ummaya 0.2.3 → 0.2.5

package/src/ummaya/tools/documents/formats/passive.py

@@ -0,0 +1,766 @@
+# SPDX-License-Identifier: Apache-2.0
+"""Known-only passive adapters for non-promoted document families."""
+
+from __future__ import annotations
+
+import csv
+import gzip
+import io
+import json
+import tarfile
+import zipfile
+from html.parser import HTMLParser
+from pathlib import Path
+from typing import TYPE_CHECKING
+from xml.etree import ElementTree as StdElementTree
+
+import yaml
+from defusedxml import ElementTree  # type: ignore[import-untyped]
+
+from ummaya.tools.documents.models import (
+    DocumentExtraction,
+    DocumentFormat,
+    ImageReference,
+    KnownDocumentFormat,
+    MetadataValue,
+    ParagraphBlock,
+    TableBlock,
+    TableCell,
+)
+
+if TYPE_CHECKING:
+    from ummaya.tools.documents.tool_defs import DocumentFieldPatch
+
+
+_ODF_FORMATS: tuple[KnownDocumentFormat, ...] = (
+    KnownDocumentFormat.odt,
+    KnownDocumentFormat.ods,
+    KnownDocumentFormat.odp,
+)
+_DATA_FORMATS: tuple[KnownDocumentFormat, ...] = (
+    KnownDocumentFormat.csv,
+    KnownDocumentFormat.tsv,
+    KnownDocumentFormat.xml,
+    KnownDocumentFormat.rdf,
+    KnownDocumentFormat.ttl,
+    KnownDocumentFormat.lod,
+    KnownDocumentFormat.json,
+    KnownDocumentFormat.jsonl,
+    KnownDocumentFormat.yaml,
+    KnownDocumentFormat.yml,
+    KnownDocumentFormat.geojson,
+    KnownDocumentFormat.gpx,
+    KnownDocumentFormat.kml,
+    KnownDocumentFormat.fasta,
+    KnownDocumentFormat.sgml,
+    KnownDocumentFormat.dtd,
+    KnownDocumentFormat.hml,
+    KnownDocumentFormat.etc,
+)
+_TEXT_WEB_FORMATS: tuple[KnownDocumentFormat, ...] = (
+    KnownDocumentFormat.html,
+    KnownDocumentFormat.htm,
+    KnownDocumentFormat.txt,
+    KnownDocumentFormat.rtf,
+    KnownDocumentFormat.md,
+)
+_LEGACY_OFFICE_FORMATS: tuple[KnownDocumentFormat, ...] = (
+    KnownDocumentFormat.doc,
+    KnownDocumentFormat.xls,
+    KnownDocumentFormat.ppt,
+)
+_CODE_FORMATS: tuple[KnownDocumentFormat, ...] = (KnownDocumentFormat.python,)
+_IMAGE_FORMATS: tuple[KnownDocumentFormat, ...] = (
+    KnownDocumentFormat.png,
+    KnownDocumentFormat.jpg,
+    KnownDocumentFormat.jpeg,
+    KnownDocumentFormat.gif,
+    KnownDocumentFormat.tif,
+    KnownDocumentFormat.tiff,
+    KnownDocumentFormat.bmp,
+    KnownDocumentFormat.webp,
+)
+_GEOSPATIAL_FORMATS: tuple[KnownDocumentFormat, ...] = (
+    KnownDocumentFormat.shp,
+    KnownDocumentFormat.shx,
+    KnownDocumentFormat.dbf,
+    KnownDocumentFormat.prj,
+    KnownDocumentFormat.stl,
+)
+_MEDIA_FORMATS: tuple[KnownDocumentFormat, ...] = (
+    KnownDocumentFormat.wav,
+    KnownDocumentFormat.mp3,
+    KnownDocumentFormat.mp4,
+)
+_ARCHIVE_FORMATS: tuple[KnownDocumentFormat, ...] = (
+    KnownDocumentFormat.epub,
+    KnownDocumentFormat.zip,
+    KnownDocumentFormat.seven_z,
+    KnownDocumentFormat.tar,
+    KnownDocumentFormat.gz,
+)
+
+_KNOWN_BY_EXTENSION = {
+    ".odt": KnownDocumentFormat.odt,
+    ".ods": KnownDocumentFormat.ods,
+    ".odp": KnownDocumentFormat.odp,
+    ".doc": KnownDocumentFormat.doc,
+    ".xls": KnownDocumentFormat.xls,
+    ".ppt": KnownDocumentFormat.ppt,
+    ".csv": KnownDocumentFormat.csv,
+    ".tsv": KnownDocumentFormat.tsv,
+    ".xml": KnownDocumentFormat.xml,
+    ".rdf": KnownDocumentFormat.rdf,
+    ".ttl": KnownDocumentFormat.ttl,
+    ".lod": KnownDocumentFormat.lod,
+    ".json": KnownDocumentFormat.json,
+    ".jsonl": KnownDocumentFormat.jsonl,
+    ".yaml": KnownDocumentFormat.yaml,
+    ".yml": KnownDocumentFormat.yml,
+    ".geojson": KnownDocumentFormat.geojson,
+    ".gpx": KnownDocumentFormat.gpx,
+    ".kml": KnownDocumentFormat.kml,
+    ".fasta": KnownDocumentFormat.fasta,
+    ".sgml": KnownDocumentFormat.sgml,
+    ".dtd": KnownDocumentFormat.dtd,
+    ".hml": KnownDocumentFormat.hml,
+    ".etc": KnownDocumentFormat.etc,
+    ".py": KnownDocumentFormat.python,
+    ".html": KnownDocumentFormat.html,
+    ".htm": KnownDocumentFormat.htm,
+    ".txt": KnownDocumentFormat.txt,
+    ".rtf": KnownDocumentFormat.rtf,
+    ".md": KnownDocumentFormat.md,
+    ".png": KnownDocumentFormat.png,
+    ".jpg": KnownDocumentFormat.jpg,
+    ".jpeg": KnownDocumentFormat.jpeg,
+    ".gif": KnownDocumentFormat.gif,
+    ".tif": KnownDocumentFormat.tif,
+    ".tiff": KnownDocumentFormat.tiff,
+    ".bmp": KnownDocumentFormat.bmp,
+    ".webp": KnownDocumentFormat.webp,
+    ".shp": KnownDocumentFormat.shp,
+    ".shx": KnownDocumentFormat.shx,
+    ".dbf": KnownDocumentFormat.dbf,
+    ".prj": KnownDocumentFormat.prj,
+    ".stl": KnownDocumentFormat.stl,
+    ".wav": KnownDocumentFormat.wav,
+    ".mp3": KnownDocumentFormat.mp3,
+    ".mp4": KnownDocumentFormat.mp4,
+    ".epub": KnownDocumentFormat.epub,
+    ".zip": KnownDocumentFormat.zip,
+    ".7z": KnownDocumentFormat.seven_z,
+    ".tar": KnownDocumentFormat.tar,
+    ".gz": KnownDocumentFormat.gz,
+}
+
+_IMAGE_MIME = {
+    KnownDocumentFormat.png: "image/png",
+    KnownDocumentFormat.jpg: "image/jpeg",
+    KnownDocumentFormat.jpeg: "image/jpeg",
+    KnownDocumentFormat.gif: "image/gif",
+    KnownDocumentFormat.tif: "image/tiff",
+    KnownDocumentFormat.tiff: "image/tiff",
+    KnownDocumentFormat.bmp: "image/bmp",
+    KnownDocumentFormat.webp: "image/webp",
+}
+
+
+class _KnownOnlyAdapterBase:
+    """Shared known-only adapter behavior."""
+
+    adapter_id: str
+    known_formats: tuple[KnownDocumentFormat, ...]
+    promoted_formats: tuple[DocumentFormat, ...] = ()
+
+    @property
+    def engine_id(self) -> str:
+        """Return adapter id for diagnostics."""
+        return self.adapter_id
+
+    def normalize_fill_patches(
+        self,
+        patches: tuple[DocumentFieldPatch, ...],
+        *,
+        extraction: DocumentExtraction | None,
+    ) -> tuple[DocumentFieldPatch, ...]:
+        """No passive adapter normalizes fill patches because writes are not promoted."""
+        _ = extraction
+        return patches
+
+
+class OdfDocumentAdapter(_KnownOnlyAdapterBase):
+    """Read-only ODF package candidate backed by ZIP/XML inspection."""
+
+    adapter_id = "odf-package-read-only-adapter"
+    known_formats = _ODF_FORMATS
+
+    def inspect(self, path: Path, *, artifact_id: str) -> DocumentExtraction:
+        """Extract ODF package text from content.xml without claiming mutation."""
+        known_format = _known_format(path)
+        paragraphs: list[ParagraphBlock] = []
+        warnings: list[str] = []
+        metadata: dict[str, MetadataValue] = _base_metadata(
+            path,
+            known_format=known_format,
+            adapter_id=self.adapter_id,
+            mutation_policy="read_only_odf_candidate",
+        )
+        try:
+            with zipfile.ZipFile(path) as archive:
+                metadata["package_entry_count"] = len(archive.infolist())
+                if "content.xml" in archive.namelist():
+                    root = ElementTree.fromstring(archive.read("content.xml"))
+                    paragraphs = _paragraphs_from_text_lines(
+                        artifact_id,
+                        _xml_text_lines(root),
+                        source_prefix="content.xml",
+                    )
+                else:
+                    warnings.append("ODF package does not contain content.xml.")
+        except zipfile.BadZipFile:
+            warnings.append("ODF read-only candidate could not open the package as ZIP.")
+
+        return DocumentExtraction(
+            artifact_id=artifact_id,
+            paragraphs=paragraphs,
+            metadata=metadata,
+            warnings=warnings,
+        )
+
+
+class DataFileDocumentAdapter(_KnownOnlyAdapterBase):
+    """Read-only data-file adapter with serializer round-trip evidence."""
+
+    adapter_id = "data-file-read-only-adapter"
+    known_formats = _DATA_FORMATS
+
+    def inspect(self, path: Path, *, artifact_id: str) -> DocumentExtraction:
+        """Parse structured data files through safe local serializers."""
+        known_format = _known_format(path)
+        if known_format in {KnownDocumentFormat.csv, KnownDocumentFormat.tsv}:
+            return _inspect_delimited(path, artifact_id=artifact_id, known_format=known_format)
+        if known_format in {
+            KnownDocumentFormat.json,
+            KnownDocumentFormat.jsonl,
+            KnownDocumentFormat.geojson,
+        }:
+            return _inspect_json(path, artifact_id=artifact_id, known_format=known_format)
+        if known_format in {KnownDocumentFormat.yaml, KnownDocumentFormat.yml}:
+            return _inspect_yaml(path, artifact_id=artifact_id, known_format=known_format)
+        if known_format in {
+            KnownDocumentFormat.xml,
+            KnownDocumentFormat.rdf,
+            KnownDocumentFormat.gpx,
+            KnownDocumentFormat.kml,
+            KnownDocumentFormat.hml,
+        }:
+            return _inspect_xml(path, artifact_id=artifact_id, known_format=known_format)
+        if known_format in {
+            KnownDocumentFormat.ttl,
+            KnownDocumentFormat.lod,
+            KnownDocumentFormat.fasta,
+            KnownDocumentFormat.sgml,
+            KnownDocumentFormat.dtd,
+            KnownDocumentFormat.etc,
+        }:
+            return _inspect_text_data(path, artifact_id=artifact_id, known_format=known_format)
+        return DocumentExtraction(
+            artifact_id=artifact_id,
+            metadata=_base_metadata(
+                path,
+                known_format=known_format,
+                adapter_id=self.adapter_id,
+                mutation_policy="read_only_data_file",
+            ),
+            warnings=[f"No passive data parser is implemented for {known_format.value}."],
+        )
+
+
+class LegacyOfficeDocumentAdapter(_KnownOnlyAdapterBase):
+    """Metadata-only adapter for pre-OOXML Office binaries."""
+
+    adapter_id = "legacy-office-metadata-only-adapter"
+    known_formats = _LEGACY_OFFICE_FORMATS
+
+    def inspect(self, path: Path, *, artifact_id: str) -> DocumentExtraction:
+        """Identify legacy Office documents without parsing binary internals."""
+        known_format = _known_format(path)
+        return DocumentExtraction(
+            artifact_id=artifact_id,
+            metadata=_base_metadata(
+                path,
+                known_format=known_format,
+                adapter_id=self.adapter_id,
+                mutation_policy="conversion_required_legacy_office",
+            ),
+            warnings=[
+                "Legacy Office binary inspection is metadata-only until an explicit local "
+                "conversion bridge is approved."
+            ],
+        )
+
+
+class TextWebExportAdapter(_KnownOnlyAdapterBase):
+    """Read-only HTML, text, RTF, and Markdown export adapter."""
+
+    adapter_id = "text-web-export-read-only-adapter"
+    known_formats = _TEXT_WEB_FORMATS
+
+    def inspect(self, path: Path, *, artifact_id: str) -> DocumentExtraction:
+        """Extract visible text lines from text and web-export formats."""
+        known_format = _known_format(path)
+        payload = path.read_text(encoding="utf-8", errors="replace")
+        lines = (
+            _html_text_lines(payload)
+            if known_format in {KnownDocumentFormat.html, KnownDocumentFormat.htm}
+            else _plain_text_lines(_strip_minimal_rtf(payload))
+        )
+        return DocumentExtraction(
+            artifact_id=artifact_id,
+            paragraphs=_paragraphs_from_text_lines(
+                artifact_id,
+                lines,
+                source_prefix=path.name,
+            ),
+            metadata=_base_metadata(
+                path,
+                known_format=known_format,
+                adapter_id=self.adapter_id,
+                mutation_policy="read_only_text_export",
+            ),
+        )
+
+
+class CodeFileDocumentAdapter(_KnownOnlyAdapterBase):
+    """Read-only source-code export adapter for public-data attachments."""
+
+    adapter_id = "code-file-read-only-adapter"
+    known_formats = _CODE_FORMATS
+
+    def inspect(self, path: Path, *, artifact_id: str) -> DocumentExtraction:
+        """Extract source lines for context without using the document writer."""
+        payload = path.read_text(encoding="utf-8", errors="replace")
+        return DocumentExtraction(
+            artifact_id=artifact_id,
+            paragraphs=_paragraphs_from_text_lines(
+                artifact_id,
+                _plain_text_lines(payload)[:200],
+                source_prefix=path.name,
+            ),
+            metadata=_base_metadata(
+                path,
+                known_format=_known_format(path),
+                adapter_id=self.adapter_id,
+                mutation_policy="read_only_code_file",
+            ),
+            warnings=["Code files are not public-form documents and cannot be mutated here."],
+        )
+
+
+class ImageScanDocumentAdapter(_KnownOnlyAdapterBase):
+    """Extraction-only image/scan adapter."""
+
+    adapter_id = "image-scan-extraction-only-adapter"
+    known_formats = _IMAGE_FORMATS
+
+    def inspect(self, path: Path, *, artifact_id: str) -> DocumentExtraction:
+        """Return an image reference without claiming OCR or write support."""
+        known_format = _known_format(path)
+        return DocumentExtraction(
+            artifact_id=artifact_id,
+            images=[
+                ImageReference(
+                    image_id=f"image-{known_format.value}",
+                    source_path=str(path),
+                    content_type=_IMAGE_MIME.get(known_format, "image/unknown"),
+                )
+            ],
+            metadata=_base_metadata(
+                path,
+                known_format=known_format,
+                adapter_id=self.adapter_id,
+                mutation_policy="extraction_only",
+            ),
+            warnings=["Image scan adapter does not mutate raster originals."],
+        )
+
+
+class GeospatialDocumentAdapter(_KnownOnlyAdapterBase):
+    """Metadata-only geospatial and 3D model adapter."""
+
+    adapter_id = "geospatial-metadata-only-adapter"
+    known_formats = _GEOSPATIAL_FORMATS
+
+    def inspect(self, path: Path, *, artifact_id: str) -> DocumentExtraction:
+        """Classify GIS/model artifacts without claiming document editing."""
+        known_format = _known_format(path)
+        paragraphs = (
+            _paragraphs_from_text_lines(
+                artifact_id,
+                _plain_text_lines(path.read_text(encoding="utf-8", errors="replace"))[:40],
+                source_prefix=path.name,
+            )
+            if known_format in {KnownDocumentFormat.prj, KnownDocumentFormat.stl}
+            else []
+        )
+        return DocumentExtraction(
+            artifact_id=artifact_id,
+            paragraphs=paragraphs,
+            metadata=_base_metadata(
+                path,
+                known_format=known_format,
+                adapter_id=self.adapter_id,
+                mutation_policy="metadata_only_geospatial_asset",
+            ),
+            warnings=[
+                "Geospatial and 3D geometry files are classified for routing, not mutated "
+                "as public documents."
+            ],
+        )
+
+
+class MediaAssetDocumentAdapter(_KnownOnlyAdapterBase):
+    """Metadata-only audio/video adapter."""
+
+    adapter_id = "media-asset-metadata-only-adapter"
+    known_formats = _MEDIA_FORMATS
+
+    def inspect(self, path: Path, *, artifact_id: str) -> DocumentExtraction:
+        """Classify media attachments without transcription or mutation claims."""
+        return DocumentExtraction(
+            artifact_id=artifact_id,
+            metadata=_base_metadata(
+                path,
+                known_format=_known_format(path),
+                adapter_id=self.adapter_id,
+                mutation_policy="metadata_only_media_asset",
+            ),
+            warnings=[
+                "Media files need a dedicated transcription or extraction adapter before "
+                "content can be written into a public document derivative."
+            ],
+        )
+
+
+class ArchiveDocumentSetAdapter(_KnownOnlyAdapterBase):
+    """Read-only archive enumerator for secure child routing."""
+
+    adapter_id = "archive-document-set-read-only-adapter"
+    known_formats = _ARCHIVE_FORMATS
+
+    def __init__(
+        self,
+        known_formats: tuple[KnownDocumentFormat, ...] | None = None,
+    ) -> None:
+        self.known_formats = known_formats or _ARCHIVE_FORMATS
+
+    def inspect(self, path: Path, *, artifact_id: str) -> DocumentExtraction:
+        """Enumerate archive members without mutating children in place."""
+        known_format = _known_format(path)
+        names, warnings = _archive_member_names(path, known_format=known_format)
+        metadata = _base_metadata(
+            path,
+            known_format=known_format,
+            adapter_id=self.adapter_id,
+            mutation_policy="archive_read_only",
+        )
+        metadata["entry_count"] = len(names)
+        metadata["child_mutation_policy"] = "route_children_as_derivatives"
+        return DocumentExtraction(
+            artifact_id=artifact_id,
+            paragraphs=_paragraphs_from_text_lines(
+                artifact_id,
+                names,
+                source_prefix=path.name,
+            ),
+            metadata=metadata,
+            warnings=warnings,
+        )
+
+
+def _inspect_delimited(
+    path: Path,
+    *,
+    artifact_id: str,
+    known_format: KnownDocumentFormat,
+) -> DocumentExtraction:
+    delimiter = "\t" if known_format is KnownDocumentFormat.tsv else ","
+    payload = path.read_text(encoding="utf-8-sig", errors="replace")
+    rows = list(csv.reader(io.StringIO(payload), delimiter=delimiter))
+    serialized = io.StringIO()
+    writer = csv.writer(serialized, delimiter=delimiter, lineterminator="\n")
+    writer.writerows(rows)
+    reparsed = list(csv.reader(io.StringIO(serialized.getvalue()), delimiter=delimiter))
+    metadata = _base_metadata(
+        path,
+        known_format=known_format,
+        adapter_id=DataFileDocumentAdapter.adapter_id,
+        mutation_policy="read_only_data_file",
+    )
+    metadata.update(
+        {
+            "serializer": known_format.value,
+            "round_trip_passed": rows == reparsed,
+            "row_count": len(rows),
+            "column_count": max((len(row) for row in rows), default=0),
+        }
+    )
+    return DocumentExtraction(
+        artifact_id=artifact_id,
+        tables=[_table_from_rows(rows, source_path=path.name)],
+        paragraphs=_paragraphs_from_text_lines(
+            artifact_id,
+            [",".join(row) for row in rows],
+            source_prefix=path.name,
+        ),
+        metadata=metadata,
+    )
+
+
+def _inspect_json(
+    path: Path,
+    *,
+    artifact_id: str,
+    known_format: KnownDocumentFormat,
+) -> DocumentExtraction:
+    payload = path.read_text(encoding="utf-8")
+    if known_format is KnownDocumentFormat.jsonl:
+        values = [json.loads(line) for line in payload.splitlines() if line.strip()]
+        serialized = "\n".join(
+            json.dumps(value, ensure_ascii=False, sort_keys=True) for value in values
+        )
+        reparsed: object = [json.loads(line) for line in serialized.splitlines()]
+        parsed: object = values
+    else:
+        parsed = json.loads(payload)
+        serialized = json.dumps(parsed, ensure_ascii=False, sort_keys=True)
+        reparsed = json.loads(serialized)
+    return _structured_data_extraction(
+        artifact_id,
+        path=path,
+        known_format=known_format,
+        serializer=known_format.value,
+        parsed=parsed,
+        round_trip_passed=parsed == reparsed,
+    )
+
+
+def _inspect_yaml(
+    path: Path,
+    *,
+    artifact_id: str,
+    known_format: KnownDocumentFormat,
+) -> DocumentExtraction:
+    parsed = yaml.safe_load(path.read_text(encoding="utf-8"))
+    serialized = yaml.safe_dump(parsed, allow_unicode=True, sort_keys=True)
+    return _structured_data_extraction(
+        artifact_id,
+        path=path,
+        known_format=known_format,
+        serializer="yaml.safe_load/safe_dump",
+        parsed=parsed,
+        round_trip_passed=parsed == yaml.safe_load(serialized),
+    )
+
+
+def _inspect_xml(
+    path: Path,
+    *,
+    artifact_id: str,
+    known_format: KnownDocumentFormat,
+) -> DocumentExtraction:
+    root = ElementTree.fromstring(path.read_bytes())
+    serialized = StdElementTree.tostring(root, encoding="unicode")
+    reparsed = ElementTree.fromstring(serialized.encode("utf-8"))
+    return _structured_data_extraction(
+        artifact_id,
+        path=path,
+        known_format=known_format,
+        serializer="defusedxml.ElementTree",
+        parsed={"root_tag": _local_name(root.tag), "text": " ".join(_xml_text_lines(root))},
+        round_trip_passed=_local_name(root.tag) == _local_name(reparsed.tag),
+    )
+
+
+def _inspect_text_data(
+    path: Path,
+    *,
+    artifact_id: str,
+    known_format: KnownDocumentFormat,
+) -> DocumentExtraction:
+    payload = path.read_text(encoding="utf-8", errors="replace")
+    lines = _plain_text_lines(payload)[:200]
+    metadata = _base_metadata(
+        path,
+        known_format=known_format,
+        adapter_id=DataFileDocumentAdapter.adapter_id,
+        mutation_policy="read_only_data_file",
+    )
+    metadata["serializer"] = "plain-text-preview"
+    metadata["round_trip_passed"] = True
+    metadata["line_count"] = len(lines)
+    return DocumentExtraction(
+        artifact_id=artifact_id,
+        paragraphs=_paragraphs_from_text_lines(
+            artifact_id,
+            lines,
+            source_prefix=path.name,
+        ),
+        metadata=metadata,
+    )
+
+
+def _structured_data_extraction(
+    artifact_id: str,
+    *,
+    path: Path,
+    known_format: KnownDocumentFormat,
+    serializer: str,
+    parsed: object,
+    round_trip_passed: bool,
+) -> DocumentExtraction:
+    metadata = _base_metadata(
+        path,
+        known_format=known_format,
+        adapter_id=DataFileDocumentAdapter.adapter_id,
+        mutation_policy="read_only_data_file",
+    )
+    metadata.update(
+        {
+            "serializer": serializer,
+            "round_trip_passed": round_trip_passed,
+            "root_type": type(parsed).__name__,
+        }
+    )
+    lines = _structured_preview_lines(parsed)
+    return DocumentExtraction(
+        artifact_id=artifact_id,
+        paragraphs=_paragraphs_from_text_lines(
+            artifact_id,
+            lines,
+            source_prefix=path.name,
+        ),
+        metadata=metadata,
+    )
+
+
+def _table_from_rows(rows: list[list[str]], *, source_path: str) -> TableBlock:
+    cells: list[TableCell] = []
+    for row_index, row in enumerate(rows):
+        for column_index, value in enumerate(row):
+            cells.append(
+                TableCell(
+                    row_index=row_index,
+                    column_index=column_index,
+                    text=value,
+                    source_path=f"{source_path}#r{row_index + 1}c{column_index + 1}",
+                )
+            )
+    return TableBlock(block_id="data-table-001", source_path=source_path, cells=cells)
+
+
+def _archive_member_names(
+    path: Path,
+    *,
+    known_format: KnownDocumentFormat,
+) -> tuple[list[str], list[str]]:
+    if known_format is KnownDocumentFormat.zip:
+        with zipfile.ZipFile(path) as archive:
+            return _safe_member_names(archive.namelist()), []
+    if known_format is KnownDocumentFormat.tar:
+        with tarfile.open(path) as archive:
+            return _safe_member_names(archive.getnames()), []
+    if known_format is KnownDocumentFormat.gz:
+        with gzip.open(path) as payload:
+            payload.read(1)
+        return [path.with_suffix("").name or path.name], [
+            "Gzip payload is treated as one compressed child candidate."
+        ]
+    return [], ["7z archive enumeration is known but not promoted without a 7z runtime."]
+
+
+def _safe_member_names(names: list[str]) -> list[str]:
+    return sorted(name for name in names if name and not name.startswith("/") and ".." not in name)
+
+
+def _html_text_lines(payload: str) -> list[str]:
+    parser = _VisibleTextParser()
+    parser.feed(payload)
+    return parser.lines
+
+
+def _strip_minimal_rtf(payload: str) -> str:
+    if not payload.lstrip().startswith("{\\rtf"):
+        return payload
+    stripped = payload.replace("\\par", "\n")
+    return "".join(ch for ch in stripped if ch not in "{}")
+
+
+def _plain_text_lines(payload: str) -> list[str]:
+    return [line.strip() for line in payload.splitlines() if line.strip()]
+
+
+def _xml_text_lines(root: StdElementTree.Element) -> list[str]:
+    return [text.strip() for text in root.itertext() if text and text.strip()]
+
+
+def _structured_preview_lines(value: object) -> list[str]:
+    if isinstance(value, dict):
+        return [f"{key}: {preview}" for key, preview in list(value.items())[:20]]
+    if isinstance(value, list):
+        return [json.dumps(item, ensure_ascii=False, sort_keys=True) for item in value[:20]]
+    return [str(value)]
+
+
+def _paragraphs_from_text_lines(
+    artifact_id: str,
+    lines: list[str],
+    *,
+    source_prefix: str,
+) -> list[ParagraphBlock]:
+    return [
+        ParagraphBlock(
+            block_id=f"{artifact_id}-line-{index:03d}",
+            text=line,
+            source_path=f"{source_prefix}#line[{index}]",
+        )
+        for index, line in enumerate(lines, start=1)
+        if line
+    ]
+
+
+def _base_metadata(
+    path: Path,
+    *,
+    known_format: KnownDocumentFormat,
+    adapter_id: str,
+    mutation_policy: str,
+) -> dict[str, MetadataValue]:
+    return {
+        "adapter_id": adapter_id,
+        "known_format": known_format.value,
+        "mutation_policy": mutation_policy,
+        "byte_size": path.stat().st_size,
+    }
+
+
+def _known_format(path: Path) -> KnownDocumentFormat:
+    return _KNOWN_BY_EXTENSION.get(path.suffix.lower(), KnownDocumentFormat.txt)
+
+
+def _local_name(tag: str) -> str:
+    return tag.rsplit("}", 1)[-1] if "}" in tag else tag
+
+
+class _VisibleTextParser(HTMLParser):
+    """Small HTML text extractor for passive public-form exports."""
+
+    def __init__(self) -> None:
+        super().__init__(convert_charrefs=True)
+        self.lines: list[str] = []
+
+    def handle_data(self, data: str) -> None:
+        text = data.strip()
+        if text:
+            self.lines.append(text)