From b84eaad74856848406ab9d9ff06863b3186c78e8 Mon Sep 17 00:00:00 2001
From: Nikhil Tanwar <2002nikhiltanwar@gmail.com>
Date: Sun, 14 Aug 2022 22:11:24 +0530
Subject: [PATCH 1/6] Extract parseQuery()

Extracts the duplicate code from publisherQuery() and nameQuery() into a new function parseQuery().
---
 src/library.cpp | 24 ++++++++++++------------
 1 file changed, 12 insertions(+), 12 deletions(-)

diff --git a/src/library.cpp b/src/library.cpp
index 07eb428cb..c5f4bce3b 100644
--- a/src/library.cpp
+++ b/src/library.cpp
@@ -521,6 +521,16 @@ Xapian::Query buildXapianQueryFromFilterQuery(const Filter& filter)
   return queryParser.parse_query(normalizeText(filter.getQuery()), flags);
 }
 
+Xapian::Query parseQuery(const std::string& query, const std::string& prefix)
+{
+  Xapian::QueryParser queryParser;
+  queryParser.set_default_op(Xapian::Query::OP_OR);
+  queryParser.set_stemming_strategy(Xapian::QueryParser::STEM_NONE);
+  const auto flags = 0;
+  const auto q = queryParser.parse_query(normalizeText(query), flags, prefix);
+  return Xapian::Query(Xapian::Query::OP_PHRASE, q.get_terms_begin(), q.get_terms_end(), q.get_length());
+}
+
 Xapian::Query nameQuery(const std::string& name)
 {
   return Xapian::Query("XN" + normalizeText(name));
@@ -538,22 +548,12 @@ Xapian::Query langQuery(const std::string& lang)
 
 Xapian::Query publisherQuery(const std::string& publisher)
 {
-  Xapian::QueryParser queryParser;
-  queryParser.set_default_op(Xapian::Query::OP_OR);
-  queryParser.set_stemming_strategy(Xapian::QueryParser::STEM_NONE);
-  const auto flags = 0;
-  const auto q = queryParser.parse_query(normalizeText(publisher), flags, "XP");
-  return Xapian::Query(Xapian::Query::OP_PHRASE, q.get_terms_begin(), q.get_terms_end(), q.get_length());
+  return parseQuery(publisher, "XP");
 }
 
 Xapian::Query creatorQuery(const std::string& creator)
 {
-  Xapian::QueryParser queryParser;
-  queryParser.set_default_op(Xapian::Query::OP_OR);
-  queryParser.set_stemming_strategy(Xapian::QueryParser::STEM_NONE);
-  const auto flags = 0;
-  const auto q = queryParser.parse_query(normalizeText(creator), flags, "A");
-  return Xapian::Query(Xapian::Query::OP_PHRASE, q.get_terms_begin(), q.get_terms_end(), q.get_length());
+  return parseQuery(creator, "A");
 }
 
 Xapian::Query tagsQuery(const Filter::Tags& acceptTags, const Filter::Tags& rejectTags)

From 48a0b3bdc7debb436a5e0c933c8cd7b8284962c2 Mon Sep 17 00:00:00 2001
From: Nikhil Tanwar <2002nikhiltanwar@gmail.com>
Date: Mon, 15 Aug 2022 20:17:46 +0530
Subject: [PATCH 2/6] Add catalog filtering using ZIM aliasname

Adds mechanism to get a ZIM using its alias name.
To make a search, one needs to visit:
`TLD/?book=aliasNameHere`
---
 include/library.h             |  5 +++++
 src/library.cpp               | 25 +++++++++++++++++++++++++
 src/server/internalServer.cpp |  3 +++
 3 files changed, 33 insertions(+)

diff --git a/include/library.h b/include/library.h
index 87b0315ea..a663a9b3f 100644
--- a/include/library.h
+++ b/include/library.h
@@ -67,6 +67,7 @@ class Filter {
     std::string _query;
     bool _queryIsPartial;
     std::string _name;
+    std::string _aliasName;
 
   public: // functions
     Filter();
@@ -112,6 +113,7 @@ class Filter {
     Filter& maxSize(size_t size);
     Filter& query(std::string query, bool partial=true);
     Filter& name(std::string name);
+    Filter& aliasName(std::string aliasName);
 
     bool hasQuery() const;
     const std::string& getQuery() const { return _query; }
@@ -135,6 +137,9 @@ class Filter {
     const Tags& getAcceptTags() const { return _acceptTags; }
     const Tags& getRejectTags() const { return _rejectTags; }
 
+    bool hasAliasName() const;
+    const std::string& getAliasName() const { return _aliasName; }
+
 private: // functions
     friend class Library;
 
diff --git a/src/library.cpp b/src/library.cpp
index c5f4bce3b..88fdd880e 100644
--- a/src/library.cpp
+++ b/src/library.cpp
@@ -461,6 +461,9 @@ void Library::updateBookDB(const Book& book)
   indexer.index_text(normalizeText(book.getPublisher()), 1, "XP");
   indexer.index_text(normalizeText(book.getName()),      1, "XN");
   indexer.index_text(normalizeText(book.getCategory()),  1, "XC");
+  const auto bookName = book.getHumanReadableIdFromPath();
+  const auto aliasName = replaceRegex(bookName, "", "_[[:digit:]]{4}-[[:digit:]]{2}$");
+  indexer.index_text(normalizeText(aliasName), 1, "XF");
 
   for ( const auto& tag : split(normalizeText(book.getTags()), ";") ) {
     doc.add_boolean_term("XT" + tag);
@@ -505,6 +508,7 @@ Xapian::Query buildXapianQueryFromFilterQuery(const Filter& filter)
   queryParser.add_prefix("publisher", "XP");
   queryParser.add_prefix("creator", "A");
   queryParser.add_prefix("tag", "XT");
+  queryParser.add_prefix("filename", "XF");
   const auto partialQueryFlag = filter.queryIsPartial()
                               ? Xapian::QueryParser::FLAG_PARTIAL
                               : 0;
@@ -541,6 +545,11 @@ Xapian::Query categoryQuery(const std::string& category)
   return Xapian::Query("XC" + normalizeText(category));
 }
 
+Xapian::Query aliasNameQuery(const std::string& fileName)
+{
+  return parseQuery(fileName, "XF");
+}
+
 Xapian::Query langQuery(const std::string& lang)
 {
   return Xapian::Query("L" + normalizeText(lang));
@@ -593,6 +602,9 @@ Xapian::Query buildXapianQuery(const Filter& filter)
     const auto tq = tagsQuery(filter.getAcceptTags(), filter.getRejectTags());
     q = Xapian::Query(Xapian::Query::OP_AND, q, tq);;
   }
+  if ( filter.hasAliasName() ) {
+    q = Xapian::Query(Xapian::Query::OP_AND, q, aliasNameQuery(filter.getAliasName()));
+  }
   return q;
 }
 
@@ -742,6 +754,7 @@ enum filterTypes {
   QUERY = FLAG(12),
   NAME = FLAG(13),
   CATEGORY = FLAG(14),
+  ALIASNAME = FLAG(15),
 };
 
 Filter& Filter::local(bool accept)
@@ -844,6 +857,13 @@ Filter& Filter::name(std::string name)
   return *this;
 }
 
+Filter& Filter::aliasName(std::string aliasName)
+{
+  _aliasName = aliasName;
+  activeFilters |= ALIASNAME;  
+  return *this;
+}
+
 #define ACTIVE(X) (activeFilters & (X))
 #define FILTER(TAG, TEST) if (ACTIVE(TAG) && !(TEST)) { return false; }
 bool Filter::hasQuery() const
@@ -856,6 +876,11 @@ bool Filter::hasName() const
   return ACTIVE(NAME);
 }
 
+bool Filter::hasAliasName() const
+{
+  return ACTIVE(ALIASNAME);
+}
+
 bool Filter::hasCategory() const
 {
   return ACTIVE(CATEGORY);
diff --git a/src/server/internalServer.cpp b/src/server/internalServer.cpp
index c2c40cf5a..5e3ab598e 100644
--- a/src/server/internalServer.cpp
+++ b/src/server/internalServer.cpp
@@ -119,6 +119,9 @@ Filter get_search_filter(const RequestContext& request, const std::string& prefi
     try {
       filter.rejectTags(kiwix::split(request.get_argument(prefix+"notag"), ";"));
     } catch (...) {}
+    try {
+      filter.aliasName(request.get_argument(prefix + "book"));
+    } catch (...) {}
     return filter;
 }
 

From bd38ea97f9137936b7a5aec0a329b7ff5b63f826 Mon Sep 17 00:00:00 2001
From: Nikhil Tanwar <2002nikhiltanwar@gmail.com>
Date: Mon, 15 Aug 2022 22:12:47 +0530
Subject: [PATCH 3/6] Multivalue support for book query

Adds support for putting multiple `book` query parameter.
---
 include/library.h             |  8 ++++----
 src/library.cpp               | 28 +++++++++++++++-------------
 src/server/internalServer.cpp |  2 +-
 3 files changed, 20 insertions(+), 18 deletions(-)

diff --git a/include/library.h b/include/library.h
index a663a9b3f..8ac4fb66a 100644
--- a/include/library.h
+++ b/include/library.h
@@ -54,6 +54,7 @@ enum supportedListMode {
 class Filter {
   public: // types
     using Tags = std::vector<std::string>;
+    using AliasNames = std::vector<std::string>;
 
   private: // data
     uint64_t activeFilters;
@@ -67,7 +68,7 @@ class Filter {
     std::string _query;
     bool _queryIsPartial;
     std::string _name;
-    std::string _aliasName;
+    AliasNames _aliasNames;
 
   public: // functions
     Filter();
@@ -113,7 +114,7 @@ class Filter {
     Filter& maxSize(size_t size);
     Filter& query(std::string query, bool partial=true);
     Filter& name(std::string name);
-    Filter& aliasName(std::string aliasName);
+    Filter& aliasNames(const AliasNames& aliasNames);
 
     bool hasQuery() const;
     const std::string& getQuery() const { return _query; }
@@ -137,8 +138,7 @@ class Filter {
     const Tags& getAcceptTags() const { return _acceptTags; }
     const Tags& getRejectTags() const { return _rejectTags; }
 
-    bool hasAliasName() const;
-    const std::string& getAliasName() const { return _aliasName; }
+    const AliasNames& getAliasNames() const { return _aliasNames; }
 
 private: // functions
     friend class Library;
diff --git a/src/library.cpp b/src/library.cpp
index 88fdd880e..e45d1d488 100644
--- a/src/library.cpp
+++ b/src/library.cpp
@@ -545,9 +545,16 @@ Xapian::Query categoryQuery(const std::string& category)
   return Xapian::Query("XC" + normalizeText(category));
 }
 
-Xapian::Query aliasNameQuery(const std::string& fileName)
+Xapian::Query aliasNamesQuery(const Filter::AliasNames& aliasNames)
 {
-  return parseQuery(fileName, "XF");
+  Xapian::Query q = Xapian::Query(std::string());
+  std::vector<Xapian::Query> queryVec;
+  for (const auto& aliasName : aliasNames) {
+    queryVec.push_back(parseQuery(aliasName, "XF"));
+  }
+  Xapian::Query combinedQuery(Xapian::Query::OP_OR, queryVec.begin(), queryVec.end());
+  q = Xapian::Query(Xapian::Query::OP_FILTER, q, combinedQuery);
+  return q;
 }
 
 Xapian::Query langQuery(const std::string& lang)
@@ -602,8 +609,8 @@ Xapian::Query buildXapianQuery(const Filter& filter)
     const auto tq = tagsQuery(filter.getAcceptTags(), filter.getRejectTags());
     q = Xapian::Query(Xapian::Query::OP_AND, q, tq);;
   }
-  if ( filter.hasAliasName() ) {
-    q = Xapian::Query(Xapian::Query::OP_AND, q, aliasNameQuery(filter.getAliasName()));
+  if ( !filter.getAliasNames().empty() ) {
+    q = Xapian::Query(Xapian::Query::OP_AND, q, aliasNamesQuery(filter.getAliasNames()));
   }
   return q;
 }
@@ -754,7 +761,7 @@ enum filterTypes {
   QUERY = FLAG(12),
   NAME = FLAG(13),
   CATEGORY = FLAG(14),
-  ALIASNAME = FLAG(15),
+  ALIASNAMES = FLAG(15),
 };
 
 Filter& Filter::local(bool accept)
@@ -857,10 +864,10 @@ Filter& Filter::name(std::string name)
   return *this;
 }
 
-Filter& Filter::aliasName(std::string aliasName)
+Filter& Filter::aliasNames(const AliasNames& aliasNames)
 {
-  _aliasName = aliasName;
-  activeFilters |= ALIASNAME;  
+  _aliasNames = aliasNames;
+  activeFilters |= ALIASNAMES;  
   return *this;
 }
 
@@ -876,11 +883,6 @@ bool Filter::hasName() const
   return ACTIVE(NAME);
 }
 
-bool Filter::hasAliasName() const
-{
-  return ACTIVE(ALIASNAME);
-}
-
 bool Filter::hasCategory() const
 {
   return ACTIVE(CATEGORY);
diff --git a/src/server/internalServer.cpp b/src/server/internalServer.cpp
index 5e3ab598e..5aa1b53ee 100644
--- a/src/server/internalServer.cpp
+++ b/src/server/internalServer.cpp
@@ -120,7 +120,7 @@ Filter get_search_filter(const RequestContext& request, const std::string& prefi
       filter.rejectTags(kiwix::split(request.get_argument(prefix+"notag"), ";"));
     } catch (...) {}
     try {
-      filter.aliasName(request.get_argument(prefix + "book"));
+      filter.aliasNames(request.get_arguments(prefix + "book"));
     } catch (...) {}
     return filter;
 }

From 956c597e80fba8dd8362f7eb68d2a031f955e065 Mon Sep 17 00:00:00 2001
From: Nikhil Tanwar <2002nikhiltanwar@gmail.com>
Date: Wed, 31 Aug 2022 20:27:16 +0530
Subject: [PATCH 4/6] fixup! Extract parseQuery()

---
 src/library.cpp | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/src/library.cpp b/src/library.cpp
index e45d1d488..1bc360903 100644
--- a/src/library.cpp
+++ b/src/library.cpp
@@ -525,7 +525,7 @@ Xapian::Query buildXapianQueryFromFilterQuery(const Filter& filter)
   return queryParser.parse_query(normalizeText(filter.getQuery()), flags);
 }
 
-Xapian::Query parseQuery(const std::string& query, const std::string& prefix)
+Xapian::Query makePhraseQuery(const std::string& query, const std::string& prefix)
 {
   Xapian::QueryParser queryParser;
   queryParser.set_default_op(Xapian::Query::OP_OR);
@@ -550,7 +550,7 @@ Xapian::Query aliasNamesQuery(const Filter::AliasNames& aliasNames)
   Xapian::Query q = Xapian::Query(std::string());
   std::vector<Xapian::Query> queryVec;
   for (const auto& aliasName : aliasNames) {
-    queryVec.push_back(parseQuery(aliasName, "XF"));
+    queryVec.push_back(makePhraseQuery(aliasName, "XF"));
   }
   Xapian::Query combinedQuery(Xapian::Query::OP_OR, queryVec.begin(), queryVec.end());
   q = Xapian::Query(Xapian::Query::OP_FILTER, q, combinedQuery);
@@ -564,12 +564,12 @@ Xapian::Query langQuery(const std::string& lang)
 
 Xapian::Query publisherQuery(const std::string& publisher)
 {
-  return parseQuery(publisher, "XP");
+  return makePhraseQuery(publisher, "XP");
 }
 
 Xapian::Query creatorQuery(const std::string& creator)
 {
-  return parseQuery(creator, "A");
+  return makePhraseQuery(creator, "A");
 }
 
 Xapian::Query tagsQuery(const Filter::Tags& acceptTags, const Filter::Tags& rejectTags)

From 8a3a0b08c246d3b8e4c95987ea7b1b677e92f5b1 Mon Sep 17 00:00:00 2001
From: Nikhil Tanwar <2002nikhiltanwar@gmail.com>
Date: Thu, 1 Sep 2022 21:17:49 +0530
Subject: [PATCH 5/6] fixup! Add catalog filtering using ZIM aliasname

---
 include/name_mapper.h | 1 +
 src/library.cpp       | 3 ++-
 src/name_mapper.cpp   | 6 +++++-
 3 files changed, 8 insertions(+), 2 deletions(-)

diff --git a/include/name_mapper.h b/include/name_mapper.h
index 4247c5c3c..8b8bde620 100644
--- a/include/name_mapper.h
+++ b/include/name_mapper.h
@@ -54,6 +54,7 @@ class HumanReadableNameMapper : public NameMapper {
     virtual ~HumanReadableNameMapper() = default;
     virtual std::string getNameForId(const std::string& id) const;
     virtual std::string getIdForName(const std::string& name) const;
+    static std::string removeDateFromBookId(const std::string& bookId);
 };
 
 class UpdatableNameMapper : public NameMapper {
diff --git a/src/library.cpp b/src/library.cpp
index 1bc360903..7b0db0776 100644
--- a/src/library.cpp
+++ b/src/library.cpp
@@ -28,6 +28,7 @@
 #include "tools/stringTools.h"
 #include "tools/otherTools.h"
 #include "tools/concurrent_cache.h"
+#include "name_mapper.h"
 
 #include <pugixml.hpp>
 #include <algorithm>
@@ -462,7 +463,7 @@ void Library::updateBookDB(const Book& book)
   indexer.index_text(normalizeText(book.getName()),      1, "XN");
   indexer.index_text(normalizeText(book.getCategory()),  1, "XC");
   const auto bookName = book.getHumanReadableIdFromPath();
-  const auto aliasName = replaceRegex(bookName, "", "_[[:digit:]]{4}-[[:digit:]]{2}$");
+  const auto aliasName = HumanReadableNameMapper::removeDateFromBookId(bookName);
   indexer.index_text(normalizeText(aliasName), 1, "XF");
 
   for ( const auto& tag : split(normalizeText(book.getTags()), ";") ) {
diff --git a/src/name_mapper.cpp b/src/name_mapper.cpp
index dccf40c9b..6877e229b 100644
--- a/src/name_mapper.cpp
+++ b/src/name_mapper.cpp
@@ -34,7 +34,7 @@ HumanReadableNameMapper::HumanReadableNameMapper(kiwix::Library& library, bool w
     if (!withAlias)
       continue;
 
-    auto aliasName = replaceRegex(bookName, "", "_[[:digit:]]{4}-[[:digit:]]{2}$");
+    auto aliasName = removeDateFromBookId(bookName);
     if (aliasName == bookName) {
       continue;
     }
@@ -51,6 +51,10 @@ HumanReadableNameMapper::HumanReadableNameMapper(kiwix::Library& library, bool w
   }
 }
 
+std::string HumanReadableNameMapper::removeDateFromBookId(const std::string& bookId) {
+  return replaceRegex(bookId, "", "_[[:digit:]]{4}-[[:digit:]]{2}$");
+}
+
 std::string HumanReadableNameMapper::getNameForId(const std::string& id) const {
   return m_idToName.at(id);
 }

From 2cd057941eafd1315cb76d1518429e8bf93dc0de Mon Sep 17 00:00:00 2001
From: Nikhil Tanwar <2002nikhiltanwar@gmail.com>
Date: Fri, 2 Sep 2022 01:24:19 +0530
Subject: [PATCH 6/6] Add test for aliasName filtering

Adds test to check if filtering by alias name works.
---
 test/library.cpp | 18 ++++++++++++++++++
 1 file changed, 18 insertions(+)

diff --git a/test/library.cpp b/test/library.cpp
index ef41c4d7c..92db1a57d 100644
--- a/test/library.cpp
+++ b/test/library.cpp
@@ -500,6 +500,24 @@ TEST_F(LibraryTest, filterByTags)
   );
 }
 
+TEST_F(LibraryTest, filterByAliasNames)
+{
+  // filtering for one book
+  EXPECT_FILTER_RESULTS(kiwix::Filter().aliasNames({"zimfile"}),
+  "Ray Charles"
+  );
+
+  // filerting for more than one book
+  EXPECT_FILTER_RESULTS(kiwix::Filter().aliasNames({"zimfile", "example"}),
+  "An example ZIM archive",
+  "Ray Charles"
+  );
+
+  // filtering by alias name requires full text match
+  EXPECT_FILTER_RESULTS(kiwix::Filter().aliasNames({"wrong_name"}),
+  /* no results */
+  );
+}
 
 TEST_F(LibraryTest, filterByQuery)
 {