Added new Docmap queries

[Issue: #13]
source-data · Oct 5, 2021 · 28910a9 · 28910a9
1 parent df5a070
commit 28910a9
Show file tree

Hide file tree

Showing 2 changed files with 76 additions and 28 deletions.
diff --git a/neoflask/queries.py b/neoflask/queries.py
@@ -425,21 +425,18 @@ class DOCMAP_BY_DOI(_DOCMAP):
  '''
  map_docmap_filtering = {'doi': {'req_param': 'doi', 'default': ''}}
 
-class DOCMAPS_BY_REVSERVICE_AND_INTERVAL(_DOCMAP):
+class DOCMAPS_FROM_REVSERVICE_IN_INTERVAL(_DOCMAP):
  code_docmap_filtering = '''
-MATCH (
- col:VizCollection {name: "refereed-preprints"}
-)-[:HasSubCol]->(
- subcol:VizSubCollection
-)-[:HasPaper]->(
- paper:VizPaper
-)-[:HasReviewDate]->(
- revdate:VizReviewDate
-)
-WHERE DATETIME(revdate.date) >= DATETIME($start_date)
- AND DATETIME(revdate.date) < DATETIME($end_date)
+MATCH
+ (col:VizCollection)-[:HasSubCol]->(subcol:VizSubCollection)-[:HasPaper]->(paper:VizPaper),
+ (preprint:Preprint)-[:inputs]->(Step)<-[:actions]-(Action)<-[:outputs]-(review:RefereeReport)
+WHERE col.name = "refereed-preprints"
  AND subcol.name = $reviewing_service
-WITH COLLECT(DISTINCT paper.doi) AS doiList
+ AND paper.doi = preprint.doi
+WITH preprint, DATETIME(MAX(review.published)) AS publish_date_newest_review
+WHERE publish_date_newest_review >= DATETIME($start_date)
+ AND publish_date_newest_review < DATETIME($end_date)
+WITH COLLECT(DISTINCT preprint.doi) AS doiList
 UNWIND doiList AS doi
  '''
  map_docmap_filtering = {
@@ -457,6 +454,27 @@ class DOCMAPS_BY_REVSERVICE_AND_INTERVAL(_DOCMAP):
  },
  }
 
+class DOCMAPS_IN_INTERVAL(_DOCMAP):
+ code_docmap_filtering = '''
+MATCH
+ (preprint:Preprint)-[:inputs]->(Step)<-[:actions]-(Action)<-[:outputs]-(review:RefereeReport)
+WITH preprint, DATETIME(MAX(review.published)) AS publish_date_newest_review
+WHERE publish_date_newest_review >= DATETIME($start_date)
+ AND publish_date_newest_review < DATETIME($end_date)
+WITH COLLECT(DISTINCT preprint.doi) AS doiList
+UNWIND doiList AS doi
+ '''
+ map_docmap_filtering = {
+ 'start_date': {
+ 'req_param': 'start_date',
+ 'default': '1900-01-01',
+ },
+ 'end_date': {
+ 'req_param': 'end_date',
+ 'default': '2900-01-01',
+ },
+ }
+
 class SEARCH_REVIEWS(Query):
  code = '''
 MATCH (

diff --git a/neoflask/views.py b/neoflask/views.py
@@ -1,3 +1,4 @@
+from datetime import date, timedelta
 from flask import (
  abort,
  jsonify,
@@ -12,7 +13,8 @@
  STATS, BY_DOI, FIG_BY_DOI_IDX,
  DESCRIBE_REVIEWING_SERVICES,
  REVIEW_PROCESS_BY_DOI, REVIEW_MATERIAL_BY_ID,
- DOCMAP_BY_DOI, DOCMAPS_BY_REVSERVICE_AND_INTERVAL,
+ DOCMAP_BY_DOI, DOCMAPS_FROM_REVSERVICE_IN_INTERVAL,
+ DOCMAPS_IN_INTERVAL,
  BY_AUTO_TOPICS, BY_REVIEWING_SERVICE, AUTOMAGIC,
  LUCENE_SEARCH, SEARCH_DOI,
  COVID19, REFEREED_PREPRINTS,
@@ -254,31 +256,59 @@ def docmap_semantic_doi(doi: str):
  j = ask_neo(DOCMAP_BY_DOI(), doi=doi, root=root)
  return jsonify(j)
 
-@app.route('/api/v2/<reviewing_service>/docmap/<start_date>/<end_date>/<int:pagination>', methods=['GET', 'POST'])
-def docmap_search(reviewing_service: str, start_date: str, end_date: str, pagination: int):
- app.logger.info(f"Searching for docmaps with parameters {reviewing_service}, {start_date}, {end_date}, {pagination}")
- if pagination < 0:
+def do_paginated_docmap_query(query, page=0, page_size=100, **kwargs):
+ if page < 0:
  abort(400) # pagination paramater must be 0 or greater
- page_size = 100
- offset = pagination * page_size
+
+ offset = page * page_size
  root = url_for('root', _external=True)
 
  result = ask_neo(
- DOCMAPS_BY_REVSERVICE_AND_INTERVAL(),
- reviewing_service=reviewing_service,
- start_date=start_date,
- end_date=end_date,
+ query,
+ root=root,
  offset=offset,
  page_size=page_size,
- root=root,
+ **kwargs,
  )
 
  return jsonify(result)
 
-# @app.route('/api/v2/docmap/<start_date>/<end_date>/<int:pagination>', methods=['GET', 'POST'])
+@app.route('/api/v2/<reviewing_service>/docmap/<start_date>/<end_date>/<int:pagination>', methods=['GET', 'POST'])
+def docmaps_from_revservice_in_interval(reviewing_service: str, start_date: str, end_date: str, pagination: int):
+ app.logger.info(f"Getting docmaps for reviewing service \"{reviewing_service}\" from {start_date} to {end_date}, page {pagination}")
+ return do_paginated_docmap_query(
+ DOCMAPS_FROM_REVSERVICE_IN_INTERVAL(),
+ reviewing_service=reviewing_service,
+ start_date=start_date,
+ end_date=end_date,
+ page=pagination,
+ )
+
+@app.route('/api/v2/docmap/<start_date>/<end_date>/<int:pagination>', methods=['GET', 'POST'])
+def docmaps_in_interval(start_date: str, end_date: str, pagination: int):
+ app.logger.info(f"Getting docmaps from {start_date} to {end_date}, page {pagination}")
+ return do_paginated_docmap_query(
+ DOCMAPS_IN_INTERVAL(),
+ start_date=start_date,
+ end_date=end_date,
+ page=pagination,
+ )
 
-# @app.route('/api/v2/<reviewing_service>/docmap/<int:N_most_recent>/<int:pagination>', methods=['GET', 'POST'])
+@app.route('/api/v2/<reviewing_service>/docmap/<int:days>d/<int:pagination>', methods=['GET', 'POST'])
+def docmaps_from_revservice_in_last_days(reviewing_service: str, days: int, pagination: int):
+ app.logger.info(f"Getting docmaps for reviewing service \"{reviewing_service}\" with reviews in the last {days} days, page {pagination}")
+ # The Docmap query is exclusive for the end of the interval. We want any reviews published today so add a day here.
+ end_date = date.today() + timedelta(days=1)
+ # With this start date we get any reviews published today and in the last n-1 days, so reviews from a total of n days.
+ start_date = end_date - timedelta(days=days)
+ return do_paginated_docmap_query(
+ DOCMAPS_FROM_REVSERVICE_IN_INTERVAL(),
+ reviewing_service=reviewing_service,
+ start_date=str(start_date),
+ end_date=str(end_date),
+ page=pagination,
+ )
 
-# @app.route('/api/v2/<reviewing_service>/docmap/<int:days>d/<int:pagination>', methods=['GET', 'POST'])
+# @app.route('/api/v2/<reviewing_service>/docmap/<int:n_most_recent>/<int:pagination>', methods=['GET', 'POST'])
 
 # date of docmap creation and date of docmap update. filter on publishing data or on docmap creation date.