Renamed basedir to archivedir.
This commit is contained in:
parent
0b96aeb49d
commit
a346118c13
21
crawler
21
crawler
|
@ -51,14 +51,14 @@ def update_support(dir, verbose, ext_id):
|
||||||
return True
|
return True
|
||||||
|
|
||||||
|
|
||||||
def update_extension(basedir, verbose, forums, ext_id):
|
def update_extension(archivedir, verbose, forums, ext_id):
|
||||||
sys.stdout.write(" Update Extension: " + ext_id + "\n")
|
sys.stdout.write(" Update Extension: " + ext_id + "\n")
|
||||||
if verbose:
|
if verbose:
|
||||||
sys.stdout.write(" Updating {}".format(ext_id))
|
sys.stdout.write(" Updating {}".format(ext_id))
|
||||||
if forums:
|
if forums:
|
||||||
sys.stdout.write(" (including forums)")
|
sys.stdout.write(" (including forums)")
|
||||||
sys.stdout.write("\n")
|
sys.stdout.write("\n")
|
||||||
dir = basedir + "/" + (
|
dir = archivedir + "/" + (
|
||||||
ExtensionCrawler.archive.get_local_archive_dir(ext_id))
|
ExtensionCrawler.archive.get_local_archive_dir(ext_id))
|
||||||
os.makedirs(dir, exist_ok=True)
|
os.makedirs(dir, exist_ok=True)
|
||||||
update_overview(dir, verbose, ext_id)
|
update_overview(dir, verbose, ext_id)
|
||||||
|
@ -68,19 +68,19 @@ def update_extension(basedir, verbose, forums, ext_id):
|
||||||
update_support(dir, verbose, ext_id)
|
update_support(dir, verbose, ext_id)
|
||||||
|
|
||||||
|
|
||||||
def update_extensions(basedir, verbose, forums_ext_ids, ext_ids):
|
def update_extensions(archivedir, verbose, forums_ext_ids, ext_ids):
|
||||||
def update_forums(ext_id):
|
def update_forums(ext_id):
|
||||||
return (ext_id in forums_ext_ids)
|
return (ext_id in forums_ext_ids)
|
||||||
foo = list(map(lambda ext_id: update_extension(basedir, verbose, update_forums(ext_id), ext_id), ext_ids))
|
foo = list(map(lambda ext_id: update_extension(archivedir, verbose, update_forums(ext_id), ext_id), ext_ids))
|
||||||
return foo
|
return foo
|
||||||
|
|
||||||
|
|
||||||
def get_existing_ids(basedir, verbose):
|
def get_existing_ids(archivedir, verbose):
|
||||||
byte = '[0-9a-z][0-9a-z][0-9a-z][0-9a-z][0-9a-z][0-9a-z][0-9a-z][0-9a-z]'
|
byte = '[0-9a-z][0-9a-z][0-9a-z][0-9a-z][0-9a-z][0-9a-z][0-9a-z][0-9a-z]'
|
||||||
word = byte + byte + byte + byte
|
word = byte + byte + byte + byte
|
||||||
return list(
|
return list(
|
||||||
map(lambda d: re.sub("^.*\/", "", d),
|
map(lambda d: re.sub("^.*\/", "", d),
|
||||||
glob.glob(os.path.join(basedir, "*", word))))
|
glob.glob(os.path.join(archivedir, "*", word))))
|
||||||
|
|
||||||
|
|
||||||
def get_forum_ext_ids(confdir, verbose):
|
def get_forum_ext_ids(confdir, verbose):
|
||||||
|
@ -91,14 +91,15 @@ def get_forum_ext_ids(confdir, verbose):
|
||||||
|
|
||||||
|
|
||||||
def main():
|
def main():
|
||||||
basedir = "./archive"
|
basedir = "."
|
||||||
confdir = "./conf"
|
archivedir = os.path.join(basedir,"archive")
|
||||||
|
confdir = os.path.join(basedir,"conf")
|
||||||
verbose = True
|
verbose = True
|
||||||
|
|
||||||
sys.stdout.write("Crawling ID\n")
|
sys.stdout.write("Crawling ID\n")
|
||||||
discovered_ids = [] # ExtensionCrawler.discover.crawl_nearly_all_of_ext_ids()
|
discovered_ids = [] # ExtensionCrawler.discover.crawl_nearly_all_of_ext_ids()
|
||||||
forum_ext_ids = get_forum_ext_ids(confdir, verbose)
|
forum_ext_ids = get_forum_ext_ids(confdir, verbose)
|
||||||
existing_ids = get_existing_ids(basedir, verbose)
|
existing_ids = get_existing_ids(archivedir, verbose)
|
||||||
existing_ids = list(set(existing_ids) | set(forum_ext_ids))
|
existing_ids = list(set(existing_ids) | set(forum_ext_ids))
|
||||||
new_ids = list(set(discovered_ids) - set(existing_ids))
|
new_ids = list(set(discovered_ids) - set(existing_ids))
|
||||||
|
|
||||||
|
@ -109,7 +110,7 @@ def main():
|
||||||
str(len(new_ids)), str(len(existing_ids)), str(len(
|
str(len(new_ids)), str(len(existing_ids)), str(len(
|
||||||
forum_ext_ids))))
|
forum_ext_ids))))
|
||||||
|
|
||||||
update_extensions(basedir, verbose, forum_ext_ids, existing_ids + new_ids)
|
update_extensions(archivedir, verbose, forum_ext_ids, existing_ids + new_ids)
|
||||||
|
|
||||||
|
|
||||||
main()
|
main()
|
||||||
|
|
Loading…
Reference in New Issue