Improve parsing of HTTP headers

Parse HTTP headers using raw byte arrays instead of strings. This allows us to apply different encodings for different parts.
This change is backward compatible and should not affect any existing operation, so WebAPI version bump is not required.

PR #23083.

src/base/http/requestparser.cpp

@@ -31,12 +31,13 @@
 #include "requestparser.h"
 
 #include <algorithm>
+#include <optional>
 #include <utility>
 
+#include <QByteArrayList>
 #include <QByteArrayView>
 #include <QDebug>
 #include <QRegularExpression>
-#include <QStringList>
 #include <QUrl>
 #include <QUrlQuery>
 
@@ -59,21 +60,19 @@ namespace
         return in;
     }
 
-    bool parseHeaderLine(const QStringView line, HeaderMap &out)
+    std::optional<QStringPair> parseHeaderLine(const QByteArrayView line)
     {
         // [rfc7230] 3.2. Header Fields
         const int i = line.indexOf(u':');
         if (i <= 0)
         {
             qWarning() << Q_FUNC_INFO << "invalid http header:" << line;
-            return false;
+            return std::nullopt;
         }
 
-        const QString name = line.first(i).trimmed().toString().toLower();
-        const QString value = line.sliced(i + 1).trimmed().toString();
-        out[name] = value;
-
-        return true;
+        const QString name = QString::fromLatin1(line.first(i).trimmed()).toLower();
+        const QString value = QString::fromLatin1(line.sliced(i + 1).trimmed());
+        return {{name, value}};
     }
 }
 
@@ -93,7 +92,7 @@ RequestParser::ParseResult RequestParser::doParse(const QByteArrayView data)
         return {ParseStatus::Incomplete, Request(), 0};
     }
 
-    const QString httpHeaders = QString::fromLatin1(data.constData(), headerEnd);
+    const QByteArrayView httpHeaders = data.first(headerEnd);
     if (!parseStartLines(httpHeaders))
     {
         qWarning() << Q_FUNC_INFO << "header parsing error";
@@ -152,36 +151,40 @@ RequestParser::ParseResult RequestParser::doParse(const QByteArrayView data)
     return {ParseStatus::BadMethod, m_request, 0};
 }
 
-bool RequestParser::parseStartLines(const QStringView data)
+bool RequestParser::parseStartLines(const QByteArrayView data)
 {
     // we don't handle malformed request which uses `LF` for newline
-    const QList<QStringView> lines = data.split(QString::fromLatin1(CRLF), Qt::SkipEmptyParts);
+    const QList<QByteArrayView> lines = splitToViews(data, CRLF, Qt::SkipEmptyParts);
 
     // [rfc7230] 3.2.2. Field Order
-    QStringList requestLines;
+    QByteArrayList requestLines;
     for (const auto &line : lines)
     {
-        if (line.at(0).isSpace() && !requestLines.isEmpty())
+        if (QChar::fromLatin1(line.at(0)).isSpace() && !requestLines.isEmpty())
         {
             // continuation of previous line
             requestLines.last() += line;
         }
         else
         {
-            requestLines += line.toString();
+            requestLines += line.toByteArray();
         }
     }
 
     if (requestLines.isEmpty())
         return false;
 
-    if (!parseRequestLine(requestLines[0]))
+    if (!parseRequestLine(QString::fromLatin1(requestLines[0])))
         return false;
 
     for (auto i = ++(requestLines.begin()); i != requestLines.end(); ++i)
     {
-        if (!parseHeaderLine(*i, m_request.headers))
+        const std::optional<QStringPair> header = parseHeaderLine(*i);
+        if (!header.has_value())
             return false;
+
+        const auto [name, value] = header.value();
+        m_request.headers[name] = value;
     }
 
     return true;
@@ -310,17 +313,23 @@ bool RequestParser::parseFormData(const QByteArrayView data)
         return false;
     }
 
-    const QString headers = QString::fromLatin1(data.first(eohPos));
+    const QByteArrayView headers = data.first(eohPos);
     const QByteArrayView payload = viewWithoutEndingWith(data.sliced((eohPos + EOH.size())), CRLF);
 
     HeaderMap headersMap;
-    const QList<QStringView> headerLines = QStringView(headers).split(QString::fromLatin1(CRLF), Qt::SkipEmptyParts);
+    const QList<QByteArrayView> headerLines = splitToViews(headers, CRLF, Qt::SkipEmptyParts);
     for (const auto &line : headerLines)
     {
-        if (line.trimmed().startsWith(HEADER_CONTENT_DISPOSITION, Qt::CaseInsensitive))
+        const std::optional<QStringPair> header = parseHeaderLine(line);
+        if (!header.has_value())
+            return false;
+
+        const auto [name, value] = header.value();
+
+        if (name == HEADER_CONTENT_DISPOSITION)
         {
             // extract out filename & name
-            const QList<QStringView> directives = line.split(u';', Qt::SkipEmptyParts);
+            const QList<QByteArrayView> directives = splitToViews(line, ";", Qt::SkipEmptyParts);
 
             for (const auto &directive : directives)
             {
@@ -328,15 +337,14 @@ bool RequestParser::parseFormData(const QByteArrayView data)
                 if (idx < 0)
                     continue;
 
-                const QString name = directive.first(idx).trimmed().toString().toLower();
-                const QString value = Utils::String::unquote(directive.sliced(idx + 1).trimmed()).toString();
+                const QString name = QString::fromLatin1(directive.first(idx).trimmed()).toLower();
+                const QString value = QString::fromLatin1(unquote(directive.sliced(idx + 1).trimmed()));
                 headersMap[name] = value;
             }
         }
         else
         {
-            if (!parseHeaderLine(line, headersMap))
-                return false;
+            headersMap[name] = value;
         }
     }
 

src/base/http/requestparser.h

@@ -61,7 +61,7 @@ namespace Http
         RequestParser() = default;
 
         ParseResult doParse(QByteArrayView data);
-        bool parseStartLines(QStringView data);
+        bool parseStartLines(QByteArrayView data);
         bool parseRequestLine(const QString &line);
 
         bool parsePostMessage(QByteArrayView data);

src/base/utils/bytearray.cpp

@@ -31,7 +31,6 @@
 
 #include <QByteArray>
 #include <QByteArrayMatcher>
-#include <QByteArrayView>
 #include <QList>
 
 QList<QByteArrayView> Utils::ByteArray::splitToViews(const QByteArrayView in, const QByteArrayView sep, const Qt::SplitBehavior behavior)

src/base/utils/bytearray.h

@@ -31,9 +31,9 @@
 
 #include <Qt>
 #include <QtContainerFwd>
+#include <QByteArrayView>
 
 class QByteArray;
-class QByteArrayView;
 
 namespace Utils::ByteArray
 {
@@ -42,4 +42,19 @@ namespace Utils::ByteArray
     QByteArray asQByteArray(QByteArrayView view);
 
     QByteArray toBase32(const QByteArray &in);
+
+    template <typename T>
+    T unquote(const T &arr, const QByteArrayView quotes = "\"")
+    {
+        if (arr.length() < 2)
+            return arr;
+
+        for (const char quote : quotes)
+        {
+            if (arr.startsWith(quote) && arr.endsWith(quote))
+                return arr.sliced(1, (arr.length() - 2));
+        }
+
+        return arr;
+    }
 }

test/testutilsbytearray.cpp

@@ -28,6 +28,7 @@
  */
 
 #include <QByteArray>
+#include <QByteArrayView>
 #include <QLatin1StringView>
 #include <QObject>
 #include <QTest>
@@ -123,6 +124,33 @@ private slots:
         QCOMPARE(Utils::ByteArray::toBase32("0000000000"), "GAYDAMBQGAYDAMBQ");
         QCOMPARE(Utils::ByteArray::toBase32("1"), "GE======");
     }
+
+    void testUnquote() const
+    {
+        const auto test = []<typename T>()
+        {
+            QCOMPARE(Utils::ByteArray::unquote<T>({}), {});
+            QCOMPARE(Utils::ByteArray::unquote<T>("abc"), "abc");
+            QCOMPARE(Utils::ByteArray::unquote<T>("\"abc\""), "abc");
+            QCOMPARE(Utils::ByteArray::unquote<T>("\"a b c\""), "a b c");
+            QCOMPARE(Utils::ByteArray::unquote<T>("\"abc"), "\"abc");
+            QCOMPARE(Utils::ByteArray::unquote<T>("abc\""), "abc\"");
+            QCOMPARE(Utils::ByteArray::unquote<T>(" \"abc\" "), " \"abc\" ");
+            QCOMPARE(Utils::ByteArray::unquote<T>("\"a\"bc\""), "a\"bc");
+            QCOMPARE(Utils::ByteArray::unquote<T>("'abc'", "'"), "abc");
+            QCOMPARE(Utils::ByteArray::unquote<T>("'abc'", "\"'"), "abc");
+            QCOMPARE(Utils::ByteArray::unquote<T>("\"'abc'\"", "\"'"), "'abc'");
+            QCOMPARE(Utils::ByteArray::unquote<T>("\"'abc'\"", "'\""), "'abc'");
+            QCOMPARE(Utils::ByteArray::unquote<T>("\"'abc'\"", "'"), "\"'abc'\"");
+            QCOMPARE(Utils::ByteArray::unquote<T>("\"abc'", "'"), "\"abc'");
+            QCOMPARE(Utils::ByteArray::unquote<T>("'abc\"", "'"), "'abc\"");
+            QCOMPARE(Utils::ByteArray::unquote<T>("\"\""), "");
+            QCOMPARE(Utils::ByteArray::unquote<T>("\""), "\"");
+        };
+
+        test.template operator()<QByteArray>();
+        test.template operator()<QByteArrayView>();
+    }
 };
 
 QTEST_APPLESS_MAIN(TestUtilsByteArray)