Add E2E tests to aggregate analysis endpoint

ooniapi/services/oonimeasurements/src/oonimeasurements/routers/data/aggregate_analysis.py

@@ -5,7 +5,7 @@
 from fastapi import APIRouter, Depends, Query
 from pydantic import BaseModel
 
-from .utils import get_measurement_start_day_agg, TimeGrains
+from .utils import get_measurement_start_day_agg, TimeGrains, parse_probe_asn
 from ...dependencies import (
     get_clickhouse_session,
 )
@@ -30,6 +30,7 @@
     "probe_cc",
     "probe_asn",
     "test_name",
+    "input"
 ]
 
 
@@ -47,13 +48,15 @@ class AggregationEntry(BaseModel):
     ok_count: float
     measurement_count: float
 
-    measurement_start_day: date
+    measurement_start_day: Optional[datetime] = None
     outcome_label: str
     outcome_value: float
 
     domain: Optional[str] = None
     probe_cc: Optional[str] = None
     probe_asn: Optional[int] = None
+    test_name: Optional[str] = None
+    input: Optional[str] = None
 
 
 class AggregationResponse(BaseModel):
@@ -67,7 +70,6 @@ class AggregationResponse(BaseModel):
 async def get_aggregation_analysis(
     axis_x: Annotated[AggregationKeys, Query()] = "measurement_start_day",
     axis_y: Annotated[Optional[AggregationKeys], Query()] = None,
-    category_code: Annotated[Optional[str], Query()] = None,
     test_name: Annotated[Optional[str], Query()] = None,
     domain: Annotated[Optional[str], Query()] = None,
     input: Annotated[Optional[str], Query()] = None,
@@ -97,8 +99,7 @@ async def get_aggregation_analysis(
         extra_cols[axis_x] = axis_x
 
     if probe_asn is not None:
-        if isinstance(probe_asn, str) and probe_asn.startswith("AS"):
-            probe_asn = int(probe_asn[2:])
+        probe_asn = parse_probe_asn(probe_asn)
         q_args["probe_asn"] = probe_asn
         and_clauses.append("probe_asn = %(probe_asn)d")
         extra_cols["probe_asn"] = "probe_asn"
@@ -259,7 +260,6 @@ async def get_aggregation_analysis(
     results: List[AggregationEntry] = []
     if rows and isinstance(rows, list):
         for row in rows:
-            print(row)
             d = dict(zip(list(extra_cols.keys()) + fixed_cols, row))
             outcome_value = d["outcome_value"]
             outcome_label = d["outcome_label"]
@@ -285,12 +285,14 @@ async def get_aggregation_analysis(
                 failure_count=failure_count,
                 ok_count=ok_count,
                 measurement_count=1.0,
-                measurement_start_day=d["measurement_start_day"],
+                measurement_start_day=d.get("measurement_start_day"),
                 outcome_label=outcome_label,
                 outcome_value=outcome_value,
                 domain=d.get("domain"),
                 probe_cc=d.get("probe_cc"),
                 probe_asn=d.get("probe_asn"),
+                test_name=d.get("test_name"),
+                input=d.get("input"),
             )
             results.append(entry)
     return AggregationResponse(

ooniapi/services/oonimeasurements/tests/test_oonidata_aggregate_analysis.py

@@ -0,0 +1,142 @@
+import pytest
+
+route = "api/v1/aggregation/analysis"
+since = "2024-11-01"
+until = "2024-11-10"
+
+
+def test_oonidata_aggregation_analysis(client):
+    response = client.get(route)
+
+    json = response.json()
+    assert isinstance(json["results"], list), json
+    assert len(json["results"]) == 0
+
+
+def test_oonidata_aggregation_analysis_with_since_and_until(client):
+    response = client.get(f"{route}?since={since}&until={until}")
+
+    json = response.json()
+    assert isinstance(json["results"], list), json
+    assert len(json["results"]) > 0
+
+    for result in json["results"]:
+        assert "anomaly_count" in result, result
+        assert "domain" in result, result
+
+
+@pytest.mark.parametrize(
+    "filter_param, filter_value",
+    [
+        ("domain", "zh.wikipedia.org"),
+        ("probe_cc", "IR"),
+        ("probe_asn", 45758),
+        ("test_name", "whatsapp"),
+        ("input", "stun://stun.voys.nl:3478"),
+    ]
+)
+def test_oonidata_aggregation_analysis_with_filters(client, filter_param, filter_value):
+    params = {
+        "since": since,
+        "until": until
+    }
+    params[filter_param] = filter_value
+
+    response = client.get(route, params=params)
+
+    json = response.json()
+    assert isinstance(json["results"], list), json
+    assert len(json["results"]) > 0
+    for result in json["results"]:
+        assert result[filter_param] == filter_value, result
+
+
+def test_oonidata_aggregation_analysis_filtering_by_probe_asn_as_a_string_with_since_and_until(client):
+    probe_asn = 45758
+    probe_asn_as_a_string =  "AS" + str(probe_asn)
+
+    response = client.get(f"{route}?probe_asn={probe_asn_as_a_string}&since={since}&until={until}")
+
+    json = response.json()
+    assert isinstance(json["results"], list), json
+    assert len(json["results"]) > 0
+    for result in json["results"]:
+        assert result["probe_asn"] == probe_asn, result
+
+
+@pytest.mark.parametrize(
+    "field", [
+        "measurement_start_day",
+        "domain",
+        "probe_cc",
+        "probe_asn",
+        "test_name",
+        "input",
+    ]
+)
+def test_oonidata_aggregation_analysis_with_axis_x(client, field):
+    params = {
+        "since": since,
+        "until": until,
+        "axis_x": field
+    }
+
+    response = client.get(route, params=params)
+
+    json = response.json()
+    assert isinstance(json["results"], list), json
+    assert len(json["results"]) > 0
+    for result in json["results"]:
+        assert result[field] is not None, result
+
+
+@pytest.mark.parametrize(
+    "field", [
+        "measurement_start_day",
+        "domain",
+        "probe_cc",
+        "probe_asn",
+        "test_name",
+        "input",
+    ]
+)
+def test_oonidata_aggregation_analysis_axis_y(client, field):
+    params = {
+        "since": since,
+        "until": until,
+        "axis_y": field
+    }
+
+    response = client.get(route, params=params)
+
+    json = response.json()
+    assert isinstance(json["results"], list), json
+    assert len(json["results"]) > 0
+    for result in json["results"]:
+        assert result[field] is not None, result
+
+
+@pytest.mark.parametrize(
+    "time_grain, total",
+    [
+        ("hour", 216),
+        ("day", 9),
+        ("week", 2),
+        ("month", 1),
+        ("year", 1),
+        ("auto", 9),
+    ]
+)
+def test_oonidata_aggregation_analysis_time_grain(client, time_grain, total):
+    params = {
+        "since": since,
+        "until": until,
+        "time_grain": time_grain
+    }
+
+    response = client.get(route, params=params)
+
+    json = response.json()
+    assert isinstance(json["results"], list), json
+    assert len(json["results"]) == total
+