The script converts data with PDT 1.0 and similar analytical annotation in CSTS, FS or any other format supported by TrEd to PML.

By default, the tool creates PML instances for the a-layer (only if available in the source), m-layer, and w-layer. It is also possible to restrict the conversion to a certain layer only by calling the appropriate macro of the script, e.g.

$ btred -m old2pml.btred -e m-data files ...

only creates the corresponding m-layer instances.

The output m-file is by default populated only with the manual morphological annotation. Whether the tagger markup and/or morphology analysis should also be included in the m-file as alternatives to the annotator's markup, is controlled by using --with-MD and --with-MM script flags respectively (script flags are passed to btred between -o and --).

The names of output files are based on the input files as follows: the input filename is first transformed according to command-line options -a|-s|-r|-p passed to btred or ntred. If the resulting filename contains a .gz suffix, it is stripped. Finally, according to the layer of annotation, one of the suffixes .a, .m, and .w is appended.

The script converts PDT 2.0 a-data (after knitting them with the corresponding m-data and w-data) to CSTS with the following limitations on the resulting CSTS (especially with respect to the content of PML w-data and m-data):

The names of the output files are based on the input files as follows: the input filename is first transformed according to command-line options -a|-s|-r|-p passed to btred or ntred. If the resulting filename contains a .gz suffix, it is stripped. Finally, the .csts suffix is appended.

This conversion is implemented as an XSL transformation and requires an XSLT processor such as xsltproc, saxon. A very simplistic conversion of PDT 2.0 m-data to CSTS is provided by this XSLT stylesheet. It suffers from the following limitations:

This script transforms a-data or t-data from PML instances to a corresponding FS file suitable for use with the Netgraph server. In case of a-data, only minor changes to the naming of attributes are done (m-data and w-data are naturally embedded). In case of t-data the conversion is more complex: for every a-node referred to in the element a of a t-node, a phantom copy is created and planted as a hidden child-node of the referring t-node. This copy also embeds complete m-layer and w-layer information of the original a-node. Node attributes of the a-layer start with the prefix a/, those of m-layer resp. w-layer start with m/. resp. with w/.

Unlike the above conversions, this script actually modifies the file loaded in btred, which means that when applied using ntred, an explicit ntred --save-files command must be issued before ntred --quit.

The names of the output files are based on the input files as follows: possible .pls or .gz suffixes are removed, .fs suffix is appended and the result is transformed according to command-line options -a|-s|-r|-p passed to btred or ntred --save.

This script transforms given PML files to a binary format based on the Storable Perl module. This format allows applications from the TrEd toolkit to retrieve the data with extreme speed.

Apart from simple saving the data as Storable objects, the script tries to achieve internal consistency of references from t-data to a-data by changing the internal reference to a-data file in the t-data file accordingly. For example, a reference to an a-data file ref_filename in a t-data file processed with this script is changed as to refer to ref_filename.pls.gz. Hence, it is required to apply this script both on a-data and t-data files.

This script modifies the file loaded in btred, which means that when it is applied using ntred, an explicit ntred --save-files command must be issued before ntred --quit.

The filenames of the referenced files are internally changed as follows: the directory path is completely stripped off, a possible .gz suffix is removed, .pls.gz suffix is appended and the result is transformed according to command-line options -a|-s|-r|-p (in case of ntred, these must be repeated also with ntred -m pml2pls.btred command).

The names of the output files are based on the input files as follows: a possible .gz suffix is removed, .pls.gz suffix is appended and the result is transformed according to command-line options -a|-s|-r|-p (in case of ntred, these are the options used with ntred --save).

A typical usage (converting both a-data and t-data files to a gzipped Storable format *.pls.gz in one step):

$ btred -Y -m pml2pls.btred -g '*.t' '*.a'

The -Y flag suppresses possible doubled loading of the a-data as secondary files required by the t-data.

The same case using ntred but without gzipping and moving the files form source/ directory to target/:

$ ntred --init -Y -g '*.t' '*.a'
$ ntred -m pml2pls.btred  -s .gz
$ ntred --save-files  -p source/ -r target/