Preview-generator Documentation

Presentation

preview-generator is a library for generating preview - thumbnails, pdf, text and json overview for all your file-based content. This module gives you access to jpeg, pdf, text, html and json preview of virtually any kind of file. It also includes a cache mechanism so you do not have to care about preview storage.

By creating this module, the goal was to delegate the responsibility of building preview of files managed by tracim.

The changelog is available in a separate file.

Supported file formats

Supported mimetypes depends on builders and dependencies installed.

full supported list file format here: supported mimetypes list .

Preview-generator has builders for:

Images

• Raster graphic image formats : png, jpeg, bmp, tiff, gif, …
• Vectorial graphic format: svg
• Camera raw format: dng, arw, …

Office/Text Document

• simple text file : txt, json, …
• office document: odt, doc, docx
• pdf document

Those file formats are generated using libreoffice. The preview generation has a default timeout of 60 seconds. It is possible to change this timeout by setting the LIBREOFFICE_PROCESS_TIMEOUT environment variable to a number of seconds. Setting a zero or negative value for this variable will disable the timeout.

Archive file

• zip, tar, …

3D File

• support for geometric file: .ply, .obj, .stl, .gltf/.glb

Video format

• support for multiples image preview of video : .mp4, .flv, .webm, .ogv, .wmv, etc.

Installation

Mandatory Dependencies:

On debian :

apt-get install poppler-utils libfile-mimeinfo-perl libimage-exiftool-perl ghostscript libsecret-1-0 zlib1g-dev libjpeg-dev imagemagick libmagic1 webp

install preview_generator without external addons:

pip install preview-generator

To install all previews builders dependencies:

pip install preview-generator[all]
sudo apt-get install libreoffice inkscape ffmpeg xvfb
DRAWIO_VERSION="15.7.3" && curl -LO https://github.com/jgraph/drawio-desktop/releases/download/v${DRAWIO_VERSION}/drawio-x86_64-${DRAWIO_VERSION}.AppImage && mv drawio-x86_64-${DRAWIO_VERSION}.AppImage /usr/local/bin/drawio

To check dependencies, you can run:

preview --check-dependencies

Office Files (LibreOffice)

apt-get install libreoffice

DTP(Scribus)

If you need to preview scribus .sla files you will need scribus >= 1.5.

On debian :

apt-get install scribus xvfb
pip install preview-generator[scribus]

If scribus >=1.5 is not available in your distribution you can use an AppImage.

Download the last AppImage from the official website https://www.scribus.net/downloads/unstable-branch/

mv /path/to/image/scribus-x.y.appimage /usr/local/bin/scribus
chmod +x /usr/local/bin/scribus

Vector Images (Inkscape)

on debian:

apt-get install inkscape

Vector Images (cairosvg)

pip install preview-generator[cairosvg]

Video(ffmpeg)

On debian :

apt-get install ffmpeg
pip install preview-generator[video]

RAW Images(ufraw-batch)

On debian :

pip install preview-generator[raw]

Diagram(DrawIO)

1. install xvfb
2. install draw-io package:
3. install python specific dependencies:

on debian:

apt install xvfb
DRAWIO_VERSION="15.7.3" && curl -LO https://github.com/jgraph/drawio-desktop/releases/download/v${DRAWIO_VERSION}/drawio-x86_64-${DRAWIO_VERSION}.AppImage && mv drawio-x86_64-${DRAWIO_VERSION}.AppImage /usr/local/bin/drawio
pip install preview-generator[drawio]

3D files (VTK)

On debian :

pip install preview-generator[3D]

HEIC support

Building ImageMagick with heic support: Building ImageMagick with heic support .

EPS support

You need to edit the policies of ImageMagick in /etc/ImageMagick-*/policy.xml.

<policy domain="coder" rights="none" pattern="ESP" />

Just wrap it between <!-- and --> to comment it.

Usage

Here are some examples of code

Basic Usage

Most basic usage, create a jpeg from a png, default size 256x256

from preview_generator.manager import PreviewManager

cache_path = '/tmp/preview_cache'
file_to_preview_path = '/tmp/an_image.png'

manager = PreviewManager(cache_path, create_folder= True)
path_to_preview_image = manager.get_jpeg_preview(file_to_preview_path)

Preview an image with a specific size

You can choose the size of your image using params width and height.

from preview_generator.manager import PreviewManager

cache_path = '/tmp/preview_cache'
file_to_preview_path = '/tmp/an_image.png'

manager = PreviewManager(cache_path, create_folder= True)
path_to_preview_image = manager.get_jpeg_preview(file_to_preview_path, width=1000, height=500)

Preview a pdf or an office document as a jpeg

from preview_generator.manager import PreviewManager

cache_path = '/tmp/preview_cache'
pdf_or_odt_to_preview_path = '/tmp/a_pdf.pdf'

manager = PreviewManager(cache_path, create_folder= True)
path_to_preview_image = manager.get_jpeg_preview(pdf_or_odt_to_preview_path)

By default it will generate the preview of the first page of the document. Using params page, you can you pick the page you want to preview.

page number starts at 0, if you want to preview the second page of your document then the argument will be 1 `page=1`

from preview_generator.manager import PreviewManager

cache_path = '/tmp/preview_cache'
pdf_or_odt_to_preview_path = '/tmp/a_pdf.pdf'

manager = PreviewManager(cache_path, create_folder= True)
path_to_preview_image = manager.get_jpeg_preview(pdf_or_odt_to_preview_path, page=1)

Generate a pdf preview of a libreoffice text document

from preview_generator.manager import PreviewManager
manager = PreviewManager('/tmp/cache/', create_folder= True)
pdf_file_path = manager.get_pdf_preview('/home/user/Documents/report.odt', page=2)
print('Preview created at path : ', thumbnail_file_path)

For Office types into PDF :

cache_path = '/tmp/previews'
preview_manager = PreviewManager(cache_path, create_folder= True)
path_to_preview = preview_manager.get_pdf_preview(file_path,page=page_id)

-> Will create a preview from an office file into a pdf file

args :

file_path : the String of the path where is the file you want to get the preview

page : the int of the page you want to get. If not mentioned all the pages will be returned. First page is page 0

returns :

str: path to the preview file

For images(GIF, BMP, PNG, JPEG, PDF) into jpeg :

cache_path = '/tmp/previews'
preview_manager = PreviewManager(cache_path, create_folder= True)
path_to_preview = preview_manager.get_jpeg_preview(file_path,height=1024,width=526)

-> Will create a preview from an image file into a jpeg file of size 1024 * 526

args :

file_path : the String of the path where is the file you want to get the preview

height : height of the preview in pixels

width : width of the preview in pixels. If not mentioned, width will be the same as height

returns :

str: path to the preview file

Other conversions :

The principle is the same as above

Zip to text or html : will build a list of files into text/html inside the json

Office to jpeg : will build the pdf out of the office file and then build the jpeg.

Text to text : mainly just a copy stored in the cache

Command Line

For test purposes, you can use preview from the command line, giving the file to preview as a parameter:

preview demo.pdf

Or multiple files:

preview *.pdf

Cache mechanism

Naming :

The name of the preview generated in the cache directory will be :

{file_name}-[{size}-]{file_md5sum}[({page})]{extension}

file_name = the name of the file you asked for a preview without the extension.

size = the size you asked for the preview. In case of a Jpeg preview.

file_md5sum = the md5sum of the entire path of the file. To avoid conflicts like files that have the same name but are in different directory.

page = the page asked in case of pdf or office document preview.

extensions = the extension of the preview (.jpeg for a jpeg, .txt for a text, etc)

Example :

These scripts :

GIF to JPEG :

import os
from preview_generator.manager import PreviewManager
current_dir = os.path.dirname(os.path.abspath(__file__)) +'/'

manager = PreviewManager(path=current_dir + 'cache')
path_to_preview = manager.get_jpeg_preview(
    file_path=current_dir + 'the_gif.gif',
    height=512,
    width=512,
)

print('Preview created at path : ', path_to_preview)

will print

Preview created at path : the_gif-512x512-60dc9ef46936cc4fff2fe60bb07d4260.jpeg

ODT to JPEG :

import os
from preview_generator.manager import PreviewManager
current_dir = os.path.dirname(os.path.abspath(__file__)) +'/'

manager = PreviewManager(path=current_dir + 'cache')
path_to_file = manager.get_jpeg_preview(
    file_path=current_dir + 'the_odt.odt',
    page=1,
    height=1024,
    width=1024,
)

print('Preview created at path : ', path_to_preview)

will print

Preview created at path : the_odt-1024x1024-c8b37debbc45fa96466e5e1382f6bd2e-page1.jpeg

ZIP to Text :

import os
from preview_generator.manager import PreviewManager
current_dir = os.path.dirname(os.path.abspath(__file__)) +'/'

manager = PreviewManager(path=current_dir + 'cache')
path_to_file = manager.get_text_preview(
    file_path=current_dir + 'the_zip.zip',
)

print('Preview created at path : ', path_to_file)

will print

Preview created at path : the_zip-a733739af8006558720be26c4dc5569a.txt

Known Issues

InputExtensionNotFound or UnsupportedMimeType

The current mimetype/file-extension database of preview-generator may differ between systems which means you can struggle to get previews depending on your OS or you installed software.

In case you get one of these exceptions and the mimetype/extension you tried is marked as supported by preview generator, you should: - check the version of PG you are using. - check you have the proper dependencies to make you're builder work as expected. - check if the mimetype of the format you are using is handled by preview_generator mimetype_storage (python console):

from preview_generator.extension import mimetypes_storage
mimetypes_storage.guess_all_extensions('application/vnd.openxmlformats-officedocument.wordprocessingml.document')
['.docx']
mimetypes_storage.guess_type('toto.docx', strict=False)
('application/vnd.openxmlformats-officedocument.wordprocessingml.document', None)

In case you don't get proper result for your file_extension/mimetype, you can work around the issue this way:

from preview_generator.extension import mimetypes_storage
mimetypes_storage.add_type(
    'application/vnd.openxmlformats-officedocument.wordprocessingml.document', '.docx')

Feel free to propose an upstream patch to add the proper MimetypeMapping to the builder you're using.

Support for 3D file on headless server

Standard vtk library require an X server to run properly. To make 3D file preview work on a headless server, you have two option:

1. Untested: Compile/find a vtk version correctly compiled with flags to disable x requirement (with OSMesa support)
2. Use an x framebuffer like xvfb

For the last one, this configuration is known to work:

1. install package xvfb.
2. launch Xvfb :99 -screen 0 1x1x16 > /dev/null 2>&1 & (note: this use a very small x framebuffer 1x1 pixel in 16 color depth to limit video usage).
3. run python code with env var DISPLAY=:99.0

Process Stuck during generation of big spreadsheet preview

With libreoffice version like 6.0.7.3, the generation of some big spreadsheet files (like xlsx) takes a very long time, until it reaches a 1 minute timeout. The solution is to update libreoffice.

During our internal tests, we found out that the spreadsheet file that failed to be generated in 6.0.7.3 is properly generated in 6.4.2.2. We advise you to use at least this version to avoid this kind of issues.

Contribute and Developer’s Kit

see here .

License

See LICENSE