OCR

app.py

@@ -9,6 +9,9 @@ import psutil
 import secrets
 import hashlib
 
+import re
+import tempfile
+
 from pillow_heif import register_heif_opener
 
 from datetime import datetime, timedelta, UTC, timezone
@@ -49,6 +52,11 @@ from sqlalchemy import func, extract
 from collections import defaultdict, deque
 from functools import wraps
 
+# OCR
+from collections import Counter
+import pytesseract
+
+
 app = Flask(__name__)
 app.config.from_object(Config)
 register_heif_opener()  # pillow_heif dla HEIC
@@ -335,6 +343,89 @@ def _receipt_error(message):
     return redirect(request.referrer or url_for("main_page"))
 
 
+############# OCR ###########################
+
+
+def preprocess_image_for_tesseract(pil_image):
+    import cv2
+    import numpy as np
+    from PIL import Image
+
+    # Konwersja PIL.Image → NumPy grayscale
+    image = np.array(pil_image.convert("L"))
+
+    # Zwiększenie skali dla lepszej czytelności OCR
+    image = cv2.resize(image, None, fx=2.0, fy=2.0, interpolation=cv2.INTER_LINEAR)
+
+    # Adaptacyjne progowanie (lepsze niż THRESH_BINARY przy nierównym tle)
+    image = cv2.adaptiveThreshold(
+        image, 255, cv2.ADAPTIVE_THRESH_MEAN_C, cv2.THRESH_BINARY, blockSize=15, C=10
+    )
+
+    # Konwersja z powrotem na PIL.Image (dla pytesseract)
+    return Image.fromarray(image)
+
+
+def extract_total_tesseract(image):
+
+    text = pytesseract.image_to_string(image, lang="pol", config="--psm 6")
+    lines = text.splitlines()
+    candidates = []
+
+    fuzzy_regex = re.compile(r"[\dOo][.,:;g9zZ][\d]{2}")
+
+    for line in lines:
+        if not line.strip():
+            continue
+
+        matches = re.findall(r"\d{1,4}[.,]\d{2}", line)
+        for match in matches:
+            try:
+                val = float(match.replace(",", "."))
+                if 0.1 <= val <= 100000:
+                    candidates.append((val, line))
+            except:
+                continue
+
+        fuzzy_matches = fuzzy_regex.findall(line)
+        for match in fuzzy_matches:
+            cleaned = (
+                match.replace("O", "0")
+                .replace("o", "0")
+                .replace(":", ".")
+                .replace(";", ".")
+                .replace(",", ".")
+                .replace("g", "9")
+                .replace("z", "9")
+                .replace("Z", "9")
+            )
+            try:
+                val = float(cleaned)
+                if 0.1 <= val <= 100:
+                    candidates.append((val, line))
+            except:
+                continue
+
+    preferred = [
+        val
+        for val, line in candidates
+        if re.search(r"sum[aąo]?|razem|zapłaty", line.lower())
+    ]
+
+    if preferred:
+        max_val = round(max(preferred), 2)
+        return max_val, lines
+
+    if candidates:
+        max_val = round(max([val for val, _ in candidates]), 2)
+        return max_val, lines
+
+    return 0.0, lines
+
+
+############# END OCR #######################
+
+
 # zabezpieczenie logowani do systemu - błędne hasła
 def is_ip_blocked(ip):
     now = time.time()
@@ -1037,6 +1128,59 @@ def reorder_items():
     return jsonify(success=True)
 
 
+# OCR
+@app.route("/lists/<int:list_id>/analyze", methods=["POST"])
+@login_required
+def analyze_receipts_for_list(list_id):
+    list_obj = db.session.get(ShoppingList, list_id)
+    if not list_obj or list_obj.owner_id != current_user.id:
+        return jsonify({"error": "Brak dostępu"}), 403
+
+    receipt_objs = Receipt.query.filter_by(list_id=list_id).all()
+    results = []
+    total = 0.0
+
+    for receipt in receipt_objs:
+        filepath = os.path.join(app.config["UPLOAD_FOLDER"], receipt.filename)
+        if not os.path.exists(filepath):
+            continue
+
+        temp_path = None
+
+        try:
+            if filepath.lower().endswith(".webp"):
+
+                raw_image = Image.open(filepath).convert("RGB")
+                image = preprocess_image_for_tesseract(raw_image)
+            else:
+
+                raw_image = Image.open(filepath).convert("RGB")
+                image = preprocess_image_for_tesseract(raw_image)
+
+            value, lines = extract_total_tesseract(image)
+
+        except Exception as e:
+            print(f"OCR error for {receipt.filename}: {e}")
+            value = 0.0
+            lines = []
+
+        finally:
+            if temp_path and os.path.exists(temp_path):
+                os.unlink(temp_path)
+
+        results.append(
+            {
+                "id": receipt.id,
+                "filename": receipt.filename,
+                "amount": round(value, 2),
+                "debug_text": lines,
+            }
+        )
+        total += value
+
+    return jsonify({"results": results, "total": round(total, 2)})
+
+
 @app.route("/admin")
 @login_required
 @admin_required