Add GetExtensionOrDomain()

Blake-Madden · Apr 14, 2024 · 1791153 · 1791153
1 parent 227d356
commit 1791153
Show file tree

Hide file tree

Showing 3 changed files with 79 additions and 4 deletions.
diff --git a/src/util/fileutil.cpp b/src/util/fileutil.cpp
@@ -810,3 +810,49 @@ wxString CreateNewFileName(const wxString& filePath)
 
     return wxString{};
     }
+
+//---------------------------------------------------
+wxString GetExtensionOrDomain(const wxString& url)
+    {
+    if (url.empty())
+        {
+        return wxString{};
+        }
+
+    const size_t lastSlash = url.rfind(L'/');
+    const size_t startPos = ((lastSlash != wxString::npos) ? (lastSlash + 1) : 0);
+
+    // Any sort of page with a query.
+    // Note that some pages are malformed and missing the variable assignment,
+    // so only look for the initial query (i.e., the '?') and go back from there.
+    if (const size_t queryPos = url.find(L'?', startPos); queryPos != wxString::npos)
+        {
+        // might be a JS, CSS, or other extension, so get the real extension
+        // in front of the query...
+        const wxFileName fn(url.substr(startPos, queryPos - startPos));
+        if (fn.HasExt())
+            {
+            return fn.GetExt();
+            }
+        // sometimes, the "webpage" name is simply "js" or something like that,
+        // so treat it as such if it has a query being passed to it
+        else if (fn.GetName().CmpNoCase(L"js") == 0 || fn.GetName().CmpNoCase(L"css") == 0)
+            {
+            return fn.GetName();
+            }
+        }
+    else
+        {
+        if (lastSlash == url.length() - 1)
+            {
+            return wxString{};
+            }
+        const wxFileName fn(url.substr(startPos));
+        if (fn.HasExt())
+            {
+            return fn.GetExt();
+            }
+        }
+
+    return wxString{};
+    }
diff --git a/src/util/fileutil.h b/src/util/fileutil.h
@@ -307,6 +307,11 @@ bool MoveDirectory(const wxString& fromDirectory, const wxString& toDirectory);
 [[nodiscard]]
 wxString CreateNewFileName(const wxString& filePath);
 
+/** @returns The extension (or simply domain) from an URL.
+    @param url The URL to parse.*/
+[[nodiscard]]
+wxString GetExtensionOrDomain(const wxString& url);
+
 /** @}*/
 
 #endif //__FILE_UTIL_H__
diff --git a/tests/gui-tests/fileutiltests.cpp b/tests/gui-tests/fileutiltests.cpp
@@ -2,17 +2,41 @@
 #include <catch2/matchers/catch_matchers_floating_point.hpp>
 #include "../../src/util/fileutil.h"
 
+TEST_CASE("Web page extensions", "[file]")
+    {
+    CHECK(GetExtensionOrDomain(L"").empty());
+    CHECK(GetExtensionOrDomain(L"/").empty());
+    CHECK(GetExtensionOrDomain(L"//").empty());
+    CHECK(GetExtensionOrDomain(L"business").empty());
+    CHECK(GetExtensionOrDomain(L"/business").empty());
+    CHECK(GetExtensionOrDomain(L"org").empty());
+
+    CHECK(GetExtensionOrDomain(L"business.org") == wxString{ "org" });
+    CHECK(GetExtensionOrDomain(L"/business.org") == wxString{ "org" });
+    CHECK(GetExtensionOrDomain(L"www.mycompany.com/business.doc") ==
+          wxString{ "doc" });
+    CHECK(GetExtensionOrDomain(L"www.mycompany.com/business").empty());
+
+    CHECK(GetExtensionOrDomain(L"www.mycompany.com/business.doc?5145") == wxString{ "doc" });
+    CHECK(GetExtensionOrDomain(L"www.mycompany.com/business.js?5145") == wxString{ "js" });
+    CHECK(GetExtensionOrDomain(L"www.mycompany.com/business.js?id=501") == wxString{ "js" });
+    CHECK(GetExtensionOrDomain(L"www.mycompany.com/business.js?id=501;l=en") == wxString{ "js" });
+    CHECK(GetExtensionOrDomain(L"www.mycompany.com/business.js.php?id=501;l=en") == wxString{ "php" });
+    CHECK(GetExtensionOrDomain(L"www.mycompany.com/js?5") == wxString{ "js" });
+    CHECK(GetExtensionOrDomain(L"www.mycompany.com/css?en") == wxString{ "css" });
+    }
+
 TEST_CASE("Parse title from file name", "[parse-title]")
     {
     SECTION("Local File")
         {
-        CHECK(ParseTitleFromFileName(L"c:\\files\\file.txt") == wxT("file"));
-        CHECK(ParseTitleFromFileName(L"/users/files/file.txt") == wxT("file"));
+        CHECK(ParseTitleFromFileName(L"c:\\files\\file.txt") == wxString("file"));
+        CHECK(ParseTitleFromFileName(L"/users/files/file.txt") == wxString("file"));
         }
     SECTION("Local File Illegal Chars")
         {
-        CHECK(ParseTitleFromFileName(L"c:\\files\\fi?l*e.txt") == wxT("file"));
-        CHECK(ParseTitleFromFileName(L"/users/files/fi?l*e.txt") == wxT("file"));
+        CHECK(ParseTitleFromFileName(L"c:\\files\\fi?l*e.txt") == wxString("file"));
+        CHECK(ParseTitleFromFileName(L"/users/files/fi?l*e.txt") == wxString("file"));
         }
     SECTION("Url Ending With Slash")
         {