create

dataset.py

@@ -0,0 +1,92 @@
+import os
+import json
+import datasets
+
+logger = datasets.logging.get_logger(__name__)
+
+_CITATION = """\
+@dataset{gotthatdata_stargate_2024,
+  title  = {STARGATE: CIA Remote Viewing Archive},
+  author = {GotThatData},
+  year   = {2024},
+  url    = {https://huggingface.co/datasets/GotThatData/STARGATE}
+}
+"""
+
+_DESCRIPTION = """\
+STARGATE is a dataset of 12,000+ declassified CIA PDFs related to remote viewing (RV), extrasensory perception (ESP), and anomalous cognition. 
+This loader includes structured metadata and binary access to the original scanned PDFs.
+"""
+
+_HOMEPAGE = "https://huggingface.co/datasets/GotThatData/STARGATE"
+
+_LICENSE = "CC-BY-4.0"
+
+class StargatePDFConfig(datasets.BuilderConfig):
+    def __init__(self, **kwargs):
+        super(StargatePDFConfig, self).__init__(**kwargs)
+
+class StargatePDFDataset(datasets.GeneratorBasedBuilder):
+    VERSION = datasets.Version("1.0.0")
+    BUILDER_CONFIGS = [
+        StargatePDFConfig(name="default", version=VERSION, description="STARGATE raw PDFs with metadata")
+    ]
+
+    def _info(self):
+        return datasets.DatasetInfo(
+            description=_DESCRIPTION,
+            features=datasets.Features({
+                "filename": datasets.Value("string"),
+                "document_id": datasets.Value("string"),
+                "page_count": datasets.Value("int32"),
+                "image_count": datasets.Value("int32"),
+                "processed_at": datasets.Value("string"),
+                "ocr_status": datasets.Value("string"),
+                "text_extracted": datasets.Value("bool"),
+                "source": datasets.Value("string"),
+                "tags": datasets.Sequence(datasets.Value("string")),
+                "pdf": datasets.Value("binary"),
+            }),
+            supervised_keys=None,
+            homepage=_HOMEPAGE,
+            license=_LICENSE,
+            citation=_CITATION,
+        )
+
+    def _split_generators(self, dl_manager):
+        # Path to local or downloaded files
+        archive_path = dl_manager.download_and_extract("./")  # Change if remote
+        metadata_path = os.path.join(archive_path, "metadata.json")
+        data_dir = os.path.join(archive_path, "data")
+
+        return [
+            datasets.SplitGenerator(
+                name=datasets.Split.TRAIN,
+                gen_kwargs={"metadata_path": metadata_path, "data_dir": data_dir}
+            )
+        ]
+
+    def _generate_examples(self, metadata_path, data_dir):
+        logger.info(f"⏳ Loading metadata from {metadata_path}")
+        with open(metadata_path, "r", encoding="utf-8") as f:
+            records = json.load(f)
+
+        for idx, record in enumerate(records):
+            pdf_path = os.path.join(data_dir, record["filename"])
+            if not os.path.isfile(pdf_path):
+                logger.warning(f"🚫 Missing PDF: {pdf_path}")
+                continue
+
+            with open(pdf_path, "rb") as pdf_file:
+                yield idx, {
+                    "filename": record.get("filename"),
+                    "document_id": record.get("document_id", record["filename"].replace(".pdf", "")),
+                    "page_count": record.get("page_count", 0),
+                    "image_count": record.get("image_count", 0),
+                    "processed_at": record.get("processed_at", ""),
+                    "ocr_status": record.get("ocr_status", "pending"),
+                    "text_extracted": record.get("text_extracted", False),
+                    "source": record.get("source", "CIA Stargate Archive"),
+                    "tags": record.get("tags", []),
+                    "pdf": pdf_file.read(),
+                }