Merge from vscode 81d7885dc2e9dc617e1522697a2966bc4025a45d (#5949)

* Merge from vscode 81d7885dc2e9dc617e1522697a2966bc4025a45d * Fix vs unit tests and hygiene issue * Fix strict null check issue
2026-02-16 10:58:30 -05:00 · 2019-06-10 18:27:09 -07:00
parent ff38bc8143
commit d15a3fcc98
926 changed files with 19529 additions and 11383 deletions
--- a/src/vs/base/common/uri.ts
+++ b/src/vs/base/common/uri.ts
@@ -96,6 +96,16 @@ const _empty = '';
 const _slash = '/';
 const _regexp = /^(([^:/?#]+?):)?(\/\/([^/?#]*))?([^?#]*)(\?([^#]*))?(#(.*))?/;

+function _isQueryStringScheme(scheme: string) {
+	switch (scheme.toLowerCase()) {
+		case 'http':
+		case 'https':
+		case 'ftp':
+			return true;
+	}
+	return false;
+}
+
 /**
 * Uniform Resource Identifier (URI) http://tools.ietf.org/html/rfc3986.
 * This class is a simple parser which creates the basic component parts
@@ -282,14 +292,14 @@ export class URI implements UriComponents {
 	static parse(value: string, _strict: boolean = false): URI {
 		const match = _regexp.exec(value);
 		if (!match) {
-			return new _URI(_empty, _empty, _empty, _empty, _empty);
+			return new _URI(_empty, _empty, _empty, _empty, _empty, _strict);
 		}
 		return new _URI(
 			match[2] || _empty,
-			decodeURIComponent(match[4] || _empty),
-			decodeURIComponent(match[5] || _empty),
-			decodeURIComponent(match[7] || _empty),
-			decodeURIComponent(match[9] || _empty),
+			decodeURIComponentFast(match[4] || _empty, false, false),
+			decodeURIComponentFast(match[5] || _empty, true, false),
+			decodeURIComponentFast(match[7] || _empty, false, _isQueryStringScheme(match[2])),
+			decodeURIComponentFast(match[9] || _empty, false, false),
 			_strict
 		);
 	}
@@ -321,7 +331,7 @@ export class URI implements UriComponents {

 		// normalize to fwd-slashes on windows,
 		// on other systems bwd-slashes are valid
-		// filename character, eg /f\oo/ba\r.txt
+		// filename character, e.g. /f\oo/ba\r.txt
 		if (isWindows) {
 			path = path.replace(/\\/g, _slash);
 		}
@@ -385,8 +395,8 @@ export class URI implements UriComponents {
 			return data;
 		} else {
 			const result = new _URI(data);
-			result._fsPath = (<UriState>data).fsPath;
 			result._formatted = (<UriState>data).external;
+			result._fsPath = (<UriState>data)._sep === _pathSepMarker ? (<UriState>data).fsPath : null;
 			return result;
 		}
 	}
@@ -402,10 +412,12 @@ export interface UriComponents {

 interface UriState extends UriComponents {
 	$mid: number;
-	fsPath: string;
 	external: string;
+	fsPath: string;
+	_sep: 1 | undefined;
 }

+const _pathSepMarker = isWindows ? 1 : undefined;

 // tslint:disable-next-line:class-name
 class _URI extends URI {
@@ -439,6 +451,7 @@ class _URI extends URI {
 		// cached state
 		if (this._fsPath) {
 			res.fsPath = this._fsPath;
+			res._sep = _pathSepMarker;
 		}
 		if (this._formatted) {
 			res.external = this._formatted;
@@ -463,6 +476,84 @@ class _URI extends URI {
 	}
 }

+function isHex(value: string, pos: number): boolean {
+	if (pos >= value.length) {
+		return false;
+	}
+	const code = value.charCodeAt(pos);
+	return (code >= CharCode.Digit0 && code <= CharCode.Digit9)// 0-9
+		|| (code >= CharCode.a && code <= CharCode.f) //a-f
+		|| (code >= CharCode.A && code <= CharCode.F); //A-F
+}
+
+
+function decodeURIComponentFast(uriComponent: string, isPath: boolean, isQueryString: boolean): string {
+
+	let res: string | undefined;
+	let nativeDecodePos = -1;
+
+	for (let pos = 0; pos < uriComponent.length; pos++) {
+		const code = uriComponent.charCodeAt(pos);
+
+		// decoding needed
+		if (code === CharCode.PercentSign && isHex(uriComponent, pos + 1) && isHex(uriComponent, pos + 2)) {
+
+			const chA = uriComponent.charCodeAt(pos + 1);
+			const chB = uriComponent.charCodeAt(pos + 2);
+
+			// when in a path -> check and accept %2f and %2F (fwd slash)
+			// when in a query string -> check and accept %3D, %26, and %3B (equals, ampersand, semi-colon)
+			if (
+				(isPath && chA === CharCode.Digit2 && (chB === CharCode.F || chB === CharCode.f))
+				||
+				(isQueryString && (
+					(chA === CharCode.Digit2 && chB === CharCode.Digit6) // %26
+					||
+					(chA === CharCode.Digit3 && (chB === CharCode.B || chB === CharCode.b || chB === CharCode.D || chB === CharCode.d)) // %3D, %3D
+				))
+			) {
+				if (nativeDecodePos !== -1) {
+					res += decodeURIComponent(uriComponent.substring(nativeDecodePos, pos));
+					nativeDecodePos = -1;
+				}
+
+				if (res !== undefined) {
+					res += uriComponent.substr(pos, 3);
+				}
+
+				pos += 2;
+				continue;
+			}
+
+			if (res === undefined) {
+				res = uriComponent.substring(0, pos);
+			}
+			if (nativeDecodePos === -1) {
+				nativeDecodePos = pos;
+			}
+
+			pos += 2;
+
+		} else {
+
+			if (nativeDecodePos !== -1) {
+				res += decodeURIComponent(uriComponent.substring(nativeDecodePos, pos));
+				nativeDecodePos = -1;
+			}
+
+			if (res !== undefined) {
+				res += String.fromCharCode(code);
+			}
+		}
+	}
+
+	if (nativeDecodePos !== -1) {
+		res += decodeURIComponent(uriComponent.substr(nativeDecodePos));
+	}
+
+	return res !== undefined ? res : uriComponent;
+}
+
 // reserved characters: https://tools.ietf.org/html/rfc3986#section-2.2
 const encodeTable: { [ch: number]: string } = {
 	[CharCode.Colon]: '%3A', // gen-delims
@@ -488,7 +579,7 @@ const encodeTable: { [ch: number]: string } = {
 	[CharCode.Space]: '%20',
 };

-function encodeURIComponentFast(uriComponent: string, allowSlash: boolean): string {
+function encodeURIComponentFast(uriComponent: string, isPath: boolean, isQueryString: boolean): string {
 	let res: string | undefined = undefined;
 	let nativeEncodePos = -1;

@@ -504,7 +595,8 @@ function encodeURIComponentFast(uriComponent: string, allowSlash: boolean): stri
 			|| code === CharCode.Period
 			|| code === CharCode.Underline
 			|| code === CharCode.Tilde
-			|| (allowSlash && code === CharCode.Slash)
+			|| (isPath && code === CharCode.Slash) // path => allow slash AS-IS
+			|| (isQueryString && (code === CharCode.Equals || code === CharCode.Ampersand || code === CharCode.Semicolon)) // query string => allow &=;
 		) {
 			// check if we are delaying native encode
 			if (nativeEncodePos !== -1) {
@@ -516,6 +608,20 @@ function encodeURIComponentFast(uriComponent: string, allowSlash: boolean): stri
 				res += uriComponent.charAt(pos);
 			}

+		} else if (code === CharCode.PercentSign && isHex(uriComponent, pos + 1) && isHex(uriComponent, pos + 2)) {
+			// at percentage encoded value
+
+			// check if we are delaying native encode
+			if (nativeEncodePos !== -1) {
+				res += encodeURIComponent(uriComponent.substring(nativeEncodePos, pos));
+				nativeEncodePos = -1;
+			}
+			// check if we write into a new string (by default we try to return the param)
+			if (res !== undefined) {
+				res += uriComponent.substr(pos, 3);
+			}
+			pos += 2;
+
 		} else {
 			// encoding needed, we need to allocate a new string
 			if (res === undefined) {
@@ -604,6 +710,7 @@ function _asFormatted(uri: URI, skipEncoding: boolean): string {

 	let res = '';
 	let { scheme, authority, path, query, fragment } = uri;
+
 	if (scheme) {
 		res += scheme;
 		res += ':';
@@ -620,22 +727,22 @@ function _asFormatted(uri: URI, skipEncoding: boolean): string {
 			authority = authority.substr(idx + 1);
 			idx = userinfo.indexOf(':');
 			if (idx === -1) {
-				res += encoder(userinfo, false);
+				res += encoder(userinfo, false, false);
 			} else {
 				// <user>:<pass>@<auth>
-				res += encoder(userinfo.substr(0, idx), false);
+				res += encoder(userinfo.substr(0, idx), false, false);
 				res += ':';
-				res += encoder(userinfo.substr(idx + 1), false);
+				res += encoder(userinfo.substr(idx + 1), false, false);
 			}
 			res += '@';
 		}
 		authority = authority.toLowerCase();
 		idx = authority.indexOf(':');
 		if (idx === -1) {
-			res += encoder(authority, false);
+			res += encoder(authority, false, false);
 		} else {
 			// <auth>:<port>
-			res += encoder(authority.substr(0, idx), false);
+			res += encoder(authority.substr(0, idx), false, false);
 			res += authority.substr(idx);
 		}
 	}
@@ -653,15 +760,15 @@ function _asFormatted(uri: URI, skipEncoding: boolean): string {
 			}
 		}
 		// encode the rest of the path
-		res += encoder(path, true);
+		res += encoder(path, true, false);
 	}
 	if (query) {
 		res += '?';
-		res += encoder(query, false);
+		res += encoder(query, false, _isQueryStringScheme(scheme));
 	}
 	if (fragment) {
 		res += '#';
-		res += !skipEncoding ? encodeURIComponentFast(fragment, false) : fragment;
+		res += !skipEncoding ? encodeURIComponentFast(fragment, false, false) : fragment;
 	}
 	return res;
 }