Introduces a search term extraction mechanism working for arbitrary search providers.

chrome/browser/search_engines/template_url.h

Index: chrome/browser/search_engines/template_url.h
diff --git a/chrome/browser/search_engines/template_url.h b/chrome/browser/search_engines/template_url.h
index 4241f529d46fc025ec40df4b17eca3fd4d4a5bc7..c0c52f1299fa5455da2f521509bbaa76cafcaec7 100644
--- a/chrome/browser/search_engines/template_url.h
+++ b/chrome/browser/search_engines/template_url.h
@@ -12,6 +12,7 @@
 #include "base/time.h"
 #include "chrome/browser/search_engines/template_url_id.h"
 #include "googleurl/src/gurl.h"
+#include "googleurl/src/url_parse.h"
 
 class Profile;
 class SearchTermsData;
@@ -40,11 +41,13 @@ class TemplateURLRef {
   };
 
   // Which kind of URL within our owner we are.  This allows us to get at the
-  // correct string field.
+  // correct string field. Use |INDEXED| to indicate that the numerical |index_|
+  // should be used instead.
   enum Type {
     SEARCH,
     SUGGEST,
     INSTANT,
+    INDEXED
   };
 
   // This struct encapsulates arguments passed to
@@ -70,6 +73,7 @@ class TemplateURLRef {
   };
 
   TemplateURLRef(TemplateURL* owner, Type type);
+  TemplateURLRef(TemplateURL* owner, size_t index_in_owner);
   ~TemplateURLRef();
 
   // Returns the raw URL. None of the parameters will have been replaced.
@@ -128,6 +132,10 @@ class TemplateURLRef {
   // {google:baseURL} or {google:baseSuggestURL}.
   bool HasGoogleBaseURLs() const;
 
+  // Extract keywords from the provided |url| using the template URL referred
+  // to. In case of failure, an empty string is returned.
+  string16 ExtractSearchTermsFromURL(const GURL& url) const;
+
  private:
   friend class TemplateURL;
   FRIEND_TEST_ALL_PREFIXES(TemplateURLTest, SetPrepopulatedAndParse);
@@ -206,12 +214,20 @@ class TemplateURLRef {
   void ParseHostAndSearchTermKey(
       const SearchTermsData& search_terms_data) const;
 
+  // Extract query key and host given a list of parameters coming from the URL
+  // query or ref.
+  void FindSearchTermsKey(const std::string& params) const;
+
   // The TemplateURL that contains us.  This should outlive us.
   TemplateURL* const owner_;
 
   // What kind of URL we are.
   const Type type_;
 
+  // If |type_| is |INDEXED|, this |index_| is used instead to refer to a URL
+  // within our owner.
+  const size_t index_in_owner_;
+
   // Whether the URL has been parsed.
   mutable bool parsed_;
 
@@ -229,11 +245,12 @@ class TemplateURLRef {
   // into the string, and may be empty.
   mutable Replacements replacements_;
 
-  // Host, path and key of the search term. These are only set if the url
-  // contains one search term.
+  // Host, path, key and location of the search term. These are only set if the
+  // url contains one search term.
   mutable std::string host_;
   mutable std::string path_;
   mutable std::string search_term_key_;
+  mutable url_parse::Parsed::ComponentType search_term_key_location_;
 
   // Whether the contained URL is a pre-populated URL.
   bool prepopulated_;
@@ -265,6 +282,18 @@ struct TemplateURLData {
   void SetURL(const std::string& url);
   const std::string& url() const { return url_; }
 
+  // Alternate URL patterns that can be used to match search terms.
+  // The URLs must not contain commas.
+  const std::vector<std::string>& alternate_urls() const {
+    return alternate_urls_;
+  }
+
+  // Returns a serialized version of |alternate_urls| as a comma separated list.
+  std::string SerializeAlternateURLs() const;
+
+  // Deserialize a comma separated list of URLs into |alternate_urls|.
+  void DeserializeAndSetAlternateURLs(const std::string& alternate_urls);
+
   // Optional additional raw URLs.
   std::string suggestions_url;
   std::string instant_url;
@@ -326,10 +355,18 @@ struct TemplateURLData {
   std::string sync_guid;
 
  private:
+  FRIEND_TEST_ALL_PREFIXES(TemplateURLTest, SerializeAlternateURLs);
+
   // Private so we can enforce using the setters and thus enforce that these
   // fields are never empty.
   string16 keyword_;
   std::string url_;
+
+  // A list of URL patterns that can be used, in addition to |url_| and
+  // |instant_url|, to extract search terms from a URL. Enforce use of a setter

[dhollowa, 2012/09/19 19:51:03]

Do we actually extract from |instant_url|?  My understanding of the instant code
is that it puts the search url into history for search terms extraction to work
correctly.

[beaudoin, 2012/09/22 06:55:45]

Not sure I fully understand your comment, but still updated the comment since we
are no longer using |instant_url|

+  // since |alternate_urls| cannot contain commas as they are serialized as a
+  // comma-separated list.
+  std::vector<std::string> alternate_urls_;
 };
 
 
@@ -369,6 +406,9 @@ class TemplateURL {
   const std::string& url() const { return data_.url(); }
   const std::string& suggestions_url() const { return data_.suggestions_url; }
   const std::string& instant_url() const { return data_.instant_url; }
+  const std::vector<std::string>& alternate_urls() const {
+    return data_.alternate_urls();
+  }
   const GURL& favicon_url() const { return data_.favicon_url; }
 
   const GURL& originating_url() const { return data_.originating_url; }
@@ -422,6 +462,26 @@ class TemplateURL {
   std::string GetExtensionId() const;
   bool IsExtensionKeyword() const;
 
+  // Returns a serialized version of |alternate_urls| as a comma separated list.
+  std::string SerializeAlternateURLs() const {
+    return data_.SerializeAlternateURLs();
+  }
+
+  // Returns the total number of URL comprised in this template, including

[dhollowa, 2012/09/19 19:51:03]

nit: s/URL/URLs/

[beaudoin, 2012/09/22 06:55:45]

Done.

+  // search, instant and alternate URLs.
+  size_t URLCount() const;
+
+  // Obtain the URL given an |index|. Alternate URLS start at index 0, followed
+  // by instant and then the regular search URL. This allows us to prioritize
+  // some pattern, so if a search term is present both in the query and the ref,
+  // we can prioritize the ref one. The |index| must be less than URLCount().
+  const std::string& GetURL(size_t index) const;
+
+  // Use the various URL comprised in this template to match the provided |url|

[dhollowa, 2012/09/19 19:51:03]

nit: s/URL/URLs/

[beaudoin, 2012/09/22 06:55:45]

Done.

+  // and extract keywords from it. If successful the search terms are rerturned.
+  // In case of failure, an empty string is returned.
+  string16 ExtractSearchTermsFromURL(const GURL& url);
+
  private:
   friend class TemplateURLService;