Adds QueryParser::ExtractQueryWords. For consistent this should

probably be ParseQuery, but I couldn't bring myself to another
ParseQuery method. I'm going to need this for bookmark queries.

BUG=4065
TEST=none

Review URL: http://codereview.chromium.org/13296

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@6604 0039d316-1c4b-4281-b951-d872f2087c98

3 files changed, 71 insertions, 15 deletions

diff --git a/chrome/browser/history/query_parser.cc b/chrome/browser/history/query_parser.cc
index ef40601..3ebaad9 100644
--- a/chrome/browser/history/query_parser.cc
+++ b/chrome/browser/history/query_parser.cc
@@ -97,6 +97,7 @@ class QueryNodeWord : public QueryNode {
                           Snippet::MatchPositions* match_positions) const;
 
   virtual bool Matches(const std::wstring& word, bool exact) const;
+  virtual void AppendWords(std::vector<std::wstring>* words) const;
 
  private:
   std::wstring word_;
@@ -124,6 +125,10 @@ bool QueryNodeWord::Matches(const std::wstring& word, bool exact) const {
          (word_.compare(0, word_.size(), word, 0, word_.size()) == 0);
 }
 
+void QueryNodeWord::AppendWords(std::vector<std::wstring>* words) const {
+  words->push_back(word_);
+}
+
 int QueryNodeWord::AppendToSQLiteQuery(std::wstring* query) const {
   query->append(word_);
 
@@ -162,6 +167,7 @@ class QueryNodeList : public QueryNode {
     NOTREACHED();
     return false;
   }
+  virtual void AppendWords(std::vector<std::wstring>* words) const;
 
  protected:
   int AppendChildrenToString(std::wstring* query) const;
@@ -175,6 +181,26 @@ QueryNodeList::~QueryNodeList() {
     delete *node;
 }
 
+void QueryNodeList::RemoveEmptySubnodes() {
+  for (size_t i = 0; i < children_.size(); ++i) {
+    if (children_[i]->IsWord())
+      continue;
+
+    QueryNodeList* list_node = static_cast<QueryNodeList*>(children_[i]);
+    list_node->RemoveEmptySubnodes();
+    if (list_node->children()->empty()) {
+      children_.erase(children_.begin() + i);
+      --i;
+      delete list_node;
+    }
+  }
+}
+
+void QueryNodeList::AppendWords(std::vector<std::wstring>* words) const {
+  for (size_t i = 0; i < children_.size(); ++i)
+    children_[i]->AppendWords(words);
+}
+
 int QueryNodeList::AppendChildrenToString(std::wstring* query) const {
   int num_words = 0;
   for (QueryNodeVector::const_iterator node = children_.begin();
@@ -259,6 +285,15 @@ void QueryParser::ParseQuery(const std::wstring& query,
     nodes->swap(*root.children());
 }
 
+
+void QueryParser::ExtractQueryWords(const std::wstring& query,
+                                    std::vector<std::wstring>* words) {
+  QueryNodeList root;
+  if (!ParseQueryImpl(query, &root))
+    return;
+  root.AppendWords(words);  
+}
+
 bool QueryParser::DoesQueryMatch(const std::wstring& text,
                                  const std::vector<QueryNode*>& query_nodes,
                                  Snippet::MatchPositions* match_positions) {
@@ -345,18 +380,3 @@ void QueryParser::ExtractQueryWords(const std::wstring& text,
     }
   }
 }
-
-void QueryNodeList::RemoveEmptySubnodes() {
-  for (size_t i = 0; i < children_.size(); ++i) {
-    if (children_[i]->IsWord())
-      continue;
-
-    QueryNodeList* list_node = static_cast<QueryNodeList*>(children_[i]);
-    list_node->RemoveEmptySubnodes();
-    if (list_node->children()->empty()) {
-      children_.erase(children_.begin() + i);
-      --i;
-      delete list_node;
-    }
-  }
-}
diff --git a/chrome/browser/history/query_parser.h b/chrome/browser/history/query_parser.h
index d374ba4..ed7bcf2 100644
--- a/chrome/browser/history/query_parser.h
+++ b/chrome/browser/history/query_parser.h
@@ -48,6 +48,9 @@ class QueryNode {
   // giving the matching region.
   virtual bool HasMatchIn(const std::vector<QueryWord>& words,
                           Snippet::MatchPositions* match_positions) const = 0;
+
+  // Appends the words that make up this node in |words|.
+  virtual void AppendWords(std::vector<std::wstring>* words) const = 0;
 };
 
 
@@ -67,6 +70,13 @@ class QueryParser {
   void ParseQuery(const std::wstring& query,
                   std::vector<QueryNode*>* nodes);
 
+  // Parses a query returning the words that make up the query. Any words in
+  // quotes are put in |words| without the quotes. For example, the query text
+  // "foo bar" results in two entries being added to words, one for foo and one
+  // for bar.
+  void ExtractQueryWords(const std::wstring& query,
+                         std::vector<std::wstring>* words);
+
   // Returns true if the string text matches the query nodes created by a call
   // to ParseQuery. If the query does match each of the matching positions in
   // the text is added to |match_positions|.
diff --git a/chrome/browser/history/query_parser_unittest.cc b/chrome/browser/history/query_parser_unittest.cc
index 50cb343..3f75989 100644
--- a/chrome/browser/history/query_parser_unittest.cc
+++ b/chrome/browser/history/query_parser_unittest.cc
@@ -131,3 +131,29 @@ TEST_F(QueryParserTest, ParseQueryNodesAndMatch) {
     }
   }
 }
+
+TEST_F(QueryParserTest, ExtractQueryWords) {
+  struct TestData2 {
+    const std::wstring text;
+    const std::wstring w1;
+    const std::wstring w2;
+    const std::wstring w3;
+    const size_t word_count;
+  } data[] = {
+    { L"foo",           L"foo", L"",    L"",  1 },
+    { L"foo bar",       L"foo", L"bar", L"",  2 },
+    { L"\"foo bar\"",   L"foo", L"bar", L"",  2 },
+    { L"\"foo bar\" a", L"foo", L"bar", L"a", 3 },
+  };
+  for (size_t i = 0; i < ARRAYSIZE_UNSAFE(data); ++i) {
+    std::vector<std::wstring> results;
+    QueryParser parser;
+    parser.ExtractQueryWords(data[i].text, &results);
+    ASSERT_EQ(data[i].word_count, results.size());
+    EXPECT_EQ(data[i].w1, results[0]);
+    if (results.size() == 2)
+      EXPECT_EQ(data[i].w2, results[1]);
+    if (results.size() == 3)
+      EXPECT_EQ(data[i].w3, results[2]);
+  }
+}