guessfilename.py

This Python script tries to come up with a new file name for each file from command line argument.

It does this with several methods: first, the current file name is analyzed and any [[https://en.wikipedia.org/wiki/Iso_date][ISO date/timestamp]] and [[https://github.com/novoid/filetags/][filetags]] are re-used. Secondly, if the parsing of the file name did not lead to any new file name, the content of the file is analyzed. Following file types are supported by now:

PDF files

The script accepts an arbitrary number of files (see your shell for possible length limitations).

Target group: users who are able to use command line tools and who
are using tags in file names.
Hosted on github: [[https://github.com/novoid/guess-filename.py]] and PyPi: [[https://pypi.org/project/guessfilename/]]

** Why

I do scan almost all paper mail. Many of those documents are sent to me regularily. Such documents are bills or insurance informations, for example.

Being too lazy to name those files manually with high chances of getting many variants for the same document type, I came up with a method to derive file names from either the old file name (cues I enter without knowing the exact target file name) or the file content.

Analyzing the content enables this script to recognize bills via customer numbers or phone numbers, amounts to pay, and so on.

** Examples

Here are some examples that demonstrate the purpose of this script. The generated file names are following [[https://www.karl-voit.at/managing-digital-photographs/][my file name convention]].

For better user experience, I like to define an abbreviation in [[https://karl-voit.at/apps-I-am-using/][my shell]] which also makes the examples easier to read:

: alias gf=guessfilename.py

A very simple example is a simple bill:

: gf "2016-03-05 phone 12,34 €.pdf" : → "2016-03-05 COMPANY landline 12,34€ -- scan bill.pdf"

Some mobile apps generate weird formatted file names. Here is some recording:

: gf "rec_20171129-0902 A nice recording .wav" : → "2017-11-29T09.02 A nice recording.wav"

Android screenshot files tend to look like that:

: gf "Screenshot_2017-11-29_10-32-12.png" : → "2017-11-29T10.32.12 -- screenshots.png"

Android photographs are handled similarly:

: gf "IMG_20190118_133928.jpg" : → "2019-01-18T13.39.28.jpg"

Files saved from [[https://signal.org/][Signal]] do have strange default names as well:

: gf "signal-2018-03-08-102332.jpg" : → "2018-03-08T10.23.32.jpg"

Many companies like to generate really silly file names. This is from my bank:

: gf "C110014365208EUR20150930001.pdf" : → "2015-09-30 Bank statement 2015-001 10014365208.pdf"

This script is able to parse content of PDF file in order to get meta-data to generate the new file name. This can be applied to you salary, for example:

: gf "2020-03-04 salary.pdf" : → "2020-02-29 MYCOMPANY salary for February 1234,56€ -- finance.pdf"

As you can see, "guessfilename" makes your digital life easier when you do have recurring file rename tasks.

** Usage

#+BEGIN_SRC sh :results output :wrap src guessfilename --help #+END_SRC

#+BEGIN_src Usage: guessfilename []

This little Python script tries to rename files according to pre-defined rules.

It does this with several methods: first, the current file name is analyzed and any ISO date/timestamp and filetags are re-used. Secondly, if the parsing of the file name did not lead to any new file name, the content of the file is analyzed.

You have to adapt the rules in the Python script to meet your requirements. The default rule-set follows the filename convention described on http://karl-voit.at/managing-digital-photographs/

©️ (c) by Karl Voit :license: GPL v3 or any later version :URL: https://github.com/novoid/guess-filename.py :bugreports: via github or [email protected]

Options: -h, --help show this help message and exit -d, --dryrun enable dryrun mode: just simulate what would happen, do not modify files -v, --verbose enable verbose mode -q, --quiet enable quiet mode --version display version and exit #+END_src

** Pixel Images and Videos :PROPERTIES: :CREATED: [2020-11-15 Sun 17:07] :END:

I added handling for [[https://karl-voit.at/2020/11/15/pixel4a-migration/][my Pixel 4a]] camera results: JPEG images and MP4 videos.

Due to [[https://www.reddit.com/r/Pixel4a/comments/jubshe/fixing_the_messy_timestamps_of_pixel_4a_camera/][a somewhat messy meta data situation]] I had to use the =File:FileModifyDate= [[https://en.wikipedia.org/wiki/Exif][Exif]] meta-data in order to get time-stamps from the local time zone. If you happen to apply guessfilename after modifying the file due to copying or editing, you will get wrong time-stamps. Therefore, use [[https://syncthing.net/][Syncthing]] or similar synchronzation tools that preserve file modification time to get the files from the mobile to your computer. Apply guessfilename before modifying the files any further.

Furthermore, you will need to install [[https://exiftool.org/][ExifTool]] as an external dependency. I was not able to find a Python-only Exif library that provided me read access to advanced Exif values the Pixel is using.

** MediathekView :PROPERTIES: :CREATED: [2018-05-10 Thu 17:03] :END:

When downloading TV shows using [[https://github.com/mediathekview/MediathekView][MediathekView]], you should use the following download pattern:

MediathekView v11: : %DT%d %s - %t - %T -ORIGINAL- %N.mp4
MediathekView v13:
- Einstellungen > Aufzeichnen und Abspielen > Set bearbeiten
  - [Set-Name] > Hilfsprogramme:
    - ffmpeg > Zieldateiname > =%DT%d %s - %t - %T -ORIGINALhd- %N.mp4=
    - ffmpeg > Schalter > =-user_agent "Mozilla" -i %f -c copy -bsf:a aac_adtstoasc **=

When applying =guessfilename= on the resulting files, you will get something like this:

#+BEGIN_EXAMPLE 20180509T235000 ORF - ZIB 24 - Auswirkungen nach US-Aus für Atomdeal -ORIGINAL- 2018-05-09_2350_tl_01_ZIB-24_Auswirkungen-na__13976363__o__1735069995__s14297628_8__BCK1HD_23514710P_23540405P_Q4A.mp4 ... → 2018-05-09T23.51.47 ORF - ZIB 24 - Auswirkungen nach US-Aus für Atomdeal -- lowquality.mp4

20180509T235000 ORF - ZIB 24 - Hirntoter Bub plötzlich aufgewacht -ORIGINAL- 2018-05-09_2350_tl_01_ZIB-24_Hirntoter-Bub-p__13976363__o__5119815115__s14297631_1__BCK1HD_00045915P_00072303P_Q4A.mp4 ... → 2018-05-09T00.04.59 ORF - ZIB 24 - Hirntoter Bub plötzlich aufgewacht -- lowquality.mp4

20180509T235000 ORF - ZIB 24 - Meldungen -ORIGINAL- 2018-05-09_2350_tl_01_ZIB-24_Meldungen__13976363__o__1117657593__s14297632_2__BCK1HD_00072303P_00085816P_Q4A.mp4 ... → 2018-05-09T00.07.23 ORF - ZIB 24 - Meldungen -- lowquality.mp4

20180509T235000 ORF - ZIB 24 - Neuerung bei Filmfestspielen in Cannes -ORIGINAL- 2018-05-09_2350_tl_01_ZIB-24_Neuerung-bei-Fi__13976363__o__1941003027__s14297634_4__BCK1HD_00085816P_00111715P_Q4A.mp4 ... → 2018-05-09T00.08.58 ORF - ZIB 24 - Neuerung bei Filmfestspielen in Cannes -- lowquality.mp4

20180509T235000 ORF - ZIB 24 - Trumps CIA-Kandidatin umstritten -ORIGINAL- 2018-05-09_2350_tl_01_ZIB-24_Trumps-Kandidat__13976363__o__1488806017__s14297630_0__BCK1HD_00020922P_00045915P_Q4A.mp4 ... → 2018-05-09T00.02.09 ORF - ZIB 24 - Trumps CIA-Kandidatin umstritten -- lowquality.mp4

20180509T235000 ORF - ZIB 24 - Wetter -ORIGINAL- 2018-05-09_2350_tl_01_ZIB-24_Wetter__13976363__o__2966973785__s14297635_5__BCK1HD_00111715P_00120000P_Q4A.mp4 ... → 2018-05-09T00.11.17 ORF - ZIB 24 - Wetter -- lowquality.mp4 #+END_EXAMPLE

As you can see, the temporal order of the chunks is extracted so that the files are in their correct order.

Please note that this does not work with a show whose chunks do cross midnight since the date is always taken from the start of the show and the time from the actual time being shown.

** .info.json Meta-Data Files :PROPERTIES: :CREATED: [2019-10-19 Sat 15:21] :END:

If you do download a media file and its associated separate =.info.json= file (both base-names without file extension need to match), this tool is able to parse the meta-data to derive a new file name.

Currently, there are two meta-data formats supported: ORG TVthek and YouTube, both via http://rg3.github.io/youtube-dl/

: youtube-dl --write-info-json

This results, for example, with files like these:

: Durchbruch bei Brexit-Verhandlungen-14577219.info.json : Durchbruch bei Brexit-Verhandlungen-14577219.mp4 : Isolierte Familie - 58-jähriger Österreicher in U-Haft-14577221.info.json : Isolierte Familie - 58-jähriger Österreicher in U-Haft-14577221.mp4 : The Star7 PDA Prototype-Ahg8OBYixL0.info.json : The Star7 PDA Prototype-Ahg8OBYixL0.mp4

Please notice the associated =mp4= files as well as the =info.json= files.

Applying guess-filename on these files look like this:

#+BEGIN_EXAMPLE vk@sherri ~tmp % guessfilename *mp4

Durchbruch bei Brexit-Verhandlungen-14577219.mp4 ... → 2019-10-17T16.59.07 ORF - ZIB 17 00 - Durchbruch bei Brexit-Verhandlungen -- highquality.mp4

Isolierte Familie - 58-jähriger Österreicher in U-Haft-14577221.mp4 ... → 2019-10-17T17.01.44 ORF - ZIB 17 00 - Isolierte Familie: 58-jähriger Österreicher in U-Haft -- highquality.mp4

The Star7 PDA Prototype-Ahg8OBYixL0.mp4 ... → 2007-09-13 youtube - The Star7 PDA Prototype - Ahg8OBYixL0.mp4 #+END_EXAMPLE

The =info.json= files are not removed or renamed.

** Extending with your own regular expressions

The structure of the script is like the following:

general header, command-line argument parser, ...
=handle_logging()=
=error_exit()=
=FileSizePlausibilityException()=
=class GuessFilename()=
- a long list of regular expression definitions
- =derive_new_filename_from_old_filename()=
  - here, you can add code to interpret the regular expressions
- =derive_new_filename_from_content()=
  - if you want to parse PDF content, add your code here
- =derive_new_filename_from_json_metadata()=
  - this handles the JSON meta-data files generated by [[https://ytdl-org.github.io/youtube-dl/index.html][youtube-dl]] (see above)
- =handle_file()=
  - the function that loops over all files is probing for new file names until a function is returning with a new name:
    1. =derive_new_filename_from_old_filename()=
    2. =derive_new_filename_from_content()=
    3. =derive_new_filename_from_json_metadata()=
    4. if no name returned until here: prints out a warning that no new name could be derived
- The rest of the class consist of a bunch of tool functions, e.g., for parsing and querying:
- =adding_tags()=
- =split_filename_entities()=
- =contains_one_of()=
- =contains_all_of()=
- =fuzzy_contains_one_of()=
- =fuzzy_contains_all_of()=
- =has_euro_charge()=
- =get_euro_charge()=
- =get_euro_charge_from_context_or_basename()=
- =get_euro_charge_from_context()=
- =rename_file()=
- =get_datetime_string_from_named_groups()=
- =get_date_string_from_named_groups()=
- =get_datetime_description_extension_filename()=
- =get_date_description_extension_filename()=
- =NumToMonth()=
- =translate_ORF_quality_string_to_tag()=
- =get_file_size()=
- =warn_if_ORF_file_seems_to_small_according_to_duration_and_quality_indicator()=
=move_to_success_dir()=
=move_to_error_dir()=
=main()=

For the most basic pattern matching, you just have to add regular expressions to the =GuessFilename()= class and add the regex matching code to =derive_new_filename_from_old_filename()=.

Do not forget to add simple tests to =guessfilename_test.py= as well!

Related tools and workflows

This tool is part of a tool-set which I use to manage my digital files such as photographs. My work-flows are described in [[http://karl-voit.at/managing-digital-photographs/][this blog posting]] you might like to read.

In short:

For tagging, please refer to [[https://github.com/novoid/filetags][filetags]] and its documentation.

See [[https://github.com/novoid/date2name][date2name]] for easily adding ISO time-stamps or date-stamps to files.

For easily naming and tagging files within file browsers that allow integration of external tools, see [[https://github.com/novoid/appendfilename][appendfilename]] (once more) and [[https://github.com/novoid/filetags][filetags]].

Moving to the archive folders is done using [[https://github.com/novoid/move2archive][move2archive]].

Having tagged photographs gives you many advantages. For example, I automatically [[https://github.com/novoid/set_desktop_background_according_to_season][choose my desktop background image according to the current season]].

Files containing an ISO time/date-stamp gets indexed by the filename-module of [[https://github.com/novoid/Memacs][Memacs]].

[[http://www.jonasjberg.com/][Jonas Sjöberg]] took my idea and developed the much more advanced (and thus a bit more complicated) [[https://github.com/jonasjberg/autonameow][autonameow]]. It uses rule-based renaming, analyzes content of plain text, epub, pdf and rtf files, extracts meta-data from many different file formats via [[https://www.sno.phy.queensu.ca/%257Ephil/exiftool/][exiftool]] and so forth.

[[https://www.reddit.com/r/datacurator/comments/f6ku5p/building_an_auto_file_sorter_need_requirements/][This reddit thread]] brought me to [[https://github.com/unreadablewxy/fs-curator][fs-curator]] whose [[https://github.com/unreadablewxy/fs-curator/wiki][documentation]] looks promising. I did not test it and it's still in an early stage. However, it could be a future user-friendly part of a workflow that watches folders for file changes and applies processes like guessfilename.

Alternatives

I you don't need the full power of a programming language, [[https://github.com/tfeldmann/organize][organize]] might do the trick for you. Instead of coding Python, you define your rules within a text file.