From 7eec83e7762cc39653202f8076bffcf4e617546a Mon Sep 17 00:00:00 2001
From: vng <viktor.govako@gmail.com>
Date: Thu, 8 Aug 2013 00:15:42 +0300
Subject: [PATCH] Added geocoding for ArticleInfo indexer.

---
 builder/genindex/main.cpp   |  7 +++--
 storage/article_info.cpp    | 16 ++++++++++-
 storage/article_info.hpp    |  1 +
 storage/storage_builder.cpp | 56 ++++++++++++++++++++++++++++++++++++-
 storage/storage_builder.hpp |  3 +-
 5 files changed, 77 insertions(+), 6 deletions(-)

diff --git a/builder/genindex/main.cpp b/builder/genindex/main.cpp
index 150b18a..dc36698 100644
--- a/builder/genindex/main.cpp
+++ b/builder/genindex/main.cpp
@@ -5,15 +5,16 @@
 
 int main(int argc, char const * argv[])
 {
-  if (argc == 4)
+  if (argc == 5)
   {
     StorageBuilder builder;
     builder.ParseEntries(argv[1]);
     builder.ParseRedirects(argv[2]);
-    builder.Save(argv[3]);
+    builder.ParseGeocodes(argv[3]);
+    builder.Save(argv[4]);
   }
   else
-    cout << "Usage: <info file> <redirect file> <result file>" << endl;
+    cout << "Usage: <info file> <redirect file> <geocodes file> <result file>" << endl;
 
   return 0;
 }
diff --git a/storage/article_info.cpp b/storage/article_info.cpp
index 6196d68..4522a3a 100644
--- a/storage/article_info.cpp
+++ b/storage/article_info.cpp
@@ -4,10 +4,13 @@
 #include "../env/strings.hpp"
 #include "../env/writer.hpp"
 #include "../env/reader.hpp"
+#include "../env/latlon.hpp"
+#include "../env/assert.hpp"
 
 #include "../std/iterator.hpp"
 #include "../std/algorithm.hpp"
 #include "../std/cmath.hpp"
+#include "../std/static_assert.hpp"
 
 
 void ArticleInfo::GenerateKey()
@@ -15,6 +18,17 @@ void ArticleInfo::GenerateKey()
   m_key = str::MakeNormalizeAndLowerUtf8(m_title);
 }
 
+bool ArticleInfo::IsValidCoordinates() const
+{
+  STATIC_ASSERT(EMPTY_COORD > 200);
+  if (m_lat < 200.0 && m_lon < 200.0)
+  {
+    ASSERT(ll::ValidLat(m_lat) && ll::ValidLon(m_lon), ());
+    return true;
+  }
+  return false;
+}
+
 namespace
 {
 
@@ -64,7 +78,7 @@ void ArticleInfo::Read(rd::Reader & r)
 
 double ArticleInfo::Score(double currLat, double currLon) const
 {
-  if (m_lat != EMPTY_COORD && m_lon != EMPTY_COORD)
+  if (IsValidCoordinates())
     return earth::Distance(m_lat, m_lon, currLat, currLon);
   else
     return 0.0;
diff --git a/storage/article_info.hpp b/storage/article_info.hpp
index eaca2fa..419a7e9 100644
--- a/storage/article_info.hpp
+++ b/storage/article_info.hpp
@@ -46,6 +46,7 @@ public:
   int32_t m_parentIndex;  // NO_PARENT is the root article
 
   double m_lat, m_lon;
+  bool IsValidCoordinates() const;
 
   bool m_redirect;
 
diff --git a/storage/storage_builder.cpp b/storage/storage_builder.cpp
index bb77d8a..d914b4f 100644
--- a/storage/storage_builder.cpp
+++ b/storage/storage_builder.cpp
@@ -3,6 +3,7 @@
 #include "../env/writer.hpp"
 #include "../env/assert.hpp"
 #include "../env/logging.hpp"
+#include "../env/latlon.hpp"
 
 #include "../std/fstream.hpp"
 #include "../std/iterator.hpp"
@@ -26,7 +27,7 @@ void ProcessEntriesFile(string const & path, ToDo & toDo)
       continue;
 
     entries.clear();
-    str::Tokenize(str, "\t", back_inserter(entries));
+    str::Tokenize(str, "\t ", back_inserter(entries));
 
     toDo(entries);
   }
@@ -78,6 +79,42 @@ public:
   }
 };
 
+class DoAddGeocodes
+{
+  StorageBuilder & m_storage;
+
+  static double ToDouble(string const & s)
+  {
+    char * stop;
+    double const d = strtod(s.c_str(), &stop);
+    CHECK(stop && *stop == 0, (s));
+    return d;
+  }
+
+public:
+  DoAddGeocodes(StorageBuilder & storage) : m_storage(storage) {}
+
+  void operator() (vector<string> const & entries)
+  {
+    CHECK(entries.size() == 3, (entries));
+
+    ArticleInfoBuilder * p = m_storage.GetArticle(entries[0]);
+    if (p)
+    {
+      double const lat = ToDouble(entries[1]);
+      double const lon = ToDouble(entries[2]);
+
+      if (ll::ValidLat(lat) && ll::ValidLon(lon))
+      {
+        p->m_lat = lat;
+        p->m_lon = lon;
+      }
+      else
+        LOG(WARNING, ("Bad Lat, Lon:", entries[1], entries[2]));
+    }
+  }
+};
+
 }
 
 void StorageBuilder::ParseEntries(string const & path)
@@ -92,6 +129,23 @@ void StorageBuilder::ParseRedirects(string const & path)
   ProcessEntriesFile(path, doAdd);
 }
 
+void StorageBuilder::ParseGeocodes(string const & path)
+{
+  DoAddGeocodes doAdd(*this);
+  ProcessEntriesFile(path, doAdd);
+
+  for (size_t i = 0; i < m_info.size(); ++i)
+  {
+    if (m_info[i].m_redirect)
+    {
+      ArticleInfoBuilder const * p = GetArticle(m_info[i].m_url);
+      CHECK(p, ());
+      m_info[i].m_lat = p->m_lat;
+      m_info[i].m_lon = p->m_lon;
+    }
+  }
+}
+
 void StorageBuilder::Add(ArticleInfoBuilder const & info)
 {
   m_info.push_back(info);
diff --git a/storage/storage_builder.hpp b/storage/storage_builder.hpp
index b7df4eb..82fd190 100644
--- a/storage/storage_builder.hpp
+++ b/storage/storage_builder.hpp
@@ -40,6 +40,7 @@ class StorageBuilder
 public:
   void ParseEntries(string const & path);
   void ParseRedirects(string const & path);
+  void ParseGeocodes(string const & path);
 
   void Add(ArticleInfoBuilder const & info);
 
@@ -49,7 +50,7 @@ public:
 
   bool operator == (Storage const & s) const;
 
-  ArticleInfoBuilder const * GetArticle(string const & url) const
+  ArticleInfoBuilder * GetArticle(string const & url)
   {
     map<string, size_t>::const_iterator i = m_url2info.find(url);
     return (i == m_url2info.end() ? 0 : &m_info[i->second]);