Adam Lesinski | 6f6ceb7 | 2014-11-14 14:48:12 -0800 | [diff] [blame] | 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 | |
Adam Lesinski | 1ab598f | 2015-08-14 14:26:04 -0700 | [diff] [blame] | 17 | #include "util/Maybe.h" |
| 18 | #include "util/Util.h" |
Adam Lesinski | 467f171 | 2015-11-16 17:35:44 -0800 | [diff] [blame] | 19 | #include "xml/XmlPullParser.h" |
| 20 | #include "xml/XmlUtil.h" |
Adam Lesinski | 1ab598f | 2015-08-14 14:26:04 -0700 | [diff] [blame] | 21 | |
Adam Lesinski | 6f6ceb7 | 2014-11-14 14:48:12 -0800 | [diff] [blame] | 22 | #include <iostream> |
| 23 | #include <string> |
| 24 | |
Adam Lesinski | 6f6ceb7 | 2014-11-14 14:48:12 -0800 | [diff] [blame] | 25 | namespace aapt { |
Adam Lesinski | 467f171 | 2015-11-16 17:35:44 -0800 | [diff] [blame] | 26 | namespace xml { |
Adam Lesinski | 6f6ceb7 | 2014-11-14 14:48:12 -0800 | [diff] [blame] | 27 | |
| 28 | constexpr char kXmlNamespaceSep = 1; |
| 29 | |
Adam Lesinski | 1ab598f | 2015-08-14 14:26:04 -0700 | [diff] [blame] | 30 | XmlPullParser::XmlPullParser(std::istream& in) : mIn(in), mEmpty(), mDepth(0) { |
Adam Lesinski | 6f6ceb7 | 2014-11-14 14:48:12 -0800 | [diff] [blame] | 31 | mParser = XML_ParserCreateNS(nullptr, kXmlNamespaceSep); |
| 32 | XML_SetUserData(mParser, this); |
| 33 | XML_SetElementHandler(mParser, startElementHandler, endElementHandler); |
| 34 | XML_SetNamespaceDeclHandler(mParser, startNamespaceHandler, endNamespaceHandler); |
| 35 | XML_SetCharacterDataHandler(mParser, characterDataHandler); |
| 36 | XML_SetCommentHandler(mParser, commentDataHandler); |
| 37 | mEventQueue.push(EventData{ Event::kStartDocument, 0, mDepth++ }); |
| 38 | } |
| 39 | |
Adam Lesinski | 1ab598f | 2015-08-14 14:26:04 -0700 | [diff] [blame] | 40 | XmlPullParser::~XmlPullParser() { |
Adam Lesinski | 6f6ceb7 | 2014-11-14 14:48:12 -0800 | [diff] [blame] | 41 | XML_ParserFree(mParser); |
| 42 | } |
| 43 | |
Adam Lesinski | 1ab598f | 2015-08-14 14:26:04 -0700 | [diff] [blame] | 44 | XmlPullParser::Event XmlPullParser::next() { |
Adam Lesinski | 6f6ceb7 | 2014-11-14 14:48:12 -0800 | [diff] [blame] | 45 | const Event currentEvent = getEvent(); |
| 46 | if (currentEvent == Event::kBadDocument || currentEvent == Event::kEndDocument) { |
| 47 | return currentEvent; |
| 48 | } |
| 49 | |
| 50 | mEventQueue.pop(); |
| 51 | while (mEventQueue.empty()) { |
| 52 | mIn.read(mBuffer, sizeof(mBuffer) / sizeof(*mBuffer)); |
| 53 | |
| 54 | const bool done = mIn.eof(); |
| 55 | if (mIn.bad() && !done) { |
| 56 | mLastError = strerror(errno); |
| 57 | mEventQueue.push(EventData{ Event::kBadDocument }); |
| 58 | continue; |
| 59 | } |
| 60 | |
| 61 | if (XML_Parse(mParser, mBuffer, mIn.gcount(), done) == XML_STATUS_ERROR) { |
| 62 | mLastError = XML_ErrorString(XML_GetErrorCode(mParser)); |
| 63 | mEventQueue.push(EventData{ Event::kBadDocument }); |
| 64 | continue; |
| 65 | } |
| 66 | |
| 67 | if (done) { |
| 68 | mEventQueue.push(EventData{ Event::kEndDocument, 0, 0 }); |
| 69 | } |
| 70 | } |
| 71 | |
Adam Lesinski | 24aad16 | 2015-04-24 19:19:30 -0700 | [diff] [blame] | 72 | Event event = getEvent(); |
| 73 | |
| 74 | // Record namespace prefixes and package names so that we can do our own |
| 75 | // handling of references that use namespace aliases. |
| 76 | if (event == Event::kStartNamespace || event == Event::kEndNamespace) { |
Adam Lesinski | 467f171 | 2015-11-16 17:35:44 -0800 | [diff] [blame] | 77 | Maybe<ExtractedPackage> result = extractPackageFromNamespace(getNamespaceUri()); |
Adam Lesinski | 24aad16 | 2015-04-24 19:19:30 -0700 | [diff] [blame] | 78 | if (event == Event::kStartNamespace) { |
| 79 | if (result) { |
Adam Lesinski | 467f171 | 2015-11-16 17:35:44 -0800 | [diff] [blame] | 80 | mPackageAliases.emplace_back( |
| 81 | PackageDecl{ getNamespacePrefix(), std::move(result.value()) }); |
Adam Lesinski | 24aad16 | 2015-04-24 19:19:30 -0700 | [diff] [blame] | 82 | } |
| 83 | } else { |
| 84 | if (result) { |
Adam Lesinski | 24aad16 | 2015-04-24 19:19:30 -0700 | [diff] [blame] | 85 | mPackageAliases.pop_back(); |
| 86 | } |
| 87 | } |
| 88 | } |
| 89 | |
| 90 | return event; |
Adam Lesinski | 6f6ceb7 | 2014-11-14 14:48:12 -0800 | [diff] [blame] | 91 | } |
| 92 | |
Adam Lesinski | 1ab598f | 2015-08-14 14:26:04 -0700 | [diff] [blame] | 93 | XmlPullParser::Event XmlPullParser::getEvent() const { |
Adam Lesinski | 6f6ceb7 | 2014-11-14 14:48:12 -0800 | [diff] [blame] | 94 | return mEventQueue.front().event; |
| 95 | } |
| 96 | |
Adam Lesinski | 1ab598f | 2015-08-14 14:26:04 -0700 | [diff] [blame] | 97 | const std::string& XmlPullParser::getLastError() const { |
Adam Lesinski | 6f6ceb7 | 2014-11-14 14:48:12 -0800 | [diff] [blame] | 98 | return mLastError; |
| 99 | } |
| 100 | |
Adam Lesinski | 1ab598f | 2015-08-14 14:26:04 -0700 | [diff] [blame] | 101 | const std::u16string& XmlPullParser::getComment() const { |
Adam Lesinski | e78fd61 | 2015-10-22 12:48:43 -0700 | [diff] [blame] | 102 | return mEventQueue.front().data1; |
Adam Lesinski | 6f6ceb7 | 2014-11-14 14:48:12 -0800 | [diff] [blame] | 103 | } |
| 104 | |
Adam Lesinski | 1ab598f | 2015-08-14 14:26:04 -0700 | [diff] [blame] | 105 | size_t XmlPullParser::getLineNumber() const { |
Adam Lesinski | 6f6ceb7 | 2014-11-14 14:48:12 -0800 | [diff] [blame] | 106 | return mEventQueue.front().lineNumber; |
| 107 | } |
| 108 | |
Adam Lesinski | 1ab598f | 2015-08-14 14:26:04 -0700 | [diff] [blame] | 109 | size_t XmlPullParser::getDepth() const { |
Adam Lesinski | 6f6ceb7 | 2014-11-14 14:48:12 -0800 | [diff] [blame] | 110 | return mEventQueue.front().depth; |
| 111 | } |
| 112 | |
Adam Lesinski | 1ab598f | 2015-08-14 14:26:04 -0700 | [diff] [blame] | 113 | const std::u16string& XmlPullParser::getText() const { |
Adam Lesinski | 6f6ceb7 | 2014-11-14 14:48:12 -0800 | [diff] [blame] | 114 | if (getEvent() != Event::kText) { |
| 115 | return mEmpty; |
| 116 | } |
| 117 | return mEventQueue.front().data1; |
| 118 | } |
| 119 | |
Adam Lesinski | 1ab598f | 2015-08-14 14:26:04 -0700 | [diff] [blame] | 120 | const std::u16string& XmlPullParser::getNamespacePrefix() const { |
Adam Lesinski | 6f6ceb7 | 2014-11-14 14:48:12 -0800 | [diff] [blame] | 121 | const Event currentEvent = getEvent(); |
| 122 | if (currentEvent != Event::kStartNamespace && currentEvent != Event::kEndNamespace) { |
| 123 | return mEmpty; |
| 124 | } |
| 125 | return mEventQueue.front().data1; |
| 126 | } |
| 127 | |
Adam Lesinski | 1ab598f | 2015-08-14 14:26:04 -0700 | [diff] [blame] | 128 | const std::u16string& XmlPullParser::getNamespaceUri() const { |
Adam Lesinski | 6f6ceb7 | 2014-11-14 14:48:12 -0800 | [diff] [blame] | 129 | const Event currentEvent = getEvent(); |
| 130 | if (currentEvent != Event::kStartNamespace && currentEvent != Event::kEndNamespace) { |
| 131 | return mEmpty; |
| 132 | } |
| 133 | return mEventQueue.front().data2; |
| 134 | } |
| 135 | |
Adam Lesinski | 467f171 | 2015-11-16 17:35:44 -0800 | [diff] [blame] | 136 | Maybe<ExtractedPackage> XmlPullParser::transformPackageAlias( |
| 137 | const StringPiece16& alias, const StringPiece16& localPackage) const { |
| 138 | if (alias.empty()) { |
| 139 | return ExtractedPackage{ localPackage.toString(), false /* private */ }; |
Adam Lesinski | 1ab598f | 2015-08-14 14:26:04 -0700 | [diff] [blame] | 140 | } |
| 141 | |
Adam Lesinski | 24aad16 | 2015-04-24 19:19:30 -0700 | [diff] [blame] | 142 | const auto endIter = mPackageAliases.rend(); |
| 143 | for (auto iter = mPackageAliases.rbegin(); iter != endIter; ++iter) { |
Adam Lesinski | 467f171 | 2015-11-16 17:35:44 -0800 | [diff] [blame] | 144 | if (alias == iter->prefix) { |
| 145 | if (iter->package.package.empty()) { |
| 146 | return ExtractedPackage{ localPackage.toString(), |
| 147 | iter->package.privateNamespace }; |
Adam Lesinski | 24aad16 | 2015-04-24 19:19:30 -0700 | [diff] [blame] | 148 | } |
Adam Lesinski | 467f171 | 2015-11-16 17:35:44 -0800 | [diff] [blame] | 149 | return iter->package; |
Adam Lesinski | 24aad16 | 2015-04-24 19:19:30 -0700 | [diff] [blame] | 150 | } |
| 151 | } |
Adam Lesinski | 1ab598f | 2015-08-14 14:26:04 -0700 | [diff] [blame] | 152 | return {}; |
Adam Lesinski | 24aad16 | 2015-04-24 19:19:30 -0700 | [diff] [blame] | 153 | } |
| 154 | |
Adam Lesinski | 1ab598f | 2015-08-14 14:26:04 -0700 | [diff] [blame] | 155 | const std::u16string& XmlPullParser::getElementNamespace() const { |
Adam Lesinski | 6f6ceb7 | 2014-11-14 14:48:12 -0800 | [diff] [blame] | 156 | const Event currentEvent = getEvent(); |
| 157 | if (currentEvent != Event::kStartElement && currentEvent != Event::kEndElement) { |
| 158 | return mEmpty; |
| 159 | } |
| 160 | return mEventQueue.front().data1; |
| 161 | } |
| 162 | |
Adam Lesinski | 1ab598f | 2015-08-14 14:26:04 -0700 | [diff] [blame] | 163 | const std::u16string& XmlPullParser::getElementName() const { |
Adam Lesinski | 6f6ceb7 | 2014-11-14 14:48:12 -0800 | [diff] [blame] | 164 | const Event currentEvent = getEvent(); |
| 165 | if (currentEvent != Event::kStartElement && currentEvent != Event::kEndElement) { |
| 166 | return mEmpty; |
| 167 | } |
| 168 | return mEventQueue.front().data2; |
| 169 | } |
| 170 | |
Adam Lesinski | 1ab598f | 2015-08-14 14:26:04 -0700 | [diff] [blame] | 171 | XmlPullParser::const_iterator XmlPullParser::beginAttributes() const { |
Adam Lesinski | 6f6ceb7 | 2014-11-14 14:48:12 -0800 | [diff] [blame] | 172 | return mEventQueue.front().attributes.begin(); |
| 173 | } |
| 174 | |
Adam Lesinski | 1ab598f | 2015-08-14 14:26:04 -0700 | [diff] [blame] | 175 | XmlPullParser::const_iterator XmlPullParser::endAttributes() const { |
Adam Lesinski | 6f6ceb7 | 2014-11-14 14:48:12 -0800 | [diff] [blame] | 176 | return mEventQueue.front().attributes.end(); |
| 177 | } |
| 178 | |
Adam Lesinski | 1ab598f | 2015-08-14 14:26:04 -0700 | [diff] [blame] | 179 | size_t XmlPullParser::getAttributeCount() const { |
Adam Lesinski | 6f6ceb7 | 2014-11-14 14:48:12 -0800 | [diff] [blame] | 180 | if (getEvent() != Event::kStartElement) { |
| 181 | return 0; |
| 182 | } |
| 183 | return mEventQueue.front().attributes.size(); |
| 184 | } |
| 185 | |
| 186 | /** |
| 187 | * Extracts the namespace and name of an expanded element or attribute name. |
| 188 | */ |
| 189 | static void splitName(const char* name, std::u16string& outNs, std::u16string& outName) { |
| 190 | const char* p = name; |
| 191 | while (*p != 0 && *p != kXmlNamespaceSep) { |
| 192 | p++; |
| 193 | } |
| 194 | |
| 195 | if (*p == 0) { |
| 196 | outNs = std::u16string(); |
| 197 | outName = util::utf8ToUtf16(name); |
| 198 | } else { |
| 199 | outNs = util::utf8ToUtf16(StringPiece(name, (p - name))); |
| 200 | outName = util::utf8ToUtf16(p + 1); |
| 201 | } |
| 202 | } |
| 203 | |
Adam Lesinski | 1ab598f | 2015-08-14 14:26:04 -0700 | [diff] [blame] | 204 | void XMLCALL XmlPullParser::startNamespaceHandler(void* userData, const char* prefix, |
Adam Lesinski | 6f6ceb7 | 2014-11-14 14:48:12 -0800 | [diff] [blame] | 205 | const char* uri) { |
Adam Lesinski | 1ab598f | 2015-08-14 14:26:04 -0700 | [diff] [blame] | 206 | XmlPullParser* parser = reinterpret_cast<XmlPullParser*>(userData); |
Adam Lesinski | 6f6ceb7 | 2014-11-14 14:48:12 -0800 | [diff] [blame] | 207 | std::u16string namespaceUri = uri != nullptr ? util::utf8ToUtf16(uri) : std::u16string(); |
| 208 | parser->mNamespaceUris.push(namespaceUri); |
| 209 | parser->mEventQueue.push(EventData{ |
| 210 | Event::kStartNamespace, |
| 211 | XML_GetCurrentLineNumber(parser->mParser), |
| 212 | parser->mDepth++, |
| 213 | prefix != nullptr ? util::utf8ToUtf16(prefix) : std::u16string(), |
| 214 | namespaceUri |
| 215 | }); |
| 216 | } |
| 217 | |
Adam Lesinski | 1ab598f | 2015-08-14 14:26:04 -0700 | [diff] [blame] | 218 | void XMLCALL XmlPullParser::startElementHandler(void* userData, const char* name, |
Adam Lesinski | 6f6ceb7 | 2014-11-14 14:48:12 -0800 | [diff] [blame] | 219 | const char** attrs) { |
Adam Lesinski | 1ab598f | 2015-08-14 14:26:04 -0700 | [diff] [blame] | 220 | XmlPullParser* parser = reinterpret_cast<XmlPullParser*>(userData); |
Adam Lesinski | 6f6ceb7 | 2014-11-14 14:48:12 -0800 | [diff] [blame] | 221 | |
| 222 | EventData data = { |
| 223 | Event::kStartElement, XML_GetCurrentLineNumber(parser->mParser), parser->mDepth++ |
| 224 | }; |
| 225 | splitName(name, data.data1, data.data2); |
| 226 | |
| 227 | while (*attrs) { |
| 228 | Attribute attribute; |
| 229 | splitName(*attrs++, attribute.namespaceUri, attribute.name); |
| 230 | attribute.value = util::utf8ToUtf16(*attrs++); |
| 231 | |
| 232 | // Insert in sorted order. |
| 233 | auto iter = std::lower_bound(data.attributes.begin(), data.attributes.end(), attribute); |
| 234 | data.attributes.insert(iter, std::move(attribute)); |
| 235 | } |
| 236 | |
| 237 | // Move the structure into the queue (no copy). |
| 238 | parser->mEventQueue.push(std::move(data)); |
| 239 | } |
| 240 | |
Adam Lesinski | 1ab598f | 2015-08-14 14:26:04 -0700 | [diff] [blame] | 241 | void XMLCALL XmlPullParser::characterDataHandler(void* userData, const char* s, int len) { |
| 242 | XmlPullParser* parser = reinterpret_cast<XmlPullParser*>(userData); |
Adam Lesinski | 6f6ceb7 | 2014-11-14 14:48:12 -0800 | [diff] [blame] | 243 | |
| 244 | parser->mEventQueue.push(EventData{ |
| 245 | Event::kText, |
| 246 | XML_GetCurrentLineNumber(parser->mParser), |
| 247 | parser->mDepth, |
| 248 | util::utf8ToUtf16(StringPiece(s, len)) |
| 249 | }); |
| 250 | } |
| 251 | |
Adam Lesinski | 1ab598f | 2015-08-14 14:26:04 -0700 | [diff] [blame] | 252 | void XMLCALL XmlPullParser::endElementHandler(void* userData, const char* name) { |
| 253 | XmlPullParser* parser = reinterpret_cast<XmlPullParser*>(userData); |
Adam Lesinski | 6f6ceb7 | 2014-11-14 14:48:12 -0800 | [diff] [blame] | 254 | |
| 255 | EventData data = { |
| 256 | Event::kEndElement, XML_GetCurrentLineNumber(parser->mParser), --(parser->mDepth) |
| 257 | }; |
| 258 | splitName(name, data.data1, data.data2); |
| 259 | |
| 260 | // Move the data into the queue (no copy). |
| 261 | parser->mEventQueue.push(std::move(data)); |
| 262 | } |
| 263 | |
Adam Lesinski | 1ab598f | 2015-08-14 14:26:04 -0700 | [diff] [blame] | 264 | void XMLCALL XmlPullParser::endNamespaceHandler(void* userData, const char* prefix) { |
| 265 | XmlPullParser* parser = reinterpret_cast<XmlPullParser*>(userData); |
Adam Lesinski | 6f6ceb7 | 2014-11-14 14:48:12 -0800 | [diff] [blame] | 266 | |
| 267 | parser->mEventQueue.push(EventData{ |
| 268 | Event::kEndNamespace, |
| 269 | XML_GetCurrentLineNumber(parser->mParser), |
| 270 | --(parser->mDepth), |
| 271 | prefix != nullptr ? util::utf8ToUtf16(prefix) : std::u16string(), |
| 272 | parser->mNamespaceUris.top() |
| 273 | }); |
| 274 | parser->mNamespaceUris.pop(); |
| 275 | } |
| 276 | |
Adam Lesinski | 1ab598f | 2015-08-14 14:26:04 -0700 | [diff] [blame] | 277 | void XMLCALL XmlPullParser::commentDataHandler(void* userData, const char* comment) { |
| 278 | XmlPullParser* parser = reinterpret_cast<XmlPullParser*>(userData); |
Adam Lesinski | 6f6ceb7 | 2014-11-14 14:48:12 -0800 | [diff] [blame] | 279 | |
| 280 | parser->mEventQueue.push(EventData{ |
| 281 | Event::kComment, |
| 282 | XML_GetCurrentLineNumber(parser->mParser), |
| 283 | parser->mDepth, |
| 284 | util::utf8ToUtf16(comment) |
| 285 | }); |
| 286 | } |
| 287 | |
Adam Lesinski | 467f171 | 2015-11-16 17:35:44 -0800 | [diff] [blame] | 288 | Maybe<StringPiece16> findAttribute(const XmlPullParser* parser, const StringPiece16& name) { |
| 289 | auto iter = parser->findAttribute(u"", name); |
| 290 | if (iter != parser->endAttributes()) { |
| 291 | return StringPiece16(util::trimWhitespace(iter->value)); |
| 292 | } |
| 293 | return {}; |
| 294 | } |
| 295 | |
| 296 | Maybe<StringPiece16> findNonEmptyAttribute(const XmlPullParser* parser, const StringPiece16& name) { |
| 297 | auto iter = parser->findAttribute(u"", name); |
| 298 | if (iter != parser->endAttributes()) { |
| 299 | StringPiece16 trimmed = util::trimWhitespace(iter->value); |
| 300 | if (!trimmed.empty()) { |
| 301 | return trimmed; |
| 302 | } |
| 303 | } |
| 304 | return {}; |
| 305 | } |
| 306 | |
| 307 | } // namespace xml |
Adam Lesinski | 6f6ceb7 | 2014-11-14 14:48:12 -0800 | [diff] [blame] | 308 | } // namespace aapt |