This tool converts Finereader extended XML documents into a basic TEI file for further editing. It maps some basic format declartion and uncertain characters into the TEI structure.
Download the release from the release tab, open the command line and execute
java -jar ocr2tei.jar with the following options
Usage: Finereader XML to basic TEI converter [command] [command options]
Commands:
file Process a single XML File
Usage: file [options]
Options:
-f, --foreign
Add foreign elements to output
Default: false
-h, --help
print this message
* -i, --input
The source XML OCR File from ABBYY
-lb, --linebreak
Add line breaks to p elements
Default: true
* -o, --output
The path and filename to the generated TEI output file
dir Process recursively a whole directory
Usage: dir [options]
Options:
* -d, --dir
The start path of the directory for processing the files
-f, --foreign
Add foreign elements to output
Default: false
-h, --help
print this message
-lb, --nolinebreak
Add line break to p elements
Default: true
-p, --pattern
The regex pattern for matching the OCR xml files
Default: ocr_(.*).xml
-s, --specific
Do very project specific directory scanning (outside of this
project do NOT set this option to true)
Default: false
Example for processing a single file
java -jar ocr2tei.jar file -i input.xml -o output.xml
Example for processing a whole directory
java -jar ocr2tei.jar dir -i -d /path/to/directory -p "(.*).xml"
Note that the pattern uses a Java regular expression, a simple *.xml will NOT work.
The linebreak element just controls the output of the XML p blocks. Semantically they are the same.
No line break
<p>This is the first line. This is the second line</p>
Line breaks
<p>This is the first line.
This is the second line</p>
The vendor provides a zip file with some samples. Extract it and use any of the DemoImage1_ABBYY_Char* files. The Basic file is not supported (see next headline). If you want to process the directory, use the option -p "DemoImage1_ABBYY(.*).xml" as pattern.
This tool works only if the ocr xml has been generated with the character by character generation, so not with the basic format.
Currently only text elements get processed, image and table elements are ignored. The following text attributes are transfered to TEI:
- italic ->
<hi rend="italic"> italic></hi> - bold ->
<hi rend="bold">bold></hi> - underline ->
<hi rend="underline">underline></hi> - strikethrough ->
<hi rend="strikethrough">strikethrough></hi> - super
<hi rend="super">super</hi> - unclear characters ->
<unclear cert="unknown" reason="illegible">l</unclear> - document language ->
<TEI lang="en"> - other languages as foreign element if -f parameter has been set ->
<foreign lang="lat">...</foreign>
Just call mvn to build it.
This tool uses JAXB to read and write the XML files. A newer input XSD can be added by using
xjc -d ocr2tei/src/main/java -p de.sub.goettingen.arendt.input.generated FineReader10-schema-v1.xml
The output does not contain the whole TEI schema. Instead a simple outputTemplate.xml has been created (see resources folder), an XSD has been generated from the XML and the JAXB classes have been created with
xjc -d ocr2tei/src/main/java -p de.sub.goettingen.arendt.output.generated outputTemplate.xsd
Afterwards you have to adapt the classes under de.sub.goettingen.arendt.ocrmapping for the changed structure.
- Support for XPath expression before processing
- extracting of main document language
- each TEI document gets an ID
- support for foreign element if specified via command line option
- improved whitespace handling. The OCR output is missing some whitespaces, this version tries the best to fix the whitespaces
- changed XML id for Arendt specific directory processing
The documentation of the Finereader XML File can be found here, a examplanation of the schema is located here.
Apache 2.0.
