Added anonymization patterns for URLs and email addresses

components/feedback/anonymizer_tool.h

Index: components/feedback/anonymizer_tool.h
diff --git a/components/feedback/anonymizer_tool.h b/components/feedback/anonymizer_tool.h
index 54a690fe9fa677eeb3ebb3a0a4d2b533d7a1d89f..554d89342b7119410acc641aa64e52d030d59f1e 100644
--- a/components/feedback/anonymizer_tool.h
+++ b/components/feedback/anonymizer_tool.h
@@ -10,9 +10,23 @@
 #include <vector>
 
 #include <base/macros.h>
+#include <base/memory/scoped_ptr.h>
+
+namespace re2 {
+class RE2;
+}
 
 namespace feedback {
 
+struct CustomPatternWithoutContext {

[vasilii, 2016/01/08 17:02:26]

Does moving to cc work for you?

[battre, 2016/01/11 09:02:20]

This is used in the unit test.

+  // A string literal used in anonymized tests. Matches to the |pattern| are
+  // replaced with <|alias|: 1>, <|alias|: 2>, ...
+  const char* alias;
+  // A RE2 regexp with exactly one capture group. Matches will be replaced by
+  // the alias reference described above.
+  const char* pattern;
+};
+
 class AnonymizerTool {
  public:
   AnonymizerTool();
@@ -23,14 +37,20 @@ class AnonymizerTool {
   std::string Anonymize(const std::string& input);
 
  private:
+  re2::RE2* GetRegExp(const std::string& pattern);

[vasilii, 2016/01/08 17:02:26]

friend declaration first, then methods.

[battre, 2016/01/11 09:02:20]

Done.

+
   friend class AnonymizerToolTest;
 
   std::string AnonymizeMACAddresses(const std::string& input);
   std::string AnonymizeCustomPatterns(std::string input);
-  static std::string AnonymizeCustomPattern(
+  std::string AnonymizeCustomPatternWithContext(
       const std::string& input,
       const std::string& pattern,
       std::map<std::string, std::string>* identifier_space);
+  std::string AnonymizeCustomPatternWithoutContext(
+      const std::string& input,
+      const CustomPatternWithoutContext& pattern,
+      std::map<std::string, std::string>* identifier_space);
 
   // Map of MAC addresses discovered in anonymized strings to anonymized
   // representations. 11:22:33:44:55:66 gets anonymized to 11:22:33:00:00:01,
@@ -40,9 +60,13 @@ class AnonymizerTool {
   std::map<std::string, std::string> mac_addresses_;
 
   // Like mac addresses, identifiers in custom patterns are anonymized.
-  // custom_patterns_[i] contains a map of original identifier to anonymized
-  // identifier for custom pattern number i.
-  std::vector<std::map<std::string, std::string>> custom_patterns_;
+  // custom_patterns_with_context_[i] contains a map of original identifier to
+  // anonymized identifier for custom pattern number i.
+  std::vector<std::map<std::string, std::string>> custom_patterns_with_context_;
+  std::vector<std::map<std::string, std::string>>
+      custom_patterns_without_context_;
+
+  std::map<std::string, scoped_ptr<re2::RE2>> regexp_cache_;

[vasilii, 2016/01/08 17:02:26]

Comment

[battre, 2016/01/11 09:02:20]

Done.

 
   DISALLOW_COPY_AND_ASSIGN(AnonymizerTool);
 };