FreDraCor (French Drama Corpus, TEI P5)

Corpus description

Based on the extensive work of Paul Fièvre, we have been working on a DraCor-ready version of Théâtre Classique. FreDraCor is intended to be a valid TEI P5 resource.

By now, the 1940 files from the source have been structurally cleaned and converted. All documents are valid against the TEI-All schema. We copied some information from the castList to the particDesc section and tried to preserve as much as possible.

Besides the fact that all texts are out of copyright, the files (including intellectual work represented as markup) is – according to the source files – licensed under a CC BY NC SA 4.0 licence.

The corpus can be explored at dracor.org/fre.

To cite FreDraCor …

… we suggest the following:

• French Drama Corpus (FreDraCor): A TEI P5 Version of Paul Fièvre's "Théâtre Classique" Corpus. Edited by Carsten Milling, Frank Fischer and Mathias Göbel. Hosted on GitHub, 2021–. https://github.com/dracor-org/fredracor

Changes, corrections and additions we have made

Here are, among others, the most significant modifications performed on the original documents:

• add TEI namespace
• add XML declaration
• replace the teiHeader with a DraCor-specific version while preserving as much as possible of original content
• refine licence statement using current version 3.0 of the given licence and adding URL
• add a particDesc
• transform the @xml:id and @who attributes into proper IDs and ID references
• transform numeral @id into @n on tei:s and tei:l
• replace @id with @corresp at castItem/role
• upper-case tei:l/@part
• remove instances of tei:l/@syll (commented)
• remove unknown attributes from tei:role (commented)
• rename docDate/@value to docDate/@when
• transform addresse element to tei:opener/tei:salute
• transform signature element to tei:signed
• remove empty @type
• add written and print dates where available
• adjust case of character names (#14)
• normalize author names
• add Wikidata IDs for authors and plays (work in progress)

For comprehensive insight into our changes see both the adjustments made on the dracor branch of the theatre-classique repository and the tc2dracor.xq transformation script.

DraCor IDs

Each FreDraCor play is given a DraCor ID (e.g. fre000784). These IDs are mapped to the Théâtre Classique documents in ids.xml. When a new play from Théâtre Classique is added to the corpus a new ID needs to be assigned and added to ids.xml.

Validation

To check the current validation status of the corpus against the tei_all schema run ./validate from the root of the repo. (You will need to have Jing installed for this to work.)

In fact, this script can be used to validate any directory of TEI documents. Just pass the directory as the first argument. For instance, if you have gerdracor checked out next to fredracor, try:

./validate ../gerdracor/tei

How to rebuild from source

For building the FreDraCor documents from the Théâtre Classique sources a scripted workflow has been set up that processes the original files with an XQuery transformation. To speed up the process multiple eXist DB instances can be started in parallel using either Podman or Docker. These are the main steps of this workflow:

1. start one or more pods (or containers) running eXist-db
2. loading the transformation XQuery tc2dracor.xq and auxiliary files (authors.xml, ids.xml) to the database(s)
3. process each source file by posting it to the transformation XQuery and storing the output to the tei directory
4. stop and remove all pods (or containers)

Usage

./tc2dracor [options] SOURCE_FILE [SOURCE_FILE...]

SOURCE_FILE

The conversion script expects one or more source files as its arguments. These would typically be files from the xml directory of the checked out dracor branch of the theatre-classique repository:

./tc2dracor ../theatre-classique/xml/*.{xml,XML}

NOTE: The dracor branch of the theatre-classique repo contains corrections and amendments to the original source files which the conversion script relies on but have not (yet) been adopted upstream.

NOTE: For the attribution of DraCor IDs to work, the file names of the source files need to match the ones of the original documents used in ids.xml (see DraCor IDs).

Options

-h, --help

Display usage information and exit.

-n, --num N

Number of pods or containers to start. Default: 1

-p, --port PORTNUMBER

As an alternative to using containers an eXist database already running on localhost can be used by passing its port number. With this option the sources will be copied to the /db/tc2dracor/sources collection of this database. No parallel processing will take place.

-o, --output DIRECTORY

Directory to write the created TEI files to. Default: ./tei

-D, --docker

By default the conversion script uses podman but falls back to docker if podman is not available. This flag allows you to force the use of docker when podman would be available.

-P, --progress

The internet does not forget. That's why the script can be run with an optional progress bar shown in the terminal.

Hint: For debugging across multiple containers you may also watch the combined log from every pod can be viewed while the conversion is running:

podman logs -f $(cat $(ls -rtd /tmp/tc2dracor-* | tail -1)/containers)

For debugging purposes the logs of all containers are also stored in a temporary working directory after the transformation has finished. Use the -v option to see the exact location of these files at the end of the script run.

authors.xml

The transformation process uses the file authors.xml to unify and enrich author information within FreDraCor. The entries in this file provide a canonical tei:author element for each author together with the matching author string in the source documents (in the name elements), e.g.:

<author>
  <author xmlns="http://www.tei-c.org/ns/1.0">
    <persName>
      <forename>Charles</forename>
      <surname>Collé</surname>
    </persName>
    <idno type="isni">0000000121258527</idno>
    <idno type="wikidata">Q2404425</idno>
  </author>
  <name>Charles COLLÉ (1709-1783)</name>
  <name>COLLE, Charles</name>
  <isni>0000 0001 2125 8527</isni>
</author>

When the transformation script discovers an author that does not yet have an entry in authors.xml, trying to properly identify the name parts and also looking up the Wikidata ID if the source TEI provides an ISNI. The new entries are written to the file authors.update.xxxx.xml (where 'xxxx' is the eXist DB port number used to run the transformation). This file should be merged manually into authors.xml.

TODO

See the list of open issues for possible future enhancements.