Support for multilang ZIMs

include/book.h

@@ -79,7 +79,9 @@ class Book
   bool isPathValid() const { return m_pathValid; }
   const std::string& getTitle() const { return m_title; }
   const std::string& getDescription() const { return m_description; }
-  const std::string& getLanguage() const { return m_language; }
+  DEPRECATED const std::string& getLanguage() const { return m_language; }
+  const std::string& getCommaSeparatedLanguages() const { return m_language; }
+  const std::vector<std::string> getLanguages() const;
   const std::string& getCreator() const { return m_creator; }
   const std::string& getPublisher() const { return m_publisher; }
   const std::string& getDate() const { return m_date; }

src/book.cpp

@@ -286,4 +286,9 @@ std::string Book::getCategoryFromTags() const
   }
 }
 
+const std::vector<std::string> Book::getLanguages() const
+{
+  return kiwix::split(m_language, ",");
+}
+
 }

src/library.cpp

@@ -373,12 +373,27 @@ std::vector<std::string> Library::getBookPropValueSet(BookStrPropMemFn p) const
 
 std::vector<std::string> Library::getBooksLanguages() const
 {
-  return getBookPropValueSet(&Book::getLanguage);
+  std::vector<std::string> langs;
+  for ( const auto& langAndCount : getBooksLanguagesWithCounts() ) {
+    langs.push_back(langAndCount.first);
+  }
+  return langs;
 }
 
 Library::AttributeCounts Library::getBooksLanguagesWithCounts() const
 {
-  return getBookAttributeCounts(&Book::getLanguage);
+  std::lock_guard<std::mutex> lock(m_mutex);
+  AttributeCounts langsWithCounts;
+
+  for (const auto& pair: mp_impl->m_books) {
+    const auto& book = pair.second;
+    if (book.getOrigId().empty()) {
+      for ( const auto& lang : book.getLanguages() ) {
+        ++langsWithCounts[lang];
+      }
+    }
+  }
+  return langsWithCounts;
 }
 
 std::vector<std::string> Library::getBooksCategories() const
@@ -440,12 +455,14 @@ void Library::updateBookDB(const Book& book)
 {
   Xapian::Stem stemmer;
   Xapian::TermGenerator indexer;
-  const std::string lang = book.getLanguage();
-  try {
-    stemmer = Xapian::Stem(iso639_3ToXapian(lang));
-    indexer.set_stemmer(stemmer);
-    indexer.set_stemming_strategy(Xapian::TermGenerator::STEM_SOME);
-  } catch (...) {}
+  const auto langs = book.getLanguages();
+  if ( langs.size() == 1 ) {
+    try {
+      stemmer = Xapian::Stem(iso639_3ToXapian(langs[0]));
+      indexer.set_stemmer(stemmer);
+      indexer.set_stemming_strategy(Xapian::TermGenerator::STEM_SOME);
+    } catch (...) {}
+  }
   Xapian::Document doc;
   indexer.set_document(doc);
 
@@ -460,7 +477,9 @@ void Library::updateBookDB(const Book& book)
   // Index all fields for field-based search
   indexer.index_text(title, 1, "S");
   indexer.index_text(desc,  1, "XD");
-  indexer.index_text(lang,  1, "L");
+  for ( const auto& lang : langs ) {
+    indexer.index_text(lang,  1, "L");
+  }
   indexer.index_text(normalizeText(book.getCreator()),   1, "A");
   indexer.index_text(normalizeText(book.getPublisher()), 1, "XP");
   indexer.index_text(normalizeText(book.getName()),      1, "XN");

src/libxml_dumper.cpp

@@ -54,7 +54,7 @@ void LibXMLDumper::handleBook(Book book, pugi::xml_node root_node) {
   if (book.getOrigId().empty()) {
     ADD_ATTR_NOT_EMPTY(entry_node, "title", book.getTitle());
     ADD_ATTR_NOT_EMPTY(entry_node, "description", book.getDescription());
-    ADD_ATTR_NOT_EMPTY(entry_node, "language", book.getLanguage());
+    ADD_ATTR_NOT_EMPTY(entry_node, "language", book.getCommaSeparatedLanguages());
     ADD_ATTR_NOT_EMPTY(entry_node, "creator", book.getCreator());
     ADD_ATTR_NOT_EMPTY(entry_node, "publisher", book.getPublisher());
     ADD_ATTR_NOT_EMPTY(entry_node, "name", book.getName());
@@ -97,7 +97,7 @@ void LibXMLDumper::handleBookmark(Bookmark bookmark, pugi::xml_node root_node) {
     auto book = library->getBookByIdThreadSafe(bookmark.getBookId());
     ADD_TEXT_ENTRY(book_node, "id", book.getId());
     ADD_TEXT_ENTRY(book_node, "title", book.getTitle());
-    ADD_TEXT_ENTRY(book_node, "language", book.getLanguage());
+    ADD_TEXT_ENTRY(book_node, "language", book.getCommaSeparatedLanguages());
     ADD_TEXT_ENTRY(book_node, "date", book.getDate());
   } catch (...) {
     ADD_TEXT_ENTRY(book_node, "id", bookmark.getBookId());

src/manager.cpp

@@ -238,7 +238,7 @@ std::string Manager::addBookFromPathAndGetId(const std::string& pathToOpen,
     }
 
     if (!checkMetaData
-        || (checkMetaData && !book.getTitle().empty() && !book.getLanguage().empty()
+        || (!book.getTitle().empty() && !book.getLanguages().empty()
             && !book.getDate().empty())) {
       book.setUrl(url);
       manipulator->addBookToLibrary(book);

src/opds_dumper.cpp

@@ -81,7 +81,7 @@ std::string fullEntryXML(const Book& book, const std::string& rootLocation, cons
       {"name", book.getName()},
       {"title", book.getTitle()},
       {"description", book.getDescription()},
-      {"language", book.getLanguage()},
+      {"language", book.getCommaSeparatedLanguages()},
       {"content_id",  urlEncode(contentId)},
       {"updated", bookDate}, // XXX: this should be the entry update datetime
       {"book_date", bookDate},

src/server/internalServer.cpp

@@ -223,7 +223,8 @@ typedef std::set<std::string> Languages;
 Languages getLanguages(const Library& lib, const Library::BookIdSet& bookIds) {
   Languages langs;
   for ( const auto& b : bookIds ) {
-    langs.insert(lib.getBookById(b).getLanguage());
+    const auto bookLangs = lib.getBookById(b).getLanguages();
+    langs.insert(bookLangs.begin(), bookLangs.end());
   }
   return langs;
 }

test/book.cpp

@@ -58,60 +58,53 @@ TEST(BookTest, updateFromXMLTest)
     EXPECT_EQ(defaultIllustration->url, "http://who.org/zara.fav");
 }
 
+namespace
+{
+
+kiwix::Book makeBook(const std::string& attr, const std::string& baseDir="")
+{
+    const XMLDoc xml("<book " + attr + "></book>");
+    kiwix::Book book;
+    book.updateFromXml(xml.child("book"), baseDir);
+    return book;
+}
+
+} // unnamed namespace
+
 TEST(BookTest, updateFromXMLCategoryHandlingTest)
 {
   {
-    const XMLDoc xml(R"(
-      <book id="abcd"
-            tags="_category:category_defined_via_tags_only"
-          >
-      </book>
+    const kiwix::Book book = makeBook(R"(
+        id="abcd"
+        tags="_category:category_defined_via_tags_only"
     )");
 
-    kiwix::Book book;
-    book.updateFromXml(xml.child("book"), "");
-
     EXPECT_EQ(book.getCategory(), "category_defined_via_tags_only");
   }
   {
-    const XMLDoc xml(R"(
-      <book id="abcd"
-            category="category_defined_via_attribute_only"
-          >
-      </book>
+    const kiwix::Book book = makeBook(R"(
+        id="abcd"
+        category="category_defined_via_attribute_only"
     )");
 
-    kiwix::Book book;
-    book.updateFromXml(xml.child("book"), "");
-
     EXPECT_EQ(book.getCategory(), "category_defined_via_attribute_only");
   }
   {
-    const XMLDoc xml(R"(
-      <book id="abcd"
-            category="category_attribute_overrides_tags"
-            tags="_category:tags_override_category_attribute"
-          >
-      </book>
+    const kiwix::Book book = makeBook(R"(
+        id="abcd"
+        category="category_attribute_overrides_tags"
+        tags="_category:tags_override_category_attribute"
     )");
 
-    kiwix::Book book;
-    book.updateFromXml(xml.child("book"), "");
-
     EXPECT_EQ(book.getCategory(), "category_attribute_overrides_tags");
   }
   {
-    const XMLDoc xml(R"(
-      <book id="abcd"
-            tags="_category:tags_override_category_attribute"
-            category="category_attribute_overrides_tags"
-          >
-      </book>
+    const kiwix::Book book = makeBook(R"(
+        id="abcd"
+        tags="_category:tags_override_category_attribute"
+        category="category_attribute_overrides_tags"
     )");
 
-    kiwix::Book book;
-    book.updateFromXml(xml.child("book"), "");
-
     EXPECT_EQ(book.getCategory(), "category_attribute_overrides_tags");
   }
 }
@@ -126,10 +119,7 @@ TEST(BookTest, setTagsDoesntAffectCategory)
 
 TEST(BookTest, updateCopiesCategory)
 {
-    const XMLDoc xml(R"(<book id="abcd" category="ted"></book>)");
-
-    kiwix::Book book;
-    book.updateFromXml(xml.child("book"), "");
+    const kiwix::Book book = makeBook(R"(id="abcd" category="ted")");
 
     kiwix::Book newBook;
     newBook.setId("abcd");
@@ -140,20 +130,15 @@ TEST(BookTest, updateCopiesCategory)
 
 TEST(BookTest, updateTest)
 {
-    const XMLDoc xml(R"(
-      <book id="xyz"
-            path="/home/user/Downloads/skin-of-color-society_en_all_2019-11.zim"
-            url="book-url"
-            name="skin-of-color-society_en_all"
-            tags="youtube;_videos:yes;_ftindex:yes;_ftindex:yes;_pictures:yes;_details:yes"
-            favicon="Ym9vay1mYXZpY29u"
-            faviconMimeType="book-favicon-mimetype"
-          >
-      </book>
-    )");
-
-    kiwix::Book book;
-    book.updateFromXml(xml.child("book"), "/data/zim");
+    kiwix::Book book = makeBook(R"(
+        id="xyz"
+        path="/home/user/Downloads/skin-of-color-society_en_all_2019-11.zim"
+        url="book-url"
+        name="skin-of-color-society_en_all"
+        tags="youtube;_videos:yes;_ftindex:yes;_ftindex:yes;_pictures:yes;_details:yes"
+        favicon="Ym9vay1mYXZpY29u"
+        faviconMimeType="book-favicon-mimetype"
+    )", "/data/zim");
 
     book.setReadOnly(false);
     book.setPathValid(true);
@@ -210,3 +195,22 @@ TEST(BookTest, getHumanReadableIdFromPath)
 #endif
   EXPECT_EQ("3plus2",  path2HumanReadableId("3+2.zim"));
 }
+
+TEST(BookTest, getLanguages)
+{
+  typedef std::vector<std::string> Langs;
+
+  {
+    const kiwix::Book book = makeBook(R"(id="abcd" language="fra")");
+
+    EXPECT_EQ(book.getCommaSeparatedLanguages(), "fra");
+    EXPECT_EQ(book.getLanguages(), Langs{ "fra" });
+  }
+
+  {
+    const kiwix::Book book = makeBook(R"(id="abcd" language="eng,ong,ing")");
+
+    EXPECT_EQ(book.getCommaSeparatedLanguages(), "eng,ong,ing");
+    EXPECT_EQ(book.getLanguages(), Langs({ "eng", "ong", "ing" }));
+  }
+}

test/data/library.xml

@@ -23,7 +23,7 @@
         url="https://github.com/kiwix/libkiwix/raw/master/test/data/zimfile.zim"
         title="Ray (uncategorized) Charles"
         description="No category is assigned to this library entry."
-        language="rus"
+        language="rus,eng"
         creator="Wikipedia"
         publisher="Kiwix"
         date="2020-03-31"

test/library.cpp

@@ -69,7 +69,7 @@ const char * sampleOpdsStream = R"(
     <id>urn:uuid:0ea1cde6-441d-6c58-f2c7-21c2838e659f</id>
     <icon>/meta?name=favicon&amp;content=wikiquote_fr_all_nopic_2019-06</icon>
     <updated>2019-06-05T00:00::00:Z</updated>
-    <language>fra</language>
+    <language>fra,ita</language>
     <summary>Une page de Wikiquote, le recueil des citations libres.</summary>
     <category>category_defined_via_category_element_only</category>
     <tags>wikiquote;nopic</tags>
@@ -199,7 +199,7 @@ const char sampleLibraryXML[] = R"(
         url="https://github.com/kiwix/libkiwix/raw/master/test/data/zimfile.zim"
         title="Ray Charles"
         description="Wikipedia articles about Ray Charles"
-        language="eng"
+        language="eng,spa"
         creator="Wikipedia"
         publisher="Kiwix"
         date="2020-03-31"
@@ -234,6 +234,8 @@ const char sampleLibraryXML[] = R"(
 namespace
 {
 
+typedef std::vector<std::string> Langs;
+
 TEST(LibraryOpdsImportTest, allInOne)
 {
   kiwix::Library lib;
@@ -248,7 +250,8 @@ TEST(LibraryOpdsImportTest, allInOne)
   EXPECT_EQ(book1.getTitle(), "Encyclopédie de la Tunisie");
   EXPECT_EQ(book1.getName(), "wikipedia_fr_tunisie_novid_2018-10");
   EXPECT_EQ(book1.getFlavour(), "unforgettable");
-  EXPECT_EQ(book1.getLanguage(), "fra");
+  EXPECT_EQ(book1.getLanguages(), Langs{ "fra" });
+  EXPECT_EQ(book1.getCommaSeparatedLanguages(), "fra");
   EXPECT_EQ(book1.getDate(), "8 Oct 2018");
   EXPECT_EQ(book1.getDescription(), "Le meilleur de Wikipédia sur la Tunisie");
   EXPECT_EQ(book1.getCreator(), "Wikipedia");
@@ -272,7 +275,8 @@ TEST(LibraryOpdsImportTest, allInOne)
   EXPECT_EQ(book2.getTitle(), "TED talks - Business");
   EXPECT_EQ(book2.getName(), "");
   EXPECT_EQ(book2.getFlavour(), "");
-  EXPECT_EQ(book2.getLanguage(), "eng");
+  EXPECT_EQ(book2.getLanguages(), Langs{ "eng" });
+  EXPECT_EQ(book2.getCommaSeparatedLanguages(), "eng");
   EXPECT_EQ(book2.getDate(), "2018-07-23");
   EXPECT_EQ(book2.getDescription(), "Ideas worth spreading");
   EXPECT_EQ(book2.getCreator(), "TED");
@@ -344,7 +348,7 @@ TEST_F(LibraryTest, sanityCheck)
 {
   EXPECT_EQ(lib.getBookCount(true, true), 12U);
   EXPECT_EQ(lib.getBooksLanguages(),
-            std::vector<std::string>({"deu", "eng", "fra"})
+            std::vector<std::string>({"deu", "eng", "fra", "ita", "spa"})
   );
   EXPECT_EQ(lib.getBooksCreators(), std::vector<std::string>({
       "Islam Stack Exchange",