Add and test uri detection.

3 files changed, 34 insertions, 34 deletions

diff --git a/searchlib/src/tests/index/docbuilder/docbuilder_test.cpp b/searchlib/src/tests/index/docbuilder/docbuilder_test.cpp
index 75cccb0d573..9f1027d0522 100644
--- a/searchlib/src/tests/index/docbuilder/docbuilder_test.cpp
+++ b/searchlib/src/tests/index/docbuilder/docbuilder_test.cpp
@@ -13,8 +13,7 @@ LOG_SETUP("docbuilder_test");
 using namespace document;
 using search::index::schema::CollectionType;
 
-namespace search {
-namespace index {
+namespace search::index {
 
 namespace
 {
@@ -26,15 +25,8 @@ namespace linguistics
 const vespalib::string SPANTREE_NAME("linguistics");
 }
 
-class Test : public vespalib::TestApp {
-private:
-    void testBuilder();
-public:
-    int Main() override;
-};
 
-void
-Test::testBuilder()
+TEST("test docBuilder")
 {
     Schema s;
     s.addIndexField(Schema::IndexField("ia", schema::DataType::STRING));
@@ -415,7 +407,7 @@ Test::testBuilder()
         EXPECT_EQUAL("</af>", *itr++);
         EXPECT_EQUAL("</document>", *itr++);
         EXPECT_TRUE(itr == lines.end());
-#if 1
+#if 0
         std::cout << "onedoc xml start -----" << std::endl <<
             xml << std::endl <<
             "-------" << std::endl;
@@ -479,7 +471,7 @@ Test::testBuilder()
         expSpans.push_back(Span(15, 9));
         expSpans.push_back(Span(15, 9));
         ASSERT_TRUE(expSpans == spans);
-#if 1
+#if 0
         std::cout << "onedoc xml start -----" << std::endl <<
             xml << std::endl <<
             "-------" << std::endl;
@@ -490,18 +482,21 @@ Test::testBuilder()
     }
 }
 
-int
-Test::Main()
-{
-    TEST_INIT("docbuilder_test");
-
-    testBuilder();
-
-    TEST_DONE();
+TEST("test if index names are valid uri parts") {
+    EXPECT_FALSE(UriField::mightBePartofUri("all"));
+    EXPECT_FALSE(UriField::mightBePartofUri("fragment"));
+    EXPECT_FALSE(UriField::mightBePartofUri(".all"));
+    EXPECT_FALSE(UriField::mightBePartofUri("all.b"));
+    EXPECT_TRUE(UriField::mightBePartofUri("b.all"));
+    EXPECT_TRUE(UriField::mightBePartofUri("b.scheme"));
+    EXPECT_TRUE(UriField::mightBePartofUri("b.host"));
+    EXPECT_TRUE(UriField::mightBePartofUri("b.port"));
+    EXPECT_TRUE(UriField::mightBePartofUri("b.hostname"));
+    EXPECT_TRUE(UriField::mightBePartofUri("b.path"));
+    EXPECT_TRUE(UriField::mightBePartofUri("b.query"));
+    EXPECT_TRUE(UriField::mightBePartofUri("b.fragment"));
 }
 
 }
-}
-
-TEST_APPHOOK(search::index::Test);
 
+TEST_MAIN() { TEST_RUN_ALL(); }
diff --git a/searchlib/src/vespa/searchlib/index/uri_field.cpp b/searchlib/src/vespa/searchlib/index/uri_field.cpp
index daf0e6e685e..070afc94837 100644
--- a/searchlib/src/vespa/searchlib/index/uri_field.cpp
+++ b/searchlib/src/vespa/searchlib/index/uri_field.cpp
@@ -18,9 +18,7 @@ UriField::UriField()
 }
 
 bool
-UriField::valid(const Schema &schema,
-                uint32_t fieldId,
-                const Schema::CollectionType &collectionType)
+UriField::valid(const Schema &schema, uint32_t fieldId, const Schema::CollectionType &collectionType)
 {
     if (fieldId == Schema::UNKNOWN_FIELD_ID) {
         return false;
@@ -36,9 +34,7 @@ UriField::valid(const Schema &schema,
 }
 
 bool
-UriField::broken(const Schema &schema,
-                 const Schema::CollectionType &
-                 collectionType) const
+UriField::broken(const Schema &schema, const Schema::CollectionType & collectionType) const
 {
     return !valid(schema, _all, collectionType) &&
         valid(schema, _scheme, collectionType) &&
@@ -50,9 +46,7 @@ UriField::broken(const Schema &schema,
 }
 
 bool
-UriField::valid(const Schema &schema,
-                const Schema::CollectionType &
-                collectionType) const
+UriField::valid(const Schema &schema, const Schema::CollectionType & collectionType) const
 {
     return valid(schema, _all, collectionType) &&
         valid(schema, _scheme, collectionType) &&
@@ -64,8 +58,7 @@ UriField::valid(const Schema &schema,
 }
 
 void
-UriField::setup(const Schema &schema,
-                const vespalib::string &field)
+UriField::setup(const Schema &schema, const vespalib::string &field)
 {
     _all = schema.getIndexFieldId(field);
     _scheme = schema.getIndexFieldId(field + ".scheme");
@@ -77,6 +70,17 @@ UriField::setup(const Schema &schema,
     _hostname = schema.getIndexFieldId(field + ".hostname");
 }
 
+bool
+UriField::mightBePartofUri(vespalib::stringref name) {
+    size_t dotPos = name.find('.');
+    if ((dotPos != 0) && (dotPos != vespalib::string::npos)) {
+        vespalib::stringref suffix = name.substr(dotPos + 1);
+        return ((suffix == "all") || (suffix == "scheme") || (suffix == "host") || (suffix == "port") ||
+                (suffix == "path") || (suffix == "query") || (suffix == "fragment") || (suffix == "hostname"));
+    }
+    return false;
+}
+
 void
 UriField::markUsed(UsedFieldsMap &usedFields, uint32_t field)
 {
diff --git a/searchlib/src/vespa/searchlib/index/uri_field.h b/searchlib/src/vespa/searchlib/index/uri_field.h
index 9b8e4e72b7c..70bf8c01a8c 100644
--- a/searchlib/src/vespa/searchlib/index/uri_field.h
+++ b/searchlib/src/vespa/searchlib/index/uri_field.h
@@ -35,6 +35,7 @@ public:
     bool valid(const Schema &schema, const Schema::CollectionType &collectionType) const;
     void setup(const Schema &schema, const vespalib::string &field);
     void markUsed(UsedFieldsMap &usedFields) const;
+    static bool mightBePartofUri(vespalib::stringref name);
 };
 
 }