nodetype);
2908
2909 if (!nodetype)
2910 nodetype = &type;
2911
2912 switch (This->state)
2913 {
2914 case XmlReadState_Closed:
2915 hr = S_FALSE;
2916 break;
2917 case XmlReadState_Error:
2918 hr = This->error;
2919 break;
2920 default:
2921 hr = reader_parse_nextnode(This);
2922 if (SUCCEEDED(hr) && oldtype == XmlNodeType_None && This->nodetype != oldtype)
2923 This->state = XmlReadState_Interactive;
2924
2925 if (FAILED(hr))
2926 {
2927 This->state = XmlReadState_Error;
2928 This->nodetype = XmlNodeType_None;
2929 This->depth = 0;
2930 This->error = hr;
2931 }
2932 }
2933
2934 TRACE("node type %s\n", debugstr_nodetype(This->nodetype));
2935 *nodetype = This->nodetype;
2936
2937 return hr;
2938 }
2939
2940 static HRESULT WINAPI xmlreader_GetNodeType(IXmlReader* iface, XmlNodeType *node_type)
2941 {
2942 xmlreader *This = impl_from_IXmlReader(iface);
2943
2944 TRACE("(%p)->(%p)\n", This, node_type);
2945
2946 if (!node_type)
2947 return E_INVALIDARG;
2948
2949 *node_type = reader_get_nodetype(This);
2950 return This->state == XmlReadState_Closed ? S_FALSE : S_OK;
2951 }
2952
2953 static void reader_set_current_attribute(xmlreader *reader, struct attribute *attr)
2954 {
2955 reader->attr = attr;
2956 reader->chunk_read_off = 0;
2957 reader_set_strvalue(reader, StringValue_Prefix, &attr->prefix);
2958 reader_set_strvalue(reader, StringValue_QualifiedName, &attr->qname);
2959 reader_set_strvalue(reader, StringValue_Value, &attr->value);
2960 }
2961
2962 static HRESULT reader_move_to_first_attribute(xmlreader *reader)
2963 {
2964 if (!reader->attr_count)
2965 return S_FALSE;
2966
2967 if (!reader->attr)
2968 reader_inc_depth(reader);
2969
2970 reader_set_current_attribute(reader, LIST_ENTRY(list_head(&reader->attrs), struct attribute, entry));
2971
2972 return S_OK;
2973 }
2974
2975 static HRESULT WINAPI xmlreader_MoveToFirstAttribute(IXmlReader* iface)
2976 {
2977 xmlreader *This = impl_from_IXmlReader(iface);
2978
2979 TRACE("(%p)\n", This);
2980
2981 return reader_move_to_first_attribute(This);
2982 }
2983
2984 static HRESULT WINAPI xmlreader_MoveToNextAttribute(IXmlReader* iface)
2985 {
2986 xmlreader *This = impl_from_IXmlReader(iface);
2987 const struct list *next;
2988
2989 TRACE("(%p)\n", This);
2990
2991 if (!This->attr_count) return S_FALSE;
2992
2993 if (!This->attr)
2994 return reader_move_to_first_attribute(This);
2995
2996 next = list_next(&This->attrs, &This->attr->entry);
2997 if (next)
2998 reader_set_current_attribute(This, LIST_ENTRY(next, struct attribute, entry));
2999
3000 return next ? S_OK : S_FALSE;
3001 }
3002
3003 static void reader_get_attribute_ns_uri(xmlreader *reader, struct attribute *attr, const WCHAR **uri, UINT *len)
3004 {
3005 static const WCHAR xmlns_uriW[] = {'h','t','t','p',':','/','/','w','w','w','.','w','3','.','o','r','g','/',
3006 '2','0','0','0','/','x','m','l','n','s','/',0};
3007 static const WCHAR xml_uriW[] = {'h','t','t','p',':','/','/','w','w','w','.','w','3','.','o','r','g','/',
3008 'X','M','L','/','1','9','9','8','/','n','a','m','e','s','p','a','c','e',0};
3009
3010 /* Check for reserved prefixes first */
3011 if ((strval_eq(reader, &attr->prefix, &strval_empty) && strval_eq(reader, &attr->localname, &strval_xmlns)) ||
3012 strval_eq(reader, &attr->prefix, &strval_xmlns))
3013 {
3014 *uri = xmlns_uriW;
3015 *len = ARRAY_SIZE(xmlns_uriW) - 1;
3016 }
3017 else if (strval_eq(reader, &attr->prefix, &strval_xml))
3018 {
3019 *uri = xml_uriW;
3020 *len = ARRAY_SIZE(xml_uriW) - 1;
3021 }
3022 else
3023 {
3024 *uri = NULL;
3025 *len = 0;
3026 }
3027
3028 if (!*uri)
3029 {
3030 struct ns *ns;
3031
3032 if ((ns = reader_lookup_ns(reader, &attr->prefix)))
3033 {
3034 *uri = ns->uri.str;
3035 *len = ns->uri.len;
3036 }
3037 else
3038 {
3039 *uri = emptyW;
3040 *len = 0;
3041 }
3042 }
3043 }
3044
3045 static void reader_get_attribute_local_name(xmlreader *reader, struct attribute *attr, const WCHAR **name, UINT *len)
3046 {
3047 if (attr->flags & ATTRIBUTE_DEFAULT_NS_DEFINITION)
3048 {
3049 *name = xmlnsW;
3050 *len = 5;
3051 }
3052 else if (attr->flags & ATTRIBUTE_NS_DEFINITION)
3053 {
3054 const struct ns *ns = reader_lookup_ns(reader, &attr->localname);
3055 *name = ns->prefix.str;
3056 *len = ns->prefix.len;
3057 }
3058 else
3059 {
3060 *name = attr->localname.str;
3061 *len = attr->localname.len;
3062 }
3063 }
3064
3065 static HRESULT WINAPI xmlreader_MoveToAttributeByName(IXmlReader* iface,
3066 const WCHAR *local_name, const WCHAR *namespace_uri)
3067 {
3068 xmlreader *This = impl_from_IXmlReader(iface);
3069 UINT target_name_len, target_uri_len;
3070 struct attribute *attr;
3071
3072 TRACE("(%p)->(%s %s)\n", This, debugstr_w(local_name), debugstr_w(namespace_uri));
3073
3074 if (!local_name)
3075 return E_INVALIDARG;
3076
3077 if (!This->attr_count)
3078 return S_FALSE;
3079
3080 if (!namespace_uri)
3081 namespace_uri = emptyW;
3082
3083 target_name_len = lstrlenW(local_name);
3084 target_uri_len = lstrlenW(namespace_uri);
3085
3086 LIST_FOR_EACH_ENTRY(attr, &This->attrs, struct attribute, entry)
3087 {
3088 UINT name_len, uri_len;
3089 const WCHAR *name, *uri;
3090
3091 reader_get_attribute_local_name(This, attr, &name, &name_len);
3092 reader_get_attribute_ns_uri(This, attr, &uri, &uri_len);
3093
3094 if (name_len == target_name_len && uri_len == target_uri_len &&
3095 !wcscmp(name, local_name) && !wcscmp(uri, namespace_uri))
3096 {
3097 reader_set_current_attribute(This, attr);
3098 return S_OK;
3099 }
3100 }
3101
3102 return S_FALSE;
3103 }
3104
3105 static HRESULT WINAPI xmlreader_MoveToElement(IXmlReader* iface)
3106 {
3107 xmlreader *This = impl_from_IXmlReader(iface);
3108
3109 TRACE("(%p)\n", This);
3110
3111 if (!This->attr_count) return S_FALSE;
3112
3113 if (This->attr)
3114 reader_dec_depth(This);
3115
3116 This->attr = NULL;
3117
3118 /* FIXME: support other node types with 'attributes' like DTD */
3119 if (This->is_empty_element) {
3120 reader_set_strvalue(This, StringValue_Prefix, &This->empty_element.prefix);
3121 reader_set_strvalue(This, StringValue_QualifiedName, &This->empty_element.qname);
3122 }
3123 else {
3124 struct element *element = LIST_ENTRY(list_head(&This->elements), struct element, entry);
3125 if (element) {
3126 reader_set_strvalue(This, StringValue_Prefix, &element->prefix);
3127 reader_set_strvalue(This, StringValue_QualifiedName, &element->qname);
3128 }
3129 }
3130 This->chunk_read_off = 0;
3131 reader_set_strvalue(This, StringValue_Value, &strval_empty);
3132
3133 return S_OK;
3134 }
3135
3136 static HRESULT WINAPI xmlreader_GetQualifiedName(IXmlReader* iface, LPCWSTR *name, UINT *len)
3137 {
3138 xmlreader *This = impl_from_IXmlReader(iface);
3139 struct attribute *attribute = This->attr;
3140 struct element *element;
3141 UINT length;
3142
3143 TRACE("(%p)->(%p %p)\n", This, name, len);
3144
3145 if (!len)
3146 len = &length;
3147
3148 switch (reader_get_nodetype(This))
3149 {
3150 case XmlNodeType_Text:
3151 case XmlNodeType_CDATA:
3152 case XmlNodeType_Comment:
3153 case XmlNodeType_Whitespace:
3154 *name = emptyW;
3155 *len = 0;
3156 break;
3157 case XmlNodeType_Element:
3158 case XmlNodeType_EndElement:
3159 element = reader_get_element(This);
3160 if (element->prefix.len)
3161 {
3162 *name = element->qname.str;
3163 *len = element->qname.len;
3164 }
3165 else
3166 {
3167 *name = element->localname.str;
3168 *len = element->localname.len;
3169 }
3170 break;
3171 case XmlNodeType_Attribute:
3172 if (attribute->flags & ATTRIBUTE_DEFAULT_NS_DEFINITION)
3173 {
3174 *name = xmlnsW;
3175 *len = 5;
3176 } else if (attribute->prefix.len)
3177 {
3178 *name = This->strvalues[StringValue_QualifiedName].str;
3179 *len = This->strvalues[StringValue_QualifiedName].len;
3180 }
3181 else
3182 {
3183 *name = attribute->localname.str;
3184 *len = attribute->localname.len;
3185 }
3186 break;
3187 default:
3188 *name = This->strvalues[StringValue_QualifiedName].str;
3189 *len = This->strvalues[StringValue_QualifiedName].len;
3190 break;
3191 }
3192
3193 return S_OK;
3194 }
3195
3196 static struct ns *reader_lookup_nsdef(xmlreader *reader)
3197 {
3198 if (list_empty(&reader->nsdef))
3199 return NULL;
3200
3201 return LIST_ENTRY(list_head(&reader->nsdef), struct ns, entry);
3202 }
3203
3204 static HRESULT WINAPI xmlreader_GetNamespaceUri(IXmlReader* iface, const WCHAR **uri, UINT *len)
3205 {
3206 xmlreader *This = impl_from_IXmlReader(iface);
3207 const strval *prefix = &This->strvalues[StringValue_Prefix];
3208 XmlNodeType nodetype;
3209 struct ns *ns;
3210 UINT length;
3211
3212 TRACE("(%p %p %p)\n", iface, uri, len);
3213
3214 if (!len)
3215 len = &length;
3216
3217 switch ((nodetype = reader_get_nodetype(This)))
3218 {
3219 case XmlNodeType_Attribute:
3220 reader_get_attribute_ns_uri(This, This->attr, uri, len);
3221 break;
3222 case XmlNodeType_Element:
3223 case XmlNodeType_EndElement:
3224 {
3225 ns = reader_lookup_ns(This, prefix);
3226
3227 /* pick top default ns if any */
3228 if (!ns)
3229 ns = reader_lookup_nsdef(This);
3230
3231 if (ns) {
3232 *uri = ns->uri.str;
3233 *len = ns->uri.len;
3234 }
3235 else {
3236 *uri = emptyW;
3237 *len = 0;
3238 }
3239 }
3240 break;
3241 case XmlNodeType_Text:
3242 case XmlNodeType_CDATA:
3243 case XmlNodeType_ProcessingInstruction:
3244 case XmlNodeType_Comment:
3245 case XmlNodeType_Whitespace:
3246 case XmlNodeType_XmlDeclaration:
3247 *uri = emptyW;
3248 *len = 0;
3249 break;
3250 default:
3251 FIXME("Unhandled node type %d\n", nodetype);
3252 *uri = NULL;
3253 *len = 0;
3254 return E_NOTIMPL;
3255 }
3256
3257 return S_OK;
3258 }
3259
3260 static HRESULT WINAPI xmlreader_GetLocalName(IXmlReader* iface, LPCWSTR *name, UINT *len)
3261 {
3262 xmlreader *This = impl_from_IXmlReader(iface);
3263 struct element *element;
3264 UINT length;
3265
3266 TRACE("(%p)->(%p %p)\n", This, name, len);
3267
3268 if (!len)
3269 len = &length;
3270
3271 switch (reader_get_nodetype(This))
3272 {
3273 case XmlNodeType_Text:
3274 case XmlNodeType_CDATA:
3275 case XmlNodeType_Comment:
3276 case XmlNodeType_Whitespace:
3277 *name = emptyW;
3278 *len = 0;
3279 break;
3280 case XmlNodeType_Element:
3281 case XmlNodeType_EndElement:
3282 element = reader_get_element(This);
3283 *name = element->localname.str;
3284 *len = element->localname.len;
3285 break;
3286 case XmlNodeType_Attribute:
3287 reader_get_attribute_local_name(This, This->attr, name, len);
3288 break;
3289 default:
3290 *name = This->strvalues[StringValue_LocalName].str;
3291 *len = This->strvalues[StringValue_LocalName].len;
3292 break;
3293 }
3294
3295 return S_OK;
3296 }
3297
3298 static HRESULT WINAPI xmlreader_GetPrefix(IXmlReader* iface, const WCHAR **ret, UINT *len)
3299 {
3300 xmlreader *This = impl_from_IXmlReader(iface);
3301 XmlNodeType nodetype;
3302 UINT length;
3303
3304 TRACE("(%p)->(%p %p)\n", This, ret, len);
3305
3306 if (!len)
3307 len = &length;
3308
3309 *ret = emptyW;
3310 *len = 0;
3311
3312 switch ((nodetype = reader_get_nodetype(This)))
3313 {
3314 case XmlNodeType_Element:
3315 case XmlNodeType_EndElement:
3316 case XmlNodeType_Attribute:
3317 {
3318 const strval *prefix = &This->strvalues[StringValue_Prefix];
3319 struct ns *ns;
3320
3321 if (strval_eq(This, prefix, &strval_xml))
3322 {
3323 *ret = xmlW;
3324 *len = 3;
3325 }
3326 else if (strval_eq(This, prefix, &strval_xmlns))
3327 {
3328 *ret = xmlnsW;
3329 *len = 5;
3330 }
3331 else if ((ns = reader_lookup_ns(This, prefix)))
3332 {
3333 *ret = ns->prefix.str;
3334 *len = ns->prefix.len;
3335 }
3336
3337 break;
3338 }
3339 default:
3340 ;
3341 }
3342
3343 return S_OK;
3344 }
3345
3346 static const strval *reader_get_value(xmlreader *reader, BOOL ensure_allocated)
3347 {
3348 strval *val;
3349
3350 switch (reader_get_nodetype(reader))
3351 {
3352 case XmlNodeType_XmlDeclaration:
3353 case XmlNodeType_EndElement:
3354 case XmlNodeType_None:
3355 return &strval_empty;
3356 case XmlNodeType_Attribute:
3357 /* For namespace definition attributes return values from namespace list */
3358 if (reader->attr->flags & (ATTRIBUTE_NS_DEFINITION | ATTRIBUTE_DEFAULT_NS_DEFINITION))
3359 {
3360 struct ns *ns;
3361
3362 if (!(ns = reader_lookup_ns(reader, &reader->attr->localname)))
3363 ns = reader_lookup_nsdef(reader);
3364
3365 return &ns->uri;
3366 }
3367 return &reader->attr->value;
3368 default:
3369 break;
3370 }
3371
3372 val = &reader->strvalues[StringValue_Value];
3373 if (!val->str && ensure_allocated)
3374 {
3375 WCHAR *ptr = reader_alloc(reader, (val->len+1)*sizeof(WCHAR));
3376 if (!ptr) return NULL;
3377 memcpy(ptr, reader_get_strptr(reader, val), val->len*sizeof(WCHAR));
3378 ptr[val->len] = 0;
3379 val->str = ptr;
3380 }
3381
3382 return val;
3383 }
3384
3385 static HRESULT WINAPI xmlreader_GetValue(IXmlReader* iface, const WCHAR **value, UINT *len)
3386 {
3387 xmlreader *reader = impl_from_IXmlReader(iface);
3388 const strval *val = &reader->strvalues[StringValue_Value];
3389 UINT off;
3390
3391 TRACE("(%p)->(%p %p)\n", reader, value, len);
3392
3393 *value = NULL;
3394
3395 if ((reader->nodetype == XmlNodeType_Comment && !val->str && !val->len) || is_reader_pending(reader))
3396 {
3397 XmlNodeType type;
3398 HRESULT hr;
3399
3400 hr = IXmlReader_Read(iface, &type);
3401 if (FAILED(hr)) return hr;
3402
3403 /* return if still pending, partially read values are not reported */
3404 if (is_reader_pending(reader)) return E_PENDING;
3405 }
3406
3407 val = reader_get_value(reader, TRUE);
3408 if (!val)
3409 return E_OUTOFMEMORY;
3410
3411 off = abs(reader->chunk_read_off);
3412 assert(off <= val->len);
3413 *value = val->str + off;
3414 if (len) *len = val->len - off;
3415 reader->chunk_read_off = -off;
3416 return S_OK;
3417 }
3418
3419 static HRESULT WINAPI xmlreader_ReadValueChunk(IXmlReader* iface, WCHAR *buffer, UINT chunk_size, UINT *read)
3420 {
3421 xmlreader *reader = impl_from_IXmlReader(iface);
3422 const strval *val;
3423 UINT len = 0;
3424
3425 TRACE("(%p)->(%p %u %p)\n", reader, buffer, chunk_size, read);
3426
3427 val = reader_get_value(reader, FALSE);
3428
3429 /* If value is already read by GetValue, chunk_read_off is negative and chunked reads are not possible. */
3430 if (reader->chunk_read_off >= 0)
3431 {
3432 assert(reader->chunk_read_off <= val->len);
3433 len = min(val->len - reader->chunk_read_off, chunk_size);
3434 }
3435 if (read) *read = len;
3436
3437 if (len)
3438 {
3439 memcpy(buffer, reader_get_strptr(reader, val) + reader->chunk_read_off, len*sizeof(WCHAR));
3440 reader->chunk_read_off += len;
3441 }
3442
3443 return len || !chunk_size ? S_OK : S_FALSE;
3444 }
3445
3446 static HRESULT WINAPI xmlreader_GetBaseUri(IXmlReader* iface,
3447 LPCWSTR *baseUri,
3448 UINT *baseUri_length)
3449 {
3450 FIXME("(%p %p %p): stub\n", iface, baseUri, baseUri_length);
3451 return E_NOTIMPL;
3452 }
3453
3454 static BOOL WINAPI xmlreader_IsDefault(IXmlReader* iface)
3455 {
3456 FIXME("(%p): stub\n", iface);
3457 return FALSE;
3458 }
3459
3460 static BOOL WINAPI xmlreader_IsEmptyElement(IXmlReader* iface)
3461 {
3462 xmlreader *This = impl_from_IXmlReader(iface);
3463 TRACE("(%p)\n", This);
3464 /* Empty elements are not placed in stack, it's stored as a global reader flag that makes sense
3465 when current node is start tag of an element */
3466 return (reader_get_nodetype(This) == XmlNodeType_Element) ? This->is_empty_element : FALSE;
3467 }
3468
3469 static HRESULT WINAPI xmlreader_GetLineNumber(IXmlReader* iface, UINT *line_number)
3470 {
3471 xmlreader *This = impl_from_IXmlReader(iface);
3472 const struct element *element;
3473
3474 TRACE("(%p %p)\n", This, line_number);
3475
3476 if (!line_number)
3477 return E_INVALIDARG;
3478
3479 switch (reader_get_nodetype(This))
3480 {
3481 case XmlNodeType_Element:
3482 case XmlNodeType_EndElement:
3483 element = reader_get_element(This);
3484 *line_number = element->position.line_number;
3485 break;
3486 case XmlNodeType_Attribute:
3487 *line_number = This->attr->position.line_number;
3488 break;
3489 case XmlNodeType_Whitespace:
3490 case XmlNodeType_XmlDeclaration:
3491 *line_number = This->empty_element.position.line_number;
3492 break;
3493 default:
3494 *line_number = This->position.line_number;
3495 break;
3496 }
3497
3498 return This->state == XmlReadState_Closed ? S_FALSE : S_OK;
3499 }
3500
3501 static HRESULT WINAPI xmlreader_GetLinePosition(IXmlReader* iface, UINT *line_position)
3502 {
3503 xmlreader *This = impl_from_IXmlReader(iface);
3504 const struct element *element;
3505
3506 TRACE("(%p %p)\n", This, line_position);
3507
3508 if (!line_position)
3509 return E_INVALIDARG;
3510
3511 switch (reader_get_nodetype(This))
3512 {
3513 case XmlNodeType_Element:
3514 case XmlNodeType_EndElement:
3515 element = reader_get_element(This);
3516 *line_position = element->position.line_position;
3517 break;
3518 case XmlNodeType_Attribute:
3519 *line_position = This->attr->position.line_position;
3520 break;
3521 case XmlNodeType_Whitespace:
3522 case XmlNodeType_XmlDeclaration:
3523 *line_position = This->empty_element.position.line_position;
3524 break;
3525 default:
3526 *line_position = This->position.line_position;
3527 break;
3528 }
3529
3530 return This->state == XmlReadState_Closed ? S_FALSE : S_OK;
3531 }
3532
3533 static HRESULT WINAPI xmlreader_GetAttributeCount(IXmlReader* iface, UINT *count)
3534 {
3535 xmlreader *This = impl_from_IXmlReader(iface);
3536
3537 TRACE("(%p)->(%p)\n", This, count);
3538
3539 if (!count) return E_INVALIDARG;
3540
3541 *count = This->attr_count;
3542 return S_OK;
3543 }
3544
3545 static HRESULT WINAPI xmlreader_GetDepth(IXmlReader* iface, UINT *depth)
3546 {
3547 xmlreader *This = impl_from_IXmlReader(iface);
3548 TRACE("(%p)->(%p)\n", This, depth);
3549 *depth = This->depth;
3550 return S_OK;
3551 }
3552
3553 static BOOL WINAPI xmlreader_IsEOF(IXmlReader* iface)
3554 {
3555 xmlreader *This = impl_from_IXmlReader(iface);
3556 TRACE("(%p)\n", iface);
3557 return This->state == XmlReadState_EndOfFile;
3558 }
3559
3560 static const struct IXmlReaderVtbl xmlreader_vtbl =
3561 {
3562 xmlreader_QueryInterface,
3563 xmlreader_AddRef,
3564 xmlreader_Release,
3565 xmlreader_SetInput,
3566 xmlreader_GetProperty,
3567 xmlreader_SetProperty,
3568 xmlreader_Read,
3569 xmlreader_GetNodeType,
3570 xmlreader_MoveToFirstAttribute,
3571 xmlreader_MoveToNextAttribute,
3572 xmlreader_MoveToAttributeByName,
3573 xmlreader_MoveToElement,
3574 xmlreader_GetQualifiedName,
3575 xmlreader_GetNamespaceUri,
3576 xmlreader_GetLocalName,
3577 xmlreader_GetPrefix,
3578 xmlreader_GetValue,
3579 xmlreader_ReadValueChunk,
3580 xmlreader_GetBaseUri,
3581 xmlreader_IsDefault,
3582 xmlreader_IsEmptyElement,
3583 xmlreader_GetLineNumber,
3584 xmlreader_GetLinePosition,
3585 xmlreader_GetAttributeCount,
3586 xmlreader_GetDepth,
3587 xmlreader_IsEOF
3588 };
3589
3590 /** IXmlReaderInput **/
3591 static HRESULT WINAPI xmlreaderinput_QueryInterface(IXmlReaderInput *iface, REFIID riid, void** ppvObject)
3592 {
3593 xmlreaderinput *This = impl_from_IXmlReaderInput(iface);
3594
3595 TRACE("(%p)->(%s %p)\n", This, debugstr_guid(riid), ppvObject);
3596
3597 if (IsEqualGUID(riid, &IID_IXmlReaderInput) ||
3598 IsEqualGUID(riid, &IID_IUnknown))
3599 {
3600 *ppvObject = iface;
3601 }
3602 else
3603 {
3604 WARN("interface %s not implemented\n", debugstr_guid(riid));
3605 *ppvObject = NULL;
3606 return E_NOINTERFACE;
3607 }
3608
3609 IUnknown_AddRef(iface);
3610
3611 return S_OK;
3612 }
3613
3614 static ULONG WINAPI xmlreaderinput_AddRef(IXmlReaderInput *iface)
3615 {
3616 xmlreaderinput *This = impl_from_IXmlReaderInput(iface);
3617 ULONG ref = InterlockedIncrement(&This->ref);
3618 TRACE("(%p)->(%d)\n", This, ref);
3619 return ref;
3620 }
3621
3622 static ULONG WINAPI xmlreaderinput_Release(IXmlReaderInput *iface)
3623 {
3624 xmlreaderinput *This = impl_from_IXmlReaderInput(iface);
3625 LONG ref = InterlockedDecrement(&This->ref);
3626
3627 TRACE("(%p)->(%d)\n", This, ref);
3628
3629 if (ref == 0)
3630 {
3631 IMalloc *imalloc = This->imalloc;
3632 if (This->input) IUnknown_Release(This->input);
3633 if (This->stream) ISequentialStream_Release(This->stream);
3634 if (This->buffer) free_input_buffer(This->buffer);
3635 readerinput_free(This, This->baseuri);
3636 readerinput_free(This, This);
3637 if (imalloc) IMalloc_Release(imalloc);
3638 }
3639
3640 return ref;
3641 }
3642
3643 static const struct IUnknownVtbl xmlreaderinputvtbl =
3644 {
3645 xmlreaderinput_QueryInterface,
3646 xmlreaderinput_AddRef,
3647 xmlreaderinput_Release
3648 };
3649
3650 HRESULT WINAPI CreateXmlReader(REFIID riid, void **obj, IMalloc *imalloc)
3651 {
3652 xmlreader *reader;
3653 HRESULT hr;
3654 int i;
3655
3656 TRACE("(%s, %p, %p)\n", wine_dbgstr_guid(riid), obj, imalloc);
3657
3658 if (imalloc)
3659 reader = IMalloc_Alloc(imalloc, sizeof(*reader));
3660 else
3661 reader = heap_alloc(sizeof(*reader));
3662 if (!reader)
3663 return E_OUTOFMEMORY;
3664
3665 memset(reader, 0, sizeof(*reader));
3666 reader->IXmlReader_iface.lpVtbl = &xmlreader_vtbl;
3667 reader->ref = 1;
3668 reader->state = XmlReadState_Closed;
3669 reader->instate = XmlReadInState_Initial;
3670 reader->resumestate = XmlReadResumeState_Initial;
3671 reader->dtdmode = DtdProcessing_Prohibit;
3672 reader->imalloc = imalloc;
3673 if (imalloc) IMalloc_AddRef(imalloc);
3674 reader->nodetype = XmlNodeType_None;
3675 list_init(&reader->attrs);
3676 list_init(&reader->nsdef);
3677 list_init(&reader->ns);
3678 list_init(&reader->elements);
3679 reader->max_depth = 256;
3680
3681 reader->chunk_read_off = 0;
3682 for (i = 0; i < StringValue_Last; i++)
3683 reader->strvalues[i] = strval_empty;
3684
3685 hr = IXmlReader_QueryInterface(&reader->IXmlReader_iface, riid, obj);
3686 IXmlReader_Release(&reader->IXmlReader_iface);
3687
3688 TRACE("returning iface %p, hr %#x\n", *obj, hr);
3689
3690 return hr;
3691 }
3692
3693 HRESULT WINAPI CreateXmlReaderInputWithEncodingName(IUnknown *stream,
3694 IMalloc *imalloc,
3695 LPCWSTR encoding,
3696 BOOL hint,
3697 LPCWSTR base_uri,
3698 IXmlReaderInput **ppInput)
3699 {
3700 xmlreaderinput *readerinput;
3701 HRESULT hr;
3702
3703 TRACE("%p %p %s %d %s %p\n", stream, imalloc, wine_dbgstr_w(encoding),
3704 hint, wine_dbgstr_w(base_uri), ppInput);
3705
3706 if (!stream || !ppInput) return E_INVALIDARG;
3707
3708 if (imalloc)
3709 readerinput = IMalloc_Alloc(imalloc, sizeof(*readerinput));
3710 else
3711 readerinput = heap_alloc(sizeof(*readerinput));
3712 if(!readerinput) return E_OUTOFMEMORY;
3713
3714 readerinput->IXmlReaderInput_iface.lpVtbl = &xmlreaderinputvtbl;
3715 readerinput->ref = 1;
3716 readerinput->imalloc = imalloc;
3717 readerinput->stream = NULL;
3718 if (imalloc) IMalloc_AddRef(imalloc);
3719 readerinput->encoding = parse_encoding_name(encoding, -1);
3720 readerinput->hint = hint;
3721 readerinput->baseuri = readerinput_strdupW(readerinput, base_uri);
3722 readerinput->pending = 0;
3723
3724 hr = alloc_input_buffer(readerinput);
3725 if (hr != S_OK)
3726 {
3727 readerinput_free(readerinput, readerinput->baseuri);
3728 readerinput_free(readerinput, readerinput);
3729 if (imalloc) IMalloc_Release(imalloc);
3730 return hr;
3731 }
3732 IUnknown_QueryInterface(stream, &IID_IUnknown, (void**)&readerinput->input);
3733
3734 *ppInput = &readerinput->IXmlReaderInput_iface;
3735
3736 TRACE("returning iface %p\n", *ppInput);
3737
3738 return S_OK;
3739 }
Zebediah Figura's local tree