Releases: korpling/annatto
Releases · korpling/annatto
0.18.0 - 2024-10-15
Release Notes
Added
- Add new
table importer for CSV files.
- Allow to set the order of output columns in
table export with the parameter
column_names and to skip the column header with the skip_header param.
remove_match option in revise now allows to delete the annotation but not
the referenced node.
Fixed
- Export token in
table exporter instead of ignoring them. You can disable
exporting the token with the skip_token parameter.
- Fixed
find_connected calls with Bound::Included(usize::MAX), which can
lead to invalid results when using the linear graph storage. Replaced with the
correct Bound::Unbounded.
Download annatto 0.18.0
0.17.0 - 2024-09-27
Release Notes
Added
- Add
meta as export format
Fixed
xlsx importer did not give the correct node name to segmentation token. Due
to this inconsistency, span annotations on segmentation nodes where not
connected to the segmentation token.- unknown keys in toml configurations are now denied not only in config context, but globally in a workflow file
Download annatto 0.17.0
0.16.0 - 2024-09-02
Release Notes
Added
table export has feature to customize n/a-value, which by default is the empty string- Add
conllu as export format
- import of
conllu now supports enhanced dependencies
- Adds
saltxml export format
- Adds
time graph op to add or enrich time annotations
- The
table exporter now supports the id_column parameter to
enable/disable the ID column.
- Importers that map directories to (sub)-corpora and files to documents can now also importt the
corpus if the path argument points to a single file.
xlsx importer now maps columns as spans if the column is not configured to
be a token_column.
Changed
exmaralda import now ranks order of tlis higher than sorting by time value (more compatible with
modern EXMARaLDA files)xlsx importer will connect spans to their corresponding segmentation node
with coverage edges instead of connecting them with the base tokens generated
for the timeline items. Thus, the configured connection between spans and base
text is not lost.
Fixed
exmaralda import keeps events with missing time values
Download annatto 0.16.0
0.14.0 - 2024-07-24
Release Notes
Added
revise now offers to delete nodes that match a query using list entry [[remove_match]] with keys query and remove.
Changed
- internal changes in deserialization of annotation components and annotation keys (keys can be provided as string or in map notation), which changes the api and the way some workflow configurations are organised. Use
annatto info [module] for more details. It does not affect behaviour once older workflows are adapted to the new interface.
- when an exmaralda file contains an empty url attributed for a referenced file, this does not raise a warning anymore, as this is the way exmaralda encodes it, when no media file is used.
Fixed
textgrid export creates intervals from global xmin to global xmax for all tiers
Download annatto 0.14.0
0.13.1 - 2024-07-11
Release Notes
Fixed
textgrid export did not generate intermediate empty intervals when xmax of an interval did not match xmin of the subsequent interval, which leads to hardly editable intervals in praat. This has been fixed.
Download annatto 0.13.1
0.13.0 - 2024-07-11
Release Notes
Added
visualize graph operation that allows to output the current graph (somehwere
in the conversion process) to SVG or DOT for debugging.
Fixed
Changed
collapse uses deserializable component, thus attributes ctype, layer, and name are now under key componentcollapse only keeps annotations with namespace annis for nodes that were terminals in the collapsed components when transferring to the merged nodes to keep node status intact (e. g. token vs. not a token in terms of annis::tok).
Download annatto 0.13.0
0.12.0 - 2024-07-08
Release Notes
Added
textgrid export considers time annotations of covered nodes as well
Fixed
textgrid export can now handle annis::time intervals with an undefined right boundary (such intervals will be skipped)
Changed
collapse now also transfers annotations with namespace "annis" with the exception of "annis::node_name". This could lead to unstable results in case of conflicting values, such as for "annis::layer", but for most use cases this is not relevant yet. Not adding many of the before dropped annotations, though, was much more severe.
Download annatto 0.12.0
0.11.0 - 2024-07-05
Release Notes
Added
textgrid export now creates PRAAT TextGrid files from annotation graphstextgrid export can be configured for a desired order of tiers in the output files; the order of tiers can be incomplete, attribute ignore_others can be used to interprete the order as an allowlisttextgrid export also looks into point_tiers ignore_others is on, since it is a reasonable expection the user could have. Thus, setting ignore_others = true with an empty tier_order would result in an export of all point tiers if at least one is set.exmaralda export can now be configured for the annotation key that provides a clue to which subgraph is relevant for a file
Fixed
- code is more robust and more transparent to the user in case of unexpected data
textgrid import now allows correct file type specification for short files
Changed
revise now deserialized components directly and uses different syntax. They are provided as a list of from and to component specifications.
Download annatto 0.11.0
0.10.0 - 2024-06-25
Release Notes
Fixed
- preconfiguration of
arch_dependency via guess_vis field of graphml export now only sets node_key mapping for named orderings. Setting it with an empty value did not address annis::tok contrary to what was expected to happen.
- some bare unwraps have been removed, thus exporting graphml is now more robust.
Addded
- New
annatto document <OUTPUT_DIR> command that allows to generate markdown
files with the module documentation in a given output directory. This command
is executed in every pull request to keep the documentation up to date.
conllu format now properly imports sentence comments, i. e. sentence level annotations that are not delimited by "=". This also requires such annotations to not contain a "=" at all. Such comments will be by default imported as values of conll::comment annotations.
The annotation name can be adapted using attribute comment_anno of toml type map with keys ns and name (a serialization of graphANNIS' AnnoKey).
Download annatto 0.10.0
