deep_classiflie_db.yaml

db:
  update_db: True
  debug_enabled: False
  db_commit_freq: 500
  use_chrome: False
  db_env_file: "" # defaults to $HOME/.dcdb_env or .dcdb_env_dev if in DEV_MODE
scraping:
  factbase:
    init_wait: 0.2
    max_pages: 1000
    connection_timeout: 120
    max_retries: 8
    request_timeouts:
      connect: 5
      read: 30
    sql:
      latest_statement: 'select max(t_date) from fbase_transcripts'
      check_tid: 'select tid from fbase_transcripts where tid in'
      txcpt_invariant: >-
        INTO fbase_transcripts
        (tid, t_date, transcript_type, transcript_url, transcript_sent)
        VALUES (%s, %s, %s, %s, %s )
      stmt_invariant: >-
        INTO fbase_statements
        (tid, sid, statement_text, sentiment)
        VALUES (%s, %s, %s, %s)
  wapo:
    init_wait: 10
    max_retries: 4
    sql:
      latest_statement: 'select max(s_date) from wp_statements where iid=1'
      stmt_invariant: >-
        INTO wp_statements
        (iid,statement_text, repeats, topic, source, pinnochios, s_date, statement_hash)
        VALUES (%s, %s, %s, %s, %s, %s, %s, SHA1(%s))
  twitter:
    default_latest_stmt: 0  # need to test init load logic by temp changing latest_statement sql
    sql:
      latest_statement: 'select max(end_thread_tweet_id) from dcbot_tweets'
      stmt_invariant: >-
        INTO dcbot_tweets
        (end_thread_tweet_id, statement_text, t_start_date, t_end_date, retweet, statement_hash)
        VALUES (%s, %s, %s, %s, %s, SHA1(%s))
dirs:
  base_dir: "" # defaults to $HOME
  log_dir: "" # defaults to {base_dir}/temp/{constants.APP_NAME}/logs