broken inference source code for 'hi_res', AttributeError: 'list' object has no attribute 'element_coords', the same code worked with previous versions of unstructured

[Arslan-Mehmood1]

[<ipython-input-2-808188905a87>](https://localhost:8080/#) in extract_data_from_pdf(pdf_path)
     57 # Function to extract text using the unstructured library
     58 def extract_data_from_pdf(pdf_path):
---> 59     elements = partition_pdf(filename=pdf_path, strategy='hi_res')
     60     text_data = ' '.join([str(el) for el in elements])  # Concatenate all elements
     61     return text_data

[/usr/local/lib/python3.10/dist-packages/unstructured/documents/elements.py](https://localhost:8080/#) in wrapper(*args, **kwargs)
    576         @functools.wraps(func)
    577         def wrapper(*args: _P.args, **kwargs: _P.kwargs) -> list[Element]:
--> 578             elements = func(*args, **kwargs)
    579             call_args = get_call_args_applying_defaults(func, *args, **kwargs)
    580 

[/usr/local/lib/python3.10/dist-packages/unstructured/file_utils/filetype.py](https://localhost:8080/#) in wrapper(*args, **kwargs)
    723         @functools.wraps(func)
    724         def wrapper(*args: _P.args, **kwargs: _P.kwargs) -> list[Element]:
--> 725             elements = func(*args, **kwargs)
    726 
    727             for element in elements:

[/usr/local/lib/python3.10/dist-packages/unstructured/file_utils/filetype.py](https://localhost:8080/#) in wrapper(*args, **kwargs)
    681     @functools.wraps(func)
    682     def wrapper(*args: _P.args, **kwargs: _P.kwargs) -> list[Element]:
--> 683         elements = func(*args, **kwargs)
    684         call_args = get_call_args_applying_defaults(func, *args, **kwargs)
    685 

[/usr/local/lib/python3.10/dist-packages/unstructured/chunking/dispatch.py](https://localhost:8080/#) in wrapper(*args, **kwargs)
     72 
     73         # -- call the partitioning function to get the elements --
---> 74         elements = func(*args, **kwargs)
     75 
     76         # -- look for a chunking-strategy argument --

[/usr/local/lib/python3.10/dist-packages/unstructured/partition/pdf.py](https://localhost:8080/#) in partition_pdf(filename, file, include_page_breaks, strategy, infer_table_structure, ocr_languages, languages, metadata_filename, metadata_last_modified, chunking_strategy, hi_res_model_name, extract_images_in_pdf, extract_image_block_types, extract_image_block_output_dir, extract_image_block_to_payload, starting_page_number, extract_forms, form_extraction_skip_tables, **kwargs)
    198     languages = check_language_args(languages or [], ocr_languages)
    199 
--> 200     return partition_pdf_or_image(
    201         filename=filename,
    202         file=file,

[/usr/local/lib/python3.10/dist-packages/unstructured/partition/pdf.py](https://localhost:8080/#) in partition_pdf_or_image(filename, file, is_image, include_page_breaks, strategy, infer_table_structure, languages, metadata_last_modified, hi_res_model_name, extract_images_in_pdf, extract_image_block_types, extract_image_block_output_dir, extract_image_block_to_payload, starting_page_number, extract_forms, form_extraction_skip_tables, **kwargs)
    294         with warnings.catch_warnings():
    295             warnings.simplefilter("ignore")
--> 296             elements = _partition_pdf_or_image_local(
    297                 filename=filename,
    298                 file=spooled_to_bytes_io_if_needed(file),

[/usr/local/lib/python3.10/dist-packages/unstructured/utils.py](https://localhost:8080/#) in wrapper(*args, **kwargs)
    215         def wrapper(*args: _P.args, **kwargs: _P.kwargs):
    216             run_check()
--> 217             return func(*args, **kwargs)
    218 
    219         @wraps(func)

[/usr/local/lib/python3.10/dist-packages/unstructured/partition/pdf.py](https://localhost:8080/#) in _partition_pdf_or_image_local(filename, file, is_image, infer_table_structure, include_page_breaks, languages, ocr_languages, ocr_mode, model_name, hi_res_model_name, pdf_image_dpi, metadata_last_modified, pdf_text_extractable, extract_images_in_pdf, extract_image_block_types, extract_image_block_output_dir, extract_image_block_to_payload, analysis, analyzed_image_output_dir_path, starting_page_number, extract_forms, form_extraction_skip_tables, pdf_hi_res_max_pages, **kwargs)
    625             )
    626 
--> 627             final_document_layout = process_file_with_ocr(
    628                 filename,
    629                 merged_document_layout,

[/usr/local/lib/python3.10/dist-packages/unstructured/utils.py](https://localhost:8080/#) in wrapper(*args, **kwargs)
    215         def wrapper(*args: _P.args, **kwargs: _P.kwargs):
    216             run_check()
--> 217             return func(*args, **kwargs)
    218 
    219         @wraps(func)

[/usr/local/lib/python3.10/dist-packages/unstructured/partition/pdf_image/ocr.py](https://localhost:8080/#) in process_file_with_ocr(filename, out_layout, extracted_layout, is_image, infer_table_structure, ocr_languages, ocr_mode, pdf_image_dpi, ocr_layout_dumper)
    176     except Exception as e:
    177         if os.path.isdir(filename) or os.path.isfile(filename):
--> 178             raise e
    179         else:
    180             raise FileNotFoundError(f'File "{filename}" not found!') from e

[/usr/local/lib/python3.10/dist-packages/unstructured/partition/pdf_image/ocr.py](https://localhost:8080/#) in process_file_with_ocr(filename, out_layout, extracted_layout, is_image, infer_table_structure, ocr_languages, ocr_mode, pdf_image_dpi, ocr_layout_dumper)
    163                     extracted_regions = extracted_layout[i] if i < len(extracted_layout) else None
    164                     with PILImage.open(image_path) as image:
--> 165                         merged_page_layout = supplement_page_layout_with_ocr(
    166                             page_layout=out_layout.pages[i],
    167                             image=image,

[/usr/local/lib/python3.10/dist-packages/unstructured/utils.py](https://localhost:8080/#) in wrapper(*args, **kwargs)
    215         def wrapper(*args: _P.args, **kwargs: _P.kwargs):
    216             run_check()
--> 217             return func(*args, **kwargs)
    218 
    219         @wraps(func)

[/usr/local/lib/python3.10/dist-packages/unstructured/partition/pdf_image/ocr.py](https://localhost:8080/#) in supplement_page_layout_with_ocr(page_layout, image, infer_table_structure, ocr_languages, ocr_mode, extracted_regions, ocr_layout_dumper)
    204         if ocr_layout_dumper:
    205             ocr_layout_dumper.add_ocred_page(ocr_layout)
--> 206         page_layout.elements[:] = merge_out_layout_with_ocr_layout(
    207             out_layout=cast(List["LayoutElement"], page_layout.elements),
    208             ocr_layout=ocr_layout,

[/usr/local/lib/python3.10/dist-packages/unstructured/partition/pdf_image/ocr.py](https://localhost:8080/#) in merge_out_layout_with_ocr_layout(out_layout, ocr_layout, supplement_with_ocr_elements)
    359 
    360     final_layout = (
--> 361         supplement_layout_with_ocr_elements(out_layout, ocr_layout)
    362         if supplement_with_ocr_elements
    363         else out_layout

[/usr/local/lib/python3.10/dist-packages/unstructured/utils.py](https://localhost:8080/#) in wrapper(*args, **kwargs)
    215         def wrapper(*args: _P.args, **kwargs: _P.kwargs):
    216             run_check()
--> 217             return func(*args, **kwargs)
    218 
    219         @wraps(func)

[/usr/local/lib/python3.10/dist-packages/unstructured/partition/pdf_image/ocr.py](https://localhost:8080/#) in supplement_layout_with_ocr_elements(layout, ocr_layout, subregion_threshold)
    438     ocr_regions_to_add = [region for region in ocr_layout if region not in ocr_regions_to_remove]
    439     if ocr_regions_to_add:
--> 440         ocr_elements_to_add = build_layout_elements_from_ocr_regions(ocr_regions_to_add)
    441         final_layout = layout + ocr_elements_to_add
    442     else:

[/usr/local/lib/python3.10/dist-packages/unstructured/partition/pdf_image/inference_utils.py](https://localhost:8080/#) in build_layout_elements_from_ocr_regions(ocr_regions, ocr_text, group_by_ocr_text)
     69             grouped_regions.append(regions)
     70     else:
---> 71         grouped_regions = partition_groups_from_regions(ocr_regions)
     72 
     73     merged_regions = [merge_text_regions(group) for group in grouped_regions]

[/usr/local/lib/python3.10/dist-packages/unstructured_inference/inference/layoutelement.py](https://localhost:8080/#) in partition_groups_from_regions(regions)
    316     if len(regions) == 0:
    317         return []
--> 318     padded_coords = regions.element_coords.copy()
    319     v_pad = (regions.y2 - regions.y1) * inference_config.ELEMENTS_V_PADDING_COEF
    320     h_pad = (regions.x2 - regions.x1) * inference_config.ELEMENTS_H_PADDING_COEF

AttributeError: 'list' object has no attribute 'element_coords'

[amysudarat]

What version that is still working? Thank you @Arslan-Mehmood1

[christinestraub]

What versions of unstructured and unstructured-inference libraries are you using?

[theogiraudon]

I've got the same error with unstructured@0.15.14 and unstructured-inference@0.7.40

[Qwedon]

The problem occurs from unstructured-inference@0.7.37.
With unstructured-inference@0.7.36 and below everything works.

Snippet of code in which error occurs:

from unstructured.partition.pdf import partition_pdf
filename = "your_pdf.pdf"

elements = partition_pdf(
    filename=filename,
    strategy="hi_res", 
    infer_table_structure=True, 
    model_name="yolox"
)

[chung-codes]

I've got the same error with unstructured@0.15.14 and unstructured-inference@0.7.40

i have same issue with same condition

[Arslan-Mehmood1]

What version that is still working? Thank you @Arslan-Mehmood1
@amysudarat amysudarat

unstructured==0.15.10
unstructured_inference==0.7.36

Actually, the problem was with unstructured_inference. I traced back to the time the code worked for me and then I check the available version of unstructured_inference that was released around that time.

[Arslan-Mehmood1]

What versions of unstructured and unstructured-inference libraries are you using?
@christinestraub
Code working with following:
unstructured==0.15.10
unstructured_inference==0.7.36

[Arslan-Mehmood1]

I've got the same error with unstructured@0.15.14 and unstructured-inference@0.7.40
@theogiraudon

Code working with following:
unstructured==0.15.10
unstructured_inference==0.7.36

[Arslan-Mehmood1]

The problem occurs from unstructured-inference@0.7.37. With unstructured-inference@0.7.36 and below everything works.

Snippet of code in which error occurs:

from unstructured.partition.pdf import partition_pdf
filename = "your_pdf.pdf"

elements = partition_pdf(
    filename=filename,
    strategy="hi_res", 
    infer_table_structure=True, 
    model_name="yolox"
)

@Qwedon
Code working with following:
unstructured==0.15.10
unstructured_inference==0.7.36

[Arslan-Mehmood1]

I've got the same error with unstructured@0.15.14 and unstructured-inference@0.7.40

i have same issue with same condition

@chung-codes
Code working with following:
unstructured==0.15.10
unstructured_inference==0.7.36

[inderjeetsingh1704]

i faced the same issue with the following versions.
unstructured==0.15.14
unstructured-inference==0.7.41

Reverted it back to which is working.
unstructured==0.15.10
unstructured-inference==0.7.36

[YunghuiHsu]

I have same issue with same condition:

Problem occurs with following:
unstructured==0.15.14
unstructured_inference==0.8.0

Code worked in :
unstructured==0.15.9
unstructured_inference==0.7.36

[simaotwx]

I'm getting AttributeError: 'list' object has no attribute 'element_coords' for

        partitioned = partition_pdf(
            filename=file,
            strategy=self.args.doc_load_strategy,
            extract_images_in_pdf=False,
            languages=["deu", "eng"],
            password="",
            kwargs={
                "check_extractable": False
            }
        )

My versions:

[[package]]
name = "unstructured"
version = "0.14.10"

[[package]]
name = "unstructured-inference"
version = "0.8.1"

I was using 0.14.10 because of another issue, so I'm upgrading to 0.16.5 to see if it helps.

[nlrahimi]

I am having problems that I didn't have two weeks ago.
Floating text misclassification: Floating text elements are being incorrectly recognized and classified as tables. element['type'] is parsed incorrectly. This is a new behavior that didn't occur in previous versions of Unstructured.

Floating text merging: Floating text (such as headers or side notes) is not being processed as individual elements. Instead, it's being merged into the same paragraph with nearby content. As you can see in attached image.
unstructured ==0.16.3
unstructured-client == 0.26.1
unstructured-inference == 0.8.1
unstructured-ingest == 0.1.1
unstructured.pytesseract == 0.3.13

Parsed document, where the floating text is underlined: