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 <rtl/ustrbuf.hxx>
  15 #include <osl/file.hxx>
  16 #include <osl/thread.h>
  17 #include <algorithm>
  18 #include <memory>
  19
  20 #include "LuceneHelper.hxx"
  21
  22 using namespace lucene::document;
  23
  24 HelpIndexer::HelpIndexer(OUString const &lang, OUString const &module,
  25     OUString const &srcDir, OUString const &outDir)
  26     : d_lang(lang), d_module(module)
  27 {
  28     d_indexDir = OUStringBuffer(outDir).append('/').
  29         append(module).append(".idxl").makeStringAndClear();
  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 (std::set<OUString>::iterator i = d_files.begin(); i != d_files.end(); ++i) {
  64             helpDocument(*i, &doc);
  65             writer.addDocument(&doc);
  66             doc.clear();
  67         }
  68         writer.optimize();
  69
  70         // Optimize the index
  71         writer.optimize();
  72     }
  73     catch (CLuceneError &e)
  74     {
  75         d_error = OUString::createFromAscii(e.what());
  76         return false;
  77     }
  78
  79     return true;
  80 }
  81
  82
  83 bool HelpIndexer::scanForFiles() {
  84     if (!scanForFiles(d_contentDir)) {
  85         return false;
  86     }
  87     if (!scanForFiles(d_captionDir)) {
  88         return false;
  89     }
  90     return true;
  91 }
  92
  93 bool HelpIndexer::scanForFiles(OUString const & path) {
  94
  95     osl::Directory dir(path);
  96     if (osl::FileBase::E_None != dir.open()) {
  97         d_error = "Error reading directory " + path;
  98         return true;
  99     }
 100
 101     osl::DirectoryItem item;
 102     osl::FileStatus fileStatus(osl_FileStatus_Mask_FileName | osl_FileStatus_Mask_Type);
 103     while (dir.getNextItem(item) == osl::FileBase::E_None) {
 104         item.getFileStatus(fileStatus);
 105         if (fileStatus.getFileType() == osl::FileStatus::Regular) {
 106             d_files.insert(fileStatus.getFileName());
 107         }
 108     }
 109
 110     return true;
 111 }
 112
 113 void HelpIndexer::helpDocument(OUString const & fileName, Document *doc) {
 114     // Add the help path as an indexed, untokenized field.
 115
 116     OUString path = "#HLP#" + d_module + "/" + fileName;
 117     std::vector<TCHAR> aPath(OUStringToTCHARVec(path));
 118     doc->add(*_CLNEW Field(_T("path"), &aPath[0], Field::STORE_YES | Field::INDEX_UNTOKENIZED));
 119
 120     OUString sEscapedFileName =
 121         rtl::Uri::encode(fileName,
 122         rtl_UriCharClassUric, rtl_UriEncodeIgnoreEscapes, RTL_TEXTENCODING_UTF8);
 123
 124     // Add the caption as a field.
 125     OUString captionPath = d_captionDir + "/" + sEscapedFileName;
 126     doc->add(*_CLNEW Field(_T("caption"), helpFileReader(captionPath), Field::STORE_NO | Field::INDEX_TOKENIZED));
 127
 128     // Add the content as a field.
 129     OUString contentPath = d_contentDir + "/" + sEscapedFileName;
 130     doc->add(*_CLNEW Field(_T("content"), helpFileReader(contentPath), Field::STORE_NO | Field::INDEX_TOKENIZED));
 131 }
 132
 133 lucene::util::Reader *HelpIndexer::helpFileReader(OUString const & path) {
 134     osl::File file(path);
 135     if (osl::FileBase::E_None == file.open(osl_File_OpenFlag_Read)) {
 136         file.close();
 137         OUString ustrSystemPath;
 138         osl::File::getSystemPathFromFileURL(path, ustrSystemPath);
 139         OString pathStr = OUStringToOString(ustrSystemPath, osl_getThreadTextEncoding());
 140         return _CLNEW lucene::util::FileReader(pathStr.getStr(), "UTF-8");
 141     } else {
 142         return _CLNEW lucene::util::StringReader(L"");
 143     }
 144 }
 145
 146 /* vim:set shiftwidth=4 softtabstop=4 expandtab: */