| from difflib import SequenceMatcher |
| from typing import Optional |
|
|
| CATEGORY_KEYWORDS: dict[str, list[str]] = { |
| "Damaged Road Issues": [ |
| "road", "damage", "damaged", "broken", "crack", "cracked", "pavement", |
| "asphalt", "street", "highway", "lane", "surface", "rough", "uneven" |
| ], |
| "Pothole Issues": [ |
| "pothole", "hole", "pit", "crater", "dip", "depression", "bump", |
| "cavity", "hollow", "gap" |
| ], |
| "Illegal Parking Issues": [ |
| "parking", "parked", "car", "vehicle", "illegal", "wrong", "blocking", |
| "obstruction", "no parking", "double park", "sidewalk" |
| ], |
| "Broken Road Sign Issues": [ |
| "sign", "signboard", "traffic sign", "road sign", "broken sign", |
| "fallen sign", "damaged sign", "missing sign", "bent" |
| ], |
| "Fallen Trees": [ |
| "tree", "fallen", "branch", "trunk", "uprooted", "collapsed", |
| "blocking", "storm", "wind", "timber" |
| ], |
| "Littering/Garbage on Public Places": [ |
| "garbage", "trash", "litter", "waste", "rubbish", "dump", "dirty", |
| "filth", "debris", "plastic", "pile", "mess", "junk", "disposal" |
| ], |
| "Vandalism Issues": [ |
| "vandal", "graffiti", "spray", "paint", "defaced", "broken", |
| "smashed", "destroyed", "damaged property", "torn" |
| ], |
| "Dead Animal Pollution": [ |
| "dead", "animal", "carcass", "body", "corpse", "rotting", "smell", |
| "stink", "dog", "cat", "bird", "cow", "roadkill" |
| ], |
| "Damaged Concrete Structures": [ |
| "concrete", "structure", "wall", "pillar", "bridge", "flyover", |
| "footpath", "sidewalk", "curb", "crack", "broken" |
| ], |
| "Damaged Electric Wires and Poles": [ |
| "electric", "wire", "pole", "cable", "power", "electricity", |
| "hanging", "exposed", "sparking", "transformer", "light pole" |
| ], |
| } |
|
|
|
|
| def normalize_text(text: str) -> str: |
| return text.lower().strip() |
|
|
|
|
| def calculate_similarity(s1: str, s2: str) -> float: |
| return SequenceMatcher(None, s1.lower(), s2.lower()).ratio() |
|
|
|
|
| def fuzzy_match_word(word: str, keywords: list[str], threshold: float = 0.7) -> bool: |
| word = normalize_text(word) |
| for keyword in keywords: |
| if word == keyword: |
| return True |
| if len(word) >= 4 and calculate_similarity(word, keyword) >= threshold: |
| return True |
| if keyword in word or word in keyword: |
| return True |
| return False |
|
|
|
|
| def match_description_to_category( |
| description: Optional[str], |
| detected_category: str, |
| threshold: float = 0.6 |
| ) -> tuple[bool, float, list[str]]: |
| if not description: |
| return False, 0.0, [] |
| |
| keywords = CATEGORY_KEYWORDS.get(detected_category, []) |
| if not keywords: |
| return False, 0.0, [] |
| |
| words = normalize_text(description).replace(",", " ").replace(".", " ").split() |
| |
| matched_words = [] |
| for word in words: |
| if len(word) < 3: |
| continue |
| if fuzzy_match_word(word, keywords): |
| matched_words.append(word) |
| |
| if not words: |
| return False, 0.0, [] |
| |
| match_score = len(matched_words) / max(len(words), 1) |
| is_match = len(matched_words) >= 1 or match_score >= threshold |
| |
| return is_match, match_score, matched_words |
|
|
|
|
| def auto_validate_issue( |
| description: Optional[str], |
| detected_categories: list[str], |
| confidence_threshold: float = 0.5 |
| ) -> tuple[bool, str]: |
| if not description or not detected_categories: |
| return False, "No description or no detections for auto-validation" |
| |
| for category in detected_categories: |
| is_match, score, matched_words = match_description_to_category( |
| description, category |
| ) |
| if is_match: |
| return True, f"Auto-validated: '{category}' matched with keywords: {matched_words}" |
| |
| return False, f"Manual verification required: no match between description and detected categories {detected_categories}" |
|
|
