qdataurl: make parsing the content-type more robust

Don't assume `charset` is the first parameter. The parameters
(attribute=value pairs) are delemited by a `;`. The order of the
parameters isn't specified, (except for `;base64` which is the last
one).

Add more tests. Add a test for image/png (.png file copied from
src/widgets/styles/images/arrow-down-16.png).

Change-Id: Ie3e45c607c093695d0c180e9a9783b2b02d7ef70
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>

src/corelib/io/qdataurl.cpp

@@ -17,6 +17,15 @@ using namespace Qt::Literals;
 */
 Q_CORE_EXPORT bool qDecodeDataUrl(const QUrl &uri, QString &mimeType, QByteArray &payload)
 {
+    /* https://www.rfc-editor.org/rfc/rfc2397.html
+
+        data:[<mediatype>][;base64],<data>
+        dataurl    := "data:" [ mediatype ] [ ";base64" ] "," data
+        mediatype  := [ type "/" subtype ] *( ";" parameter )
+        data       := *urlchar
+        parameter  := attribute "=" value
+    */
+
     if (uri.scheme() != "data"_L1 || !uri.host().isEmpty())
         return false;
 
@@ -48,20 +57,36 @@ Q_CORE_EXPORT bool qDecodeDataUrl(const QUrl &uri, QString &mimeType, QByteArray
             data.chop(base64.size());
         }
 
-        QLatin1StringView textPlain;
+        QLatin1StringView mime;
+        QLatin1StringView charsetParam;
         constexpr auto charset = "charset"_L1;
-        if (data.startsWith(charset, Qt::CaseInsensitive)) {
-            QLatin1StringView copy = data.sliced(charset.size());
-            while (copy.startsWith(u' '))
-                copy.slice(1);
-            if (copy.startsWith(u'='))
-                textPlain = "text/plain;"_L1;
+        bool first = true;
+        for (auto part : qTokenize(data, u';', Qt::SkipEmptyParts)) {
+            part = part.trimmed();
+            if (first) {
+                if (part.contains(u'/'))
+                    mime = part;
+                first = false;
+            }
+            // Minimal changes, e.g. if it's "charset=;" or "charset;" without
+            // an encoding, leave it as-is
+            if (part.startsWith(charset, Qt::CaseInsensitive))
+                charsetParam = part;
+
+            if (!mime.isEmpty() && !charsetParam.isEmpty())
+                break;
         }
 
-        if (!data.isEmpty())
-            mimeType = textPlain + data.trimmed();
+        if (mime.isEmpty()) {
+            mime = "text/plain"_L1;
+            if (charsetParam.isEmpty())
+                charsetParam = "charset=US-ASCII"_L1;
+        }
+        if (!charsetParam.isEmpty())
+            mimeType = mime + u';' + charsetParam;
         else
-            mimeType = QStringLiteral("text/plain;charset=US-ASCII");
+            mimeType = mime;
+
         return true;
     }
 

tests/auto/corelib/io/qdataurl/CMakeLists.txt

@@ -16,4 +16,5 @@ qt_internal_add_test(tst_qdataurl
         tst_qdataurl.cpp
     LIBRARIES
         Qt::CorePrivate
+    TESTDATA "arrow-down-16.png"
 )

tests/auto/corelib/io/qdataurl/tst_qdataurl.cpp

@@ -32,10 +32,14 @@ void tst_QDataUrl::decode_data()
     row("malformed-host2", "data://text/plain;charset=ISO-8859-1", false);
     row("malformed-host3", "data://test.com/,", false);
     row("malformed-no-comma", "data:text/plain", false);
+
+    constexpr auto defaultMimeType = "text/plain;charset=US-ASCII"_L1;
+
     row("emptyData-default-mimetype", "data:,", true,
         "text/plain;charset=US-ASCII"_L1, "");
     row("emptyData-only-charset", "data:charset=ISO-8859-1,", true,
         "text/plain;charset=ISO-8859-1"_L1, "");
+
     row("alreadyPercentageEncoded", "data:text/plain,%E2%88%9A", true,
         "text/plain"_L1, QByteArray::fromPercentEncoding("%E2%88%9A"));
     row("everythingIsCaseInsensitive", "Data:texT/PlaiN;charSet=iSo-8859-1;Base64,SGVsbG8=", true,
@@ -43,7 +47,28 @@ void tst_QDataUrl::decode_data()
     row("spacesAroundCharset", "data:%20charset%20%20=%20UTF-8,Hello", true,
         "text/plain;charset  = UTF-8"_L1, QByteArrayLiteral("Hello"));
     row("prematureCharsetEnd", "data:charset,", true,
-        "charset", ""); // nonsense result, but don't crash
+        "text/plain;charset"_L1, ""); // nonsense result, but don't crash
+    row("prematureCharsetEnd-no-encoding", "data:charset=,", true,
+        "text/plain;charset="_L1, ""); // nonsense result, but don't crash
+    row("charset-value-as-quoted-string", "data:charset=%22UTF-8%22,Hello", true,
+        "text/plain;charset=\"UTF-8\""_L1, "Hello"_ba);
+
+    row("extraparameter-before-charset", "data:;extraparameter=foo;charset=ISO-8859-1,Hello", true,
+        "text/plain;charset=ISO-8859-1", "Hello");
+    row("extraparameter-after-charset", "data:charset=ISO-8859-1;extraparameter=foo,Hello", true,
+        "text/plain;charset=ISO-8859-1", "Hello");
+    row("content-type-parsing-slash-in-mimetype",
+        "data:charset=UTF-8;alternate=\"application/octet-stream\";bar=baz,", true,
+        "text/plain;charset=UTF-8", "");
+    row("not-real-charset", "data:incharsetter=true,", true, defaultMimeType, "");
+
+    QString path = QFINDTESTDATA("arrow-down-16.png");
+    QFile img(path);
+    QVERIFY(img.open(QFile::ReadOnly));
+    QByteArray imageData = img.readAll();
+    QByteArray base64 = imageData.toBase64();
+    row("image-png", "data:image/png;base64," + base64, true,
+        "image/png", imageData);
 }
 
 void tst_QDataUrl::decode()