Keep TableAnnotation type when lines are merged

Dockerfile

@@ -1,17 +1,14 @@
-ARG REPOSITORY="docker.io"
-FROM dedocproject/dedoc_jammy_p3.10_base:version_2025_09_11
-ARG LANGUAGES=""
+FROM dedocproject/dedoc_p3.10_base:version_2026_07_29
+ARG LANGUAGES="rus"
 RUN for lang in $LANGUAGES; do apt install -y tesseract-ocr-$(echo $lang | tr "_" "-"); done
 
 ENV PYTHONPATH "${PYTHONPATH}:/dedoc_root"
 ENV RESOURCES_PATH "/dedoc_root/resources"
 
 COPY requirements.txt .
 RUN pip3 install --no-cache-dir -r requirements.txt
+RUN pip3 install --no-cache-dir torch~=2.6.0 torchvision~=0.21.0 --index-url https://download.pytorch.org/whl/cpu
 RUN pip3 install --no-cache-dir transformers~=4.49.0
-RUN apt-get update && apt-get install -y --fix-missing --no-install-recommends fontforge
-RUN apt install -y libutf8proc-dev
-RUN ln -s /usr/lib/x86_64-linux-gnu/libutf8proc.so /usr/lib/libutf8proc.so.1
 
 RUN mkdir /dedoc_root
 RUN mkdir /dedoc_root/dedoc

dedoc/data_structures/tree_node.py

@@ -111,9 +111,19 @@ def add_text(self, line: LineWithMeta) -> None:
 
     @staticmethod
     def __shift_annotations(line: LineWithMeta, text_length: int) -> List[Annotation]:
+        from dedoc.data_structures.concrete_annotations.attach_annotation import AttachAnnotation
+        from dedoc.data_structures.concrete_annotations.table_annotation import TableAnnotation
+
         new_annotations = []
         for annotation in line.annotations:
-            new_annotation = Annotation(start=annotation.start + text_length, end=annotation.end + text_length, name=annotation.name, value=annotation.value)
+            start = annotation.start + text_length
+            end = annotation.end + text_length
+            if annotation.name == TableAnnotation.name:
+                new_annotation = TableAnnotation(value=annotation.value, start=start, end=end)
+            elif annotation.name == AttachAnnotation.name:
+                new_annotation = AttachAnnotation(attach_uid=annotation.value, start=start, end=end)
+            else:
+                new_annotation = Annotation(start=start, end=end, name=annotation.name, value=annotation.value, is_mergeable=annotation.is_mergeable)
             new_annotations.append(new_annotation)
         return new_annotations
 

dedoc/extensions.py

@@ -46,8 +46,9 @@
     archive_like_format={},
     image_like_format={
         "image/bmp", "image/x-ms-bmp", "image/dib", "image/x-eps", "application/postscript", "image/gif", "image/jpeg", "image/jp2", "image/x-jp2", "image/jpg",
-        "image/x-portable-pixmap", "image/x-portable-anymap", "image/x-portable-graymap", "image/x-portable-bitmap", "image/x-pcx", "image/x-pict",
-        "image/ras", "image/x-sun-raster", "image/sgi", "image/x-sgi", "image/tiff", "image/webp", "image/x-cmu-raster", "image/x-jp2-codestream"
+        "image/x-portable-pixmap", "image/x-portable-anymap", "image/x-portable-graymap", "image/x-portable-bitmap", "image/x-pcx", "image/vnd.zbrush.pcx",
+        "image/x-pict", "image/ras", "image/x-sun-raster", "image/sgi", "image/x-sgi", "image/tiff", "image/webp", "image/x-cmu-raster",
+        "image/x-jp2-codestream"
     },
     pdf_like_format={"image/vnd.djvu"},
     csv_like_format={},
@@ -79,7 +80,7 @@
     mhtml_like_format={"message/rfc822"},
     archive_like_format={
         "application/zip", "application/x-gzip", "application/gzip", "application/x-tar", "application/x-rar-compressed", "application/rar",
-        "application/x-rar", "application/x-7z-compressed"
+        "application/vnd.rar", "application/x-rar", "application/x-7z-compressed"
     },
     image_like_format={"image/png"},
     pdf_like_format={"application/pdf"},
@@ -105,13 +106,13 @@
     "message/rfc822": ".eml",
     # archive
     "application/x-gzip": ".gz", "application/gzip": ".gz", "application/x-rar-compressed": ".rar", "application/rar": ".rar", "application/x-tar": ".tar",
-    "application/x-rar": ".rar", "application/x-7z-compressed": ".7z", "application/zip": ".zip",
+    "application/vnd.rar": ".rar", "application/x-rar": ".rar", "application/x-7z-compressed": ".7z", "application/zip": ".zip",
     # image
     "image/x-ms-bmp": ".bmp", "image/sgi": ".sgi", "image/jpeg": ".jpg", "image/x-jp2": ".jp2", "image/x-portable-graymap": ".pgm", "image/x-pict": ".pic",
     "image/gif": ".gif", "image/x-sgi": ".sgi", "image/x-sun-raster": ".ras", "image/x-eps": ".eps", "image/bmp": ".bmp", "image/x-portable-bitmap": ".pbm",
     "image/webp": ".webp", "application/postscript": ".eps", "image/jpg": ".jpg", "image/x-jp2-codestream": "jp2", "image/x-portable-pixmap": ".ppm",
-    "image/jp2": ".jp2", "image/dib": ".dib", "image/x-pcx": ".pcx", "image/tiff": ".tiff", "image/x-cmu-raster": ".ras", "image/ras": ".ras",
-    "image/x-portable-anymap": ".pnm", "image/png": ".png",
+    "image/jp2": ".jp2", "image/dib": ".dib", "image/x-pcx": ".pcx", "image/vnd.zbrush.pcx": ".pcx", "image/tiff": ".tiff", "image/x-cmu-raster": ".ras",
+    "image/ras": ".ras", "image/x-portable-anymap": ".pnm", "image/png": ".png",
     # pdf
     "image/vnd.djvu": ".djvu", "application/pdf": ".pdf",
     # csv

docker_gpu/Dockerfile

@@ -1,25 +1,23 @@
-ARG REPOSITORY="docker.io"
-FROM dedocproject/dedoc_jammy_p3.10_base:version_2025_09_11
+FROM dedocproject/dedoc_p3.10_base:version_2026_07_29
+ARG LANGUAGES="rus"
+RUN for lang in $LANGUAGES; do apt install -y tesseract-ocr-$(echo $lang | tr "_" "-"); done
 
 ENV PYTHONPATH "${PYTHONPATH}:/dedoc_root"
 ENV RESOURCES_PATH "/dedoc_root/resources"
 
-ADD requirements.txt .
+COPY requirements.txt .
 RUN pip3 install --no-cache-dir -r requirements.txt
 RUN pip install torch==1.11.0+cu113 torchvision==0.12.0+cu113 -f https://download.pytorch.org/whl/torch_stable.html
 RUN pip3 install --no-cache-dir transformers~=4.49.0
 
 RUN mkdir /dedoc_root
 RUN mkdir /dedoc_root/dedoc
-ADD dedoc/config.py /dedoc_root/dedoc/config.py
-ADD dedoc/download_models.py /dedoc_root/dedoc/download_models.py
+COPY dedoc/config.py /dedoc_root/dedoc/config.py
+COPY dedoc/download_models.py /dedoc_root/dedoc/download_models.py
 RUN python3 /dedoc_root/dedoc/download_models.py
 
-ADD dedoc /dedoc_root/dedoc
-ADD VERSION /dedoc_root
+COPY dedoc /dedoc_root/dedoc
+COPY VERSION /dedoc_root
 RUN echo "__version__ = \"$(cat /dedoc_root/VERSION)\"" > /dedoc_root/dedoc/version.py
 
-ADD tests /dedoc_root/tests
-ADD resources /dedoc_root/resources
-
-CMD ["python3", "/dedoc_root/dedoc/main.py"]
+CMD ["python3", "/dedoc_root/dedoc/main.py"]

labeling/Dockerfile

@@ -1,11 +1,12 @@
-ARG REPOSITORY="docker.io"
-FROM dedocproject/dedoc_jammy_p3.10_base:version_2025_09_11
+FROM dedocproject/dedoc_p3.10_base:version_2026_07_29
+RUN apt-get install -y tesseract-ocr-rus
 
 ENV PYTHONPATH "${PYTHONPATH}:/labeling_root:/labeling_root/labeling"
 ENV RESOURCES_PATH "/labeling_root/resources"
 
 ADD requirements.txt .
 RUN pip3 install --no-cache-dir -r requirements.txt
+RUN pip3 install --no-cache-dir torch~=2.6.0 torchvision~=0.21.0 --index-url https://download.pytorch.org/whl/cpu
 RUN pip3 install Jinja2==3.1.2
 
 RUN mkdir /labeling_root

requirements.txt

@@ -8,7 +8,6 @@ huggingface-hub>=0.14.1,<1.0
 imutils==0.5.4
 itsdangerous>=2.1.0,<=2.1.2
 levenshtein>=0.12.2,<1
-numpy>=1.22.0,<2.0
 olefile~=0.46
 opencv-python>=4.5.5.64,<4.6.0
 orjson>=3.8.11,<4.0
@@ -22,7 +21,6 @@ py7zr~=1.0
 pypdf>=4.0,<6.0
 pytesseract==0.3.10
 python-docx==0.8.11
-python-logstash-async>=2.5.0,<=2.7.0
 python-magic<1.0
 python-multipart==0.0.6
 rarfile==4.0

tests/Dockerfile

@@ -1,13 +1,13 @@
-ARG REPOSITORY="docker.io"
-FROM dedocproject/dedoc_jammy_p3.10_base:version_2025_09_11
+FROM dedocproject/dedoc_p3.10_base:version_2026_07_29
+RUN apt-get install -y tesseract-ocr-rus
 
 ENV PYTHONPATH "${PYTHONPATH}:/dedoc_root:/dedoc_root/tests:/dedoc_root/langchain"
 
 COPY requirements.txt .
 RUN pip3 install --no-cache-dir "langchain-community<1.0"
 RUN pip3 install --no-cache-dir -r requirements.txt
+RUN pip3 install --no-cache-dir torch~=2.6.0 torchvision~=0.21.0 --index-url https://download.pytorch.org/whl/cpu
 RUN pip3 install --no-cache-dir transformers~=4.49.0
-RUN apt-get update && apt-get install -y --fix-missing --no-install-recommends fontforge
 
 RUN mkdir /dedoc_root
 COPY docs/source/_static/code_examples/langchain /dedoc_root/langchain

tests/unit_tests/test_misc_tree_node.py

@@ -2,10 +2,12 @@
 
 from dedoc.data_structures.concrete_annotations.bold_annotation import BoldAnnotation
 from dedoc.data_structures.concrete_annotations.italic_annotation import ItalicAnnotation
+from dedoc.data_structures.concrete_annotations.table_annotation import TableAnnotation
 from dedoc.data_structures.hierarchy_level import HierarchyLevel
 from dedoc.data_structures.line_metadata import LineMetadata
 from dedoc.data_structures.line_with_meta import LineWithMeta
 from dedoc.data_structures.tree_node import TreeNode
+from dedoc.utils.annotation_merger import AnnotationMerger
 
 
 class TestTreeNode(TestCase):
@@ -34,3 +36,21 @@ def test_root_node_annotations(self) -> None:
         self.assertEqual("True", italic.value)
         self.assertEqual(10, italic.start)
         self.assertEqual(22, italic.end)
+
+    def test_add_text_keeps_multiple_tables_after_merge(self) -> None:
+        """Two tables on a glued line must both survive add_text and AnnotationMerger."""
+        hl = HierarchyLevel.create_raw_text()
+        first = LineWithMeta(line="Начало ", metadata=LineMetadata(hierarchy_level=hl, page_id=0, line_id=0))
+        tables = [TableAnnotation(value="t1", start=0, end=5), TableAnnotation(value="t2", start=0, end=5)]
+        second = LineWithMeta(line="текст", metadata=LineMetadata(hierarchy_level=hl, page_id=0, line_id=1), annotations=tables)
+        node = TreeNode.create(lines=[first])
+        node.add_text(second)
+
+        table_anns = [ann for ann in node.annotations if ann.name == TableAnnotation.name]
+        self.assertEqual(2, len(table_anns))
+        self.assertTrue(all(isinstance(ann, TableAnnotation) for ann in table_anns))
+        self.assertTrue(all(ann.is_mergeable is False for ann in table_anns))
+
+        merged = AnnotationMerger().merge_annotations(node.annotations, node.text)
+        merged_tables = [ann for ann in merged if ann.name == TableAnnotation.name]
+        self.assertEqual({"t1", "t2"}, {ann.value for ann in merged_tables})