From 0c03890545af5eae05697885f283f493d08be4df Mon Sep 17 00:00:00 2001
From: Nicolas Giard <github@ngpixel.com>
Date: Mon, 17 Feb 2025 17:00:34 -0500
Subject: [PATCH] feat: repeated words check validation

---
 src/components/DrawerChecks.vue | 31 +++++++++++++++
 src/stores/editor.js            |  6 ++-
 src/tools/repeated-words.js     | 69 +++++++++++++++++++++++++++++++++
 3 files changed, 104 insertions(+), 2 deletions(-)
 create mode 100644 src/tools/repeated-words.js

diff --git a/src/components/DrawerChecks.vue b/src/components/DrawerChecks.vue
index 0609391..abd7840 100644
--- a/src/components/DrawerChecks.vue
+++ b/src/components/DrawerChecks.vue
@@ -103,6 +103,7 @@ q-list
 import { onBeforeUnmount, onMounted } from 'vue'
 import { useQuasar } from 'quasar'
 import { checkArticles } from 'src/tools/articles'
+import { checkRepeatedWords } from 'src/tools/repeated-words'
 import { checkHyphenation } from 'src/tools/hyphenation'
 import { checkInclusiveLanguage } from 'src/tools/inclusive-language'
 import { checkNonAscii } from 'src/tools/non-ascii'
@@ -153,6 +154,13 @@ const valChecks = [
     description: 'Check for common placeholders',
     icon: 'mdi-select-remove',
     click: () => placeholdersCheck()
+  },
+  {
+    key: 'repeatedWords',
+    title: 'Repeated Words Check',
+    description: 'Check for accidental repeated terms',
+    icon: 'mdi-repeat',
+    click: () => repeatedWordsCheck()
   }
 ]
 
@@ -251,6 +259,25 @@ function placeholdersCheck (silent = false) {
   }
 }
 
+function repeatedWordsCheck (silent) {
+  const results = checkRepeatedWords(modelStore[docsStore.activeDocument.id].getValue())
+  if (results.count < 1) {
+    editorStore.setValidationCheckState('repeatedWords', 1)
+    editorStore.setValidationCheckDetails('repeatedWords', [])
+    if (!silent) {
+      $q.notify({
+        message: 'Looks good!',
+        caption: 'No repeated terms found.',
+        color: 'positive',
+        icon: 'mdi-repeat'
+      })
+    }
+  } else {
+    editorStore.setValidationCheckState('repeatedWords', -2)
+    editorStore.setValidationCheckDetails('repeatedWords', results)
+  }
+}
+
 function runAllChecks () {
   editorStore.clearErrors()
   articlesCheck(true)
@@ -258,6 +285,7 @@ function runAllChecks () {
   inclusiveLangCheck(true)
   nonAsciiCheck(true)
   placeholdersCheck(true)
+  repeatedWordsCheck(true)
 
   if (editorStore.errors.length < 1) {
     $q.notify({
@@ -290,6 +318,9 @@ function runSelectedCheck (key) {
     case 'placeholders':
       placeholdersCheck(true)
       break
+    case 'repeatedWords':
+      repeatedWordsCheck(true)
+      break
   }
 }
 
diff --git a/src/stores/editor.js b/src/stores/editor.js
index cf55e16..90ebbed 100644
--- a/src/stores/editor.js
+++ b/src/stores/editor.js
@@ -72,14 +72,16 @@ export const useEditorStore = defineStore('editor', {
       hyphenation: 0,
       inclusiveLanguage: 0,
       nonAscii: 0,
-      placeholders: 0
+      placeholders: 0,
+      repeatedWords: 0
     },
     validationChecksDetails: {
       articles: [],
       hyphenation: [],
       inclusiveLanguage: [],
       nonAscii: [],
-      placeholders: []
+      placeholders: [],
+      repeatedWords: []
     },
     wordWrap: true,
     workingDirectory: '',
diff --git a/src/tools/repeated-words.js b/src/tools/repeated-words.js
new file mode 100644
index 0000000..36b57cd
--- /dev/null
+++ b/src/tools/repeated-words.js
@@ -0,0 +1,69 @@
+import { sortBy } from 'lodash-es'
+import { decorationsStore } from 'src/stores/models'
+
+export function checkRepeatedWords (text) {
+  const matchRgx = /\b(\w+)\s+\1\b/gi
+  const textLines = text.split('\n')
+
+  const decorations = []
+  const occurences = []
+  const details = []
+  const termCount = {}
+  for (const [lineIdx, line] of textLines.entries()) {
+    for (const match of line.matchAll(matchRgx)) {
+      const term = match[1].toLowerCase()
+      let occIdx = occurences.indexOf(term)
+      if (occIdx < 0) {
+        occIdx = occurences.push(term) - 1
+      }
+      decorations.push({
+        options: {
+          hoverMessage: {
+            value: `Repeated term "${match[1]}" detected.`
+          },
+          className: 'dec-warning',
+          minimap: {
+            position: 1
+          },
+          glyphMarginClassName: 'dec-warning-margin'
+        },
+        range: {
+          startLineNumber: lineIdx + 1,
+          startColumn: match.index + 1,
+          endLineNumber: lineIdx + 1,
+          endColumn: match.index + 1 + match[0].length
+        }
+      })
+      details.push({
+        key: crypto.randomUUID(),
+        group: occIdx + 1,
+        message: match[1].toLowerCase(),
+        range: {
+          startLineNumber: lineIdx + 1,
+          startColumn: match.index + 1,
+          endLineNumber: lineIdx + 1,
+          endColumn: match.index + 1 + match[0].length
+        }
+      })
+      if (termCount[term]) {
+        termCount[term]++
+      } else {
+        termCount[term] = 1
+      }
+    }
+  }
+
+  decorationsStore.get('repeatedWords').set(decorations)
+
+  return {
+    count: decorations.length,
+    details: sortBy(details, d => d.range.startLineNumber),
+    hasTextOutput: true,
+    getTextOutput: () => {
+      return `Repeated Words
+-------------
+${Object.entries(termCount).map(([k, v]) => `${k} (${v})`).join('\n')}
+`
+    }
+  }
+}