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 025b88e31eb43dce7097799a84ab878fe5c1ad99..ce6ac0c5b1d9e61b0f0927dbebdf19ed994e4bf3 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/common/url_database/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_|

[dhollowa, 2012/09/28 18:06:29]

nit: |index_in_owner_|

[beaudoin, 2012/10/02 17:43:29]

Good catch! :)

Done.

+  // 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 +132,12 @@ class TemplateURLRef {
   // {google:baseURL} or {google:baseSuggestURL}.
   bool HasGoogleBaseURLs() const;
 
+  // Use the alternate URLs and the search URL to match the provided |url|
+  // and extract |search_terms| from it. Returns false and an empty
+  // |search_terms| if no search terms can be matched.
+  bool ExtractSearchTermsFromURL(const GURL& url,
+                                 string16* search_terms) const;
+
  private:
   friend class TemplateURL;
   FRIEND_TEST_ALL_PREFIXES(TemplateURLTest, SetPrepopulatedAndParse);
@@ -206,12 +216,20 @@ 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

[dhollowa, 2012/09/28 18:06:29]

nit: |index_in_owner_|

[beaudoin, 2012/10/02 17:43:29]

Done.

+  // within our owner.
+  const size_t index_in_owner_;
+
   // Whether the URL has been parsed.
   mutable bool parsed_;
 
@@ -229,11 +247,12 @@ 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_;
@@ -325,7 +344,13 @@ struct TemplateURLData {
   // regardless of whether they have been associated with Sync.
   std::string sync_guid;
 
+  // A list of URL patterns that can be used, in addition to |url_|, to extract
+  // search terms from a URL.
+  std::vector<std::string> alternate_urls;
+
  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_;
@@ -369,6 +394,9 @@ 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 +450,23 @@ class TemplateURL {
   std::string GetExtensionId() const;
   bool IsExtensionKeyword() const;
 
+  // Returns the total number of URLs comprised in this template, including
+  // search and alternate URLs.
+  size_t URLCount() const;
+
+  // Obtain the URL given an |index|. Alternate URLS start at index 0, followed
+  // by 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 alternate URLs and the search URL to match the provided |url|
+  // and extract |search_terms| from it. Returns false and an empty
+  // |search_terms| if no search terms can be matched or if this is not an
+  // instant extended URL.
+  bool ExtractSearchTermsFromInstantExtendedURL(const GURL& url,
+                                                string16* search_terms);
+
  private:
   friend class TemplateURLService;