fix: do not check binary files and add Ada dictionary (#1011)

* fix: add Ada dictionary * fix: do not load binary files when spell checking. * fix: add Ada to known file types. * ci: Add integration test for Ada repo * fix: Issue with glob root resolution. * ci: Update Azure integration test
streetsidesoftware · Feb 26, 2021 · af04ead · af04ead
1 parent c748f09
commit af04ead
Show file tree

Hide file tree

Showing 15 changed files with 2,933 additions and 106 deletions.
diff --git a/.github/workflows/integration-test.yml b/.github/workflows/integration-test.yml
@@ -35,6 +35,7 @@ jobs:
           - wireapp/wire-desktop wireapp/wire-webapp
           - django/django
           - TheAlgorithms/Python
+          - AdaDoom3/AdaDoom3
 
     steps:
       - uses: actions/checkout@v2

diff --git a/integration-tests/config/config.json b/integration-tests/config/config.json
@@ -158,11 +158,11 @@
     {
       "path": "Azure/azure-rest-api-specs",
       "url": "https://github.com/Azure/azure-rest-api-specs.git",
-      "commit": "80f368e639e71088d201d235a9174a5d434c6ff6",
       "args": [
         "--config=cSpell.json",
         "**/*.{md,ts,js}"
-      ]
+      ],
+      "commit": "a06d912bc912010bdc0aa5d1d3c4b28ee1d89905"
     },
     {
       "path": "pycontribs/jira",
@@ -221,6 +221,14 @@
         "**/*.*"
       ],
       "commit": "1171f431bbf6b1d84556fb19de54626f59deb0d6"
+    },
+    {
+      "path": "AdaDoom3/AdaDoom3",
+      "url": "https://github.com/AdaDoom3/AdaDoom3.git",
+      "args": [
+        "**/*.*"
+      ],
+      "commit": "06416b73f27740b29581df0acbc6dd1ced9a8d7c"
     }
   ]
-}
+}
diff --git a/integration-tests/repositories/cspell.yaml b/integration-tests/repositories/cspell.yaml
@@ -1,2 +1,18 @@
+"$schema": "https://raw.githubusercontent.com/streetsidesoftware/cspell/master/cspell.schema.json"
+version: 0.2
+id: integration-test-default
 words:
   - octokit
+
+ignorePaths:
+  - "*.ktx"
+
+overrides:
+  - filename: "temp/TheAlgorithms/Python/**/*.py"
+    languageSettings:
+      - languageId: python
+        allowCompoundWords: true
+  - filename: "temp/AdaDoom3/AdaDoom3/**/*.py"
+    languageSettings:
+      - languageId: python
+        allowCompoundWords: false
diff --git a/integration-tests/snapshots/AdaDoom3/AdaDoom3/snapshot.txt b/integration-tests/snapshots/AdaDoom3/AdaDoom3/snapshot.txt
diff --git a/integration-tests/snapshots/Azure/azure-rest-api-specs/snapshot.txt b/integration-tests/snapshots/Azure/azure-rest-api-specs/snapshot.txt
diff --git a/packages/cspell-bundled-dicts/cspell-default.json b/packages/cspell-bundled-dicts/cspell-default.json
@@ -1,6 +1,7 @@
 {
-    "version": "0.1",
-    "name": "cspell default json",
+    "$schema": "https://raw.githubusercontent.com/streetsidesoftware/cspell/master/cspell.schema.json",
+    "version": "0.2",
+    "name": "cspell default settings",
     "id": "cspell-default",
     "language": "en",
     "description": "Default cspell configuration.",
@@ -78,6 +79,7 @@
         }
     ],
     "import": [
+        "@cspell/dict-ada/cspell-ext.json",
         "@cspell/dict-aws/cspell-ext.json",
         "@cspell/dict-bash/cspell-ext.json",
         "@cspell/dict-companies/cspell-ext.json",

diff --git a/packages/cspell-bundled-dicts/package-lock.json b/packages/cspell-bundled-dicts/package-lock.json
diff --git a/packages/cspell-bundled-dicts/package.json b/packages/cspell-bundled-dicts/package.json
@@ -31,9 +31,10 @@
   },
   "homepage": "https://github.com/streetsidesoftware/cspell#readme",
   "dependencies": {
+    "@cspell/dict-ada": "^1.1.1",
     "@cspell/dict-aws": "^1.0.13",
     "@cspell/dict-bash": "^1.0.11",
-    "@cspell/dict-companies": "^1.0.35",
+    "@cspell/dict-companies": "^1.0.36",
     "@cspell/dict-cpp": "^1.1.37",
     "@cspell/dict-cryptocurrencies": "^1.0.10",
     "@cspell/dict-csharp": "^1.0.10",

diff --git a/packages/cspell-lib/src/LanguageIds.ts b/packages/cspell-lib/src/LanguageIds.ts
@@ -22,6 +22,8 @@ export type LanguageExtensionDefinitions = LanguageExtensionDefinition[];
 export type ExtensionToLanguageIdMap = Map<string, Set<string>>;
 
 export const languageExtensionDefinitions: LanguageExtensionDefinitions = [
+    { id: 'ada', extensions: ['.adb', '.ads'] },
+    { id: 'apiblueprint', extensions: ['.apib', '.apiblueprint'] },
     { id: 'asciidoc', extensions: ['.adoc', '.asc', '.asciidoc'] },
     { id: 'bat', extensions: ['.bat', '.cmd'] },
     { id: 'clojure', extensions: ['.clj', '.cljs', '.cljx', '.clojure', '.edn'] },
@@ -56,6 +58,7 @@ export const languageExtensionDefinitions: LanguageExtensionDefinitions = [
         extensions: ['.json', '.jsonc', '.bowerrc', '.jshintrc', '.jscsrc', '.eslintrc', '.babelrc', '.webmanifest'],
     },
     { id: 'less', extensions: ['.less'] },
+    { id: 'literate haskell', extensions: ['.lhs'] },
     { id: 'lua', extensions: ['.lua'] },
     { id: 'makefile', extensions: ['.mk'] },
     { id: 'markdown', extensions: ['.md', '.mdown', '.markdown', '.markdn'] },
@@ -168,13 +171,16 @@ export const languageExtensionDefinitions: LanguageExtensionDefinitions = [
     { id: 'yaml', extensions: ['.eyaml', '.eyml', '.yaml', '.yml'] },
     { id: 'latex', extensions: ['.tex'] },
     { id: 'map', extensions: ['.map'] },
+
+    //
+    // Special file types used to prevent spell checking.
+    //
     { id: 'image', extensions: ['.jpg', '.png', '.jpeg', '.tiff', '.bmp', '.gif'] },
     // cspell:ignore woff
     {
         id: 'binary',
-        extensions: ['.gz', '.exe', '.dll', '.lib', '.obj', '.o', '.tff', '.woff', '.woff2', '.eot', '.ico'],
+        extensions: ['.gz', '.exe', '.dll', '.lib', '.obj', '.o', '.tff', '.woff', '.woff2', '.eot', '.ico', '.cur'],
     },
-    { id: 'apiblueprint', extensions: ['.apib', '.apiblueprint'] },
 ];
 
 export const languageIds: string[] = languageExtensionDefinitions.map(({ id }) => id);

diff --git a/packages/cspell-lib/src/Settings/CSpellSettingsServer.ts b/packages/cspell-lib/src/Settings/CSpellSettingsServer.ts
@@ -614,11 +614,12 @@ export function extractImportErrors(settings: CSpellSettings): ImportFileRefWith
 function resolveGlobRoot(settings: CSpellSettings, pathToSettingsFile: string): string {
     const settingsFileDir = path.dirname(pathToSettingsFile);
     const envGlobRoot = process.env[ENV_CSPELL_GLOB_ROOT];
+    const cwd = envGlobRoot || process.cwd();
     const defaultGlobRoot = envGlobRoot ?? '${cwd}';
     const rawRoot =
         settings.globRoot ??
         (settings.version === '0.1' || (envGlobRoot && !settings.version) ? defaultGlobRoot : settingsFileDir);
-    const globRoot = path.resolve(settingsFileDir, rawRoot.replace('${cwd}', process.cwd()));
+    const globRoot = path.resolve(settingsFileDir, rawRoot.replace('${cwd}', cwd));
     return globRoot;
 }
 

diff --git a/packages/cspell-lib/src/Settings/DefaultSettings.test.ts b/packages/cspell-lib/src/Settings/DefaultSettings.test.ts
@@ -8,6 +8,6 @@ describe('Validate Default Settings', () => {
 
     test('tests the default setting file is loaded', () => {
         const defaultSetting = DefaultSettings.getDefaultSettings();
-        expect(defaultSetting.name).toBe('cspell default json');
+        expect(defaultSetting.name).toBe('cspell default settings');
     });
 });
diff --git a/packages/cspell-lib/src/Settings/DefaultSettings.ts b/packages/cspell-lib/src/Settings/DefaultSettings.ts
@@ -75,6 +75,7 @@ export const _defaultSettings: CSpellSettingsWithSourceTrace = {
     name: 'Static Defaults',
     enabled: true,
     enabledLanguageIds: [
+        'ada',
         'csharp',
         'go',
         'javascript',
@@ -88,6 +89,7 @@ export const _defaultSettings: CSpellSettingsWithSourceTrace = {
         'text',
         'typescript',
         'typescriptreact',
+        'haskell',
         'html',
         'css',
         'less',

diff --git a/packages/cspell-lib/src/spellCheckFile.test.ts b/packages/cspell-lib/src/spellCheckFile.test.ts
@@ -122,8 +122,15 @@ describe('Validate Spell Checking Documents', () => {
         return URI.file(s(file)).toString();
     }
 
-    function d(uri: string, text?: string): Document {
-        return text === undefined ? { uri } : { uri, text };
+    function d(uri: string | Document, text?: string, languageId?: string): Document {
+        const doc: Document = typeof uri === 'string' ? { uri } : { ...uri };
+        if (text) {
+            doc.text = text;
+        }
+        if (languageId) {
+            doc.languageId = languageId;
+        }
+        return doc;
     }
 
     type ArrayType<T> = T extends (infer R)[] ? R : never;
@@ -136,22 +143,24 @@ describe('Validate Spell Checking Documents', () => {
 
     // cspell:ignore texxt
     test.each`
-        uri                     | text            | settings                       | options                                       | expected
-        ${f('src/not_found.c')} | ${''}           | ${{}}                          | ${{}}                                         | ${{ checked: false, errors: [errNoEnt('src/not_found.c')] }}
-        ${f('src/sample.c')}    | ${''}           | ${{}}                          | ${{}}                                         | ${{ checked: true, issues: [], localConfigFilepath: s('.cspell.json'), errors: undefined }}
-        ${f('src/sample.c')}    | ${''}           | ${{}}                          | ${{ noConfigSearch: true }}                   | ${{ checked: true, localConfigFilepath: undefined, errors: undefined }}
-        ${f('src/sample.c')}    | ${''}           | ${{ noConfigSearch: true }}    | ${{}}                                         | ${{ checked: true, localConfigFilepath: undefined, errors: undefined }}
-        ${f('src/sample.c')}    | ${''}           | ${{}}                          | ${{ configFile: s('../cspell.config.json') }} | ${{ checked: true, localConfigFilepath: s('../cspell.config.json'), errors: undefined }}
-        ${f('src/sample.c')}    | ${''}           | ${{ noConfigSearch: true }}    | ${{ configFile: s('../cspell.config.json') }} | ${{ checked: true, localConfigFilepath: s('../cspell.config.json'), errors: undefined }}
-        ${f('src/sample.c')}    | ${''}           | ${{ noConfigSearch: true }}    | ${{ noConfigSearch: false }}                  | ${{ checked: true, localConfigFilepath: s('.cspell.json'), errors: undefined }}
-        ${f('src/sample.c')}    | ${''}           | ${{}}                          | ${{}}                                         | ${{ document: oc(d(f('src/sample.c'))), errors: undefined }}
-        ${f('src/sample.c')}    | ${''}           | ${{}}                          | ${{ configFile: s('../cSpell.json') }}        | ${{ checked: false, localConfigFilepath: s('../cSpell.json'), errors: [eFailed(s('../cSpell.json'))] }}
-        ${f('src/not_found.c')} | ${''}           | ${{}}                          | ${{}}                                         | ${{ checked: false, errors: [errNoEnt('src/not_found.c')] }}
-        ${f(__filename)}        | ${''}           | ${{}}                          | ${{}}                                         | ${{ checked: true, localConfigFilepath: s('../cspell.config.json'), errors: undefined }}
-        ${'stdin:///'}          | ${'some text'}  | ${{ languageId: 'plaintext' }} | ${{}}                                         | ${{ checked: true, issues: [], localConfigFilepath: undefined, errors: undefined }}
-        ${'stdin:///'}          | ${'some text'}  | ${{ languageId: 'plaintext' }} | ${{}}                                         | ${{ document: oc(d('stdin:///')) }}
-        ${'stdin:///'}          | ${'some texxt'} | ${{ languageId: 'plaintext' }} | ${{}}                                         | ${{ checked: true, issues: i('texxt'), localConfigFilepath: undefined, errors: undefined }}
-        ${'stdin:///'}          | ${''}           | ${{ languageId: 'plaintext' }} | ${{}}                                         | ${{ checked: false, issues: [], localConfigFilepath: undefined, errors: [err('Unsupported schema: "stdin", open "stdin:/"')] }}
+        uri                                               | text            | settings                       | options                                       | expected
+        ${f('src/not_found.c')}                           | ${''}           | ${{}}                          | ${{}}                                         | ${{ checked: false, errors: [errNoEnt('src/not_found.c')] }}
+        ${f('src/sample.c')}                              | ${''}           | ${{}}                          | ${{}}                                         | ${{ checked: true, issues: [], localConfigFilepath: s('.cspell.json'), errors: undefined }}
+        ${f('src/sample.c')}                              | ${''}           | ${{}}                          | ${{ noConfigSearch: true }}                   | ${{ checked: true, localConfigFilepath: undefined, errors: undefined }}
+        ${f('src/sample.c')}                              | ${''}           | ${{ noConfigSearch: true }}    | ${{}}                                         | ${{ checked: true, localConfigFilepath: undefined, errors: undefined }}
+        ${f('src/sample.c')}                              | ${''}           | ${{}}                          | ${{ configFile: s('../cspell.config.json') }} | ${{ checked: true, localConfigFilepath: s('../cspell.config.json'), errors: undefined }}
+        ${f('src/sample.c')}                              | ${''}           | ${{ noConfigSearch: true }}    | ${{ configFile: s('../cspell.config.json') }} | ${{ checked: true, localConfigFilepath: s('../cspell.config.json'), errors: undefined }}
+        ${f('src/sample.c')}                              | ${''}           | ${{ noConfigSearch: true }}    | ${{ noConfigSearch: false }}                  | ${{ checked: true, localConfigFilepath: s('.cspell.json'), errors: undefined }}
+        ${f('src/sample.c')}                              | ${''}           | ${{}}                          | ${{}}                                         | ${{ document: oc(d(f('src/sample.c'))), errors: undefined }}
+        ${f('src/sample.c')}                              | ${''}           | ${{}}                          | ${{ configFile: s('../cSpell.json') }}        | ${{ checked: false, localConfigFilepath: s('../cSpell.json'), errors: [eFailed(s('../cSpell.json'))] }}
+        ${f('src/not_found.c')}                           | ${''}           | ${{}}                          | ${{}}                                         | ${{ checked: false, errors: [errNoEnt('src/not_found.c')] }}
+        ${f(__filename)}                                  | ${''}           | ${{}}                          | ${{}}                                         | ${{ checked: true, localConfigFilepath: s('../cspell.config.json'), errors: undefined }}
+        ${'stdin:///'}                                    | ${'some text'}  | ${{ languageId: 'plaintext' }} | ${{}}                                         | ${{ checked: true, issues: [], localConfigFilepath: undefined, errors: undefined }}
+        ${'stdin:///'}                                    | ${'some text'}  | ${{ languageId: 'plaintext' }} | ${{}}                                         | ${{ document: oc(d('stdin:///')) }}
+        ${'stdin:///'}                                    | ${'some texxt'} | ${{ languageId: 'plaintext' }} | ${{}}                                         | ${{ checked: true, issues: i('texxt'), localConfigFilepath: undefined, errors: undefined }}
+        ${'stdin:///'}                                    | ${''}           | ${{ languageId: 'plaintext' }} | ${{}}                                         | ${{ checked: false, issues: [], localConfigFilepath: undefined, errors: [err('Unsupported schema: "stdin", open "stdin:/"')] }}
+        ${f('src/big_image.jpeg')}                        | ${''}           | ${{}}                          | ${{}}                                         | ${{ checked: false, errors: undefined }}
+        ${d(f('src/big_image.txt'), undefined, 'binary')} | ${''}           | ${{}}                          | ${{}}                                         | ${{ checked: false, errors: undefined }}
     `(
         'spellCheckFile $uri $settings $options',
         async ({ uri, text, settings, options, expected }: TestSpellCheckFile) => {

diff --git a/packages/cspell-lib/src/spellCheckFile.ts b/packages/cspell-lib/src/spellCheckFile.ts
@@ -88,6 +88,17 @@ export async function spellCheckDocument(
     options: SpellCheckFileOptions,
     settings: CSpellUserSettings
 ): Promise<SpellCheckFileResult> {
+    if (isBinaryDoc(document)) {
+        return {
+            document,
+            options,
+            settingsUsed: settings,
+            localConfigFilepath: undefined,
+            issues: [],
+            checked: false,
+            errors: undefined,
+        };
+    }
     try {
         return spellCheckFullDocument(await resolveDocument(document), options, settings);
     } catch (e) {
@@ -240,6 +251,19 @@ export function determineFinalDocumentSettings(
     };
 }
 
+function isBinaryDoc(document: Document): boolean {
+    return isBinaryFile(URI.parse(document.uri).fsPath, document.languageId);
+}
+
+function isBinaryFile(filename: string, languageId?: string | string[]): boolean {
+    const ext = path.extname(filename);
+    languageId = languageId || [];
+    languageId = typeof languageId === 'string' ? languageId.split(',') : languageId;
+    languageId = languageId.map((a) => a.trim());
+    const languageIds = new Set(languageId.length ? languageId : getLanguagesForExt(ext));
+    return languageIds.has('binary') || languageIds.has('image');
+}
+
 export function fileToDocument(file: string): Document;
 export function fileToDocument(file: string, text: string, languageId?: string, locale?: string): DocumentWithText;
 export function fileToDocument(

diff --git a/packages/cspell/src/lint.ts b/packages/cspell/src/lint.ts
@@ -170,7 +170,7 @@ export function runLint(cfg: CSpellApplicationConfiguration): Promise<RunResult>
     }
 
     async function run(): Promise<RunResult> {
-        if (cfg.root) {
+        if (cfg.options.root) {
             process.env[cspell.ENV_CSPELL_GLOB_ROOT] = cfg.root;
         }