diff --git a/include/searcher.h b/include/searcher.h index e5549778f..c9db988dc 100644 --- a/include/searcher.h +++ b/include/searcher.h @@ -46,17 +46,21 @@ class Result virtual std::string get_title() = 0; virtual int get_score() = 0; virtual std::string get_snippet() = 0; + virtual std::string get_content() = 0; virtual int get_wordCount() = 0; virtual int get_size() = 0; + virtual int get_readerIndex() = 0; }; struct SearcherInternal; class Searcher { public: + Searcher(); Searcher(const string& xapianDirectoryPath, Reader* reader); ~Searcher(); + void add_reader(Reader* reader, const std::string& humanReaderName); void search(std::string& search, unsigned int resultStart, unsigned int resultEnd, @@ -82,7 +86,8 @@ class Searcher const unsigned int resultEnd, const bool verbose = false); - Reader* reader; + std::vector readers; + std::vector humanReaderNames; SearcherInternal* internal; std::string searchPattern; std::string protocolPrefix; diff --git a/include/xapianSearcher.h b/include/xapianSearcher.h index 907ca733e..8c0cb3a71 100644 --- a/include/xapianSearcher.h +++ b/include/xapianSearcher.h @@ -43,8 +43,10 @@ class XapianResult : public Result virtual std::string get_title(); virtual int get_score(); virtual std::string get_snippet(); + virtual std::string get_content(); virtual int get_wordCount(); virtual int get_size(); + virtual int get_readerIndex() { return 0; }; private: XapianSearcher* searcher; diff --git a/src/android/kiwix.cpp b/src/android/kiwix.cpp index fa8daafb7..119b37e2b 100644 --- a/src/android/kiwix.cpp +++ b/src/android/kiwix.cpp @@ -486,12 +486,19 @@ JNIEXPORT jboolean JNICALL Java_org_kiwix_kiwixlib_JNIKiwix_loadFulltextIndex( std::string cPath = jni2c(path, env); pthread_mutex_lock(&searcherLock); - searcher = NULL; try { if (searcher != NULL) { delete searcher; } - searcher = new kiwix::Searcher(cPath, reader); + if (!reader || !reader->hasFulltextIndex()) { + // Use old API (no embedded full text index). + searcher = new kiwix::Searcher(cPath, reader); + } else { + // Use the new API. We don't care about the human readable name as + // we don't use it (in android). + searcher = new kiwix::Searcher(); + searcher->add_reader(reader, ""); + } } catch (...) { searcher = NULL; retVal = JNI_FALSE; diff --git a/src/searcher.cpp b/src/searcher.cpp index 4bfeab355..aae7a5267 100644 --- a/src/searcher.cpp +++ b/src/searcher.cpp @@ -45,8 +45,10 @@ class _Result : public Result virtual std::string get_title(); virtual int get_score(); virtual std::string get_snippet(); + virtual std::string get_content(); virtual int get_wordCount(); virtual int get_size(); + virtual int get_readerIndex(); private: Searcher* searcher; @@ -72,8 +74,7 @@ struct SearcherInternal { /* Constructor */ Searcher::Searcher(const string& xapianDirectoryPath, Reader* reader) - : reader(reader), - internal(new SearcherInternal()), + : internal(new SearcherInternal()), searchPattern(""), protocolPrefix("zim://"), searchProtocolPrefix("search://?"), @@ -89,11 +90,32 @@ Searcher::Searcher(const string& xapianDirectoryPath, Reader* reader) } } +Searcher::Searcher() + : internal(new SearcherInternal()), + searchPattern(""), + protocolPrefix("zim://"), + searchProtocolPrefix("search://?"), + resultCountPerPage(0), + estimatedResultCount(0), + resultStart(0), + resultEnd(0) +{ + template_ct2 = RESOURCE::results_ct2; + loadICUExternalTables(); +} + /* Destructor */ Searcher::~Searcher() { delete internal; } + +void Searcher::add_reader(Reader* reader, const std::string& humanReadableName) +{ + this->readers.push_back(reader); + this->humanReaderNames.push_back(humanReadableName); +} + /* Search strings in the database */ void Searcher::search(std::string& search, unsigned int resultStart, @@ -133,8 +155,15 @@ void Searcher::search(std::string& search, this->estimatedResultCount = internal->_xapianSearcher->results.get_matches_estimated(); } else { - internal->_search = this->reader->getZimFileHandler()->search( - unaccentedSearch, resultStart, resultEnd); + std::vector zims; + for (auto current = this->readers.begin(); current != this->readers.end(); + current++) { + zims.push_back((*current)->getZimFileHandler()); + } + zim::Search* search = new zim::Search(zims); + search->set_query(unaccentedSearch); + search->set_range(resultStart, resultEnd); + internal->_search = search; internal->current_iterator = internal->_search->begin(); this->estimatedResultCount = internal->_search->get_matches_estimated(); } @@ -190,8 +219,16 @@ void Searcher::suggestions(std::string& search, const bool verbose) * We do not support that. */ this->estimatedResultCount = 0; } else { - internal->_search = this->reader->getZimFileHandler()->suggestions( - unaccentedSearch, resultStart, resultEnd); + std::vector zims; + for (auto current = this->readers.begin(); current != this->readers.end(); + current++) { + zims.push_back((*current)->getZimFileHandler()); + } + zim::Search* search = new zim::Search(zims); + search->set_query(unaccentedSearch); + search->set_range(resultStart, resultEnd); + search->set_suggestion_mode(true); + internal->_search = search; internal->current_iterator = internal->_search->begin(); this->estimatedResultCount = internal->_search->get_matches_estimated(); } @@ -241,6 +278,13 @@ std::string _Result::get_snippet() { return iterator.get_snippet(); } +std::string _Result::get_content() +{ + if (iterator->good()) { + return iterator->getData(); + } + return ""; +} int _Result::get_size() { return iterator.get_size(); @@ -249,6 +293,10 @@ int _Result::get_wordCount() { return iterator.get_wordCount(); } +int _Result::get_readerIndex() +{ + return iterator.get_fileIndex(); +} #ifdef ENABLE_CTPP2 string Searcher::getHtml() @@ -266,6 +314,7 @@ string Searcher::getHtml() result["title"] = p_result->get_title(); result["url"] = p_result->get_url(); result["snippet"] = p_result->get_snippet(); + result["contentId"] = humanReaderNames[p_result->get_readerIndex()]; if (p_result->get_size() >= 0) { result["size"] = kiwix::beautifyInteger(p_result->get_size()); diff --git a/src/xapianSearcher.cpp b/src/xapianSearcher.cpp index aa0223d99..206888155 100644 --- a/src/xapianSearcher.cpp +++ b/src/xapianSearcher.cpp @@ -177,11 +177,10 @@ std::string XapianResult::get_snippet() We parse it and use the html dump to avoid remove html tags in the content and be able to nicely cut the text at random place. */ MyHtmlParser htmlParser; - std::string content; - unsigned int contentLength; - std::string contentType; - searcher->reader->getContentByUrl( - get_url(), content, contentLength, contentType); + std::string content = get_content(); + if (content.empty()) { + return content; + } try { htmlParser.parse_html(content, "UTF-8", true); } catch (...) { @@ -189,6 +188,19 @@ std::string XapianResult::get_snippet() return searcher->results.snippet(htmlParser.dump, 500); } +std::string XapianResult::get_content() +{ + if (!searcher->reader) { + return ""; + } + std::string content; + unsigned int contentLength; + std::string contentType; + searcher->reader->getContentByUrl( + get_url(), content, contentLength, contentType); + return content; +} + int XapianResult::get_size() { if (searcher->valuesmap.empty()) { diff --git a/static/results.tmpl b/static/results.tmpl index 1d95b8012..88f2f9499 100644 --- a/static/results.tmpl +++ b/static/results.tmpl @@ -92,36 +92,68 @@ Search: <TMPL_var searchPattern> - - + +
- Results - of for No result were found for + + Results + + - + of + + for + + + + No result were found for +
-
    +
      -
    • - ... -
      words
       -
      • +
    • + + + + + + ... + + + +
      words
      +       +
      • -
    +
- -
- - + + +