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 <iostream>
18 #include <string>
19 
20 #include "Maybe.h"
21 #include "SourceXmlPullParser.h"
22 #include "Util.h"
23 
24 namespace aapt {
25 
26 constexpr char kXmlNamespaceSep = 1;
27 
SourceXmlPullParser(std::istream & in)28 SourceXmlPullParser::SourceXmlPullParser(std::istream& in) : mIn(in), mEmpty(), mDepth(0) {
29     mParser = XML_ParserCreateNS(nullptr, kXmlNamespaceSep);
30     XML_SetUserData(mParser, this);
31     XML_SetElementHandler(mParser, startElementHandler, endElementHandler);
32     XML_SetNamespaceDeclHandler(mParser, startNamespaceHandler, endNamespaceHandler);
33     XML_SetCharacterDataHandler(mParser, characterDataHandler);
34     XML_SetCommentHandler(mParser, commentDataHandler);
35     mEventQueue.push(EventData{ Event::kStartDocument, 0, mDepth++ });
36 }
37 
~SourceXmlPullParser()38 SourceXmlPullParser::~SourceXmlPullParser() {
39     XML_ParserFree(mParser);
40 }
41 
next()42 SourceXmlPullParser::Event SourceXmlPullParser::next() {
43     const Event currentEvent = getEvent();
44     if (currentEvent == Event::kBadDocument || currentEvent == Event::kEndDocument) {
45         return currentEvent;
46     }
47 
48     mEventQueue.pop();
49     while (mEventQueue.empty()) {
50         mIn.read(mBuffer, sizeof(mBuffer) / sizeof(*mBuffer));
51 
52         const bool done = mIn.eof();
53         if (mIn.bad() && !done) {
54             mLastError = strerror(errno);
55             mEventQueue.push(EventData{ Event::kBadDocument });
56             continue;
57         }
58 
59         if (XML_Parse(mParser, mBuffer, mIn.gcount(), done) == XML_STATUS_ERROR) {
60             mLastError = XML_ErrorString(XML_GetErrorCode(mParser));
61             mEventQueue.push(EventData{ Event::kBadDocument });
62             continue;
63         }
64 
65         if (done) {
66             mEventQueue.push(EventData{ Event::kEndDocument, 0, 0 });
67         }
68     }
69 
70     Event event = getEvent();
71 
72     // Record namespace prefixes and package names so that we can do our own
73     // handling of references that use namespace aliases.
74     if (event == Event::kStartNamespace || event == Event::kEndNamespace) {
75         Maybe<std::u16string> result = util::extractPackageFromNamespace(getNamespaceUri());
76         if (event == Event::kStartNamespace) {
77             if (result) {
78                 mPackageAliases.emplace_back(getNamespacePrefix(), result.value());
79             }
80         } else {
81             if (result) {
82                 assert(mPackageAliases.back().second == result.value());
83                 mPackageAliases.pop_back();
84             }
85         }
86     }
87 
88     return event;
89 }
90 
getEvent() const91 SourceXmlPullParser::Event SourceXmlPullParser::getEvent() const {
92     return mEventQueue.front().event;
93 }
94 
getLastError() const95 const std::string& SourceXmlPullParser::getLastError() const {
96     return mLastError;
97 }
98 
getComment() const99 const std::u16string& SourceXmlPullParser::getComment() const {
100     return mEventQueue.front().comment;
101 }
102 
getLineNumber() const103 size_t SourceXmlPullParser::getLineNumber() const {
104     return mEventQueue.front().lineNumber;
105 }
106 
getDepth() const107 size_t SourceXmlPullParser::getDepth() const {
108     return mEventQueue.front().depth;
109 }
110 
getText() const111 const std::u16string& SourceXmlPullParser::getText() const {
112     if (getEvent() != Event::kText) {
113         return mEmpty;
114     }
115     return mEventQueue.front().data1;
116 }
117 
getNamespacePrefix() const118 const std::u16string& SourceXmlPullParser::getNamespacePrefix() const {
119     const Event currentEvent = getEvent();
120     if (currentEvent != Event::kStartNamespace && currentEvent != Event::kEndNamespace) {
121         return mEmpty;
122     }
123     return mEventQueue.front().data1;
124 }
125 
getNamespaceUri() const126 const std::u16string& SourceXmlPullParser::getNamespaceUri() const {
127     const Event currentEvent = getEvent();
128     if (currentEvent != Event::kStartNamespace && currentEvent != Event::kEndNamespace) {
129         return mEmpty;
130     }
131     return mEventQueue.front().data2;
132 }
133 
applyPackageAlias(std::u16string * package,const std::u16string & defaultPackage) const134 bool SourceXmlPullParser::applyPackageAlias(std::u16string* package,
135                                             const std::u16string& defaultPackage) const {
136     const auto endIter = mPackageAliases.rend();
137     for (auto iter = mPackageAliases.rbegin(); iter != endIter; ++iter) {
138         if (iter->first == *package) {
139             if (iter->second.empty()) {
140                 *package = defaultPackage;
141             } else {
142                 *package = iter->second;
143             }
144             return true;
145         }
146     }
147     return false;
148 }
149 
getElementNamespace() const150 const std::u16string& SourceXmlPullParser::getElementNamespace() const {
151     const Event currentEvent = getEvent();
152     if (currentEvent != Event::kStartElement && currentEvent != Event::kEndElement) {
153         return mEmpty;
154     }
155     return mEventQueue.front().data1;
156 }
157 
getElementName() const158 const std::u16string& SourceXmlPullParser::getElementName() const {
159     const Event currentEvent = getEvent();
160     if (currentEvent != Event::kStartElement && currentEvent != Event::kEndElement) {
161         return mEmpty;
162     }
163     return mEventQueue.front().data2;
164 }
165 
beginAttributes() const166 XmlPullParser::const_iterator SourceXmlPullParser::beginAttributes() const {
167     return mEventQueue.front().attributes.begin();
168 }
169 
endAttributes() const170 XmlPullParser::const_iterator SourceXmlPullParser::endAttributes() const {
171     return mEventQueue.front().attributes.end();
172 }
173 
getAttributeCount() const174 size_t SourceXmlPullParser::getAttributeCount() const {
175     if (getEvent() != Event::kStartElement) {
176         return 0;
177     }
178     return mEventQueue.front().attributes.size();
179 }
180 
181 /**
182  * Extracts the namespace and name of an expanded element or attribute name.
183  */
splitName(const char * name,std::u16string & outNs,std::u16string & outName)184 static void splitName(const char* name, std::u16string& outNs, std::u16string& outName) {
185     const char* p = name;
186     while (*p != 0 && *p != kXmlNamespaceSep) {
187         p++;
188     }
189 
190     if (*p == 0) {
191         outNs = std::u16string();
192         outName = util::utf8ToUtf16(name);
193     } else {
194         outNs = util::utf8ToUtf16(StringPiece(name, (p - name)));
195         outName = util::utf8ToUtf16(p + 1);
196     }
197 }
198 
startNamespaceHandler(void * userData,const char * prefix,const char * uri)199 void XMLCALL SourceXmlPullParser::startNamespaceHandler(void* userData, const char* prefix,
200         const char* uri) {
201     SourceXmlPullParser* parser = reinterpret_cast<SourceXmlPullParser*>(userData);
202     std::u16string namespaceUri = uri != nullptr ? util::utf8ToUtf16(uri) : std::u16string();
203     parser->mNamespaceUris.push(namespaceUri);
204     parser->mEventQueue.push(EventData{
205             Event::kStartNamespace,
206             XML_GetCurrentLineNumber(parser->mParser),
207             parser->mDepth++,
208             prefix != nullptr ? util::utf8ToUtf16(prefix) : std::u16string(),
209             namespaceUri
210     });
211 }
212 
startElementHandler(void * userData,const char * name,const char ** attrs)213 void XMLCALL SourceXmlPullParser::startElementHandler(void* userData, const char* name,
214         const char** attrs) {
215     SourceXmlPullParser* parser = reinterpret_cast<SourceXmlPullParser*>(userData);
216 
217     EventData data = {
218             Event::kStartElement, XML_GetCurrentLineNumber(parser->mParser), parser->mDepth++
219     };
220     splitName(name, data.data1, data.data2);
221 
222     while (*attrs) {
223         Attribute attribute;
224         splitName(*attrs++, attribute.namespaceUri, attribute.name);
225         attribute.value = util::utf8ToUtf16(*attrs++);
226 
227         // Insert in sorted order.
228         auto iter = std::lower_bound(data.attributes.begin(), data.attributes.end(), attribute);
229         data.attributes.insert(iter, std::move(attribute));
230     }
231 
232     // Move the structure into the queue (no copy).
233     parser->mEventQueue.push(std::move(data));
234 }
235 
characterDataHandler(void * userData,const char * s,int len)236 void XMLCALL SourceXmlPullParser::characterDataHandler(void* userData, const char* s, int len) {
237     SourceXmlPullParser* parser = reinterpret_cast<SourceXmlPullParser*>(userData);
238 
239     parser->mEventQueue.push(EventData{
240             Event::kText,
241             XML_GetCurrentLineNumber(parser->mParser),
242             parser->mDepth,
243             util::utf8ToUtf16(StringPiece(s, len))
244     });
245 }
246 
endElementHandler(void * userData,const char * name)247 void XMLCALL SourceXmlPullParser::endElementHandler(void* userData, const char* name) {
248     SourceXmlPullParser* parser = reinterpret_cast<SourceXmlPullParser*>(userData);
249 
250     EventData data = {
251             Event::kEndElement, XML_GetCurrentLineNumber(parser->mParser), --(parser->mDepth)
252     };
253     splitName(name, data.data1, data.data2);
254 
255     // Move the data into the queue (no copy).
256     parser->mEventQueue.push(std::move(data));
257 }
258 
endNamespaceHandler(void * userData,const char * prefix)259 void XMLCALL SourceXmlPullParser::endNamespaceHandler(void* userData, const char* prefix) {
260     SourceXmlPullParser* parser = reinterpret_cast<SourceXmlPullParser*>(userData);
261 
262     parser->mEventQueue.push(EventData{
263             Event::kEndNamespace,
264             XML_GetCurrentLineNumber(parser->mParser),
265             --(parser->mDepth),
266             prefix != nullptr ? util::utf8ToUtf16(prefix) : std::u16string(),
267             parser->mNamespaceUris.top()
268     });
269     parser->mNamespaceUris.pop();
270 }
271 
commentDataHandler(void * userData,const char * comment)272 void XMLCALL SourceXmlPullParser::commentDataHandler(void* userData, const char* comment) {
273     SourceXmlPullParser* parser = reinterpret_cast<SourceXmlPullParser*>(userData);
274 
275     parser->mEventQueue.push(EventData{
276             Event::kComment,
277             XML_GetCurrentLineNumber(parser->mParser),
278             parser->mDepth,
279             util::utf8ToUtf16(comment)
280     });
281 }
282 
283 } // namespace aapt
284