update ingest request schema

This is mostly changing ingest_type from 'file' to 'pdf', and adding
'link_source'/'link_source_id', plus some small cleanups.

2 files changed, 14 insertions, 8 deletions

diff --git a/python/fatcat_web/forms.py b/python/fatcat_web/forms.py
index bd4e4bbd..5539cc20 100644
--- a/python/fatcat_web/forms.py
+++ b/python/fatcat_web/forms.py
@@ -386,18 +386,16 @@ class SavePaperNowForm(FlaskForm):
         choices=release_stage_options,
         default='')
 
-    def to_ingest_request(self, release, actor='savepapernow-web'):
+    def to_ingest_request(self, release, ingest_request_source='savepapernow'):
         base_url = self.base_url.data
         ext_ids = release.ext_ids.to_dict()
         # by default this dict has a bunch of empty values
         ext_ids = dict([(k, v) for (k, v) in ext_ids.items() if v])
         ingest_request = {
             'ingest_type': self.ingest_type.data,
-            'ingest_request_source': actor, # TODO: deprecate?
-            'actor': actor,
+            'ingest_request_source': ingest_request_source,
             'base_url': base_url,
             'fatcat': {
-                'release_stage': release.release_stage,
                 'release_ident': release.ident,
                 'work_ident': release.work_id,
             },
@@ -405,8 +403,12 @@ class SavePaperNowForm(FlaskForm):
         }
         if self.release_stage.data:
             ingest_request['release_stage'] = self.release_stage.data
+
         if release.ext_ids.doi and base_url == "https://doi.org/{}".format(release.ext_ids.doi):
-            ingest_request['source'] = 'doi'
-            ingest_request['source_id'] = release.ext_ids.doi
+            ingest_request['link_source'] = 'doi'
+            ingest_request['link_source_id'] = release.ext_ids.doi
+        elif release.ext_ids.arxiv and base_url == "https://arxiv.org/pdf/{}.pdf".format(release.ext_ids.arxiv):
+            ingest_request['link_source'] = 'arxiv'
+            ingest_request['link_source_id'] = release.ext_ids.arxiv
         return ingest_request
 
diff --git a/python/fatcat_web/routes.py b/python/fatcat_web/routes.py
index cc0af5cc..8583d255 100644
--- a/python/fatcat_web/routes.py
+++ b/python/fatcat_web/routes.py
@@ -648,11 +648,11 @@ def release_save(ident):
     if form.is_submitted():
         if form.validate_on_submit():
             # got a valid spn request! try to send to kafka-pixy
-            msg = form.to_ingest_request(release)
+            msg = form.to_ingest_request(release, ingest_request_source="savepapernow-web")
             try:
                 kafka_pixy_produce(
                     Config.KAFKA_SAVEPAPERNOW_TOPIC,
-                    json.dumps(msg),
+                    json.dumps(msg, sort_keys=True),
                 )
             except Exception as e:
                 print(e, file=sys.stderr)
@@ -666,6 +666,10 @@ def release_save(ident):
         form.release_stage.data = release.release_stage
     if release.ext_ids.doi:
         form.base_url.data = "https://doi.org/{}".format(release.ext_ids.doi)
+    elif release.ext_ids.arxiv:
+        form.base_url.data = "https://arxiv.org/pdf/{}.pdf".format(release.ext_ids.arxiv)
+    elif release.ext_ids.pmcid:
+        form.base_url.data = "http://europepmc.org/backend/ptpmcrender.fcgi?accid={}&blobtype=pdf".format(release.ext_ids.pmcid)
     return render_template('release_save.html', entity=release, form=form), 200
 
 ### Search ##################################################################