Moved regular expressions used in hot paths to global constants so they don't need to be reconstructed each time. (#10506)

Author: erictraut

packages/pyright-internal/src/analyzer/docStringUtils.ts

@@ -9,9 +9,12 @@
  * (https://www.python.org/dev/peps/pep-0257/).
  */
 
+const docStringCrRegEx = /\r/g;
+const docStringTabRegEx = /\t/g;
+
 export function cleanAndSplitDocString(rawString: string): string[] {
     // Remove carriage returns and replace tabs.
-    const unescaped = rawString.replace(/\r/g, '').replace(/\t/g, '        ');
+    const unescaped = rawString.replace(docStringCrRegEx, '').replace(docStringTabRegEx, '        ');
 
     // Split into lines.
     const lines = unescaped.split('\n');

packages/pyright-internal/src/analyzer/importStatementUtils.ts

@@ -38,6 +38,9 @@ import { ImportResult, ImportType } from './importResult';
 import { getTokenAfter, getTokenAt } from './parseTreeUtils';
 import * as SymbolNameUtils from './symbolNameUtils';
 
+const underscoreRegEx = /_/g;
+const indentTextRegEx = /^\s*$/;
+
 export interface ImportStatement {
     node: ImportNode | ImportFromNode;
     subnode?: ImportAsNode;
@@ -238,8 +241,8 @@ function _compareImportNames(name1: string, name2: string) {
     // This can't be reproduced by a normal string compare in TypeScript, since '_' > 'A'.
     // Replace all '_' with '=' which guarantees '=' < 'A'.
     // Safe to do as '=' is an invalid char in Python names.
-    const name1toCompare = name1.replace(/_/g, '=');
-    const name2toCompare = name2.replace(/_/g, '=');
+    const name1toCompare = name1.replace(underscoreRegEx, '=');
+    const name2toCompare = name2.replace(underscoreRegEx, '=');
     return compareStringsCaseSensitive(name1toCompare, name2toCompare);
 }
 
@@ -293,7 +296,7 @@ function _getTextEditsForAutoImportSymbolAddition(
             indentText = parseFileResults.text.substr(firstSymbolLineRange.start, firstSymbolPos.character);
 
             // Is the indent text composed of whitespace only?
-            if (/^\s*$/.test(indentText)) {
+            if (indentTextRegEx.test(indentText)) {
                 useOnePerLineFormatting = true;
             }
         }

packages/pyright-internal/src/analyzer/tracePrinter.ts

@@ -19,6 +19,8 @@ import * as ParseTreeUtils from './parseTreeUtils';
 import { Symbol } from './symbol';
 import { OverloadedType, Type, TypeBase, TypeCategory } from './types';
 
+const separatorRegExp = /[\\/]/g;
+
 export type PrintableType = ParseNode | Declaration | Symbol | Type | undefined;
 
 export interface TracePrinter {
@@ -36,7 +38,6 @@ export function createTracePrinter(roots: Uri[], includeRoots: boolean = false):
     // ex) d:/root/.env/lib/site-packages, d:/root/.env
     roots = roots.sort((a, b) => a.key.localeCompare(b.key)).reverse();
 
-    const separatorRegExp = /[\\/]/g;
     function printFileOrModuleName(fileUriOrModule: Uri | AbsoluteModuleDescriptor | undefined) {
         if (fileUriOrModule) {
             if (Uri.is(fileUriOrModule)) {

packages/pyright-internal/src/analyzer/typePrinterUtils.ts

@@ -7,10 +7,10 @@
  * Simple utility functions used by the type printer.
  */
 
-export function printStringLiteral(value: string, quotation = '"'): string {
-    const singleTickRegEx = /'/g;
-    const escapedDoubleQuoteRegEx = /\\"/g;
+const singleTickRegEx = /'/g;
+const escapedDoubleQuoteRegEx = /\\"/g;
 
+export function printStringLiteral(value: string, quotation = '"'): string {
     // JSON.stringify will perform proper escaping for " case.
     // So, we only need to do our own escaping for ' case.
     let literalStr = JSON.stringify(value).toString();

packages/pyright-internal/src/common/pathUtils.ts

@@ -32,6 +32,7 @@ export interface FileSpec {
 }
 
 const _includeFileRegex = /\.pyi?$/;
+const _wildcardRootRegex = /[*?]/;
 
 export namespace FileSpec {
     export function is(value: any): value is FileSpec {
@@ -166,10 +167,10 @@ export function getRelativePath(dirPath: string, relativeTo: string) {
     return relativePath;
 }
 
+const separatorRegExp = /[\\/]/g;
 const getInvalidSeparator = (sep: string) => (sep === '/' ? '\\' : '/');
 export function normalizeSlashes(pathString: string, sep = path.sep): string {
     if (pathString.includes(getInvalidSeparator(sep))) {
-        const separatorRegExp = /[\\/]/g;
         return pathString.replace(separatorRegExp, sep);
     }
 
@@ -583,7 +584,7 @@ export function getWildcardRoot(rootPath: string, fileSpec: string): string {
         if (component === '**') {
             break;
         } else {
-            if (component.match(/[*?]/)) {
+            if (component.match(_wildcardRootRegex)) {
                 break;
             }
 

packages/pyright-internal/src/common/realFileSystem.ts

@@ -10,6 +10,7 @@ import * as fs from 'fs';
 import * as tmp from 'tmp';
 import { isMainThread } from 'worker_threads';
 
+import { Disposable } from 'vscode-jsonrpc';
 import { CaseSensitivityDetector } from './caseSensitivityDetector';
 import { ConsoleInterface, NullConsole } from './console';
 import { randomBytesHex } from './crypto';
@@ -26,7 +27,6 @@ import { combinePaths, getRootLength } from './pathUtils';
 import { FileUri, FileUriSchema } from './uri/fileUri';
 import { Uri } from './uri/uri';
 import { getRootUri } from './uri/uriUtils';
-import { Disposable } from 'vscode-jsonrpc';
 
 // Automatically remove files created by tmp at process exit.
 tmp.setGracefulCleanup();
@@ -49,6 +49,8 @@ const DOT_ZIP = `.zip`;
 const DOT_EGG = `.egg`;
 const DOT_JAR = `.jar`;
 
+const zipPathRegEx = /[^\\/]\.(?:egg|zip|jar)[\\/]/;
+
 // Exactly the same as ZipOpenFS's getArchivePart, but supporting .egg files.
 // https://github.com/yarnpkg/berry/blob/64a16b3603ef2ccb741d3c44f109c9cfc14ba8dd/packages/yarnpkg-fslib/sources/ZipOpenFS.ts#L23
 function getArchivePart(path: string) {
@@ -464,7 +466,7 @@ export class RealFileSystem implements FileSystem {
 
     isInZip(uri: Uri): boolean {
         const path = uri.getFilePath();
-        return /[^\\/]\.(?:egg|zip|jar)[\\/]/.test(path) && yarnFS.isZip(path);
+        return zipPathRegEx.test(path) && yarnFS.isZip(path);
     }
 }
 

packages/pyright-internal/src/common/uri/baseUri.ts

@@ -8,11 +8,13 @@
 
 import { some } from '../collectionUtils';
 import { getRootLength, getShortenedFileName } from '../pathUtils';
-import { Uri } from './uri';
 import { cacheProperty } from './memoization';
+import { Uri } from './uri';
 
 export type JsonObjType = any;
 
+const backslashRegEx = /\\/g;
+
 export abstract class BaseUri implements Uri {
     protected constructor(private readonly _key: string) {}
 
@@ -245,7 +247,7 @@ export abstract class BaseUri implements Uri {
 
     protected normalizeSlashes(path: string): string {
         if (path.includes('\\')) {
-            return path.replace(/\\/g, '/');
+            return path.replace(backslashRegEx, '/');
         }
         return path;
     }

packages/pyright-internal/src/common/uri/uri.ts

@@ -120,6 +120,9 @@ export interface Uri {
     toJsonObj(): any;
 }
 
+const _dosPathRegex = /^\/[a-zA-Z]:\//;
+const _win32NormalizationRegex = /\//g;
+
 // Returns just the fsPath path portion of a vscode URI.
 function getFilePath(uri: URI): string {
     let filePath: string | undefined;
@@ -136,14 +139,14 @@ function getFilePath(uri: URI): string {
 
     // If this is a DOS-style path with a drive letter, remove
     // the leading slash.
-    if (filePath.match(/^\/[a-zA-Z]:\//)) {
+    if (filePath.match(_dosPathRegex)) {
         filePath = filePath.slice(1);
     }
 
     // vscode.URI normalizes the path to use the correct path separators.
     // We need to do the same.
     if (process?.platform === 'win32') {
-        filePath = filePath.replace(/\//g, '\\');
+        filePath = filePath.replace(_win32NormalizationRegex, '\\');
     }
 
     return filePath;
@@ -164,16 +167,16 @@ function normalizeUri(uri: string | URI): { uri: URI; str: string } {
     return { uri: finalURI, str: finalString };
 }
 
+const windowsUriRegEx = /^[a-zA-Z]:\\?/;
+const uriRegEx = /^[a-zA-Z][a-zA-Z0-9+.-]*:\/?\/?/;
+
 export namespace Uri {
     export interface IServiceProvider {
         get<T>(key: ServiceKey<T>): T;
     }
 
     export function maybeUri(value: string) {
-        const windows = /^[a-zA-Z]:\\?/;
-        const uri = /^[a-zA-Z][a-zA-Z0-9+.-]*:\/?\/?/;
-
-        return uri.test(value) && !windows.test(value);
+        return uriRegEx.test(value) && !windowsUriRegEx.test(value);
     }
 
     export function create(value: string, serviceProvider: IServiceProvider, checkRelative?: boolean): Uri;

packages/pyright-internal/src/common/uri/uriUtils.ts

@@ -8,16 +8,16 @@
 
 import type { Dirent } from 'fs';
 
+import { CaseSensitivityDetector } from '../caseSensitivityDetector';
 import { FileSystem, ReadOnlyFileSystem, Stats } from '../fileSystem';
 import {
     getRegexEscapedSeparator,
     isDirectoryWildcardPatternPresent,
     stripTrailingDirectorySeparator,
 } from '../pathUtils';
-import { Uri } from './uri';
 import { ServiceKeys } from '../serviceKeys';
-import { CaseSensitivityDetector } from '../caseSensitivityDetector';
 import { ServiceProvider } from '../serviceProvider';
+import { Uri } from './uri';
 
 export interface FileSpec {
     // File specs can contain wildcard characters (**, *, ?). This
@@ -36,6 +36,7 @@ export interface FileSpec {
 }
 
 const _includeFileRegex = /\.pyi?$/;
+const _wildcardRegex = /[*?]/;
 
 export namespace FileSpec {
     export function is(value: any): value is FileSpec {
@@ -264,7 +265,7 @@ export function getWildcardRoot(root: Uri, fileSpec: string): Uri {
         if (component === '**') {
             break;
         } else {
-            if (/[*?]/.test(component)) {
+            if (_wildcardRegex.test(component)) {
                 break;
             }
 

packages/pyright-internal/src/parser/parser.ts

@@ -160,6 +160,9 @@ interface SubscriptListResult {
     trailingComma: boolean;
 }
 
+const commentRegEx = /^(\s*#\s*type:\s*)([^\r\n]*)/;
+const ignoreCommentRegEx = /^ignore(\s|\[|$)/;
+
 export class ParseOptions {
     isStubFile: boolean;
     pythonVersion: PythonVersion;
@@ -4760,7 +4763,6 @@ export class Parser {
         }
 
         const interTokenContents = this._fileContents!.slice(curToken.start + curToken.length, nextToken.start);
-        const commentRegEx = /^(\s*#\s*type:\s*)([^\r\n]*)/;
         const match = interTokenContents.match(commentRegEx);
         if (!match) {
             return undefined;
@@ -4773,7 +4775,7 @@ export class Parser {
         // expression because mypy supports ignore comments of the
         // form ignore[errorCode, ...]. We'll treat these as regular
         // ignore statements (as though no errorCodes were included).
-        if (typeString.trim().match(/^ignore(\s|\[|$)/)) {
+        if (typeString.trim().match(ignoreCommentRegEx)) {
             return undefined;
         }