start container coverage page

4 files changed, 104 insertions, 2 deletions

diff --git a/python/fatcat_web/routes.py b/python/fatcat_web/routes.py
index f313fce0..3f5af621 100644
--- a/python/fatcat_web/routes.py
+++ b/python/fatcat_web/routes.py
@@ -230,6 +230,10 @@ def generic_editgroup_entity_view(editgroup_id, entity_type, ident, view_templat
 def container_view(ident):
     return generic_entity_view('container', ident, 'container_view.html')
 
+@app.route('/container/<ident>/coverage', methods=['GET'])
+def container_view_coverage(ident):
+    return generic_entity_view('container', ident, 'container_view_coverage.html')
+
 @app.route('/container/<ident>/metadata', methods=['GET'])
 def container_view_metadata(ident):
     return generic_entity_view('container', ident, 'entity_view_metadata.html')
@@ -718,6 +722,21 @@ def container_ident_stats(ident):
         abort(503)
     return jsonify(stats)
 
+@app.route('/container/<ident>/ia_coverage_years.json', methods=['GET', 'OPTIONS'])
+@crossdomain(origin='*',headers=['access-control-allow-origin','Content-Type'])
+def container_ident_ia_coverage_years_json(ident):
+    try:
+        container = api.get_container(ident)
+    except ApiException as ae:
+        abort(ae.status)
+    try:
+        histogram = get_elastic_container_histogram(container.ident)
+    except Exception as ae:
+        app.log.error(ae)
+        abort(503)
+    histogram = [dict(year=h[0], in_ia=h[1], count=h[2]) for h in histogram]
+    return jsonify({'container_id': ident, "histogram": histogram})
+
 @app.route('/release/<ident>.bib', methods=['GET'])
 def release_bibtex(ident):
     try:
diff --git a/python/fatcat_web/search.py b/python/fatcat_web/search.py
index 94246329..523269ce 100644
--- a/python/fatcat_web/search.py
+++ b/python/fatcat_web/search.py
@@ -274,7 +274,7 @@ def get_elastic_container_random_releases(ident, limit=5):
     #print(resp.json())
     resp.raise_for_status()
     resp = resp.json()
-    print(resp)
+    #print(resp)
     hits = [h['_source'] for h in resp['hits']['hits']]
     for h in hits:
         # Handle surrogate strings that elasticsearch returns sometimes,
@@ -285,3 +285,69 @@ def get_elastic_container_random_releases(ident, limit=5):
                 h[key] = h[key].encode('utf8', 'ignore').decode('utf8')
 
     return hits
+
+def get_elastic_container_histogram(ident):
+    """
+    Fetches a stacked histogram of 
+
+    Filters to the past 500 years (at most), or about 1000 vaules.
+
+    Returns a list of tuples:
+        (year, in_ia, count)
+    """
+
+    query = {
+        "aggs": {
+            "year_in_ia": {
+                "composite": {
+                    "size": 1000,
+                    "sources": [
+                        {"year": {
+                            "histogram": {
+                                "field": "release_year",
+                                "interval": 1,
+                        }}},
+                        {"in_ia": {
+                            "terms": {
+                                "field": "in_ia",
+                        }}},
+                    ],
+                },
+            },
+        },
+        "size": 0,
+        "query": {
+            "bool": {
+                "must": [{
+                    "range": {
+                        "release_year": {
+                            "gte": datetime.datetime.today().year - 499,
+                            "lte": datetime.datetime.today().year,
+                        }
+                    }
+                }],
+                "filter": [{
+                    "bool": {
+                        "should": [{
+                            "match": {
+                                "container_id": ident
+                            }
+                        }],
+                        "minimum_should_match": 1,
+                    },
+                }],
+            }
+        }
+    }
+    resp = requests.get(
+        "{}/fatcat_release/_search".format(app.config['ELASTICSEARCH_BACKEND']),
+        json=query,
+        params=dict(request_cache="true"))
+    resp.raise_for_status()
+    # TODO: abort()
+    resp = resp.json()
+    print(resp)
+    vals = [(h['key']['year'], h['key']['in_ia'], h['doc_count'])
+            for h in resp['aggregations']['year_in_ia']['buckets']]
+    vals = sorted(vals)
+    return vals
diff --git a/python/fatcat_web/templates/container_view_coverage.html b/python/fatcat_web/templates/container_view_coverage.html
new file mode 100644
index 00000000..eb9dba8a
--- /dev/null
+++ b/python/fatcat_web/templates/container_view_coverage.html
@@ -0,0 +1,15 @@
+{% set container = entity %}
+{% set entity_view = "coverage" %}
+{% set entity_type = "container" %}
+{% import "entity_macros.html" as entity_macros %}
+{% extends "entity_base.html" %}
+
+{% block entity_main %}
+
+<h3>Preservation Coverage By Year</h3>
+
+<img src="/container/{{ container.ident }}/ia_coverage_years.svg">
+<br><a href="/container/{{ container.ident }}/ia_coverage_years.json">Download as JSON</a>
+
+{% endblock %}
+
diff --git a/python/fatcat_web/templates/entity_base.html b/python/fatcat_web/templates/entity_base.html
index bba95d9d..48f51ec6 100644
--- a/python/fatcat_web/templates/entity_base.html
+++ b/python/fatcat_web/templates/entity_base.html
@@ -78,7 +78,9 @@
         <div style="min-width: 40em;">
           <div class="ui small tabular compact menu">
             {{ entity_tab("overview", "Overview", "") }}
-            {% if entity_type == "release" %}
+            {% if entity_type == "container" and entity.state == 'active' and not editgroup %}
+              {{ entity_tab("coverage", "Coverage", "/coverage") }}
+            {% elif entity_type == "release" %}
               {{ entity_tab("contribs", "Authors", "/contribs", entity._authors|count ) }}
               {{ entity_tab("references", "References", "/references", entity.refs|count) }}
             {% endif %}