Skip to main content

Simple, Pythonic extraction of images, text, and shapes from PDFs

Project description

Travis CI build status (Linux) Coverage Status Req-status

minecart is a Python package that simplifies the extraction of text, images, and shapes from a PDF document. It provides a very Pythonic interface to extract positioning, color, and font metadata for all of the objects in the PDF. It is a pure-Python package (it depends on pdfminer for the low-level parsing). minecart takes inspiration from Tim McNamara’s slate, but aims to provide more detailed information:

>>> pdffile = open('example.pdf', 'rb')
>>> doc = minecart.Document(pdffile)
>>> page = doc.get_page(3)
>>> for shape in page.shapes.iter_in_bbox((0, 0, 100, 200)):
...     print shape.path, shape.fill.color.as_rgb()
>>> im = page.images[0].as_pil()  # requires pillow
>>> im.show()

Installation

As of version 0.3.0, only Python 3 is support, using pdfminer3k.

  1. The easy way: pip install minecart
  2. The hard way: download the source code, change into the working directory, and run python setup.py install

For CJK languages: Supporting the CJK languages requires an addtional step, as detailed in pdfminer.

Features

  • Shapes: You can extract path information, bounding box, stroke parameters, and stroke/fill colors. Color support is fairly robust, allowing the simple .as_rgb() in most cases. (To be concrete, minecart supports the DeviceRGB, DeviceCMYK, DeviceGray, and CIE-based color spaces. Indexed colors are supported if they index into one of the above.)
  • Images: minecart can easily extract images to PIL.Image objects.
  • Text: (Called Lettering in the source) In addition to extracting plain text from the PDF, you can access the position/bounding box information and the font used.

If there’s a feature you’d like to extract from a PDF that’s not currently supported, open up an issue or submit a pull request! I’m especially interested in hearing whether there are many PDFs using color spaces outside of the ones currently supported.

Documentation

The main entry point will always be minecart.Document, which accepts a single parameter, an open file-like object which will be read to create the document. The Document has two primary methods for accessing its contents: .get_page(num) and .iter_pages(). Both methods return minecart.Page objects, which provide access to the graphical elements found on the page. Page objects have three main attributes:

  • .images: A list of all the minecart.Image objects found on the page.
  • .letterings: A list of all the text objects found on the page, as Lettering objects. Lettering is a unicode subclass which adds bounding box and font information (using .get_bbox() or .font).
  • .shapes: A list of all the squares, circles, lines, etc. found on the page as Shape objects. Shape objects have three main attributes of interest:
    • .stroke: An object containing the stroke parameters used to draw the shape. .stroke has .color, .linewidth, .linecap, .linejoin, .miterlimit, and .dash attributes. If the shape was not stroked, .stroke will be None.
    • .fill: An object containing the fill parameters used to draw the shape. Right now, .fill only has a .colorparameter.
    • .path: A list with the coordinates used to defined the shape, as well as the type of line segment each set of coordinates defines. Refer to the minecart.Shape documentation for more details

Note on color: The PDF spec spends a fair amount of time dealing with color specifications, defining color spaces, and transforms and the like. minecart’s approach is to simplify things down with sensible defaults, so that every color has an .as_rgb() method, which returns a 3-tuple with component values between 0 (black) and 1 (white). If you are interested in extracting colorspace families and parameters, you can do that too, though!

We try to keep docstrings complete and up to date, so you can read through the source or use dir and help to see what methods are available.

Support

If you are having trouble working with minecart, feel free to create a new issue.

Contributing

Bug reports are always welcome (using the GitHub tracker) as are feature requests. The PDF spec has so many corners, it is hard to prioritize which features to implement. If there’s something you’d like to extract from a document but isn’t currently supported, please create a new issue.

If you’d like to contribute code, you can either create an issue and include a patch (if the changes are small) or fork the project and create a pull request.

License

This project is licensed under the MIT license.

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Filename, size & hash SHA256 hash help File type Python version Upload date
minecart-0.3.0-py3-none-any.whl (23.0 kB) Copy SHA256 hash SHA256 Wheel py3
minecart-0.3.0.tar.gz (18.2 kB) Copy SHA256 hash SHA256 Source None

Supported by

Elastic Elastic Search Pingdom Pingdom Monitoring Google Google BigQuery Sentry Sentry Error logging AWS AWS Cloud computing DataDog DataDog Monitoring Fastly Fastly CDN SignalFx SignalFx Supporter DigiCert DigiCert EV certificate StatusPage StatusPage Status page