Path	Lines of Code
services/admin/poetry.toml	3
services/admin/pyproject.toml	58
services/admin/src/admin/__init__.py	1
services/admin/src/admin/app.py	184
services/admin/src/admin/authentication.py	68
services/admin/src/admin/config.py	90
services/admin/src/admin/main.py	3
services/admin/src/admin/routes/__init__.py	1
services/admin/src/admin/routes/blocked_datasets.py	32
services/admin/src/admin/routes/cache_reports.py	47
services/admin/src/admin/routes/cache_reports_with_content.py	51
services/admin/src/admin/routes/dataset_status.py	47
services/admin/src/admin/routes/force_refresh.py	104
services/admin/src/admin/routes/healthcheck.py	6
services/admin/src/admin/routes/metrics.py	25
services/admin/src/admin/routes/num_dataset_infos_by_builder_name.py	77
services/admin/src/admin/routes/pending_jobs.py	37
services/admin/src/admin/routes/recreate_dataset.py	74
services/api/poetry.toml	3
services/api/pyproject.toml	58
services/api/src/api/__init__.py	1
services/api/src/api/app.py	104
services/api/src/api/config.py	81
services/api/src/api/main.py	3
services/api/src/api/routes/__init__.py	1
services/api/src/api/routes/endpoint.py	187
services/reverse-proxy/nginx-templates/404.html	5
services/rows/poetry.toml	3
services/rows/pyproject.toml	59
services/rows/src/rows/__init__.py	1
services/rows/src/rows/app.py	105
services/rows/src/rows/config.py	43
services/rows/src/rows/main.py	3
services/rows/src/rows/routes/__init__.py	1
services/rows/src/rows/routes/rows.py	139
services/search/poetry.toml	3
services/search/pyproject.toml	55
services/search/src/search/__init__.py	1
services/search/src/search/app.py	133
services/search/src/search/config.py	69
services/search/src/search/duckdb_connection.py	15
services/search/src/search/main.py	3
services/search/src/search/routes/__init__.py	1
services/search/src/search/routes/filter.py	195
services/search/src/search/routes/search.py	225
services/sse-api/poetry.toml	3
services/sse-api/pyproject.toml	61
services/sse-api/src/sse_api/__init__.py	1
services/sse-api/src/sse_api/app.py	66
services/sse-api/src/sse_api/config.py	25
services/sse-api/src/sse_api/constants.py	1
services/sse-api/src/sse_api/main.py	3
services/sse-api/src/sse_api/routes/hub_cache.py	44
services/sse-api/src/sse_api/watcher.py	140
services/webhook/poetry.toml	3
services/webhook/pyproject.toml	57
services/webhook/src/webhook/__init__.py	1
services/webhook/src/webhook/app.py	87
services/webhook/src/webhook/config.py	40
services/webhook/src/webhook/main.py	3
services/webhook/src/webhook/routes/__init__.py	1
services/webhook/src/webhook/routes/webhook.py	182
services/worker/poetry.toml	3
services/worker/pyproject.toml	86
services/worker/src/worker/__init__.py	2
services/worker/src/worker/config.py	297
services/worker/src/worker/dtos.py	248
services/worker/src/worker/executor.py	191
services/worker/src/worker/job_manager.py	206
services/worker/src/worker/job_runner.py	23
services/worker/src/worker/job_runner_factory.py	215
services/worker/src/worker/job_runners/__init__.py	1
services/worker/src/worker/job_runners/_job_runner_with_cache.py	49
services/worker/src/worker/job_runners/_job_runner_with_datasets_cache.py	44
services/worker/src/worker/job_runners/config/__init__.py	1
services/worker/src/worker/job_runners/config/config_job_runner.py	40
services/worker/src/worker/job_runners/config/info.py	29
services/worker/src/worker/job_runners/config/is_valid.py	62
services/worker/src/worker/job_runners/config/opt_in_out_urls_count.py	67
services/worker/src/worker/job_runners/config/parquet.py	29
services/worker/src/worker/job_runners/config/parquet_and_info.py	1039
services/worker/src/worker/job_runners/config/parquet_metadata.py	123
services/worker/src/worker/job_runners/config/size.py	82
services/worker/src/worker/job_runners/config/split_names.py	91
services/worker/src/worker/job_runners/dataset/__init__.py	1
services/worker/src/worker/job_runners/dataset/compatible_libraries.py	646
services/worker/src/worker/job_runners/dataset/config_names.py	81
services/worker/src/worker/job_runners/dataset/croissant_crumbs.py	198
services/worker/src/worker/job_runners/dataset/dataset_job_runner.py	41
services/worker/src/worker/job_runners/dataset/filetypes.py	80
services/worker/src/worker/job_runners/dataset/hub_cache.py	114
services/worker/src/worker/job_runners/dataset/info.py	62
services/worker/src/worker/job_runners/dataset/is_valid.py	62
services/worker/src/worker/job_runners/dataset/modalities.py	264
services/worker/src/worker/job_runners/dataset/opt_in_out_urls_count.py	74
services/worker/src/worker/job_runners/dataset/parquet.py	80
services/worker/src/worker/job_runners/dataset/presidio_entities_count.py	86
services/worker/src/worker/job_runners/dataset/size.py	113
services/worker/src/worker/job_runners/dataset/split_names.py	75
services/worker/src/worker/job_runners/split/__init__.py	1
services/worker/src/worker/job_runners/split/descriptive_statistics.py	251
services/worker/src/worker/job_runners/split/first_rows.py	225
services/worker/src/worker/job_runners/split/image_url_columns.py	72
services/worker/src/worker/job_runners/split/is_valid.py	68
services/worker/src/worker/job_runners/split/opt_in_out_urls_count.py	40
services/worker/src/worker/job_runners/split/opt_in_out_urls_scan_from_streaming.py	208
services/worker/src/worker/job_runners/split/presidio_scan.py	579
services/worker/src/worker/job_runners/split/split_job_runner.py	59
services/worker/src/worker/loop.py	98
services/worker/src/worker/main.py	59
services/worker/src/worker/resources.py	31
services/worker/src/worker/routes/__init__.py	1
services/worker/src/worker/routes/healthcheck.py	6
services/worker/src/worker/routes/metrics.py	16
services/worker/src/worker/start_web_app.py	22
services/worker/src/worker/start_worker_loop.py	59
services/worker/src/worker/utils.py	233