Upload dcad_2000.py

dcad_2000.py

@@ -0,0 +1,128 @@
+import os
+import json
+import datasets
+
+
+_CITATION = ""
+_DESCRIPTION = "DCAD-2000 flexible loader supporting 2000+ languages without defining configs."
+_HOMEPAGE = ""
+_LICENSE = ""
+
+
+class DCAD2000Config(datasets.BuilderConfig):
+    """Flexible BuilderConfig for arbitrary language name."""
+    def __init__(self, **kwargs):
+        super().__init__(version=datasets.Version("1.0.0"), **kwargs)
+
+
+class DCAD2000(datasets.GeneratorBasedBuilder):
+    BUILDER_CONFIG_CLASS = DCAD2000Config
+
+    def _info(self):
+        return datasets.DatasetInfo(
+            description=_DESCRIPTION,
+            features=datasets.Features({
+                "text": datasets.Value("string"),
+                "source_file": datasets.Value("string"),
+            }),
+            citation=_CITATION,
+            homepage=_HOMEPAGE,
+            license=_LICENSE,
+        )
+
+    @staticmethod
+    def _get_available_languages(data_dir):
+        """Automatically detect all language folders."""
+        if not os.path.isdir(data_dir):
+            return []
+        return [
+            d for d in os.listdir(data_dir)
+            if os.path.isdir(os.path.join(data_dir, d))
+        ]
+
+    @classmethod
+    def _builder_configs(cls):
+        """
+        Dynamically create configs for any language folder in repo.
+        """
+
+        data_dir = os.getenv("DATASETS_DATA_DIR")  # Set by HF automatically
+        if data_dir is None:
+            # No detection possible (run-time local load)
+            return []
+
+        langs = cls._get_available_languages(data_dir)
+
+        return [
+            DCAD2000Config(name=lang, description=f"Language folder: {lang}")
+            for lang in langs
+        ]
+
+    def _split_generators(self, dl_manager):
+        """
+        Supports only split='keep' → load files containing 'keep'
+        """
+        data_dir = dl_manager.download_and_extract(self.config.data_dir)
+
+        lang_dir = os.path.join(data_dir, self.config.name)
+
+        if not os.path.isdir(lang_dir):
+            raise FileNotFoundError(f"Language directory not found: {lang_dir}")
+
+        split = self.config.split
+        if split != "keep":
+            raise ValueError(f"Only split='keep' is supported. Got {split}")
+
+        files = [
+            os.path.join(lang_dir, f)
+            for f in os.listdir(lang_dir)
+            if "keep" in f
+        ]
+
+        return [
+            datasets.SplitGenerator(
+                name=split,
+                gen_kwargs={"files": files},
+            )
+        ]
+
+    def _generate_examples(self, files):
+        idx = 0
+        for path in files:
+            filename = os.path.basename(path)
+            ext = filename.lower()
+
+            if ext.endswith(".txt"):
+                with open(path, encoding="utf-8") as f:
+                    for line in f:
+                        line = line.strip()
+                        if line:
+                            yield idx, {"text": line, "source_file": filename}
+                            idx += 1
+
+            elif ext.endswith(".jsonl"):
+                with open(path, encoding="utf-8") as f:
+                    for line in f:
+                        obj = json.loads(line)
+                        yield idx, {
+                            "text": obj.get("text", ""),
+                            "source_file": filename
+                        }
+                        idx += 1
+
+            elif ext.endswith(".json"):
+                with open(path, encoding="utf-8") as f:
+                    obj = json.load(f)
+                    if isinstance(obj, list):
+                        for item in obj:
+                            yield idx, {
+                                "text": item.get("text", ""),
+                                "source_file": filename
+                            }
+                            idx += 1
+                    elif isinstance(obj, dict):
+                        yield idx, {
+                            "text": obj.get("text", ""),
+                            "source_file": filename
+                        }
+                        idx += 1