Upgraded sciencbeam-trainer-delft and with it Python

.python-version

@@ -1 +1 @@
-3.9.17
+3.11

Dockerfile

@@ -1,4 +1,4 @@
-FROM ghcr.io/astral-sh/uv:python3.9-bookworm AS base
+FROM ghcr.io/astral-sh/uv:python3.11-bookworm AS base
 
 
 # shared between builder and runtime image
@@ -22,7 +22,7 @@ RUN apt-get update \
 
 # set and check UNO_PATH, UNO_PYTHON_PATH and UNO_OFFICE_BINARY_PATH
 ENV UNO_PATH=/usr/lib/python3/dist-packages
-ENV UNO_PYTHON_PATH=/usr/local/bin/python3.9
+ENV UNO_PYTHON_PATH=/usr/local/bin/python3.11
 ENV UNO_OFFICE_BINARY_PATH=/usr/lib/libreoffice/program/soffice.bin
 RUN \
   echo "UNO_PATH: ${UNO_PATH}" \
@@ -105,8 +105,8 @@ ENV LC_ALL=C
 FROM dev AS python-dist-builder
 
 ARG python_package_version
-RUN echo "Setting version to: $version" && \
-    uv version "$python_package_version"
+RUN echo "Setting version to: $python_package_version" && \
+    if [ -n "$python_package_version" ]; then uv version "$python_package_version"; fi
 RUN python scripts/dev/update_readme.py \
     --source=./doc/python_library.md \
     --target=./doc/generated_python_library.md \

pyproject.toml

@@ -3,7 +3,7 @@ name = "sciencebeam-parser"
 version = "0.0.0"
 description = "ScienceBeam Parser, parse scientific documents."
 readme = "doc/generated_python_library.md"
-requires-python = ">=3.9,<3.10"
+requires-python = ">=3.10"
 dependencies = [
     "fastapi[standard]>=0.124.0",
     "fsspec>=2022.1.0",
@@ -15,11 +15,11 @@ dependencies = [
 
 [project.optional-dependencies]
 cpu = [
-  "torch==1.10.1+cpu",
-  "torchvision==0.11.2+cpu"
+  "torch>=2.5.1",
+  "torchvision>=0.20.1"
 ]
 delft = [
-    "sciencebeam-trainer-delft[delft]>=0.0.34",
+    "sciencebeam-trainer-delft[delft]>=0.0.36",
 ]
 cv = [
     "layoutparser==0.3.2",
@@ -39,7 +39,7 @@ ocr = [
     "tesserocr==2.5.2",
 ]
 tf = [
-    "tensorflow==2.9.3",
+    "tensorflow>=2.17.1",
 ]
 
 [tool.uv.sources]

sciencebeam_parser/document/semantic_document.py

@@ -21,7 +21,6 @@
 from typing_extensions import Protocol
 
 from sciencebeam_parser.document.layout_document import (
-    EMPTY_BLOCK,
     LayoutBlock,
     LayoutGraphic,
     LayoutToken
@@ -59,7 +58,7 @@ def merged_block(self) -> LayoutBlock:
 
 @dataclass
 class SemanticSimpleContentWrapper(SemanticContentWrapper):
-    content: LayoutBlock = EMPTY_BLOCK
+    content: LayoutBlock = field(default_factory=lambda: LayoutBlock(lines=[]))
 
     layout_block: dataclasses.InitVar[LayoutBlock] = None
 

sciencebeam_parser/models/delft_model_impl.py

@@ -2,10 +2,7 @@
 from typing import Optional, List, Tuple
 
 from sciencebeam_trainer_delft.embedding.manager import EmbeddingManager
-from sciencebeam_trainer_delft.sequence_labelling.wrapper import (
-    DEFAULT_EMBEDDINGS_PATH,
-    Sequence
-)
+from sciencebeam_trainer_delft.sequence_labelling.wrapper import Sequence
 
 from sciencebeam_parser.app.context import AppContext
 from sciencebeam_parser.models.model_impl import ModelImpl
@@ -15,6 +12,9 @@
 LOGGER = logging.getLogger(__name__)
 
 
+REGISTRY_REGISTRY_PATH = 'delft/resources-registry.json'
+
+
 class DelftModelImpl(ModelImpl):
     def __init__(self, model_url: str, app_context: AppContext):
         self.model_url = model_url
@@ -27,9 +27,8 @@ def __repr__(self) -> str:
         )
 
     def _load_model(self) -> Sequence:
-        embedding_registry_path = DEFAULT_EMBEDDINGS_PATH
         embedding_manager = EmbeddingManager(
-            path=embedding_registry_path,
+            path=REGISTRY_REGISTRY_PATH,
             download_manager=self.app_context.download_manager
         )
         model = Sequence(