commit 88d7e65fc09ecf5009d2e1294af7fde9c9691ce9 Author: MambetniyazovAmir Date: Fri Aug 21 11:00:56 2020 +0500 testvoc update diff --git a/apertium-uzb.uzb.lexc b/apertium-uzb.uzb.lexc index a02a92e..a33b332 100644 --- a/apertium-uzb.uzb.lexc +++ b/apertium-uzb.uzb.lexc @@ -24634,7 +24634,6 @@ senka:senka N1 ; ! 0.9999910702008322 senka:senka N1 ; ! 0.9999999950489525 sen%-men:sen%-men N1 ; ! 0.9976975672104723 sensatsion:sensatsion A2 ; ! 0.9999005421465287 -sen:sen IJ ; ! 0.6776063718726091 sensira:sensira V-IV ; ! 0.9999992117946327 sensirash:sensirash V-IV-COOP ; ! 0.8231863502539652 sentimental:sentimental A2 ; ! 0.9296210512073955 @@ -32475,8 +32474,7 @@ harfxoʻr:harfxoʻr N1 ; ! 0.9999987485485815 harf:harf N1 ; ! 0.9998707768876696 harf:harf N1 ; ! "letter" har:har A1 ; ! -har:har DET-DEM ; ! "each" -har:har N1 ; ! 0.5070062450926205 +har:har DET-QNT ; ! "each" har% holda:har% holda CA ; ! "" har% holda:har% holda CA ; ! "after all" har% holda:har% holda ADV ; ! "presumably" diff --git a/test-scripts/uzb.test b/test-scripts/uzb.test new file mode 100644 index 0000000..c8fd6ac --- /dev/null +++ b/test-scripts/uzb.test @@ -0,0 +1,30 @@ +#!/bin/bash + +# Expands the dictionary and compares it to the expansion of the previous +# state of the dictionary (before you made changes). +# +# Currently expanding uzb.lexc with hfst-fst2strings takes too long, so we +# concatenate txt.gz files from apertium-uzb/tests/morphotactics (expansions of +# prefixed transducers) into one file in apertium-uzb/expanded and use that +# instead. +# +# Both 'previous' and 'current' files are stored in ../expanded directory +# sorted and compressed. + +# compile the transducer and regenerate .txt.gz files in tests/morphotactics +make +cd tests/morphotactics +make +cd ../.. + +echo "" > /tmp/expanded.txt + +for file in tests/morphotactics/*.txt.gz +do + zcat $file >> /tmp/expanded.txt +done + +sort /tmp/expanded.txt | gzip -c > expanded/current-state.txt.gz + +diff -U0 <(zcat expanded/previous-state.txt.gz) <(zcat expanded/current-state.txt.gz) | +dwdiff --diff-input | less diff --git a/tests/morphotactics/.deps/any-symbol.hfst b/tests/morphotactics/.deps/any-symbol.hfst new file mode 100644 index 0000000..fdcd4b7 Binary files /dev/null and b/tests/morphotactics/.deps/any-symbol.hfst differ diff --git a/tests/morphotactics/.deps/uzb.automorf.hfst b/tests/morphotactics/.deps/uzb.automorf.hfst new file mode 100644 index 0000000..b19b9fa Binary files /dev/null and b/tests/morphotactics/.deps/uzb.automorf.hfst differ