# import pdfplumber # import re # import json # import os # import logging # # ---------------- CONFIG ---------------- # # request_payload = { # "JL No.": ["63", "124","190"], # "daag No": ["371", "261","519"] # } # # --------------------------------------- # # # 🔕 Suppress pdfminer font warnings # logging.getLogger("pdfminer").setLevel(logging.ERROR) # PDF_FOLDER = "/var/www/html/land-ocr/input_pdfs" # folder containing PDFs # JL_SET = set(request_payload["JL No."]) # DAAG_SET = set(request_payload["daag No"]) # results = [] # def scan_pdf(pdf_path): # output = { # "pdf": os.path.basename(pdf_path), # "matches": [] # } # with pdfplumber.open(pdf_path) as pdf: # for page_no, page in enumerate(pdf.pages, start=1): # words = page.extract_words(use_text_flow=True) or [] # page_text = " ".join(w["text"] for w in words) # found_jl = [] # found_daag = [] # # JL No detection # for jl in JL_SET: # if re.search(rf"\b{jl}\b", page_text): # found_jl.append(jl) # # Daag No detection # for daag in DAAG_SET: # if re.search(rf"\b{daag}\b", page_text): # found_daag.append(daag) # if found_jl or found_daag: # output["matches"].append({ # "page": page_no, # "JL_No": found_jl, # "Daag_No": found_daag # }) # if output["matches"]: # return output # return None # for file in os.listdir(PDF_FOLDER): # if file.lower().endswith(".pdf"): # result = scan_pdf(os.path.join(PDF_FOLDER, file)) # if result: # results.append(result) # print(json.dumps(results, indent=2, ensure_ascii=False)) # import pdfplumber # import pytesseract # import re # import os # import json # import logging # from pdf2image import convert_from_path # # ---------------- CONFIG ---------------- # # PDF_FOLDER = "/var/www/html/land-ocr/input_pdfs" # request_payload = { # "JL No.": ["63", "124", "999"], # "daag No": ["261", "371", "888"] # } # OCR_LANG = "ben+eng" # TEXT_MIN_LEN = 30 # # -------------------------------------- # # logging.getLogger("pdfminer").setLevel(logging.ERROR) # REQ_JL = set(request_payload["JL No."]) # REQ_DAAG = set(request_payload["daag No"]) # FOUND_JL = set() # FOUND_DAAG = set() # def extract_text_pdf(page): # words = page.extract_words(use_text_flow=True) or [] # return " ".join(w["text"] for w in words) # def extract_text_ocr(pdf_path, page_no): # images = convert_from_path( # pdf_path, # first_page=page_no, # last_page=page_no, # dpi=300 # ) # return pytesseract.image_to_string(images[0], lang=OCR_LANG) # def search(text): # for jl in REQ_JL: # if re.search(rf"\b{jl}\b", text): # FOUND_JL.add(jl) # for daag in REQ_DAAG: # if re.search(rf"\b{daag}\b", text): # FOUND_DAAG.add(daag) # for file in os.listdir(PDF_FOLDER): # if not file.lower().endswith(".pdf"): # continue # pdf_path = os.path.join(PDF_FOLDER, file) # with pdfplumber.open(pdf_path) as pdf: # for page_no, page in enumerate(pdf.pages, start=1): # text = extract_text_pdf(page) # if not text or len(text) < TEXT_MIN_LEN: # text = extract_text_ocr(pdf_path, page_no) # search(text) # # ---------------- FINAL OUTPUT ---------------- # # output = { # "found": { # "JL_No": sorted(FOUND_JL), # "Daag_No": sorted(FOUND_DAAG) # }, # "not_found": { # "JL_No": sorted(REQ_JL - FOUND_JL), # "Daag_No": sorted(REQ_DAAG - FOUND_DAAG) # } # } # print(output) # import pdfplumber # import pytesseract # import re # import os # import requests # import tempfile # import logging # from pdf2image import convert_from_path # # ---------------- INPUT ---------------- # # pdf_urls = [ # "https://example.com/261__BRAHMANBAHARA__MAYURESWAR-2.pdf", # "https://example.com/371(Plot No)_Suahanta Mondal.pdf" # ] # request_payload = { # "JL No.": ["63", "124", "999"], # "daag No": ["261", "371", "888"] # } # # ------------------------------------- # # OCR_LANG = "ben+eng" # TEXT_MIN_LEN = 30 # logging.getLogger("pdfminer").setLevel(logging.ERROR) # REQ_JL = set(request_payload["JL No."]) # REQ_DAAG = set(request_payload["daag No"]) # FOUND_JL = set() # FOUND_DAAG = set() # def download_pdf(url, dest_folder): # local_path = os.path.join(dest_folder, os.path.basename(url)) # r = requests.get(url, timeout=30) # r.raise_for_status() # with open(local_path, "wb") as f: # f.write(r.content) # return local_path # def extract_text_pdf(page): # words = page.extract_words(use_text_flow=True) or [] # return " ".join(w["text"] for w in words) # def extract_text_ocr(pdf_path, page_no): # images = convert_from_path( # pdf_path, # first_page=page_no, # last_page=page_no, # dpi=300 # ) # return pytesseract.image_to_string(images[0], lang=OCR_LANG) # def search(text): # for jl in REQ_JL: # if re.search(rf"\b{jl}\b", text): # FOUND_JL.add(jl) # for daag in REQ_DAAG: # if re.search(rf"\b{daag}\b", text): # FOUND_DAAG.add(daag) # # ---------------- MAIN ---------------- # # with tempfile.TemporaryDirectory() as tmpdir: # for url in pdf_urls: # try: # pdf_path = download_pdf(url, tmpdir) # with pdfplumber.open(pdf_path) as pdf: # for page_no, page in enumerate(pdf.pages, start=1): # text = extract_text_pdf(page) # if not text or len(text) < TEXT_MIN_LEN: # text = extract_text_ocr(pdf_path, page_no) # search(text) # except Exception as e: # print(f"Skipped {url}: {e}") # # ---------------- OUTPUT ---------------- # # output = { # "found": { # "JL_No": sorted(FOUND_JL), # "Daag_No": sorted(FOUND_DAAG) # }, # "not_found": { # "JL_No": sorted(REQ_JL - FOUND_JL), # "Daag_No": sorted(REQ_DAAG - FOUND_DAAG) # } # } # print(output) # from fastapi import FastAPI, HTTPException,Request # from pydantic import BaseModel # import pytesseract # from pytesseract import Output # import re # import requests # import tempfile # import pdfplumber # import os # from pdf2image import convert_from_path # from PIL import Image # from fastapi.responses import JSONResponse # OCR_LANG = "ben+eng" # DPI = 400 # app = FastAPI(title="Land PDF Search API") # class SearchRequest(BaseModel): # pdf_urls: list[str] # JL_No: list[str] # Daag_No: list[str] # BN_TO_EN = str.maketrans("০১২৩৪৫৬৭৮৯", "0123456789") # def normalize_text(text: str) -> str: # text = text.translate(BN_TO_EN) # text = re.sub(r"\s+", " ", text) # return text # def extract_jl_numbers(text: str) -> set[str]: # return set(re.findall(r"\b\d{2,4}\b", text)) # # def extract_daag_numbers(text: str) -> set[str]: # # text = re.sub(r"দা\s*গ", "দাগ", text) # # matches = re.findall(r"দাগ(?:\s*নং)?\s*(\d{2,4})", text) # # return set(matches) # def extract_daag_from_pdf_text(pdf_path: str) -> set[str]: # results = set() # with pdfplumber.open(pdf_path) as pdf: # for page in pdf.pages: # text = page.extract_text() or "" # text = text.translate(BN_TO_EN) # text = re.sub(r"\s+", " ", text) # # Match rows like: 261 সৈ য়ম 2.17 Click Here # matches = re.findall(r"\b(\d{2,4})\s+[^0-9]+?\s+\d+\.\d+\s+Click", text) # for m in matches: # results.add(m) # return results # def extract_daag_numbers(text: str) -> set[str]: # """ # Extract Daag numbers from Daag ROWS like: # '261 সৈ য়ম 2.17 Click Here' # """ # results = set() # # Normalize digits & Bengali spacing # text = text.translate(BN_TO_EN) # text = re.sub(r"দা\s*গ", "দাগ", text) # # Split into logical lines # lines = text.splitlines() # for line in lines: # line = line.strip() # # Heuristic: Daag rows always contain "Click" # if "Click" in line: # match = re.search(r"\b\d{2,4}\b", line) # if match: # results.add(match.group(0)) # return results # def download_pdf(url: str, folder: str) -> str: # path = os.path.join(folder, os.path.basename(url)) # r = requests.get(url, timeout=30) # r.raise_for_status() # with open(path, "wb") as f: # f.write(r.content) # return path # @app.post("/extract") # def extract_land_data(payload: SearchRequest): # req_jl = set(payload.JL_No) # req_daag = set(payload.Daag_No) # found_jl = set() # found_daag = set() # with tempfile.TemporaryDirectory() as tmpdir: # for url in payload.pdf_urls: # try: # pdf_path = download_pdf(url, tmpdir) # images = convert_from_path(pdf_path, dpi=DPI) # for img in images: # text = pytesseract.image_to_string( # img, # lang=OCR_LANG, # config="--psm 6" # ) # text = normalize_text(text) # found_jl.update(req_jl & extract_jl_numbers(text)) # # 1️⃣ Try PDF text-layer # daag_from_text = extract_daag_from_pdf_text(pdf_path) # found_daag.update(req_daag & daag_from_text) # # 2️⃣ OCR fallback only if needed # if req_daag - found_daag: # found_daag.update(req_daag & extract_daag_numbers(text)) # except Exception: # continue # response = { # "found": { # "JL_No": sorted(found_jl), # "Daag_No": sorted(found_daag) # }, # "not_found": { # "JL_No": sorted(req_jl - found_jl), # "Daag_No": sorted(req_daag - found_daag) # } # } # if not response["not_found"]["JL_No"] and not response["not_found"]["Daag_No"]: # return response # raise HTTPException(status_code=422, detail=response) import os import re import tempfile import requests import pdfplumber from fastapi import FastAPI, HTTPException from pydantic import BaseModel from pdf2image import convert_from_path import pytesseract # ================= CONFIG ================= OCR_LANG = "ben+eng" DPI = 300 # Optimized (400 is unnecessary) app = FastAPI(title="Land PDF Search API") # ================= REQUEST MODEL ================= class SearchRequest(BaseModel): pdf_urls: list[str] JL_No: list[str] Daag_No: list[str] # ================= UTILS ================= BN_TO_EN = str.maketrans("০১২৩৪৫৬৭৮৯", "0123456789") def normalize_text(text: str) -> str: text = text.translate(BN_TO_EN) text = re.sub(r"\s+", " ", text) return text # 🔒 JL logic (UNCHANGED) def extract_jl_numbers(text: str) -> set[str]: return set(re.findall(r"\b\d{2,4}\b", text)) # ================= DAAG: PDF TEXT LAYER ================= def extract_daag_from_pdf_text(pdf_path: str) -> set[str]: """ Extract Daag numbers from PDF text-layer like: '261 সৈ য়ম 2.17 Click Here' """ results = set() with pdfplumber.open(pdf_path) as pdf: for page in pdf.pages: text = page.extract_text() or "" text = text.translate(BN_TO_EN) text = re.sub(r"\s+", " ", text) matches = re.findall( r"\b(\d{2,4})\s+[^0-9]+?\s+\d+\.\d+\s+Click", text ) results.update(matches) return results # ================= DAAG: OCR FALLBACK ================= def extract_daag_numbers(text: str) -> set[str]: """ OCR fallback for Daag rows like: '261 সৈ য়ম 2.17 Click Here' """ results = set() text = text.translate(BN_TO_EN) for line in text.splitlines(): if "Click" in line: match = re.search(r"\b\d{2,4}\b", line) if match: results.add(match.group(0)) return results # ================= FILE DOWNLOAD ================= def download_pdf(url: str, folder: str) -> str: path = os.path.join(folder, os.path.basename(url)) r = requests.get(url, timeout=30) r.raise_for_status() with open(path, "wb") as f: f.write(r.content) return path # ================= API ================= @app.post("/extract") def extract_land_data(payload: SearchRequest): req_jl = set(payload.JL_No) req_daag = set(payload.Daag_No) found_jl = set() found_daag = set() with tempfile.TemporaryDirectory() as tmpdir: for url in payload.pdf_urls: try: pdf_path = download_pdf(url, tmpdir) # 1️⃣ FAST: PDF text-layer for Daag daag_from_text = extract_daag_from_pdf_text(pdf_path) found_daag.update(req_daag & daag_from_text) need_jl_ocr = bool(req_jl - found_jl) need_daag_ocr = bool(req_daag - found_daag) # 2️⃣ OCR ONLY IF NEEDED if need_jl_ocr or need_daag_ocr: images = convert_from_path( pdf_path, dpi=DPI, first_page=1, last_page=1 ) for img in images: text = pytesseract.image_to_string( img, lang=OCR_LANG, config="--psm 6" ) text = normalize_text(text) if need_jl_ocr: found_jl.update(req_jl & extract_jl_numbers(text)) if need_daag_ocr: found_daag.update(req_daag & extract_daag_numbers(text)) # 3️⃣ EARLY EXIT (BIG SPEED GAIN) if not (req_jl - found_jl or req_daag - found_daag): break except Exception: continue response = { "found": { "JL_No": sorted(found_jl), "Daag_No": sorted(found_daag) }, "not_found": { "JL_No": sorted(req_jl - found_jl), "Daag_No": sorted(req_daag - found_daag) } } if not response["not_found"]["JL_No"] and not response["not_found"]["Daag_No"]: return response raise HTTPException(status_code=422, detail=response) # @app.post("/quick_check") # async def quick_check(request: Request): # return JSONResponse( # status_code=200, # content={ # "found": { # "JL_No": ["63", "124"], # "Daag_No": ["261", "371"] # }, # "not_found": { # "JL_No": ["999"], # "Daag_No": ["888"] # } # } # )