Convert dataset to Parquet (#1)

- Convert dataset to Parquet (c06146153a29c5c8d3a4cd27878ea7a2919193a2)
- Add 'strains' config data files (1db987caa6bc9a8806375afd5ec5fbe27311bd4d)
- Delete loading script (1ee40cc7d7e54eb60574afff4f9e0559ff802c48)
- Delete data file (725df46de952c2efd5b64e3c712dbd703dbb18aa)

.gitignore

@@ -1,2 +0,0 @@
-flake.nix
-flake.lock

README.md

@@ -37,7 +37,7 @@ dataset_info:
   - name: train
     num_bytes: 50995544
     num_examples: 505
-  download_size: 21777460
+  download_size: 11318610
   dataset_size: 50995544
 - config_name: strains
   features:
@@ -70,6 +70,16 @@ dataset_info:
   - name: train
     num_bytes: 41153821
     num_examples: 400
-  download_size: 21777460
+  download_size: 8836267
   dataset_size: 41153821
+configs:
+- config_name: pathogens
+  data_files:
+  - split: train
+    path: pathogens/train-*
+  default: true
+- config_name: strains
+  data_files:
+  - split: train
+    path: strains/train-*
 ---

pathogens/train-00000-of-00001.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c3d6e32db9118f41876ac2144fa5fc951c9985da44eaef244ea5fdb81e8d7a88
+size 11318610

strains/train-00000-of-00001.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fd15182d90d212137f5c72604a1021969e7ac266d395312bd9db692ed02693c1
+size 8836267

tea_curated.py

@@ -1,238 +0,0 @@
-# -*- coding: utf-8 -*-
-"""
-TEA: Taxonomic Entity Augmentation (curated raw corpus)
-
-Usage
------
->>> from datasets import load_dataset
->>> ds = load_dataset("tznurmin/tea_curated",
-...                   "pathogens",
-...                   trust_remote_code=True,
-...                   split="train")
-"""
-
-import json
-import logging
-import re
-from pathlib import Path
-
-import datasets
-
-logger = logging.getLogger(__name__)
-
-# ---------------------------------------------------------------------------
-# Hub metadata
-# ---------------------------------------------------------------------------
-_HOMEPAGE = "https://github.com/tznurmin/TEA_curated_data"
-
-_LICENSE = (
-    "MIT (curation JSON annotations) and mixed Creative-Commons licences "
-    "(CC-BY / CC-BY-SA / CC-BY-NC / CC-BY-NC-SA) for article texts"
-)
-
-_CITATION = r"""
-@misc{nurminen2025tea,
-  author       = {Nurminen, Toni and Kim, John and Si, Tong and Chang, Matthew Wook},
-  title        = {TEA: Taxonomic Entity Augmentation for Biological Literature},
-  year         = {2025},
-  howpublished = {\url{https://github.com/tznurmin/TEA_curated_data}},
-  note         = {Dataset available at \url{https://github.com/tznurmin/TEA_curated_data}}
-}
-"""
-
-_DESCRIPTION = (
-    "Taxonomic Entity Augmentation (TEA) curated dataset. It contains the full "
-    "text of biomedical articles (under their original Creative-Commons licences) "
-    "and curation JSON files that mark pathogen / strain / species entities."
-    "Two configs are provided:\n\n"
-    "1: **pathogens** – commensals, probiotics, opportunistics, negatives, strains, pathogens\n"
-    "2: **strains**   – strains, species, negatives\n\n"
-    "Each token is annotated in **IOB2** format (B-XXX, I-XXX, O). Per-record "
-    "licence information and attribution text are included."
-)
-
-_URL = "https://github.com/tznurmin/TEA_curated_data/archive/refs/tags/v1.1.tar.gz"
-
-# ---------------------------------------------------------------------------
-# IOB2 label sets
-# ---------------------------------------------------------------------------
-_LABELS = {
-    "pathogens": [
-        "O",
-        "B-COMM", "I-COMM",
-        "B-PROB", "I-PROB",
-        "B-OPPO", "I-OPPO",
-        "B-NEGA", "I-NEGA",
-        "B-STRA", "I-STRA",
-        "B-PATH", "I-PATH",
-    ],
-    "strains": [
-        "O",
-        "B-STRA", "I-STRA",
-        "B-SPEC", "I-SPEC",
-        "B-NEGA", "I-NEGA",
-    ],
-}
-
-# Map raw JSON tags (colour suffix stripped) to 4-letter codes
-_CODE4 = {
-    "commensals":     "COMM",
-    "probiotics":     "PROB",
-    "opportunistics": "OPPO",
-    "negatives":      "NEGA",
-    "strains":        "STRA",
-    "pathogens":      "PATH",
-    "species":        "SPEC",
-}
-
-# ---------------------------------------------------------------------------
-# DatasetBuilder
-# ---------------------------------------------------------------------------
-class TeaCurated(datasets.GeneratorBasedBuilder):
-    VERSION = datasets.Version("1.0.0")
-
-    BUILDER_CONFIGS = [
-        datasets.BuilderConfig(
-            name="pathogens",
-            version=VERSION,
-            description="Pathogen curation data",
-        ),
-        datasets.BuilderConfig(
-            name="strains",
-            version=VERSION,
-            description="Strain curation data",
-        ),
-    ]
-    DEFAULT_CONFIG_NAME = "pathogens"
-
-
-    def _info(self):
-        return datasets.DatasetInfo(
-            description=_DESCRIPTION,
-            citation=_CITATION,
-            homepage=_HOMEPAGE,
-            license=_LICENSE,
-            features=datasets.Features(
-                {
-                    "hash": datasets.Value("string"),
-                    "tokens": datasets.Sequence(datasets.Value("string")),
-                    "ner_tags": datasets.Sequence(
-                        datasets.ClassLabel(names=_LABELS[self.config.name])
-                    ),
-                    "locations": datasets.Value("string"),
-                    "license": datasets.Value("string"),
-                    "license_version": datasets.Value("string"),
-                    "license_url": datasets.Value("string"),
-                    "attribution_text": datasets.Value("string"),
-                }
-            ),
-            supervised_keys=None,
-        )
-
-
-    def _split_generators(self, dl_manager):
-        extracted = Path(dl_manager.download_and_extract(_URL))
-        base_dir = extracted / "TEA_curated_data-1.1"
-
-        curation_json = (
-            base_dir / "curation_data" / self.config.name / f"{self.config.name}.json"
-        )
-        articles_dir = base_dir / "source_articles"
-
-        return [
-            datasets.SplitGenerator(
-                name=datasets.Split.TRAIN,
-                gen_kwargs={
-                    "curation_path": curation_json,
-                    "articles_dir": articles_dir,
-                },
-            )
-        ]
-
-
-    def _generate_examples(self, curation_path: Path, articles_dir: Path):
-        with open(curation_path, encoding="utf-8") as f:
-            curation = json.load(f)
-
-        attribution = _parse_attribution_file(articles_dir.parent / "attribution.txt")
-
-        for art_hash, annotation in curation.items():
-            if art_hash not in attribution:
-                raise ValueError(f"Attribution missing for article hash {art_hash!r}")
-
-            txt_path = articles_dir / art_hash / f"{art_hash}.txt"
-            text = txt_path.read_text(encoding="utf-8")
-            text = re.sub(r"\$/i\$", "", text)
-            text = re.sub(r"\$i\$", "", text)
-            text = re.sub(r"\s+", " ", text)
-
-            tokens = text.split()
-            ner_tags = ["O"] * len(tokens)
-
-            for raw_tag, locs in annotation.items():
-                code = _CODE4[raw_tag.split("/")[0]]
-                for loc in locs:                      # e.g. "353+2"
-                    start, length = map(int, loc.split("+"))
-                    for i in range(start, start + length):
-                        if i < len(tokens):
-                            ner_tags[i] = ("B-" if i == start else "I-") + code
-
-            meta = attribution[art_hash]
-
-            yield art_hash, {
-                "hash": art_hash,
-                "tokens": tokens,
-                "ner_tags": ner_tags,
-                "locations": json.dumps(annotation),
-                "license": meta["license"],
-                "license_version": meta["license_version"],
-                "license_url": meta["license_url"],
-                "attribution_text": meta["text"],
-            }
-
-# ---------------------------------------------------------------------------
-# Attribution parser
-# ---------------------------------------------------------------------------
-def _parse_attribution_file(file_path: Path) -> dict[str, dict[str, str]]:
-    """
-    Parse attribution.txt into
-    {checksum: {license, license_version, license_url, text}}.
-    """
-    out = {}
-    csum = lic = lic_url = lic_ver = None
-    temp_lines = []
-
-    with open(file_path, encoding="utf-8") as fp:
-        for line in fp:
-            if "Checksum: " in line:
-                csum = line.split("Checksum: ")[1].strip()
-            elif "is obtained by the following licence: " in line:
-                lic_url = line.split("is obtained by the following licence: ")[1].strip()
-                lic_parts = (
-                    line.split(
-                        "is obtained by the following licence: http://creativecommons.org/licenses/"
-                    )[1]
-                    .strip()
-                    .split("/")
-                )
-                lic_ver = lic_parts[1]
-                lic = f"CC-{lic_parts[0].upper()}"
-                out[csum] = {
-                    "license": lic,
-                    "license_version": lic_ver,
-                    "license_url": lic_url,
-                }
-            else:
-                if len(line.strip()) > 0:
-                    temp_lines.append(line.strip())
-                else:
-                    # blank line marks end of a record
-                    out[csum]["text"] = " ".join(temp_lines).strip()
-                    csum = lic = lic_url = lic_ver = None
-                    temp_lines = []
-
-    # flush last record if file does not end with newline
-    if csum and "text" not in out[csum]:
-        out[csum]["text"] = " ".join(temp_lines).strip()
-
-    return out