From cac2bc677fb7dfca9230ec4cee343b983e31fc96 Mon Sep 17 00:00:00 2001 From: Martin Czygan Date: Thu, 11 Feb 2021 13:53:26 +0100 Subject: update notes --- fuzzycat/__main__.py | 14 ++++++++++---- fuzzycat/refs.py | 12 ++++++------ 2 files changed, 16 insertions(+), 10 deletions(-) diff --git a/fuzzycat/__main__.py b/fuzzycat/__main__.py index 67ffa40..9121bd8 100644 --- a/fuzzycat/__main__.py +++ b/fuzzycat/__main__.py @@ -6,6 +6,7 @@ COMMANDS cluster verify verify_single + verify_ref release_match Run, e.g. fuzzycat cluster --help for more options. @@ -38,6 +39,10 @@ EXAMPLES ] } + Verify clustered refs: + + $ python -m fuzzycat verify-ref + Release match (non-bulk). $ python -m fuzzycat release_match -q "hello world" @@ -66,8 +71,8 @@ from fuzzycat.cluster import (Cluster, release_key_title, release_key_title_ngra release_key_title_sandcrawler) from fuzzycat.entities import entity_to_dict from fuzzycat.matching import anything_to_entity, match_release_fuzzy -from fuzzycat.utils import random_idents_from_query, random_word from fuzzycat.refs import RefsGroupVerifier +from fuzzycat.utils import random_idents_from_query, random_word from fuzzycat.verify import GroupVerifier, verify logging.getLogger("requests").setLevel(logging.WARNING) @@ -161,11 +166,12 @@ def run_verify_single(args): }) print(json.dumps(result)) + def run_ref_verify(args): - verifier = RefsGroupVerifier(iterable=fileinput.input(files=args.files), - verbose=args.verbose) + verifier = RefsGroupVerifier(iterable=fileinput.input(files=args.files), verbose=args.verbose) verifier.run() + def run_release_match(args): """ Given a release, return similar releases. @@ -253,7 +259,7 @@ if __name__ == '__main__': sub_verify_single.add_argument('-b', help='ident or url to release') sub_verify_single.set_defaults(func=run_verify_single) - sub_ref_verify = subparsers.add_parser('verify', help='verify ref groups', parents=[parser]) + sub_ref_verify = subparsers.add_parser('verify_ref', help='verify ref groups', parents=[parser]) sub_ref_verify.add_argument('-f', '--files', default="-", help='input files') sub_ref_verify.set_defaults(func=run_ref_verify) diff --git a/fuzzycat/refs.py b/fuzzycat/refs.py index 04420d0..689d800 100644 --- a/fuzzycat/refs.py +++ b/fuzzycat/refs.py @@ -4,10 +4,11 @@ import json import operator import sys -from fuzzycat.verify import verify -from fuzzycat.common import Reason, Status from glom import PathAccessError, glom +from fuzzycat.common import Reason, Status +from fuzzycat.verify import verify + def find_release_entity(docs): """ @@ -23,6 +24,7 @@ def find_release_entity(docs): raise ValueError("docs do not contain any release") + def ref_entities(docs): """ Genator yielding ref entities only. @@ -34,14 +36,13 @@ def ref_entities(docs): except PathAccessError: continue + class RefsGroupVerifier: """ A specific verifier for grouped releases and references. We do not need to pair-wise compare, just compare one release to all references. """ - def __init__(self, - iterable: collections.abc.Iterable, - verbose=False): + def __init__(self, iterable: collections.abc.Iterable, verbose=False): self.iterable: collections.abc.Iterable = iterable self.verbose: bool = verbose self.counter: Counter = collections.Counter() @@ -64,4 +65,3 @@ class RefsGroupVerifier: "https://fatcat.wiki/release/{}".format(b["ident"]), result, reason) self.counter["total"] = sum(v for _, v in self.counter.items()) - -- cgit v1.2.3