""" OCR Converter — extract text from images and scanned PDFs using Tesseract. System requirement: tesseract-ocr must be installed. Ubuntu/Debian: sudo apt install tesseract-ocr Fedora/RHEL: sudo dnf install tesseract macOS: brew install tesseract Python requirement: pip install pytesseract pdf2image python-docx PyMuPDF Supported input: jpg, jpeg, png, tiff, tif, bmp, gif, webp, pdf Supported output: txt, pdf (searchable), docx """ import os import uuid import json SUPPORTED_INPUT_EXTS = { ".jpg", ".jpeg", ".png", ".tiff", ".tif", ".bmp", ".gif", ".webp", ".pdf", } SUPPORTED_OUTPUT = {"txt", "pdf", "docx"} def _parse_url_entry(url_entry): if isinstance(url_entry, str): try: parsed = json.loads(url_entry) if isinstance(parsed, dict): return parsed except (json.JSONDecodeError, TypeError): name, ext = os.path.splitext(os.path.basename(url_entry)) return {"path": url_entry, "name": name, "ext": ext} elif isinstance(url_entry, dict): return url_entry return None def _require_pytesseract(): try: import pytesseract return pytesseract except ImportError: raise RuntimeError( "pytesseract is not installed. Run: pip install pytesseract " "Also install the system package: sudo apt install tesseract-ocr" ) def _get_images(full_path, ext_lower): """Return a list of PIL Images from the input file.""" if ext_lower == ".pdf": try: from pdf2image import convert_from_path except ImportError: raise RuntimeError( "pdf2image is not installed. Run: pip install pdf2image " "Also install: sudo apt install poppler-utils" ) return convert_from_path(full_path, dpi=200) else: from PIL import Image img = Image.open(full_path) img.load() return [img] def _run_ocr_text(images): """OCR a list of PIL Images; return list of text strings (one per image).""" tsr = _require_pytesseract() texts = [] for img in images: try: text = tsr.image_to_string(img) except Exception as e: text = f"[OCR error: {e}]" texts.append(text) return texts def _write_txt(texts, output_path): with open(output_path, "w", encoding="utf-8") as f: for i, text in enumerate(texts): if i > 0: f.write("\n\n--- Page Break ---\n\n") f.write(text) def _write_docx(texts, output_path): try: from docx import Document except ImportError: raise RuntimeError( "python-docx is not installed. Run: pip install python-docx" ) doc = Document() for i, text in enumerate(texts): if i > 0: doc.add_page_break() for line in text.splitlines(): doc.add_paragraph(line) doc.save(output_path) def _write_searchable_pdf(images, output_path): """Create a searchable PDF (image + invisible text layer) via Tesseract.""" tsr = _require_pytesseract() page_pdfs = [] for img in images: pdf_bytes = tsr.image_to_pdf_or_hocr(img, extension="pdf") page_pdfs.append(pdf_bytes) if len(page_pdfs) == 1: with open(output_path, "wb") as f: f.write(page_pdfs[0]) return # Merge pages using PyMuPDF (preferred) or PyPDF2 (fallback) try: import fitz merged = fitz.open() for page_bytes in page_pdfs: sub = fitz.open("pdf", page_bytes) merged.insert_pdf(sub) sub.close() merged.save(output_path) merged.close() return except ImportError: pass try: import io from PyPDF2 import PdfWriter, PdfReader writer = PdfWriter() for page_bytes in page_pdfs: reader = PdfReader(io.BytesIO(page_bytes)) for page in reader.pages: writer.add_page(page) with open(output_path, "wb") as f: writer.write(f) return except ImportError: pass # Last resort: write the first page only with open(output_path, "wb") as f: f.write(page_pdfs[0]) def _convert_single(file_object, target_format, upload_dir): file_path = file_object.get("path", "") file_name = file_object.get("name", "") or os.path.splitext( os.path.basename(file_path) )[0] file_ext = file_object.get("ext", os.path.splitext(file_path)[1]) if not file_path: raise ValueError("No file path provided") full_path = os.path.join(upload_dir, file_path) if not os.path.exists(full_path): full_path = file_path if not os.path.exists(full_path): raise FileNotFoundError(f"File not found: {os.path.basename(file_path)}") ext_lower = file_ext.lower() if file_ext.startswith(".") else f".{file_ext.lower()}" if ext_lower not in SUPPORTED_INPUT_EXTS: raise ValueError(f"Unsupported input format for OCR: {file_ext}") images = _get_images(full_path, ext_lower) output_folder = uuid.uuid4().hex output_dir = os.path.join(upload_dir, output_folder) os.makedirs(output_dir, exist_ok=True) fmt = target_format.lower() output_path = os.path.join(output_dir, f"{file_name}.{fmt}") if fmt == "txt": texts = _run_ocr_text(images) _write_txt(texts, output_path) elif fmt == "docx": texts = _run_ocr_text(images) _write_docx(texts, output_path) elif fmt == "pdf": _write_searchable_pdf(images, output_path) else: raise ValueError(f"Unsupported OCR output format: {target_format}") if not os.path.exists(output_path) or os.path.getsize(output_path) == 0: raise RuntimeError("OCR output file was not created or is empty.") return output_path def convert(urls, target_format, options, config): try: upload_dir = config.get("UPLOAD_DIR", "static/uploads") fmt = target_format.lower() if fmt not in SUPPORTED_OUTPUT: return { "error": True, "message": ( f"Unsupported OCR output format: {target_format}. " "Supported formats: txt, pdf, docx" ), } results = [] errors = [] for url_entry in urls: file_object = _parse_url_entry(url_entry) if not file_object: continue file_path = file_object.get("path", "") if not file_path or file_path == "empty": continue try: output_path = _convert_single(file_object, fmt, upload_dir) results.append(output_path) except Exception as e: fname = os.path.basename(file_path) if file_path else "unknown" errors.append(f"OCR failed for {fname}: {str(e)}") if not results and errors: return {"error": True, "message": "; ".join(errors)} if not results and not errors: return {"error": True, "message": "No files were provided for OCR."} return { "error": False, "results": results, "output_path": results[0] if results else "", "errors": errors if errors else None, } except Exception as e: return {"error": True, "message": f"OCR conversion failed: {str(e)}"}