| """ |
| Data Loader for Kalim Chatbot |
| Utilities for loading and accessing the processed knowledge base |
| """ |
| import json |
| from typing import Dict, List, Optional, Any |
| from functools import lru_cache |
|
|
| from config.settings import ( |
| MAJORS_JSON_FILE, |
| LOCATIONS_JSON_FILE, |
| HOLLAND_MAPPING_FILE, |
| HOLLAND_CODES, |
| SECONDARY_TRACKS, |
| GOVERNORATES |
| ) |
|
|
|
|
| @lru_cache(maxsize=1) |
| def load_majors() -> List[Dict[str, Any]]: |
| """Load majors data from JSON file.""" |
| if not MAJORS_JSON_FILE.exists(): |
| raise FileNotFoundError( |
| f"Majors file not found: {MAJORS_JSON_FILE}\n" |
| "Run 'python -m src.data.processor' first to generate the data." |
| ) |
|
|
| with open(MAJORS_JSON_FILE, 'r', encoding='utf-8') as f: |
| data = json.load(f) |
|
|
| return data.get('majors', []) |
|
|
| @lru_cache(maxsize=1) |
| def load_holland_mapping() -> Dict[str, Any]: |
| """Load Holland RIASEC mapping from JSON file.""" |
| if not HOLLAND_MAPPING_FILE.exists(): |
| |
| return { |
| "codes": HOLLAND_CODES, |
| "code_list": ["R", "I", "A", "S", "E", "C"], |
| "dropdown_options": [ |
| {"code": "R", "label": "R - الواقعي (Realistic)"}, |
| {"code": "I", "label": "I - الباحث (Investigative)"}, |
| {"code": "A", "label": "A - الفني (Artistic)"}, |
| {"code": "S", "label": "S - الاجتماعي (Social)"}, |
| {"code": "E", "label": "E - المغامر (Enterprising)"}, |
| {"code": "C", "label": "C - التقليدي (Conventional)"}, |
| ] |
| } |
|
|
| with open(HOLLAND_MAPPING_FILE, 'r', encoding='utf-8') as f: |
| return json.load(f) |
|
|
|
|
| def get_major_by_id(major_id: int) -> Optional[Dict[str, Any]]: |
| """Get a specific major by ID.""" |
| majors = load_majors() |
| for major in majors: |
| if major['id'] == major_id: |
| return major |
| return None |
|
|
| def get_governorates() -> List[str]: |
| """Get list of all governorates.""" |
| return list(GOVERNORATES.keys()) |
|
|
|
|
| def get_districts(governorate: str) -> List[str]: |
| """Get districts for a specific governorate.""" |
| return GOVERNORATES.get(governorate, []) |
|
|
|
|
| @lru_cache(maxsize=1) |
| def get_dataset_governorates() -> List[str]: |
| """Governorates that actually occur in majors.json, in GOVERNORATES order. |
| |
| The UI must offer only these. `GOVERNORATES` follows Lebanon's newer |
| eight-governorate division, in which عكار and بعلبك-الهرمل are governorates |
| of their own; the dataset follows the older six, filing حلبا under |
| الشمال/عكار and دورس under البقاع/بعلبك. Offering the config's list let a |
| student from Akkar finish the whole wizard and be told no major matched — |
| a dead end that reads as "the university has nothing for you" when the |
| branch exists one level down. |
| """ |
| present = { |
| loc.get('governorate') |
| for major in load_majors() |
| for loc in major.get('locations', []) |
| if loc.get('governorate') |
| } |
| return [g for g in GOVERNORATES if g != "الكل" and g in present] |
|
|
|
|
| @lru_cache(maxsize=1) |
| def _dataset_districts_by_gov() -> Dict[str, List[str]]: |
| by_gov: Dict[str, set] = {} |
| for major in load_majors(): |
| for loc in major.get('locations', []): |
| gov, district = loc.get('governorate'), loc.get('district') |
| if gov and district: |
| by_gov.setdefault(gov, set()).add(district) |
| return {gov: sorted(names) for gov, names in by_gov.items()} |
|
|
|
|
| def get_dataset_districts(governorate: str) -> List[str]: |
| """Districts of this governorate present in majors.json. |
| |
| Kept in the GOVERNORATES reference order where possible, with anything the |
| config map does not list appended — that is how عكار (a district of الشمال |
| in the data, a governorate in the config) still reaches the dropdown. |
| """ |
| present = set(_dataset_districts_by_gov().get(governorate, [])) |
| ordered = [d for d in GOVERNORATES.get(governorate, []) if d in present] |
| return ordered + sorted(present - set(ordered)) |
|
|
| |
| __all__ = [ |
| 'load_majors', |
| 'load_holland_mapping', |
| 'get_major_by_id', |
| 'get_governorates', |
| 'get_districts', |
| 'get_dataset_governorates', |
| 'get_dataset_districts', |
| ] |
|
|
