config.yaml

# =============================================================================
# UnlockEgypt Parser Configuration
# =============================================================================
# This file contains all configurable parameters for the parser.
# Modify values here instead of changing the source code.
# =============================================================================

# Application metadata
app:
  name: "UnlockEgypt Site Researcher"
  version: "3.4.0"
  author: "UnlockEgypt Team"

# Target website configuration
website:
  base_url: "https://egymonuments.gov.eg"
  allowed_domain: "egymonuments.gov.eg"
  page_types:
    - archaeological-sites
    - monuments
    - museums
    - sunken-monuments

# Browser/Selenium settings
browser:
  headless: true
  window_width: 1920
  window_height: 1080
  user_agent: "Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36"

# Timing configuration (in seconds)
timing:
  implicit_wait_timeout: 10
  page_load_wait: 5
  click_wait: 2
  scroll_wait: 2
  arabic_page_wait: 2
  show_more_wait: 3
  geocoding_rate_limit: 1
  http_timeout: 15

# Retry configuration
retry:
  max_retries: 3
  initial_delay: 1.0
  backoff_multiplier: 2.0

# Content extraction thresholds
content:
  min_paragraph_length: 40
  min_sentence_length: 30
  max_sub_locations: 5
  summary_char_limit: 200
  max_load_iterations: 20

# Geographic bounds for Egypt (for coordinate validation)
geography:
  egypt_lat_min: 21
  egypt_lat_max: 32
  egypt_lon_min: 24
  egypt_lon_max: 37

# Geocoding service configuration
geocoding:
  nominatim_url: "https://nominatim.openstreetmap.org/search"
  user_agent: "UnlockEgyptParser/3.4 (educational project)"

# Output configuration
output:
  default_filename: "parsed_sites.json"
  indent: 2
  ensure_ascii: false

# Historical era detection keywords
era_keywords:
  Pre-Dynastic:
    - predynastic
    - pre-dynastic
  Old Kingdom:
    - old kingdom
    - early dynastic
  Middle Kingdom:
    - middle kingdom
  New Kingdom:
    - new kingdom
    - 18th dynasty
    - 19th dynasty
    - 20th dynasty
  Late Period:
    - late period
  Ptolemaic:
    - ptolemaic
    - ptolemy
    - greek
    - hellenistic
  Roman:
    - roman
    - byzantine
    - coptic
  Islamic:
    - islamic
    - fatimid
    - mamluk
    - ayyubid
    - ottoman
  Modern:
    - muhammad ali
    - modern
    - 19th century
    - 20th century

# Valid classification values
classifications:
  eras:
    - Pre-Dynastic
    - Old Kingdom
    - Middle Kingdom
    - New Kingdom
    - Late Period
    - Ptolemaic
    - Roman
    - Islamic
    - Modern
  tourism_types:
    - Pharaonic
    - Greco-Roman
    - Coptic
    - Islamic
    - Modern
  place_types:
    - Pyramid
    - Temple
    - Tomb
    - Museum
    - Mosque
    - Church
    - Fortress
    - Market
    - Monument
    - Ruins

# Place type detection keywords
place_keywords:
  Pyramid:
    - pyramid
  Temple:
    - temple
  Tomb:
    - tomb
    - cemetery
    - necropolis
    - burial
  Museum:
    - museum
  Mosque:
    - mosque
    - madrasa
  Church:
    - church
    - cathedral
    - monastery
    - convent
  Fortress:
    - fortress
    - citadel
    - fort
    - castle
  Market:
    - market
    - bazaar
    - khan
  Monument:
    - amphitheatre
    - theater
    - theatre
    - obelisk
    - statue
    - colossus

# City-specific visitor tips
city_tips:
  Alexandria: "Bring sunscreen and a hat - the Mediterranean sun can be intense."
  Cairo: "Start early to avoid crowds and the midday heat."
  Giza: "Hire an official guide at the entrance for the best experience."
  Luxor: "Consider visiting early morning or late afternoon to avoid peak heat."
  Aswan: "Carry plenty of water, especially during summer months."

# Arabic vocabulary phrases by place type
arabic_phrases:
  Temple:
    - english: "Temple"
      arabic: "معبد"
      pronunciation: "Ma'bad"
    - english: "Pharaoh"
      arabic: "فرعون"
      pronunciation: "Fir'awn"
    - english: "God/Goddess"
      arabic: "إله/إلهة"
      pronunciation: "Ilah/Ilaha"
    - english: "Columns"
      arabic: "أعمدة"
      pronunciation: "A'mida"
    - english: "Hieroglyphics"
      arabic: "هيروغليفية"
      pronunciation: "Hieroglyphia"
  Pyramid:
    - english: "Pyramid"
      arabic: "هرم"
      pronunciation: "Haram"
    - english: "King"
      arabic: "ملك"
      pronunciation: "Malik"
    - english: "Chamber"
      arabic: "غرفة"
      pronunciation: "Ghorfa"
    - english: "Stone"
      arabic: "حجر"
      pronunciation: "Hagar"
  Tomb:
    - english: "Tomb"
      arabic: "مقبرة"
      pronunciation: "Maqbara"
    - english: "Mummy"
      arabic: "مومياء"
      pronunciation: "Mumya"
    - english: "Coffin"
      arabic: "تابوت"
      pronunciation: "Taboot"
    - english: "Noble"
      arabic: "نبيل"
      pronunciation: "Nabeel"
    - english: "Cemetery"
      arabic: "جبانة"
      pronunciation: "Gabbana"
  Mosque:
    - english: "Mosque"
      arabic: "مسجد"
      pronunciation: "Masgid"
    - english: "Minaret"
      arabic: "مئذنة"
      pronunciation: "Me'zana"
    - english: "Dome"
      arabic: "قبة"
      pronunciation: "Qobba"
    - english: "Prayer"
      arabic: "صلاة"
      pronunciation: "Salah"
  Museum:
    - english: "Museum"
      arabic: "متحف"
      pronunciation: "Mat-haf"
    - english: "Artifact"
      arabic: "أثر"
      pronunciation: "Athar"
    - english: "Statue"
      arabic: "تمثال"
      pronunciation: "Timthal"
    - english: "Collection"
      arabic: "مجموعة"
      pronunciation: "Magmoo'a"
    - english: "Display"
      arabic: "عرض"
      pronunciation: "Ard"
  Monument:
    - english: "Monument"
      arabic: "نصب تذكاري"
      pronunciation: "Nasb Tizkari"
    - english: "Amphitheatre"
      arabic: "مدرج"
      pronunciation: "Mudarrag"
    - english: "Theater"
      arabic: "مسرح"
      pronunciation: "Masrah"
    - english: "Roman"
      arabic: "روماني"
      pronunciation: "Romani"
    - english: "Mosaic"
      arabic: "فسيفساء"
      pronunciation: "Fusayfisa'"
  Ruins:
    - english: "Ruins"
      arabic: "آثار"
      pronunciation: "Athar"
    - english: "Ancient"
      arabic: "قديم"
      pronunciation: "Adeem"
    - english: "Excavation"
      arabic: "حفريات"
      pronunciation: "Hafriyat"
    - english: "Archaeology"
      arabic: "علم الآثار"
      pronunciation: "Ilm al-Athar"
  Fortress:
    - english: "Fortress"
      arabic: "قلعة"
      pronunciation: "Qal'a"
    - english: "Tower"
      arabic: "برج"
      pronunciation: "Borg"
    - english: "Wall"
      arabic: "سور"
      pronunciation: "Soor"
    - english: "Gate"
      arabic: "بوابة"
      pronunciation: "Bawwaba"
  Church:
    - english: "Church"
      arabic: "كنيسة"
      pronunciation: "Kaneesa"
    - english: "Coptic"
      arabic: "قبطي"
      pronunciation: "Qibti"
    - english: "Monastery"
      arabic: "دير"
      pronunciation: "Deir"
    - english: "Icon"
      arabic: "أيقونة"
      pronunciation: "Ayqoona"

# CSS selectors for web scraping
selectors:
  list_item: "a.listItem"
  show_more_button: '[class*="showMore"], .showMoreBtn'
  location: ".location p"
  description: ".details > p"
  title: "h1, .title h1, .pageTitle"
  gallery_images: ".gallery img, .slider img, .monumentGallery img, article img, .swiper-slide img"

# Content filtering patterns
content_filters:
  footer_indicators:
    - copyright
    - developed by
    - all rights reserved
    - ministry of
  navigation_indicators:
    - read more
    - click here
    - show more
    - back to
    - home >

# Logging configuration
logging:
  level: "INFO"
  format: "%(asctime)s - %(name)s - %(levelname)s - %(message)s"
  date_format: "%Y-%m-%d %H:%M:%S"