components/query_parser/query_parser_unittest.cc

   1 // Copyright 2014 The Chromium Authors. All rights reserved.
   2 // Use of this source code is governed by a BSD-style license that can be
   3 // found in the LICENSE file.
   4
   5 #include "base/basictypes.h"
   6 #include "base/memory/scoped_vector.h"
   7 #include "base/strings/utf_string_conversions.h"
   8 #include "components/query_parser/query_parser.h"
   9 #include "testing/gtest/include/gtest/gtest.h"
  10
  11 namespace query_parser {
  12
  13 class QueryParserTest : public testing::Test {
  14  public:
  15   struct TestData {
  16     const char* input;
  17     const int expected_word_count;
  18   };
  19
  20   std::string QueryToString(const std::string& query);
  21
  22  protected:
  23   QueryParser query_parser_;
  24 };
  25
  26 // Test helper: Convert a user query string in 8-bit (for hardcoding
  27 // convenience) to a SQLite query string.
  28 std::string QueryParserTest::QueryToString(const std::string& query) {
  29   base::string16 sqlite_query;
  30   query_parser_.ParseQuery(base::UTF8ToUTF16(query), &sqlite_query);
  31   return base::UTF16ToUTF8(sqlite_query);
  32 }
  33
  34 // Basic multi-word queries, including prefix matching.
  35 TEST_F(QueryParserTest, SimpleQueries) {
  36   EXPECT_EQ("", QueryToString(" "));
  37   EXPECT_EQ("singleword*", QueryToString("singleword"));
  38   EXPECT_EQ("spacedout*", QueryToString("  spacedout "));
  39   EXPECT_EQ("foo* bar*", QueryToString("foo bar"));
  40   // Short words aren't prefix matches. For Korean Hangul
  41   // the minimum is 2 while for other scripts, it's 3.
  42   EXPECT_EQ("f b", QueryToString(" f b"));
  43   // KA JANG
  44   EXPECT_EQ(base::WideToUTF8(L"\xAC00 \xC7A5"),
  45             QueryToString(base::WideToUTF8(L" \xAC00 \xC7A5")));
  46   EXPECT_EQ("foo* bar*", QueryToString(" foo   bar "));
  47   // KA-JANG BICH-GO
  48   EXPECT_EQ(base::WideToUTF8(L"\xAC00\xC7A5* \xBE5B\xACE0*"),
  49             QueryToString(base::WideToUTF8(L"\xAC00\xC7A5 \xBE5B\xACE0")));
  50 }
  51
  52 // Quoted substring parsing.
  53 TEST_F(QueryParserTest, Quoted) {
  54   // ASCII quotes
  55   EXPECT_EQ("\"Quoted\"", QueryToString("\"Quoted\""));
  56   // Missing end quotes
  57   EXPECT_EQ("\"miss end\"", QueryToString("\"miss end"));
  58   // Missing begin quotes
  59   EXPECT_EQ("miss* beg*", QueryToString("miss beg\""));
  60   // Weird formatting
  61   EXPECT_EQ("\"Many\" \"quotes\"", QueryToString("\"Many   \"\"quotes"));
  62 }
  63
  64 // Apostrophes within words should be preserved, but otherwise stripped.
  65 TEST_F(QueryParserTest, Apostrophes) {
  66   EXPECT_EQ("foo* bar's*", QueryToString("foo bar's"));
  67   EXPECT_EQ("l'foo*", QueryToString("l'foo"));
  68   EXPECT_EQ("foo*", QueryToString("'foo"));
  69 }
  70
  71 // Special characters.
  72 TEST_F(QueryParserTest, SpecialChars) {
  73   EXPECT_EQ("foo* the* bar*", QueryToString("!#:/*foo#$*;'* the!#:/*bar"));
  74 }
  75
  76 TEST_F(QueryParserTest, NumWords) {
  77   TestData data[] = {
  78     { "blah",                  1 },
  79     { "foo \"bar baz\"",       3 },
  80     { "foo \"baz\"",           2 },
  81     { "foo \"bar baz\"  blah", 4 },
  82   };
  83
  84   for (size_t i = 0; i < arraysize(data); ++i) {
  85     base::string16 query_string;
  86     EXPECT_EQ(data[i].expected_word_count,
  87               query_parser_.ParseQuery(base::UTF8ToUTF16(data[i].input),
  88                                        &query_string));
  89   }
  90 }
  91
  92 TEST_F(QueryParserTest, ParseQueryNodesAndMatch) {
  93   struct TestData2 {
  94     const std::string query;
  95     const std::string text;
  96     const bool matches;
  97     const size_t m1_start;
  98     const size_t m1_end;
  99     const size_t m2_start;
 100     const size_t m2_end;
 101   } data[] = {
 102     { "foo",           "fooey foo",        true,  0, 3, 6, 9 },
 103     { "foo foo",       "foo",              true,  0, 3, 0, 0 },
 104     { "foo fooey",     "fooey",            true,  0, 5, 0, 0 },
 105     { "fooey foo",     "fooey",            true,  0, 5, 0, 0 },
 106     { "foo fooey bar", "bar fooey",        true,  0, 3, 4, 9 },
 107     { "blah",          "blah",             true,  0, 4, 0, 0 },
 108     { "blah",          "foo",              false, 0, 0, 0, 0 },
 109     { "blah",          "blahblah",         true,  0, 4, 0, 0 },
 110     { "blah",          "foo blah",         true,  4, 8, 0, 0 },
 111     { "foo blah",      "blah",             false, 0, 0, 0, 0 },
 112     { "foo blah",      "blahx foobar",     true,  0, 4, 6, 9 },
 113     { "\"foo blah\"",  "foo blah",         true,  0, 8, 0, 0 },
 114     { "\"foo blah\"",  "foox blahx",       false, 0, 0, 0, 0 },
 115     { "\"foo blah\"",  "foo blah",         true,  0, 8, 0, 0 },
 116     { "\"foo blah\"",  "\"foo blah\"",     true,  1, 9, 0, 0 },
 117     { "foo blah",      "\"foo bar blah\"", true,  1, 4, 9, 13 },
 118   };
 119   for (size_t i = 0; i < ARRAYSIZE_UNSAFE(data); ++i) {
 120     QueryParser parser;
 121     ScopedVector<QueryNode> query_nodes;
 122     parser.ParseQueryNodes(base::UTF8ToUTF16(data[i].query),
 123                            &query_nodes.get());
 124     Snippet::MatchPositions match_positions;
 125     ASSERT_EQ(data[i].matches,
 126               parser.DoesQueryMatch(base::UTF8ToUTF16(data[i].text),
 127                                     query_nodes.get(),
 128                                     &match_positions));
 129     size_t offset = 0;
 130     if (data[i].m1_start != 0 || data[i].m1_end != 0) {
 131       ASSERT_TRUE(match_positions.size() >= 1);
 132       EXPECT_EQ(data[i].m1_start, match_positions[0].first);
 133       EXPECT_EQ(data[i].m1_end, match_positions[0].second);
 134       offset++;
 135     }
 136     if (data[i].m2_start != 0 || data[i].m2_end != 0) {
 137       ASSERT_TRUE(match_positions.size() == 1 + offset);
 138       EXPECT_EQ(data[i].m2_start, match_positions[offset].first);
 139       EXPECT_EQ(data[i].m2_end, match_positions[offset].second);
 140     }
 141   }
 142 }
 143
 144 TEST_F(QueryParserTest, ParseQueryWords) {
 145   struct TestData2 {
 146     const std::string text;
 147     const std::string w1;
 148     const std::string w2;
 149     const std::string w3;
 150     const size_t word_count;
 151   } data[] = {
 152     { "foo",           "foo", "",    "",  1 },
 153     { "foo bar",       "foo", "bar", "",  2 },
 154     { "\"foo bar\"",   "foo", "bar", "",  2 },
 155     { "\"foo bar\" a", "foo", "bar", "a", 3 },
 156   };
 157   for (size_t i = 0; i < ARRAYSIZE_UNSAFE(data); ++i) {
 158     std::vector<base::string16> results;
 159     QueryParser parser;
 160     parser.ParseQueryWords(base::UTF8ToUTF16(data[i].text), &results);
 161     ASSERT_EQ(data[i].word_count, results.size());
 162     EXPECT_EQ(data[i].w1, base::UTF16ToUTF8(results[0]));
 163     if (results.size() == 2)
 164       EXPECT_EQ(data[i].w2, base::UTF16ToUTF8(results[1]));
 165     if (results.size() == 3)
 166       EXPECT_EQ(data[i].w3, base::UTF16ToUTF8(results[2]));
 167   }
 168 }
 169
 170 }  // namespace query_parser