From d6a8071da4f47bceebf20a8ed6bea9332e76fe5d Mon Sep 17 00:00:00 2001
From: Jacob Dufault <jacobdufault@gmail.com>
Date: Wed, 13 Sep 2017 23:39:32 -0700
Subject: [PATCH] Be more aggressive about only indexing files once.

This should help fix perf regressions with long import times after syncing.
---
 src/command_line.cc |  9 ++++++++-
 src/utils.h         | 14 +++++++++++---
 2 files changed, 19 insertions(+), 4 deletions(-)

diff --git a/src/command_line.cc b/src/command_line.cc
index 548c10e4..9a958510 100644
--- a/src/command_line.cc
+++ b/src/command_line.cc
@@ -896,6 +896,10 @@ std::vector<Index_DoIdMap> DoParseFile(
   IndexFile* previous_index = cache_loader->TryLoad(path);
   if (previous_index) {
     // If none of the dependencies have changed, skip parsing and just load from cache.
+
+    // Checks if |path| needs to be reparsed. This will modify cached state
+    // such that calling this function twice with the same path may return true
+    // the first time but will return false the second.
     auto file_needs_parse = [&](const std::string& path) {
       optional<int64_t> modification_timestamp = GetLastModificationTime(path);
       if (!modification_timestamp)
@@ -905,6 +909,7 @@ std::vector<Index_DoIdMap> DoParseFile(
 
       if (!last_cached_modification || modification_timestamp != *last_cached_modification) {
         file_consumer_shared->Reset(path);
+        timestamp_manager->UpdateCachedModificationTime(path, *modification_timestamp);
         return FileParseQuery::NeedsParse;
       }
       return FileParseQuery::DoesNotNeedParse;
@@ -1284,7 +1289,9 @@ bool QueryDb_ImportMain(Config* config, QueryDatabase* db, ImportManager* import
 
     time.Reset();
     db->ApplyIndexUpdate(&response->update);
-    time.ResetAndPrint("Applying index update");
+    time.ResetAndPrint("Applying index update for " + StringJoinMap(response->update.files_def_update, [](const QueryFile::DefUpdate& value) {
+      return value.path;
+    }));
   }
 
   return did_work;
diff --git a/src/utils.h b/src/utils.h
index 0375be63..23b722e5 100644
--- a/src/utils.h
+++ b/src/utils.h
@@ -32,19 +32,27 @@ std::vector<std::string> SplitString(const std::string& str, const std::string&
 
 std::string LowerPathIfCaseInsensitive(const std::string& path);
 
-template <typename TValues>
-std::string StringJoin(const TValues& values) {
+
+template <typename TValues, typename TMap>
+std::string StringJoinMap(const TValues& values, const TMap& map) {
   std::string result;
   bool first = true;
   for (auto& entry : values) {
     if (!first)
       result += ", ";
     first = false;
-    result += entry;
+    result += map(entry);
   }
   return result;
 }
 
+template <typename TValues>
+std::string StringJoin(const TValues& values) {
+  return StringJoinMap(values, [](const std::string& entry) {
+    return entry;
+  });
+}
+
 // Finds all files in the given folder. This is recursive.
 std::vector<std::string> GetFilesInFolder(std::string folder, bool recursive, bool add_folder_to_path);
 void GetFilesInFolder(std::string folder, bool recursive, bool add_folder_to_path, const std::function<void(const std::string&)>& handler);