From 55d0f1c27fc406da2acc2e041c6f651139e4a0e4 Mon Sep 17 00:00:00 2001
From: vng <viktor.govako@gmail.com>
Date: Tue, 3 Dec 2013 14:24:08 +0100
Subject: [PATCH] Show more diagnostics in case of invalid categories.txt file.

---
 indexer/categories_holder.cpp | 14 +++++++++++---
 indexer/classificator.cpp     | 12 ++++++++++--
 indexer/classificator.hpp     |  7 ++++++-
 3 files changed, 27 insertions(+), 6 deletions(-)

diff --git a/indexer/categories_holder.cpp b/indexer/categories_holder.cpp
index 5714359ba2..d22ec2ddc4 100644
--- a/indexer/categories_holder.cpp
+++ b/indexer/categories_holder.cpp
@@ -70,8 +70,10 @@ void CategoriesHolder::LoadFromStream(istream & s)
 
   Classificator const & c = classif();
 
+  int lineNumber = 0;
   while (s.good())
   {
+    ++lineNumber;
     getline(s, line);
     strings::SimpleTokenizer iter(line, ":|");
 
@@ -88,7 +90,12 @@ void CategoriesHolder::LoadFromStream(istream & s)
           strings::Tokenize(*iter, "-", MakeBackInsertFunctor(v));
 
           // get classificator type
-          types.push_back(c.GetTypeByPath(v));
+          uint32_t const type = c.GetTypeByPathSafe(v);
+          if (type != 0)
+            types.push_back(type);
+          else
+            LOG(LWARNING, ("Invalid type:", v, "at line:", lineNumber));
+
           ++iter;
         }
 
@@ -104,10 +111,11 @@ void CategoriesHolder::LoadFromStream(istream & s)
           state = EParseTypes;
           continue;
         }
+
         int8_t const langCode = StringUtf8Multilang::GetLangIndex(*iter);
         if (langCode == StringUtf8Multilang::UNSUPPORTED_LANGUAGE_CODE)
         {
-          LOG(LWARNING, ("Invalid language code:", *iter));
+          LOG(LWARNING, ("Invalid language code:", *iter, "at line:", lineNumber));
           continue;
         }
 
@@ -119,7 +127,7 @@ void CategoriesHolder::LoadFromStream(istream & s)
 
           if (name.m_name.empty())
           {
-            LOG(LWARNING, ("Empty category name"));
+            LOG(LWARNING, ("Empty category name at line:", lineNumber));
             continue;
           }
 
diff --git a/indexer/classificator.cpp b/indexer/classificator.cpp
index 01efd1174c..0e1288c9b6 100644
--- a/indexer/classificator.cpp
+++ b/indexer/classificator.cpp
@@ -363,7 +363,7 @@ void Classificator::SortClassificator()
   GetMutableRoot()->Sort();
 }
 
-uint32_t Classificator::GetTypeByPath(vector<string> const & path) const
+uint32_t Classificator::GetTypeByPathSafe(vector<string> const & path) const
 {
   ClassifObject const * p = GetRoot();
 
@@ -373,7 +373,8 @@ uint32_t Classificator::GetTypeByPath(vector<string> const & path) const
   while (i < path.size())
   {
     ClassifObjectPtr ptr = p->BinaryFind(path[i]);
-    ASSERT ( ptr, ("Invalid path in Classificator::GetTypeByPath", path) );
+    if (!ptr)
+      return 0;
 
     ftype::PushValue(type, ptr.GetIndex());
 
@@ -384,6 +385,13 @@ uint32_t Classificator::GetTypeByPath(vector<string> const & path) const
   return type;
 }
 
+uint32_t Classificator::GetTypeByPath(vector<string> const & path) const
+{
+  uint32_t const type = GetTypeByPathSafe(path);
+  ASSERT_NOT_EQUAL(type, 0, (path));
+  return type;
+}
+
 void Classificator::ReadTypesMapping(istream & s)
 {
   m_mapping.Load(s);
diff --git a/indexer/classificator.hpp b/indexer/classificator.hpp
index 1e36c7ee06..b30a1f3607 100644
--- a/indexer/classificator.hpp
+++ b/indexer/classificator.hpp
@@ -181,9 +181,14 @@ public:
 
   void Clear();
 
-  /// Return type by path in classificator tree, example:
+  /// Return type by path in classificator tree, for example
   /// path = ["natural", "caostline"].
+  //@{
+  /// @return 0 in case of nonexisting type
+  uint32_t GetTypeByPathSafe(vector<string> const & path) const;
+  /// Shows ASSERT in case of nonexisting type
   uint32_t GetTypeByPath(vector<string> const & path) const;
+  //@}
 
   uint32_t GetIndexForType(uint32_t t) const { return m_mapping.GetIndex(t); }
   uint32_t GetTypeForIndex(uint32_t i) const { return m_mapping.GetType(i); }