PDF ekleme ve metin düzenleme eklendi

2025-11-18 00:07:17 +03:00
parent b6155c3fcb
commit 49024fa6c9
5 changed files with 262 additions and 63 deletions
--- a/package.json
+++ b/package.json
@@ -17,6 +17,7 @@
    "@mui/material": "^6.1.1",
    "@supabase/supabase-js": "^2.81.1",
    "jszip": "^3.10.1",
+    "pdfjs-dist": "^5.4.394",
    "react": "^18.3.1",
    "react-dom": "^18.3.1",
    "react-dropzone": "^14.2.3",
--- a/src/components/OcrStep.jsx
+++ b/src/components/OcrStep.jsx
@@ -5,6 +5,7 @@ import {
  Button,
  LinearProgress,
  Stack,
+  TextField,
  Typography,
 } from '@mui/material';
 import { useNavigate } from 'react-router-dom';
@@ -22,9 +23,11 @@ const OcrStep = () => {
  const clearTranslation = useAppStore((state) => state.clearTranslation);
  const bookMetadata = useAppStore((state) => state.bookMetadata);
  const epubImports = useAppStore((state) => state.epubImports);
+  const pdfImports = useAppStore((state) => state.pdfImports);
  const [status, setStatus] = useState('idle');
  const [currentIndex, setCurrentIndex] = useState(0);
  const [previewText, setPreviewText] = useState('');
+  const [isEditing, setIsEditing] = useState(false);
  const total = croppedImages.length;
  const abortRef = useRef(false);

@@ -39,6 +42,12 @@ const OcrStep = () => {
  const [workerReady, setWorkerReady] = useState(false);
  const previewRef = useRef(null);
  // removed auto navigation to translation
+  const handlePreviewChange = (event) => {
+    if (!isEditing) return;
+    const value = event.target.value;
+    setPreviewText(value);
+    setOcrText(value);
+  };

  const orderedImages = useMemo(
    () => [...croppedImages].sort((a, b) => (a.order ?? 0) - (b.order ?? 0)),
@@ -119,13 +128,15 @@ const OcrStep = () => {
  }, [assetBase, isDev, orderedImages.length, setError]);

  useEffect(() => {
-    if (!orderedImages.length) return;
+    if (!orderedImages.length && !(epubImports.length || pdfImports.length)) return;
    setStatus('idle');
    setCurrentIndex(0);
-    setPreviewText('');
+    setPreviewText(ocrText || '');
+    if (!ocrText) {
      setOcrText('');
+    }
    clearTranslation();
-  }, [clearTranslation, orderedImages, setOcrText]);
+  }, [clearTranslation, epubImports.length, orderedImages, pdfImports.length, ocrText, setOcrText]);

  useEffect(() => {
    if (previewRef.current) {
@@ -176,7 +187,7 @@ const OcrStep = () => {
  }, [orderedImages, setError, setOcrText, status, total, workerReady]);


-  if (!orderedImages.length && !epubImports.length) {
+  if (!orderedImages.length && !epubImports.length && !pdfImports.length) {
    return (
      <Stack spacing={2}>
        <Alert severity="info">Önce görselleri cropla.</Alert>
@@ -187,7 +198,8 @@ const OcrStep = () => {
    );
  }

-  if (!orderedImages.length && epubImports.length) {
+  if (!orderedImages.length && (epubImports.length || pdfImports.length)) {
+    const documentCount = epubImports.length + pdfImports.length;
    return (
      <Stack spacing={4}>
        {bookMetadata && (
@@ -197,15 +209,34 @@ const OcrStep = () => {
          </Typography>
        )}
        <Box textAlign="center">
-          <Typography variant="h5">EPUB metni ayrıştırıldı</Typography>
+          <Typography variant="h5">Belge metni ayrıştırıldı</Typography>
          <Typography color="text.secondary">
-            {epubImports.length > 1
-              ? `${epubImports.length} EPUB dosyasından metin çıkarıldı.`
-              : 'EPUB dosyasındaki metin çıkarıldı.'}
+            {documentCount > 1
+              ? `${documentCount} belge dosyasından metin çıkarıldı.`
+              : 'Yüklediğin belgedeki metin çıkarıldı.'}
          </Typography>
        </Box>
        <Box sx={{ p: 2, borderRadius: 2, bgcolor: 'background.default' }}>
+          <Stack direction="row" alignItems="center" justifyContent="space-between">
            <Typography variant="subtitle1">Ön izleme</Typography>
+            <Button
+              size="small"
+              variant={isEditing ? 'contained' : 'outlined'}
+              onClick={() => setIsEditing((prev) => !prev)}
+            >
+              {isEditing ? 'Görünüme geç' : 'Metni düzenle'}
+            </Button>
+          </Stack>
+          {isEditing ? (
+            <TextField
+              fullWidth
+              multiline
+              minRows={6}
+              value={previewText}
+              onChange={handlePreviewChange}
+              sx={{ mt: 1 }}
+            />
+          ) : (
            <Box
              ref={previewRef}
              sx={{
@@ -219,8 +250,9 @@ const OcrStep = () => {
                pr: 1,
              }}
            >
-            {ocrText?.trim() || 'Metin ayrıştırılıyor...'}
+              {previewText || ocrText?.trim() || 'Metin ayrıştırılıyor...'}
            </Box>
+          )}
        </Box>
        <Stack direction={{ xs: 'column', sm: 'row' }} spacing={2} justifyContent="space-between">
          <Button variant="contained" onClick={() => navigate('/')}>
@@ -272,7 +304,26 @@ const OcrStep = () => {
        </Typography>
      </Box>
      <Box sx={{ p: 2, borderRadius: 2, bgcolor: 'background.default' }}>
+        <Stack direction="row" alignItems="center" justifyContent="space-between">
          <Typography variant="subtitle1">Ön izleme</Typography>
+          <Button
+            size="small"
+            variant={isEditing ? 'contained' : 'outlined'}
+            onClick={() => setIsEditing((prev) => !prev)}
+          >
+            {isEditing ? 'Görünüme geç' : 'Metni düzenle'}
+          </Button>
+        </Stack>
+        {isEditing ? (
+          <TextField
+            fullWidth
+            multiline
+            minRows={6}
+            value={previewText}
+            onChange={handlePreviewChange}
+            sx={{ mt: 1 }}
+          />
+        ) : (
          <Box
            ref={previewRef}
            sx={{
@@ -288,6 +339,7 @@ const OcrStep = () => {
          >
            {previewText || 'Metin bekleniyor'}
          </Box>
+        )}
      </Box>
      <Stack direction={{ xs: 'column', sm: 'row' }} spacing={2} justifyContent="space-between">
        <Button variant="contained" onClick={() => navigate('/bulk-crop')}>
--- a/src/components/UploadStep.jsx
+++ b/src/components/UploadStep.jsx
@@ -19,6 +19,7 @@ import {
 import { useNavigate } from 'react-router-dom';
 import { useAppStore } from '../store/useAppStore';
 import { extractTextFromEpub } from '../utils/epubImport';
+import { extractTextFromPdf } from '../utils/pdfImport';

 const dropzoneStyle = {
  border: '2px dashed rgba(108, 155, 207, 0.7)',
@@ -42,6 +43,8 @@ const UploadStep = () => {
  const setBookMetadata = useAppStore((state) => state.setBookMetadata);
  const epubImports = useAppStore((state) => state.epubImports);
  const setEpubImports = useAppStore((state) => state.setEpubImports);
+  const pdfImports = useAppStore((state) => state.pdfImports);
+  const setPdfImports = useAppStore((state) => state.setPdfImports);
  const setOcrText = useAppStore((state) => state.setOcrText);
  const clearTranslation = useAppStore((state) => state.clearTranslation);
  const setError = useAppStore((state) => state.setError);
@@ -51,12 +54,12 @@ const UploadStep = () => {
  const [selectedBookId, setSelectedBookId] = useState(bookMetadata?.id || null);
  const skipSearchRef = useRef(false);
  const [showResults, setShowResults] = useState(false);
-  const [epubProcessing, setEpubProcessing] = useState(false);
+  const [docProcessing, setDocProcessing] = useState(false);

  const onDrop = useCallback(
    async (acceptedFiles) => {
      if (!acceptedFiles.length) return;
-      setEpubProcessing(true);
+      setDocProcessing(true);
      const preservedMetadata = bookMetadata;
      const preservedTitle = bookTitle;
      resetFromStep('upload');
@@ -71,11 +74,19 @@ const UploadStep = () => {

      const imageFiles = [];
      const epubFiles = [];
+      const pdfFiles = [];
      acceptedFiles.forEach((file) => {
+        const lowerName = file.name?.toLowerCase() || '';
        const isEpub =
-          file.type === 'application/epub+zip' || file.name?.toLowerCase().endsWith('.epub');
+          file.type === 'application/epub+zip' ||
+          file.type === 'application/zip' ||
+          file.type === 'application/x-zip-compressed' ||
+          lowerName.endsWith('.epub');
+        const isPdf = file.type === 'application/pdf' || lowerName.endsWith('.pdf');
        if (isEpub) {
          epubFiles.push(file);
+        } else if (isPdf) {
+          pdfFiles.push(file);
        } else {
          imageFiles.push(file);
        }
@@ -92,13 +103,13 @@ const UploadStep = () => {
        setUploadedImages([...uploadedImages, ...mapped]);
      }

-      const importedEntries = [];
+      const importedEpubEntries = [];
      if (epubFiles.length) {
        for (const file of epubFiles) {
          try {
            // eslint-disable-next-line no-await-in-loop
            const parsed = await extractTextFromEpub(file);
-            importedEntries.push({
+            importedEpubEntries.push({
              id: crypto.randomUUID(),
              filename: file.name,
              size: file.size,
@@ -109,23 +120,54 @@ const UploadStep = () => {
            setError(error.message || `${file.name} okunamadı.`);
          }
        }
-        setEpubImports(importedEntries);
+        setEpubImports(importedEpubEntries);
+      } else {
+        setEpubImports([]);
+      }
+
+      const importedPdfEntries = [];
+      if (pdfFiles.length) {
+        for (const file of pdfFiles) {
+          try {
+            // eslint-disable-next-line no-await-in-loop
+            const parsed = await extractTextFromPdf(file);
+            importedPdfEntries.push({
+              id: crypto.randomUUID(),
+              filename: file.name,
+              size: file.size,
+              text: parsed.text,
+              metadata: parsed.metadata,
+            });
+          } catch (error) {
+            setError(error.message || `${file.name} okunamadı.`);
+          }
+        }
+        setPdfImports(importedPdfEntries);
+      } else {
+        setPdfImports([]);
+      }
+
+      if (importedEpubEntries.length || importedPdfEntries.length) {
        clearTranslation();
-        const combinedText = importedEntries.map((entry) => entry.text).filter(Boolean).join('\n\n');
+        const combinedText = [...importedEpubEntries, ...importedPdfEntries]
+          .map((entry) => entry.text)
+          .filter(Boolean)
+          .join('\n\n');
        if (combinedText) {
          setOcrText(combinedText);
        }
-        if (!preservedMetadata && importedEntries[0]?.metadata) {
-          const meta = importedEntries[0].metadata;
+        const firstMeta = importedEpubEntries[0]?.metadata || importedPdfEntries[0]?.metadata;
+        if (!preservedMetadata && firstMeta) {
+          const meta = firstMeta;
          setBookMetadata({
-            id: `epub-${crypto.randomUUID()}`,
-            title: meta.title || bookTitle || 'İsimsiz EPUB',
+            id: `doc-${crypto.randomUUID()}`,
+            title: meta.title || bookTitle || 'İsimsiz belge',
            subtitle: '',
            authors: meta.authors || [],
            publisher: meta.publisher || '',
            publishedDate: meta.publishedDate || '',
            description: meta.description || '',
-            pageCount: null,
+            pageCount: meta.pageCount || null,
            categories: meta.categories || [],
            averageRating: null,
            ratingsCount: null,
@@ -139,10 +181,10 @@ const UploadStep = () => {
            setBookTitle(meta.title);
          }
        }
-      } else {
-        setEpubImports([]);
+      } else if (!preservedMetadata) {
+        setBookMetadata(null);
      }
-      setEpubProcessing(false);
+      setDocProcessing(false);
    },
    [
      bookMetadata,
@@ -152,6 +194,7 @@ const UploadStep = () => {
      setBookMetadata,
      setBookTitle,
      setEpubImports,
+      setPdfImports,
      setError,
      setOcrText,
      setUploadedImages,
@@ -259,6 +302,7 @@ const UploadStep = () => {
      'application/zip': ['.epub'],
      'application/x-zip-compressed': ['.epub'],
      'application/octet-stream': ['.epub'],
+      'application/pdf': ['.pdf'],
    },
    multiple: true,
  });
@@ -303,8 +347,8 @@ const UploadStep = () => {
  }, [bookMetadata]);

  const hasImages = uploadedImages.length > 0;
-  const hasEpubImports = epubImports.length > 0;
-  const canProceed = hasImages || hasEpubImports;
+  const hasDocumentImports = epubImports.length > 0 || pdfImports.length > 0;
+  const canProceed = hasImages || hasDocumentImports;
  const nextPath = hasImages ? '/crop' : '/ocr';

  return (
@@ -466,10 +510,10 @@ const UploadStep = () => {
      <Box {...getRootProps()} sx={dropzoneStyle}>
        <input {...getInputProps()} />
        <Typography variant="h5" gutterBottom>
-          Görselleri veya EPUB dosyasını sürükleyip bırak ya da tıkla
+          Görselleri, EPUB veya PDF dosyalarını sürükleyip bırak ya da tıkla
        </Typography>
        <Typography color="text.secondary" gutterBottom>
-          .png, .jpg, .jpeg formatlarında çoklu görsel ya da .epub dosyaları yükleyebilirsin.
+          .png, .jpg, .jpeg, .webp görsellerinin yanı sıra .epub ve .pdf belgeleri yükleyebilirsin.
        </Typography>
        <Button variant="contained" color="primary">
          Dosya seç
@@ -479,9 +523,9 @@ const UploadStep = () => {
            Bırak ve yükleyelim!
          </Typography>
        )}
-        {epubProcessing && (
+        {docProcessing && (
          <Typography mt={2} color="text.secondary">
-            EPUB içeriği ayrıştırılıyor...
+            Belge içeriği ayrıştırılıyor...
          </Typography>
        )}
      </Box>
@@ -524,6 +568,44 @@ const UploadStep = () => {
        </Box>
      )}

+      {pdfImports.length > 0 && (
+        <Box>
+          <Typography variant="h6" gutterBottom>
+            Yüklenen PDF dosyaları ({pdfImports.length})
+          </Typography>
+          <Stack spacing={1.5}>
+            {pdfImports.map((item) => (
+              <Paper key={item.id} variant="outlined" sx={{ p: 2, borderRadius: 2 }}>
+                <Typography variant="subtitle1" sx={{ fontWeight: 600 }}>
+                  {item.metadata?.title || item.filename}
+                </Typography>
+                {item.metadata?.authors?.length ? (
+                  <Typography variant="body2" color="text.secondary" sx={{ fontStyle: 'italic' }}>
+                    {item.metadata.authors.join(', ')}
+                  </Typography>
+                ) : (
+                  <Typography variant="body2" color="text.secondary">
+                    Yazar bilgisi bulunamadı
+                  </Typography>
+                )}
+                <Typography variant="caption" color="text.secondary" display="block" mt={0.5}>
+                  {[
+                    item.metadata?.publisher,
+                    item.metadata?.language ? item.metadata.language.toUpperCase() : null,
+                    item.metadata?.pageCount ? `${item.metadata.pageCount} sayfa` : null,
+                  ]
+                    .filter(Boolean)
+                    .join(' • ')}
+                </Typography>
+                <Typography variant="caption" color="text.secondary" display="block">
+                  {item.filename} • {(item.size / (1024 * 1024)).toFixed(2)} MB
+                </Typography>
+              </Paper>
+            ))}
+          </Stack>
+        </Box>
+      )}
+
      <Box>
        <Typography variant="h6" gutterBottom>
          Yüklenen görseller ({uploadedImages.length})
--- a/src/store/useAppStore.js
+++ b/src/store/useAppStore.js
@@ -45,6 +45,7 @@ export const useAppStore = create((set) => ({
  croppedCoverImage: null,
  ocrText: '',
  epubImports: [],
+  pdfImports: [],
  bookTitle: '',
  bookMetadata: null,
  translatedText: '',
@@ -76,6 +77,7 @@ export const useAppStore = create((set) => ({
    }),
  setOcrText: (text) => set({ ocrText: text }),
  setEpubImports: (imports) => set({ epubImports: imports }),
+  setPdfImports: (imports) => set({ pdfImports: imports }),
  setBookTitle: (title) => set({ bookTitle: title }),
  setBookMetadata: (metadata) => set({ bookMetadata: metadata }),
  setTranslatedText: (text) => set({ translatedText: text }),
@@ -154,6 +156,7 @@ export const useAppStore = create((set) => ({
        draft.croppedCoverImage = null;
        draft.ocrText = '';
        draft.epubImports = [];
+        draft.pdfImports = [];
        draft.bookTitle = '';
        draft.bookMetadata = null;
        draft.translatedText = '';
--- a/src/utils/pdfImport.js
+++ b/src/utils/pdfImport.js
@@ -0,0 +1,61 @@
+import { GlobalWorkerOptions, getDocument } from 'pdfjs-dist/build/pdf.mjs';
+import pdfWorker from 'pdfjs-dist/build/pdf.worker.mjs?url';
+
+GlobalWorkerOptions.workerSrc = pdfWorker;
+
+const cleanText = (value = '') =>
+  value
+    .replace(/\r/g, '')
+    .replace(/\t/g, ' ')
+    .replace(/\s+\n/g, '\n')
+    .replace(/\n{3,}/g, '\n\n')
+    .replace(/[ \u00A0]{2,}/g, ' ')
+    .trim();
+
+const extractPageText = async (page) => {
+  const content = await page.getTextContent();
+  const strings = content.items
+    .map((item) => ('str' in item ? item.str : item?.unicode))
+    .filter(Boolean)
+    .map((text) => text.replace(/\s+/g, ' ').trim());
+  return cleanText(strings.join(' '));
+};
+
+export const extractTextFromPdf = async (file) => {
+  const arrayBuffer = await file.arrayBuffer();
+  const pdf = await getDocument({ data: arrayBuffer }).promise;
+  const texts = [];
+  for (let pageNumber = 1; pageNumber <= pdf.numPages; pageNumber += 1) {
+    // eslint-disable-next-line no-await-in-loop
+    const page = await pdf.getPage(pageNumber);
+    // eslint-disable-next-line no-await-in-loop
+    const text = await extractPageText(page);
+    if (text) {
+      texts.push(text);
+    }
+  }
+  let info = {};
+  try {
+    // pdf.getMetadata() resolves with info + metadata.
+    // eslint-disable-next-line no-await-in-loop
+    const metadata = await pdf.getMetadata();
+    info = metadata?.info || {};
+  } catch {
+    info = {};
+  }
+  return {
+    text: texts.join('\n\n').trim(),
+    metadata: {
+      title: info.Title || '',
+      authors: info.Author ? [info.Author] : [],
+      publisher: info.Producer || '',
+      language: info.Language || '',
+      publishedDate: info.CreationDate || '',
+      description: '',
+      identifiers: info.Identifier ? [{ identifier: info.Identifier }] : [],
+      filename: file.name,
+      fileSize: file.size,
+      pageCount: pdf.numPages,
+    },
+  };
+};