dev: Catch forbidden whole words (#2000)

Fix #1953
streetsidesoftware · Nov 16, 2021 · b76fd12 · b76fd12
1 parent 7c360c6
commit b76fd12
Show file tree

Hide file tree

Showing 4 changed files with 78 additions and 26 deletions.
diff --git a/packages/cspell-lib/samples/forbid-words/README.md b/packages/cspell-lib/samples/forbid-words/README.md
@@ -0,0 +1,12 @@
+# Test Flagged Words
+
+[`flagWords` does not mark all flagged words. · Issue #1953 · streetsidesoftware/cspell](https://github.com/streetsidesoftware/cspell/issues/1953)
+
+Here is a list of words that should be flagged:
+
+- therefor
+- they'd
+- they'll
+- they're
+- they've
+- therefor
diff --git a/packages/cspell-lib/samples/forbid-words/cspell.json b/packages/cspell-lib/samples/forbid-words/cspell.json
@@ -0,0 +1,13 @@
+{
+    "version": "0.2",
+    "language": "en",
+    "words": [],
+    "flagWords": [
+        "therefor",
+        "they'd",
+        "they'll",
+        "they're",
+        "they've",
+        "therefor"
+    ]
+}
diff --git a/packages/cspell-lib/src/textValidator.test.ts b/packages/cspell-lib/src/textValidator.test.ts
@@ -209,15 +209,20 @@ describe('Validate textValidator functions', () => {
     });
 
     test.each`
-        text                           | ignoreWords   | flagWords       | expected
-        ${'red'}                       | ${[]}         | ${undefined}    | ${[]}
-        ${'color'}                     | ${[]}         | ${undefined}    | ${[ov({ text: 'color', isFound: false })]}
-        ${'colour'}                    | ${[]}         | ${undefined}    | ${[ov({ text: 'colour', isFlagged: true })]}
-        ${'colour'}                    | ${['colour']} | ${undefined}    | ${[]}
-        ${'The ant ate the antelope.'} | ${[]}         | ${['fbd']}      | ${[]}
-        ${'The ant ate the antelope.'} | ${[]}         | ${['ate']}      | ${[ov({ text: 'ate', isFlagged: true })]}
-        ${'theANT_ateThe_antelope.'}   | ${[]}         | ${['ate']}      | ${[ov({ text: 'ate', isFlagged: true })]}
-        ${'The ant ate the antelope.'} | ${[]}         | ${['antelope']} | ${[ov({ text: 'antelope', isFlagged: true })]}
+        text                            | ignoreWords   | flagWords        | expected
+        ${'red'}                        | ${[]}         | ${undefined}     | ${[]}
+        ${'color'}                      | ${[]}         | ${undefined}     | ${[ov({ text: 'color', isFound: false })]}
+        ${'colour'}                     | ${[]}         | ${undefined}     | ${[ov({ text: 'colour', isFlagged: true })]}
+        ${'colour'}                     | ${['colour']} | ${undefined}     | ${[]}
+        ${'The ant ate the antelope.'}  | ${[]}         | ${['fbd']}       | ${[]}
+        ${'The ant ate the antelope.'}  | ${[]}         | ${['ate']}       | ${[ov({ text: 'ate', isFlagged: true })]}
+        ${'theANT_ateThe_antelope.'}    | ${[]}         | ${['ate']}       | ${[ov({ text: 'ate', isFlagged: true })]}
+        ${'The ant ate the antelope.'}  | ${[]}         | ${['antelope']}  | ${[ov({ text: 'antelope', isFlagged: true })]}
+        ${'This should be ok'}          | ${[]}         | ${[]}            | ${[]}
+        ${"This should've been ok"}     | ${[]}         | ${[]}            | ${[]}
+        ${"This should've not been ok"} | ${[]}         | ${["should've"]} | ${[ov({ text: "should've", isFlagged: true })]}
+        ${"They'll be allowed"}         | ${[]}         | ${[]}            | ${[]}
+        ${"They'll not be allowed"}     | ${[]}         | ${["they'll"]}   | ${[ov({ text: "They'll", isFlagged: true })]}
     `('Validate forbidden words', ({ text, ignoreWords, expected, flagWords }) => {
         const dict = getSpellingDictionaryCollectionSync({ ignoreWords });
         const result = [...validateText(text, dict, { ignoreCase: false, flagWords })];
@@ -265,29 +270,38 @@ const fruit = [
 const animals = ['ape', 'lion', 'tiger', 'Elephant', 'monkey', 'gazelle', 'antelope', 'aardvark', 'hyena'];
 const insects = ['ant', 'snail', 'beetle', 'worm', 'stink bug', 'centipede', 'millipede', 'flea', 'fly'];
 const words = [
-    'the',
+    'allowed',
     'and',
-    'is',
-    'has',
     'ate',
-    'light',
-    'dark',
-    'little',
+    'be',
+    'been',
+    'better',
     'big',
-    'we',
+    'dark',
+    'done',
+    'fixes',
+    'has',
     'have',
-    'published',
+    'is',
+    'known',
+    'light',
+    'little',
     'multiple',
-    'fixes',
-    'to',
-    'the',
+    'not',
     'problems',
-    'better',
-    'done',
-    'known',
-    "shouldn't",
+    'published',
+    'should',
+    'the',
+    'they',
+    'this',
+    'to',
+    'we',
+    "'ll",
     "couldn't",
     "should've",
+    "shouldn't",
+    "they'll",
+    "they've",
 ];
 
 const forbiddenWords = ['!colour', '!favour'];

diff --git a/packages/cspell-lib/src/textValidator.ts b/packages/cspell-lib/src/textValidator.ts
@@ -124,10 +124,14 @@ function lineValidator(dict: SpellingDictionary, options: ValidationOptions): Li
         return dict.isNoSuggestWord(word, options);
     }
 
-    function checkFlagWords(word: ValidationResult): ValidationResult {
+    function isWordFlagged(word: TextOffset): boolean {
         const isIgnored = isWordIgnored(word.text);
         const isFlagged = !isIgnored && testForFlaggedWord(word);
-        word.isFlagged = isFlagged;
+        return isFlagged;
+    }
+
+    function checkFlagWords(word: ValidationResult): ValidationResult {
+        word.isFlagged = isWordFlagged(word);
         return word;
     }
 
@@ -175,6 +179,15 @@ function lineValidator(dict: SpellingDictionary, options: ValidationOptions): Li
         }
 
         function checkPossibleWords(possibleWord: TextOffset) {
+            if (isWordFlagged(possibleWord)) {
+                const vr: ValidationResult = {
+                    ...possibleWord,
+                    line: lineSegment,
+                    isFlagged: true,
+                };
+                return [vr];
+            }
+
             const mismatches: ValidationResult[] = Text.extractWordsFromTextOffset(possibleWord)
                 .filter(filterAlreadyChecked)
                 .map((wo) => ({ ...wo, line: lineSegment }))