Trim version and extraneus parts from AntiVirus product names.

chrome/browser/metrics/antivirus_metrics_provider_win.cc

Index: chrome/browser/metrics/antivirus_metrics_provider_win.cc
diff --git a/chrome/browser/metrics/antivirus_metrics_provider_win.cc b/chrome/browser/metrics/antivirus_metrics_provider_win.cc
index ada7c8a7cb55c818c32112f70b28ed83d687dc03..b1432209705b01f4714208b205c2aa80efd8f305 100644
--- a/chrome/browser/metrics/antivirus_metrics_provider_win.cc
+++ b/chrome/browser/metrics/antivirus_metrics_provider_win.cc
@@ -10,6 +10,7 @@
 #include <windows.h>
 #include <wscapi.h>
 
+#include <algorithm>
 #include <string>
 
 #include "base/bind.h"
@@ -21,6 +22,7 @@
 #include "base/metrics/field_trial.h"
 #include "base/metrics/histogram_macros.h"
 #include "base/path_service.h"
+#include "base/strings/string_split.h"
 #include "base/strings/string_util.h"
 #include "base/strings/sys_string_conversions.h"
 #include "base/task_runner_util.h"
@@ -57,6 +59,23 @@ struct PRODUCT_STATE {
 
 static_assert(sizeof(PRODUCT_STATE) == 4, "Wrong packing!");
 
+// Filter any part of a product string that looks like it might be a version
+// number.
+bool ShouldFilterPart(const std::string& str) {
+  if (str.empty())
+    return false;
+  // Special case for "360" (used by Norton) and "365" (used by Kaspersky).
+  if (str == "365" || str == "360")
+    return false;
+  for (const auto ch : str) {
+    if (ch == '.')
+      return true;

[rkaplow, 2017/01/12 23:18:56]

shouldn't this not return right away? i.e. this would mark "abc." as a version
string.
Could we verify that all elements of the string are digits/periods/etc?

[Will Harris, 2017/01/12 23:23:59]

this was specifically to handle stuff like

"nProtect Anti-Virus/Spyware V4.0"
"ESET NOD32 Antivirus 9.0.349.15P"

to be honest it doesn't actually matter too much if we're overzealous here as
long as some semblance of the name is retained.

I could add 'V' and 'P' to the blacklist but then it starts to get sort of
crazy... Happy to do what you think here...

[Will Harris, 2017/01/12 23:27:05]

hmm turns out this algorithm doesn't handle "nProtect Anti-Virus/Spyware V4.0"
anyway... working on it...

[Will Harris, 2017/01/12 23:39:06]

Done.

+    if (ch != '_' && !isdigit(ch))
+      return false;
+  }
+  return true;
+}
+
 bool ShouldReportFullNames() {
   // The expectation is that this will be disabled for the majority of users,
   // but this allows a small group to be enabled on other channels if there are
@@ -185,6 +204,22 @@ AntiVirusMetricsProvider::GetAntiVirusProductsOnFileThread() {
   return av_products;
 }
 
+std::string AntiVirusMetricsProvider::TrimVersionOfAvProductName(
+    const std::string& av_product) {
+  auto av_product_parts = base::SplitString(
+      av_product, " ", base::TRIM_WHITESPACE, base::SPLIT_WANT_NONEMPTY);
+
+  if (av_product_parts.size() >= 2) {
+    // Skipping first element, remove any that look like version numbers.
+    av_product_parts.erase(
+        std::remove_if(av_product_parts.begin() + 1, av_product_parts.end(),
+                       ShouldFilterPart),
+        av_product_parts.end());
+  }
+
+  return base::JoinString(av_product_parts, " ");
+}
+
 void AntiVirusMetricsProvider::GotAntiVirusProducts(
     const base::Closure& done_callback,
     const std::vector<AvProduct>& av_products) {