imgtk
Image file manipulation toolkit
This tool provides image file manipulation subcommands. Currently supported commands are as follows:
| sub command | Description |
|---|---|
| join | Join multiple image files into one PDF file. |
| sort | Sort image files into folders using exif date |
| dedup | Find duplicate images in folder using image hash |
Installation
> pip install imgtk
Note: Strongly recomend using pipx to isolate package dependencies.
Usage : join
Convert set of images to PDF.
This tool is intended to merge multiple scanned images into one PDF file. The tool will take multiple images in order, a directory of images or both and merge them into a PDF file in the order it is given. Images in the directory will be searched by specified extensions and sorted by name.
Usage: imgtk join [OPTIONS] [IMG]...
Join images into one PDF file
Options:
-c, --config <cfg> Configuration File (default: imgtk-join.yml)
-o, --out <out> output filename for the generated PDF
-t, --toc <toc> toc file to populate PDF outline
-e, --ext <ext> file extensions to pick up when parsing
directories
-d, --dpi <dpi> pixel density of input image in dpi
-P, --page-layout <layout> page layout (/PageLayout) in the PDF document
catalog: SinglePage, OneColumn, TwoColumnLeft,
TwoColumnRight, TwoPageLeft, TwoPageRight, none
(default: TwoPageRight)
-D, --direction <dir> reading order (/Direction) in the PDF viewer
preferences: L2R, R2L, none (default: R2L)
-m, --page-mode <mode> initial view mode (/PageMode) in the PDF
document catalog: UseOutlines, UseNone, UseOC,
UseAttachments, FullScreen, none (default: UseOutlines
when a TOC outline was written, otherwise
no entry)
-z, --zoom <zoom> initial magnification (/OpenAction Go To on the
first page): default, none, actual, NN% such as
150%, fit_page, fit_width, fit_height, fit_visible,
fit_visible_width, fit_visible_height (default:
default, i.e. do not write an /OpenAction)
-v, --verbose output in verbose mode
-h, --help Show this message and exit.
Options (join)
-c, --config <cfg>- Configuration File (default: imgtk-join.yml)
- the name is resolved relative to the current working directory. If the file does not exist, join writes the defaults shown below into it and then reads them back.
-o, --out <out>- output filename for the generated PDF
-t, --toc <toc>- toc file to populate PDF outline (TBD)
-e, --ext <ext>- file extensions to pick up when parsing directories. directories will be searched for all files using this extensions.
- can set multiple items.
-d, --dpi <dpi>- pixel density of input image in dpi. used to calculate image width and height in mm.
-P, --page-layout <layout>- page layout (
/PageLayout) written into the PDF document catalog.<layout>is one ofSinglePage,OneColumn,TwoColumnLeft,TwoColumnRight,TwoPageLeft,TwoPageRightornone. - default:
TwoPageRight(a two-page spread, matching theR2Lreading order). noneremoves the/PageLayoutentry from the PDF, including theOneColumnthe PDF generator would otherwise write. Leaving the option out entirely is different: it selects the default. See "PDF viewer defaults (document catalog)".
- page layout (
-D, --direction <dir>- predominant reading order (
/Direction) written into theViewerPreferencesdictionary of the PDF document catalog.<dir>is one ofL2R,R2Lornone. - default:
R2L(right to left), which makes viewers such as Acrobat open the document as a right-to-left book. noneremoves/Directionfrom the PDF, letting the PDF default (L2R) apply. Leaving the option out selects the default instead.
- predominant reading order (
-m, --page-mode <mode>- initial view mode (
/PageMode) written into the PDF document catalog.<mode>is one ofUseOutlines,UseNone,UseOC,UseAttachments,FullScreenornone. - default:
UseOutlineswhen the run wrote a PDF outline (i.e. atoc.yml-based outline was emitted), otherwise the entry is left out of the PDF entirely. noneremoves/PageModefrom the PDF even when an outline was written. Leaving the option out selects the auto default above instead.
- initial view mode (
-z, --zoom <zoom>- initial magnification written as an
/OpenActionGo To action on the first page (PDF 32000-2008 section 3.3.6).<zoom>is one of:default(do not keep the PDF generator's default /OpenAction),none(same, remove it),actual(open at 100%),NN%(fixed magnification such as150%or75%), or one of the following:fit_page(/Fit),fit_width(/FitH),fit_height(/FitV),fit_visible(/FitB),fit_visible_width(/FitBH),fit_visible_height(/FitBV). - default:
default(no/OpenActionin the output, so the viewer decides; the PDF generator would otherwise have written a full-width/OpenAction). noneanddefaulthave the same effect.
- initial magnification written as an
-v, --verbose- output in verbose mode
- specify twice to get debug messages
-h, --help- Show help message
Configuration file
YAML based configuration file (default file name: imgtk-join.yml) is used to store default settings for the tool. The name is resolved relative to the current working directory, and the file is created with the defaults shown below when it is missing. The command line options will take precedence over the configuration parameters.
Following shows the default settings of the configuration
out: output.pdf
toc: toc.yml
dpi: 96
ext:
- jpg
- jpeg
- png
page_layout: TwoPageRight
direction: R2L
PDF viewer defaults (document catalog)
join writes four entries into the PDF document catalog so viewers such as
Adobe Acrobat open the generated book the way the scanned pages are meant to be
read. The keys follow the PDF 32000-2008 specification and can be set in the
configuration file or on the command line (command line wins):
| config key | command line | PDF catalog entry | default |
|---|---|---|---|
page_layout |
-P, --page-layout |
/PageLayout (Table 3.2) |
TwoPageRight |
direction |
-D, --direction |
/ViewerPreferences /Direction (Table 3.22) |
R2L |
page_mode |
-m, --page-mode |
/PageMode (Table 3.2) |
UseOutlines if the run wrote an outline, otherwise omitted |
zoom |
-z, --zoom |
/OpenAction (Table 3.3.6) Go To on page 1 |
no entry |
Valid page_layout values are SinglePage, OneColumn, TwoColumnLeft,
TwoColumnRight, TwoPageLeft and TwoPageRight; the two TwoPage* values
raise the generated file to PDF 1.5 as the specification requires.
Valid direction values are L2R and R2L.
Valid page_mode values are UseOutlines, UseNone, UseOC,
UseAttachments and FullScreen.
Valid zoom values are default, none, actual, NN% (e.g. 150%,
75%) and the six magnification keywords fit_page, fit_width,
fit_height, fit_visible, fit_visible_width, fit_visible_height.
Each key has three states, and they are not all the same thing:
- omitted from the configuration file and not given on the command line: the tool default shown in the table above is written to the PDF.
- set to a value: that value is written to the PDF. Command-line values win over the configuration file.
none(configuration file or command line): the entry is removed from the PDF, which is not the same as leaving it out of the configuration file./PageLayoutin particular is removed, overriding theOneColumnthat the PDF generator would otherwise write.
Configuration values are matched case-insensitively and the separators -,
_ and spaces are ignored, so page_layout: two page left,
page_layout: two_page_left and page_layout: TwoPageLeft all mean the same.
A bad configuration value is reported and makes join exit unsuccessfully; an
invalid command-line value is rejected by the option parser before any file is
written.
The PDF is written by pikepdf into a temporary file next to the target and then renamed into place, so a failing run never leaves a half-written PDF behind, and the temporary files are always cleaned up. Because pikepdf (qpdf) re-serializes the document, the generated file is not byte-identical to what the PDF generator alone would produce.
These catalog entries are applied with pikepdf, so pikepdf is a runtime
dependency of join.
TOC File (default name: toc.yml)
Consist of dictionary with following structure, where level is the
outline level (starting from 0), and title is the title of the
outline and will be inserted at pagenum. The keys are case
sensitive.
pagenum:
- level: Num
title: 'Title'
Note, each entry inside the page must be an array of dictionaries. This is to allow multiple ToC item to be inserted in one page.
Example may be as follows:
1:
- level: 0
title: 'Getting Started'
- level: 1
title: 'Installation'
3:
- level: 1
title: 'Hello, World'
5:
- level: 0
title: 'Programming a Guessing Game'
Usage : sort
Command to sort image files in one folder to another by date of image (exif meta-data or file creation date).
Moves/Copies the images inside SRCDIR into sorted folders under TGTDIR. Target folders will be sorted based on the date the image was created (Referes to Exif meta-data or the file creation date if exif is not available).
Usage: imgtk sort [OPTIONS] SRCDIR TGTDIR
Sort and Move/Copy images from SRCDIR to TGTDIR.
Options:
--copy Copy files from SRCDIR to TGTDIR (Default)
--move Move files from SRCDIR to TGTDIR
-r, --recurse Search for images in SRCDIR recursively
-o, --overwrite Overwrite files in TGTDIR
-H, --hierarch Create a hierarchical directory for Year, Month and Date in
TGTDIR
-d, --dry-run Dry run the command without moving
-f, --fmt <fmt> Format target directory name using date format strftime
string give in <fmt> (default: %Y-%m-%d). Ignored when
--hierarch is specified.
-v, --verbose output in verbose mode
-h, --help Show this message and exit.
Options (sort)
--copy/--move- specify whether the file should be copied or moved. Files will be copied if not specified.
-r, --recurse- search images in SRCDIR and any subdirectories.
-o, --overwrite- overwrite any image files in TGTDIR
-H, --hierarch- create a folder for year, month and date in a hierarchical manner instead of one folder for each date.
-d, --dry-run- dry run the command. nothing will be moved
-f, --fmt DATEFORMAT- format the folder name using date format string specified by DATEFORMAT
- ignored when "-H/--hierarch" is specified. --hierarch always nests the target as %Y/%m/%d regardless of the format string.
-v, --verbose- verbose mode
Usage : dedup
This is a tool to find similar images in a set of images using imagehash library. If two or more similar images are found (same hash value), the images are moved to a subfolder for each hash value.
Usage: imgtk dedup [OPTIONS] PATH
Find duplicates in PATH using <hashfunc>
<hashfunc> can be any of the following:
ahash : Average hash
phash : Perceptual hash
dhash : Difference hash (default)
haar : Haar wavelet hash
db4 : Daubechies wavelet hash
color : HSV color hash
crop : crop-resistant hash
Options:
-s, --hash-size <size> hash size to use for image hashing (default: 8;
accepted but not implemented)
-f, --hash-func <hashfunc> hash function to use for comparing images
-v, --verbose output in verbose mode
-h, --help Show this message and exit.
Options (dedup)
PATH- PATH with the images to check for (will not be recursed)
-s, --hash-size SIZE- specify the hash size to use. (Default: 8)
- NOT IMPLEMENTED: the value is only range-checked and is never passed to the hash functions, so every hash function runs at its own built-in default size and the option does not change which images are grouped. See Known Issues.
-f, --hash-func- ahash
- Use Average hash to find duplicates in PATH
- color
- Use HSV color hash to find duplicates in PATH
- crop
- Use Crop resistant hash to find similar images in PATH
- db4
- Use Daubechies wavelet hash to find duplicates in PATH
- dhash
- Use Difference hash to find duplicates in PATH
- haar
- Use Haar wavelet hash to find duplicates in PATH
- phash
- Use Perceptual hash to find duplicates in PATH
- ahash
-v, --verbose- specify verbosity. specify twice to get debug message
If two or more images with the same hash is found, they are put into the same folder with tha hash value as the folder name.
Known Issues
dedup --hash-size/-sis accepted but not implemented. The value is only range-checked ("Hash size must be a positive integer"); it is never forwarded to the hashing helpers, so every hash function runs at imagehash's built-in default size (8) and-s 4selects exactly the same duplicate groups as-s 8. Marked with aTODO:inimgtk/deduplicate/deduplicate.py.dedupdecides which files are images by name only. As well as the documented extensions, any file whose name merely contains.jpg(for examplescan.jpg.bak) is hashed and reports an unreadable-image error, and.svgis listed but Pillow cannot decode it without the SVG plugin. Unreadable files are reported and skipped; the run still succeeds.sort's image filter listsheicwithout a leading dot, so any file whose name ends inheic(notes-heic,archive-scheme-style names) is treated as an image and moved/copied using its file date. Genuine.heicfiles also need Pillow's HEIF/HEIC plugin to be readable.- See github issues for latest issues.
Development
Building an Executable
Install pyinstaller and package the project. May want to use venv when executing the pyinstaller.
First, enter venv and install the local package and pyinstaller
>. .venv/Scripts/activate
(.venv) >pip install .
Processing /path/to/proj/imgtk
~snip~
Installing collected packages: imgtk
Running setup.py install for imgtk ... done
Successfully installed imgtk-0.1.0
(.venv) >pip install pyinstaller
~snip~
Successfully installed pyinstaller-3.6
Use pyinstaller to build the exe file.
(.venv) >pyinstaller imgtk\cli.py --onefile --name imgtk
~snip~
13691 INFO: Building EXE from EXE-00.toc completed successfully.
Executable should be ready in dist/imgtk.exe
Versioning
The project will follow the semver2.0 versioning scheme. With initial development phase starting at 0.1.0 and increasing minor/patch versions until we deploy the tool to production (and reach 1.0.0). The interface relevant to versioning is whatever defined in this document's "Usage" section (includes all (sub)commands, their cli arguments, and the format of the configuration file "imgtk-join.yml").
Version History
| Date | Version | Changes |
|---|---|---|
| 2026.10.20 | 1.3.0 | implement test |
| implement --page-layout in join subcommand | ||
| implement --direction in join subcommand | ||
| implement --page-mode in join subcommand | ||
| implement --zoom in join subcommand | ||
| 2026.04.28 | 1.2.2 | update dependencies |
| 2026.01.13 | 1.2.1 | Fix call to FPDF failing |
| 2025.09.12 | 1.2.0 | Parallelize dedup feature (x6 speedup on 8core) |
| 2025.07.10 | 1.1.1 | fix Version History |
| 2025.07.10 | 1.1.0 | Bump Pillow to 11.x |
| 2023.11.28 | 1.0.1 | fix issues with Readme. |
| 2023.11.12 | 1.0.0 | Bump to 1.0.0. Update Pillow to latest |
| 2023.01.26 | 0.1.0 | First Release |
Metadata
Release files for imgtk 1.3.0
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | |
|---|---|---|---|
| imgtk-1.3.0.tar.gz | 45.9 kB | Details |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Reset |
|---|---|---|---|---|
| imgtk-1.3.0-py3-none-any.whl | Python 3 | none | any | Details |
Total release size: 78.2 kB
Release files / imgtk-1.3.0.tar.gz
| Download URL | imgtk-1.3.0.tar.gz |
|---|---|
| Size | 45.9 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
7c0a7a7b1a135d2a15010d58793caaff654e8c717d0bcd6e68673dd740b33644
|
|
BLAKE2b-256 checksum How to use checksums |
a2e6d76e33a0050b6e747111125e2c36b8f63645ba8bec3d2a0a3c7ec22ddf7f
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/7.0.0 CPython/3.14.7
|
Release files / imgtk-1.3.0-py3-none-any.whl
| Download URL | imgtk-1.3.0-py3-none-any.whl |
|---|---|
| Size | 32.2 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
f09938626fbd3d6f80c2dab92f443849826b14bfc17c77aed9c27ae283bfd5eb
|
|
BLAKE2b-256 checksum How to use checksums |
b857d02653fd97a00d23be5ab9d6f01bafed99b64f5be56fc9013637e0d85f4e
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/7.0.0 CPython/3.14.7
|