diff options
| author | Bryan Newbold <bnewbold@archive.org> | 2022-07-12 15:03:29 -0700 |
|---|---|---|
| committer | Bryan Newbold <bnewbold@archive.org> | 2022-07-12 15:03:29 -0700 |
| commit | c15432c0ce52c48efabcd7e3221a5d625ef3e9d0 (patch) | |
| tree | 948d8e72a7235d07a5ea5d778e8ee388003418e6 /python/sandcrawler/html_metadata.py | |
| parent | 8f85ab294eae50e31efa9e31bb0bca1bca76cf8b (diff) | |
| download | sandcrawler-bnewbold-refactor-loggging.tar.gz sandcrawler-bnewbold-refactor-loggging.zip | |
WIP: refactor logging calls in ingest pipelinesbnewbold-refactor-loggging
Diffstat (limited to 'python/sandcrawler/html_metadata.py')
| -rw-r--r-- | python/sandcrawler/html_metadata.py | 6 |
1 files changed, 3 insertions, 3 deletions
diff --git a/python/sandcrawler/html_metadata.py b/python/sandcrawler/html_metadata.py index 2fb500c..cee3b00 100644 --- a/python/sandcrawler/html_metadata.py +++ b/python/sandcrawler/html_metadata.py @@ -1,5 +1,5 @@ import datetime -import sys +import logging import urllib.parse from typing import Any, Dict, List, Optional, Tuple @@ -726,7 +726,7 @@ def html_extract_fulltext_url( continue return (val, pattern.get("technique", "unknown")) if self_doc_url: - print(" WARN: returning fulltext URL pointing to self", file=sys.stderr) + logging.warn(f"returning fulltext URL pointing to self {self_doc_url=}") return self_doc_url return None @@ -736,7 +736,7 @@ def html_extract_biblio(doc_url: str, doc: HTMLParser) -> Optional[BiblioMetadat meta: Any = dict() head = doc.css_first("head") if not head: - print(f"WARN: empty <head>? {doc_url}", file=sys.stderr) + logging.warn(f"empty HTML head {doc_url=}") return None for field, patterns in HEAD_META_PATTERNS.items(): |