### Description
Exposes the endpoint url as an access kwarg when using the s3 filesystem
library via the fsspec abstraction. This allows for any non-aws data
providers that support the s3 protocol to be used with the s3 connector
(i.e. minio)
Closes out https://github.com/Unstructured-IO/unstructured/issues/950
---------
Co-authored-by: ryannikolaidis <1208590+ryannikolaidis@users.noreply.github.com>
Co-authored-by: rbiseck3 <rbiseck3@users.noreply.github.com>
- bump `unstructured-inference` to `0.6.6`
- specify default model name for element detection to be
`detectron2_onnx` to keep current behavior
- NOTE: the updated inference package by default would use yolox as
element detection model; this will be evaluated and enabled in a
separated PR
---------
Co-authored-by: ryannikolaidis <1208590+ryannikolaidis@users.noreply.github.com>
Co-authored-by: badGarnet <badGarnet@users.noreply.github.com>
Occasionally the es test can fail because the index fail to be created
on the first try. Experiments show adding timeout doesn't help but add
retry mitigates the issue. See history of commits in branch:
yao/bump-inference-to-0.6.6
https://github.com/Unstructured-IO/unstructured/pull/1563
---------
Co-authored-by: ryannikolaidis <1208590+ryannikolaidis@users.noreply.github.com>
Co-authored-by: badGarnet <badGarnet@users.noreply.github.com>
Fixes
docker exec unstructured-smoke-test /bin/bash -c
/home/notebook-user/test_unstructured_ingest/test-ingest-wikipedia.sh
/home/notebook-user/test_unstructured_ingest/test-ingest-wikipedia.sh:
line 10: python: command not found
in
https://github.com/Unstructured-IO/unstructured/blob/6ad4971/scripts/docker-smoke-test.sh#L43
that was preventing docker images from being built.
Closes GH Issue #1233.
### Summary
- add functionality to shrink all bounding boxes along x and y axes
(still centered around the same center point) before running xy-cut sort
### Evaluation
Run the followin gcommand for this
[PDF](https://utic-dev-tech-fixtures.s3.us-east-2.amazonaws.com/pastebin/patent-11723901-page2.pdf).
PYTHONPATH=. python examples/custom-layout-order/evaluate_xy_cut_sorting.py <file_path> <strategy>
- resolves an issue where occasionally deltalake writer results in
SIGABRT event though the writer finished writing table properly on linux
- this is first observed in ingest test
- Putting the writer into a process mitigates this problem by forcing
python to finish the deltalake rust backend to finish its tasks
## test
To test this it is best to setup an instance on a Linux system since the
problem has only been observed on Linux so far. Run
```bash
PYTHONPATH=. ./unstructured/ingest/main.py delta-table --num-processes 2 --metadata-exclude coordinates,filename,file_directory,metadata.data_source.date_processed,metadata.last_modified,metadata.date_created,metadata.detection_class_prob,metadata.parent_id,metadata.category_depth --table-uri ../tables/delta/ --preserve-downloads --verbose delta-table --write-column json_data --mode overwrite --table-uri file:///tmp/delta
```
Without this fix occasionally we'd encounter `SIGABTR`.
---------
Co-authored-by: ryannikolaidis <1208590+ryannikolaidis@users.noreply.github.com>
### Description
Optionally adds in chunking to the CLI which adds a flag to trigger
chunking and exposes the parameters used by the `chunk_by_title` method.
Runs chunking before the embedding step.
Opened to replace original PR
https://github.com/Unstructured-IO/unstructured/pull/1531
**Executive Summary**
Adds PDF functionality to capture hyperlink (external or internal) for
pdf fast strategy along with associate text.
**Technical Details**
- `pdfminer` associates `annotation` (links and uris) with bounding box
rather than text. Therefore, the link and text matching is not a perfect
pair but rather a logic-based and calculation matching from bounding box
overlapping.
- There is no word-level bounding box. Only character-level (access
using `LTChar`). Thus in order to get to word-level, there is a window
slicing through the text. The words are captured in alphanumeric and
non-alphanumeric separately, meaning it will split the word if contains
both, on the first encounter of non-alphanumeric.)
- The bounding box calculation is calculated using start and stop
coordinates for the corresponding word calculated from above. The
calculation is simply using distance between two dots.
The result now contains `links` in `metadata` as shown below:
```
"links": [
{
"text": "link",
"url": "https://github.com/Unstructured-IO/unstructured",
"start_index": 12
},
{
"text": "email",
"url": "mailto:unstructuredai@earlygrowth.com",
"start_index": 30
},
{
"text": "phone number",
"url": "tel:6505124019",
"start_index": 49
}
]
```
---------
Co-authored-by: ryannikolaidis <1208590+ryannikolaidis@users.noreply.github.com>
Co-authored-by: Klaijan <Klaijan@users.noreply.github.com>
### Description
This PR is two-fold:
**Embeddings:**
* Embeddings incorporated into the sharepoint source connector, which
will now call out to OpenAI and create embeddings if the flag is passed
in and the api key provided.
**Writing vector content (embeddings) to Azure cognitive search index:**
* The schema for the index expected to exist in Azure has been updated
to include the vector field type and a test script has been added to
test the new content being produced from the Sharepoint connector to
push the embedding content.
Some important notes about other changes in here:
* The embedding code had to be updated to patch the `to_dict` method on
elements to add `embeddings` to the dict output if that was added. While
the code originally added the embedding content, when `to_dict` was
called to save the content as json, this was lost.
### Summary
Uses `langdetect` to detect all languages present in the input document.
### Details
- Converts all language codes (whether user inputted or detected using
`langdetect`) to a standard ISO 639-3 code.
- Adds `languages` field to the metadata
- Will revisit how to nonstandardly represent simplified vs traditional
Chinese scripts internally (separate PR).
- Update ingest test results to add `languages` field to documents. Some
other side effects are changes in order of some elements and changes in
element categorization
### Test
You can test the detect_languages function individually by importing the
function and inputting a text sample and optionally a language:
```
text = "My lubimy mleko i chleb."
doc_langs = detect_languages(text)
print(doc_langs)
```
-> ['ces', 'pol', 'slk']
---------
Co-authored-by: Newel H <37004249+newelh@users.noreply.github.com>
Co-authored-by: ryannikolaidis <1208590+ryannikolaidis@users.noreply.github.com>
Co-authored-by: shreyanid <shreyanid@users.noreply.github.com>
Co-authored-by: Trevor Bossert <37596773+tabossert@users.noreply.github.com>
Co-authored-by: Ronny H <138828701+ron-unstructured@users.noreply.github.com>
* Partitions Salesforce data as xlm instead of text for improved detail and flexibility
* Partitions htmlbody instead of textbody for Salesforce emails
### Description
New [Azure Cognitive
Search](https://azure.microsoft.com/en-us/products/ai-services/cognitive-search)
destination connector added. Writes each json element from the created
json files via partition and writes that content to an index.
**Bonus bug fix:** Due to a recent change where the default version of
python used in the repo was bumped to `3.10` from `3.8`, this means
running `pip-compile` now runs it against that version rather than the
lowest we support which is still `3.8`. This breaks the setup for those
lower versions because some of the versions pulled in by `pip-compile`
exist for `3.10` but not `3.8`. `pip-compile` was updates to run as a
script that checks the version of python being used first, which helps
guarantee that all dependencies meet the minimum python version
requirement.
Closes out https://github.com/Unstructured-IO/unstructured/issues/1466
Addresses
[#1332](https://github.com/Unstructured-IO/unstructured/issues/1332)
with `unstructured-inference` PR
[#208](https://github.com/Unstructured-IO/unstructured-inference/pull/208).
### Summary
- Add `image_path` to element metadata
- Pass parameters related to extracting images in PDF
- Preserve image elements ignored due to garbage text if
`el.metadata.image_path` is `True`
### Testing
from unstructured.partition.pdf import partition_pdf
f_path = "example-docs/embedded-images.pdf"
# default image output directory
elements = partition_pdf(
f_path,
strategy=strategy,
extract_images_in_pdf=True,
)
# specific image output directory
elements = partition_pdf(
f_path,
strategy=strategy,
extract_images_in_pdf=True,
image_output_dir_path=<directory path>,
)
This bump removes the preprocessing before table structure extraction
and improves the OCR results for tables.
---------
Co-authored-by: yuming-long <yuming-long@users.noreply.github.com>
## **Summary**
By adding hierarchy to unstructured elements, users will have more
information for implementing vector db/LLM chunking strategies. For
example, text elements could be queried by their preceding title
element. The hierarchy is implemented by a parent_id tag in the
element's metadata.
### Features
- Introduces a parent_id to ElementMetadata (The id of the parent
element, not a pointer)
- Creates a rule set for assigning hierarchies. Sensible default is
assigned, with an optional override parameter
- Sets element parent ids if there isn't an existing parent id or
matches the ruleset
### How it works
Hierarchies are assigned via a parent id field in element metadata.
Elements are read sequentially and evaluated against a ruleset. For
example take the following elements:
1. Title, "This is the Title"
2. Text, "this is the text"
And the ruleset: `{"title": ["text"]}`. When evaluated, the parent_id of
2 will be the id of 1. The algorithm for determining this is more
complex and resolves several edge cases, so please read the code for
further details.
### Schema Changes
```
@dataclass
class ElementMetadata:
coordinates: Optional[CoordinatesMetadata] = None
data_source: Optional[DataSourceMetadata] = None
filename: Optional[str] = None
file_directory: Optional[str] = None
last_modified: Optional[str] = None
filetype: Optional[str] = None
attached_to_filename: Optional[str] = None
+ parent_id: Optional[Union[str, uuid.UUID, NoID, UUID]] = None
+ category_depth: Optional[int] = None
...
```
### Testing
```
from unstructured.partition.auto import partition
from typing import List
elements = partition(filename="./unstructured/example-docs/fake-html.html", strategy="auto")
for element in elements:
print(
f"Category: {getattr(element, 'category', '')}\n"\
f"Text: {getattr(element, 'text', '')}\n"
f"ID: {element.id}\n" \
f"Parent ID: {element.metadata.parent_id}\n"\
f"Depth: {element.metadata.category_depth}\n" \
)
```
### Additional Notes
Implementing this feature revealed a possibly undesired side-effect in
how element metadata are processed. In
`unstructured/partition/common.py` the `_add_element_metadata` is
invoked as part of the `add_metadata_with_filetype` decorator for
filetype partitioning. This method is intended to add additional
information to the metadata generated with the element including
filename and filetype, however the existing metadata is merged into a
newly created metadata object rather than the other way around. Because
of the way it's structured, new metadata fields can easily be forgotten
and pose debugging challenges to developers. This likely warrants a new
issue.
I'm guessing that the implementation is done this way to avoid issues
with deserializing elements, but could be wrong.
---------
Co-authored-by: Benjamin Torres <benjats07@users.noreply.github.com>
**Summary**
Adds logic to combine broken numbered list for pdf fast strategy.
**Details**
Previously the document reads the numbered list items part of the
`layout-parser-paper-fast.pdf` file as:
```
'1. An off-the-shelf toolkit for applying DL models for layout detection, character'
'recognition, and other DIA tasks (Section 3)'
'2. A rich repository of pre-trained neural network models (Model Zoo) that'
'underlies the off-the-shelf usage'
'3. Comprehensive tools for efficient document image data annotation and model'
'tuning to support different levels of customization'
'4. A DL model hub and community platform for the easy sharing, distribu- tion, and discussion of DIA models and pipelines, to promote reusability, reproducibility, and extensibility (Section 4)'
```
Now it reads:
```
'1. An off-the-shelf toolkit for applying DL models for layout detection, character recognition, and other DIA tasks (Section 3)'
'2. A rich repository of pre-trained neural network models (Model Zoo) that underlies the off-the-shelf usage'
'3. Comprehensive tools for efficient document image data annotation and model' tuning to support different levels of customization'
'4. A DL model hub and community platform for the easy sharing, distribu- tion, and discussion of DIA models and pipelines, to promote reusability, reproducibility, and extensibility (Section 4)'
```
The added logic leverages `ElementType` and `coordinates` to determine
whether the following lines is a part of the previously detected
`ListItem` or not.
**Test**
Add test that checks the element length less than original version with
broken numbered list. The test also checks whether the first detected
numbered list ends with previously broken line.
---------
Co-authored-by: ryannikolaidis <1208590+ryannikolaidis@users.noreply.github.com>
Co-authored-by: Klaijan <Klaijan@users.noreply.github.com>
Currently there are some cases when `partition_pdf` is run using the
`hi_res` strategy, in which elements can come back with category
`UncategorizedText`. This happens when the detection model fails to
detect an element, but we're able to find it anyway either because it
was embedded in the PDF, or we found it using OCR.
This commit is to allow for attempting to categorize these uncategorized
elements using our text-based classification function,
`element_from_text`.
### Summary
In order to support language functionality other than Tesseract OCR, we
want to represent languages provided for either partitioning accuracy or
OCR as a standard list of langcodes as strings.
### Details
Adds `languages` (a list of strings) as a parameter to pdf partitioning
functions. Marks `ocr_languages` for deprecation. Adds a new file
`lang.py` for language-related helper functions.
Coming up: langcode standardization, language detection
### Test
Call `partition_pdf` or `partition_pdf_or_image` with a variety of
strategies, languages, or `ocr_languages`.
- inclusion of `ocr_languages` as a parameter should display a
deprecation warning
- the other valid call outputs should be no different from the current
outputs.
ex:
```
from unstructured.partition.pdf import partition_pdf
elements = partition_pdf(filename="example-docs/DA-1p.pdf", strategy="hi_res", languages=["eng", "spa"])
print("\n\n".join([str(el) for el in elements]))
```
Adding table extraction to HTML partitioning.
This PR utilizes 'table' HTML elements to extract and parse HTML tables
and return them in partitioning.
```
# checkout this branch, go into ipython shell
In [1]: from unstructured.partition.html import partition_html
In [2]: path_to_html = "{html sample file with table}"
In [3]: elements = partition_html(path_to_html)
```
you should see the table in the elements list!
### Description
Update all other connectors to use the new downstream architecture that
was recently introduced for the s3 connector.
Closes#1313 and #1311
The default sorting algorithm for PDF's, "xycut," would cause an error
when partitioning a document if Y coordinate points were negative. This
change checks for that condition (or more broadly, any negative
coordinates) and falls back to the "basic" sort if that is the case.
This PR does not address the underlying issue of "bad points" which
still should be investigated. However, the sorting code should be less
brittle to unexpected bounding boxes in the first case.
Resolves: https://github.com/Unstructured-IO/unstructured/issues/1296
Bumps unstructured-inference==05.23 to pull in @christinestraub's fix:
https://github.com/Unstructured-IO/unstructured-inference/pull/198 , so
embedded Images
in PDF's are now included in partition results ("hi_res").
From the perspective of elements with clean text, this is not a big win
as a lot of the images have OCR garbage. However, it is important to
preserve image elements for other downstream use cases, so overall this
is a step forward.
If a layout model is used from unstructured-inference, you get back
class probabilities in the element metadata from partition.
extra-pdf-image-in in requirements already has the newest version of
unstructured-inference in there without a pinned version. Is there any
place else that the unstructured-inference version needs to be updated
to the required release version, 0.5.22?
### Summary
Closes#1230. Updates `partition_html` to split on `<br>` tags that
appear within text elements.
### Testing
The following is code previously produced one giant element on `main`.
```python
from unstructured.partition.html import partition_html
filename = "example-docs/ideas-page.html"
elements = partition_html(filename=filename)
len(elements) # Should be 4
print("\n\n".join([str(el) for el in elements)])
```
The output should be:
```python
January 2023
(Someone fed my essays into GPT to make something that could answer
questions based on them, then asked it where good ideas come from. The
answer was ok, but not what I would have said. This is what I would have said.)
The way to get new ideas is to notice anomalies: what seems strange,
or missing, or broken? You can see anomalies in everyday life (much
of standup comedy is based on this), but the best place to look for
them is at the frontiers of knowledge.
Knowledge grows fractally.
From a distance its edges look smooth, but when you learn enough
to get close to one, you'll notice it's full of gaps. These gaps
will seem obvious; it will seem inexplicable that no one has tried
x or wondered about y. In the best case, exploring such gaps yields
whole new fractal buds.
```
This connector:
- takes a Jira Cloud URL, user email and api token; to authenticate into
Jira Cloud
- ingests:
- either all issues in all projects in a Jira Cloud Organization
- or
- issues in user specified projects, boards
- user specified issues
- processes this kind of data:
- text fields such as issue summary, description, and comments
- dropdown fields such as issue type, status, priority, assignee,
reporter, labels, and components
- other data such as issue id, issue key, project id, information on
subtasks
- notes down attachment URLs, however does not process attachments
- stores each downloaded issue in a txt file, in a predefined template
form (consisting of the data above)
- then processes each downloaded issue document into elements using
unstructured library
- related to: https://github.com/Unstructured-IO/unstructured/issues/263
To test the changes, make the necessary setups and run the relevant
ingest test scripts.
---------
Co-authored-by: ryannikolaidis <1208590+ryannikolaidis@users.noreply.github.com>
Co-authored-by: ahmetmeleq <ahmetmeleq@users.noreply.github.com>
The CustomError that we use to wrap custom ingest errors inherits from
BaseException rather than Exception (as we should, per specification
[here](https://docs.python.org/3/library/exceptions.html#BaseException)).
This resulted in exceptions not properly raising as expected. This PR
changes the inheritance which resolves the known issue.
Additionally, our base definition for get_file on IngestDoc was wrapped
with SourceConnectionError, however this must be explicitly decorating
each subclass definition in order to function. This PR does that.
## Testing
Some unit test coverage was added for the error wrapping class, however
this wasn't properly recreating the issue we are seeing when running
ingest tests.
To recreate that issue one can intentionally raise an exception in the
[partition_file](https://github.com/Unstructured-IO/unstructured/blob/main/unstructured/ingest/interfaces.py#L214C9-L214C23)
definition and then run any ingest test. Prior to this change: the code
and logs suggest that everything ran without exception, but the
partitioned output was not generated (as a result the test will fail
without any clues as to what went wrong). With this update, the expected
custom partition error, error message, and stack trace will be visible.
---------
Co-authored-by: Ahmet Melek <39141206+ahmetmeleq@users.noreply.github.com>
- revert the layout parser fast pdf file to original with just two pages
- add a new file that has one empty page and one page says "this page is
intentionally left blank" for tests
### Description
Convert s3 cli code to also support writing to s3. Writers are added as
optional subcommands to the parent command with their own arguments.
Custom `click.Group` introduced to add some custom formatting and text
in help messages.
To limit the scope of this PR, most existing files were not touched but
instead new files were added for the new flow. This allowed _only_ the
s3 connector to be updated without breaking any other ones.
