diff options
| author | manuel <manuel@mausz.at> | 2020-10-19 00:52:24 +0200 |
|---|---|---|
| committer | manuel <manuel@mausz.at> | 2020-10-19 00:52:24 +0200 |
| commit | be933ef2241d79558f91796cc5b3a161f72ebf9c (patch) | |
| tree | fe3ab2f130e20c99001f2d7a81d610c78c96a3f4 /xbmc/utils/RssReader.cpp | |
| parent | 5f8335c1e49ce108ef3481863833c98efa00411b (diff) | |
| download | kodi-pvr-build-be933ef2241d79558f91796cc5b3a161f72ebf9c.tar.gz kodi-pvr-build-be933ef2241d79558f91796cc5b3a161f72ebf9c.tar.bz2 kodi-pvr-build-be933ef2241d79558f91796cc5b3a161f72ebf9c.zip | |
sync with upstream
Diffstat (limited to 'xbmc/utils/RssReader.cpp')
| -rw-r--r-- | xbmc/utils/RssReader.cpp | 413 |
1 files changed, 413 insertions, 0 deletions
diff --git a/xbmc/utils/RssReader.cpp b/xbmc/utils/RssReader.cpp new file mode 100644 index 0000000..ecebc93 --- /dev/null +++ b/xbmc/utils/RssReader.cpp | |||
| @@ -0,0 +1,413 @@ | |||
| 1 | /* | ||
| 2 | * Copyright (C) 2005-2018 Team Kodi | ||
| 3 | * This file is part of Kodi - https://kodi.tv | ||
| 4 | * | ||
| 5 | * SPDX-License-Identifier: GPL-2.0-or-later | ||
| 6 | * See LICENSES/README.md for more information. | ||
| 7 | */ | ||
| 8 | |||
| 9 | #include "RssReader.h" | ||
| 10 | |||
| 11 | #include "CharsetConverter.h" | ||
| 12 | #include "ServiceBroker.h" | ||
| 13 | #include "URL.h" | ||
| 14 | #include "filesystem/CurlFile.h" | ||
| 15 | #include "filesystem/File.h" | ||
| 16 | #include "guilib/GUIRSSControl.h" | ||
| 17 | #include "guilib/LocalizeStrings.h" | ||
| 18 | #include "log.h" | ||
| 19 | #include "network/Network.h" | ||
| 20 | #include "settings/AdvancedSettings.h" | ||
| 21 | #include "settings/SettingsComponent.h" | ||
| 22 | #include "threads/SingleLock.h" | ||
| 23 | #include "threads/SystemClock.h" | ||
| 24 | #include "utils/HTMLUtil.h" | ||
| 25 | #include "utils/XTimeUtils.h" | ||
| 26 | |||
| 27 | #define RSS_COLOR_BODY 0 | ||
| 28 | #define RSS_COLOR_HEADLINE 1 | ||
| 29 | #define RSS_COLOR_CHANNEL 2 | ||
| 30 | |||
| 31 | using namespace XFILE; | ||
| 32 | |||
| 33 | ////////////////////////////////////////////////////////////////////// | ||
| 34 | // Construction/Destruction | ||
| 35 | ////////////////////////////////////////////////////////////////////// | ||
| 36 | |||
| 37 | CRssReader::CRssReader() : CThread("RSSReader") | ||
| 38 | { | ||
| 39 | m_pObserver = NULL; | ||
| 40 | m_spacesBetweenFeeds = 0; | ||
| 41 | m_bIsRunning = false; | ||
| 42 | m_savedScrollPixelPos = 0; | ||
| 43 | m_rtlText = false; | ||
| 44 | m_requestRefresh = false; | ||
| 45 | } | ||
| 46 | |||
| 47 | CRssReader::~CRssReader() | ||
| 48 | { | ||
| 49 | if (m_pObserver) | ||
| 50 | m_pObserver->OnFeedRelease(); | ||
| 51 | StopThread(); | ||
| 52 | for (unsigned int i = 0; i < m_vecTimeStamps.size(); i++) | ||
| 53 | delete m_vecTimeStamps[i]; | ||
| 54 | } | ||
| 55 | |||
| 56 | void CRssReader::Create(IRssObserver* aObserver, const std::vector<std::string>& aUrls, const std::vector<int> ×, int spacesBetweenFeeds, bool rtl) | ||
| 57 | { | ||
| 58 | CSingleLock lock(m_critical); | ||
| 59 | |||
| 60 | m_pObserver = aObserver; | ||
| 61 | m_spacesBetweenFeeds = spacesBetweenFeeds; | ||
| 62 | m_vecUrls = aUrls; | ||
| 63 | m_strFeed.resize(aUrls.size()); | ||
| 64 | m_strColors.resize(aUrls.size()); | ||
| 65 | // set update times | ||
| 66 | m_vecUpdateTimes = times; | ||
| 67 | m_rtlText = rtl; | ||
| 68 | m_requestRefresh = false; | ||
| 69 | |||
| 70 | // update each feed on creation | ||
| 71 | for (unsigned int i = 0; i < m_vecUpdateTimes.size(); ++i) | ||
| 72 | { | ||
| 73 | AddToQueue(i); | ||
| 74 | KODI::TIME::SystemTime* time = new KODI::TIME::SystemTime; | ||
| 75 | KODI::TIME::GetLocalTime(time); | ||
| 76 | m_vecTimeStamps.push_back(time); | ||
| 77 | } | ||
| 78 | } | ||
| 79 | |||
| 80 | void CRssReader::requestRefresh() | ||
| 81 | { | ||
| 82 | m_requestRefresh = true; | ||
| 83 | } | ||
| 84 | |||
| 85 | void CRssReader::AddToQueue(int iAdd) | ||
| 86 | { | ||
| 87 | CSingleLock lock(m_critical); | ||
| 88 | if (iAdd < (int)m_vecUrls.size()) | ||
| 89 | m_vecQueue.push_back(iAdd); | ||
| 90 | if (!m_bIsRunning) | ||
| 91 | { | ||
| 92 | StopThread(); | ||
| 93 | m_bIsRunning = true; | ||
| 94 | CThread::Create(false); | ||
| 95 | } | ||
| 96 | } | ||
| 97 | |||
| 98 | void CRssReader::OnExit() | ||
| 99 | { | ||
| 100 | m_bIsRunning = false; | ||
| 101 | } | ||
| 102 | |||
| 103 | int CRssReader::GetQueueSize() | ||
| 104 | { | ||
| 105 | CSingleLock lock(m_critical); | ||
| 106 | return m_vecQueue.size(); | ||
| 107 | } | ||
| 108 | |||
| 109 | void CRssReader::Process() | ||
| 110 | { | ||
| 111 | while (GetQueueSize()) | ||
| 112 | { | ||
| 113 | CSingleLock lock(m_critical); | ||
| 114 | |||
| 115 | int iFeed = m_vecQueue.front(); | ||
| 116 | m_vecQueue.erase(m_vecQueue.begin()); | ||
| 117 | |||
| 118 | m_strFeed[iFeed].clear(); | ||
| 119 | m_strColors[iFeed].clear(); | ||
| 120 | |||
| 121 | CCurlFile http; | ||
| 122 | http.SetUserAgent(CServiceBroker::GetSettingsComponent()->GetAdvancedSettings()->m_userAgent); | ||
| 123 | http.SetTimeout(2); | ||
| 124 | std::string strXML; | ||
| 125 | std::string strUrl = m_vecUrls[iFeed]; | ||
| 126 | lock.Leave(); | ||
| 127 | |||
| 128 | int nRetries = 3; | ||
| 129 | CURL url(strUrl); | ||
| 130 | std::string fileCharset; | ||
| 131 | |||
| 132 | // we wait for the network to come up | ||
| 133 | if ((url.IsProtocol("http") || url.IsProtocol("https")) && | ||
| 134 | !CServiceBroker::GetNetwork().IsAvailable()) | ||
| 135 | { | ||
| 136 | CLog::Log(LOGWARNING, "RSS: No network connection"); | ||
| 137 | strXML = "<rss><item><title>"+g_localizeStrings.Get(15301)+"</title></item></rss>"; | ||
| 138 | } | ||
| 139 | else | ||
| 140 | { | ||
| 141 | XbmcThreads::EndTime timeout(15000); | ||
| 142 | while (!m_bStop && nRetries > 0) | ||
| 143 | { | ||
| 144 | if (timeout.IsTimePast()) | ||
| 145 | { | ||
| 146 | CLog::Log(LOGERROR, "Timeout while retrieving rss feed: %s", strUrl.c_str()); | ||
| 147 | break; | ||
| 148 | } | ||
| 149 | nRetries--; | ||
| 150 | |||
| 151 | if (!url.IsProtocol("http") && !url.IsProtocol("https")) | ||
| 152 | { | ||
| 153 | CFile file; | ||
| 154 | auto_buffer buffer; | ||
| 155 | if (file.LoadFile(strUrl, buffer) > 0) | ||
| 156 | { | ||
| 157 | strXML.assign(buffer.get(), buffer.length()); | ||
| 158 | break; | ||
| 159 | } | ||
| 160 | } | ||
| 161 | else | ||
| 162 | { | ||
| 163 | if (http.Get(strUrl, strXML)) | ||
| 164 | { | ||
| 165 | fileCharset = http.GetProperty(XFILE::FILE_PROPERTY_CONTENT_CHARSET); | ||
| 166 | CLog::Log(LOGDEBUG, "Got rss feed: %s", strUrl.c_str()); | ||
| 167 | break; | ||
| 168 | } | ||
| 169 | else if (nRetries > 0) | ||
| 170 | CThread::Sleep(5000); // Network problems? Retry, but not immediately. | ||
| 171 | else | ||
| 172 | CLog::Log(LOGERROR, "Unable to obtain rss feed: %s", strUrl.c_str()); | ||
| 173 | } | ||
| 174 | } | ||
| 175 | http.Cancel(); | ||
| 176 | } | ||
| 177 | if (!strXML.empty() && m_pObserver) | ||
| 178 | { | ||
| 179 | // erase any <content:encoded> tags (also unsupported by tinyxml) | ||
| 180 | size_t iStart = strXML.find("<content:encoded>"); | ||
| 181 | size_t iEnd = 0; | ||
| 182 | while (iStart != std::string::npos) | ||
| 183 | { | ||
| 184 | // get <content:encoded> end position | ||
| 185 | iEnd = strXML.find("</content:encoded>", iStart) + 18; | ||
| 186 | |||
| 187 | // erase the section | ||
| 188 | strXML = strXML.erase(iStart, iEnd - iStart); | ||
| 189 | |||
| 190 | iStart = strXML.find("<content:encoded>"); | ||
| 191 | } | ||
| 192 | |||
| 193 | if (Parse(strXML, iFeed, fileCharset)) | ||
| 194 | CLog::Log(LOGDEBUG, "Parsed rss feed: %s", strUrl.c_str()); | ||
| 195 | } | ||
| 196 | } | ||
| 197 | UpdateObserver(); | ||
| 198 | } | ||
| 199 | |||
| 200 | void CRssReader::getFeed(vecText &text) | ||
| 201 | { | ||
| 202 | text.clear(); | ||
| 203 | // double the spaces at the start of the set | ||
| 204 | for (int j = 0; j < m_spacesBetweenFeeds; j++) | ||
| 205 | text.push_back(L' '); | ||
| 206 | for (unsigned int i = 0; i < m_strFeed.size(); i++) | ||
| 207 | { | ||
| 208 | for (int j = 0; j < m_spacesBetweenFeeds; j++) | ||
| 209 | text.push_back(L' '); | ||
| 210 | |||
| 211 | for (unsigned int j = 0; j < m_strFeed[i].size(); j++) | ||
| 212 | { | ||
| 213 | character_t letter = m_strFeed[i][j] | ((m_strColors[i][j] - 48) << 16); | ||
| 214 | text.push_back(letter); | ||
| 215 | } | ||
| 216 | } | ||
| 217 | } | ||
| 218 | |||
| 219 | void CRssReader::AddTag(const std::string &aString) | ||
| 220 | { | ||
| 221 | m_tagSet.push_back(aString); | ||
| 222 | } | ||
| 223 | |||
| 224 | void CRssReader::AddString(std::wstring aString, int aColour, int iFeed) | ||
| 225 | { | ||
| 226 | if (m_rtlText) | ||
| 227 | m_strFeed[iFeed] = aString + m_strFeed[iFeed]; | ||
| 228 | else | ||
| 229 | m_strFeed[iFeed] += aString; | ||
| 230 | |||
| 231 | size_t nStringLength = aString.size(); | ||
| 232 | |||
| 233 | for (size_t i = 0;i < nStringLength;i++) | ||
| 234 | aString[i] = static_cast<char>(48 + aColour); | ||
| 235 | |||
| 236 | if (m_rtlText) | ||
| 237 | m_strColors[iFeed] = aString + m_strColors[iFeed]; | ||
| 238 | else | ||
| 239 | m_strColors[iFeed] += aString; | ||
| 240 | } | ||
| 241 | |||
| 242 | void CRssReader::GetNewsItems(TiXmlElement* channelXmlNode, int iFeed) | ||
| 243 | { | ||
| 244 | HTML::CHTMLUtil html; | ||
| 245 | |||
| 246 | TiXmlElement * itemNode = channelXmlNode->FirstChildElement("item"); | ||
| 247 | std::map<std::string, std::wstring> mTagElements; | ||
| 248 | typedef std::pair<std::string, std::wstring> StrPair; | ||
| 249 | std::list<std::string>::iterator i; | ||
| 250 | |||
| 251 | // Add the title tag in if we didn't pass any tags in at all | ||
| 252 | // Represents default behaviour before configurability | ||
| 253 | |||
| 254 | if (m_tagSet.empty()) | ||
| 255 | AddTag("title"); | ||
| 256 | |||
| 257 | while (itemNode != nullptr) | ||
| 258 | { | ||
| 259 | TiXmlNode* childNode = itemNode->FirstChild(); | ||
| 260 | mTagElements.clear(); | ||
| 261 | while (childNode != nullptr) | ||
| 262 | { | ||
| 263 | std::string strName = childNode->ValueStr(); | ||
| 264 | |||
| 265 | for (i = m_tagSet.begin(); i != m_tagSet.end(); ++i) | ||
| 266 | { | ||
| 267 | if (!childNode->NoChildren() && *i == strName) | ||
| 268 | { | ||
| 269 | std::string htmlText = childNode->FirstChild()->ValueStr(); | ||
| 270 | |||
| 271 | // This usually happens in right-to-left languages where they want to | ||
| 272 | // specify in the RSS body that the text should be RTL. | ||
| 273 | // <title> | ||
| 274 | // <div dir="RTL">��� ����: ���� �� �����</div> | ||
| 275 | // </title> | ||
| 276 | if (htmlText == "div" || htmlText == "span") | ||
| 277 | htmlText = childNode->FirstChild()->FirstChild()->ValueStr(); | ||
| 278 | |||
| 279 | std::wstring unicodeText, unicodeText2; | ||
| 280 | |||
| 281 | g_charsetConverter.utf8ToW(htmlText, unicodeText2, m_rtlText); | ||
| 282 | html.ConvertHTMLToW(unicodeText2, unicodeText); | ||
| 283 | |||
| 284 | mTagElements.insert(StrPair(*i, unicodeText)); | ||
| 285 | } | ||
| 286 | } | ||
| 287 | childNode = childNode->NextSibling(); | ||
| 288 | } | ||
| 289 | |||
| 290 | int rsscolour = RSS_COLOR_HEADLINE; | ||
| 291 | for (i = m_tagSet.begin(); i != m_tagSet.end(); ++i) | ||
| 292 | { | ||
| 293 | std::map<std::string, std::wstring>::iterator j = mTagElements.find(*i); | ||
| 294 | |||
| 295 | if (j == mTagElements.end()) | ||
| 296 | continue; | ||
| 297 | |||
| 298 | std::wstring& text = j->second; | ||
| 299 | AddString(text, rsscolour, iFeed); | ||
| 300 | rsscolour = RSS_COLOR_BODY; | ||
| 301 | text = L" - "; | ||
| 302 | AddString(text, rsscolour, iFeed); | ||
| 303 | } | ||
| 304 | itemNode = itemNode->NextSiblingElement("item"); | ||
| 305 | } | ||
| 306 | } | ||
| 307 | |||
| 308 | bool CRssReader::Parse(const std::string& data, int iFeed, const std::string& charset) | ||
| 309 | { | ||
| 310 | m_xml.Clear(); | ||
| 311 | m_xml.Parse(data, charset); | ||
| 312 | |||
| 313 | CLog::Log(LOGDEBUG, "RSS feed encoding: %s", m_xml.GetUsedCharset().c_str()); | ||
| 314 | |||
| 315 | return Parse(iFeed); | ||
| 316 | } | ||
| 317 | |||
| 318 | bool CRssReader::Parse(int iFeed) | ||
| 319 | { | ||
| 320 | TiXmlElement* rootXmlNode = m_xml.RootElement(); | ||
| 321 | |||
| 322 | if (!rootXmlNode) | ||
| 323 | return false; | ||
| 324 | |||
| 325 | TiXmlElement* rssXmlNode = NULL; | ||
| 326 | |||
| 327 | std::string strValue = rootXmlNode->ValueStr(); | ||
| 328 | if (strValue.find("rss") != std::string::npos || | ||
| 329 | strValue.find("rdf") != std::string::npos) | ||
| 330 | rssXmlNode = rootXmlNode; | ||
| 331 | else | ||
| 332 | { | ||
| 333 | // Unable to find root <rss> or <rdf> node | ||
| 334 | return false; | ||
| 335 | } | ||
| 336 | |||
| 337 | TiXmlElement* channelXmlNode = rssXmlNode->FirstChildElement("channel"); | ||
| 338 | if (channelXmlNode) | ||
| 339 | { | ||
| 340 | TiXmlElement* titleNode = channelXmlNode->FirstChildElement("title"); | ||
| 341 | if (titleNode && !titleNode->NoChildren()) | ||
| 342 | { | ||
| 343 | std::string strChannel = titleNode->FirstChild()->Value(); | ||
| 344 | std::wstring strChannelUnicode; | ||
| 345 | g_charsetConverter.utf8ToW(strChannel, strChannelUnicode, m_rtlText); | ||
| 346 | AddString(strChannelUnicode, RSS_COLOR_CHANNEL, iFeed); | ||
| 347 | |||
| 348 | AddString(L":", RSS_COLOR_CHANNEL, iFeed); | ||
| 349 | AddString(L" ", RSS_COLOR_CHANNEL, iFeed); | ||
| 350 | } | ||
| 351 | |||
| 352 | GetNewsItems(channelXmlNode,iFeed); | ||
| 353 | } | ||
| 354 | |||
| 355 | GetNewsItems(rssXmlNode,iFeed); | ||
| 356 | |||
| 357 | // avoid trailing ' - ' | ||
| 358 | if (m_strFeed[iFeed].size() > 3 && m_strFeed[iFeed].substr(m_strFeed[iFeed].size() - 3) == L" - ") | ||
| 359 | { | ||
| 360 | if (m_rtlText) | ||
| 361 | { | ||
| 362 | m_strFeed[iFeed].erase(0, 3); | ||
| 363 | m_strColors[iFeed].erase(0, 3); | ||
| 364 | } | ||
| 365 | else | ||
| 366 | { | ||
| 367 | m_strFeed[iFeed].erase(m_strFeed[iFeed].length() - 3); | ||
| 368 | m_strColors[iFeed].erase(m_strColors[iFeed].length() - 3); | ||
| 369 | } | ||
| 370 | } | ||
| 371 | return true; | ||
| 372 | } | ||
| 373 | |||
| 374 | void CRssReader::SetObserver(IRssObserver *observer) | ||
| 375 | { | ||
| 376 | m_pObserver = observer; | ||
| 377 | } | ||
| 378 | |||
| 379 | void CRssReader::UpdateObserver() | ||
| 380 | { | ||
| 381 | if (!m_pObserver) | ||
| 382 | return; | ||
| 383 | |||
| 384 | vecText feed; | ||
| 385 | getFeed(feed); | ||
| 386 | if (!feed.empty()) | ||
| 387 | { | ||
| 388 | CSingleLock lock(CServiceBroker::GetWinSystem()->GetGfxContext()); | ||
| 389 | if (m_pObserver) // need to check again when locked to make sure observer wasnt removed | ||
| 390 | m_pObserver->OnFeedUpdate(feed); | ||
| 391 | } | ||
| 392 | } | ||
| 393 | |||
| 394 | void CRssReader::CheckForUpdates() | ||
| 395 | { | ||
| 396 | KODI::TIME::SystemTime time; | ||
| 397 | KODI::TIME::GetLocalTime(&time); | ||
| 398 | |||
| 399 | for (unsigned int i = 0;i < m_vecUpdateTimes.size(); ++i ) | ||
| 400 | { | ||
| 401 | if (m_requestRefresh || ((time.day * 24 * 60) + (time.hour * 60) + time.minute) - | ||
| 402 | ((m_vecTimeStamps[i]->day * 24 * 60) + | ||
| 403 | (m_vecTimeStamps[i]->hour * 60) + m_vecTimeStamps[i]->minute) > | ||
| 404 | m_vecUpdateTimes[i]) | ||
| 405 | { | ||
| 406 | CLog::Log(LOGDEBUG, "Updating RSS"); | ||
| 407 | KODI::TIME::GetLocalTime(m_vecTimeStamps[i]); | ||
| 408 | AddToQueue(i); | ||
| 409 | } | ||
| 410 | } | ||
| 411 | |||
| 412 | m_requestRefresh = false; | ||
| 413 | } | ||
