Skip to main content

A local, offline document archive

Project description


filecabinet is a minimal document management system for your computer. It has metadata per document and supports fulltext search in various document types.


The easiest way to install is to use pip:

    pip install filecabinet

Alternatively you can get the source code at codeberg:

    git clone
    pip install filecabinet


filecabinet requires the xapian python bindings which can not be installed through pip!

Other automatically installed required dependencies are:

Even though optional, I strongly recommend installing Tesseract OCR to enable fulltext search in scanned documents.

Quick start

To initialize your file cabinet, run filecabinet init and provide a new path where you would like to store your documents:

    filecabinet init ~/Documents/cabinet

Now you can start either copying files into ~/Documents/cabinet/inbox and run

    filecabinet pickup

to process them, or add files manually via

    filecabinet add ~/some_scanned_document.jpg

To get a basic overview of documents, you can use the Shell.

Workflow / Use cases

Here’s the usual worflow with filecabinet:

  1. Put some documents (PDF, scanned documents, etc) into the inbox folder of your cabinet
  2. Run filecabinet pickup
  3. List all new documents with filecabinet list new

Other use cases are:

  • Search for a specific document with filecabinet find "searchterm" "other search term"
  • Edit the metadata of a document through the shell filecabinet shell (see next section)


There’s a basic shell that allows you to inspect indexed documents, edit their metadata (by means of an external text editor), or view the documents.

To open the shell, run

    filecabinet shell

Try help inside the shell to see what your options are.

Metadata editing

If you want to use a specific text editor to modify metadata, consider updating your configuration file’s Shell section and add a document_editor, like this:

    editor = subl -w

In this example we set up SublimeText as the external editor. Note that the -w option is necessary to make filecabinet wait until you’re done editing the file before returning into the shell.
Visual Studio Code uses the -W or --wait flag to accomplish the same behaviour.


Searching for tags is done case-insensitive and is done using tag:. For example if you're looking for a document that's tagged with banana, you can search for it by tag:banana.

Searching new documents is accomplished by searching for tag:new. If you only want to find documents that are not new, you can also search for -tag:new. Unless specified, a search will ignore whether or not a document is new.

You can search for any metadata value, like title, author, or language, by searching with the metadata name and a colon like title:gravity.

Everything else that does not match the special search terms will be used in the fulltext search.

If you want to search for terms with whitespaces, you can use quotes: title:"brain surgery".


The title contains "brain", is from author "Gumby" and it was set to some time before August 2005: title:brain author:gumby date:2015-08-01

Looking for a newly added document with the title "The Larch": title:larch tag:new

Grouping of pages

Sometimes you will have a scanned document in form of multiple pages, each page a .jpg file, like page1.jpg, page2.jpg, page3.jpg.

Of course all these pages form the same document.

To tell filecabinet that these files all belong to the same document, you can put them in a folder inside the inbox before running pickup:

  • inbox/doc/page1.jpg
  • inbox/doc/page2.jpg
  • inbox/doc/page3.jpg

This will tell filecabinet that they all belong to the same document.

Here’s also where you can hint to the language of the document for OCR (see Language hinting in the next section) by calling the folder, for example, doc-nl to indicate that all pages are written in the dutch language.


filecabinet can use Tesseract OCR to do character recognition on pictures and scanned PDFs, so you can search the text of images.

In order for that to work, you have to install Tesseract and some language packages, depending on the languages of the documents you wish to scan.

If you don't have Tesseract OCR installed, filecabinet will still work, but be much less useful.

Language hinting

You can tell filecabinet what language a document has even as it is in the inbox by adding its language as a suffix: hyphen followed by language code (ISO-639).

A few examples will help. Consider these files:

  • page-1.jpg
  • contract.png

Suppose your default language is set to english (default-lanugage = eng in the configuration file); page-1.jpg is in English but contract.png is in German.

OCR will likely have difficulties with letters like öäü in contract.png unless you tell it what language the document is in:

  • contract-ger.png

ger is one of the ISO-639 language codes for German (others are de and deu; see wikipedia for the long listing).

With this -ger suffix, filecabinet will use the correct language packet (if you have it installed) and the OCR will yield much better results.

Rule based tagging

By using metaindex, filecabinet inherits the powerful rule based tagging. This allows you to automatically add metadata tags to documents based on their text (which might have come from OCR).

Rules are defined in text files and you have to point filecabinet to the rule files that you want it to use. To do that, add a section [Rules] to your configuration file (usually at ~/.config/filecabinet/filecabinet.conf) and list your rule files like this:

    base = ~/.config/filecabinet/basic_rules.txt
    companies = ~/Document/company_rules.txt

The names (before the =) are somewhat free-form descriptors.

To understand how to write these rule files, please have a look at the metaindex documentation.

To test your rules on documents, you can use the filecabinet test-rules command. It will run all indexers on a file and show you what tags have been found by your rules.

When using test-rules the tested document will not be added to your cabinet.

Cabinet Directory Structure

Assuming a cabinet is set up at ~/cabinet, the directory structure is:

     ├── inbox
     ├── metaindex.conf
     ├── metaindex.log
     └── documents
          └── <partial document id>
               └── <full document id>
                    ├── <document id>.yaml
                    ├── <document id>.<suffix>
                    └── <document id>.txt
  • inbox will be processed (and emptied) when filecabinet pickup is being run
  • documents contains the documents
  • <document id>.yaml contains the metadata
  • <document id>.<suffix> is the original document (usually a PDF)
  • <document id>.txt is the extracted full text, if it could be extracted
  • metaindex.conf, the configuration file for filecabinet's metaindexserver
  • metaindex.log, the log file of file cabinet's metaindexserver


filecabinet itself as well as each individual cabinet can be configured through the user’s configuration file (usually in ~/.config/filecabinet/filecabinet.conf).

See example.conf for all configuration options!

Usage from Python

To use filecabinet from Python, you can use this boilerplate:

    from filecabinet import Manager

    manager = Manager()

    session = manager.new_session()

session will be an instance of Session which, together with manager, allows manipulation of metadata and querying of documents.

Project details

Download files

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

Source Distribution

filecabinet-2.1.0.tar.gz (25.0 kB view hashes)

Uploaded Source

Built Distribution

filecabinet-2.1.0-py3-none-any.whl (22.0 kB view hashes)

Uploaded Python 3

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page