Creating a `create.py`script for retsinformation and updating the data, adding 300M tokens.

README.md

@@ -211,9 +211,9 @@ https://github.com/huggingface/datasets/blob/main/templates/README_guide.md
 
 <!-- START-DESC-STATS -->
 - **Language**: dan, dansk, Danish
-- **Number of samples**: 891.35K
-- **Number of tokens (Llama 3)**: 4.37B
-- **Average document length (characters)**: 15083.99
+- **Number of samples**: 927.89K
+- **Number of tokens (Llama 3)**: 4.67B
+- **Average document length (characters)**: 15475.06
 <!-- END-DESC-STATS -->
 
 
@@ -314,8 +314,8 @@ Below follows a brief overview of the sources in the corpus along with their ind
 | Source              | Description                                                                                                                                                                            | Domain       | N. Tokens   | License                |
 |:--------------------|:---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|:-------------|:------------|:-----------------------|
 | [cellar]            | The official digital repository for European Union legal documents and open data                                                                                                       | Legal        | 1.15B       | [CC-BY-SA 4.0]         |
+| [retsinformationdk] | [retsinformation.dk](https://www.retsinformation.dk) (legal-information.dk) the official legal information system of Denmark                                                           | Legal        | 818.25M     | [Danish Copyright Law] |
 | [ncc_books]         | Danish books extracted from the [Norwegian Colossal Corpus](https://huggingface.co/datasets/NbAiLab/NCC) derived from OCR                                                              | Books        | 531.97M     | [CC-0]                 |
-| [retsinformationdk] | [retsinformation.dk](https://www.retsinformation.dk) (legal-information.dk) the official legal information system of Denmark                                                           | Legal        | 516.35M     | [Danish Copyright Law] |
 | [hest]              | Samples from the Danish debate forum www.heste-nettet.dk                                                                                                                               | Social Media | 389.32M     | [CC-0]                 |
 | [ncc_parliament]    | Collections from the Norwegian parliament in Danish. Extracted from the [Norwegian Colossal Corpus](https://huggingface.co/datasets/NbAiLab/NCC) derived from ocr                      | Other        | 338.87M     | [NLOD 2.0]             |
 | [opensubtitles]     | Danish subsection of [OpenSubtitles](https://opus.nlpl.eu/OpenSubtitles/corpus/version/OpenSubtitles)                                                                                  | Conversation | 271.60M     | [CC-0]                 |
@@ -347,7 +347,7 @@ Below follows a brief overview of the sources in the corpus along with their ind
 | [naat]              | Danish speeches from 1930-2022                                                                                                                                                         | Conversation | 286.68K     | [CC-0]                 |
 | [depbank]           | The Danish subsection of the [Universal Dependencies Treebank](https://github.com/UniversalDependencies/UD_Danish-DDT)                                                                 | Other        | 185.45K     | [CC-BY-SA 4.0]         |
 | [synne]             | Dataset collected from [synnejysk forening's website](https://www.synnejysk.dk), covering the Danish dialect sønderjysk                                                                | Other        | 52.02K      | [CC-0]                 |
-| **Total**           |                                                                                                                                                                                        |              | 4.37B       |                        |
+| **Total**           |                                                                                                                                                                                        |              | 4.67B       |                        |
 
 [ai-aktindsigt]: data/ai-aktindsigt/ai-aktindsigt.md
 [cellar]: data/cellar/cellar.md

data/retsinformationdk/create.py

@@ -0,0 +1,166 @@
+# /// script
+# requires-python = ">=3.12"
+# dependencies = [
+#     "datasets==3.2.0",
+#     "pandas",
+#     "requests",
+#     "trafilatura",
+#     "dynaword"
+# ]
+# [tool.uv.sources]
+# dynaword = { git = "https://huggingface.co/datasets/danish-foundation-models/danish-dynaword", rev = "00e7f2aee7f7ad2da423419f77ecbb9c0536de0d" }
+# ///
+
+from datetime import date, datetime
+from io import StringIO
+import logging
+from pathlib import Path
+import pandas as pd
+import requests
+from requests.adapters import HTTPAdapter
+from urllib3 import Retry
+from trafilatura import extract
+from datasets import Dataset
+from tqdm import tqdm
+
+from dynaword.process_dataset import (
+    add_token_count,
+    ensure_column_order,
+    remove_duplicate_text,
+    remove_empty_texts,
+)
+
+TMP_DIR = Path(__file__).parent / "tmp"
+
+BASE_URL = "https://www.retsinformation.dk/api/document/eli"
+
+logger = logging.getLogger(__name__)
+today = date.today()
+
+
+def create_session_with_retries(retries=2, backoff_factor=0.5):
+    session = requests.Session()
+    retry_strategy = Retry(
+        total=retries,
+        backoff_factor=backoff_factor,
+        status_forcelist=[500, 502, 503, 504],
+        allowed_methods=["GET"],
+        respect_retry_after_header=True,
+    )
+    adapter = HTTPAdapter(max_retries=retry_strategy)
+    session.mount("http://", adapter)
+    session.mount("https://", adapter)
+    return session
+
+
+def fetch_document_list():
+    download = True
+    csv_content = ""
+
+    df: pd.DataFrame = pd.DataFrame()
+
+    if TMP_DIR.exists():
+        files = list(TMP_DIR.glob("*.csv"))
+        file = sorted(files, reverse=True)[0]
+
+        file_date = datetime.strptime(file.stem, "%Y-%m-%d").date()
+
+        if (today - file_date).days < 180:
+            download = False
+            df = pd.read_csv(file)
+
+    if download:
+        logger.info("Downloading list of files from Retsinformation.dk")
+        response = requests.get(
+            "https://www.retsinformation.dk/api/documentsearch/csv?dt=10&dt=1480&dt=20&dt=30&dt=40&dt=50&dt=90&dt=120&dt=270&dt=60&dt=100&dt=80&dt=110&dt=130&dt=140&dt=150&dt=160&dt=170&dt=180&dt=200&dt=210&dt=220&dt=1510&dt=1490&dt=-10&dt=230&dt=240&dt=250&dt=260&dt=980&dt=360&dt=400&dt=380&dt=420&dt=1530&dt=440&dt=450&dt=430&dt=1540&dt=460&dt=410&dt=370&dt=480&dt=390&dt=500&dt=510&dt=520&dt=490&dt=300&dt=310&dt=320&dt=330&dt=340&dt=350&o=40"
+        )
+        # response = requests.get(url, headers=headers)
+        response.raise_for_status()  # Raise error for bad responses
+
+        # The response is a gzip-compressed CSV in plain text
+        csv_content = response.content.decode("utf-16", errors="replace")
+        logger.info("Downloaded list of documents")
+
+        # Optionally parse with pandas
+        df = pd.read_csv(StringIO(csv_content), sep=";")  # Assuming semicolon separator
+
+        df.to_csv(TMP_DIR / (today.strftime("%Y-%m-%d") + ".csv"), index=False)
+
+    return df[
+        [
+            "DokumentType",
+            "DokumentId",
+            "Titel",
+            "Ressort",
+            "Historisk",
+            "PubliceretTidspunkt",
+            "EliUrl",
+        ]
+    ]
+
+
+def fetch_document(doc_info: pd.Series, session: requests.Session) -> dict:
+    url = BASE_URL + doc_info["EliUrl"].strip().split("eli")[1]
+
+    response = session.post(
+        url,
+        headers={
+            "Accept": "application/json",
+            "Content-Type": "application/json",
+        },
+        json={},
+    )
+    response.raise_for_status()
+
+    return response.json()[0]
+
+
+def main():
+    save_path = Path(__file__).parent / "retsinformationdk.parquet"
+    documents = fetch_document_list()
+
+    logger.info(f"Found {len(documents)} documents from retsinformationdk")
+
+    session = create_session_with_retries()
+    docs = []
+    for idx, doc_info in tqdm(documents.iterrows(), total=len(documents)):
+        if doc_info["Historisk"]:
+            continue
+        try:
+            doc = fetch_document(doc_info, session)
+            text = extract(doc["documentHtml"], output_format="markdown")
+            docs.append(
+                {
+                    "id": doc_info["DokumentId"],
+                    "text": text if text else "",
+                    "source": "retsinformationdk",
+                    "added": today.strftime("%Y-%m-%d"),
+                    "created": f"{date.fromisoformat(str(doc_info['PubliceretTidspunkt'])).strftime('%Y-%m-%d')}, {date.fromisoformat(str(doc_info['PubliceretTidspunkt'])).strftime('%Y-%m-%d')}",
+                }
+            )
+        except Exception as e:
+            logger.error(f"Ran in to error: {e}")
+            logger.error(f"Skipping doc {doc_info['DokumentId']}")
+
+    ds = Dataset.from_list(docs)
+
+    # quality checks and processing
+    ds = remove_empty_texts(ds)
+    ds = remove_duplicate_text(ds)
+    ds = add_token_count(ds)
+    ds = ensure_column_order(ds)
+
+    ds.to_parquet(save_path)
+
+
+if __name__ == "__main__":
+    log_path = Path(__file__).parent / "retsinformationdk.log"
+    logging.basicConfig(
+        level=logging.INFO,
+        format="%(asctime)s - %(levelname)s - %(message)s",
+        handlers=[
+            logging.StreamHandler(),
+            logging.FileHandler(log_path),
+        ],
+    )
+    main()

data/retsinformationdk/descriptive_stats.json

@@ -1,6 +1,6 @@
 {
-  "number_of_samples": 63979,
-  "average_document_length": 22262.417246283938,
-  "number_of_tokens": 516352722,
-  "revision": "1546256ca9562ecef403e433276c36770859089e"
+  "number_of_samples": 100524,
+  "average_document_length": 23265.030191794995,
+  "number_of_tokens": 818252220,
+  "revision": "2c91001b440e33497c34fbfa9b40dfffffa25620"
 }

data/retsinformationdk/retsinformationdk.md

@@ -39,9 +39,9 @@ It serves as a central repository for Danish legislation, administrative regulat
 <!-- START-DESC-STATS -->
 - **Language**: dan, dansk, Danish
 - **Domains**: Legal
-- **Number of samples**: 63.98K
-- **Number of tokens (Llama 3)**: 516.35M
-- **Average document length (characters)**: 22262.42
+- **Number of samples**: 100.52K
+- **Number of tokens (Llama 3)**: 818.25M
+- **Average document length (characters)**: 23265.03
 <!-- END-DESC-STATS -->
 
 
@@ -52,12 +52,12 @@ An example from the dataset looks as follows.
 <!-- START-SAMPLE -->
 ```py
 {
-  "id": "retsinformationdk_173889",
-  "text": "Den fulde tekst Pressenævnets kendelse i sag nr. 15-70-00822\nResumé\nForeningen for Skånsomt Kystfisk[...]",
+  "id": "AA014851",
+  "text": "Indsamlingsnævnets afgørelse i sag nr. 22-730-00015\n\nIndsamlingsnævnet fandt det kritisabelt, at Gad[...]",
   "source": "retsinformationdk",
-  "added": "2019-11-22",
-  "created": "2000-01-01, 2022-01-01",
-  "token_count": 900
+  "added": "2025-06-26",
+  "created": "2025-06-25, 2025-06-25",
+  "token_count": 4062
 }
 ```
 

data/retsinformationdk/retsinformationdk.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b5949575b39426d8084c9001213d838d94b71b739fa63725445402ef73c4c1f2
-size 648707629
+oid sha256:191bab8a3e7ae419394a622b74ae0fe64e9b5033066eeab4a3b3d2960153d48a
+size 1017748370

descriptive_stats.json

@@ -1,6 +1,6 @@
 {
-  "number_of_samples": 891348,
-  "average_document_length": 15083.994534121353,
-  "number_of_tokens": 4369589385,
-  "revision": "8d056aba9953ef0cf4c402ccb9deff745d8307af"
+  "number_of_samples": 927893,
+  "average_document_length": 15475.058174811104,
+  "number_of_tokens": 4671403830,
+  "revision": "2c91001b440e33497c34fbfa9b40dfffffa25620"
 }

test_results.log

@@ -1,5 +1,5 @@
 ============================= test session starts ==============================
-platform darwin -- Python 3.12.9, pytest-8.3.4, pluggy-1.5.0
+platform darwin -- Python 3.12.0, pytest-8.3.4, pluggy-1.5.0
 rootdir: /Users/kristianjensen/Documents/danish-dynaword
 configfile: pyproject.toml
 collected 310 items
@@ -21,4 +21,4 @@ src/tests/test_quality/test_short_texts.py: 34 warnings
   /Users/kristianjensen/Documents/danish-dynaword/.venv/lib/python3.12/site-packages/datasets/utils/_dill.py:385: DeprecationWarning: co_lnotab is deprecated, use co_lines instead.
 
 -- Docs: https://docs.pytest.org/en/stable/how-to/capture-warnings.html
-================= 309 passed, 1 skipped, 34 warnings in 46.75s =================
+================= 309 passed, 1 skipped, 34 warnings in 44.54s =================