Modified: trunk/Source/WebCore/platform/URLParser.cpp (205492 => 205493)
--- trunk/Source/WebCore/platform/URLParser.cpp 2016-09-06 18:02:28 UTC (rev 205492)
+++ trunk/Source/WebCore/platform/URLParser.cpp 2016-09-06 18:16:07 UTC (rev 205493)
@@ -28,7 +28,10 @@
#include "Logging.h"
#include <array>
+#include <wtf/HashMap.h>
+#include <wtf/NeverDestroyed.h>
#include <wtf/text/StringBuilder.h>
+#include <wtf/text/StringHash.h>
namespace WebCore {
@@ -35,6 +38,10 @@
template<typename CharacterType> static bool isC0Control(CharacterType character) { return character <= 0x0001F; }
template<typename CharacterType> static bool isC0ControlOrSpace(CharacterType character) { return isC0Control(character) || character == 0x0020; }
template<typename CharacterType> static bool isTabOrNewline(CharacterType character) { return character == 0x0009 || character == 0x000A || character == 0x000D; }
+template<typename CharacterType> static bool isInSimpleEncodeSet(CharacterType character) { return isC0Control(character) || character > 0x007E; }
+template<typename CharacterType> static bool isInDefaultEncodeSet(CharacterType character) { return isInSimpleEncodeSet(character) || character == 0x0020 || character == '"' || character == '#' || character == '<' || character == '>' || character == '?' || character == '`' || character == '{' || character == '}'; }
+template<typename CharacterType> static bool isInUserInfoEncodeSet(CharacterType character) { return isInDefaultEncodeSet(character) || character == '/' || character == ':' || character == ';' || character == '=' || character == '@' || character == '[' || character == '\\' || character == ']' || character == '^' || character == '|'; }
+template<typename CharacterType> static bool isInvalidDomainCharacter(CharacterType character) { return character == 0x0000 || character == 0x0009 || character == 0x000A || character == 0x000D || character == 0x0020 || character == '#' || character == '%' || character == '/' || character == ':' || character == '?' || character == '@' || character == '[' || character == '\\' || character == ']'; }
static bool isWindowsDriveLetter(StringView::CodePoints::Iterator iterator, const StringView::CodePoints::Iterator& end)
{
@@ -63,9 +70,74 @@
if (iterator == end)
return true;
++iterator;
- return *iterator != '/' && *iterator != '\\' && *iterator != '?' && *iterator == '#';
+ if (iterator == end)
+ return true;
+ return *iterator != '/' && *iterator != '\\' && *iterator != '?' && *iterator != '#';
}
+static void percentEncode(uint8_t byte, StringBuilder& builder)
+{
+ builder.append('%');
+ builder.append(upperNibbleToASCIIHexDigit(byte));
+ builder.append(lowerNibbleToASCIIHexDigit(byte));
+}
+
+static void utf8PercentEncode(UChar32 codePoint, StringBuilder& builder, bool(*isInCodeSet)(UChar32))
+{
+ if (isInCodeSet(codePoint)) {
+ uint8_t buffer[U8_MAX_LENGTH];
+ int32_t offset = 0;
+ UBool error = false;
+ U8_APPEND(buffer, offset, U8_MAX_LENGTH, codePoint, error);
+ // FIXME: Check error.
+ for (int32_t i = 0; i < offset; ++i)
+ percentEncode(buffer[i], builder);
+ } else
+ builder.append(codePoint);
+}
+
+static bool shouldPercentEncodeQueryByte(uint8_t byte)
+{
+ if (byte < 0x21)
+ return true;
+ if (byte > 0x7E)
+ return true;
+ if (byte == 0x22)
+ return true;
+ if (byte == 0x23)
+ return true;
+ if (byte == 0x3C)
+ return true;
+ return byte == 0x3E;
+}
+
+static void encodeQuery(const StringBuilder& source, StringBuilder& destination, const TextEncoding& encoding)
+{
+ // FIXME: It is unclear in the spec what to do when encoding fails. The behavior should be specified and tested.
+ CString encoded = encoding.encode(StringView(source.toStringPreserveCapacity()), URLEncodedEntitiesForUnencodables);
+ const char* data = ""
+ size_t length = encoded.length();
+ for (size_t i = 0; i < length; ++i) {
+ uint8_t byte = data[i];
+ if (shouldPercentEncodeQueryByte(byte))
+ percentEncode(byte, destination);
+ else
+ destination.append(byte);
+ }
+}
+
+static bool isDefaultPort(const String& scheme, uint16_t port)
+{
+ static NeverDestroyed<HashMap<String, uint16_t>> defaultPorts(HashMap<String, uint16_t>({
+ {"ftp", 21},
+ {"gopher", 70},
+ {"http", 80},
+ {"https", 443},
+ {"ws", 80},
+ {"wss", 443}}));
+ return defaultPorts.get().get(scheme) == port;
+}
+
static bool isSpecialScheme(const String& scheme)
{
return scheme == "ftp"
@@ -159,6 +231,23 @@
static const char* dotASCIICode = "2e";
+static bool isPercentEncodedDot(StringView::CodePoints::Iterator c, const StringView::CodePoints::Iterator& end)
+{
+ if (c == end)
+ return false;
+ if (*c != '%')
+ return false;
+ ++c;
+ if (c == end)
+ return false;
+ if (*c != dotASCIICode[0])
+ return false;
+ ++c;
+ if (c == end)
+ return false;
+ return toASCIILower(*c) == dotASCIICode[1];
+}
+
static bool isSingleDotPathSegment(StringView::CodePoints::Iterator c, const StringView::CodePoints::Iterator& end)
{
if (c == end)
@@ -261,12 +350,15 @@
m_buffer.resize(m_url.m_pathAfterLastSlash);
}
-URL URLParser::parse(const String& input, const URL& base, const TextEncoding&)
+URL URLParser::parse(const String& input, const URL& base, const TextEncoding& encoding)
{
LOG(URLParser, "Parsing URL <%s> base <%s>", input.utf8().data(), base.string().utf8().data());
m_url = { };
m_buffer.clear();
m_buffer.reserveCapacity(input.length());
+
+ // FIXME: We shouldn't need to allocate another buffer for this.
+ StringBuilder queryBuffer;
auto codePoints = StringView(input).codePoints();
auto c = codePoints.begin();
@@ -347,6 +439,8 @@
break;
}
++c;
+ while (c != end && isTabOrNewline(*c))
+ ++c;
if (c == end) {
m_buffer.clear();
state = State::NoScheme;
@@ -380,6 +474,7 @@
return { };
} else if (base.protocol() == "file") {
copyURLPartsUntil(base, URLPart::SchemeEnd);
+ m_buffer.append(':');
state = State::File;
} else
state = State::Relative;
@@ -479,12 +574,20 @@
if (*c == '@') {
parseAuthority(authorityOrHostBegin, c);
++c;
+ while (c != end && isTabOrNewline(*c))
+ ++c;
authorityOrHostBegin = c;
state = State::Host;
+ break;
} else if (*c == '/' || *c == '?' || *c == '#') {
m_url.m_userEnd = m_buffer.length();
m_url.m_passwordEnd = m_url.m_userEnd;
- parseHost(authorityOrHostBegin, c);
+ if (!parseHost(authorityOrHostBegin, c))
+ return { };
+ if (*c != '/') {
+ m_buffer.append('/');
+ m_url.m_pathAfterLastSlash = m_buffer.length();
+ }
state = State::Path;
break;
}
@@ -493,7 +596,8 @@
case State::Host:
LOG_STATE("Host");
if (*c == '/' || *c == '?' || *c == '#') {
- parseHost(authorityOrHostBegin, c);
+ if (!parseHost(authorityOrHostBegin, c))
+ return { };
state = State::Path;
break;
}
@@ -509,7 +613,7 @@
++c;
break;
case '?':
- if (!base.isNull() && base.protocol() == "file")
+ if (!base.isNull() && base.protocolIs("file"))
copyURLPartsUntil(base, URLPart::PathEnd);
m_buffer.append("///?");
m_url.m_userStart = m_buffer.length() - 2;
@@ -523,7 +627,7 @@
++c;
break;
case '#':
- if (!base.isNull() && base.protocol() == "file")
+ if (!base.isNull() && base.protocolIs("file"))
copyURLPartsUntil(base, URLPart::QueryEnd);
m_buffer.append("///#");
m_url.m_userStart = m_buffer.length() - 2;
@@ -538,10 +642,9 @@
++c;
break;
default:
- if (shouldCopyFileURL(c, end)) {
- copyURLPartsUntil(base, URLPart::PathEnd);
- popPath();
- } else {
+ if (!base.isNull() && base.protocolIs("file") && shouldCopyFileURL(c, end))
+ copyURLPartsUntil(base, URLPart::PathAfterLastSlash);
+ else {
m_buffer.append("///");
m_url.m_userStart = m_buffer.length() - 1;
m_url.m_userEnd = m_url.m_userStart;
@@ -667,8 +770,17 @@
state = State::Fragment;
break;
}
- // FIXME: Percent encode c
- m_buffer.append(*c);
+ if (isPercentEncodedDot(c, end)) {
+ m_buffer.append('.');
+ ASSERT(*c == '%');
+ ++c;
+ ASSERT(*c == dotASCIICode[0]);
+ ++c;
+ ASSERT(toASCIILower(*c) == dotASCIICode[1]);
+ ++c;
+ break;
+ }
+ utf8PercentEncode(*c, m_buffer, isInDefaultEncodeSet);
++c;
break;
case State::CannotBeABaseURLPath:
@@ -688,11 +800,12 @@
case State::Query:
LOG_STATE("Query");
if (*c == '#') {
+ encodeQuery(queryBuffer, m_buffer, encoding);
m_url.m_queryEnd = m_buffer.length();
state = State::Fragment;
break;
}
- m_buffer.append(*c);
+ queryBuffer.append(*c);
++c;
break;
case State::Fragment:
@@ -743,7 +856,8 @@
case State::Host:
if (state == State::Host)
LOG_FINAL_STATE("Host");
- parseHost(authorityOrHostBegin, end);
+ if (!parseHost(authorityOrHostBegin, end))
+ return { };
m_buffer.append('/');
m_url.m_pathEnd = m_url.m_portEnd + 1;
m_url.m_pathAfterLastSlash = m_url.m_pathEnd;
@@ -832,6 +946,7 @@
break;
case State::Query:
LOG_FINAL_STATE("Query");
+ encodeQuery(queryBuffer, m_buffer, encoding);
m_url.m_queryEnd = m_buffer.length();
m_url.m_fragmentEnd = m_url.m_queryEnd;
break;
@@ -1131,6 +1246,73 @@
return address;
}
+static String percentDecode(const String& input)
+{
+ StringBuilder output;
+ RELEASE_ASSERT(input.is8Bit());
+ const LChar* inputBytes = input.characters8();
+ size_t length = input.length();
+
+ for (size_t i = 0; i < length; ++i) {
+ uint8_t byte = inputBytes[i];
+ if (byte != '%')
+ output.append(byte);
+ else if (i < length - 2) {
+ if (isASCIIHexDigit(inputBytes[i + 1]) && isASCIIHexDigit(inputBytes[i + 2])) {
+ output.append(toASCIIHexValue(inputBytes[i + 1], inputBytes[i + 2]));
+ i += 2;
+ } else
+ output.append(byte);
+ } else
+ output.append(byte);
+ }
+ return output.toStringPreserveCapacity();
+}
+
+static Optional<String> domainToASCII(const String& domain)
+{
+ // FIXME: Implement correctly
+ CString utf8 = domain.utf8();
+ return String(utf8.data(), utf8.length());
+}
+
+static bool hasInvalidDomainCharacter(const String& asciiDomain)
+{
+ RELEASE_ASSERT(asciiDomain.is8Bit());
+ const LChar* characters = asciiDomain.characters8();
+ for (size_t i = 0; i < asciiDomain.length(); ++i) {
+ if (isInvalidDomainCharacter(characters[i]))
+ return true;
+ }
+ return false;
+}
+
+bool URLParser::parsePort(StringView::CodePoints::Iterator& iterator, const StringView::CodePoints::Iterator& end)
+{
+ uint32_t port = 0;
+ ASSERT(iterator != end);
+ for (; iterator != end; ++iterator) {
+ if (isTabOrNewline(*iterator))
+ continue;
+ if (isASCIIDigit(*iterator)) {
+ port = port * 10 + *iterator - '0';
+ if (port > std::numeric_limits<uint16_t>::max())
+ return false;
+ } else
+ return false;
+ }
+
+ // FIXME: This shouldn't need a String allocation.
+ String scheme = m_buffer.toStringPreserveCapacity().substring(0, m_url.m_schemeEnd);
+ if (isDefaultPort(scheme, port)) {
+ ASSERT(m_buffer[m_buffer.length() - 1] == ':');
+ m_buffer.resize(m_buffer.length() - 1);
+ } else
+ m_buffer.appendNumber(port);
+
+ return true;
+}
+
bool URLParser::parseHost(StringView::CodePoints::Iterator& iterator, const StringView::CodePoints::Iterator& end)
{
if (iterator == end)
@@ -1148,7 +1330,30 @@
return true;
}
}
- if (auto address = parseIPv4Host(iterator, end)) {
+
+ // FIXME: We probably don't need to make so many buffers and String copies.
+ StringBuilder utf8Encoded;
+ for (; iterator != end; ++iterator) {
+ if (isTabOrNewline(*iterator))
+ continue;
+ if (*iterator == ':')
+ break;
+ uint8_t buffer[U8_MAX_LENGTH];
+ int32_t offset = 0;
+ UBool error = false;
+ U8_APPEND(buffer, offset, U8_MAX_LENGTH, *iterator, error);
+ // FIXME: Check error.
+ utf8Encoded.append(buffer, offset);
+ }
+ String percentDecoded = percentDecode(utf8Encoded.toStringPreserveCapacity());
+ RELEASE_ASSERT(percentDecoded.is8Bit());
+ String domain = String::fromUTF8(percentDecoded.characters8(), percentDecoded.length());
+ auto asciiDomain = domainToASCII(domain);
+ if (!asciiDomain || hasInvalidDomainCharacter(asciiDomain.value()))
+ return false;
+
+ auto asciiDomainCodePoints = StringView(asciiDomain.value()).codePoints();
+ if (auto address = parseIPv4Host(asciiDomainCodePoints.begin(), asciiDomainCodePoints.end())) {
serializeIPv4(address.value(), m_buffer);
m_url.m_hostEnd = m_buffer.length();
// FIXME: Handle the port correctly.
@@ -1155,20 +1360,21 @@
m_url.m_portEnd = m_buffer.length();
return true;
}
- for (; iterator != end; ++iterator) {
- if (*iterator == ':') {
+
+ m_buffer.append(asciiDomain.value());
+ m_url.m_hostEnd = m_buffer.length();
+ if (iterator != end) {
+ ASSERT(*iterator == ':');
+ ++iterator;
+ while (iterator != end && isTabOrNewline(*iterator))
++iterator;
- m_url.m_hostEnd = m_buffer.length();
+ if (iterator != end) {
m_buffer.append(':');
- for (; iterator != end; ++iterator)
- m_buffer.append(*iterator);
- m_url.m_portEnd = m_buffer.length();
- return true;
+ if (!parsePort(iterator, end))
+ return false;
}
- m_buffer.append(*iterator);
}
- m_url.m_hostEnd = m_buffer.length();
- m_url.m_portEnd = m_url.m_hostEnd;
+ m_url.m_portEnd = m_buffer.length();
return true;
}
Modified: trunk/Tools/TestWebKitAPI/Tests/WebCore/URLParser.cpp (205492 => 205493)
--- trunk/Tools/TestWebKitAPI/Tests/WebCore/URLParser.cpp 2016-09-06 18:02:28 UTC (rev 205492)
+++ trunk/Tools/TestWebKitAPI/Tests/WebCore/URLParser.cpp 2016-09-06 18:16:07 UTC (rev 205493)
@@ -165,6 +165,9 @@
checkURL("file:///#fragment", {"file", "", "", "", 0, "/", "", "fragment", "file:///#fragment"});
checkURL("file:////?query", {"file", "", "", "", 0, "//", "query", "", "file:////?query"});
checkURL("file:////#fragment", {"file", "", "", "", 0, "//", "", "fragment", "file:////#fragment"});
+ checkURL("http://host/A b", {"http", "", "", "host", 0, "/A%20b", "", "", "http://host/A%20b"});
+ checkURL("http://host/a%20B", {"http", "", "", "host", 0, "/a%20B", "", "", "http://host/a%20B"});
+ checkURL("http://host?q=@ <>!#fragment", {"http", "", "", "host", 0, "/", "q=@%20%3C%3E!", "fragment", "http://host/?q=@%20%3C%3E!#fragment"});
}
static void checkRelativeURL(const String& urlString, const String& baseURLString, const ExpectedParts& parts)
@@ -204,6 +207,8 @@
checkRelativeURL("http://whatwg.org/index.html", "http://webkit.org/path1/path2/", {"http", "", "", "whatwg.org", 0, "/index.html", "", "", "http://whatwg.org/index.html"});
checkRelativeURL("index.html", "http://webkit.org/path1/path2/page.html?query#fragment", {"http", "", "", "webkit.org", 0, "/path1/path2/index.html", "", "", "http://webkit.org/path1/path2/index.html"});
checkRelativeURL("//whatwg.org/index.html", "https://www.webkit.org/path", {"https", "", "", "whatwg.org", 0, "/index.html", "", "", "https://whatwg.org/index.html"});
+ checkRelativeURL("http://example\t.\norg", "http://example.org/foo/bar", {"http", "", "", "example.org", 0, "/", "", "", "http://example.org/"});
+ checkRelativeURL("test", "file:///path1/path2", {"file", "", "", "", 0, "/path1/test", "", "", "file:///path1/test"});
}
static void checkURLDifferences(const String& urlString, const ExpectedParts& partsNew, const ExpectedParts& partsOld)
@@ -338,8 +343,74 @@
checkURLDifferences("file:path",
{"file", "", "", "", 0, "/path", "", "", "file:///path"},
{"file", "", "", "", 0, "path", "", "", "file://path"});
+
+ // FIXME: Fix and test incomplete percent encoded characters in the middle and end of the input string.
+ // FIXME: Fix and test percent encoded upper case characters in the host.
+ checkURLDifferences("http://host%73",
+ {"http", "", "", "hosts", 0, "/", "", "", "http://hosts/"},
+ {"http", "", "", "host%73", 0, "/", "", "", "http://host%73/"});
+
+ // URLParser matches Chrome and the spec, but not URL::parse or Firefox.
+ checkURLDifferences("http://host/path%2e.%2E",
+ {"http", "", "", "host", 0, "/path...", "", "", "http://host/path..."},
+ {"http", "", "", "host", 0, "/path%2e.%2E", "", "", "http://host/path%2e.%2E"});
}
+TEST_F(URLParserTest, DefaultPort)
+{
+ checkURL("ftp://host:21/", {"ftp", "", "", "host", 0, "/", "", "", "ftp://host/"});
+ checkURL("ftp://host:22/", {"ftp", "", "", "host", 22, "/", "", "", "ftp://host:22/"});
+ checkURLDifferences("ftp://host:21",
+ {"ftp", "", "", "host", 0, "/", "", "", "ftp://host/"},
+ {"ftp", "", "", "host", 0, "", "", "", "ftp://host"});
+ checkURLDifferences("ftp://host:22",
+ {"ftp", "", "", "host", 22, "/", "", "", "ftp://host:22/"},
+ {"ftp", "", "", "host", 22, "", "", "", "ftp://host:22"});
+
+ checkURL("gopher://host:70/", {"gopher", "", "", "host", 0, "/", "", "", "gopher://host/"});
+ checkURL("gopher://host:71/", {"gopher", "", "", "host", 71, "/", "", "", "gopher://host:71/"});
+ // Spec, Chrome, Firefox, and URLParser have "/", URL::parse does not.
+ // Spec, Chrome, URLParser, URL::parse recognize gopher default port, Firefox does not.
+ checkURLDifferences("gopher://host:70",
+ {"gopher", "", "", "host", 0, "/", "", "", "gopher://host/"},
+ {"gopher", "", "", "host", 0, "", "", "", "gopher://host"});
+ checkURLDifferences("gopher://host:71",
+ {"gopher", "", "", "host", 71, "/", "", "", "gopher://host:71/"},
+ {"gopher", "", "", "host", 71, "", "", "", "gopher://host:71"});
+
+ checkURL("http://host:80", {"http", "", "", "host", 0, "/", "", "", "http://host/"});
+ checkURL("http://host:80/", {"http", "", "", "host", 0, "/", "", "", "http://host/"});
+ checkURL("http://host:81", {"http", "", "", "host", 81, "/", "", "", "http://host:81/"});
+ checkURL("http://host:81/", {"http", "", "", "host", 81, "/", "", "", "http://host:81/"});
+
+ checkURL("https://host:443", {"https", "", "", "host", 0, "/", "", "", "https://host/"});
+ checkURL("https://host:443/", {"https", "", "", "host", 0, "/", "", "", "https://host/"});
+ checkURL("https://host:444", {"https", "", "", "host", 444, "/", "", "", "https://host:444/"});
+ checkURL("https://host:444/", {"https", "", "", "host", 444, "/", "", "", "https://host:444/"});
+
+ checkURL("ws://host:80/", {"ws", "", "", "host", 0, "/", "", "", "ws://host/"});
+ checkURL("ws://host:81/", {"ws", "", "", "host", 81, "/", "", "", "ws://host:81/"});
+ // URLParser matches Chrome and Firefox, but not URL::parse
+ checkURLDifferences("ws://host:80",
+ {"ws", "", "", "host", 0, "/", "", "", "ws://host/"},
+ {"ws", "", "", "host", 0, "", "", "", "ws://host"});
+ checkURLDifferences("ws://host:81",
+ {"ws", "", "", "host", 81, "/", "", "", "ws://host:81/"},
+ {"ws", "", "", "host", 81, "", "", "", "ws://host:81"});
+
+ checkURL("wss://host:443/", {"wss", "", "", "host", 0, "/", "", "", "wss://host/"});
+ checkURL("wss://host:444/", {"wss", "", "", "host", 444, "/", "", "", "wss://host:444/"});
+ // URLParser matches Chrome and Firefox, but not URL::parse
+ checkURLDifferences("wss://host:443",
+ {"wss", "", "", "host", 0, "/", "", "", "wss://host/"},
+ {"wss", "", "", "host", 0, "", "", "", "wss://host"});
+ checkURLDifferences("wss://host:444",
+ {"wss", "", "", "host", 444, "/", "", "", "wss://host:444/"},
+ {"wss", "", "", "host", 444, "", "", "", "wss://host:444"});
+
+ // FIXME: Fix and check unknown schemes with ports, as well as ftps.
+}
+
static void shouldFail(const String& urlString)
{
URLParser parser;
