1 /*
2 * Copyright (C) 2015 The Android Open Source Project
3 *
4 * Licensed under the Apache License, Version 2.0 (the "License");
5 * you may not use this file except in compliance with the License.
6 * You may obtain a copy of the License at
7 *
8 * http://www.apache.org/licenses/LICENSE-2.0
9 *
10 * Unless required by applicable law or agreed to in writing, software
11 * distributed under the License is distributed on an "AS IS" BASIS,
12 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
13 * See the License for the specific language governing permissions and
14 * limitations under the License.
15 */
16
17 #include "XmlDom.h"
18
19 #include <expat.h>
20
21 #include <memory>
22 #include <stack>
23 #include <string>
24 #include <tuple>
25
26 #include "android-base/logging.h"
27
28 #include "ResourceUtils.h"
29 #include "trace/TraceBuffer.h"
30 #include "XmlPullParser.h"
31 #include "util/Util.h"
32
33 using ::android::InputStream;
34 using ::android::StringPiece;
35 using ::android::StringPiece16;
36
37 namespace aapt {
38 namespace xml {
39
40 constexpr char kXmlNamespaceSep = 1;
41
42 struct Stack {
43 std::unique_ptr<xml::Element> root;
44 std::stack<xml::Element*> node_stack;
45 std::unique_ptr<xml::Element> pending_element;
46 std::string pending_comment;
47 std::unique_ptr<xml::Text> last_text_node;
48 };
49
50 // Extracts the namespace and name of an expanded element or attribute name.
SplitName(const char * name,std::string * out_ns,std::string * out_name)51 static void SplitName(const char* name, std::string* out_ns, std::string* out_name) {
52 const char* p = name;
53 while (*p != 0 && *p != kXmlNamespaceSep) {
54 p++;
55 }
56
57 if (*p == 0) {
58 out_ns->clear();
59 out_name->assign(name);
60 } else {
61 out_ns->assign(name, (p - name));
62 out_name->assign(p + 1);
63 }
64 }
65
FinishPendingText(Stack * stack)66 static void FinishPendingText(Stack* stack) {
67 if (stack->last_text_node != nullptr) {
68 if (!stack->last_text_node->text.empty()) {
69 CHECK(!stack->node_stack.empty());
70 stack->node_stack.top()->AppendChild(std::move(stack->last_text_node));
71 } else {
72 // Drop an empty text node.
73 }
74 stack->last_text_node = nullptr;
75 }
76 }
77
StartNamespaceHandler(void * user_data,const char * prefix,const char * uri)78 static void XMLCALL StartNamespaceHandler(void* user_data, const char* prefix, const char* uri) {
79 XML_Parser parser = reinterpret_cast<XML_Parser>(user_data);
80 Stack* stack = reinterpret_cast<Stack*>(XML_GetUserData(parser));
81 FinishPendingText(stack);
82
83 NamespaceDecl decl;
84 decl.line_number = XML_GetCurrentLineNumber(parser);
85 decl.column_number = XML_GetCurrentColumnNumber(parser);
86 decl.prefix = prefix ? prefix : "";
87 decl.uri = uri ? uri : "";
88
89 if (stack->pending_element == nullptr) {
90 stack->pending_element = util::make_unique<Element>();
91 }
92 stack->pending_element->namespace_decls.push_back(std::move(decl));
93 }
94
EndNamespaceHandler(void * user_data,const char *)95 static void XMLCALL EndNamespaceHandler(void* user_data, const char* /*prefix*/) {
96 XML_Parser parser = reinterpret_cast<XML_Parser>(user_data);
97 Stack* stack = reinterpret_cast<Stack*>(XML_GetUserData(parser));
98 FinishPendingText(stack);
99 }
100
less_attribute(const Attribute & lhs,const Attribute & rhs)101 static bool less_attribute(const Attribute& lhs, const Attribute& rhs) {
102 return std::tie(lhs.namespace_uri, lhs.name, lhs.value) <
103 std::tie(rhs.namespace_uri, rhs.name, rhs.value);
104 }
105
StartElementHandler(void * user_data,const char * name,const char ** attrs)106 static void XMLCALL StartElementHandler(void* user_data, const char* name, const char** attrs) {
107 XML_Parser parser = reinterpret_cast<XML_Parser>(user_data);
108 Stack* stack = reinterpret_cast<Stack*>(XML_GetUserData(parser));
109 FinishPendingText(stack);
110
111 std::unique_ptr<Element> el;
112 if (stack->pending_element != nullptr) {
113 el = std::move(stack->pending_element);
114 } else {
115 el = util::make_unique<Element>();
116 }
117
118 el->line_number = XML_GetCurrentLineNumber(parser);
119 el->column_number = XML_GetCurrentColumnNumber(parser);
120 el->comment = std::move(stack->pending_comment);
121
122 SplitName(name, &el->namespace_uri, &el->name);
123
124 while (*attrs) {
125 Attribute attribute;
126 SplitName(*attrs++, &attribute.namespace_uri, &attribute.name);
127 attribute.value = *attrs++;
128 el->attributes.push_back(std::move(attribute));
129 }
130
131 // Sort the attributes.
132 std::sort(el->attributes.begin(), el->attributes.end(), less_attribute);
133
134 // Add to the stack.
135 Element* this_el = el.get();
136 if (!stack->node_stack.empty()) {
137 stack->node_stack.top()->AppendChild(std::move(el));
138 } else {
139 stack->root = std::move(el);
140 }
141 stack->node_stack.push(this_el);
142 }
143
EndElementHandler(void * user_data,const char * name)144 static void XMLCALL EndElementHandler(void* user_data, const char* name) {
145 XML_Parser parser = reinterpret_cast<XML_Parser>(user_data);
146 Stack* stack = reinterpret_cast<Stack*>(XML_GetUserData(parser));
147 FinishPendingText(stack);
148
149 CHECK(!stack->node_stack.empty());
150 // stack->nodeStack.top()->comment = std::move(stack->pendingComment);
151 stack->node_stack.pop();
152 }
153
CharacterDataHandler(void * user_data,const char * s,int len)154 static void XMLCALL CharacterDataHandler(void* user_data, const char* s, int len) {
155 XML_Parser parser = reinterpret_cast<XML_Parser>(user_data);
156 Stack* stack = reinterpret_cast<Stack*>(XML_GetUserData(parser));
157
158 const StringPiece str(s, len);
159 if (str.empty()) {
160 return;
161 }
162
163 // See if we can just append the text to a previous text node.
164 if (stack->last_text_node != nullptr) {
165 stack->last_text_node->text.append(str.data(), str.size());
166 return;
167 }
168
169 stack->last_text_node = util::make_unique<Text>();
170 stack->last_text_node->line_number = XML_GetCurrentLineNumber(parser);
171 stack->last_text_node->column_number = XML_GetCurrentColumnNumber(parser);
172 stack->last_text_node->text.assign(str);
173 }
174
CommentDataHandler(void * user_data,const char * comment)175 static void XMLCALL CommentDataHandler(void* user_data, const char* comment) {
176 XML_Parser parser = reinterpret_cast<XML_Parser>(user_data);
177 Stack* stack = reinterpret_cast<Stack*>(XML_GetUserData(parser));
178 FinishPendingText(stack);
179
180 if (!stack->pending_comment.empty()) {
181 stack->pending_comment += '\n';
182 }
183 stack->pending_comment += comment;
184 }
185
Inflate(InputStream * in,android::IDiagnostics * diag,const android::Source & source)186 std::unique_ptr<XmlResource> Inflate(InputStream* in, android::IDiagnostics* diag,
187 const android::Source& source) {
188 Stack stack;
189
190 std::unique_ptr<std::remove_pointer<XML_Parser>::type, decltype(XML_ParserFree)*> parser = {
191 XML_ParserCreateNS(nullptr, kXmlNamespaceSep), XML_ParserFree};
192 XML_SetUserData(parser.get(), &stack);
193 XML_UseParserAsHandlerArg(parser.get());
194 XML_SetElementHandler(parser.get(), StartElementHandler, EndElementHandler);
195 XML_SetNamespaceDeclHandler(parser.get(), StartNamespaceHandler, EndNamespaceHandler);
196 XML_SetCharacterDataHandler(parser.get(), CharacterDataHandler);
197 XML_SetCommentHandler(parser.get(), CommentDataHandler);
198
199 const char* buffer = nullptr;
200 size_t buffer_size = 0;
201 while (in->Next(reinterpret_cast<const void**>(&buffer), &buffer_size)) {
202 if (XML_Parse(parser.get(), buffer, buffer_size, false) == XML_STATUS_ERROR) {
203 diag->Error(android::DiagMessage(source.WithLine(XML_GetCurrentLineNumber(parser.get())))
204 << XML_ErrorString(XML_GetErrorCode(parser.get())));
205 return {};
206 }
207 }
208
209 if (in->HadError()) {
210 diag->Error(android::DiagMessage(source) << in->GetError());
211 return {};
212 } else {
213 // Finish off the parsing.
214 if (XML_Parse(parser.get(), nullptr, 0u, true) == XML_STATUS_ERROR) {
215 diag->Error(android::DiagMessage(source.WithLine(XML_GetCurrentLineNumber(parser.get())))
216 << XML_ErrorString(XML_GetErrorCode(parser.get())));
217 return {};
218 }
219 }
220 return util::make_unique<XmlResource>(ResourceFile{{}, {}, ResourceFile::Type::kUnknown, source},
221 android::StringPool{}, std::move(stack.root));
222 }
223
CopyAttributes(Element * el,android::ResXMLParser * parser,android::StringPool * out_pool)224 static void CopyAttributes(Element* el, android::ResXMLParser* parser,
225 android::StringPool* out_pool) {
226 const size_t attr_count = parser->getAttributeCount();
227 if (attr_count > 0) {
228 el->attributes.reserve(attr_count);
229 for (size_t i = 0; i < attr_count; i++) {
230 Attribute attr;
231 size_t len;
232 const char16_t* str16 = parser->getAttributeNamespace(i, &len);
233 if (str16) {
234 attr.namespace_uri = android::util::Utf16ToUtf8(StringPiece16(str16, len));
235 }
236
237 str16 = parser->getAttributeName(i, &len);
238 if (str16) {
239 attr.name = android::util::Utf16ToUtf8(StringPiece16(str16, len));
240 }
241
242 uint32_t res_id = parser->getAttributeNameResID(i);
243 if (res_id > 0) {
244 attr.compiled_attribute = AaptAttribute(::aapt::Attribute(), {res_id});
245 }
246
247 str16 = parser->getAttributeStringValue(i, &len);
248 if (str16) {
249 attr.value = android::util::Utf16ToUtf8(StringPiece16(str16, len));
250 }
251
252 android::Res_value res_value;
253 if (parser->getAttributeValue(i, &res_value) > 0) {
254 // Only compile the value if it is not a string, or it is a string that differs from
255 // the raw attribute value.
256 int32_t raw_value_idx = parser->getAttributeValueStringID(i);
257 if (res_value.dataType != android::Res_value::TYPE_STRING || raw_value_idx < 0 ||
258 static_cast<uint32_t>(raw_value_idx) != res_value.data) {
259 attr.compiled_value = ResourceUtils::ParseBinaryResValue(
260 ResourceType::kAnim, {}, parser->getStrings(), res_value, out_pool);
261 }
262 }
263
264 el->attributes.push_back(std::move(attr));
265 }
266 }
267 }
268
Inflate(const void * data,size_t len,std::string * out_error)269 std::unique_ptr<XmlResource> Inflate(const void* data, size_t len, std::string* out_error) {
270 TRACE_CALL();
271 // We import the android namespace because on Windows NO_ERROR is a macro, not
272 // an enum, which causes errors when qualifying it with android::
273 using namespace android;
274
275 std::unique_ptr<XmlResource> xml_resource = util::make_unique<XmlResource>();
276
277 std::stack<Element*> node_stack;
278 std::unique_ptr<Element> pending_element;
279
280 ResXMLTree tree;
281 if (tree.setTo(data, len) != NO_ERROR) {
282 if (out_error != nullptr) {
283 *out_error = "failed to initialize ResXMLTree";
284 }
285 return {};
286 }
287
288 ResXMLParser::event_code_t code;
289 while ((code = tree.next()) != ResXMLParser::BAD_DOCUMENT && code != ResXMLParser::END_DOCUMENT) {
290 std::unique_ptr<Node> new_node;
291 switch (code) {
292 case ResXMLParser::START_NAMESPACE: {
293 NamespaceDecl decl;
294 decl.line_number = tree.getLineNumber();
295
296 size_t len;
297 const char16_t* str16 = tree.getNamespacePrefix(&len);
298 if (str16) {
299 decl.prefix = android::util::Utf16ToUtf8(StringPiece16(str16, len));
300 }
301
302 str16 = tree.getNamespaceUri(&len);
303 if (str16) {
304 decl.uri = android::util::Utf16ToUtf8(StringPiece16(str16, len));
305 }
306
307 if (pending_element == nullptr) {
308 pending_element = util::make_unique<Element>();
309 }
310 // pending_element is not nullptr
311 // NOLINTNEXTLINE(bugprone-use-after-move)
312 pending_element->namespace_decls.push_back(std::move(decl));
313 break;
314 }
315
316 case ResXMLParser::START_TAG: {
317 std::unique_ptr<Element> el;
318 if (pending_element != nullptr) {
319 el = std::move(pending_element);
320 } else {
321 el = util::make_unique<Element>();
322 }
323 el->line_number = tree.getLineNumber();
324
325 size_t len;
326 const char16_t* str16 = tree.getElementNamespace(&len);
327 if (str16) {
328 el->namespace_uri = android::util::Utf16ToUtf8(StringPiece16(str16, len));
329 }
330
331 str16 = tree.getElementName(&len);
332 if (str16) {
333 el->name = android::util::Utf16ToUtf8(StringPiece16(str16, len));
334 }
335
336 Element* this_el = el.get();
337 CopyAttributes(el.get(), &tree, &xml_resource->string_pool);
338
339 if (!node_stack.empty()) {
340 node_stack.top()->AppendChild(std::move(el));
341 } else {
342 xml_resource->root = std::move(el);
343 }
344 node_stack.push(this_el);
345 break;
346 }
347
348 case ResXMLParser::TEXT: {
349 std::unique_ptr<Text> text = util::make_unique<Text>();
350 text->line_number = tree.getLineNumber();
351 size_t len;
352 const char16_t* str16 = tree.getText(&len);
353 if (str16) {
354 text->text = android::util::Utf16ToUtf8(StringPiece16(str16, len));
355 }
356 CHECK(!node_stack.empty());
357 node_stack.top()->AppendChild(std::move(text));
358 break;
359 }
360
361 case ResXMLParser::END_NAMESPACE:
362 break;
363
364 case ResXMLParser::END_TAG:
365 CHECK(!node_stack.empty());
366 node_stack.pop();
367 break;
368
369 default:
370 LOG(FATAL) << "unhandled XML chunk type";
371 break;
372 }
373 }
374 return xml_resource;
375 }
376
Clone() const377 std::unique_ptr<XmlResource> XmlResource::Clone() const {
378 std::unique_ptr<XmlResource> cloned = util::make_unique<XmlResource>(file);
379 CloningValueTransformer cloner(&cloned->string_pool);
380 if (root != nullptr) {
381 cloned->root = root->CloneElement([&](const xml::Element& src, xml::Element* dst) {
382 dst->attributes.reserve(src.attributes.size());
383 for (const xml::Attribute& attr : src.attributes) {
384 xml::Attribute cloned_attr;
385 cloned_attr.name = attr.name;
386 cloned_attr.namespace_uri = attr.namespace_uri;
387 cloned_attr.value = attr.value;
388 cloned_attr.compiled_attribute = attr.compiled_attribute;
389 if (attr.compiled_value != nullptr) {
390 cloned_attr.compiled_value = attr.compiled_value->Transform(cloner);
391 }
392 dst->attributes.push_back(std::move(cloned_attr));
393 }
394 });
395 }
396 return cloned;
397 }
398
FindRootElement(Node * node)399 Element* FindRootElement(Node* node) {
400 if (node == nullptr) {
401 return nullptr;
402 }
403
404 while (node->parent != nullptr) {
405 node = node->parent;
406 }
407 return NodeCast<Element>(node);
408 }
409
AppendChild(std::unique_ptr<Node> child)410 void Element::AppendChild(std::unique_ptr<Node> child) {
411 child->parent = this;
412 children.push_back(std::move(child));
413 }
414
InsertChild(size_t index,std::unique_ptr<Node> child)415 void Element::InsertChild(size_t index, std::unique_ptr<Node> child) {
416 child->parent = this;
417 children.insert(children.begin() + index, std::move(child));
418 }
419
FindAttribute(StringPiece ns,StringPiece name)420 Attribute* Element::FindAttribute(StringPiece ns, StringPiece name) {
421 return const_cast<Attribute*>(static_cast<const Element*>(this)->FindAttribute(ns, name));
422 }
423
FindAttribute(StringPiece ns,StringPiece name) const424 const Attribute* Element::FindAttribute(StringPiece ns, StringPiece name) const {
425 for (const auto& attr : attributes) {
426 if (ns == attr.namespace_uri && name == attr.name) {
427 return &attr;
428 }
429 }
430 return nullptr;
431 }
432
RemoveAttribute(StringPiece ns,StringPiece name)433 void Element::RemoveAttribute(StringPiece ns, StringPiece name) {
434 auto new_attr_end = std::remove_if(attributes.begin(), attributes.end(),
435 [&](const Attribute& attr) -> bool {
436 return ns == attr.namespace_uri && name == attr.name;
437 });
438
439 attributes.erase(new_attr_end, attributes.end());
440 }
441
FindOrCreateAttribute(StringPiece ns,StringPiece name)442 Attribute* Element::FindOrCreateAttribute(StringPiece ns, StringPiece name) {
443 Attribute* attr = FindAttribute(ns, name);
444 if (attr == nullptr) {
445 attributes.push_back(Attribute{std::string(ns), std::string(name)});
446 attr = &attributes.back();
447 }
448 return attr;
449 }
450
FindChild(StringPiece ns,StringPiece name)451 Element* Element::FindChild(StringPiece ns, StringPiece name) {
452 return FindChildWithAttribute(ns, name, {}, {}, {});
453 }
454
FindChild(StringPiece ns,StringPiece name) const455 const Element* Element::FindChild(StringPiece ns, StringPiece name) const {
456 return FindChildWithAttribute(ns, name, {}, {}, {});
457 }
458
FindChildWithAttribute(StringPiece ns,StringPiece name,StringPiece attr_ns,StringPiece attr_name,StringPiece attr_value)459 Element* Element::FindChildWithAttribute(StringPiece ns, StringPiece name, StringPiece attr_ns,
460 StringPiece attr_name, StringPiece attr_value) {
461 return const_cast<Element*>(static_cast<const Element*>(this)->FindChildWithAttribute(
462 ns, name, attr_ns, attr_name, attr_value));
463 }
464
FindChildWithAttribute(StringPiece ns,StringPiece name,StringPiece attr_ns,StringPiece attr_name,StringPiece attr_value) const465 const Element* Element::FindChildWithAttribute(StringPiece ns, StringPiece name,
466 StringPiece attr_ns, StringPiece attr_name,
467 StringPiece attr_value) const {
468 for (const auto& child : children) {
469 if (const Element* el = NodeCast<Element>(child.get())) {
470 if (ns == el->namespace_uri && name == el->name) {
471 if (attr_ns.empty() && attr_name.empty()) {
472 return el;
473 }
474
475 const Attribute* attr = el->FindAttribute(attr_ns, attr_name);
476 if (attr && attr_value == attr->value) {
477 return el;
478 }
479 }
480 }
481 }
482 return nullptr;
483 }
484
GetChildElements()485 std::vector<Element*> Element::GetChildElements() {
486 std::vector<Element*> elements;
487 for (auto& child_node : children) {
488 if (Element* child = NodeCast<Element>(child_node.get())) {
489 elements.push_back(child);
490 }
491 }
492 return elements;
493 }
494
Clone(const ElementCloneFunc & el_cloner) const495 std::unique_ptr<Node> Element::Clone(const ElementCloneFunc& el_cloner) const {
496 auto el = util::make_unique<Element>();
497 el->namespace_decls = namespace_decls;
498 el->comment = comment;
499 el->line_number = line_number;
500 el->column_number = column_number;
501 el->name = name;
502 el->namespace_uri = namespace_uri;
503 el->attributes.reserve(attributes.size());
504 el_cloner(*this, el.get());
505 el->children.reserve(children.size());
506 for (const std::unique_ptr<xml::Node>& child : children) {
507 el->AppendChild(child->Clone(el_cloner));
508 }
509 return std::move(el);
510 }
511
CloneElement(const ElementCloneFunc & el_cloner) const512 std::unique_ptr<Element> Element::CloneElement(const ElementCloneFunc& el_cloner) const {
513 return std::unique_ptr<Element>(static_cast<Element*>(Clone(el_cloner).release()));
514 }
515
Accept(Visitor * visitor)516 void Element::Accept(Visitor* visitor) {
517 visitor->BeforeVisitElement(this);
518 visitor->Visit(this);
519 visitor->AfterVisitElement(this);
520 }
521
Accept(ConstVisitor * visitor) const522 void Element::Accept(ConstVisitor* visitor) const {
523 visitor->BeforeVisitElement(this);
524 visitor->Visit(this);
525 visitor->AfterVisitElement(this);
526 }
527
Clone(const ElementCloneFunc &) const528 std::unique_ptr<Node> Text::Clone(const ElementCloneFunc&) const {
529 auto t = util::make_unique<Text>();
530 t->comment = comment;
531 t->line_number = line_number;
532 t->column_number = column_number;
533 t->text = text;
534 return std::move(t);
535 }
536
Accept(Visitor * visitor)537 void Text::Accept(Visitor* visitor) {
538 visitor->Visit(this);
539 }
540
Accept(ConstVisitor * visitor) const541 void Text::Accept(ConstVisitor* visitor) const {
542 visitor->Visit(this);
543 }
544
BeforeVisitElement(Element * el)545 void PackageAwareVisitor::BeforeVisitElement(Element* el) {
546 std::vector<PackageDecl> decls;
547 for (const NamespaceDecl& decl : el->namespace_decls) {
548 if (std::optional<ExtractedPackage> maybe_package = ExtractPackageFromNamespace(decl.uri)) {
549 decls.push_back(PackageDecl{decl.prefix, std::move(maybe_package.value())});
550 }
551 }
552 package_decls_.push_back(std::move(decls));
553 }
554
AfterVisitElement(Element * el)555 void PackageAwareVisitor::AfterVisitElement(Element* el) {
556 package_decls_.pop_back();
557 }
558
TransformPackageAlias(StringPiece alias) const559 std::optional<ExtractedPackage> PackageAwareVisitor::TransformPackageAlias(
560 StringPiece alias) const {
561 if (alias.empty()) {
562 return ExtractedPackage{{}, false /*private*/};
563 }
564
565 const auto rend = package_decls_.rend();
566 for (auto iter = package_decls_.rbegin(); iter != rend; ++iter) {
567 const std::vector<PackageDecl>& decls = *iter;
568 const auto rend2 = decls.rend();
569 for (auto iter2 = decls.rbegin(); iter2 != rend2; ++iter2) {
570 const PackageDecl& decl = *iter2;
571 if (alias == decl.prefix) {
572 if (decl.package.package.empty()) {
573 return ExtractedPackage{{}, decl.package.private_namespace};
574 }
575 return decl.package;
576 }
577 }
578 }
579 return {};
580 }
581
582 } // namespace xml
583 } // namespace aapt
584