Merge pull request #1 from OpenBudget/datarecords-endpoint

Add /api/datarecords/{key} endpoint

Author: akariv

budgetkey_api/flask_app.py

@@ -94,6 +94,12 @@ def create_flask_app(session_file_dir=None, cache_dir=None, services=None):
         setup_simpledb(app, es_blueprint, db_blueprint)
         log.info("SimpleDB setup complete")
 
+    # No external dependencies, so always available
+    log.info("Setting up DataRecords")
+    from .modules.datarecords import setup_datarecords
+    setup_datarecords(app, cache)
+    log.info("DataRecords setup complete")
+
     app.before_request(logging_before)
     app.after_request(logging_after)
 

budgetkey_api/modules/datarecords.py

@@ -0,0 +1,57 @@
+import os
+
+import requests
+
+from flask import Blueprint, abort
+from flask_jsonpify import jsonpify
+
+from .caching import add_cache_header
+
+DATARECORDS_URL = os.environ.get('DATARECORDS_URL', 'https://data-input.obudget.org/api/datarecords')
+TIMEOUT = 24 * 60 * 60  # one day
+
+KEYS = [
+    'target_age_group',
+    'subject',
+    'intervention',
+    'target_audience',
+]
+
+
+class DataRecordsBlueprint(Blueprint):
+
+    def __init__(self, cache):
+        super().__init__('datarecords', 'datarecords')
+        self.cache = cache
+
+        self.add_url_rule(
+            '/datarecords/<key>',
+            'datarecords',
+            self.get_datarecords,
+            methods=['GET']
+        )
+
+    def fetch_datarecords(self, key):
+        response = requests.get(f'{DATARECORDS_URL}/{key}', timeout=60)
+        response.raise_for_status()
+        return response.json()
+
+    def get_datarecords(self, key):
+        if key not in KEYS:
+            abort(404, f'Data record {key} not found. Available keys: {", ".join(KEYS)}')
+        cache_key = f'datarecords/{key}'
+        data = self.cache.get(cache_key)
+        if data is None:
+            try:
+                data = self.fetch_datarecords(key)
+            except Exception as e:
+                abort(502, f'Failed to fetch data records for {key}: {e}')
+            self.cache.set(cache_key, data, timeout=TIMEOUT)
+        return jsonpify(data)
+
+
+def setup_datarecords(app, cache):
+    bp = DataRecordsBlueprint(cache)
+    add_cache_header(bp, TIMEOUT)
+    app.register_blueprint(bp, url_prefix='/api/')
+    return bp

tests/test_datarecords.py

@@ -0,0 +1,61 @@
+import dotenv
+
+dotenv.load_dotenv('tests/sample.env')
+
+
+def make_client(tmp_path):
+    from budgetkey_api.flask_app import create_flask_app
+
+    app = create_flask_app(session_file_dir=str(tmp_path / 'sessions'), cache_dir=str(tmp_path / 'cache'),
+                           services='none')
+    app.config.update({'TESTING': True})
+    app.testing = True
+    return app.test_client()
+
+
+def test_datarecords_bad_key(tmp_path):
+    client = make_client(tmp_path)
+    resp = client.get('/api/datarecords/no_such_key')
+    assert resp.status_code == 404
+
+
+def test_datarecords_fetches_and_caches(tmp_path, monkeypatch):
+    from budgetkey_api.modules import datarecords
+
+    calls = []
+
+    class FakeResponse:
+        def raise_for_status(self):
+            pass
+
+        def json(self):
+            return {'result': [{'key': 'subject::foo'}]}
+
+    def fake_get(url, **kwargs):
+        calls.append(url)
+        return FakeResponse()
+
+    monkeypatch.setattr(datarecords.requests, 'get', fake_get)
+
+    client = make_client(tmp_path)
+    for _ in range(2):
+        resp = client.get('/api/datarecords/subject')
+        assert resp.status_code == 200
+        assert resp.json == {'result': [{'key': 'subject::foo'}]}
+        assert resp.headers['Cache-Control'] == f'max-age={datarecords.TIMEOUT}'
+
+    # Second call is served from the cache
+    assert calls == [f'{datarecords.DATARECORDS_URL}/subject']
+
+
+def test_datarecords_upstream_failure(tmp_path, monkeypatch):
+    from budgetkey_api.modules import datarecords
+
+    def fake_get(url, **kwargs):
+        raise ConnectionError('boom')
+
+    monkeypatch.setattr(datarecords.requests, 'get', fake_get)
+
+    client = make_client(tmp_path)
+    resp = client.get('/api/datarecords/intervention')
+    assert resp.status_code == 502