alix-tz / aspyre-gt Goto Github PK

View Code? Open in Web Editor NEW

7.0 2.0 0.0 16.19 MB

A pipeline to transfer ground truth from Transkribus to eScriptorium.

License: MIT License

Python 100.00%

alto-xml alto escriptorium transkribus htr pipeline

aspyre-gt's Introduction

ASPYRE GT

A converter to help making your data compatible for import in eScriptorium.

How to use Aspyre

As a library
As a CLI
~~As a service online (GUI)~~

As a library

Aspyre is a library. To install it, simply download aspyrelib/ and make sure to install the dependencies! Use from aspyrelib import aspyre to import it in your program.

Parsing parameters with `aspyre.AspyreArgs()`

Start your project parsing all the required information with AspyreArgs() objects.

Process essential information to run Aspyre
          :param scenario: keyword describing the scenario (string)
          :param source: path to source file (string)
    [opt] :param destination: path to output (string)
    [opt] :param talkative: activate a few print commands (bool)
    [opt] :param vpadding: value to add to VPOS attr. in String nodes (int)

supported values for scenario: "tkb", "pdfalto", "limb"

vpadding is only used in PDFALTO and LIMB scenarios

Transkribus to eScriptorium scenario with `aspyre.TkbToEs()`

⚠️ really not the best way to transfer data between these two softwares.

Run Transkribus to eScriptorium (mainly resolve schema declaration, source image information).

Handle a Transkribus to eScriptorium transformation scenario
        :param args: essential information to run transformation scenario (AspyreArgs)

PDFALTO to eScriptorium scenario with `aspyre.PdfaltoToEs()`

Run PDFALTO to eScriptorium scenario (mainly resolve schema declaration, source image information and homothety)

Handle a PDFALTO to eScriptorium transformation scenario
        :param args: essential information to run transformation scenario (AspyreArgs)

As a CLI

A legacy script (run.py) from earlier stage enables you to use Aspyre as a CLI fairly easily.

Step by step (Transkribus scenario)

Export the transcriptions and the images from Transkribus; you now have a zip file

Create a virtual environment based on Python 3 and install dependencies (cf. requirements.txt)
Run aspyre/run.py (python3 aspyre/run.py) with the fitting options
See the CLI's options with --help* (python3 aspyre/run.py --help)
Aspyre will create a new ZIP that can be loaded onto eScriptorium

Example

$ virtualenv venv -p python3
$ source venv/bin/activate
(venv)$ pip install -r requirements.txt 
(venv)$ python3 aspyre/run.py -i /path/to/exported/documents

As a service online

This is no longer an option, following Heroku's decision in 2021 to stop supporting free hosting services.

You can now access Aspyre as a service online (GUI)! ➡️ go to Aspyre GUI

Step by step (Transkribus scenario)

~~Export the transcriptions and the images from Transkribus; you now have a zip file~~
~~If your archive weighs more than 500 MB, remove the images from the zip file (unzip the archive and rezip it keeping only the alto/ directory and the 'mets.xml' file)~~
~~Load the zip file onto the application and download the returned zip file~~
~~You can now directly load this new ZIP onto eScriptorium~~

Configuring the export from Transkribus

Export your data checking the “Transkribus Document” format option and checking the “Export ALTO” and “Export Image” sub-options.

Which input from PDFALTO?

Contenu minimum:

dossier(.zip)/
    - out/
        - identifiant.xml_data/
            - image-1.png
        - identifiant.xml

Pour le moment les archives tar.gz ne sont pas supportées. Seules les archives zip le sont.

Reporting Errors

If you notice unexpected errors or bugs or if you wish to add more complexity to the way Aspyre transforms the ALTO XML files, please create an issue and contribute!

Wiki

What was the problem?

aspyre-gt's People

Contributors

Stargazers

Watchers

aspyre-gt's Issues

Make Aspyre available via Pip ?

Once we have a v1.

eScriptorium fails to import the segmentation...

This can be because :

Transkribus created absurd coordinates such as "-9.999999E6" - see https://gitlab.inria.fr/dh-projects/aspyre-gt/-/issues/6

Créer un scénario pour intégrer pdfalto

Il s'agirait d'utiliser Aspyre pour créer un système de conversion pour passer des ALTO XML (3) produits par le script pdfalto en intégrant les modifications nécessaires (schéma, filename, ...) et la question de l'homothétie (qui se pose aussi pour les ALTO de Limb (#15).

<?xml version="1.0" encoding="UTF-8"?>
<!-- added manually for compatibility with eScriptorium -->
<alto xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
      xmlns="http://www.loc.gov/standards/alto/ns-v2#"
      xmlns:page="http://schema.primaresearch.org/PAGE/gts/pagecontent/2013-07-15"
      xsi:schemaLocation="http://www.loc.gov/standards/alto/ns-v2# http://www.loc.gov/standards/alto/alto.xsd">
<!-- end of added -->
<!--<alto xmlns="http://www.loc.gov/standards/alto/ns-v3#">-->
<Description>
<MeasurementUnit>pixel</MeasurementUnit>
<sourceImageInformation>
	<!-- <fileName>test_aspyre/1903 159_258 3.pdf</fileName> -->
	<!-- added manually for compatibility with eScriptorium -->
	<fileName>Annuaire_1903 161.tif</fileName>
	<!-- end of added -->
</sourceImageInformation>
<OCRProcessing ID="IdOcr">
...

Autre modification: pdfalto met des float dans les attributs comme "HEIGHT", "WIDTH", "HPOS" etc, au lieu de int.

Cette feature est une solution pour répondre à https://gitlab.inria.fr/scripta/escriptorium/-/issues/331

~~aspyre/main.py n'existe pas, c'est aspyre/run.py, je crois;~~
(du coup, dans le readme, est-ce que tout ne serait pas fait en
préservant le chemin relatif depuis la racine du dépôt?)
~~fournir un --help avec la commande ?~~ -> --help fonctionne avec run.py ; help() fonctionne sur aspyre.main()
~~TRP Export = Transkribus export ?~~ -> corrigé (+ explicit)
Ajouter la commande zip dans le README ? Important, car ne faut pas
ziper le dossier mais les fichiers directement (sinon, ça ne fonctionne
pas), et que du coup, il faut l'option -j -> à faire dans v.0.2.4

~~Also update the doc now that Aspyre is a library~~

Application error

An error occurred in the application and your page could not be served. If you are the application owner, check your logs for details. You can do this from the Heroku CLI with the command
heroku logs --tail

FF 111.0.1 (64-bit), Ubuntu 22.04

Add a period at the end of "Aspyre GUI is a simple application to make Aspyre GT available as a service online"
Change "compatible for import" to "compatible with importation" in the following sentence. "Aspyre GT is a pipeline to make Ground Truth exported from Transkribus compatible for import into eScriptorium using ALTO XML as a pivotal format."
Add "the" in the following sentence: "Use the Transkribus export option to download a ZIP file containing image files, ALTO XML files and a METS XML file"
To remain consistent with the previous step, add an "s" to the following": "you may need to remove the image files from the ZIP file in..."
Add punctuation: "initially checking the "Image" option on Transkribus**,** however**,** is necessary in order to get a properly documented METS XML file..."

alix-tz / aspyre-gt Goto Github PK

aspyre-gt's Introduction

ASPYRE GT

SUMMARY

How to use Aspyre

As a library

Parsing parameters with aspyre.AspyreArgs()

Transkribus to eScriptorium scenario with aspyre.TkbToEs()

PDFALTO to eScriptorium scenario with aspyre.PdfaltoToEs()

As a CLI

Step by step (Transkribus scenario)

Example

As a service online

Step by step (Transkribus scenario)

Configuring the export from Transkribus

Which input from PDFALTO?

Reporting Errors

Wiki

aspyre-gt's People

Contributors

Stargazers

Watchers

aspyre-gt's Issues

Recommend Projects

Recommend Topics

Recommend Org

Parsing parameters with `aspyre.AspyreArgs()`

Transkribus to eScriptorium scenario with `aspyre.TkbToEs()`

PDFALTO to eScriptorium scenario with `aspyre.PdfaltoToEs()`