helpcompiler/source/HelpIndexer.cxx

   1 /* -*- Mode: C++; tab-width: 4; indent-tabs-mode: nil; c-basic-offset: 4 -*- */
   2 /*
   3  * This file is part of the LibreOffice project.
   4  *
   5  * This Source Code Form is subject to the terms of the Mozilla Public
   6  * License, v. 2.0. If a copy of the MPL was not distributed with this
   7  * file, You can obtain one at http://mozilla.org/MPL/2.0/.
   8  */
   9
  10 #include <helpcompiler/HelpIndexer.hxx>
  11
  12 #include <rtl/string.hxx>
  13 #include <rtl/uri.hxx>
  14 #include <o3tl/runtimetooustring.hxx>
  15 #include <osl/file.hxx>
  16 #include <osl/thread.h>
  17 #include <memory>
  18
  19 #include "LuceneHelper.hxx"
  20 #include <CLucene.h>
  21 #include <CLucene/analysis/LanguageBasedAnalyzer.h>
  22
  23 using namespace lucene::document;
  24
  25 HelpIndexer::HelpIndexer(OUString const &lang, OUString const &module,
  26     OUString const &srcDir, OUString const &outDir)
  27     : d_lang(lang), d_module(module)
  28 {
  29     d_indexDir = outDir + OUStringChar('/') + module + ".idxl";
  30     d_captionDir = srcDir + "/caption";
  31     d_contentDir = srcDir + "/content";
  32 }
  33
  34 bool HelpIndexer::indexDocuments()
  35 {
  36     if (!scanForFiles())
  37         return false;
  38
  39     try
  40     {
  41         OUString sLang = d_lang.getToken(0, '-');
  42         bool bUseCJK = sLang == "ja" || sLang == "ko" || sLang == "zh";
  43
  44         // Construct the analyzer appropriate for the given language
  45         std::unique_ptr<lucene::analysis::Analyzer> analyzer;
  46         if (bUseCJK)
  47             analyzer.reset(new lucene::analysis::LanguageBasedAnalyzer(L"cjk"));
  48         else
  49             analyzer.reset(new lucene::analysis::standard::StandardAnalyzer());
  50
  51         OUString ustrSystemPath;
  52         osl::File::getSystemPathFromFileURL(d_indexDir, ustrSystemPath);
  53
  54         OString indexDirStr = OUStringToOString(ustrSystemPath, osl_getThreadTextEncoding());
  55         lucene::index::IndexWriter writer(indexDirStr.getStr(), analyzer.get(), true);
  56         //Double limit of tokens allowed, otherwise we'll get a too-many-tokens
  57         //exception for ja help. Could alternative ignore the exception and get
  58         //truncated results as per java-Lucene apparently
  59         writer.setMaxFieldLength(lucene::index::IndexWriter::DEFAULT_MAX_FIELD_LENGTH*2);
  60
  61         // Index the identified help files
  62         Document doc;
  63         for (auto const& elem : d_files)
  64         {
  65             helpDocument(elem, &doc);
  66             writer.addDocument(&doc);
  67             doc.clear();
  68         }
  69         writer.optimize();
  70
  71         // Optimize the index
  72         writer.optimize();
  73     }
  74     catch (CLuceneError &e)
  75     {
  76         d_error = o3tl::runtimeToOUString(e.what());
  77         return false;
  78     }
  79
  80     return true;
  81 }
  82
  83
  84 bool HelpIndexer::scanForFiles() {
  85     if (!scanForFiles(d_contentDir)) {
  86         return false;
  87     }
  88     if (!scanForFiles(d_captionDir)) {
  89         return false;
  90     }
  91     return true;
  92 }
  93
  94 bool HelpIndexer::scanForFiles(OUString const & path) {
  95
  96     osl::Directory dir(path);
  97     if (osl::FileBase::E_None != dir.open()) {
  98         d_error = "Error reading directory " + path;
  99         return false;
 100     }
 101
 102     osl::DirectoryItem item;
 103     osl::FileStatus fileStatus(osl_FileStatus_Mask_FileName | osl_FileStatus_Mask_Type);
 104     while (dir.getNextItem(item) == osl::FileBase::E_None) {
 105         item.getFileStatus(fileStatus);
 106         if (fileStatus.getFileType() == osl::FileStatus::Regular) {
 107             d_files.insert(fileStatus.getFileName());
 108         }
 109     }
 110
 111     return true;
 112 }
 113
 114 void HelpIndexer::helpDocument(OUString const & fileName, Document *doc) const {
 115     // Add the help path as an indexed, untokenized field.
 116
 117     OUString path = "#HLP#" + d_module + "/" + fileName;
 118     std::vector<TCHAR> aPath(OUStringToTCHARVec(path));
 119     doc->add(*_CLNEW Field(_T("path"), aPath.data(), Field::STORE_YES | Field::INDEX_UNTOKENIZED));
 120
 121     OUString sEscapedFileName =
 122         rtl::Uri::encode(fileName,
 123         rtl_UriCharClassUric, rtl_UriEncodeIgnoreEscapes, RTL_TEXTENCODING_UTF8);
 124
 125     // Add the caption as a field.
 126     OUString captionPath = d_captionDir + "/" + sEscapedFileName;
 127     doc->add(*_CLNEW Field(_T("caption"), helpFileReader(captionPath), Field::STORE_NO | Field::INDEX_TOKENIZED));
 128
 129     // Add the content as a field.
 130     OUString contentPath = d_contentDir + "/" + sEscapedFileName;
 131     doc->add(*_CLNEW Field(_T("content"), helpFileReader(contentPath), Field::STORE_NO | Field::INDEX_TOKENIZED));
 132 }
 133
 134 lucene::util::Reader *HelpIndexer::helpFileReader(OUString const & path) {
 135     osl::File file(path);
 136     if (osl::FileBase::E_None == file.open(osl_File_OpenFlag_Read)) {
 137         file.close();
 138         OUString ustrSystemPath;
 139         osl::File::getSystemPathFromFileURL(path, ustrSystemPath);
 140         OString pathStr = OUStringToOString(ustrSystemPath, osl_getThreadTextEncoding());
 141         return _CLNEW lucene::util::FileReader(pathStr.getStr(), "UTF-8");
 142     } else {
 143         return _CLNEW lucene::util::StringReader(L"");
 144     }
 145 }
 146
 147 /* vim:set shiftwidth=4 softtabstop=4 expandtab: */