Datasets:

do-me
/

EUR-LEX

Error code:   InfoError
Exception:    HfHubHTTPError
Message:      504 Server Error: Gateway Time-out for url: https://huggingface.co/api/datasets/do-me/EUR-LEX/tree/2bbef3ff0f05687e125ac26f2e5ab5e7bf1b2d96/files?expand=false&recursive=true&limit=1000&cursor=ZXlKbWFXeGxYMjVoYldVaU9pSm1hV3hsY3k4eU1ERXhMMlJoZEdGelpYUmZNakF4TVMwd015MHlNVjlsYm1jdWNHRnljWFZsZENJc0luUnlaV1ZmYjJsa0lqb2lOMkU0WkdJeVptVTROMlkzT0RVMU9XUTNOVFl3WVRBME1qUmxZMk0wT1RWbE1ERmtNRFV6TlNKOToxNDAwMA%3D%3D
Traceback:    Traceback (most recent call last):
                File "/src/services/worker/src/worker/job_runners/split/first_rows.py", line 223, in compute_first_rows_from_streaming_response
                  info = get_dataset_config_info(path=dataset, config_name=config, token=hf_token)
                         ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.12/site-packages/datasets/inspect.py", line 268, in get_dataset_config_info
                  builder = load_dataset_builder(
                            ^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.12/site-packages/datasets/load.py", line 1315, in load_dataset_builder
                  dataset_module = dataset_module_factory(
                                   ^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.12/site-packages/datasets/load.py", line 1207, in dataset_module_factory
                  raise e1 from None
                File "/usr/local/lib/python3.12/site-packages/datasets/load.py", line 1182, in dataset_module_factory
                  ).get_module()
                    ^^^^^^^^^^^^
                File "/usr/local/lib/python3.12/site-packages/datasets/load.py", line 638, in get_module
                  patterns = get_data_patterns(base_path, download_config=self.download_config)
                             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.12/site-packages/datasets/data_files.py", line 493, in get_data_patterns
                  return _get_data_files_patterns(resolver)
                         ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.12/site-packages/datasets/data_files.py", line 290, in _get_data_files_patterns
                  data_files = pattern_resolver(pattern)
                               ^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.12/site-packages/datasets/data_files.py", line 372, in resolve_pattern
                  for filepath, info in fs.glob(fs_pattern, detail=True, **glob_kwargs).items():
                                        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.12/site-packages/huggingface_hub/hf_file_system.py", line 521, in glob
                  return super().glob(path, **kwargs)
                         ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.12/site-packages/fsspec/spec.py", line 604, in glob
                  allpaths = self.find(root, maxdepth=depth, withdirs=True, detail=True, **kwargs)
                             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.12/site-packages/huggingface_hub/hf_file_system.py", line 563, in find
                  out = self._ls_tree(path, recursive=True, refresh=refresh, revision=resolved_path.revision, **kwargs)
                        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.12/site-packages/huggingface_hub/hf_file_system.py", line 446, in _ls_tree
                  self._ls_tree(
                File "/usr/local/lib/python3.12/site-packages/huggingface_hub/hf_file_system.py", line 463, in _ls_tree
                  for path_info in tree:
                                   ^^^^
                File "/usr/local/lib/python3.12/site-packages/huggingface_hub/hf_api.py", line 3140, in list_repo_tree
                  for path_info in paginate(path=tree_url, headers=headers, params={"recursive": recursive, "expand": expand}):
                                   ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.12/site-packages/huggingface_hub/utils/_pagination.py", line 46, in paginate
                  hf_raise_for_status(r)
                File "/usr/local/lib/python3.12/site-packages/huggingface_hub/utils/_http.py", line 482, in hf_raise_for_status
                  raise _format(HfHubHTTPError, str(e), response) from e
              huggingface_hub.errors.HfHubHTTPError: 504 Server Error: Gateway Time-out for url: https://huggingface.co/api/datasets/do-me/EUR-LEX/tree/2bbef3ff0f05687e125ac26f2e5ab5e7bf1b2d96/files?expand=false&recursive=true&limit=1000&cursor=ZXlKbWFXeGxYMjVoYldVaU9pSm1hV3hsY3k4eU1ERXhMMlJoZEdGelpYUmZNakF4TVMwd015MHlNVjlsYm1jdWNHRnljWFZsZENJc0luUnlaV1ZmYjJsa0lqb2lOMkU0WkdJeVptVTROMlkzT0RVMU9XUTNOVFl3WVRBME1qUmxZMk0wT1RWbE1ERmtNRFV6TlNKOToxNDAwMA%3D%3D

Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.

EUR-LEX Miner 🇪🇺

A high-performance mining tool on GitHub (https://github.com/do-me/eur-lex) for extracting text and semantic concepts from the European Commission's Cellar database.

Originally based on EuropeanParliament/Eurovoc_2025 and EuropeanParliament/Eurovoc but improved and extended for various reasons:

line breaks are not included that are quite important for semantic chunking approaches and RAG
the preprocessing was way too aggressive and removed too much information
these repos were abandoned in the past
Updates are infrequent
Most of the metadata was not included

This repo improves on all of these points and provides a convenient github action syncing with Huggingface for weekly updates.

Features

Fast SPARQL Retrieval: Custom J2-templated queries for efficient metadata fetching.
Parallel Parsing: Multi-threaded extraction from PDF, DOCX, and HTML.
Vectorized Preprocessing: Ultra-fast text cleaning powered by Polars.
Modular Design: Clean separation of concerns following modern Python package standards.
Robust Caching: Joblib-powered caching to avoid redundant downloads and expensive parsing.
Batch Discovery: Tweak network efficiency with the --days-per-request flag to fetch metadata for multiple days in one SPARQL call.
Rich Metadata Collection: Automatically extracts:
- CELEX numbers for legal uniquely indexing.
- Institution/Author labels (e.g., "European Commission").
- Procedure IDs for legislative tracking.
- Document types and Eurovoc concept IDs.

Full-Text Search

You can use the script in this GitHub repo to run full-text search in roughly 10-20s on consumer hardware after downloading the whole 6.7Gb dump with 328k rows: https://github.com/do-me/copernicus-climate-eur-lex-viz For instance you can analyze the growth of Copernicus- and climate-related records in the past years:

Installation

Ensure you have uv installed, then:

git clone <repo-url>
cd Eurovoc_2025
uv sync

Usage

Run the miner using the CLI entry point:

# General usage for all languages
uv run eur_lex_miner dataset_ --days 10

# Specific language filtering
uv run eur_lex_miner dataset_ --days 5 --lang ENG

# Keyword Matching
# Add boolean columns for specific terms (case-insensitive)
uv run eur_lex_miner dataset_ --days 5 --lang ENG --keywords "earth observation" copernicus

# Batch Discovery
# Fetch 60 days of metadata in a single SPARQL request (faster for large spans)
uv run eur_lex_miner dataset_history --days 60 --days-per-request 60

# Data Engineering Best Practices
# Use --lookback for a safety margin and --unique-on to deduplicate by ID
uv run eur_lex_miner dataset_ --lookback 14 --unique-on celex --lang ENG

Large batches might give you a slight performance boost, but probably irrelevant for most use cases. Comparison:

Past 30 days, 30 requests (40.079s) vs 1 request (38.349s)

(base) ➜  Eurovoc_2025 git:(main) time uv run eur_lex_miner dataset_history --days 30 --days-per-request 30 --lang ENG
2026-01-28 12:38:37,533 - eurovoc_miner.core - INFO - Processing 333 documents for 2025-12-30 to 2026-01-28
2025-12-30 to 2026-01-28: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 333/333 [00:12<00:00, 26.51it/s]
2026-01-28 12:38:50,320 - eurovoc_miner.cli - INFO - ✓ Saved 303 records to /Users/dome/work/jrc/Eurovoc_2025/files/dataset_history2025-12-30_to_2026-01-28_eng.parquet
uv run eur_lex_miner dataset_history --days 30 --days-per-request 30 --lang   40.44s user 1.68s system 109% cpu 38.349 total
(base) ➜  Eurovoc_2025 git:(main) time uv run eur_lex_miner dataset_ --days 30 --days-per-request 1 --lang ENG 
2026-01-28 12:39:35,265 - eurovoc_miner.core - INFO - Processing 1 documents for 2026-01-28
2026-01-28: 100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 1/1 [00:00<00:00,  6.20it/s]
2026-01-28 12:39:35,463 - eurovoc_miner.cli - INFO - ✓ Saved 1 records to /Users/dome/work/jrc/Eurovoc_2025/files/dataset_2026-01-28_eng.parquet
2026-01-28 12:39:36,307 - eurovoc_miner.core - INFO - Processing 7 documents for 2026-01-27
2026-01-27: 100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 7/7 [00:00<00:00, 37.39it/s]
2026-01-28 12:39:36,562 - eurovoc_miner.cli - INFO - ✓ Saved 7 records to /Users/dome/work/jrc/Eurovoc_2025/files/dataset_2026-01-27_eng.parquet
2026-01-28 12:39:38,917 - eurovoc_miner.core - INFO - Processing 33 documents for 2026-01-26
2026-01-26: 100%|████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 33/33 [00:00<00:00, 145.75it/s]
2026-01-28 12:39:39,305 - eurovoc_miner.cli - INFO - ✓ Saved 33 records to /Users/dome/work/jrc/Eurovoc_2025/files/dataset_2026-01-26_eng.parquet
2026-01-28 12:39:39,699 - eurovoc_miner.cli - INFO - ∅ No documents for 2026-01-25, creating empty file.
2026-01-28 12:39:39,700 - eurovoc_miner.cli - INFO - ✓ Saved empty file to /Users/dome/work/jrc/Eurovoc_2025/files/dataset_2026-01-25_eng.parquet
2026-01-28 12:39:40,177 - eurovoc_miner.cli - INFO - ∅ No documents for 2026-01-24, creating empty file.
2026-01-28 12:39:40,180 - eurovoc_miner.cli - INFO - ✓ Saved empty file to /Users/dome/work/jrc/Eurovoc_2025/files/dataset_2026-01-24_eng.parquet
2026-01-28 12:39:42,279 - eurovoc_miner.core - INFO - Processing 28 documents for 2026-01-23
2026-01-23: 100%|████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 28/28 [00:00<00:00, 118.97it/s]
2026-01-28 12:39:42,664 - eurovoc_miner.cli - INFO - ✓ Saved 23 records to /Users/dome/work/jrc/Eurovoc_2025/files/dataset_2026-01-23_eng.parquet
2026-01-28 12:39:44,256 - eurovoc_miner.core - INFO - Processing 18 documents for 2026-01-22
2026-01-22: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 18/18 [00:00<00:00, 75.58it/s]
2026-01-28 12:39:44,647 - eurovoc_miner.cli - INFO - ✓ Saved 15 records to /Users/dome/work/jrc/Eurovoc_2025/files/dataset_2026-01-22_eng.parquet
2026-01-28 12:39:47,137 - eurovoc_miner.core - INFO - Processing 36 documents for 2026-01-21
2026-01-21: 100%|████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 36/36 [00:00<00:00, 153.10it/s]
2026-01-28 12:39:47,522 - eurovoc_miner.cli - INFO - ✓ Saved 33 records to /Users/dome/work/jrc/Eurovoc_2025/files/dataset_2026-01-21_eng.parquet
2026-01-28 12:39:49,706 - eurovoc_miner.core - INFO - Processing 31 documents for 2026-01-20
2026-01-20: 100%|████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 31/31 [00:00<00:00, 135.42it/s]
2026-01-28 12:39:50,112 - eurovoc_miner.cli - INFO - ✓ Saved 29 records to /Users/dome/work/jrc/Eurovoc_2025/files/dataset_2026-01-20_eng.parquet
2026-01-28 12:39:51,489 - eurovoc_miner.core - INFO - Processing 16 documents for 2026-01-19
2026-01-19: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 16/16 [00:00<00:00, 71.30it/s]
2026-01-28 12:39:51,866 - eurovoc_miner.cli - INFO - ✓ Saved 15 records to /Users/dome/work/jrc/Eurovoc_2025/files/dataset_2026-01-19_eng.parquet
2026-01-28 12:39:52,360 - eurovoc_miner.cli - INFO - ∅ No documents for 2026-01-18, creating empty file.
2026-01-28 12:39:52,361 - eurovoc_miner.cli - INFO - ✓ Saved empty file to /Users/dome/work/jrc/Eurovoc_2025/files/dataset_2026-01-18_eng.parquet
2026-01-28 12:39:52,878 - eurovoc_miner.cli - INFO - ∅ No documents for 2026-01-17, creating empty file.
2026-01-28 12:39:52,879 - eurovoc_miner.cli - INFO - ✓ Saved empty file to /Users/dome/work/jrc/Eurovoc_2025/files/dataset_2026-01-17_eng.parquet
2026-01-28 12:39:54,190 - eurovoc_miner.core - INFO - Processing 15 documents for 2026-01-16
2026-01-16: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 15/15 [00:00<00:00, 69.88it/s]
2026-01-28 12:39:54,552 - eurovoc_miner.cli - INFO - ✓ Saved 13 records to /Users/dome/work/jrc/Eurovoc_2025/files/dataset_2026-01-16_eng.parquet
2026-01-28 12:39:56,940 - eurovoc_miner.core - INFO - Processing 33 documents for 2026-01-15
2026-01-15: 100%|████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 33/33 [00:00<00:00, 135.13it/s]
2026-01-28 12:39:57,350 - eurovoc_miner.cli - INFO - ✓ Saved 29 records to /Users/dome/work/jrc/Eurovoc_2025/files/dataset_2026-01-15_eng.parquet
2026-01-28 12:39:59,039 - eurovoc_miner.core - INFO - Processing 21 documents for 2026-01-14
2026-01-14: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 21/21 [00:00<00:00, 94.03it/s]
2026-01-28 12:39:59,419 - eurovoc_miner.cli - INFO - ✓ Saved 18 records to /Users/dome/work/jrc/Eurovoc_2025/files/dataset_2026-01-14_eng.parquet
2026-01-28 12:40:00,748 - eurovoc_miner.core - INFO - Processing 15 documents for 2026-01-13
2026-01-13: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 15/15 [00:00<00:00, 69.36it/s]
2026-01-28 12:40:01,142 - eurovoc_miner.cli - INFO - ✓ Saved 14 records to /Users/dome/work/jrc/Eurovoc_2025/files/dataset_2026-01-13_eng.parquet
2026-01-28 12:40:03,223 - eurovoc_miner.core - INFO - Processing 24 documents for 2026-01-12
2026-01-12: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 24/24 [00:00<00:00, 76.47it/s]
2026-01-28 12:40:03,763 - eurovoc_miner.cli - INFO - ✓ Saved 23 records to /Users/dome/work/jrc/Eurovoc_2025/files/dataset_2026-01-12_eng.parquet
2026-01-28 12:40:04,431 - eurovoc_miner.cli - INFO - ∅ No documents for 2026-01-11, creating empty file.
2026-01-28 12:40:04,432 - eurovoc_miner.cli - INFO - ✓ Saved empty file to /Users/dome/work/jrc/Eurovoc_2025/files/dataset_2026-01-11_eng.parquet
2026-01-28 12:40:04,957 - eurovoc_miner.cli - INFO - ∅ No documents for 2026-01-10, creating empty file.
2026-01-28 12:40:04,958 - eurovoc_miner.cli - INFO - ✓ Saved empty file to /Users/dome/work/jrc/Eurovoc_2025/files/dataset_2026-01-10_eng.parquet
2026-01-28 12:40:06,618 - eurovoc_miner.core - INFO - Processing 17 documents for 2026-01-09
2026-01-09: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 17/17 [00:00<00:00, 71.38it/s]
2026-01-28 12:40:07,024 - eurovoc_miner.cli - INFO - ✓ Saved 15 records to /Users/dome/work/jrc/Eurovoc_2025/files/dataset_2026-01-09_eng.parquet
2026-01-28 12:40:07,785 - eurovoc_miner.core - INFO - Processing 7 documents for 2026-01-08
2026-01-08: 100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 7/7 [00:00<00:00, 39.67it/s]
2026-01-28 12:40:08,036 - eurovoc_miner.cli - INFO - ✓ Saved 7 records to /Users/dome/work/jrc/Eurovoc_2025/files/dataset_2026-01-08_eng.parquet
2026-01-28 12:40:08,620 - eurovoc_miner.core - INFO - Processing 3 documents for 2026-01-07
2026-01-07: 100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 3/3 [00:00<00:00, 17.43it/s]
2026-01-28 12:40:08,852 - eurovoc_miner.cli - INFO - ✓ Saved 2 records to /Users/dome/work/jrc/Eurovoc_2025/files/dataset_2026-01-07_eng.parquet
2026-01-28 12:40:09,884 - eurovoc_miner.core - INFO - Processing 9 documents for 2026-01-06
2026-01-06: 100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 9/9 [00:00<00:00, 50.34it/s]
2026-01-28 12:40:10,164 - eurovoc_miner.cli - INFO - ✓ Saved 8 records to /Users/dome/work/jrc/Eurovoc_2025/files/dataset_2026-01-06_eng.parquet
2026-01-28 12:40:10,903 - eurovoc_miner.core - INFO - Processing 7 documents for 2026-01-05
2026-01-05: 100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 7/7 [00:00<00:00, 39.85it/s]
2026-01-28 12:40:11,156 - eurovoc_miner.cli - INFO - ✓ Saved 6 records to /Users/dome/work/jrc/Eurovoc_2025/files/dataset_2026-01-05_eng.parquet
2026-01-28 12:40:11,584 - eurovoc_miner.cli - INFO - ∅ No documents for 2026-01-04, creating empty file.
2026-01-28 12:40:11,585 - eurovoc_miner.cli - INFO - ✓ Saved empty file to /Users/dome/work/jrc/Eurovoc_2025/files/dataset_2026-01-04_eng.parquet
2026-01-28 12:40:11,911 - eurovoc_miner.cli - INFO - ∅ No documents for 2026-01-03, creating empty file.
2026-01-28 12:40:11,912 - eurovoc_miner.cli - INFO - ✓ Saved empty file to /Users/dome/work/jrc/Eurovoc_2025/files/dataset_2026-01-03_eng.parquet
2026-01-28 12:40:12,313 - eurovoc_miner.cli - INFO - ∅ No documents for 2026-01-02, creating empty file.
2026-01-28 12:40:12,314 - eurovoc_miner.cli - INFO - ✓ Saved empty file to /Users/dome/work/jrc/Eurovoc_2025/files/dataset_2026-01-02_eng.parquet
2026-01-28 12:40:12,784 - eurovoc_miner.cli - INFO - ∅ No documents for 2026-01-01, creating empty file.
2026-01-28 12:40:12,786 - eurovoc_miner.cli - INFO - ✓ Saved empty file to /Users/dome/work/jrc/Eurovoc_2025/files/dataset_2026-01-01_eng.parquet
2026-01-28 12:40:13,390 - eurovoc_miner.core - INFO - Processing 5 documents for 2025-12-31
2025-12-31: 100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 5/5 [00:00<00:00, 28.90it/s]
2026-01-28 12:40:13,631 - eurovoc_miner.cli - INFO - ✓ Saved 5 records to /Users/dome/work/jrc/Eurovoc_2025/files/dataset_2025-12-31_eng.parquet
2026-01-28 12:40:14,318 - eurovoc_miner.core - INFO - Processing 7 documents for 2025-12-30
2025-12-30: 100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 7/7 [00:00<00:00, 40.57it/s]
2026-01-28 12:40:14,568 - eurovoc_miner.cli - INFO - ✓ Saved 7 records to /Users/dome/work/jrc/Eurovoc_2025/files/dataset_2025-12-30_eng.parquet
uv run eur_lex_miner dataset_ --days 30 --days-per-request 1 --lang ENG  56.75s user 15.49s system 180% cpu 40.079 total

Keyword Matching Logic

When using the --keywords flag:

Compound Terms: Wrap terms with spaces in quotes (e.g., "earth observation").
Auto-Sanitization: Keywords are converted to clean snake_case column names (e.g., match_earth_observation).
Performance: Matching is performed using vectorized Polars operations.
Consistency: Empty daily files still contain the keyword columns to maintain schema across all Parquet files.
Filtering: Use --save-only-keyword-matches to only save records that match at least one of the provided keywords.

The output will be saved as daily .parquet files in the files/ directory.

Project Structure

src/eurovoc_miner/: Core logic and CLI.
files/: Output data storage.
cache/: Internal joblib cache (ignored by git).
tests/: Unit and integration tests. (to do)

Development

To add new parsers, refer to src/eurovoc_miner/parsers.py. To modify the SPARQL logic, see src/eurovoc_miner/fetcher.py.

Downloads last month: 12,506

Size of downloaded dataset files:

6.66 GB

Size of the auto-converted Parquet files:

6.66 GB

Number of rows:

328,025