Update app.py

app.py

@@ -21,6 +21,7 @@ reader = easyocr.Reader(['en'])
 
 class ImageRequest(BaseModel):
     image_url: str
+    document_type: str
 
 # =========================
 # DOWNLOAD IMAGE
@@ -73,7 +74,6 @@ def is_dark(image):
 
     return brightness < 50
 
-
 # =========================
 # OCR TEXT EXTRACTION
 # =========================
@@ -88,16 +88,14 @@ def extract_text(image_path):
 
     return text
 
-
 # =========================
-# DOCUMENT DETECTION
+# DOCUMENT VALIDATION
 # =========================
 
-def detect_document(text):
-
-    # =========================
-    # CLEAN TEXT
-    # =========================
+def validate_document_type(
+    text,
+    document_type
+):
 
     text = text.lower().strip()
 
@@ -107,411 +105,349 @@ def detect_document(text):
         text
     )
 
-    words = cleaned_text.split()
-
-    # =========================
-    # SCORE SYSTEM
-    # =========================
-
-    scores = {
-        "nin": 0,
-        "passport": 0,
-        "drivers_license": 0,
-        "voters_card": 0,
-        "utility_bill": 0,
-        "bank_statement": 0,
-        "tenancy_agreement": 0,
-        "vehicle_plate": 0,
-        "vehicle_image": 0
-    }
+    matched_keywords = []
 
-    matched_keywords = {
-        "nin": [],
-        "passport": [],
-        "drivers_license": [],
-        "voters_card": [],
-        "utility_bill": [],
-        "bank_statement": [],
-        "tenancy_agreement": [],
-        "vehicle_plate": [],
-        "vehicle_image": []
-    }
+    confidence = 0
 
     # =========================
-    # GARBAGE OCR DETECTION
+    # NATIONAL ID (NIN)
     # =========================
 
-    garbage_patterns = [
-        r'^[a-z0-9]{4,8}$'
-    ]
+    if document_type == "National ID (NIN)":
+
+        keywords = [
+            "national identification number",
+            "national identity",
+            "nimc",
+            "nin"
+        ]
 
-    garbage_count = 0
+        for keyword in keywords:
 
-    for pattern in garbage_patterns:
+            if keyword in cleaned_text:
 
-        for word in words:
+                matched_keywords.append(keyword)
 
-            if re.match(pattern, word):
-                garbage_count += 1
+                confidence += 25
 
     # =========================
-    # NIN
+    # INTERNATIONAL PASSPORT
     # =========================
 
-    nin_keywords = [
-        "national identification number",
-        "national identity",
-        "nimc"
-    ]
+    elif document_type == "International Passport":
 
-    for keyword in nin_keywords:
+        keywords = [
+            "passport",
+            "federal republic of nigeria",
+            "nigeria passport",
+            "international passport"
+        ]
 
-        if keyword in cleaned_text:
+        for keyword in keywords:
 
-            scores["nin"] += 5
-            matched_keywords["nin"].append(keyword)
+            if keyword in cleaned_text:
 
-    # Weak standalone nin
-    if " nin " in f" {cleaned_text} ":
+                matched_keywords.append(keyword)
 
-        scores["nin"] += 1
-        matched_keywords["nin"].append("nin")
+                confidence += 25
 
     # =========================
-    # PASSPORT
+    # DRIVER LICENSE
     # =========================
 
-    passport_keywords = [
-        "passport",
-        "federal republic of nigeria",
-        "nigeria passport",
-        "international passport"
-    ]
-
-    for keyword in passport_keywords:
+    elif document_type == "Driver License":
 
-        if keyword in cleaned_text:
-
-            scores["passport"] += 5
-            matched_keywords["passport"].append(keyword)
-
-    # =========================
-    # DRIVER LICENSE
-    # =========================
+        keywords = [
+            "driver",
+            "license",
+            "drivers licence",
+            "driver licence",
+            "frsc"
+        ]
 
-    license_keywords = [
-        "driver",
-        "license",
-        "drivers licence",
-        "driver licence",
-        "frsc"
-    ]
+        for keyword in keywords:
 
-    for keyword in license_keywords:
+            if keyword in cleaned_text:
 
-        if keyword in cleaned_text:
+                matched_keywords.append(keyword)
 
-            scores["drivers_license"] += 3
-            matched_keywords["drivers_license"].append(keyword)
+                confidence += 20
 
     # =========================
     # VOTER CARD
     # =========================
 
-    voter_keywords = [
-        "voter",
-        "inec",
-        "permanent voter",
-        "polling unit"
-    ]
+    elif document_type == "Voter Card":
+
+        keywords = [
+            "voter",
+            "inec",
+            "permanent voter",
+            "polling unit"
+        ]
+
+        for keyword in keywords:
 
-    for keyword in voter_keywords:
+            if keyword in cleaned_text:
 
-        if keyword in cleaned_text:
+                matched_keywords.append(keyword)
 
-            scores["voters_card"] += 4
-            matched_keywords["voters_card"].append(keyword)
+                confidence += 25
 
     # =========================
     # UTILITY BILL
     # =========================
 
-    electricity_keywords = [
+    elif document_type == "Utility Bill":
+
+        keywords = [
 
-        # General
-        "electricity",
-        "electric bill",
-        "power bill",
-        "meter number",
-        "meter no",
-        "token",
-        "kwh",
-        "prepaid",
-        "postpaid",
-        "energy charge",
-        "tariff",
+            # General
+            "electricity",
+            "electric bill",
+            "power bill",
+            "meter number",
+            "meter no",
+            "token",
+            "kwh",
+            "prepaid",
+            "postpaid",
+            "energy charge",
+            "tariff",
 
-        # Nigerian DISCOs
-        "ibedc",
-        "ibadan electricity",
+            # Nigerian DISCOs
+            "ibedc",
+            "ibadan electricity",
 
-        "ikedc",
-        "ikeja electric",
+            "ikedc",
+            "ikeja electric",
 
-        "ekedc",
-        "eko electric",
+            "ekedc",
+            "eko electric",
 
-        "aedc",
-        "abuja electricity",
+            "aedc",
+            "abuja electricity",
 
-        "eedc",
-        "enugu electricity",
+            "eedc",
+            "enugu electricity",
 
-        "bedc",
-        "benin electricity",
+            "bedc",
+            "benin electricity",
 
-        "jed",
-        "jos electricity",
+            "jed",
+            "jos electricity",
 
-        "kedco",
-        "kano electricity",
+            "kedco",
+            "kano electricity",
 
-        "kaedco",
-        "kaduna electric",
+            "kaedco",
+            "kaduna electric",
 
-        "phed",
-        "port harcourt electricity",
+            "phed",
+            "port harcourt electricity",
 
-        "yedc",
-        "yola electricity"
-    ]
+            "yedc",
+            "yola electricity"
+        ]
 
-    for keyword in electricity_keywords:
+        for keyword in keywords:
 
-        if keyword in cleaned_text:
+            if keyword in cleaned_text:
 
-            scores["utility_bill"] += 4
-            matched_keywords["utility_bill"].append(keyword)
+                matched_keywords.append(keyword)
+
+                confidence += 15
 
     # =========================
     # BANK STATEMENT
     # =========================
 
-    bank_keywords = [
-
-        "account statement",
-        "statement of account",
-        "transaction",
-        "balance",
-        "account number",
-        "credit",
-        "debit",
-        "withdrawal",
-        "deposit",
-
-        # Nigerian Banks
-        "access bank",
-        "gtbank",
-        "uba",
-        "zenith bank",
-        "first bank",
-        "opay",
-        "moniepoint",
-        "kuda",
-        "fcmb",
-        "sterling bank",
-        "wema bank",
-        "providus",
-        "fidelity bank",
-        "union bank"
-    ]
-
-    for keyword in bank_keywords:
-
-        if keyword in cleaned_text:
-
-            scores["bank_statement"] += 3
-            matched_keywords["bank_statement"].append(keyword)
+    elif document_type == "Bank Statement":
+
+        keywords = [
+
+            "account statement",
+            "statement of account",
+            "transaction",
+            "balance",
+            "account number",
+            "credit",
+            "debit",
+            "withdrawal",
+            "deposit",
+
+            # Nigerian Banks
+            "access bank",
+            "gtbank",
+            "uba",
+            "zenith bank",
+            "first bank",
+            "opay",
+            "moniepoint",
+            "kuda",
+            "fcmb",
+            "sterling bank",
+            "wema bank",
+            "providus",
+            "fidelity bank",
+            "union bank"
+        ]
+
+        for keyword in keywords:
+
+            if keyword in cleaned_text:
+
+                matched_keywords.append(keyword)
+
+                confidence += 15
 
     # =========================
     # TENANCY AGREEMENT
     # =========================
 
-    tenancy_keywords = [
-        "tenancy agreement",
-        "landlord",
-        "tenant",
-        "rent",
-        "property",
-        "lease agreement",
-        "rental agreement"
-    ]
+    elif document_type == "Tenancy Agreement":
 
-    for keyword in tenancy_keywords:
+        keywords = [
+            "tenancy agreement",
+            "landlord",
+            "tenant",
+            "rent",
+            "property",
+            "lease agreement",
+            "rental agreement"
+        ]
 
-        if keyword in cleaned_text:
+        for keyword in keywords:
 
-            scores["tenancy_agreement"] += 3
-            matched_keywords["tenancy_agreement"].append(keyword)
+            if keyword in cleaned_text:
 
-    # =========================
-    # VEHICLE KEYWORDS
-    # =========================
+                matched_keywords.append(keyword)
 
-    vehicle_keywords = [
-        "toyota",
-        "honda",
-        "lexus",
-        "benz",
-        "mercedes",
-        "ford",
-        "jeep",
-        "hyundai",
-        "kia",
-        "nissan",
-        "camry",
-        "corolla",
-        "rav4",
-        "pilot",
-        "highlander",
-        "vehicle",
-        "plate number"
-    ]
-
-    for keyword in vehicle_keywords:
-
-        if keyword in cleaned_text:
-
-            scores["vehicle_image"] += 3
-            matched_keywords["vehicle_image"].append(keyword)
+                confidence += 20
 
     # =========================
-    # NIGERIAN STATES
+    # VEHICLE WITH PLATE NUMBER
     # =========================
 
-    nigeria_states = [
-        "lagos",
-        "abuja",
-        "kano",
-        "kaduna",
-        "oyo",
-        "ogun",
-        "ondo",
-        "osun",
-        "kwara",
-        "imo",
-        "anambra",
-        "enugu",
-        "rivers",
-        "delta",
-        "edo",
-        "cross river",
-        "akwa ibom",
-        "bayelsa",
-        "plateau",
-        "benue",
-        "kogi",
-        "ekiti",
-        "niger",
-        "zamfara",
-        "sokoto",
-        "katsina",
-        "borno",
-        "yobe",
-        "adamawa",
-        "taraba",
-        "gombe",
-        "bauchi",
-        "jigawa",
-        "nasarawa",
-        "kebbi",
-        "ebonyi"
-    ]
-
-    for state in nigeria_states:
-
-        if state in cleaned_text:
-
-            scores["vehicle_plate"] += 1
-            matched_keywords["vehicle_plate"].append(state)
+    elif document_type == "Vehicle with Plate Number":
 
-    # =========================
-    # NIGERIAN PLATE PATTERNS
-    # =========================
+        vehicle_keywords = [
 
-    plate_patterns = [
-        r"[A-Z]{3}-?\d{3}[A-Z]{2}",
-        r"[A-Z]{2}\d{3}[A-Z]{3}",
-        r"[A-Z]{3}\s\d{3}\s[A-Z]{2}"
-    ]
+            "toyota",
+            "honda",
+            "lexus",
+            "benz",
+            "mercedes",
+            "ford",
+            "jeep",
+            "hyundai",
+            "kia",
+            "nissan",
+            "camry",
+            "corolla",
+            "rav4",
+            "pilot",
+            "highlander",
+            "vehicle",
+            "plate number"
+        ]
 
-    for pattern in plate_patterns:
+        for keyword in vehicle_keywords:
+
+            if keyword in cleaned_text:
+
+                matched_keywords.append(keyword)
+
+                confidence += 10
+
+        # Nigerian states
+
+        nigeria_states = [
+            "lagos",
+            "abuja",
+            "kano",
+            "kaduna",
+            "oyo",
+            "ogun",
+            "ondo",
+            "osun",
+            "kwara",
+            "imo",
+            "anambra",
+            "enugu",
+            "rivers",
+            "delta",
+            "edo",
+            "cross river",
+            "akwa ibom",
+            "bayelsa",
+            "plateau",
+            "benue",
+            "kogi",
+            "ekiti",
+            "niger",
+            "zamfara",
+            "sokoto",
+            "katsina",
+            "borno",
+            "yobe",
+            "adamawa",
+            "taraba",
+            "gombe",
+            "bauchi",
+            "jigawa",
+            "nasarawa",
+            "kebbi",
+            "ebonyi"
+        ]
 
-        plate_match = re.search(
-            pattern,
-            cleaned_text.upper()
-        )
+        for state in nigeria_states:
 
-        if plate_match:
+            if state in cleaned_text:
 
-            scores["vehicle_plate"] += 10
+                matched_keywords.append(state)
 
-            matched_keywords[
-                "vehicle_plate"
-            ].append(
-                plate_match.group()
-            )
+                confidence += 5
 
-    # =========================
-    # OCR GARBAGE PENALTY
-    # =========================
+        # Plate patterns
 
-    if garbage_count >= 5:
+        plate_patterns = [
+            r"[A-Z]{3}-?\d{3}[A-Z]{2}",
+            r"[A-Z]{2}\d{3}[A-Z]{3}",
+            r"[A-Z]{3}\s\d{3}\s[A-Z]{2}"
+        ]
 
-        for key in scores:
-            scores[key] -= 2
+        for pattern in plate_patterns:
 
-    # =========================
-    # BEST MATCH
-    # =========================
+            plate_match = re.search(
+                pattern,
+                cleaned_text.upper()
+            )
 
-    best_doc = max(
-        scores,
-        key=scores.get
-    )
+            if plate_match:
 
-    best_score = scores[best_doc]
+                matched_keywords.append(
+                    plate_match.group()
+                )
+
+                confidence += 50
 
     # =========================
     # LOW CONFIDENCE
     # =========================
 
-    if best_score <= 0:
-        return None
-
-    # =========================
-    # CONFIDENCE
-    # =========================
+    if confidence <= 0:
 
-    confidence = min(
-        99,
-        max(50, best_score * 5)
-    )
+        return None
 
-    # =========================
-    # RETURN RESULT
-    # =========================
+    confidence = min(confidence, 99)
 
     return {
-        "document_type": best_doc,
+        "document_type": document_type,
         "confidence": confidence,
-        "matched_keywords": (
-            matched_keywords[best_doc]
-        ),
-        "all_scores": scores
+        "matched_keywords": matched_keywords
     }
 
-
 # =========================
 # HOME ROUTE
 # =========================
@@ -534,7 +470,6 @@ def home():
         ]
     }
 
-
 # =========================
 # VALIDATION ENDPOINT
 # =========================
@@ -546,6 +481,32 @@ async def validate_document(
 
     try:
 
+        # =========================
+        # VALID DOCUMENT TYPES
+        # =========================
+
+        valid_document_types = [
+
+            "National ID (NIN)",
+            "International Passport",
+            "Driver License",
+            "Voter Card",
+            "Vehicle with Plate Number",
+            "Utility Bill",
+            "Bank Statement",
+            "Tenancy Agreement"
+        ]
+
+        if request.document_type not in valid_document_types:
+
+            return {
+                "success": False,
+                "message": "Invalid document type",
+                "supported_document_types": (
+                    valid_document_types
+                )
+            }
+
         # =========================
         # DOWNLOAD IMAGE
         # =========================
@@ -577,8 +538,9 @@ async def validate_document(
                 "success": False,
                 "message": "Invalid image",
                 "reason": (
-                    "The downloaded file could "
-                    "not be read as an image."
+                    "The downloaded file "
+                    "could not be read "
+                    "as an image."
                 ),
                 "suggestion": (
                     "Ensure the URL points "
@@ -596,7 +558,8 @@ async def validate_document(
                 "success": False,
                 "message": "Image rejected",
                 "reason": (
-                    "The uploaded image is blurry."
+                    "The uploaded image "
+                    "is blurry."
                 ),
                 "suggestion": (
                     "Retake the photo "
@@ -605,7 +568,7 @@ async def validate_document(
             }
 
         # =========================
-        # DARK IMAGE CHECK
+        # DARK CHECK
         # =========================
 
         if is_dark(image):
@@ -614,7 +577,8 @@ async def validate_document(
                 "success": False,
                 "message": "Image rejected",
                 "reason": (
-                    "The uploaded image is too dark."
+                    "The uploaded image "
+                    "is too dark."
                 ),
                 "suggestion": (
                     "Take the photo in a "
@@ -623,7 +587,7 @@ async def validate_document(
             }
 
         # =========================
-        # OCR TEXT EXTRACTION
+        # OCR EXTRACTION
         # =========================
 
         text = extract_text(image_path)
@@ -638,8 +602,9 @@ async def validate_document(
                 "success": False,
                 "message": "Document rejected",
                 "reason": (
-                    "No readable text was "
-                    "detected in the image."
+                    "No readable text "
+                    "was detected "
+                    "in the image."
                 ),
                 "suggestion": (
                     "Ensure the document "
@@ -648,15 +613,16 @@ async def validate_document(
             }
 
         # =========================
-        # DOCUMENT DETECTION
+        # VALIDATE DOCUMENT
         # =========================
 
-        document_result = detect_document(
-            text
+        document_result = validate_document_type(
+            text,
+            request.document_type
         )
 
         # =========================
-        # UNSUPPORTED DOCUMENT
+        # DOCUMENT FAILED
         # =========================
 
         if document_result is None:
@@ -665,30 +631,21 @@ async def validate_document(
                 "success": False,
                 "message": "Document rejected",
                 "reason": (
-                    "The uploaded image "
-                    "does not match any "
-                    "supported document type."
+                    f"The uploaded image "
+                    f"does not match "
+                    f"the expected "
+                    f"document type: "
+                    f"{request.document_type}"
                 ),
-                "supported_documents": [
-                    "National ID (NIN)",
-                    "International Passport",
-                    "Driver License",
-                    "Voter Card",
-                    "Vehicle with Plate Number",
-                    "Utility Bill",
-                    "Bank Statement",
-                    "Tenancy Agreement"
-                ],
+                "ocr_preview": text[:300],
                 "possible_issues": [
+                    "Wrong document uploaded",
+                    "Image is blurry",
                     "Image is cropped",
-                    "Text is unreadable",
-                    "Unsupported document uploaded",
                     "Poor lighting",
-                    "Low image quality",
-                    "Document too far from camera",
+                    "Text not readable",
                     "Document partially hidden"
-                ],
-                "ocr_preview": text[:300]
+                ]
             }
 
         # =========================
@@ -697,7 +654,9 @@ async def validate_document(
 
         return {
             "success": True,
-            "message": "Document verified successfully",
+            "message": (
+                "Document verified successfully"
+            ),
             "document_type": (
                 document_result["document_type"]
             ),
@@ -707,9 +666,6 @@ async def validate_document(
             "matched_keywords": (
                 document_result["matched_keywords"]
             ),
-            "score_breakdown": (
-                document_result["all_scores"]
-            ),
             "ocr_preview": text[:300]
         }
 
@@ -728,4 +684,5 @@ async def validate_document(
         # =========================
 
         if os.path.exists("temp.jpg"):
+
             os.remove("temp.jpg")