standalone OCR

deployment/docker-compose.yaml

@@ -18,9 +18,11 @@ services:
     build:
       context: ../application
       args:
-        # Bake the optional docling engine (OCR, structured output) into the
-        # image: set INSTALL_DOCLING=true in ../.env or the shell.
+        # Bake the optional docling engine (layout-model OCR backend, structured
+        # output) into the image: set INSTALL_DOCLING=true in ../.env or the shell.
         INSTALL_DOCLING: ${INSTALL_DOCLING:-false}
+        # The tesseract binary behind OCR_ENABLED=true; on by default (~35 MB).
+        INSTALL_TESSERACT: ${INSTALL_TESSERACT:-true}
     env_file:
       - ../.env
     environment:
@@ -47,6 +49,7 @@ services:
       context: ../application
       args:
         INSTALL_DOCLING: ${INSTALL_DOCLING:-false}
+        INSTALL_TESSERACT: ${INSTALL_TESSERACT:-true}
     # Consumes the default queue AND the dedicated `parsing` queue (read_document /
     # parse_document). Without `parsing` here the read_document await never resolves.
     # For heavy/OCR parsing run a separate worker with `-Q parsing`.