Merge pull request #164 from WorksApplications/feature/kazuma-t/build_test_dict

Build test dictionaries on the fly

Author: takaoka-k-wap

.github/workflows/build.yml

@@ -33,11 +33,8 @@ jobs:
         flake8 . --count --exit-zero --max-complexity=10 --max-line-length=127 --statistics
     - name: Check license header
       run: |
-        HEADER=`cat scripts/license-header.txt`
-        for FILE in `find setup.py sudachipy tests -name '*.py'`; do FILECONTENTS=`cat "$FILE"`; if [[ "$FILECONTENTS" != "$HEADER"* ]]; then >&2 echo "invalid license header on $FILE"; fi; done
+        scripts/checkheader.sh
     - name: Test with unittest
       run: |
-        cp .travis/system.dic.test tests/resources/system.dic
-        cp .travis/user.dic.test tests/resources/user.dic
         python setup.py build_ext --inplace
-        python -m unittest discover tests
+        scripts/test.sh

.travis/system.dic.test

@@ -0,0 +1,10 @@
+#!/usr/bin/env bash
+
+HEADER=scripts/license-header.txt
+SIZE=`wc -c < "$HEADER"`
+
+RES=`find setup.py sudachipy tests -type f -name '*.py' -exec cmp -n "$SIZE" "$HEADER" {} \;`
+if [ -n "$RES" ]; then
+    echo "$RES" | awk '{print "invalid license header on " $2}' >&2
+    exit 1
+fi

.travis/user.dic.test

@@ -5,25 +5,5 @@ cd $(dirname $0)
 flake8 --show --config=flake8.cfg ../sudachipy
 flake8 --show --config=flake8.cfg ../tests
 
-HEADER=`cat license-header.txt`
-
 cd ..
-
-array=()
-
-for FILE in `find ./sudachipy -type f -name "*.py"`; do
-    array+=( ${FILE} )
-done
-
-for FILE in `find ./tests -type f -name "*.py"`; do
-    array+=( ${FILE} )
-done
-
-array+=( ./setup.py )
-
-for FILE in ${array[@]}; do
-    FILECONTENTS=`cat ${FILE}`
-    if [[ ${FILECONTENTS} != ${HEADER}* ]]; then
-        >&2 echo "invalid license header on ${FILE}"
-    fi
-done
+scripts/checkheader.sh

scripts/checkheader.sh

@@ -5,29 +5,23 @@
 # You need to prepare system.dic in resources and tests/resources
 # see README
 
-cd $(dirname $0)
+set -e
 
-# check system.dic
-if [[ ! -f "../tests/resources/system.dic" ]]; then
-    cp ../.travis/system.dic.test ../tests/resources/system.dic
-fi
-DIFF=$(diff ../.travis/system.dic.test ../tests/resources/system.dic)
-if [[ "$DIFF" != "" ]]; then
-    cp ../.travis/system.dic.test ../tests/resources/system.dic
+# build dictionaries
+if !(type sudachipy > /dev/null 2>&1); then
+    python setup.py develop
 fi
+sudachipy build -o tests/resources/system.dic -d "the system dictionary for the unit tests" -m tests/resources/dict/matrix.def tests/resources/dict/lex.csv
+sudachipy ubuild -o tests/resources/user.dic -s tests/resources/system.dic tests/resources/dict/user.csv
 
-# check user.dic
-if [[ ! -f "../tests/resources/user.dic" ]]; then
-    cp ../.travis/user.dic.test ../tests/resources/user.dic
-fi
-DIFF=$(diff ../.travis/user.dic.test ../tests/resources/user.dic)
-if [[ "$DIFF" != "" ]]; then
-    cp ../.travis/user.dic.test ../tests/resources/user.dic
-fi
+set +e
 
 # unittest
-RES=`cd ..; python -m unittest discover tests -p '*test*.py' 2>&1`
+RES=`python -m unittest discover tests -p '*test*.py' 2>&1`
+STATUS=$?
 RES_TAIL=`echo "$RES" | tail -1`
 if [[ $RES_TAIL != "OK" ]]; then
     >&2 echo "$RES"
 fi
+
+exit $STATUS

scripts/format.sh

@@ -0,0 +1,39 @@
+た,1,1,8729,た,助動詞,*,*,*,助動詞-タ,終止形-一般,タ,た,*,A,*,*,*,*
+に,2,2,11406,に,助詞,接続助詞,*,*,*,*,ニ,に,*,A,*,*,*,*
+に,3,3,4481,に,助詞,格助詞,*,*,*,*,ニ,に,*,A,*,*,*,*
+京都,6,6,5293,京都,名詞,固有名詞,地名,一般,*,*,キョウト,京都,*,A,*,*,*,1/5
+東,7,7,4675,東,名詞,普通名詞,一般,*,*,*,ヒガシ,東,*,A,*,*,*,*
+東京,6,6,2816,東京,名詞,固有名詞,地名,一般,*,*,トウキョウ,東京,*,A,*,*,*,*
+東京都,6,8,5320,東京都,名詞,固有名詞,地名,一般,*,*,トウキョウト,東京都,*,B,5/9,*,5/9,*
+行く,4,4,5105,行く,動詞,非自立可能,*,*,五段-カ行,終止形-一般,イク,行く,*,A,*,*,*,*
+行っ,5,5,5122,行っ,動詞,非自立可能,*,*,五段-カ行,連用形-促音便,イッ,行く,7,A,*,*,*,*
+都,8,8,2914,都,名詞,普通名詞,一般,*,*,*,ト,都,*,A,*,*,*,*
+アイ,7,7,4675,アイ,名詞,普通名詞,一般,*,*,*,アイ,アイ,*,A,*,*,*,*
+アイウ,7,7,4675,アイウ,名詞,普通名詞,一般,*,*,*,アイウ,アイウ,*,A,*,*,*,*
+アイアイウ,6,6,32766,アイウ,名詞,固有名詞,地名,一般,*,*,アイアイウ,アイアイウ,*,A,*,*,*,*
+0,9,9,2478,0,名詞,数詞,*,*,*,*,ゼロ,0,*,A,*,*,*,*
+1,9,9,2478,1,名詞,数詞,*,*,*,*,イチ,1,*,A,*,*,*,*
+2,9,9,2478,2,名詞,数詞,*,*,*,*,ニ,2,*,A,*,*,*,*
+3,9,9,2478,3,名詞,数詞,*,*,*,*,サン,3,*,A,*,*,*,*
+4,9,9,2478,4,名詞,数詞,*,*,*,*,ヨン,4,*,A,*,*,*,*
+5,9,9,2478,5,名詞,数詞,*,*,*,*,ゴ,5,*,A,*,*,*,*
+6,9,9,2478,6,名詞,数詞,*,*,*,*,ロク,6,*,A,*,*,*,*
+7,9,9,2478,7,名詞,数詞,*,*,*,*,ナナ,7,*,A,*,*,*,*
+8,9,9,2478,8,名詞,数詞,*,*,*,*,ハチ,8,*,A,*,*,*,*
+9,9,9,2478,9,名詞,数詞,*,*,*,*,キュウ,9,*,A,*,*,*,*
+〇,9,9,2478,〇,名詞,数詞,*,*,*,*,ゼロ,〇,*,A,*,*,*,*
+一,9,9,2478,一,名詞,数詞,*,*,*,*,イチ,一,*,A,*,*,*,*
+二,9,9,2478,二,名詞,数詞,*,*,*,*,ニ,二,*,A,*,*,*,*
+三,9,9,2478,三,名詞,数詞,*,*,*,*,サン,三,*,A,*,*,*,*
+四,9,9,2478,四,名詞,数詞,*,*,*,*,ヨン,四,*,A,*,*,*,*
+五,9,9,2478,五,名詞,数詞,*,*,*,*,ゴ,五,*,A,*,*,*,*
+六,9,9,2478,六,名詞,数詞,*,*,*,*,ロク,六,*,A,*,*,*,*
+七,9,9,2478,七,名詞,数詞,*,*,*,*,ナナ,七,*,A,*,*,*,*
+八,9,9,2478,八,名詞,数詞,*,*,*,*,ハチ,八,*,A,*,*,*,*
+九,9,9,2478,九,名詞,数詞,*,*,*,*,キュウ,九,*,A,*,*,*,*
+六三四,6,6,0,六三四,名詞,固有名詞,地名,一般,*,*,ムサシ,六三四,*,A,*,*,*,*
+いく,4,4,5105,いく,動詞,非自立可能,*,*,五段-カ行,終止形-一般,イク,行く,*,A,*,*,*,*
+いっ,5,5,5122,いっ,動詞,非自立可能,*,*,五段-カ行,連用形-促音便,イッ,行く,34,A,*,*,*,*
+012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789,9,9,2478,012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789,名詞,数詞,*,*,*,*,ゼロイチニサンヨンゴロクナナハチキュウゼロイチニサンヨンゴロクナナハチキュウゼロイチニサンヨンゴロクナナハチキュウゼロイチニサンヨンゴロクナナハチキュウゼロイチニサンヨンゴロクナナハチキュウゼロイチニサンヨンゴロクナナハチキュウゼロイチニサンヨンゴロクナナハチキュウゼロイチニサンヨンゴロクナナハチキュウゼロイチニサンヨンゴロクナナハチキュウゼロイチニサンヨンゴロクナナハチキュウゼロイチニサンヨンゴロクナナハチキュウゼロイチニサンヨンゴロクナナハチキュウゼロイチニサンヨンゴロクナナハチキュウゼロイチニサンヨンゴロクナナハチキュウゼロイチニサンヨンゴロクナナハチキュウゼロイチニサンヨンゴロクナナハチキュウゼロイチニサンヨンゴロクナナハチキュウゼロイチニサンヨンゴロクナナハチキュウゼロイチニサンヨンゴロクナナハチキュウゼロイチニサンヨンゴロクナナハチキュウゼロイチニサンヨンゴロクナナハチキュウゼロイチニサンヨンゴロクナナハチキュウゼロイチニサンヨンゴロクナナハチキュウゼロイチニサンヨンゴロクナナハチキュウゼロイチニサンヨンゴロクナナハチキュウゼロイチニサンヨンゴロクナナハチキュウゼロイチニサンヨンゴロクナナハチキュウゼロイチニサンヨンゴロクナナハチキュウゼロイチニサンヨンゴロクナナハチキュウゼロイチニサンヨンゴロクナナハチキュウ,012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789,*,A,*,*,*,*
+特a,8,8,2914,特a,名詞,普通名詞,一般,*,*,*,トクエー,特a,*,A,*,*,*,*
+な。な,8,8,2914,な。な,名詞,普通名詞,一般,*,*,*,ナナ,な。な,*,A,*,*,*,*

scripts/test.sh

@@ -0,0 +1,101 @@
+10 10
+0 0 0
+0 1 863
+0 2 2124
+0 3 1032
+0 4 591
+0 5 -162
+0 6 -79
+0 7 887
+0 8 447
+0 9 -535
+1 0 -3689
+1 1 -3361
+1 2 -7643
+1 3 -3267
+1 4 809
+1 5 -1098
+1 6 4606
+1 7 4269
+1 8 4567
+1 9 1635
+2 0 -1959
+2 1 2457
+2 2 811
+2 3 840
+2 4 903
+2 5 -958
+2 6 517
+2 7 2037
+2 8 1392
+2 9 -193
+3 0 -2288
+3 1 1741
+3 2 487
+3 3 792
+3 4 -1474
+3 5 -3429
+3 6 126
+3 7 437
+3 8 605
+3 9 -547
+4 0 -2809
+4 1 -3584
+4 2 -6743
+4 3 -2869
+4 4 -2805
+4 5 -407
+4 6 3422
+4 7 5642
+4 8 6382
+4 9 2165
+5 0 -509
+5 1 -3665
+5 2 -3882
+5 3 -572
+5 4 -1036
+5 5 -54
+5 6 2570
+5 7 3319
+5 8 4059
+5 9 882
+6 0 101
+6 1 2933
+6 2 2198
+6 3 -2004
+6 4 4392
+6 5 4017
+6 6 569
+6 7 475
+6 8 -390
+6 9 852
+7 0 -852
+7 1 2079
+7 2 1180
+7 3 -3084
+7 4 2010
+7 5 1570
+7 6 746
+7 7 2341
+7 8 2051
+7 9 1393
+8 0 -522
+8 1 3354
+8 2 2037
+8 3 -2542
+8 4 3071
+8 5 2631
+8 6 -352
+8 7 2847
+8 8 1134
+8 9 1256
+9 0 -975
+9 1 2498
+9 2 1690
+9 3 -1523
+9 4 3023
+9 5 3139
+9 6 2562
+9 7 3962
+9 8 418
+9 9 -2490

tests/resources/dict/lex.csv

@@ -0,0 +1,4 @@
+ぴらる,8,8,-32768,ぴらる,名詞,普通名詞,一般,*,*,*,ピラル,ぴらる,*,A,*,*,*,*
+府,8,8,2914,府,名詞,普通名詞,一般,*,*,*,フ,府,*,A,*,*,*,*
+東京府,6,6,2816,東京府,名詞,固有名詞,地名,一般,*,*,トウキョウフ,東京府,*,B,5/U1,*,5/U1,1/3
+すだち,6,6,2816,すだち,被子植物門,双子葉植物綱,ムクロジ目,ミカン科,ミカン属,スダチ,スダチ,すだち,*,A,*,*,*,*

tests/resources/dict/matrix.def

@@ -0,0 +1,2 @@
+ぴさる,8,8,-32768,ぴさる,名詞,普通名詞,一般,*,*,*,ピサル,ぴさる,*,A,*,*,*,*
+かぼす,6,6,2816,かぼす,被子植物門,双子葉植物綱,ムクロジ目,ミカン科,ミカン属,カボス,カボス,かぼす,*,A,*,*,*,*