From: Peter Amstutz <peter.amstutz@curii.com>
Date: Mon, 7 Nov 2022 21:59:05 +0000 (-0500)
Subject: 19699: Tighten up etag handling a bit
X-Git-Tag: 2.5.0~37^2~4
X-Git-Url: https://git.arvados.org/arvados.git/commitdiff_plain/c6428570be58b01ce80c257adce06114cea1d88f

19699: Tighten up etag handling a bit

Arvados-DCO-1.1-Signed-off-by: Peter Amstutz <peter.amstutz@curii.com>
---

diff --git a/sdk/cwl/arvados_cwl/http.py b/sdk/cwl/arvados_cwl/http.py
index d85ae30d2d..33aa098845 100644
--- a/sdk/cwl/arvados_cwl/http.py
+++ b/sdk/cwl/arvados_cwl/http.py
@@ -88,6 +88,14 @@ def changed(url, properties, now):
 
     return True
 
+def etag_quote(etag):
+    # if it already has leading and trailing quotes, do nothing
+    if etag[0] == '"' and etag[-1] == '"':
+        return etag
+    else:
+        # Add quotes.
+        return '"' + etag + '"'
+
 def http_to_keep(api, project_uuid, url, utcnow=datetime.datetime.utcnow):
     r = api.collections().list(filters=[["properties", "exists", url]]).execute()
 
@@ -108,13 +116,13 @@ def http_to_keep(api, project_uuid, url, utcnow=datetime.datetime.utcnow):
             cr = arvados.collection.CollectionReader(item["portable_data_hash"], api_client=api)
             return "keep:%s/%s" % (item["portable_data_hash"], list(cr.keys())[0])
 
-        if "ETag" in properties:
+        if "ETag" in properties and len(properties["ETag"]) > 2:
             etags[properties["ETag"]] = item
 
     properties = {}
     headers = {}
     if etags:
-        headers['If-None-Match'] = ', '.join(['"%s"' % k for k,v in etags.items()])
+        headers['If-None-Match'] = ', '.join([etag_quote(k) for k,v in etags.items()])
     req = requests.get(url, stream=True, allow_redirects=True, headers=headers)
 
     if req.status_code not in (200, 304):