feat(workers): extract.image via Tesseract

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

workers/tests/test_image.py

@@ -0,0 +1,38 @@
+import subprocess
+from pathlib import Path
+from void_workers.handlers.image import handle as handle_image
+
+FIXTURES = Path(__file__).parent / "fixtures"
+
+
+def _reset_void_schema(conn):
+    conn.execute("DROP SCHEMA IF EXISTS public CASCADE")
+    conn.execute("CREATE SCHEMA public")
+    conn.execute("CREATE EXTENSION IF NOT EXISTS pgcrypto")
+    conn.execute("CREATE EXTENSION IF NOT EXISTS vector")
+
+
+def _run_node_migrations():
+    subprocess.run(
+        ["node", "lib/db/migrate.js", "up"],
+        cwd="/project/src/void-v2",
+        check=True
+    )
+
+
+def test_image_ocr(conn):
+    _reset_void_schema(conn)
+    _run_node_migrations()
+    sp = conn.execute(
+        "INSERT INTO spaces(slug, name) VALUES('plan4-img', 'I') RETURNING id"
+    ).fetchone()[0]
+    blob = FIXTURES / "eng_text.png"
+    ref = conn.execute(
+        "INSERT INTO refs(space_id, kind, blob_path) "
+        "VALUES(%s, 'image', %s) RETURNING id",
+        (sp, str(blob))
+    ).fetchone()[0]
+    out = handle_image({"ref_id": str(ref), "blob_path": str(blob)})
+    assert out["chars"] > 0
+    row = conn.execute("SELECT body_text FROM refs WHERE id=%s", (ref,)).fetchone()
+    assert "blackflame" in (row[0] or "").lower()

workers/void_workers/handlers/__init__.py

@@ -1,6 +1,7 @@
-from . import echo, pdf
+from . import echo, pdf, image
 
 REGISTRY = {
     echo.NAME: echo.handle,
     pdf.NAME: pdf.handle,
+    image.NAME: image.handle,
 }

workers/void_workers/handlers/image.py

@@ -0,0 +1,18 @@
+from PIL import Image
+import pytesseract
+from .. import repo
+
+NAME = "extract.image"
+
+
+def handle(job_data: dict) -> dict:
+    ref_id = job_data["ref_id"]
+    blob_path = job_data["blob_path"]
+    text = pytesseract.image_to_string(Image.open(blob_path), lang="eng").strip()
+    body_text = text[:200_000]
+    repo.update_ref(
+        ref_id,
+        body_text=body_text,
+        metadata_patch={"extract": {"method": "tesseract", "chars": len(body_text)}}
+    )
+    return {"ref_id": ref_id, "chars": len(body_text)}