Add fast-mode Uri class.

sdk/lib/core/uri.dart

Index: sdk/lib/core/uri.dart
diff --git a/sdk/lib/core/uri.dart b/sdk/lib/core/uri.dart
index 5718ca80b83b870dd578011592707b4c7d99cdc4..a89f8d4b445abc4870142572e78b459e8e293a91 100644
--- a/sdk/lib/core/uri.dart
+++ b/sdk/lib/core/uri.dart
@@ -73,6 +73,11 @@ class Uri {
   List<String> _pathSegments;
 
   /**
+   * Cache of the full normalized text representation of the URI.
+   */
+  String _text;
+
+  /**
    * Cache the computed return value of [queryParameters].
    */
   Map<String, String> _queryParameters;
@@ -158,8 +163,8 @@ class Uri {
    * general delimiters, are escaped if necessary.
    * If `fragment` is omitted or `null`, the URI has no fragment part.
    */
-  factory Uri({String scheme : "",
-               String userInfo : "",
+  factory Uri({String scheme,
+               String userInfo,
                String host,
                int port,
                String path,
@@ -394,199 +399,192 @@ class Uri {
     // query         = *( pchar / "/" / "?" )
     //
     // fragment      = *( pchar / "/" / "?" )
-    const int EOI = -1;
+    end ??= uri.length;
 
-    String scheme = "";
-    String userinfo = "";
-    String host = null;
-    int port = null;
-    String path = null;
-    String query = null;
-    String fragment = null;
-    if (end == null) end = uri.length;
-
-    int index = start;
-    int pathStart = start;
-    // End of input-marker.
-    int char = EOI;
-
-    void parseAuth() {
-      if (index == end) {
-        char = EOI;
-        return;
-      }
-      int authStart = index;
-      int lastColon = -1;
-      int lastAt = -1;
-      char = uri.codeUnitAt(index);
-      while (index < end) {
-        char = uri.codeUnitAt(index);
-        if (char == _SLASH || char == _QUESTION || char == _NUMBER_SIGN) {
-          break;
-        }
-        if (char == _AT_SIGN) {
-          lastAt = index;
-          lastColon = -1;
-        } else if (char == _COLON) {
-          lastColon = index;
-        } else if (char == _LEFT_BRACKET) {
-          lastColon = -1;
-          int endBracket = uri.indexOf(']', index + 1);
-          if (endBracket == -1) {
-            index = end;
-            char = EOI;
-            break;
-          } else {
-            index = endBracket;
-          }
-        }
-        index++;
-        char = EOI;
-      }
-      int hostStart = authStart;
-      int hostEnd = index;
-      if (lastAt >= 0) {
-        userinfo = _makeUserInfo(uri, authStart, lastAt);
-        hostStart = lastAt + 1;
-      }
-      if (lastColon >= 0) {
-        int portNumber;
-        if (lastColon + 1 < index) {
-          portNumber = 0;
-          for (int i = lastColon + 1; i < index; i++) {
-            int digit = uri.codeUnitAt(i);
-            if (_ZERO > digit || _NINE < digit) {
-              _fail(uri, i, "Invalid port number");
-            }
-            portNumber = portNumber * 10 + (digit - _ZERO);
-          }
-        }
-        port = _makePort(portNumber, scheme);
-        hostEnd = lastColon;
-      }
-      host = _makeHost(uri, hostStart, hostEnd, true);
-      if (index < end) {
-        char = uri.codeUnitAt(index);
+    // Special case data:URIs. Ignore case when testing.
+    if (end >= start + 5 &&
+        uri.codeUnitAt(start + 4) == _COLON &&
+        uri.codeUnitAt(start) | 0x20 == 0x64 /*d*/ &&
+        uri.codeUnitAt(start + 1) | 0x20 == 0x61 /*a*/ &&
+        uri.codeUnitAt(start + 2) | 0x20 == 0x74 /*t*/ &&
+        uri.codeUnitAt(start + 3) | 0x20 == 0x61 /*a*/) {
+      // Data-URI.
+      if (uri.startsWith("data", start)) {
+        // The case is right.
+        if (start > 0 || end < uri.length) uri = uri.substring(start, end);
+        return UriData._parse(uri, 5, null).uri;
       }
+      return Uriata._parse(uri.substring(start + 5, end), 0, null).uri;

[floitsch, 2016/06/28 00:28:03]

UriData

I'm guessing you didn't run the tests?

[Lasse Reichstein Nielsen, 2016/06/28 13:00:51]

I did. Turns out we never tested a data URI starting with a non-lower-case
"data" - and that we didn't accept it. Fixed now.

     }
 
-    // When reaching path parsing, the current character is known to not
-    // be part of the path.
-    const int NOT_IN_PATH = 0;
-    // When reaching path parsing, the current character is part
-    // of the a non-empty path.
-    const int IN_PATH = 1;
-    // When reaching authority parsing, authority is possible.
-    // This is only true at start or right after scheme.
-    const int ALLOW_AUTH = 2;
+    var indices = _scanUri(uri, start, end);

[floitsch, 2016/06/28 00:28:03]

Provide example of what the indices could look like.
(probably also as dartdoc on the '_scanUri' function.

[Lasse Reichstein Nielsen, 2016/06/28 13:00:52]

Done.

 
-    // Current state.
-    // Initialized to the default value that is used when exiting the
-    // scheme loop by reaching the end of input.
-    // All other breaks set their own state.
-    int state = NOT_IN_PATH;
-    int i = index;  // Temporary alias for index to avoid bug 19550 in dart2js.
-    while (i < end) {
-      char = uri.codeUnitAt(i);
-      if (char == _QUESTION || char == _NUMBER_SIGN) {
-        state = NOT_IN_PATH;
-        break;
-      }
-      if (char == _SLASH) {
-        state = (i == start) ? ALLOW_AUTH : IN_PATH;
-        break;
+    int schemeEnd     = indices[_schemeEndIndex];      // >0 if has scheme

[floitsch, 2016/06/28 00:28:04]

Finish with ".".

[Lasse Reichstein Nielsen, 2016/06/28 13:00:52]

Just removed, it's documented in the _scanUri docs.
(And it was really ">start").

+    int hostStart     = indices[_hostStartIndex] + 1;  // >0 if has authority.
+    int portStart     = indices[_portStartIndex];
+    int pathStart     = indices[_pathStartIndex];
+    int queryStart    = indices[_queryStartIndex];
+    int fragmentStart = indices[_fragmentStartIndex];
+
+    // We may discover scheme while handling special cases.
+    String scheme;
+
+    // Derive some indices that weren't set.
+    // If fragment but no query, set query to start at fragment.
+    if (fragmentStart < queryStart) queryStart = fragmentStart;

[floitsch, 2016/06/28 00:28:03]

It's not clear why this wouldn't be done by the _scanUri function.
Same for the other derived indices.

[Lasse Reichstein Nielsen, 2016/06/28 13:00:52]

It absolutely could. More to the point, the _scanUri function should really be
inlined here - it's the only place it's used, and the separation isn't logical
(as you say, this feels like it belongs closer to the scanning).

I think I'll keep it as-is for now, but add a "TODO(lrn): consider inlining
_scanUri"

+    // If scheme but no authority, the pathStart isn't set.
+    if (schemeEnd >= start && hostStart <= start) pathStart = schemeEnd + 1;
+    // If scheme or authority but pathStart isn't set.
+    if (pathStart == start && (schemeEnd >= start || hostStart > start)) {
+      pathStart = queryStart;
+    }
+    // If authority and no port.
+    // (including when user-info contains : and portStart >= 0).
+    if (portStart < hostStart) portStart = pathStart;
+
+    bool isSimple = indices[_notSimpleIndex] < start;
+    if (isSimple) {
+      // Check/do normalizations that weren't detected by the scanner.
+      // This includes removal of empty port or userInfo,
+      // or scheme specific port and path normalizations.
+      if (hostStart > schemeEnd + 3) {
+        // Always be non-simple if URI contains user-info.
+        isSimple = false;
       }

[floitsch, 2016/06/28 00:28:03]

Is it necessary to enter the 'else' below?
if not -> "else if".
Otherwise add comment or move this check into the else below.

[Lasse Reichstein Nielsen, 2016/06/28 13:00:52]

Done.

-      if (char == _COLON) {
-        if (i == start) _fail(uri, start, "Invalid empty scheme");
-        scheme = _makeScheme(uri, start, i);
-        i++;
-        if (scheme == "data") {
-          // This generates a URI that is (potentially) not path normalized.
-          // Applying part normalization to a non-hierarchial URI isn't
-          // meaningful.
-          return UriData._parse(uri, i, null).uri;
-        }
-        pathStart = i;
-        if (i == end) {
-          char = EOI;
-          state = NOT_IN_PATH;
-        } else {
-          char = uri.codeUnitAt(i);
-          if (char == _QUESTION || char == _NUMBER_SIGN) {
-            state = NOT_IN_PATH;
-          } else if (char == _SLASH) {
-            state = ALLOW_AUTH;
-          } else {
-            state = IN_PATH;
+      if (portStart > start && portStart + 1 == pathStart) {
+        // If the port is empty, it should be omitted.
+        // Path case, don't bother correcting it.
+        isSimple = false;
+      } else if (hostStart == schemeEnd + 4) {
+        // If the userInfo is empty, it should be omitted.
+        // (4 is length of "://@").
+        // Pathological case, don't bother correcting it.
+        isSimple = false;
+      } else {
+        // There are a few scheme-based normalizations that
+        // the scanner couldn't check.
+        // That means that the input is very close to simple, so just do
+        // the normalizations.
+        if (schemeEnd == start + 4) {
+          // Do scheme based normalizations for file, http.
+          if (uri.startsWith("file", start)) {
+            scheme = "file";
+            if (hostStart <= 0) {
+              // File URIs should have an authority.
+              // Paths after an authority should be absolute.
+              String insert = "//";
+              int delta = 2;
+              if (!uri.startsWith("/", pathStart)) {
+                insert = "///";
+                delta = 3;
+              }
+              uri = "${uri.substring(start, schemeEnd + 1)}$insert"
+                    "${uri.substring(pathStart, end)}";
+              schemeEnd -= start;
+              pathStart += 2 - start;
+              hostStart = schemeEnd + 3;
+              portStart = pathStart;
+              queryStart += delta;
+              fragmentStart += delta;
+              start = 0;
+              end = uri.length;
+            } else if (pathStart == queryStart) {
+              uri = "${uri.substring(start, pathStart)}/"
+                    "${uri.substring(pathStart, end)}";
+              queryStart += 1;
+              fragmentStart += 1;
+            }
+          } else if (uri.startsWith("http", start)) {
+            scheme = "http";
+            // Http URIs should not have an explicit port of 80

[floitsch, 2016/06/28 00:28:03]

Finish with ".".

[Lasse Reichstein Nielsen, 2016/06/28 13:00:51]

Done.

+            if (portStart > start && portStart + 3 == pathStart &&
+                uri.startsWith("80", portStart + 1)) {
+              uri = uri.substring(start, portStart) +
+                    uri.substring(pathStart, end);
+              schemeEnd -= start;
+              hostStart -= start;
+              portStart -= start;
+              pathStart -= 3 + start;
+              queryStart -= 3 + start;
+              fragmentStart -= 3 + start;
+              start = 0;
+              end = uri.length;
+            }
+          }
+        } else if (schemeEnd == start + 5 && uri.startsWith("https", start)) {
+          scheme = "https";
+          // Https URIs should not have an explicit port of 443
+          if (portStart > start && portStart + 4 == pathStart &&
+              uri.startsWith("443", portStart + 1)) {
+            uri = uri.substring(start, portStart) +
+                  uri.substring(pathStart, end);
+            schemeEnd -= start;
+            hostStart -= start;
+            portStart -= start;
+            pathStart -= 4 + start;
+            queryStart -= 4 + start;
+            fragmentStart -= 4 + start;
+            start = 0;
+            end = uri.length;
           }
         }
-        break;
       }
-      i++;
-      char = EOI;
     }
-    index = i;  // Remove alias when bug is fixed.
 
-    if (state == ALLOW_AUTH) {
-      assert(char == _SLASH);
-      // Have seen one slash either at start or right after scheme.
-      // If two slashes, it's an authority, otherwise it's just the path.
-      index++;
-      if (index == end) {
-        char = EOI;
-        state = NOT_IN_PATH;
-      } else {
-        char = uri.codeUnitAt(index);
-        if (char == _SLASH) {
-          index++;
-          parseAuth();
-          pathStart = index;
-        }
-        if (char == _QUESTION || char == _NUMBER_SIGN || char == EOI) {
-          state = NOT_IN_PATH;
-        } else {
-          state = IN_PATH;
+    if (isSimple) {
+      if (start > 0 || end < uri.length) {
+        uri = uri.substring(start, end);
+        if (schemeEnd >= 0) schemeEnd -= start;
+        if (hostStart > 0) {
+          hostStart -= start;
+          portStart -= start;
         }
+        pathStart -= start;
+        queryStart -= start;
+        fragmentStart -= start;
       }
+      return new _SimpleUri(uri, schemeEnd, hostStart, portStart, pathStart,
+                            queryStart, fragmentStart, scheme);
+
     }
 
-    assert(state == IN_PATH || state == NOT_IN_PATH);
-    if (state == IN_PATH) {
-      // Characters from pathStart to index (inclusive) are known
-      // to be part of the path.
-      while (++index < end) {
-        char = uri.codeUnitAt(index);
-        if (char == _QUESTION || char == _NUMBER_SIGN) {
-          break;
-        }
-        char = EOI;
+    if (scheme == null) {
+      scheme = "";
+      if (schemeEnd > start) {
+        scheme = _makeScheme(uri, start, schemeEnd);
+      } else if (schemeEnd == start) {
+        _fail(uri, start, "Invalid empty scheme");
       }
-      state = NOT_IN_PATH;
     }
-
-    assert(state == NOT_IN_PATH);
-    bool hasAuthority = (host != null);
-    path = _makePath(uri, pathStart, index, null, scheme, hasAuthority);
-
-    if (char == _QUESTION) {
-      int numberSignIndex = -1;
-      for (int i = index + 1; i < end; i++) {
-        if (uri.codeUnitAt(i) == _NUMBER_SIGN) {
-          numberSignIndex = i;
-          break;
-        }
+    String userInfo = "";
+    String host;
+    int port;
+    if (hostStart > start) {
+      int userInfoStart = schemeEnd + 3;
+      if (userInfoStart < hostStart) {
+        userInfo = _makeUserInfo(uri, userInfoStart, hostStart - 1);
       }
-      if (numberSignIndex < 0) {
-        query = _makeQuery(uri, index + 1, end, null);
-      } else {
-        query = _makeQuery(uri, index + 1, numberSignIndex, null);
-        fragment = _makeFragment(uri, numberSignIndex + 1, end);
+      host = _makeHost(uri, hostStart, portStart, false);
+      if (portStart + 1 < pathStart) {
+        // Should throw because invalid.
+        port = int.parse(uri.substring(portStart + 1, pathStart), onError: (_) {
+          throw new FormatException("Invalid port", uri, portStart + 1);
+        });
+        port = _makePort(port, scheme);
       }
-    } else if (char == _NUMBER_SIGN) {
-      fragment = _makeFragment(uri, index + 1, end);
+    }
+    String path = _makePath(uri, pathStart, queryStart, null,
+                            scheme, host != null);
+    String query;
+    if (queryStart < fragmentStart) {
+      query = _makeQuery(uri, queryStart + 1, fragmentStart, null);
+    }
+    String fragment;
+    if (fragmentStart < end) {
+      fragment = _makeFragment(uri, fragmentStart + 1, end);
     }
     return new Uri._internal(scheme,
-                             userinfo,
+                             userInfo,
                              host,
                              port,
                              path,
@@ -1333,6 +1331,17 @@ class Uri {
     }
     scheme = scheme.substring(start, end);
     if (containsUpperCase) scheme = scheme.toLowerCase();
+    return _canonicalizeScheme(scheme);
+  }
+
+  // Canonicalize a few often-used scheme strings.
+  //
+  // This improves memory usage and makes comparison faster.
+  static String _canonicalizeScheme(String scheme) {
+    if (scheme == "http") return "http";
+    if (scheme == "file") return "file";
+    if (scheme == "https") return "https";
+    if (scheme == "package") return "package";
     return scheme;
   }
 
@@ -1932,6 +1941,8 @@ class Uri {
    * If the URI cannot be converted to a file path calling this throws
    * [UnsupportedError].
    */
+  // TODO(lrn): Deprecate and move functionality to File class or similar.
+  // The core libraries should not worry about the platform.
   String toFilePath({bool windows}) {
     if (scheme != "" && scheme != "file") {
       throw new UnsupportedError(
@@ -1946,25 +1957,27 @@ class Uri {
           "Cannot extract a file path from a URI with a fragment component");
     }
     if (windows == null) windows = _isWindows;
-    return windows ? _toWindowsFilePath() : _toFilePath();
+    return windows ? _toWindowsFilePath(this) : _toFilePath();
   }
 
   String _toFilePath() {
-    if (host != "") {
+    if (hasAuthority && host != "") {
       throw new UnsupportedError(
           "Cannot extract a non-Windows file path from a file URI "
           "with an authority");
     }
+    // Use path segments to have any escapes unescaped.
+    var pathSegments = this.pathSegments;
     _checkNonWindowsPathReservedCharacters(pathSegments, false);
     var result = new StringBuffer();
-    if (_isPathAbsolute) result.write("/");
+    if (hasAbsolutePath) result.write("/");
     result.writeAll(pathSegments, "/");
     return result.toString();
   }
 
-  String _toWindowsFilePath() {
+  static String _toWindowsFilePath(Uri uri) {
     bool hasDriveLetter = false;
-    var segments = pathSegments;
+    var segments = uri.pathSegments;
     if (segments.length > 0 &&
         segments[0].length == 2 &&
         segments[0].codeUnitAt(1) == _COLON) {
@@ -1972,23 +1985,25 @@ class Uri {
       _checkWindowsPathReservedCharacters(segments, false, 1);
       hasDriveLetter = true;
     } else {
-      _checkWindowsPathReservedCharacters(segments, false);
+      _checkWindowsPathReservedCharacters(segments, false, 0);
     }
     var result = new StringBuffer();
-    if (_isPathAbsolute && !hasDriveLetter) result.write("\\");
-    if (host != "") {
-      result.write("\\");
-      result.write(host);
-      result.write("\\");
+    if (uri.hasAbsolutePath && !hasDriveLetter) result.write(r"\");
+    if (uri.hasAuthority) {
+      var host = uri.host;
+      if (host.isNotEmpty) {
+        result.write(r"\");
+        result.write(host);
+        result.write(r"\");
+      }
     }
-    result.writeAll(segments, "\\");
-    if (hasDriveLetter && segments.length == 1) result.write("\\");
+    result.writeAll(segments, r"\");
+    if (hasDriveLetter && segments.length == 1) result.write(r"\");
     return result.toString();
   }
 
   bool get _isPathAbsolute {
-    if (path == null || path.isEmpty) return false;
-    return path.startsWith('/');
+    return _path != null && _path.startsWith('/');
   }
 
   void _writeAuthority(StringSink ss) {
@@ -2014,8 +2029,9 @@ class Uri {
   UriData get data => (scheme == "data") ? new UriData.fromUri(this) : null;
 
   String toString() {
+    if (_text != null) return _text;
     StringBuffer sb = new StringBuffer();
-    _addIfNonEmpty(sb, scheme, scheme, ':');
+    if (scheme.isNotEmpty) sb..write(scheme)..write(":");
     if (hasAuthority || path.startsWith("//") || (scheme == "file")) {
       // File URIS always have the authority, even if it is empty.
       // The empty URI means "localhost".
@@ -2023,9 +2039,10 @@ class Uri {
       _writeAuthority(sb);
     }
     sb.write(path);
-    if (_query != null) { sb..write("?")..write(_query); }
-    if (_fragment != null) { sb..write("#")..write(_fragment); }
-    return sb.toString();
+    if (_query != null) sb..write("?")..write(_query);
+    if (_fragment != null) sb..write("#")..write(_fragment);
+    _text = sb.toString();
+    return _text;
   }
 
   bool operator==(other) {
@@ -2044,20 +2061,7 @@ class Uri {
   }
 
   int get hashCode {
-    int combine(part, current) {
-      // The sum is truncated to 30 bits to make sure it fits into a Smi.
-      return (current * 31 + part.hashCode) & 0x3FFFFFFF;
-    }
-    return combine(scheme, combine(userInfo, combine(host, combine(port,
-        combine(path, combine(query, combine(fragment, 1)))))));
-  }
-
-  static void _addIfNonEmpty(StringBuffer sb, String test,
-                             String first, String second) {
-    if ("" != test) {
-      sb.write(first);
-      sb.write(second);
-    }
+    return (_text ?? toString()).hashCode;
   }
 
   /**
@@ -3359,3 +3363,713 @@ class UriData {
   // This is the same characters as in a URI query (which is URI pchar plus '?')
   static const _uricTable = Uri._queryCharTable;
 }
+
+// --------------------------------------------------------------------
+// Constants used to read the scanner result.
+
+const int _nopIndex           = 0;
+const int _schemeEndIndex     = 1;
+const int _hostStartIndex     = 2;
+const int _portStartIndex     = 3;
+const int _pathStartIndex     = 4;
+const int _queryStartIndex    = 5;
+const int _fragmentStartIndex = 6;
+const int _notSimpleIndex     = 7;
+
+// Initial state for scanner.
+const int _start              = 00;
+
+final _scannerTables = _createTables();
+
+// ----------------------------------------------------------------------
+// Code to create the URI scanner table.
+
+const int _schemeOrPath       = 01;
+const int _authOrPath         = 02;
+const int _authOrPathSlash    = 03;
+const int _uinfoOrHost0       = 04;
+const int _uinfoOrHost        = 05;
+const int _uinfoOrPort0       = 06;
+const int _uinfoOrPort        = 07;
+const int _ipv6Host           = 08;
+const int _relPathSeg         = 09;
+const int _pathSeg            = 10;
+const int _path               = 11;
+const int _query              = 12;
+const int _fragment           = 13;
+const int _schemeOrPathDot    = 14;
+const int _schemeOrPathDot2   = 15;
+const int _relPathSegDot      = 16;
+const int _relPathSegDot2     = 17;
+const int _pathSegDot         = 18;
+const int _pathSegDot2        = 19;
+
+const int _scheme0            = 20;
+const int _scheme             = 21;
+
+const int _stateCount         = 22;
+const int _nonSimpleEndStates = 14;

[floitsch, 2016/06/28 00:28:04]

This needs a comment and should be separated from the other constants.

[Lasse Reichstein Nielsen, 2016/06/28 13:00:52]

Commented and moved. Most of the states have been moved into the _createTable
function.

+
+const int _nop                = _nopIndex           << 5;
+const int _schemeEnd          = _schemeEndIndex     << 5;
+const int _hostStart          = _hostStartIndex     << 5;
+const int _portStart          = _portStartIndex     << 5;
+const int _pathStart          = _pathStartIndex     << 5;
+const int _queryStart         = _queryStartIndex    << 5;
+const int _fragmentStart      = _fragmentStartIndex << 5;
+const int _notSimple          = _notSimpleIndex     << 5;
+
+
+void _setChars(Uint8List target, String chars, int value) {

[floitsch, 2016/06/28 00:28:03]

Needs dartdocs.

[Lasse Reichstein Nielsen, 2016/06/28 13:00:51]

Done.

+  for (int i = 0; i < chars.length; i++) {
+    var char = chars.codeUnitAt(i);
+    target[char ^ 0x60] = value;

[floitsch, 2016/06/28 00:28:04]

Needs explanation what the ^ 0x60 does.

[Lasse Reichstein Nielsen, 2016/06/28 13:00:52]

Now described on the _scannerTables declaration.

+  }
+}
+
+void _setRange(Uint8List target, String range, int value) {

[floitsch, 2016/06/28 00:28:03]

Needs dartdocs.

[Lasse Reichstein Nielsen, 2016/06/28 13:00:51]

Done.

+  for (int i = range.codeUnitAt(0), n = range.codeUnitAt(1); i <= n; i++) {
+    target[i ^ 0x60] = value;
+  }
+}
+
+List<Uint8List> _createTables() {

[floitsch, 2016/06/28 00:28:04]

Needs documentation.

[Lasse Reichstein Nielsen, 2016/06/28 13:00:52]

Done.

+  // TODO(lrn): Use a precomputed table.
+  var tables = new List<Uint8List>.generate(_stateCount,
+      (_) => new Uint8List(96));
+
+  const unreserved =
+      "0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz-._~"  ;
+  const subDelims = r"!$&'()*+,;=";
+  const pchar = "$unreserved$subDelims";

[floitsch, 2016/06/28 00:28:03]

what does the "p" stand for?

[Lasse Reichstein Nielsen, 2016/06/28 13:00:52]

This refers to the "pchar" character group in RFC 3986. Added documentation.

+
+  build(index, defaultValue) =>
+      tables[index]..fillRange(0, 96, defaultValue);
+
+  var b;
+  // Validate as path, if it is a scheme, we handle it later.
+  b = build(_start, _schemeOrPath | _notSimple);
+  _setChars(b, pchar, _schemeOrPath);
+  _setChars(b, ".", _schemeOrPathDot);
+  _setChars(b, "%", _schemeOrPath | _notSimple);
+  _setChars(b, ":", _authOrPath | _schemeEnd);  // Handle later.
+  _setChars(b, "/", _authOrPathSlash);
+  _setChars(b, "?", _query | _queryStart);
+  _setChars(b, "#", _fragment | _fragmentStart);
+
+  b = build(_schemeOrPathDot, _schemeOrPath | _notSimple);
+  _setChars(b, pchar, _schemeOrPath);
+  _setChars(b, ".", _schemeOrPathDot2);
+  _setChars(b, "%", _schemeOrPath | _notSimple);
+  _setChars(b, ':', _authOrPath | _schemeEnd);
+  _setChars(b, "/", _pathSeg | _notSimple);
+  _setChars(b, "?", _query | _queryStart);
+  _setChars(b, "#", _fragment | _fragmentStart);
+
+  b = build(_schemeOrPathDot2, _schemeOrPath | _notSimple);
+  _setChars(b, pchar, _schemeOrPath);
+  _setChars(b, "%", _schemeOrPath | _notSimple);
+  _setChars(b, ':', _authOrPath | _schemeEnd);
+  _setChars(b, "/", _relPathSeg);
+  _setChars(b, "?", _query | _queryStart);
+  _setChars(b, "#", _fragment | _fragmentStart);
+
+  b = build(_schemeOrPath, _schemeOrPath | _notSimple);
+  _setChars(b, pchar, _schemeOrPath);
+  _setChars(b, ':', _authOrPath | _schemeEnd);
+  _setChars(b, "/", _pathSeg);
+  _setChars(b, "?", _query | _queryStart);
+  _setChars(b, "#", _fragment | _fragmentStart);
+
+  b = build(_authOrPath, _path | _notSimple);
+  _setChars(b, pchar, _path);
+  _setChars(b, "/", _authOrPathSlash);
+  _setChars(b, ".", _pathSegDot);
+  _setChars(b, "?", _query | _queryStart);
+  _setChars(b, "#", _fragment | _fragmentStart);
+
+  b = build(_authOrPathSlash, _path | _notSimple);
+  _setChars(b, pchar, _path);
+  _setChars(b, "/", _uinfoOrHost0 | _hostStart);
+  _setChars(b, ".", _pathSegDot);
+  _setChars(b, "?", _query | _queryStart);
+  _setChars(b, "#", _fragment | _fragmentStart);
+
+  b = build(_uinfoOrHost0, _uinfoOrHost | _notSimple);
+  _setChars(b, pchar, _uinfoOrHost);
+  _setRange(b, "AZ", _uinfoOrHost | _notSimple);
+  _setChars(b, ":", _uinfoOrPort0 | _portStart);
+  _setChars(b, "@", _uinfoOrHost0 | _hostStart);
+  _setChars(b, "[", _ipv6Host | _notSimple);
+  _setChars(b, "/", _pathSeg | _pathStart);
+  _setChars(b, "?", _query | _queryStart);
+  _setChars(b, "#", _fragment | _fragmentStart);
+
+  b = build(_uinfoOrHost, _uinfoOrHost | _notSimple);
+  _setChars(b, pchar, _uinfoOrHost);
+  _setRange(b, "AZ", _uinfoOrHost | _notSimple);
+  _setChars(b, ":", _uinfoOrPort0 | _portStart);
+  _setChars(b, "@", _uinfoOrHost0 | _hostStart);
+  _setChars(b, "/", _pathSeg | _pathStart);
+  _setChars(b, "?", _query | _queryStart);
+  _setChars(b, "#", _fragment | _fragmentStart);
+
+  b = build(_uinfoOrPort0, _uinfoOrPort | _notSimple);
+  _setChars(b, "0", _uinfoOrPort | _notSimple);
+  _setRange(b, "19", _uinfoOrPort);
+  _setChars(b, "@", _uinfoOrHost0 | _hostStart);
+  _setChars(b, "/", _pathSeg | _pathStart);
+  _setChars(b, "?", _query | _queryStart);
+  _setChars(b, "#", _fragment | _fragmentStart);
+
+  b = build(_uinfoOrPort, _uinfoOrPort | _notSimple);
+  _setRange(b, "09", _uinfoOrPort);
+  _setChars(b, "@", _uinfoOrHost0 | _hostStart);
+  _setChars(b, "/", _pathSeg | _pathStart);
+  _setChars(b, "?", _query | _queryStart);
+  _setChars(b, "#", _fragment | _fragmentStart);
+
+  b = build(_ipv6Host, _ipv6Host);
+  _setChars(b, "]", _uinfoOrHost);
+
+  b = build(_relPathSeg, _path | _notSimple);
+  _setChars(b, pchar, _path);
+  _setChars(b, ".", _relPathSegDot);
+  _setChars(b, "/", _pathSeg | _notSimple);
+  _setChars(b, "?", _query | _queryStart);
+  _setChars(b, "#", _fragment | _fragmentStart);
+
+  b = build(_relPathSegDot, _path | _notSimple);
+  _setChars(b, pchar, _path);
+  _setChars(b, ".", _relPathSegDot2);
+  _setChars(b, "/", _pathSeg | _notSimple);
+  _setChars(b, "?", _query | _queryStart);
+  _setChars(b, "#", _fragment | _fragmentStart);
+
+  b = build(_relPathSegDot2, _path | _notSimple);
+  _setChars(b, pchar, _path);
+  _setChars(b, ".", _path);
+  _setChars(b, "/", _relPathSeg);
+  _setChars(b, "?", _query | _queryStart);
+  _setChars(b, "#", _fragment | _fragmentStart);
+
+  b = build(_pathSeg, _path | _notSimple);
+  _setChars(b, pchar, _path);
+  _setChars(b, ".", _pathSegDot);
+  _setChars(b, "/", _pathSeg | _notSimple);
+  _setChars(b, "?", _query | _queryStart);
+  _setChars(b, "#", _fragment | _fragmentStart);
+
+  b = build(_pathSegDot, _path | _notSimple);
+  _setChars(b, pchar, _path);
+  _setChars(b, ".", _pathSegDot2);
+  _setChars(b, "/", _pathSeg | _notSimple);
+  _setChars(b, "?", _query | _queryStart);
+  _setChars(b, "#", _fragment | _fragmentStart);
+
+  b = build(_pathSegDot2, _path | _notSimple);
+  _setChars(b, pchar, _path);
+  _setChars(b, ".", _path);
+  _setChars(b, "/", _pathSeg | _notSimple);
+  _setChars(b, "?", _query | _queryStart);
+  _setChars(b, "#", _fragment | _fragmentStart);
+
+  b = build(_path, _path | _notSimple);
+  _setChars(b, pchar, _path);
+  _setChars(b, "/", _pathSeg);
+  _setChars(b, "?", _query | _queryStart);
+  _setChars(b, "#", _fragment | _fragmentStart);
+
+  b = build(_query, _query | _notSimple);
+  _setChars(b, pchar, _query);
+  _setChars(b, "?", _query);
+  _setChars(b, "#", _fragment | _fragmentStart);
+
+  b = build(_fragment, _fragment | _notSimple);
+  _setChars(b, pchar, _fragment);
+  _setChars(b, "?", _fragment);
+
+  // A separate two-state validator for scheme names.
+  // Only accepts lower-case letters.
+  b = build(_scheme0, _scheme | _notSimple);
+  _setRange(b, "az", _scheme);
+
+  b = build(_scheme, _scheme | _notSimple);
+  _setRange(b, "az", _scheme);
+  _setRange(b, "09", _scheme);
+  _setChars(b, "+-.", _scheme);
+
+  return tables;
+}
+
+// --------------------------------------------------------------------
+// Code that uses the URI scanner table.
+
+int _scan(String uri, int start, int end, int state, List<int> indices) {
+  var tables = _scannerTables;
+  assert(end <= uri.length);
+  for (int i = start; i < end; i++) {
+    var table = tables[state];
+    // xor 0x60 to move range 0x20-0x7f into 0x00-0x5f
+    int char = uri.codeUnitAt(i) ^ 0x60;
+    // use 0x1f (nee 0x7f) to represent all unhandled characters.

[floitsch, 2016/06/28 00:28:03]

Use

[Lasse Reichstein Nielsen, 2016/06/28 13:00:52]

Done.

+    if (char > 0x5f) char = 0x1f;  // TODO: check if negating test is better.

[floitsch, 2016/06/28 00:28:03]

TODO(ldap)

[Lasse Reichstein Nielsen, 2016/06/28 13:00:52]

Or check it. It isn't faster.

+    int transition = table[char];
+    state = transition & 0x1f;
+    indices[transition >> 5] = i;
+  }
+  return state;
+}
+
+List<int> _scanUri(String uri, int start, int end) {
+  var indices = new List<int>.filled(8, start - 1);
+  indices[_portStartIndex] = start;  // equals effective pathStart if no auth.

[floitsch, 2016/06/28 00:28:04]

"Equals"

[Lasse Reichstein Nielsen, 2016/06/28 13:00:51]

Done.

+  indices[_pathStartIndex] = start;
+  indices[_queryStartIndex] = end;
+  indices[_fragmentStartIndex] = end;
+  var state = _scan(uri, start, end, _start, indices);
+  // Some states that should be non-simple, but the URI ended early.
+  if (state >= _nonSimpleEndStates) {
+    indices[_notSimpleIndex] = end;
+  }
+  int schemeEnd = indices[_schemeEndIndex];
+  if (schemeEnd >= start) {
+    // Rescan the scheme part now that we know it's not a path.
+    state = _scan(uri, start, schemeEnd, _scheme0, indices);
+    if (state == _scheme0) {
+      // Empty scheme.
+      indices[_notSimpleIndex] = schemeEnd;
+    }
+  }
+  return indices;
+}
+
+bool _isUpperCase(int char) {

[floitsch, 2016/06/28 00:28:04]

unused?

[Lasse Reichstein Nielsen, 2016/06/28 13:00:52]

Indeed. Gone.

+  return 0x41 <= char && char <= 0x5b;
+}
+
+class _SimpleUri implements Uri {
+  final String _uri;
+  final int _schemeEnd;
+  final int _hostStart;
+  final int _portStart;
+  final int _pathStart;
+  final int _queryStart;
+  final int _fragmentStart;
+  /// The scheme is often used to distinguish URIs.
+  /// To make comparisons more efficient, we cache the value, and
+  /// canonicalize a few known types.
+  String _schemeCache;
+
+  _SimpleUri(
+      this._uri,
+      this._schemeEnd,
+      this._hostStart,
+      this._portStart,
+      this._pathStart,
+      this._queryStart,
+      this._fragmentStart,
+      this._schemeCache);
+
+  bool get hasScheme => _schemeEnd > 0;
+  bool get hasAuthority => _hostStart > 0;
+  bool get hasUserInfo => _hostStart > _schemeEnd + 4;
+  bool get hasPort => _hostStart > 0 && _portStart + 1 < _pathStart;
+  bool get hasQuery => _queryStart < _fragmentStart;
+  bool get hasFragment => _fragmentStart < _uri.length;
+
+  bool get _isFile => _schemeEnd == 4 && _uri.startsWith("file");
+  bool get _isHttp => _schemeEnd == 4 && _uri.startsWith("http");
+  bool get _isHttps => _schemeEnd == 5 && _uri.startsWith("https");
+  bool get _isPackage => _schemeEnd == 7 && _uri.startsWith("package");
+  bool _isScheme(String scheme) =>
+    _schemeEnd == scheme.length && _uri.startsWith(scheme);
+
+  bool get hasAbsolutePath => _uri.startsWith("/", _pathStart);
+  bool get hasEmptyPath => _pathStart == _queryStart;
+
+  bool get isAbsolute => hasScheme && !hasFragment;
+
+  String get scheme {
+    if (_schemeEnd <= 0) return "";
+    if (_schemeCache != null) return _schemeCache;
+    if (_isHttp) {
+      _schemeCache = "http";
+    } else if (_isHttps) {
+      _schemeCache = "https";
+    } else if (_isFile) {
+      _schemeCache = "file";
+    } else if (_isPackage) {
+      _schemeCache = "package";
+    } else {
+      _schemeCache = _uri.substring(0, _schemeEnd);
+    }
+    return _schemeCache;
+  }
+  String get authority => _hostStart > 0 ?
+      _uri.substring(_schemeEnd + 3, _pathStart) : "";
+  String get userInfo => (_hostStart > _schemeEnd + 3) ?
+      _uri.substring(_schemeEnd + 3, _hostStart - 1) : "";
+  String get host =>
+      _hostStart > 0 ? _uri.substring(_hostStart, _portStart) : "";
+  int get port {
+    if (hasPort) return int.parse(_uri.substring(_portStart + 1, _pathStart));
+    if (_isHttp) return 80;
+    if (_isHttps) return 443;
+    return 0;
+  }
+  String get path =>_uri.substring(_pathStart, _queryStart);
+  String get query => (_queryStart < _fragmentStart) ?
+     _uri.substring(_queryStart + 1, _fragmentStart) : "";
+  String get fragment => (_fragmentStart < _uri.length) ?
+     _uri.substring(_fragmentStart + 1) : "";
+
+  String get origin {
+    // Check original behavior - W3C spec is wonky!
+    bool isHttp = _isHttp;
+    if (_schemeEnd < 0 || _hostStart == _portStart) {
+      throw new StateError("Cannot use origin without a scheme: $this");
+    }
+    if (!isHttp && !_isHttps) {
+      throw new StateError(
+        "Origin is only applicable schemes http and https: $this");
+    }
+    if (_hostStart == _schemeEnd + 3) {
+      return _uri.substring(0, _pathStart);
+    }
+    // Need to drop anon-empty userInfo.
+    return _uri.substring(0, _schemeEnd + 3) +
+           _uri.substring(_hostStart, _pathStart);
+  }
+
+  List<String> get pathSegments {
+    int start = _pathStart;
+    int end = _queryStart;
+    if (_uri.startsWith("/", start)) start++;
+    if (start == end) return const <String>[];
+    List<String> parts = [];
+    for (int i = start; i < end; i++) {
+      var char = _uri.codeUnitAt(i);
+      if (char == Uri._SLASH) {
+        parts.add(_uri.substring(start, i));
+        start = i + 1;
+      }
+    }
+    parts.add(_uri.substring(start, end));
+    return new List<String>.unmodifiable(parts);
+  }
+
+  Map<String, String> get queryParameters {
+    if (!hasQuery) return const <String, String>{};
+    return new UnmodifiableMapView<String, String>(
+        Uri.splitQueryString(query));
+  }
+
+  Map<String, List<String>> get queryParametersAll {
+    if (!hasQuery) return const <String, List<String>>{};
+    Map queryParameterLists = Uri._splitQueryStringAll(query);
+    for (var key in queryParameterLists.keys) {
+      queryParameterLists[key] =
+          new List<String>.unmodifiable(queryParameterLists[key]);
+    }
+    return new Map<String, List<String>>.unmodifiable(queryParameterLists);
+  }
+
+  bool _isPort(String port) {
+    int portDigitStart = _portStart + 1;
+    return portDigitStart + port.length == _pathStart &&
+           _uri.startsWith(port, portDigitStart);
+  }
+
+  Uri normalizePath() => this;
+
+  Uri removeFragment() {
+    if (!hasFragment) return this;
+    return new _SimpleUri(
+      _uri.substring(0, _fragmentStart),
+      _schemeEnd, _hostStart, _portStart,
+      _pathStart, _queryStart, _fragmentStart, _schemeCache);
+  }
+
+  Uri replace({String scheme,
+               String userInfo,
+               String host,
+               int port,
+               String path,
+               Iterable<String> pathSegments,
+               String query,
+               Map<String, dynamic/*String|Iterable<String>*/> queryParameters,
+               String fragment}) {
+    bool schemeChanged = false;
+    if (scheme != null) {
+      scheme = Uri._makeScheme(scheme, 0, scheme.length);
+      schemeChanged = !_isScheme(scheme);
+    } else {
+      scheme = this.scheme;
+    }
+    bool isFile = (scheme == "file");
+    if (userInfo != null) {
+      userInfo = Uri._makeUserInfo(userInfo, 0, userInfo.length);
+    } else if (_hostStart > 0) {
+      userInfo = _uri.substring(_schemeEnd + 3, _hostStart);
+    } else {
+      userInfo = "";
+    }
+    if (port != null) {
+      port = Uri._makePort(port, scheme);
+    } else {
+      port = this.hasPort ? this.port : null;
+      if (schemeChanged) {
+        // The default port might have changed.
+        port = Uri._makePort(port, scheme);
+      }
+    }
+    if (host != null) {
+      host = Uri._makeHost(host, 0, host.length, false);
+    } else if (_hostStart > 0) {
+      host = _uri.substring(_hostStart, _portStart);
+    } else if (userInfo.isNotEmpty || port != null || isFile) {
+      host = "";
+    }
+
+    bool hasAuthority = host != null;
+    if (path != null || pathSegments != null) {
+      path = Uri._makePath(path, 0, _stringOrNullLength(path), pathSegments,
+                           scheme, hasAuthority);
+    } else {
+      path = _uri.substring(_pathStart, _queryStart);
+      if ((isFile || (hasAuthority && !path.isEmpty)) &&
+          !path.startsWith('/')) {
+        path = "/" + path;
+      }
+    }
+
+    if (query != null || queryParameters != null) {
+      query =
+          Uri._makeQuery(query, 0, _stringOrNullLength(query), queryParameters);
+    } else if (_queryStart < _fragmentStart) {
+      query = _uri.substring(_queryStart, _fragmentStart);
+    }
+
+    if (fragment != null) {
+      fragment = Uri._makeFragment(fragment, 0, fragment.length);
+    } else if (_fragmentStart < _uri.length) {
+      fragment = _uri.substring(_fragmentStart);
+    }
+
+    return new Uri._internal(
+        scheme, userInfo, host, port, path, query, fragment);
+  }
+
+  Uri resolve(String reference) {
+    return resolveUri(Uri.parse(reference));
+  }
+
+  Uri resolveUri(Uri reference) {
+    if (reference is _SimpleUri) {
+      return _simpleMerge(this, reference);
+    }
+    return _toNonSimple().resolveUri(reference);
+  }
+
+  // Merge two simple URIs. This should always result in a prefix of
+  // one concatentated with a suffix of the other, which is again simple.
+  // In a few cases, there might be a need for extra normalization, when
+  // resolving on top of a known scheme.
+  Uri _simpleMerge(_SimpleUri base, _SimpleUri ref) {
+    if (ref.hasScheme) return ref;
+    if (ref.hasAuthority) {
+      if (!base.hasScheme) return ref;
+      bool isSimple = true;
+      if (base._isFile) {
+        isSimple = !ref.hasEmptyPath;
+      } else if (base._isHttp) {
+        isSimple = !ref._isPort("80");
+      } else if (base._isHttps) {
+        isSimple = !ref._isPort("443");
+      }
+      if (isSimple) {
+        var delta = base._schemeEnd + 1;
+        var newUri = base._uri.substring(0, base._schemeEnd + 1) +
+                     ref._uri.substring(ref._schemeEnd + 1);
+        return new _SimpleUri(newUri,
+           base._schemeEnd,
+           ref._hostStart + delta,
+           ref._portStart + delta,
+           ref._pathStart + delta,
+           ref._queryStart + delta,
+           ref._fragmentStart + delta,
+           base._schemeCache);
+      } else {
+        // Slowcase.
+        return _toNonSimple().resolveUri(ref);
+      }
+    }
+    if (ref.hasEmptyPath) {
+      if (ref.hasQuery) {
+        int delta = base._queryStart - ref._queryStart;
+        var newUri = base._uri.substring(0, base._queryStart) +
+                     ref._uri.substring(ref._queryStart);
+        return new _SimpleUri(newUri,
+           base._schemeEnd,
+           base._hostStart,
+           base._portStart,
+           base._pathStart,
+           ref._queryStart + delta,
+           ref._fragmentStart + delta,
+           base._schemeCache);
+      }
+      if (ref.hasFragment) {
+        int delta = base._fragmentStart - ref._fragmentStart;
+        var newUri = base._uri.substring(0, base._fragmentStart) +
+                     ref._uri.substring(ref._fragmentStart);
+        return new _SimpleUri(newUri,
+           base._schemeEnd,
+           base._hostStart,
+           base._portStart,
+           base._pathStart,
+           base._queryStart,
+           ref._fragmentStart + delta,
+           base._schemeCache);
+      }
+      return base.removeFragment();
+    }
+    if (ref.hasAbsolutePath) {
+      var delta = base._pathStart - ref._pathStart;
+      var newUri = base._uri.substring(0, base._pathStart) +
+                   ref._uri.substring(ref._pathStart);
+      return new _SimpleUri(newUri,
+        base._schemeEnd,
+        base._hostStart,
+        base._portStart,
+        base._pathStart,
+        ref._queryStart + delta,
+        ref._fragmentStart + delta,
+        base._schemeCache);
+    }
+    if (base.hasEmptyPath && base.hasAuthority) {
+      // ref has relative non-empty path.
+      var delta = base._pathStart - ref._pathStart + 1;
+      var newUri = "${base._uri.substring(0, base._pathStart)}/"
+                   "${ref._uri.substring(ref._pathStart)}";
+      return new _SimpleUri(newUri,
+        base._schemeEnd,
+        base._hostStart,
+        base._portStart,
+        base._pathStart,
+        ref._queryStart + delta,
+        ref._fragmentStart + delta,
+        base._schemeCache);
+    }
+    // Merge paths.
+    if (base._uri.startsWith("../", base._pathStart)) {
+      // Complex rare case, go slow.
+      return _toNonSimple().resolveUri(ref);
+    }
+    String baseUri = base._uri;
+    String refUri = ref._uri;
+    int baseStart = base._pathStart;
+    int baseEnd = base._queryStart;
+    int refStart = ref._pathStart;
+    int refEnd = ref._queryStart;
+    int backCount = 1;
+    while (refStart + 3 <= refEnd && refUri.startsWith("../", refStart)) {
+      refStart += 3;
+      backCount += 1;
+    }
+    if (refStart + 2 == refEnd && refUri.startsWith("..", refStart)) {
+      refStart += 2;
+      backCount += 1;
+    }
+
+    const slash = 0x2f;
+    while (baseEnd > baseStart) {
+      baseEnd--;
+      int char = baseUri.codeUnitAt(baseEnd);
+      if (char == slash) {
+        backCount--;
+        if (backCount == 0) break;
+      }
+    }
+    var delta;
+    var newUri;
+    if (baseEnd != baseStart || base._uri.startsWith("/", baseStart) ||
+        base.hasScheme || base.hasAuthority) {
+      delta = baseEnd - refStart + 1;
+      newUri = "${base._uri.substring(0, baseEnd)}/"
+               "${ref._uri.substring(refStart)}";
+    } else {
+      // We exactly removed all of a relative path.
+      // Example: foo:bar/baz  resolve vs. ../../qux  -> foo:qux
+      delta = baseEnd - refStart;
+      newUri = "${base._uri.substring(0, baseStart)}"
+               "${ref._uri.substring(refStart)}";
+    }
+
+    return new _SimpleUri(newUri,
+      base._schemeEnd,
+      base._hostStart,
+      base._portStart,
+      base._pathStart,
+      ref._queryStart + delta,
+      ref._fragmentStart + delta,
+      base._schemeCache);
+  }
+
+  // TODO: Deprecate Remove Kill Crush Destroy!

[floitsch, 2016/06/28 00:28:03]

TODO(ldap).
Also, make this more informative.

[Lasse Reichstein Nielsen, 2016/06/28 13:00:52]

Done. Whoops. :)

+  // Reason: requires knowing whether we are on Windows.
+  // That's a PLATFORM SPECIFIC THING.
+  // This should be File.pathFromUri, since File is platform specific already.
+  String toFilePath({bool windows}) {
+    if (_schemeEnd >= 0 && !_isFile) {
+      throw new UnsupportedError(
+          "Cannot extract a file path from a $scheme URI");
+    }
+    if (_queryStart < _uri.length) {
+      if (_queryStart < _fragmentStart) {
+        throw new UnsupportedError(
+            "Cannot extract a file path from a URI with a query component");
+      }
+      throw new UnsupportedError(
+          "Cannot extract a file path from a URI with a fragment component");
+    }
+    if (windows == null) windows = Uri._isWindows;
+    return windows ? Uri._toWindowsFilePath(this) : _toFilePath();
+  }
+
+  String _toFilePath() {
+    if (_hostStart < _portStart) {
+      // Has authority and non-empty host.
+      throw new UnsupportedError(
+        "Cannot extract a non-Windows file path from a file URI "
+        "with an authority");
+    }
+    return this.path;
+  }
+
+  UriData get data {
+    assert(scheme != "data");
+    return null;
+  }
+
+  int get hashCode => _uri.hashCode;
+
+  bool operator==(Object other) {
+    if (other is! Uri) return false;
+    return _uri == other.toString();
+  }
+
+  Uri _toNonSimple() {
+    return new Uri._internal(
+      this.scheme,
+      this.userInfo,
+      this.hasAuthority ? this.host: null,
+      this.hasPort ? this.port : null,
+      this.path,
+      this.hasQuery ? this.query : null,
+      this.hasFragment ? this.fragment : null
+    );
+  }
+
+  String toString() => _uri;
+}