From 0a1ebc1dc02d6ec4e62c7758154b940b9f35f4f2 Mon Sep 17 00:00:00 2001
From: Matthieu Gautier <mgautier@kymeria.fr>
Date: Fri, 3 Nov 2023 18:34:14 +0100
Subject: [PATCH] Introduce `getEntryByUrl` to search for entry using url and
 fuzzy rules.

---
 include/zim/archive.h | 14 ++++++++++++++
 src/archive.cpp       | 18 ++++++++++++++++++
 src/fileimpl.cpp      | 18 ++++++++++++++++++
 src/fileimpl.h        |  6 ++++++
 4 files changed, 56 insertions(+)

diff --git a/include/zim/archive.h b/include/zim/archive.h
index 7539b96c6..ebf57a400 100644
--- a/include/zim/archive.h
+++ b/include/zim/archive.h
@@ -222,6 +222,7 @@ namespace zim
        *
        *  Get an entry using its path.
        *  The path must contains the namespace.
+       *  Path must be exact. Either a entry exists with the path or not.
        *
        *  @param path The entry's path.
        *  @return The Entry.
@@ -229,6 +230,19 @@ namespace zim
        */
       Entry getEntryByPath(const std::string& path) const;
 
+      /** Get an entry using a url.
+       *
+       *  Get an entry using a full url.
+       *  A full url is composed of a path and a querystring (separated by a '?')
+       *  If the path is not in the archive, libzim will try find the entry
+       *  using a combination of the querystring and fuzzyRules stored in the archive.
+       *
+       * @param url The url (including querystring) to search for.
+       * @return The Entry.
+       * @exception EntryNotFound If no entry has been found.
+       */
+      Entry getEntryByUrl(const std::string& url) const;
+
       /** Get an entry using its "title" index.
        *
        *  Use the index of the entry to get the idx'th entry
diff --git a/src/archive.cpp b/src/archive.cpp
index 17f33157d..7f3980686 100644
--- a/src/archive.cpp
+++ b/src/archive.cpp
@@ -250,6 +250,24 @@ namespace zim
     throw EntryNotFound("Cannot find entry");
   }
 
+  Entry Archive::getEntryByUrl(const std::string& url) const
+  {
+    const auto [path, queryParams] = urlSplit(url);
+    try {
+      return getEntryByPath(path);
+    } catch (const EntryNotFound& e) {
+      for(const auto& path_to_try: m_impl->getFuzzyRules().get_fuzzy_paths(path, queryParams)) {
+        try {
+          return getEntryByPath(path_to_try);
+        } catch (const EntryNotFound& e) {
+          continue;
+        }
+      }
+    }
+
+    throw EntryNotFound("Cannot find entry");
+  }
+
   Entry Archive::getEntryByTitle(entry_index_type idx) const
   {
     return Entry(m_impl, entry_index_type(m_impl->getIndexByTitle(title_index_t(idx))));
diff --git a/src/fileimpl.cpp b/src/fileimpl.cpp
index 152f98aa0..2529dd203 100644
--- a/src/fileimpl.cpp
+++ b/src/fileimpl.cpp
@@ -36,6 +36,7 @@
 #include "envvalue.h"
 #include "md5.h"
 #include "tools.h"
+#include "fuzzy_rules.h"
 
 log_define("zim.file.impl")
 
@@ -236,6 +237,7 @@ class Grouping
     m_byTitleDirentLookup.reset(new ByTitleDirentLookup(mp_titleDirentAccessor.get()));
 
     readMimeTypes();
+    readFuzzyRules();
   }
 
   std::unique_ptr<IndirectDirentAccessor> FileImpl::getTitleAccessor(const std::string& path)
@@ -365,6 +367,22 @@ class Grouping
     }
   }
 
+  void FileImpl::readFuzzyRules() {
+    auto r = findx('M', "FuzzyRules");
+    if (!r.first) {
+      // No rules
+      return;
+    }
+    auto fuzzy_rule_dirent = getDirent(r.second);
+    if (fuzzy_rule_dirent->isRedirect()) {
+      std::cerr << "Error: 'M/FuzzyRules' is a redirect." << std::endl;
+      return;
+    }
+    auto cluster = getCluster(fuzzy_rule_dirent->getClusterNumber());
+    auto blob = cluster->getBlob(fuzzy_rule_dirent->getBlobNumber());
+    fuzzyRules = FuzzyRules(blob);
+  }
+
   FileImpl::FindxResult FileImpl::findx(char ns, const std::string& url)
   {
     return direntLookup().find(ns, url);
diff --git a/src/fileimpl.h b/src/fileimpl.h
index cc85c46d6..a7d352905 100644
--- a/src/fileimpl.h
+++ b/src/fileimpl.h
@@ -40,6 +40,7 @@
 #include "fileheader.h"
 #include "zim_types.h"
 #include "direntreader.h"
+#include "fuzzy_rules.h"
 
 
 namespace zim
@@ -68,6 +69,8 @@ namespace zim
       typedef std::vector<std::string> MimeTypes;
       MimeTypes mimeTypes;
 
+      FuzzyRules fuzzyRules;
+
       mutable std::vector<entry_index_type> m_articleListByCluster;
       mutable std::mutex m_articleListByClusterMutex;
 
@@ -148,6 +151,8 @@ namespace zim
 
       const std::string& getMimeType(uint16_t idx) const;
 
+      const FuzzyRules& getFuzzyRules() const { return fuzzyRules; };
+
       std::string getChecksum();
       bool verify();
       bool is_multiPart() const;
@@ -165,6 +170,7 @@ namespace zim
       ClusterHandle readCluster(cluster_index_t idx);
       offset_type getMimeListEndUpperLimit() const;
       void readMimeTypes();
+      void readFuzzyRules();
       void quickCheckForCorruptFile();
 
       bool checkChecksum();