refactor: organize models in submodules

docling/experimental/pipeline/threaded_layout_vlm_pipeline.py

@@ -32,13 +32,13 @@
 from docling.experimental.datamodel.threaded_layout_vlm_pipeline_options import (
     ThreadedLayoutVlmPipelineOptions,
 )
-from docling.models.api_vlm_model import ApiVlmModel
 from docling.models.base_model import BaseVlmPageModel
-from docling.models.layout_model import LayoutModel
-from docling.models.vlm_models_inline.hf_transformers_model import (
+from docling.models.stages.layout.layout_model import LayoutModel
+from docling.models.vlm_pipeline_models.api_vlm_model import ApiVlmModel
+from docling.models.vlm_pipeline_models.hf_transformers_model import (
     HuggingFaceTransformersVlmModel,
 )
-from docling.models.vlm_models_inline.mlx_model import HuggingFaceMlxModel
+from docling.models.vlm_pipeline_models.mlx_model import HuggingFaceMlxModel
 from docling.pipeline.base_pipeline import BasePipeline
 from docling.pipeline.standard_pdf_pipeline import (
     ProcessingResult,
@@ -162,7 +162,7 @@ def build_prompt(
                     vlm_options=vlm_options,
                 )
             elif vlm_options.inference_framework == InferenceFramework.VLLM:
-                from docling.models.vlm_models_inline.vllm_model import VllmVlmModel
+                from docling.models.vlm_pipeline_models.vllm_model import VllmVlmModel
 
                 self.vlm_model = VllmVlmModel(
                     enabled=True,

docling/models/plugins/defaults.py

@@ -1,10 +1,10 @@
 def ocr_engines():
-    from docling.models.auto_ocr_model import OcrAutoModel
-    from docling.models.easyocr_model import EasyOcrModel
-    from docling.models.ocr_mac_model import OcrMacModel
-    from docling.models.rapid_ocr_model import RapidOcrModel
-    from docling.models.tesseract_ocr_cli_model import TesseractOcrCliModel
-    from docling.models.tesseract_ocr_model import TesseractOcrModel
+    from docling.models.stages.ocr.auto_ocr_model import OcrAutoModel
+    from docling.models.stages.ocr.easyocr_model import EasyOcrModel
+    from docling.models.stages.ocr.ocr_mac_model import OcrMacModel
+    from docling.models.stages.ocr.rapid_ocr_model import RapidOcrModel
+    from docling.models.stages.ocr.tesseract_ocr_cli_model import TesseractOcrCliModel
+    from docling.models.stages.ocr.tesseract_ocr_model import TesseractOcrModel
 
     return {
         "ocr_engines": [
@@ -19,8 +19,12 @@ def ocr_engines():
 
 
 def picture_description():
-    from docling.models.picture_description_api_model import PictureDescriptionApiModel
-    from docling.models.picture_description_vlm_model import PictureDescriptionVlmModel
+    from docling.models.stages.picture_description.picture_description_api_model import (
+        PictureDescriptionApiModel,
+    )
+    from docling.models.stages.picture_description.picture_description_vlm_model import (
+        PictureDescriptionVlmModel,
+    )
 
     return {
         "picture_description": [
@@ -34,7 +38,7 @@ def layout_engines():
     from docling.experimental.models.table_crops_layout_model import (
         TableCropsLayoutModel,
     )
-    from docling.models.layout_model import LayoutModel
+    from docling.models.stages.layout.layout_model import LayoutModel
 
     return {
         "layout_engines": [
@@ -45,7 +49,9 @@ def layout_engines():
 
 
 def table_structure_engines():
-    from docling.models.table_structure_model import TableStructureModel
+    from docling.models.stages.table_structure.table_structure_model import (
+        TableStructureModel,
+    )
 
     return {
         "table_structure_engines": [

docling/models/auto_ocr_model.py → docling/models/stages/ocr/auto_ocr_model.py

@@ -15,9 +15,9 @@
     RapidOcrOptions,
 )
 from docling.models.base_ocr_model import BaseOcrModel
-from docling.models.easyocr_model import EasyOcrModel
-from docling.models.ocr_mac_model import OcrMacModel
-from docling.models.rapid_ocr_model import RapidOcrModel
+from docling.models.stages.ocr.easyocr_model import EasyOcrModel
+from docling.models.stages.ocr.ocr_mac_model import OcrMacModel
+from docling.models.stages.ocr.rapid_ocr_model import RapidOcrModel
 
 _log = logging.getLogger(__name__)
 

docling/models/page_assemble_model.py → docling/models/stages/page_assemble/page_assemble_model.py

@@ -17,7 +17,7 @@
 )
 from docling.datamodel.document import ConversionResult
 from docling.models.base_model import BasePageModel
-from docling.models.layout_model import LayoutModel
+from docling.models.stages.layout.layout_model import LayoutModel
 from docling.utils.profiling import TimeRecorder
 
 _log = logging.getLogger(__name__)

docling/pipeline/base_pipeline.py

@@ -28,12 +28,12 @@
 )
 from docling.datamodel.settings import settings
 from docling.models.base_model import GenericEnrichmentModel
-from docling.models.document_picture_classifier import (
+from docling.models.factories import get_picture_description_factory
+from docling.models.picture_description_base_model import PictureDescriptionBaseModel
+from docling.models.stages.picture_classifier.document_picture_classifier import (
     DocumentPictureClassifier,
     DocumentPictureClassifierOptions,
 )
-from docling.models.factories import get_picture_description_factory
-from docling.models.picture_description_base_model import PictureDescriptionBaseModel
 from docling.utils.profiling import ProfilingScope, TimeRecorder
 from docling.utils.utils import chunkify
 

docling/pipeline/extraction_vlm_pipeline.py

@@ -20,7 +20,7 @@
     VlmExtractionPipelineOptions,
 )
 from docling.datamodel.settings import settings
-from docling.models.vlm_models_inline.nuextract_transformers_model import (
+from docling.models.extraction.nuextract_transformers_model import (
     NuExtractTransformersModel,
 )
 from docling.pipeline.base_extraction_pipeline import BaseExtractionPipeline

docling/pipeline/legacy_standard_pdf_pipeline.py

@@ -14,18 +14,27 @@
 from docling.datamodel.pipeline_options import PdfPipelineOptions
 from docling.datamodel.settings import settings
 from docling.models.base_ocr_model import BaseOcrModel
-from docling.models.code_formula_model import CodeFormulaModel, CodeFormulaModelOptions
 from docling.models.factories import (
     get_layout_factory,
     get_ocr_factory,
     get_table_structure_factory,
 )
-from docling.models.page_assemble_model import PageAssembleModel, PageAssembleOptions
-from docling.models.page_preprocessing_model import (
+from docling.models.stages.code_formula.code_formula_model import (
+    CodeFormulaModel,
+    CodeFormulaModelOptions,
+)
+from docling.models.stages.page_assemble.page_assemble_model import (
+    PageAssembleModel,
+    PageAssembleOptions,
+)
+from docling.models.stages.page_preprocessing.page_preprocessing_model import (
     PagePreprocessingModel,
     PagePreprocessingOptions,
 )
-from docling.models.readingorder_model import ReadingOrderModel, ReadingOrderOptions
+from docling.models.stages.reading_order.readingorder_model import (
+    ReadingOrderModel,
+    ReadingOrderOptions,
+)
 from docling.pipeline.base_pipeline import PaginatedPipeline
 from docling.utils.model_downloader import download_models
 from docling.utils.profiling import ProfilingScope, TimeRecorder

docling/pipeline/standard_pdf_pipeline.py

@@ -40,18 +40,27 @@
 from docling.datamodel.document import ConversionResult
 from docling.datamodel.pipeline_options import ThreadedPdfPipelineOptions
 from docling.datamodel.settings import settings
-from docling.models.code_formula_model import CodeFormulaModel, CodeFormulaModelOptions
 from docling.models.factories import (
     get_layout_factory,
     get_ocr_factory,
     get_table_structure_factory,
 )
-from docling.models.page_assemble_model import PageAssembleModel, PageAssembleOptions
-from docling.models.page_preprocessing_model import (
+from docling.models.stages.code_formula.code_formula_model import (
+    CodeFormulaModel,
+    CodeFormulaModelOptions,
+)
+from docling.models.stages.page_assemble.page_assemble_model import (
+    PageAssembleModel,
+    PageAssembleOptions,
+)
+from docling.models.stages.page_preprocessing.page_preprocessing_model import (
     PagePreprocessingModel,
     PagePreprocessingOptions,
 )
-from docling.models.readingorder_model import ReadingOrderModel, ReadingOrderOptions
+from docling.models.stages.reading_order.readingorder_model import (
+    ReadingOrderModel,
+    ReadingOrderOptions,
+)
 from docling.pipeline.base_pipeline import ConvertPipeline
 from docling.utils.profiling import ProfilingScope, TimeRecorder
 from docling.utils.utils import chunkify

docling/pipeline/vlm_pipeline.py

@@ -37,11 +37,11 @@
     ResponseFormat,
 )
 from docling.datamodel.settings import settings
-from docling.models.api_vlm_model import ApiVlmModel
-from docling.models.vlm_models_inline.hf_transformers_model import (
+from docling.models.vlm_pipeline_models.api_vlm_model import ApiVlmModel
+from docling.models.vlm_pipeline_models.hf_transformers_model import (
     HuggingFaceTransformersVlmModel,
 )
-from docling.models.vlm_models_inline.mlx_model import HuggingFaceMlxModel
+from docling.models.vlm_pipeline_models.mlx_model import HuggingFaceMlxModel
 from docling.pipeline.base_pipeline import PaginatedPipeline
 from docling.utils.profiling import ProfilingScope, TimeRecorder
 
@@ -93,7 +93,7 @@ def __init__(self, pipeline_options: VlmPipelineOptions):
                     ),
                 ]
             elif vlm_options.inference_framework == InferenceFramework.VLLM:
-                from docling.models.vlm_models_inline.vllm_model import VllmVlmModel
+                from docling.models.vlm_pipeline_models.vllm_model import VllmVlmModel
 
                 self.build_pipe = [
                     VllmVlmModel(

docling/utils/model_downloader.py

@@ -15,13 +15,19 @@
     SMOLDOCLING_MLX,
     SMOLDOCLING_TRANSFORMERS,
 )
-from docling.models.code_formula_model import CodeFormulaModel
-from docling.models.document_picture_classifier import DocumentPictureClassifier
-from docling.models.easyocr_model import EasyOcrModel
-from docling.models.layout_model import LayoutModel
-from docling.models.picture_description_vlm_model import PictureDescriptionVlmModel
-from docling.models.rapid_ocr_model import RapidOcrModel
-from docling.models.table_structure_model import TableStructureModel
+from docling.models.stages.code_formula.code_formula_model import CodeFormulaModel
+from docling.models.stages.layout.layout_model import LayoutModel
+from docling.models.stages.ocr.easyocr_model import EasyOcrModel
+from docling.models.stages.ocr.rapid_ocr_model import RapidOcrModel
+from docling.models.stages.picture_classifier.document_picture_classifier import (
+    DocumentPictureClassifier,
+)
+from docling.models.stages.picture_description.picture_description_vlm_model import (
+    PictureDescriptionVlmModel,
+)
+from docling.models.stages.table_structure.table_structure_model import (
+    TableStructureModel,
+)
 from docling.models.utils.hf_model_download import download_hf_model
 
 _log = logging.getLogger(__name__)

docs/examples/enrich_doclingdocument.py

@@ -35,7 +35,7 @@
 from docling.datamodel.base_models import InputFormat, ItemAndImageEnrichmentElement
 from docling.datamodel.document import InputDocument
 from docling.models.base_model import BaseItemAndImageEnrichmentModel
-from docling.models.document_picture_classifier import (
+from docling.models.stages.picture_classifier.document_picture_classifier import (
     DocumentPictureClassifier,
     DocumentPictureClassifierOptions,
 )