Changeset 201863 in webkit

Timestamp:

Jun 9, 2016, 7:54:50 AM (10 years ago)

Author:

msaboff@apple.com

Message:

WebKitTestRunner and DumpRenderTree do not handle dangling surrogate characters
​https://bugs.webkit.org/show_bug.cgi?id=154863

Reviewed by Alexey Proskuryakov.

Source/WebKit2:

Added a non-strict verions of WKStringGetUTF8CString() that will handle dangling
surrogates called WKStringGetUTF8CStringNonStrict().

• Shared/API/c/WKString.cpp:

(WKStringGetUTF8CStringImpl):
(WKStringGetUTF8CString):
(WKStringGetUTF8CStringNonStrict):

• Shared/API/c/WKString.h:

Tools:

Added a non-strict verions of WKStringGetUTF8CString() that will handle dangling
surrogates. Changed the extraction of inner text from frames in DumpRenderTree
to use the new WKStringGetUTF8CStringNonStrict() function instead of NSString
conversion since NSString doesn't have a way to handle dangling surrogates.
The code added in DumpRenderTree matches what was changed in WebKitTestRunner.

• DumpRenderTree/mac/DumpRenderTree.mm:

(dumpFramesAsText):

• WebKitTestRunner/StringFunctions.h:

(WTR::toWTFString):

LayoutTests:

New tests.

• fast/text/dangling-surrogates-expected.txt: Added.
• fast/text/dangling-surrogates.html: Added.

Location:

trunk

Files:

• LayoutTests/ChangeLog (modified) (1 diff)
• LayoutTests/fast/text/dangling-surrogates-expected.txt (added)
• LayoutTests/fast/text/dangling-surrogates.html (added)
• Source/WebKit2/ChangeLog (modified) (1 diff)
• Source/WebKit2/Shared/API/c/WKString.cpp (modified) (3 diffs)
• Source/WebKit2/Shared/API/c/WKString.h (modified) (1 diff)
• Tools/ChangeLog (modified) (1 diff)
• Tools/DumpRenderTree/mac/DumpRenderTree.mm (modified) (2 diffs)
• Tools/WebKitTestRunner/StringFunctions.h (modified) (1 diff)

trunk/LayoutTests/ChangeLog

@@ +1 @@
+2016-06-09  Michael Saboff  <msaboff@apple.com>
+
+        WebKitTestRunner and DumpRenderTree do not handle dangling surrogate characters
+        https://bugs.webkit.org/show_bug.cgi?id=154863
+
+        Reviewed by Alexey Proskuryakov.
+
+        New tests.
+
+        * fast/text/dangling-surrogates-expected.txt: Added.
+        * fast/text/dangling-surrogates.html: Added.
+
 2016-06-09  Commit Queue  <commit-queue@webkit.org>
 

trunk/Source/WebKit2/ChangeLog

@@ +1 @@
+2016-06-09  Michael Saboff  <msaboff@apple.com>
+
+        WebKitTestRunner and DumpRenderTree do not handle dangling surrogate characters
+        https://bugs.webkit.org/show_bug.cgi?id=154863
+
+        Reviewed by Alexey Proskuryakov.
+
+        Added a non-strict verions of WKStringGetUTF8CString() that will handle dangling
+        surrogates called WKStringGetUTF8CStringNonStrict().
+
+        * Shared/API/c/WKString.cpp:
+        (WKStringGetUTF8CStringImpl):
+        (WKStringGetUTF8CString):
+        (WKStringGetUTF8CStringNonStrict):
+        * Shared/API/c/WKString.h:
+
 2016-06-09  Carlos Garcia Campos  <cgarcia@igalia.com>
 

trunk/Source/WebKit2/Shared/API/c/WKString.cpp

@@ -71 +71 @@
 }
 
-size_t WKStringGetUTF8CString(WKStringRef stringRef, char* buffer, size_t bufferSize)
+enum StrictType { NonStrict = false, Strict = true };
+
+template <StrictType strict>
+size_t WKStringGetUTF8CStringImpl(WKStringRef stringRef, char* buffer, size_t bufferSize)
 {
     if (!bufferSize)
@@ -86 +89 @@
     } else {
         const UChar* characters = stringView.characters16();
-        result = WTF::Unicode::convertUTF16ToUTF8(&characters, characters + stringView.length(), &p, p + bufferSize - 1, /* strict */ true);
+        result = WTF::Unicode::convertUTF16ToUTF8(&characters, characters + stringView.length(), &p, p + bufferSize - 1, strict);
     }
 
@@ -94 +97 @@
     *p++ = '\0';
     return p - buffer;
+}
+
+size_t WKStringGetUTF8CString(WKStringRef stringRef, char* buffer, size_t bufferSize)
+{
+    return WKStringGetUTF8CStringImpl<StrictType::Strict>(stringRef, buffer, bufferSize);
+}
+
+size_t WKStringGetUTF8CStringNonStrict(WKStringRef stringRef, char* buffer, size_t bufferSize)
+{
+    return WKStringGetUTF8CStringImpl<StrictType::NonStrict>(stringRef, buffer, bufferSize);
 }
 

trunk/Source/WebKit2/Shared/API/c/WKString.h

@@ -56 +56 @@
 WK_EXPORT size_t WKStringGetMaximumUTF8CStringSize(WKStringRef string);
 WK_EXPORT size_t WKStringGetUTF8CString(WKStringRef string, char* buffer, size_t bufferSize);
+WK_EXPORT size_t WKStringGetUTF8CStringNonStrict(WKStringRef string, char* buffer, size_t bufferSize);
 
 WK_EXPORT bool WKStringIsEqual(WKStringRef a, WKStringRef b);

trunk/Tools/ChangeLog

@@ +1 @@
+2016-06-09  Michael Saboff  <msaboff@apple.com>
+
+        WebKitTestRunner and DumpRenderTree do not handle dangling surrogate characters
+        https://bugs.webkit.org/show_bug.cgi?id=154863
+
+        Reviewed by Alexey Proskuryakov.
+
+        Added a non-strict verions of WKStringGetUTF8CString() that will handle dangling
+        surrogates.  Changed the extraction of inner text from frames in DumpRenderTree
+        to use the new WKStringGetUTF8CStringNonStrict() function instead of NSString
+        conversion since NSString doesn't have a way to handle dangling surrogates.
+        The code added in DumpRenderTree matches what was changed in WebKitTestRunner.
+
+        * DumpRenderTree/mac/DumpRenderTree.mm:
+        (dumpFramesAsText):
+        * WebKitTestRunner/StringFunctions.h:
+        (WTR::toWTFString):
+
 2016-06-08  Hunseop Jeong  <hs85.jeong@samsung.com>
 

trunk/Tools/DumpRenderTree/mac/DumpRenderTree.mm

@@ -63 +63 @@
 #import <WebKit/DOMExtensions.h>
 #import <WebKit/DOMRange.h>
+#import <WebKit/WKString.h>
+#import <WebKit/WKStringCF.h>
 #import <WebKit/WebArchive.h>
 #import <WebKit/WebBackForwardList.h>
@@ -1524 +1526 @@
         result = [NSMutableString stringWithFormat:@"\n--------\nFrame: '%@'\n--------\n", [frame name]];
 
-    [result appendFormat:@"%@\n", [documentElement innerText]];
+    NSString *innerText = [documentElement innerText];
+    // We use WKStringGetUTF8CStringNonStrict() to convert innerText to a WK String since
+    // WKStringGetUTF8CStringNonStrict() can handle dangling surrogates and the NSString
+    // conversion methods cannot. After the conversion to a buffer, we turn that buffer into
+    // a CFString via fromUTF8WithLatin1Fallback().createCFString() which can be appended to
+    // the result without any conversion.
+    WKStringRef stringRef = WKStringCreateWithCFString((CFStringRef)innerText);
+    size_t bufferSize = WKStringGetMaximumUTF8CStringSize(stringRef);
+    auto buffer = std::make_unique<char[]>(bufferSize);
+    size_t stringLength = WKStringGetUTF8CStringNonStrict(stringRef, buffer.get(), bufferSize);
+    [result appendFormat:@"%@\n", String::fromUTF8WithLatin1Fallback(buffer.get(), stringLength - 1).createCFString().get()];
 
     if (gTestRunner->dumpChildFramesAsText()) {

trunk/Tools/WebKitTestRunner/StringFunctions.h

@@ -88 +88 @@
     size_t bufferSize = WKStringGetMaximumUTF8CStringSize(string);
     auto buffer = std::make_unique<char[]>(bufferSize);
-    size_t stringLength = WKStringGetUTF8CString(string, buffer.get(), bufferSize);
+    size_t stringLength = WKStringGetUTF8CStringNonStrict(string, buffer.get(), bufferSize);
     return WTF::String::fromUTF8WithLatin1Fallback(buffer.get(), stringLength - 1);
 }