This PR adds new capabilities for drawing bboxes for each layout
(extracted, inferred, ocr and final) + OD model output dump as a json
file for better analysis.
---------
Co-authored-by: Christine Straub <christinemstraub@gmail.com>
Co-authored-by: Michal Martyniak <michal.martyniak@deepsense.ai>
The Issue:
When extracting images from pdfs, we use the metadata page number to
index into a list of the images. However, the metadata page number can
now be changed via `starting_page_number`. To get the true page index,
we need to subtract this value.
Testing:
Run this snippet in a python shell. Before the fix, this throws an
IndexError. On this branch, it will return the elements.
```
from unstructured.partition.auto import partition
filename = "example-docs/layout-parser-paper-with-table.pdf"
partition(filename, strategy="hi_res", extract_image_block_types=["Image", "Table"], starting_page_number=20)
```
---------
Co-authored-by: Matt Robinson <mrobinson@unstructuredai.io>
Co-authored-by: christinestraub <christinemstraub@gmail.com>
### Summary
- bump unstructured-inference to `0.7.35` which fixed syntax for
generated HTML tables
- update unit tests and ingest test fixtures to reflect changes in the
generated HTML tables
- cut a release for `0.14.6`
---------
Co-authored-by: ryannikolaidis <1208590+ryannikolaidis@users.noreply.github.com>
Co-authored-by: christinestraub <christinestraub@users.noreply.github.com>
This PR aims to pass `kwargs` through `fast` strategy pipeline, which
was missing as part of the previous PR -
https://github.com/Unstructured-IO/unstructured/pull/3030.
I also did some code refactoring in this PR, so I recommend reviewing
this PR commit by commit.
### Summary
- pass `kwargs` through `fast` strategy pipeline, which will allow users
to specify additional params like `sort_mode`
- refactor: code reorganization
- cut a release for `0.14.0`
### Testing
CI should pass
This PR aims to skip element sorting when determining whether embedded
text can be extracted. The extracted elements in this step are returned
as final elements only for the `fast` strategy pipeline and are never
used for other strategy pipelines (`hi_res`, `ocr`).
Removing element sorting in this step and adding it to the `fast`
strategy pipeline later will improve performance and reduce execution
time.
### Summary
- skip element sorting when determining whether embedded text can be
extracted.
- add `_partition_pdf_with_pdfparser()` function for fast` strategy
pipeline
### Testing
CI should pass.
This PR attempts to fix a memory issue, which resulted in errors like
this: https://github.com/Unstructured-IO/unstructured/issues/2931
The root cause seems to be in how ListItems are being combined, not in
how hashes or parent IDs are updated.
When `assign_and_map_hash_ids()` is called and elements (or elements'
metadata) do not have unique memory addresses, then updating the
parent_id of one element will also overwrite the parent_id of some other
element.
---------
Co-authored-by: cragwolfe <crag@unstructured.io>
Part two of: https://github.com/Unstructured-IO/unstructured/pull/2842
Main changes compared to part one:
* hash computation includes element's sequence number on page, page
number, document filename and its text
* there are more test for deterministic behavior of IDs returned by
partitioning functions + their uniqueness (guaranteed at the document
level, and high probability across multiple documents)
This PR addresses the following issue:
https://github.com/Unstructured-IO/unstructured/issues/2461
Introduce `date_from_file_object` to `partition*` functions, by default
set to `False`.
If set to `True` and file is provided via `file` parameter, partition
will attempt to infer last modified date from `file`'s contents
otherwise last modified metadata will be set to `None`.
---------
Co-authored-by: Filip Knefel <filip@unstructured.io>
Co-authored-by: Ronny H <138828701+ron-unstructured@users.noreply.github.com>
### Summary
Detects headers and footers when using `partition_pdf` with the fast
strategy. Identifies elements that are positioned in the top or bottom
5% of the page as headers or footers. If no coordinate information is
available, an element won't be detected as a header or footer.
---------
Co-authored-by: ryannikolaidis <1208590+ryannikolaidis@users.noreply.github.com>
Co-authored-by: MthwRobinson <MthwRobinson@users.noreply.github.com>
This PR is the last in a series of PRs for refactoring and fixing the
language parameters (`languages` and `ocr_languages` so we can address
incorrect input by users. See #2293
It is recommended to go though this PR commit-by-commit and note the
commit message. The most significant commit is "update
check_languages..."
- there are multiple places setting the default `hi_res_model_name` in
both `unstructured` and `unstructured-inference`
- they lead to inconsistency and unexpected behaviors
- this fix removes a helper in `unstructured` that tries to set the
default hi_res layout detection model; instead we rely on the
`unstructured-inference` to provide that default when no explicit model
name is passed in
## test
```bash
UNSTRUCTURED_INCLUDE_DEBUG_METADATA=true ipython
```
```python
from unstructured.partition.auto import partition
# find a pdf file
elements = partition("foo.pdf", strategy="hi_res")
assert elements[0].metadata.detection_origin == "yolox"
```
---------
Co-authored-by: ryannikolaidis <1208590+ryannikolaidis@users.noreply.github.com>
Co-authored-by: badGarnet <badGarnet@users.noreply.github.com>
### Summary
The goal of this PR is to keep all image elements when using "hi_res"
strategy. Previously, `Image` elements with small chunks of text were
ignored unless the image block extraction parameters
(`extract_images_in_pdf` or `extract_image_block_types`) were specified.
Now, all image elements are kept regardless of whether the image block
extraction parameters are specified.
### Testing
- on `main` branch,
```
elements = partition_pdf(
filename="example-docs/embedded-images.pdf",
strategy="hi_res",
)
image_elements = [el for el in elements if el.category == ElementType.IMAGE]
print("number of image elements: ", len(image_elements))
```
The above code will display `number of image elements: 0`.
- on this `feature` branch,
The same code will display `number of image elements: 3`
---------
Co-authored-by: ryannikolaidis <1208590+ryannikolaidis@users.noreply.github.com>
Co-authored-by: christinestraub <christinestraub@users.noreply.github.com>
Currently, we're using different kwarg names in partition() and
partition_pdf(), which has implications for the API since it goes
through partition().
### Summary
- rename `extract_element_types` -> `extract_image_block_types`
- rename `image_output_dir_path` to `extract_image_block_output_dir`
- rename `extract_to_payload` -> `extract_image_block_to_payload`
- rename `pdf_extract_images` -> `extract_images_in_pdf` in
`partition.auto`
- add unit tests to test element extraction for `pdf/image` via
`partition.auto`
### Testing
CI should pass.
Closes#2323.
### Summary
- update logic to return "hi_res" if either `extract_images_in_pdf` or
`extract_element_types` is set
- refactor: remove unused `file` parameter from
`determine_pdf_or_image_strategy()`
### Testing
```
from unstructured.partition.pdf import partition_pdf
elements = partition_pdf(
filename="example-docs/embedded-images-tables.pdf",
extract_element_types=["Image"],
extract_to_payload=True,
)
image_elements = [el for el in elements if el.category == ElementType.IMAGE]
print(image_elements)
```
Closes#2302.
### Summary
- add functionality to get a Base64 encoded string from a PIL image
- store base64 encoded image data in two metadata fields: `image_base64`
and `image_mime_type`
- update the "image element filter" logic to keep all image elements in
the output if a user specifies image extraction
### Testing
```
from unstructured.partition.pdf import partition_pdf
elements = partition_pdf(
filename="example-docs/embedded-images-tables.pdf",
strategy="hi_res",
extract_element_types=["Image", "Table"],
extract_to_payload=True,
)
```
or
```
from unstructured.partition.auto import partition
elements = partition(
filename="example-docs/embedded-images-tables.pdf",
strategy="hi_res",
pdf_extract_element_types=["Image", "Table"],
pdf_extract_to_payload=True,
)
```
Closes#2160
Explicitly adds `hi_res_model_name` as kwarg to relevant functions and
notes that `model_name` is to be deprecated.
Testing:
```
from unstructured.partition.auto import partition
filename = "example-docs/DA-1p.pdf"
elements = partition(filename, strategy="hi_res", hi_res_model_name="yolox")
```
---------
Co-authored-by: cragwolfe <crag@unstructured.io>
Co-authored-by: Steve Canny <stcanny@gmail.com>
Co-authored-by: Christine Straub <christinemstraub@gmail.com>
Co-authored-by: Yao You <yao@unstructured.io>
Co-authored-by: Yao You <theyaoyou@gmail.com>
### Summary
This PR is the second part of `pdfminer` refactor to move it from
`unstructured-inference` repo to `unstructured` repo, the first part is
done in
https://github.com/Unstructured-IO/unstructured-inference/pull/294. This
PR adds logic to merge the extracted layout with the inferred layout.
The updated workflow for the `hi_res` strategy:
* pass the document (as data/filename) to the `inference` repo to get
`inferred_layout` (DocumentLayout)
* pass the `inferred_layout` returned from the `inference` repo and the
document (as data/filename) to the `pdfminer_processing` module, which
first opens the document (create temp file/dir as needed), and splits
the document by pages
* if is_image is `True`, return the passed
inferred_layout(DocumentLayout)
* if is_image is `False`:
* get extracted_layout (TextRegions) from the passed
document(data/filename) by pdfminer
* merge `extracted_layout` (TextRegions) with the passed
`inferred_layout` (DocumentLayout)
* return the `inferred_layout `(DocumentLayout) with updated elements
(all merged LayoutElements) as merged_layout (DocumentLayout)
* pass merged_layout and the document (as data/filename) to the `OCR`
module, which first opens the document (create temp file/dir as needed),
and splits the document by pages (convert PDF pages to image pages for
PDF file)
### Note
This PR also fixes issue #2164 by using functionality similar to the one
implemented in the `fast` strategy workflow when extracting elements by
`pdfminer`.
### TODO
* image extraction refactor to move it from `unstructured-inference`
repo to `unstructured` repo
* improving natural reading order by applying the current default
`xycut` sorting to the elements extracted by `pdfminer`
### Summary
Add a procedure to repair PDF when the PDF structure is invalid for
`PDFminer` to process.
This PR handles two cases of `PSSyntaxError Invalid dictionary
construct: ...`:
* PDFminer open entire document and create pages generator on
`PDFPage.get_pages(fp)`: [sentry log
example](https://unstructuredio.sentry.io/issues/4655715023/?alert_rule_id=14681339&alert_type=issue¬ification_uuid=d8db4cf4-686f-4504-8a22-74a79a8e966f&project=4505909127086080&referrer=slack)
* PDFminer's interpreter process a single page on
`interpreter.process_page(page)`: [sentry log
example](https://unstructuredio.sentry.io/issues/4655898781/?referrer=slack¬ification_uuid=0d929d48-f490-4db8-8dad-5d431c8460bc&alert_rule_id=14681339&alert_type=issue)
**Additional tech details:**
* Add new dependency `pikepdf` in `requirements/extra-pdf-image.in`,
which is used for repairing PDF.
* Add new denpendenct `pypdf` in `requirements/extra-pdf-image.in`,
which is used to find the error page from entire document by reading the
PDF file again (can't find a way to split pdf in PDFminer).
* Refactor the `is null` check for `get_uris_from_annots`, since the
root cause is that `get_uris` passed a None `annots` to
`get_uris_from_annots`, so the Null check should happen in `get_uris`.
* Add more type protection in `get_uris_from_annots` when using any
`PDFObjRef.resolve()` as `dict` (it could still be a `PDFObjRef`). This
should fix :
* https://github.com/Unstructured-IO/unstructured/issues/1922 where
`annotation_dict` is a `PDFObjRef`
* https://github.com/Unstructured-IO/unstructured/issues/1921 where
`rect` is a `PDFObjRef`
### Test
Added three test files (both are larger than 500 KB) for unittests to
test:
* Repair entire doc
* Repair one page
* Reprocess failure after repairing one page (just return the elements
before error page in this case).
* Also seems like splitting the document into smaller pages could fix
this problem, but not sure why. For example, I saw error from reprocess
in the whole
[cancer.pdf](https://github.com/Unstructured-IO/unstructured/files/13461616/cancer.pdf)
doc, but no error when i split the pdf by error page....
* tested if i can repair the entire doc again in this case, saw other
error which means repairing is not helping imo
* PDFminer can process the whole doc after pikepdf only repaired the
entire doc in the first place, but we can't repair by pages in this way
---------
Co-authored-by: cragwolfe <crag@unstructured.io>
Closes#2059.
We've found some pdfs that throw an error in pdfminer. These files use a
ICCBased color profile but do not include an expected value `N`. As a
workaround, we can wrap pdfminer and drop any colorspace info, since we
don't need to render the document.
To verify, try to partition the document in the linked issue.
```
elements = partition(filename="google-2023-environmental-report_condensed.pdf", strategy="fast")
```
---------
Co-authored-by: cragwolfe <crag@unstructured.io>
Closes#2038.
### Summary
The `fast` strategy should not fall back to a more expensive strategy.
### Testing
For
[9493801-p17.pdf](https://github.com/Unstructured-IO/unstructured/files/13292884/9493801-p17.pdf),
the following code should return an empty list.
```
elements = partition(filename=filename, strategy="fast")
```
---------
Co-authored-by: ryannikolaidis <1208590+ryannikolaidis@users.noreply.github.com>
Co-authored-by: christinestraub <christinestraub@users.noreply.github.com>
### Summary
Closes#2011
`languages` was missing from the metadata when partitioning pdfs via
`hi_res` and `fast` strategies and missing from image partitions via
`hi_res`. This PR adds `languages` to the relevant function calls so it
is included in the resulting elements.
### Testing
On the main branch, `partition_image` will include `languages` when
`strategy='ocr_only'`, but not when `strategy='hi_res'`:
```
filename = "example-docs/english-and-korean.png"
from unstructured.partition.image import partition_image
elements = partition_image(filename, strategy="ocr_only", languages=['eng', 'kor'])
elements[0].metadata.languages
elements = partition_image(filename, strategy="hi_res", languages=['eng', 'kor'])
elements[0].metadata.languages
```
For `partition_pdf`, `'ocr_only'` will include `languages` in the
metadata, but `'fast'` and `'hi_res'` will not.
```
filename = "example-docs/korean-text-with-tables.pdf"
from unstructured.partition.pdf import partition_pdf
elements = partition_pdf(filename, strategy="ocr_only", languages=['kor'])
elements[0].metadata.languages
elements = partition_pdf(filename, strategy="fast", languages=['kor'])
elements[0].metadata.languages
elements = partition_pdf(filename, strategy="hi_res", languages=['kor'])
elements[0].metadata.languages
```
On this branch, `languages` is included in the metadata regardless of
strategy
---------
Co-authored-by: ryannikolaidis <1208590+ryannikolaidis@users.noreply.github.com>
Co-authored-by: Coniferish <Coniferish@users.noreply.github.com>
Closes#2027
Tables or pages that contain only numbers are returned as floats in a
pandas.DataFrame when the image or page is converted from
`.image_to_data()`. An AttributeError was raised downstream when trying
to `.strip()` the floats. This update converts those floats if needed
and otherwise strips the text.
Testing (note: the document used for testing is new, so you will have to
copy it to the main branch in order to see that this snippet raises an
AttributeError on the main branch, but works on this branch)
```
from unstructured.partition.pdf import partition_pdf
filename = "example-docs/all-number-table.pdf"
partition_pdf(filename, strategy="ocr_only")
```
---------
Co-authored-by: cragwolfe <crag@unstructured.io>
Fix TypeError: string indices must be integers. The `annotation_dict`
variable is conditioned to be `None` if instance type is not dict. Then
we add logic to skip the attempt if the value is `None`.
### Summary
Update `ocr_only` strategy in `partition_pdf()`. This PR adds the
functionality to get accurate coordinate data when partitioning PDFs and
Images with the `ocr_only` strategy.
- Add functionality to perform OCR region grouping based on the OCR text
taken from `pytesseract.image_to_string()`
- Add functionality to get layout elements from OCR regions (ocr_layout)
for both `tesseract` and `paddle`
- Add functionality to determine the `source` of merged text regions
when merging text regions in `merge_text_regions()`
- Merge multiple test functions related to "ocr_only" strategy into
`test_partition_pdf_with_ocr_only_strategy()`
- This PR also fixes [issue
#1792](https://github.com/Unstructured-IO/unstructured/issues/1792)
### Evaluation
```
# Image
PYTHONPATH=. python examples/custom-layout-order/evaluate_natural_reading_order.py example-docs/double-column-A.jpg ocr_only xy-cut image
# PDF
PYTHONPATH=. python examples/custom-layout-order/evaluate_natural_reading_order.py example-docs/multi-column-2p.pdf ocr_only xy-cut pdf
```
### Test
- **Before update**
All elements have the same coordinate data
- **After update**
All elements have accurate coordinate data
---------
Co-authored-by: ryannikolaidis <1208590+ryannikolaidis@users.noreply.github.com>
Co-authored-by: christinestraub <christinestraub@users.noreply.github.com>
This PR introduces `clean_pdfminer_inner_elements` , which deletes
pdfminer elements inside other detection origins such as YoloX or
detectron.
This function returns the clean document.
Also, the ingest-test fixtures were updated to reflect the new standard
output.
The best way to check that this function is working properly is check
the new test `test_clean_pdfminer_inner_elements` in
`test_unstructured/partition/utils/test_processing_elements.py`
---------
Co-authored-by: Roman Isecke <roman@unstructured.io>
Co-authored-by: ryannikolaidis <1208590+ryannikolaidis@users.noreply.github.com>
Co-authored-by: rbiseck3 <rbiseck3@users.noreply.github.com>
Co-authored-by: Roman Isecke <136338424+rbiseck3@users.noreply.github.com>
- yolox has better recall than yolox_quantized, the current default
model, for table detection
- update logic so that when `infer_table_structure=True` the default
model is `yolox` instead of `yolox_quantized`
- user can still override the default by passing in a `model_name` or
set the env variable `UNSTRUCTURED_HI_RES_MODEL_NAME`
## Test:
Partition the attached file with
```python
from unstructured.partition.pdf import partition_pdf
yolox_elements = partition_pdf(filename, strategy="hi_re", infer_table_structure=True)
yolox_quantized_elements = partition_pdf(filename, strategy="hi_re", infer_table_structure=True, model_name="yolox_quantized")
```
Compare the table elements between those two and yolox (default)
elements should have more complete table.
[AK_AK-PERS_CAFR_2008_3.pdf](https://github.com/Unstructured-IO/unstructured/files/13191198/AK_AK-PERS_CAFR_2008_3.pdf)
### Summary
A follow up ticket on
https://github.com/Unstructured-IO/unstructured/pull/1801, I forgot to
remove the lines that pass extract_tables to inference, and noted the
table regression if we only do one OCR for entire doc
**Tech details:**
* stop passing `extract_tables` parameter to inference
* added table extraction ingest test for image, which was skipped
before, and the "text_as_html" field contains the OCR output from the
table OCR refactor PR
* replaced `assert_called_once_with` with `call_args` so that the unit
tests don't need to test additional parameters
* added `error_margin` as ENV when comparing bounding boxes
of`ocr_region` with `table_element`
* added more tests for tables and noted the table regression in test for
partition pdf
### Test
* for stop passing `extract_tables` parameter to inference, run test
`test_partition_pdf_hi_res_ocr_mode_with_table_extraction` before this
branch and you will see warning like `Table OCR from get_tokens method
will be deprecated....`, which means it called the table OCR in
inference repo. This branch removed the warning.
Closes `unstructured-inference` issue
[#265](https://github.com/Unstructured-IO/unstructured-inference/issues/265).
Cleaned up the kwarg handling, taking opportunities to turn instances of
handling kwargs as dicts to just using them as normal in function
signatures.
#### Testing:
Should just pass CI.
This PR resolves#1754
- function wrapper tries to use `cast` to convert kwargs into `str` but
when a value is `None` `cast(str, None)` still returns `None`
- fix replaces the conversion to simply using `str()` function call
### Description
Currently linting only takes place over the base unstructured directory
but we support python files throughout the repo. It makes sense for all
those files to also abide by the same linting rules so the entire repo
was set to be inspected when the linters are run. Along with that
autoflake was added as a linter which has a lot of added benefits such
as removing unused imports for you that would currently break flake and
require manual intervention.
The only real relevant changes in this PR are in the `Makefile`,
`setup.cfg`, and `requirements/test.in`. The rest is the result of
running the linters.
PR to support schema changes introduced from [PR
232](https://github.com/Unstructured-IO/unstructured-inference/pull/232)
in `unstructured-inference`.
Specifically what needs to be supported is:
* Change to the way `LayoutElement` from `unstructured-inference` is
structured, specifically that this class is no longer a subclass of
`Rectangle`, and instead `LayoutElement` has a `bbox` property that
captures the location information and a `from_coords` method that allows
construction of a `LayoutElement` directly from coordinates.
* Removal of `LocationlessLayoutElement` since chipper now exports
bounding boxes, and if we need to support elements without bounding
boxes, we can make the `bbox` property mentioned above optional.
* Getting hierarchy data directly from the inference elements rather
than in post-processing
* Don't try to reorder elements received from chipper v2, as they should
already be ordered.
#### Testing:
The following demonstrates that the new version of chipper is inferring
hierarchy.
```python
from unstructured.partition.pdf import partition_pdf
elements = partition_pdf("example-docs/layout-parser-paper-fast.pdf", strategy="hi_res", model_name="chipper")
children = [el for el in elements if el.metadata.parent_id is not None]
print(children)
```
Also verify that running the traditional `hi_res` gives different
results:
```python
from unstructured.partition.pdf import partition_pdf
elements = partition_pdf("example-docs/layout-parser-paper-fast.pdf", strategy="hi_res")
```
---------
Co-authored-by: Sebastian Laverde Alfonso <lavmlk20201@gmail.com>
Co-authored-by: ryannikolaidis <1208590+ryannikolaidis@users.noreply.github.com>
Co-authored-by: christinestraub <christinemstraub@gmail.com>
Each partitioner has a test like `test_partition_x_with_json()`. What
these do is serialize the elements produced by the partitioner to JSON,
then read them back in from JSON and compare the before and after
elements.
Because our element equality (`Element.__eq__()`) is shallow, this
doesn't tell us a lot, but if we take it one more step, like
`List[Element] -> JSON -> List[Element] -> JSON` and then compare the
JSON, it gives us some confidence that the serialized elements can be
"re-hydrated" without losing any information.
This actually showed up a few problems, all in the
serialization/deserialization (serde) code that all elements share.
