KSvend Claude Happy commited on
Commit ·
f491e48
1
Parent(s): bf99ddb
feat: three-phase batch job worker (submit → poll → harvest)
Browse filesGenerated with [Claude Code](https://claude.ai/code)
via [Happy](https://happy.engineering)
Co-Authored-By: Claude <noreply@anthropic.com>
Co-Authored-By: Happy <yesreply@happy.engineering>
- app/worker.py +90 -8
- tests/test_worker.py +110 -0
app/worker.py
CHANGED
|
@@ -3,6 +3,7 @@ import asyncio
|
|
| 3 |
import json
|
| 4 |
import logging
|
| 5 |
import os
|
|
|
|
| 6 |
import traceback
|
| 7 |
from app.database import Database
|
| 8 |
from app.indicators.base import IndicatorRegistry
|
|
@@ -17,6 +18,9 @@ from app.core.email import send_completion_email
|
|
| 17 |
|
| 18 |
logger = logging.getLogger(__name__)
|
| 19 |
|
|
|
|
|
|
|
|
|
|
| 20 |
|
| 21 |
def _save_spatial_json(spatial, status_value: str, path: str) -> None:
|
| 22 |
"""Serialize spatial data to JSON for the frontend."""
|
|
@@ -51,19 +55,97 @@ async def process_job(job_id: str, db: Database, registry: IndicatorRegistry) ->
|
|
| 51 |
return
|
| 52 |
await db.update_job_status(job_id, JobStatus.PROCESSING)
|
| 53 |
try:
|
| 54 |
-
# Track spatial data per indicator for map generation
|
| 55 |
spatial_cache = {}
|
| 56 |
|
|
|
|
|
|
|
|
|
|
| 57 |
for indicator_id in job.request.indicator_ids:
|
| 58 |
-
await db.update_job_progress(job_id, indicator_id, "processing")
|
| 59 |
indicator = registry.get(indicator_id)
|
| 60 |
-
|
| 61 |
-
|
| 62 |
-
|
| 63 |
-
|
| 64 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 65 |
|
| 66 |
-
# Capture spatial data before it's lost
|
| 67 |
spatial = indicator.get_spatial_data()
|
| 68 |
if spatial is not None:
|
| 69 |
spatial_cache[indicator_id] = spatial
|
|
|
|
| 3 |
import json
|
| 4 |
import logging
|
| 5 |
import os
|
| 6 |
+
import time
|
| 7 |
import traceback
|
| 8 |
from app.database import Database
|
| 9 |
from app.indicators.base import IndicatorRegistry
|
|
|
|
| 18 |
|
| 19 |
logger = logging.getLogger(__name__)
|
| 20 |
|
| 21 |
+
BATCH_POLL_INTERVAL = 30 # seconds between status checks
|
| 22 |
+
BATCH_TIMEOUT = 1200 # 20 minutes maximum wait
|
| 23 |
+
|
| 24 |
|
| 25 |
def _save_spatial_json(spatial, status_value: str, path: str) -> None:
|
| 26 |
"""Serialize spatial data to JSON for the frontend."""
|
|
|
|
| 55 |
return
|
| 56 |
await db.update_job_status(job_id, JobStatus.PROCESSING)
|
| 57 |
try:
|
|
|
|
| 58 |
spatial_cache = {}
|
| 59 |
|
| 60 |
+
# Separate batch vs non-batch indicators
|
| 61 |
+
batch_indicators = {}
|
| 62 |
+
process_indicators = []
|
| 63 |
for indicator_id in job.request.indicator_ids:
|
|
|
|
| 64 |
indicator = registry.get(indicator_id)
|
| 65 |
+
if indicator.uses_batch:
|
| 66 |
+
batch_indicators[indicator_id] = indicator
|
| 67 |
+
else:
|
| 68 |
+
process_indicators.append((indicator_id, indicator))
|
| 69 |
+
|
| 70 |
+
# -- Phase 1: Submit batch jobs --
|
| 71 |
+
batch_submissions = {}
|
| 72 |
+
fallback_ids = set()
|
| 73 |
+
for indicator_id, indicator in batch_indicators.items():
|
| 74 |
+
await db.update_job_progress(job_id, indicator_id, "submitting")
|
| 75 |
+
try:
|
| 76 |
+
jobs = await indicator.submit_batch(
|
| 77 |
+
job.request.aoi,
|
| 78 |
+
job.request.time_range,
|
| 79 |
+
season_months=job.request.season_months(),
|
| 80 |
+
)
|
| 81 |
+
batch_submissions[indicator_id] = jobs
|
| 82 |
+
await db.update_job_progress(job_id, indicator_id, "processing on CDSE")
|
| 83 |
+
except Exception as exc:
|
| 84 |
+
logger.warning("Batch submit failed for %s, will use fallback: %s", indicator_id, exc)
|
| 85 |
+
fallback_ids.add(indicator_id)
|
| 86 |
+
|
| 87 |
+
# -- Phase 2: Poll until all batch jobs finish --
|
| 88 |
+
poll_start = time.monotonic()
|
| 89 |
+
pending = dict(batch_submissions)
|
| 90 |
+
|
| 91 |
+
while pending:
|
| 92 |
+
# Check current statuses before sleeping
|
| 93 |
+
for indicator_id in list(pending.keys()):
|
| 94 |
+
jobs = pending[indicator_id]
|
| 95 |
+
statuses = [j.status() for j in jobs]
|
| 96 |
+
if all(s == "finished" for s in statuses):
|
| 97 |
+
logger.info("Batch jobs finished for %s", indicator_id)
|
| 98 |
+
del pending[indicator_id]
|
| 99 |
+
elif any(s in ("error", "canceled") for s in statuses):
|
| 100 |
+
logger.warning("Batch job failed for %s: %s", indicator_id, statuses)
|
| 101 |
+
del pending[indicator_id]
|
| 102 |
+
|
| 103 |
+
if not pending:
|
| 104 |
+
break
|
| 105 |
+
|
| 106 |
+
elapsed = time.monotonic() - poll_start
|
| 107 |
+
if elapsed >= BATCH_TIMEOUT:
|
| 108 |
+
logger.warning("Batch poll timeout after %.0fs, remaining: %s", elapsed, list(pending.keys()))
|
| 109 |
+
fallback_ids.update(pending.keys())
|
| 110 |
+
break
|
| 111 |
+
|
| 112 |
+
await asyncio.sleep(BATCH_POLL_INTERVAL)
|
| 113 |
+
|
| 114 |
+
# -- Phase 3: Harvest batch results + process non-batch indicators --
|
| 115 |
+
for indicator_id in job.request.indicator_ids:
|
| 116 |
+
indicator = registry.get(indicator_id)
|
| 117 |
+
|
| 118 |
+
if indicator_id in fallback_ids:
|
| 119 |
+
await db.update_job_progress(job_id, indicator_id, "processing")
|
| 120 |
+
result = await indicator.process(
|
| 121 |
+
job.request.aoi,
|
| 122 |
+
job.request.time_range,
|
| 123 |
+
season_months=job.request.season_months(),
|
| 124 |
+
)
|
| 125 |
+
elif indicator_id in batch_submissions:
|
| 126 |
+
await db.update_job_progress(job_id, indicator_id, "downloading")
|
| 127 |
+
try:
|
| 128 |
+
result = await indicator.harvest(
|
| 129 |
+
job.request.aoi,
|
| 130 |
+
job.request.time_range,
|
| 131 |
+
season_months=job.request.season_months(),
|
| 132 |
+
batch_jobs=batch_submissions[indicator_id],
|
| 133 |
+
)
|
| 134 |
+
except Exception as exc:
|
| 135 |
+
logger.warning("Harvest failed for %s, using fallback: %s", indicator_id, exc)
|
| 136 |
+
result = await indicator.process(
|
| 137 |
+
job.request.aoi,
|
| 138 |
+
job.request.time_range,
|
| 139 |
+
season_months=job.request.season_months(),
|
| 140 |
+
)
|
| 141 |
+
else:
|
| 142 |
+
await db.update_job_progress(job_id, indicator_id, "processing")
|
| 143 |
+
result = await indicator.process(
|
| 144 |
+
job.request.aoi,
|
| 145 |
+
job.request.time_range,
|
| 146 |
+
season_months=job.request.season_months(),
|
| 147 |
+
)
|
| 148 |
|
|
|
|
| 149 |
spatial = indicator.get_spatial_data()
|
| 150 |
if spatial is not None:
|
| 151 |
spatial_cache[indicator_id] = spatial
|
tests/test_worker.py
CHANGED
|
@@ -1,5 +1,6 @@
|
|
| 1 |
import pytest
|
| 2 |
from datetime import date
|
|
|
|
| 3 |
from app.worker import process_job
|
| 4 |
from app.database import Database
|
| 5 |
from app.models import JobStatus, AOI, TimeRange, JobRequest, IndicatorResult, StatusLevel, TrendDirection, ConfidenceLevel
|
|
@@ -66,3 +67,112 @@ async def test_process_job_handles_unknown_indicator(temp_db_path):
|
|
| 66 |
job = await db.get_job(job_id)
|
| 67 |
assert job.status == JobStatus.FAILED
|
| 68 |
assert "nonexistent" in job.error
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
import pytest
|
| 2 |
from datetime import date
|
| 3 |
+
from unittest.mock import MagicMock
|
| 4 |
from app.worker import process_job
|
| 5 |
from app.database import Database
|
| 6 |
from app.models import JobStatus, AOI, TimeRange, JobRequest, IndicatorResult, StatusLevel, TrendDirection, ConfidenceLevel
|
|
|
|
| 67 |
job = await db.get_job(job_id)
|
| 68 |
assert job.status == JobStatus.FAILED
|
| 69 |
assert "nonexistent" in job.error
|
| 70 |
+
|
| 71 |
+
|
| 72 |
+
class MockBatchIndicator(BaseIndicator):
|
| 73 |
+
"""Batch indicator for testing the three-phase worker."""
|
| 74 |
+
id = "ndvi"
|
| 75 |
+
name = "Vegetation (NDVI)"
|
| 76 |
+
category = "D2"
|
| 77 |
+
question = "Is vegetation cover declining?"
|
| 78 |
+
estimated_minutes = 8
|
| 79 |
+
uses_batch = True
|
| 80 |
+
|
| 81 |
+
async def process(self, aoi, time_range, season_months=None):
|
| 82 |
+
return IndicatorResult(
|
| 83 |
+
indicator_id="ndvi", headline="placeholder",
|
| 84 |
+
status=StatusLevel.GREEN, trend=TrendDirection.STABLE,
|
| 85 |
+
confidence=ConfidenceLevel.LOW, map_layer_path="",
|
| 86 |
+
chart_data={"dates": ["2025"], "values": [0.3], "label": "NDVI"},
|
| 87 |
+
data_source="placeholder",
|
| 88 |
+
summary="Fallback.", methodology="Placeholder.", limitations=[],
|
| 89 |
+
)
|
| 90 |
+
|
| 91 |
+
async def submit_batch(self, aoi, time_range, season_months=None):
|
| 92 |
+
mock_job = MagicMock()
|
| 93 |
+
mock_job.job_id = "j-test"
|
| 94 |
+
mock_job.status.return_value = "finished"
|
| 95 |
+
return [mock_job, mock_job, mock_job]
|
| 96 |
+
|
| 97 |
+
async def harvest(self, aoi, time_range, season_months=None, batch_jobs=None):
|
| 98 |
+
return IndicatorResult(
|
| 99 |
+
indicator_id="ndvi", headline="Real NDVI data",
|
| 100 |
+
status=StatusLevel.GREEN, trend=TrendDirection.STABLE,
|
| 101 |
+
confidence=ConfidenceLevel.HIGH, map_layer_path="",
|
| 102 |
+
chart_data={"dates": ["2025-01"], "values": [0.45], "label": "NDVI"},
|
| 103 |
+
data_source="satellite",
|
| 104 |
+
summary="Real.", methodology="Sentinel-2.", limitations=[],
|
| 105 |
+
)
|
| 106 |
+
|
| 107 |
+
|
| 108 |
+
@pytest.mark.asyncio
|
| 109 |
+
async def test_process_job_uses_batch_flow(temp_db_path):
|
| 110 |
+
"""Worker uses submit_batch -> poll -> harvest for batch indicators."""
|
| 111 |
+
db = Database(temp_db_path)
|
| 112 |
+
await db.init()
|
| 113 |
+
reg = IndicatorRegistry()
|
| 114 |
+
reg.register(MockBatchIndicator())
|
| 115 |
+
request = JobRequest(
|
| 116 |
+
aoi=AOI(name="Test", bbox=[32.45, 15.65, 32.65, 15.80]),
|
| 117 |
+
time_range=TimeRange(start=date(2025, 3, 1), end=date(2026, 3, 1)),
|
| 118 |
+
indicator_ids=["ndvi"],
|
| 119 |
+
email="test@example.com",
|
| 120 |
+
)
|
| 121 |
+
job_id = await db.create_job(request)
|
| 122 |
+
await process_job(job_id, db, reg)
|
| 123 |
+
job = await db.get_job(job_id)
|
| 124 |
+
assert job.status == JobStatus.COMPLETE
|
| 125 |
+
assert len(job.results) == 1
|
| 126 |
+
assert job.results[0].data_source == "satellite"
|
| 127 |
+
assert job.results[0].headline == "Real NDVI data"
|
| 128 |
+
|
| 129 |
+
|
| 130 |
+
@pytest.mark.asyncio
|
| 131 |
+
async def test_process_job_mixes_batch_and_process(temp_db_path):
|
| 132 |
+
"""Worker handles batch and non-batch indicators in the same job."""
|
| 133 |
+
db = Database(temp_db_path)
|
| 134 |
+
await db.init()
|
| 135 |
+
reg = IndicatorRegistry()
|
| 136 |
+
reg.register(MockBatchIndicator())
|
| 137 |
+
reg.register(MockFiresIndicator())
|
| 138 |
+
request = JobRequest(
|
| 139 |
+
aoi=AOI(name="Test", bbox=[32.45, 15.65, 32.65, 15.80]),
|
| 140 |
+
time_range=TimeRange(start=date(2025, 3, 1), end=date(2026, 3, 1)),
|
| 141 |
+
indicator_ids=["ndvi", "fires"],
|
| 142 |
+
email="test@example.com",
|
| 143 |
+
)
|
| 144 |
+
job_id = await db.create_job(request)
|
| 145 |
+
await process_job(job_id, db, reg)
|
| 146 |
+
job = await db.get_job(job_id)
|
| 147 |
+
assert job.status == JobStatus.COMPLETE
|
| 148 |
+
assert len(job.results) == 2
|
| 149 |
+
|
| 150 |
+
ndvi_result = next(r for r in job.results if r.indicator_id == "ndvi")
|
| 151 |
+
fires_result = next(r for r in job.results if r.indicator_id == "fires")
|
| 152 |
+
assert ndvi_result.data_source == "satellite"
|
| 153 |
+
assert fires_result.headline == "3 fire events detected"
|
| 154 |
+
|
| 155 |
+
|
| 156 |
+
@pytest.mark.asyncio
|
| 157 |
+
async def test_process_job_batch_submit_failure_falls_back(temp_db_path):
|
| 158 |
+
"""If submit_batch() fails, worker falls back to process()."""
|
| 159 |
+
|
| 160 |
+
class FailingBatchIndicator(MockBatchIndicator):
|
| 161 |
+
async def submit_batch(self, aoi, time_range, season_months=None):
|
| 162 |
+
raise ConnectionError("CDSE unreachable")
|
| 163 |
+
|
| 164 |
+
db = Database(temp_db_path)
|
| 165 |
+
await db.init()
|
| 166 |
+
reg = IndicatorRegistry()
|
| 167 |
+
reg.register(FailingBatchIndicator())
|
| 168 |
+
request = JobRequest(
|
| 169 |
+
aoi=AOI(name="Test", bbox=[32.45, 15.65, 32.65, 15.80]),
|
| 170 |
+
time_range=TimeRange(start=date(2025, 3, 1), end=date(2026, 3, 1)),
|
| 171 |
+
indicator_ids=["ndvi"],
|
| 172 |
+
email="test@example.com",
|
| 173 |
+
)
|
| 174 |
+
job_id = await db.create_job(request)
|
| 175 |
+
await process_job(job_id, db, reg)
|
| 176 |
+
job = await db.get_job(job_id)
|
| 177 |
+
assert job.status == JobStatus.COMPLETE
|
| 178 |
+
assert job.results[0].data_source == "placeholder"
|