.heic files are an image filetype we have not supported.
#### Testing
```
from unstructured.partition.image import partition_image
png_filename = "example-docs/DA-1p.png"
heic_filename = "example-docs/DA-1p.heic"
png_elements = partition_image(png_filename, strategy="hi_res")
heic_elements = partition_image(heic_filename, strategy="hi_res")
for i in range(len(heic_elements)):
print(heic_elements[i].text == png_elements[i].text)
```
---------
Co-authored-by: christinestraub <christinemstraub@gmail.com>
The code makes edit to the `measure_text_extraction_accuracy` function
to allows dir of txt as well as json. The function also takes input
`output_type` to be either "json" or "txt" only, and checks if the files
under given directory/list contains only specified file type or not.
To test this feature, run the following code:
```PYTHONPATH=. python unstructured/ingest/evaluate.py measure-text-extraction-accuracy-command --output_dir <clean-text-path> --source_dir <cct-label-path> --output_type txt```
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#816
## Description
Added functionality for `partition_email` to automatically decode base64
text before passing it to `partition_text` or `partition_html`.
Also adds base64 encoded email text test cases.
Closes#2218. When a csv has commas in its content, and the delimiter is
something else, Pandas may throw an error. We can sniff the csv and get
the correct delimiter to pass to Pandas. To verify, try partitioning the
file in the linked bug.
### 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>
A DOCX header or footer is a so-called "story part" meaning like the
document body (which is also a story part) it can contain both
paragraphs and tables. The implementation of `Header.text` and
`Footer.text` gather only the paragraphs.
Add a new method to extract all content from a header or footer,
including table content, suitable for use as the `.text` attribute of
that element.
Fixes#2126.
**Summary.** The `python-docx` table API is designed for _uniform_
tables (no merged cells, no nested tables). Naive processing of DOCX
tables using this API produces duplicate text when the table has merged
cells. Add a more sophisticated parsing method that reads only "root"
cells (those with an actual `<tc>` element) and skip cells spanned by a
merge.
In the process, abandon use of the `tabulate` package for this job
(which is also designed for uniform tables) and remove the whitespace
padding it adds for visual alignment of columns. Separate the text for
each cell with a single newline ("\n").
Since it's little extra trouble, add support for nested tables such that
their text also contributes to the `Table.text` string.
The new `._iter_table_texts()` method will also be used for parsing
tables in headers and footers (where they are frequently used for layout
purposes) in a closely following PR.
Fixes#2106.
Page breaks can and often do occur within a paragraph. The full text of
the paragraph is attributed to the page (number) the paragraph starts
on.
Improve page-break fidelity such that a paragraph containing a
page-break is split into two elements, one containing the text before
the page-break and the other the text after. Emit the `PageBreak`
element between these two and assign the correct page-number (n and n+1
respectively) to the two textual elements.
This functionality is largely provided upstream by the new `python-docx`
v1.0.0 release (1.0.0 from 0.8.11 because it drops Python 2 support).
That version also makes obsolete the "include hyperlink text in
`Paragraph.text` monkey patch that we had maintained up to now. Remove
that monkey-patch.
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#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>
Page breaks are reliably indicated by `w:lastRenderedPageBreak` elements
present in the document XML. Page breaks are NOT reliably indicated by
"hard" page-breaks inserted by the author and when present are redundant
to a `w:lastRenderedPageBreak` element so cause over-counting if used.
Use rendered page-breaks only.
A DOCX document that has no sections can still contain one or more
tables. Such files are never created by Word but Word can open them just
fine. These can be and are generated by other applications.
Use the newly-added `Document.iter_inner_content()` method added
upstream in `python-docx` to capture both paragraphs and tables from a
section-less DOCX document.
This generalizes the fix for MS Teams chat-transcripts (an example of
sectionless-docx) implemented in #1825.
In DOCX, like HTML, a table cell can itself contain a table. This is not
uncommon and is typically used for formatting purposes.
When a DOCX table is nested, create nested HTML tables to reflect that
structure and create a plain-text table with captures all the text in
nested tables, formatting it as a reasonable facsimile of a table.
This implements the solution described and spiked in PR #1952.
---------
Co-authored-by: Bruno Bornsztein <bruno.bornsztein@gmail.com>
### Summary
Click decorated functions cannot (properly) be called outside of the
click interface. This makes it difficult to reuse the setup
functionality in measure_text_edit_distance or
measure_element_type_accuracy. This PR removes the click decoration and
separates it into a wrapper function purely to execute the command.
### Technical Details
- Changed as suggested in [this StackOverflow
post](https://stackoverflow.com/questions/40091347/call-another-click-command-from-a-click-command)
response
- The locations of these now distinct functions are separate: the
`_command` click-decorated functions stay in ingest/evaluate.py, and the
core functions measure_text_edit_distance and
measure_element_type_accuracy are moved into the unstructured/metrics/
folder (which is a more logical location for them).
- Initial test added for measure_text_edit_distance
### Test
`sh ./test_unstructured_ingest/evaluation-metrics.sh text-extraction`
functionality is unchanged.
---------
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>
- This PR adds a function to check if a piece of text only contains a
bullet (no text) to prevent creating an empty element.
- Also fixed a test that had a typo.
*Reviewer:* May be quicker to review commit by commit as they are quite
distinct and well-groomed to each focus on a single clean-up task.
Clean up odds-and-ends in the docx partitioner in preparation for adding
nested-tables support in a closely following PR.
1. Remove obsolete TODOs now in GitHub issues, which is probably where
they belong in future anyway.
2. Remove local DOCX "workaround" code that has been implemented
upstream and is now obsolete.
3. "Clean" the docx tests, introducing strict typing, extracting a
fixture or two, and generally tightening things up.
4. Extract docx-local versions of
`unstructured.partition.common.convert_ms_office_table_to_text()` which
will be the base for adding nested-table support. More information on
why this is required in that commit.
Courtesy @phowat, created a branch in the repo to make some changes and
merge quickly.
Closes#1486.
* **Fixes issue where tables from markdown documents were being treated
as text** Problem: Tables from markdown documents were being treated as
text, and not being extracted as tables. Solution: Enable the `tables`
extension when instantiating the `python-markdown` object. Importance:
This will allow users to extract structured data from tables in markdown
documents.
#### Testing:
On `main` run the following (run `git checkout fix/md-tables --
example-docs/simple-table.md` first to grab the example table from this
branch)
```python
from unstructured.partition.md import partition_md
elements = partition_md("example-docs/simple-table.md")
print(elements[0].category)
```
Output should be `UncategorizedText`. Then run the same code on this
branch and observe the output is `Table`.
---------
Co-authored-by: cragwolfe <crag@unstructured.io>
- add helper to run inference over an image or pdf of table and compare
it against a ground truth csv file
- this metric generates a similarity score between 1 and 0, where 1 is
perfect match and 0 is no match at all
- add example docs for testing
- NOTE: this metric is only relevant to table structure detection.
Therefore the input should be just the table area in an image/pdf file;
we are not evaluating table element detection in this metric
This PR resolves#1816
- current docx partition assumes all contents are in sections
- this is not true for MS Teams chat transcript exported to docx
- now the code checks if there are sections or not; if not then iterate
through the paragraphs and partition contents in the paragraphs
Fixes recursion limit error that was being raised when partitioning
Excel documents of a certain size.
Previously we used a recursive method to find subtables within an excel
sheet. However this would run afoul of Python's recursion depth limit
when there was a contiguous block of more than 1000 cells within a
sheet. This function has been updated to use the NetworkX library which
avoids Python recursion issues.
* Updated `_get_connected_components` to use `networkx` graph methods
rather than implementing our own algorithm for finding contiguous groups
of cells within a sheet.
* Added a test and example doc that replicates the `RecursionError`
prior to the change.
* Added `networkx` to `extra_xlsx` dependencies and `pip-compile`d.
#### Testing:
The following run from a Python terminal should raise a `RecursionError`
on `main` and succeed on this branch:
```python
import sys
from unstructured.partition.xlsx import partition_xlsx
old_recursion_limit = sys.getrecursionlimit()
try:
sys.setrecursionlimit(1000)
filename = "example-docs/more-than-1k-cells.xlsx"
partition_xlsx(filename=filename)
finally:
sys.setrecursionlimit(old_recursion_limit)
```
Note: the recursion limit is different in different contexts. Checking
my own system, the default in a notebook seems to be 3000, but in a
terminal it's 1000. The documented Python default recursion limit is
1000.
### Summary
Closes#1534 and #1535
Detects document language using `langdetect` package.
Creates new kwargs for user to set the document language (`languages`)
or detect the language at the element level instead of the default
document level (`detect_language_per_element`)
---------
Co-authored-by: shreyanid <42684285+shreyanid@users.noreply.github.com>
Co-authored-by: ryannikolaidis <1208590+ryannikolaidis@users.noreply.github.com>
Co-authored-by: Coniferish <Coniferish@users.noreply.github.com>
Co-authored-by: cragwolfe <crag@unstructured.io>
Co-authored-by: Austin Walker <austin@unstructured.io>
**Executive Summary**
Add function that returns frequency of given element types and depth.
---------
Co-authored-by: shreyanid <42684285+shreyanid@users.noreply.github.com>
## Summary
**Improve title detection in pptx documents** The default title
textboxes on a pptx slide are now categorized as titles.
**Improve hierarchy detection in pptx documents** List items, and other
slide text are properly nested under the slide title. This will enable
better chunking of pptx documents.
Hierarchy detection is improved by determining category depth via the
following:
- Check if the paragraph item has a level parameter via the python pptx
paragraph. If so, use the paragraph level as the category_depth level.
- If the shape being checked is a title shape and the item is not a
bullet or email, the element will be set as a Title with a depth
corresponding to the enumerated paragraph increment (e.g. 1st line of
title shape is depth 0, second is depth 1 etc.).
- If the shape is not a title shape but the paragraph is a title, the
increment will match the level + 1, so that all paragraph titles are at
least 1 to set them below the slide title element
**Executive Summary**
Fix bug on the `get_word_bounding_box_from_element` function that
prevent `partition_pdf` to run.
**Technical Details**
- The function originally first define `isalnum` on the first index. Now
switched to conditional on flag value.
**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>
Improves hierarchy from docx files by leveraging natural hierarchies
built into docx documents. Hierarchy can now be detected from an
indentation level for list bullets/numbers and by style name (e.g.
Heading 1, List Bullet 2, List Number).
Hierarchy detection is improved by determining category depth via the
following:
1. Check if the paragraph item has an indentation level (ilvl) xpath -
these are typically on list bullet/numbers. Return the indentation level
if it exists
2. Check the name of the paragraph style if it contains any category
depth information (e.g. Heading 1 vs Heading 2 or List Bullet vs List
Bullet 2). Return the category depth if found, else default to depth of
0.
3. Check the paragraph ilvl via the paragraph's style name. Outside of
the paragraph's metadata, docx stores default ilvls for various style
names, which requires a complex lookup. This check is yet to be
implemented, as the above methods cover most usecases but the
implementation is stubbed out.
---
Co-authored-by: Steve Canny <stcanny@gmail.com>
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>,
)
### Summary
In order to convert between incompatible language codes from packages
used for OCR, this change adds a function to map between any standard
language codes and tesseract OCR specific codes. Users can input
language information to `languages` in any Tesseract-supported langcode
or any ISO 639 standard language code.
### Details
- Introduces the
[python-iso639](https://pypi.org/project/python-iso639/) package for
matching standard language codes. Recompiles all dependencies.
- If a language is not already supplied by the user as a Tesseract
specific langcode, supplies all possible script/orthography variants of
the language to the Tesseract OCR agent.
### Test
Added many unit tests for a variety of language combinations, special
cases, and variants. For general testing, call partition functions with
any lang codes in the languages parameter (Tesseract or standard).
for example,
```
from unstructured.partition.auto import partition
elements = partition(filename="example-docs/layout-parser-paper.pdf", strategy="hi_res", languages=["en", "chi"])
print("\n\n".join([str(el) for el in elements]))
```
should supply eng+chi_sim+chi_sim_vert+chi_tra+chi_tra_vert to Tesseract
## **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>
This PR does two things:
1. Adds test case (and alters sample doc) for rtf and epub files with
table
2. Adds `xls/x` file extension to `skip_infer_table_types` default list
---------
Co-authored-by: shreyanid <42684285+shreyanid@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
This PR resolves#1247 by using the matching elements and bbox for
coordinate computation.
This PR also updates the example doc
`example-docs/layout-parser-paper-fast.pdf` so that it includes a true
blank page and a page with text "this page is intentionally left blank".
This change helps us testing:
- differences between fast and hi_res
- code handling empty pages in between pages with contents (which
triggers the bug found in #1247 )
Lastly, this PR updates the names of the variables inside
`_partition_pdf_or_image_with_ocr` so that matching inputs all starts
with `_` like `_elements`, `_text`, and `_bboxes` to improve
readability.
This change also improves partition performance for multi-page pdfs as
it reduces the amount of iterations inside
`add_pytesseract_bbox_to_elements`. Testing locally on m2 mac + Rocky
docker shows it reduces partition time for DA-619p.pdf file from around
1min to around 23s.
### Summary
Closes#1018. Enables `partition_email` and `partition_msg` to detect if
an email has PGP encrypted content. Based on the specification in [RFC
2015](https://www.ietf.org/rfc/rfc2015.txt). The test emails are based
on the example email in the spec. If PGP detected content is detected, a
warning is emitted and an empty set of lists is returned.
### Testing
```python
from unstructured.partition_email import partition_email
filename = "example-docs/eml/fake-encrypted.eml"
partition_email(filename=filename)
```
```python
from unstructured.partition_msg import partition_msg
filename = "example-docs/fake-encrypted.msg"
partition_msgl(filename=filename)
```
### Summary
Address
[#1136](https://github.com/Unstructured-IO/unstructured/issues/1136) for
`hi_res` and `fast` strategies. The `ocr_only` strategy does not include
coordinates.
- add functionality to switch sort mode between the current `basic`
sorting and the new `xy-cut` sorting for `hi_res` and `fast` strategies
- add the script to evaluate the `xy-cut` sorting approach
- add jupyter notebook to provide evaluation and visualization for the
`xy-cut` sorting approach
### Evaluation
```
export PYTHONPATH=.:$PYTHONPATH && python examples/custom-layout-order/evaluate_xy_cut_sorting.py <file_path> <strategy>
```
Here, the file should be under the project root directory. For example,
```
export PYTHONPATH=.:$PYTHONPATH && python examples/custom-layout-order/evaluate_xy_cut_sorting.py example-docs/multi-column-2p.pdf fast
```
Add test case test_partition_image_with_multipage_tiff that reads multipage TIFF file and
- confirms that the function reads all the pages in the TIFF.
- page number is added to the metadata
This PR is branched from and developed on top of 6d6be99 commit.
Handle Content-Disposition: inline and attachment without filename
* Add new email test example and test with Content-Disposition: inline.
* Move attachment_info above for loop so it is always defined
* Check if item is inline as well as attachment as these both lack an = character to split on
* Create filename if filename is not specified and write file.
* Update list_attachments with new filename
Fix attachments with = in filename
* Limit split to first match of = to prevent creating a list of more than two parts
* Add example email with attachment name and test for issue
* feat: add functionality to check if a string contains any emoji characters
* feat: add functionality to switch `html` text parser based on whether the `html` text contains emoji
* chore: add `beautifulsoup4` and `emoji` packages to `requirements/base.in` for general use
* chore: update changelog & version
* chore: update changelog & version
* chore: update dependencies
* test: update `EXPECTED_XLS_TEXT_LEN` for `test_auto_partition_xls_from_filename`
* chore: update changelog & version
