From 7a1beb5dd376f1edc037f55ffd5ff9794922002d Mon Sep 17 00:00:00 2001
From: ShareVB <sharevb@gmail.com>
Date: Sun, 28 Apr 2024 14:57:17 +0200
Subject: [PATCH] feat: add OCRized text statistics

---
 components.d.ts                   | 11 +----------
 src/tools/ocr-image/ocr-image.vue | 28 +++++++++++++++++++++++++++-
 2 files changed, 28 insertions(+), 11 deletions(-)

diff --git a/components.d.ts b/components.d.ts
index 4a61d5b6..2e08fd8e 100644
--- a/components.d.ts
+++ b/components.d.ts
@@ -132,25 +132,16 @@ declare module '@vue/runtime-core' {
     NConfigProvider: typeof import('naive-ui')['NConfigProvider']
     NDivider: typeof import('naive-ui')['NDivider']
     NEllipsis: typeof import('naive-ui')['NEllipsis']
-    NFormItem: typeof import('naive-ui')['NFormItem']
-    NGi: typeof import('naive-ui')['NGi']
-    NGrid: typeof import('naive-ui')['NGrid']
     NH1: typeof import('naive-ui')['NH1']
     NH3: typeof import('naive-ui')['NH3']
     NIcon: typeof import('naive-ui')['NIcon']
-    NInputNumber: typeof import('naive-ui')['NInputNumber']
-    NLabel: typeof import('naive-ui')['NLabel']
     NLayout: typeof import('naive-ui')['NLayout']
     NLayoutSider: typeof import('naive-ui')['NLayoutSider']
     NMenu: typeof import('naive-ui')['NMenu']
     NScrollbar: typeof import('naive-ui')['NScrollbar']
-    NSlider: typeof import('naive-ui')['NSlider']
+    NSpace: typeof import('naive-ui')['NSpace']
     NSpin: typeof import('naive-ui')['NSpin']
     NStatistic: typeof import('naive-ui')['NStatistic']
-    NSwitch: typeof import('naive-ui')['NSwitch']
-    NTable: typeof import('naive-ui')['NTable']
-    NTag: typeof import('naive-ui')['NTag']
-    NSpin: typeof import('naive-ui')['NSpin']
     NumeronymGenerator: typeof import('./src/tools/numeronym-generator/numeronym-generator.vue')['default']
     OcrImage: typeof import('./src/tools/ocr-image/ocr-image.vue')['default']
     OtpCodeGeneratorAndValidator: typeof import('./src/tools/otp-code-generator-and-validator/otp-code-generator-and-validator.vue')['default']
diff --git a/src/tools/ocr-image/ocr-image.vue b/src/tools/ocr-image/ocr-image.vue
index bbde03b4..9609d7e2 100644
--- a/src/tools/ocr-image/ocr-image.vue
+++ b/src/tools/ocr-image/ocr-image.vue
@@ -4,6 +4,7 @@ import { createWorker } from 'tesseract.js';
 import { getDocument } from 'pdfjs-dist';
 import * as pdfJS from 'pdfjs-dist';
 import pdfJSWorkerURL from 'pdfjs-dist/build/pdf.worker?url';
+import { textStatistics } from '../text-statistics/text-statistics.service';
 import TextareaCopyable from '@/components/TextareaCopyable.vue';
 import { useQueryParamOrStorage } from '@/composable/queryParams';
 
@@ -115,6 +116,7 @@ const languagesOptions = Array.from(languages.map(l => ({
 
 const language = useQueryParamOrStorage({ name: 'lang', storageName: 'ocr-image:lang', defaultValue: 'eng' });
 
+const pageSeparator = '\n=============\n';
 const ocrInProgress = ref(false);
 const fileInput = ref() as Ref<File>;
 const ocrText = computedAsync(async () => {
@@ -125,6 +127,8 @@ const ocrText = computedAsync(async () => {
     return e.toString();
   }
 });
+const stats = computed(() => textStatistics(ocrText.value?.replace(new RegExp(pageSeparator, 'g'), ' ') || ''));
+const pageCount = computed(() => ocrText.value?.split(new RegExp(pageSeparator, 'g')).length || 0);
 
 async function onUpload(file: File) {
   if (file) {
@@ -180,7 +184,7 @@ async function ocr(file: File, language: string) {
   }
   await worker.terminate();
   ocrInProgress.value = false;
-  return allTexts.join('\n=============\n');
+  return allTexts.join(pageSeparator);
 };
 </script>
 
@@ -215,6 +219,28 @@ async function ocr(file: File, language: string) {
         size="small"
       />
     </div>
+
+    <c-card v-if="!ocrInProgress && stats" title="Statistics">
+      <n-space mt-3>
+        <n-statistic label="Character count" :value="stats.chars" />
+        <n-statistic label="Word count" :value="stats.words" />
+        <n-statistic label="Line count" :value="stats.lines" />
+        <n-statistic label="Pages count" :value="pageCount" />
+        <n-statistic label="Sentences count" :value="stats.sentences" />
+      </n-space>
+
+      <n-divider />
+
+      <n-space>
+        <n-statistic label="Chars (no spaces)" :value="stats.chars_no_spaces" />
+        <n-statistic label="Uppercase chars" :value="stats.chars_upper" />
+        <n-statistic label="Lowercase chars" :value="stats.chars_lower" />
+        <n-statistic label="Digit chars" :value="stats.chars_digits" />
+        <n-statistic label="Punctuations" :value="stats.chars_puncts" />
+        <n-statistic label="Spaces chars" :value="stats.chars_spaces" />
+        <n-statistic label="Word count (no punct)" :value="stats.words_no_puncs" />
+      </n-space>
+    </c-card>
   </div>
 </template>