[core] add util::isURL() for checking whether a string starts with a URL scheme

4 files changed, 60 insertions, 2 deletions

diff --git a/cmake/test-files.cmake b/cmake/test-files.cmake
index ceff50ef64..7d8dc82756 100644
--- a/cmake/test-files.cmake
+++ b/cmake/test-files.cmake
@@ -112,5 +112,6 @@ set(MBGL_TEST_FILES
     test/util/tile_cover.test.cpp
     test/util/timer.test.cpp
     test/util/token.test.cpp
+    test/util/url.test.cpp
     test/util/work_queue.test.cpp
 )
diff --git a/src/mbgl/util/url.cpp b/src/mbgl/util/url.cpp
index bf6fc70ff5..9831bc9038 100644
--- a/src/mbgl/util/url.cpp
+++ b/src/mbgl/util/url.cpp
@@ -1,12 +1,30 @@
 #include <mbgl/util/url.hpp>
 
-#include <cctype>
 #include <iomanip>
 #include <sstream>
 #include <string>
 #include <cstdlib>
 #include <algorithm>
 
+
+namespace {
+
+// std::alnum etc. suffer from locale-dependence.
+
+inline bool isAlphaCharacter(char c) {
+    return (c >= 'a' && c <= 'z') || (c >= 'A' && c <= 'Z');
+}
+
+inline bool isAlphaNumericCharacter(char c) {
+    return isAlphaCharacter(c) || (c >= '0' && c <= '9');
+}
+
+inline bool isSchemeCharacter(char c) {
+    return isAlphaNumericCharacter(c) || c == '-' || c == '+' || c == '.';
+}
+
+} // namespace
+
 namespace mbgl {
 namespace util {
 
@@ -17,7 +35,7 @@ std::string percentEncode(const std::string& input) {
     encoded << std::hex;
 
     for (auto c : input) {
-        if (std::isalnum(c) || c == '-' || c == '_' || c == '.' || c == '~') {
+        if (isAlphaNumericCharacter(c) || c == '-' || c == '_' || c == '.' || c == '~') {
             encoded << c;
         } else {
             encoded << '%' << std::setw(2) << int(c);
@@ -47,5 +65,18 @@ std::string percentDecode(const std::string& input) {
     return decoded;
 }
 
+// Checks whether the input string contains ://, and the part before it is all alphanumeric ASCII.
+bool isURL(const std::string& input) {
+    auto it = input.begin();
+    // First character has to be alphabetic
+    if (it == input.end() || !isAlphaCharacter(*it++)) return false;
+    // The remaining characters of the scheme can be alphanumeric, or be one of +.-
+    while (it != input.end() && isSchemeCharacter(*it)) ++it;
+    // Check that :// follows
+    return (it != input.end() && *it++ == ':') &&
+           (it != input.end() && *it++ == '/') &&
+           (it != input.end() && *it++ == '/');
+}
+
 } // namespace util
 } // namespace mbgl
diff --git a/src/mbgl/util/url.hpp b/src/mbgl/util/url.hpp
index b9c66261df..2d18f7476a 100644
--- a/src/mbgl/util/url.hpp
+++ b/src/mbgl/util/url.hpp
@@ -8,6 +8,7 @@ namespace util {
 
 std::string percentEncode(const std::string&);
 std::string percentDecode(const std::string&);
+bool isURL(const std::string&);
 
 } // namespace util
 } // namespace mbgl
diff --git a/test/util/url.test.cpp b/test/util/url.test.cpp
new file mode 100644
index 0000000000..c0ee30efab
--- /dev/null
+++ b/test/util/url.test.cpp
@@ -0,0 +1,25 @@
+#include <mbgl/test/util.hpp>
+
+#include <mbgl/util/url.hpp>
+
+#include <memory>
+
+using namespace mbgl::util;
+
+TEST(URL, isURL) {
+    EXPECT_TRUE(isURL("mapbox://foo"));
+    EXPECT_TRUE(isURL("mapbox://"));
+    EXPECT_TRUE(isURL("mapbox-test-scheme://foo"));
+    EXPECT_TRUE(isURL("mapbox+style://foo"));
+    EXPECT_TRUE(isURL("mapbox-2.0://foo"));
+    EXPECT_TRUE(isURL("mapbox99://foo"));
+
+    EXPECT_FALSE(isURL("mapbox:/"));
+    EXPECT_FALSE(isURL(" mapbox://"));
+    EXPECT_FALSE(isURL("://"));
+    EXPECT_FALSE(isURL("mapbox"));
+    EXPECT_FALSE(isURL("mapbox:foo"));
+    EXPECT_FALSE(isURL("mapbox:/foo"));
+    EXPECT_FALSE(isURL("test/mapbox://foo"));
+    EXPECT_FALSE(isURL("123://foo"));
+}