From e18d48642cecb55d9f2270f9048953a7b543472e Mon Sep 17 00:00:00 2001 From: Martin Czygan Date: Mon, 10 Aug 2020 19:55:14 +0200 Subject: harvest: datacite API yields HTTP 200 with broken JSON As a first step: log response body for debugging. --- python/fatcat_tools/harvest/doi_registrars.py | 9 ++++++++- 1 file changed, 8 insertions(+), 1 deletion(-) diff --git a/python/fatcat_tools/harvest/doi_registrars.py b/python/fatcat_tools/harvest/doi_registrars.py index 2554fe96..03773c15 100644 --- a/python/fatcat_tools/harvest/doi_registrars.py +++ b/python/fatcat_tools/harvest/doi_registrars.py @@ -136,7 +136,14 @@ class HarvestCrossrefWorker: time.sleep(30.0) continue http_resp.raise_for_status() - resp = http_resp.json() + try: + resp_body = http_resp.text + resp = json.loads(resp_body) + except json.JSONDecodeError as exc: + # Datacite API returned HTTP 200, but JSON seemed unparseable. + # It might be a glitch, so we retry. + print("failed to decode body from {}: {}".format(http_resp.url, resp_body), file=sys.stderr) + raise items = self.extract_items(resp) count += len(items) print("... got {} ({} of {}), HTTP fetch took {}".format(len(items), count, -- cgit v1.2.3