[Offline Pages] Adds duplicate detection metrics.

components/offline_pages/offline_page_model_impl.cc

Index: components/offline_pages/offline_page_model_impl.cc
diff --git a/components/offline_pages/offline_page_model_impl.cc b/components/offline_pages/offline_page_model_impl.cc
index 25f2e08a68de91249597a5a83ee048cdcde3bb42..8242ff22fd1eecb57be85084221c9438566430de 100644
--- a/components/offline_pages/offline_page_model_impl.cc
+++ b/components/offline_pages/offline_page_model_impl.cc
@@ -150,7 +150,44 @@ void ReportSavePageResultHistogramAfterSave(const ClientId& client_id,
   histogram->Add(static_cast<int>(result));
 }
 
-void ReportPageHistogramAfterSave(const OfflinePageItem& offline_page) {
+bool GetMatchingURLCountAndMostRecentCreationTime(
+    const std::map<int64_t, OfflinePageItem>& offline_pages,
+    std::string name_space,
+    const GURL& url,
+    base::Time created_before,
+    int* matching_url_count,
+    base::TimeDelta* most_recent_creation_time) {
+  int count = 0;
+
+  // Create a time that is very old, so that any valid time will be newer than
+  // it.
+  base::Time latest_time;
+  bool matching_page;
+
+  for (auto& id_page_pair : offline_pages) {
+    if (id_page_pair.second.client_id.name_space == name_space &&
+        url == id_page_pair.second.url) {
+      count++;

[fgorski, 2016/09/07 22:08:35]

shouldn't this be incremented when you have a matching page only?

if so can you replace matching_page with count > 0?

[dewittj, 2016/09/08 22:49:20]

In the case of saving then it's equivalent.  But if we are deleting a page, we
want the count for all pages saved before and after.

+      base::Time page_creation_time = id_page_pair.second.creation_time;
+      if (page_creation_time < created_before &&
+          page_creation_time > latest_time) {
+        latest_time = page_creation_time;
+        matching_page = true;
+      }
+    }
+  }
+
+  if (matching_url_count != nullptr)
+    *matching_url_count = count;
+  if (most_recent_creation_time != nullptr && latest_time != base::Time())
+    *most_recent_creation_time = created_before - latest_time;
+
+  return matching_page;
+}
+
+void ReportPageHistogramAfterSave(
+    const std::map<int64_t, OfflinePageItem>& offline_pages,
+    const OfflinePageItem& offline_page) {
   // The histogram below is an expansion of the UMA_HISTOGRAM_TIMES
   // macro adapted to allow for a dynamically suffixed histogram name.
   // Note: The factory creates and owns the histogram.
@@ -168,6 +205,23 @@ void ReportPageHistogramAfterSave(const OfflinePageItem& offline_page) {
       AddHistogramSuffix(offline_page.client_id, "OfflinePages.PageSize"),
       1, 10000, 50, base::HistogramBase::kUmaTargetedHistogramFlag);
   histogram->Add(offline_page.file_size / 1024);
+
+  if (offline_page.client_id.name_space == kDownloadNamespace) {
+    int matching_url_count;
+    base::TimeDelta time_since_most_recent_duplicate;
+    if (GetMatchingURLCountAndMostRecentCreationTime(
+            offline_pages, offline_page.client_id.name_space, offline_page.url,
+            offline_page.creation_time, &matching_url_count,
+            &time_since_most_recent_duplicate)) {
+      UMA_HISTOGRAM_CUSTOM_TIMES(
+          "OfflinePages.DownloadSavedPageTimeSinceDuplicateSaved",
+          time_since_most_recent_duplicate,
+          base::TimeDelta::FromMilliseconds(1), base::TimeDelta::FromDays(7),
+          200);
+    }
+    UMA_HISTOGRAM_COUNTS_100("OfflinePages.DownloadSavedPageDuplicateCount",
+                             matching_url_count);
+  }
 }
 
 void ReportPageHistogramsAfterDelete(
@@ -178,11 +232,21 @@ void ReportPageHistogramsAfterDelete(
   int64_t total_size = 0;
   for (int64_t offline_id : deleted_offline_ids) {
     auto iter = offline_pages.find(offline_id);
-    if (iter == offline_pages.end())
+    if (iter == offline_pages.end()) {

[fgorski, 2016/09/07 22:08:35]

nit: {} not needed.

[dewittj, 2016/09/08 22:49:20]

Done.

       continue;
+    }
     total_size += iter->second.file_size;
     ClientId client_id = iter->second.client_id;
 
+    if (client_id.name_space == kDownloadNamespace) {
+      int remaining_pages_with_url;
+      GetMatchingURLCountAndMostRecentCreationTime(
+          offline_pages, iter->second.client_id.name_space, iter->second.url,
+          base::Time::Max(), &remaining_pages_with_url, nullptr);
+      UMA_HISTOGRAM_COUNTS_100("OfflinePages.DownloadDeletedPageDuplicateCount",
+                               remaining_pages_with_url);
+    }
+
     // The histograms below are an expansion of the UMA_HISTOGRAM_CUSTOM_COUNTS
     // macro adapted to allow for a dynamically suffixed histogram name.
     // Note: The factory creates and owns the histogram.
@@ -736,7 +800,7 @@ void OfflinePageModelImpl::OnAddOfflinePageDone(
   if (success) {
     offline_pages_[offline_page.offline_id] = offline_page;
     result = SavePageResult::SUCCESS;
-    ReportPageHistogramAfterSave(offline_page);
+    ReportPageHistogramAfterSave(offline_pages_, offline_page);
     offline_event_logger_.RecordPageSaved(
         offline_page.client_id.name_space, offline_page.url.spec(),
         std::to_string(offline_page.offline_id));