Hugging Face's logo

javohirmat
/
uzmorph

Uzbek
uzmorph
morphology
apertium
morphological-analysis
uzbek
Model card Files
xet
 Community
uzmorph / data /SOURCE.md
javohirmat's picture
javohirmat
uzmorph 0.9.0
1a37f17 verified
preview code
|
Raw History Blame Contribute Delete
2.28 kB

Corresponding Source for the transducers in this directory (GPL-3.0, section 6)

uzb.automorf.bin (analyser), uzb.autogen.bin (generator) and uzb_guesser.automorf.bin (guesser) are object code. They were compiled from a modified version of apertium-uzb, which is licensed under the GNU General Public License version 3 only (GPL-3.0-only; full text in LICENSE / apertium-uzb/COPYING).

Where the Corresponding Source is

  1. In the source distribution. uzmorph-0.9.0.tar.gz (the sdist, published next to every wheel) contains the complete source under apertium-uzb/. That covers the lexc, twol, rlx, guesser and coopfilter.regexp files, the Makefile.am / configure.ac / autogen.sh build scripts, upstream COPYING and AUTHORS, and CHANGES.md, which lists every modification.
  2. In the repository. The apertium-uzb/ directory of the uzmorph repository, at the tag v0.9.0: https://github.com/javohirmat/uzmorph/tree/v0.9.0/apertium-uzb
  3. Upstream base. https://github.com/apertium/apertium-uzb, commit 00522e3 (2025-05-16). Our changes are applied on top of it.

How to rebuild these exact files 

sudo apt install lttoolbox lttoolbox-dev hfst cg3 cg3-dev apertium apertium-dev autoconf automake pkg-config make gawk
cd apertium-uzb && ./autogen.sh && make
# uzb.automorf.bin, uzb.autogen.bin, uzb_guesser.automorf.bin are byte-identical to the files here
# (tested with apertium 3.9, lttoolbox 3.7, hfst 3.16 on Ubuntu)

Written offer

For at least three years after the last distribution of this version, and for as long as spare parts or customer support for it are offered, the maintainer will give anyone who possesses the object code a copy of the complete Corresponding Source. The copy comes on a durable physical medium customarily used for software interchange, or by download, for no more than the reasonable cost of physically performing the transfer. Contact: Javohir Matniyazov, javohirmatniyazovv@gmail.com.

The other files here

The *.json.gz, name_sufprior.json and names_list.tsv files are model data, not compiled from source code. They are derived from corpus statistics by the scripts in tools/models/ (see tools/README.md). The final step is tools/models/export_models.py.