paths.py 670 B

123456789101112131415
  1. import os
  2. HOME = os.getenv('HOME')
  3. DATA_DIR = os.path.join(HOME, 'data', 'tinysearch')
  4. HN_TOP_PATH = os.path.join(DATA_DIR, 'hn-top.csv')
  5. CRAWL_PREFIX = 'crawl_'
  6. CRAWL_GLOB = os.path.join(DATA_DIR, f"{CRAWL_PREFIX}*")
  7. INDEX_PATH = os.path.join(DATA_DIR, 'index.tinysearch')
  8. TEST_INDEX_PATH = os.path.join(DATA_DIR, 'index-test.tinysearch')
  9. WIKI_DATA_PATH = os.path.join(DATA_DIR, 'enwiki-20210301-pages-articles1.xml-p1p41242.bz2')
  10. WIKI_TITLES_PATH = os.path.join(DATA_DIR, 'abstract-titles-sorted.txt.gz')
  11. DOMAINS_QUEUE_NAME = 'domains-queue-fs'
  12. DOMAINS_TITLES_QUEUE_NAME = 'domains-title-queue-fs'
  13. DOMAINS_PATH = os.path.join(DATA_DIR, 'top10milliondomains.csv.gz')