srx-languagetool 0.4.0 → 0.7.0
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/.github/workflows/main.yml +2 -2
- data/.ruby-version +1 -1
- data/CHANGELOG.md +12 -0
- data/Gemfile.lock +31 -26
- data/lib/srx/languagetool/version.rb +1 -1
- data/lib/srx/segment.srx +66 -20
- data/srx-languagetool.gemspec +1 -0
- metadata +3 -2
checksums.yaml
CHANGED
@@ -1,7 +1,7 @@
|
|
1
1
|
---
|
2
2
|
SHA256:
|
3
|
-
metadata.gz:
|
4
|
-
data.tar.gz:
|
3
|
+
metadata.gz: c277febbbdec765f5eeba65cdc43dd661ddf807a53f2e2858f0e321eccdf3459
|
4
|
+
data.tar.gz: 34aa30e138bddfacd30ed8f9a48f45d4e91afa5c4e4a789d3208860c1bbd608e
|
5
5
|
SHA512:
|
6
|
-
metadata.gz:
|
7
|
-
data.tar.gz:
|
6
|
+
metadata.gz: 62a56ee4d2cb0ba5fe16181681561a829f7f59f84155753d896fa6dc6153bd238ed9de8337cea2e75ac6c83e94381d84a0ede80c2bbcff1648fe97c57f3d5fed
|
7
|
+
data.tar.gz: 66d09127f9eb76307bd38938c619a6099c0ff941bcd3c33b51e4c130c0d4373ea12cdb080bec1aab783e4070c0d5d70fa2e97a4d64d1c394390316630012e0b3
|
data/.github/workflows/main.yml
CHANGED
@@ -10,10 +10,10 @@ jobs:
|
|
10
10
|
- name: Set up Ruby
|
11
11
|
uses: ruby/setup-ruby@v1
|
12
12
|
with:
|
13
|
-
ruby-version: 2.7.
|
13
|
+
ruby-version: 2.7.6
|
14
14
|
- name: Install
|
15
15
|
run: |
|
16
|
-
gem install bundler -v 2.
|
16
|
+
gem install bundler -v 2.3.17
|
17
17
|
bundle install
|
18
18
|
- name: Type check
|
19
19
|
run: bundle exec solargraph typecheck --level typed
|
data/.ruby-version
CHANGED
@@ -1 +1 @@
|
|
1
|
-
2.7.
|
1
|
+
2.7.6
|
data/CHANGELOG.md
CHANGED
@@ -1,5 +1,17 @@
|
|
1
1
|
## [Unreleased]
|
2
2
|
|
3
|
+
## [0.7.0] - 2022-07-04
|
4
|
+
|
5
|
+
- Update rules to LanguageTool 5.8
|
6
|
+
|
7
|
+
## [0.6.0] - 2022-03-30
|
8
|
+
|
9
|
+
- Update rules to LanguageTool 5.7
|
10
|
+
|
11
|
+
## [0.5.0] - 2021-12-31
|
12
|
+
|
13
|
+
- Update rules to LanguageTool 5.6
|
14
|
+
|
3
15
|
## [0.4.0] - 2021-10-03
|
4
16
|
|
5
17
|
- Update rules to LanguageTool 5.5
|
data/Gemfile.lock
CHANGED
@@ -1,7 +1,7 @@
|
|
1
1
|
PATH
|
2
2
|
remote: .
|
3
3
|
specs:
|
4
|
-
srx-languagetool (0.
|
4
|
+
srx-languagetool (0.7.0)
|
5
5
|
srx (< 1.0)
|
6
6
|
|
7
7
|
GEM
|
@@ -9,45 +9,47 @@ GEM
|
|
9
9
|
specs:
|
10
10
|
ast (2.4.2)
|
11
11
|
backport (1.2.0)
|
12
|
-
benchmark (0.
|
12
|
+
benchmark (0.2.0)
|
13
13
|
byebug (11.1.3)
|
14
|
-
diff-lcs (1.
|
14
|
+
diff-lcs (1.5.0)
|
15
15
|
e2mmap (0.1.0)
|
16
16
|
jaro_winkler (1.5.4)
|
17
|
-
|
17
|
+
json (2.6.2)
|
18
|
+
kramdown (2.4.0)
|
18
19
|
rexml
|
19
20
|
kramdown-parser-gfm (1.1.0)
|
20
21
|
kramdown (~> 2.0)
|
21
|
-
minitest (5.
|
22
|
-
nokogiri (1.
|
22
|
+
minitest (5.16.2)
|
23
|
+
nokogiri (1.13.6-x86_64-darwin)
|
23
24
|
racc (~> 1.4)
|
24
|
-
parallel (1.
|
25
|
-
parser (3.
|
25
|
+
parallel (1.22.1)
|
26
|
+
parser (3.1.2.0)
|
26
27
|
ast (~> 2.4.1)
|
27
|
-
racc (1.
|
28
|
-
rainbow (3.
|
28
|
+
racc (1.6.0)
|
29
|
+
rainbow (3.1.1)
|
29
30
|
rake (13.0.6)
|
30
|
-
regexp_parser (2.
|
31
|
-
reverse_markdown (2.
|
31
|
+
regexp_parser (2.5.0)
|
32
|
+
reverse_markdown (2.1.1)
|
32
33
|
nokogiri
|
33
34
|
rexml (3.2.5)
|
34
|
-
rspec-expectations (3.
|
35
|
+
rspec-expectations (3.11.0)
|
35
36
|
diff-lcs (>= 1.2.0, < 2.0)
|
36
|
-
rspec-support (~> 3.
|
37
|
-
rspec-support (3.
|
38
|
-
rubocop (1.
|
37
|
+
rspec-support (~> 3.11.0)
|
38
|
+
rspec-support (3.11.0)
|
39
|
+
rubocop (1.31.1)
|
40
|
+
json (~> 2.3)
|
39
41
|
parallel (~> 1.10)
|
40
|
-
parser (>= 3.
|
42
|
+
parser (>= 3.1.0.0)
|
41
43
|
rainbow (>= 2.2.2, < 4.0)
|
42
44
|
regexp_parser (>= 1.8, < 3.0)
|
43
|
-
rexml
|
44
|
-
rubocop-ast (>= 1.
|
45
|
+
rexml (>= 3.2.5, < 4.0)
|
46
|
+
rubocop-ast (>= 1.18.0, < 2.0)
|
45
47
|
ruby-progressbar (~> 1.7)
|
46
48
|
unicode-display_width (>= 1.4.0, < 3.0)
|
47
|
-
rubocop-ast (1.
|
48
|
-
parser (>= 3.
|
49
|
+
rubocop-ast (1.18.0)
|
50
|
+
parser (>= 3.1.1.0)
|
49
51
|
ruby-progressbar (1.11.0)
|
50
|
-
solargraph (0.
|
52
|
+
solargraph (0.45.0)
|
51
53
|
backport (~> 1.2)
|
52
54
|
benchmark
|
53
55
|
bundler (>= 1.17.2)
|
@@ -64,13 +66,16 @@ GEM
|
|
64
66
|
yard (~> 0.9, >= 0.9.24)
|
65
67
|
srx (0.6.0)
|
66
68
|
nokogiri (~> 1.11)
|
67
|
-
thor (1.1
|
69
|
+
thor (1.2.1)
|
68
70
|
tilt (2.0.10)
|
69
|
-
unicode-display_width (2.
|
70
|
-
|
71
|
+
unicode-display_width (2.2.0)
|
72
|
+
webrick (1.7.0)
|
73
|
+
yard (0.9.28)
|
74
|
+
webrick (~> 1.7.0)
|
71
75
|
|
72
76
|
PLATFORMS
|
73
77
|
x86_64-darwin-20
|
78
|
+
x86_64-darwin-21
|
74
79
|
|
75
80
|
DEPENDENCIES
|
76
81
|
byebug
|
@@ -82,4 +87,4 @@ DEPENDENCIES
|
|
82
87
|
srx-languagetool!
|
83
88
|
|
84
89
|
BUNDLED WITH
|
85
|
-
2.
|
90
|
+
2.3.17
|
data/lib/srx/segment.srx
CHANGED
@@ -1134,6 +1134,14 @@
|
|
1134
1134
|
<beforebreak>\bU\.</beforebreak>
|
1135
1135
|
<afterbreak>[SK]\b</afterbreak>
|
1136
1136
|
</rule>
|
1137
|
+
<rule break="no"><!-- I.S (no dot at end) -->
|
1138
|
+
<beforebreak>\bI\.</beforebreak>
|
1139
|
+
<afterbreak>S\b</afterbreak>
|
1140
|
+
</rule>
|
1141
|
+
<rule break="no"><!-- M.Z (no dot at end) -->
|
1142
|
+
<beforebreak>\bM\.</beforebreak>
|
1143
|
+
<afterbreak>Z\b</afterbreak>
|
1144
|
+
</rule>
|
1137
1145
|
<rule break="no"><!-- URLs without "www."-->
|
1138
1146
|
<beforebreak>\b(https?|ftp|file|chrome|chromium|android|(chrome|moz)\-extension):///?[A-Za-z0-9\-]+\.</beforebreak>
|
1139
1147
|
<afterbreak>[A-Za-z0-9\-]+(\.|\b)</afterbreak>
|
@@ -1151,7 +1159,7 @@
|
|
1151
1159
|
<afterbreak>D\.?</afterbreak>
|
1152
1160
|
</rule>
|
1153
1161
|
<rule break="no"><!-- min. -->
|
1154
|
-
<beforebreak>\b([Ee]d|pp|[Vv]iz|i\.?[\s\u00A0]*e|[Vvol]|[Rr]col|maj|Lt|[Ff]ig|[Ff]igs|[Vv]iz|[Vv]ols|[Aa]pprox|[Ii]ncl?|[Aa]cc|Pres|[Cc]orp|[Ee]x|[Cc]onn|[Dd]ept|[Mm]in|
|
1162
|
+
<beforebreak>\b([Aa]vg|[Ee]d|pp|[Vv]iz|i\.?[\s\u00A0]*e|[Vvol]|[Rr]col|maj|Lt|[Ff]ig|[Ff]igs|[Vv]iz|[Vv]ols|[Aa]pprox|[Ii]ncl?|[Aa]cc|Pres|[Cc]orp|[Ee]x|[Cc]onn|[Dd]ept|[Ll]tda|[Mm]in|[Mm]ax|[Gg]ovt|[Rr]etd|lb|lbf|ft|c\.?[\s\u00A0]*f|vs|dia|lbs|\d+-(:?oz|kc|in|h[rp]|ml)|M?sec)\.[\s\u00A0]</beforebreak>
|
1155
1163
|
<afterbreak>[^\p{Lu}]|I</afterbreak>
|
1156
1164
|
</rule>
|
1157
1165
|
<rule break="no"><!-- hr. -->
|
@@ -1179,7 +1187,11 @@
|
|
1179
1187
|
<afterbreak></afterbreak>
|
1180
1188
|
</rule>
|
1181
1189
|
<rule break="no"><!-- vs. -->
|
1182
|
-
<beforebreak>\
|
1190
|
+
<beforebreak>\b[Vv]s\.[\s\u00A0]</beforebreak>
|
1191
|
+
<afterbreak></afterbreak>
|
1192
|
+
</rule>
|
1193
|
+
<rule break="no"><!-- pp. -->
|
1194
|
+
<beforebreak>\b(pp|PP)\.[\s\u00A0]</beforebreak>
|
1183
1195
|
<afterbreak></afterbreak>
|
1184
1196
|
</rule>
|
1185
1197
|
<rule break="no"><!-- esp. -->
|
@@ -1241,7 +1253,7 @@
|
|
1241
1253
|
</rule>
|
1242
1254
|
<rule break="no"><!-- https://de.wikipedia.org/wiki/VW_ID.3 -->
|
1243
1255
|
<beforebreak>ID.</beforebreak>
|
1244
|
-
<afterbreak>3|Buzz|Crozz</afterbreak>
|
1256
|
+
<afterbreak>3|4|Buzz|Crozz</afterbreak>
|
1245
1257
|
</rule>
|
1246
1258
|
<rule break="no"><!-- Ph.D. (see rule PH_D) -->
|
1247
1259
|
<beforebreak>\bP[Hh]\.?[\s\u00A0]?[Dd]\.[\s\u00A0]</beforebreak>
|
@@ -1520,6 +1532,11 @@
|
|
1520
1532
|
</languagerule>
|
1521
1533
|
<languagerule languagerulename="Dutch">
|
1522
1534
|
<rule break="no">
|
1535
|
+
<!-- sp.a -->
|
1536
|
+
<beforebreak>\b(sp|SP)</beforebreak>
|
1537
|
+
<afterbreak>\.[aA]\b</afterbreak>
|
1538
|
+
</rule>
|
1539
|
+
<rule break="no">
|
1523
1540
|
<!-- .Net -->
|
1524
1541
|
<beforebreak>\s[.]</beforebreak>
|
1525
1542
|
<afterbreak>[Nn][Ee][Tt](\b|-)</afterbreak>
|
@@ -1537,7 +1554,7 @@
|
|
1537
1554
|
<afterbreak>[A-Za-z0-9\-]+\.(com|net|org|info|de|es|edu|co|eu|nl|io|cn|uk|gov|biz|ca|tk|ru|br|jp|pl)(\.|\b)</afterbreak>
|
1538
1555
|
</rule>
|
1539
1556
|
<rule break="no">
|
1540
|
-
<beforebreak>\b(Drs|Art|Afr|Am|Ar|Br|Cie|Comp|Dhr|([Pp]rof\.)?[Dd]r|Em|Fa|Kon|Bros)\.\s</beforebreak>
|
1557
|
+
<beforebreak>\b(Drs|Art|Afr|Am|Ar|Br|Cie|Comp|Dhr|([Pp]rof\.)?[Dd]r|Em|Fa|Kon|Bros|Stb)\.\s</beforebreak>
|
1541
1558
|
<afterbreak></afterbreak>
|
1542
1559
|
</rule>
|
1543
1560
|
<rule break="no">
|
@@ -1557,6 +1574,10 @@
|
|
1557
1574
|
<afterbreak></afterbreak>
|
1558
1575
|
</rule>
|
1559
1576
|
<rule break="no">
|
1577
|
+
<beforebreak>\b(alc|bro|opm|acc)\.\s</beforebreak>
|
1578
|
+
<afterbreak></afterbreak>
|
1579
|
+
</rule>
|
1580
|
+
<rule break="no">
|
1560
1581
|
<beforebreak>\b(arch|archeol|art|bc|betr|bez|bibl|bijl|bijv)\.\s</beforebreak>
|
1561
1582
|
<afterbreak></afterbreak>
|
1562
1583
|
</rule>
|
@@ -4629,7 +4650,7 @@
|
|
4629
4650
|
<afterbreak>[XIV\d]+\b</afterbreak>
|
4630
4651
|
</rule>
|
4631
4652
|
<rule break="no">
|
4632
|
-
<beforebreak>\b([Ee]ds?|[Cc]oords?|\d+(r|n|t|è|é|ns|es)|masc|fem|sing|pl|adj|adv|g|kg|m|km|cm|ha|u|h|hrs|s|ss|alt|cant|cast|cert|com|dir|gr|nom|parc|pres|set|Sr|Jr|Admón|Adm|Inc|Co|Hnos|Vda|[VU]d[s]?)\.[\p{Pe}\p{Pf}\p{Pd}"']*[\s\u00A0]</beforebreak>
|
4653
|
+
<beforebreak>\b([Ee]ds?|[Cc]oords?|\d+(r|n|t|è|é|a|rs|ns|es)|masc|fem|sing|pl|adj|adv|g|kg|m|km|cm|ha|u|h|hrs|s|ss|alt|cant|cast|cert|com|dir|gr|nom|parc|pres|set|Sr|Jr|Admón|Adm|Inc|Co|Hnos|Vda|[VU]d[s]?)\.[\p{Pe}\p{Pf}\p{Pd}"']*[\s\u00A0]</beforebreak>
|
4633
4654
|
<afterbreak>[\-¡¿«»"'\u2018\u201C\p{Ps}\u2012\u2013\u2014\u2015\u2053]*\p{Ll}</afterbreak>
|
4634
4655
|
</rule>
|
4635
4656
|
<!-- Any word in acronyms like U.S.A.F or F. B. I. or C. or c.s.p. or p. e. -->
|
@@ -4700,6 +4721,10 @@
|
|
4700
4721
|
<beforebreak>Yahoo![\s\u00A0]</beforebreak>
|
4701
4722
|
<afterbreak>\p{Ll}</afterbreak>
|
4702
4723
|
</rule>
|
4724
|
+
<rule break="no">
|
4725
|
+
<beforebreak>40dB.[\s\u00A0]</beforebreak>
|
4726
|
+
<afterbreak>\p{Ll}</afterbreak>
|
4727
|
+
</rule>
|
4703
4728
|
<rule break="yes">
|
4704
4729
|
<beforebreak>\.\[\d+\][\s\u00A0]</beforebreak>
|
4705
4730
|
<afterbreak></afterbreak>
|
@@ -4748,6 +4773,10 @@
|
|
4748
4773
|
</rule>
|
4749
4774
|
<!-- Abbreviations that cannot finish sentences-->
|
4750
4775
|
<rule break="no">
|
4776
|
+
<beforebreak>\b((?iu)(en|febr|mzo|abr|my|jun|jul|ag|agt|set|sept|setbre|oct|nov|novbre|dic|dicbre))\.[\s\u00A0]</beforebreak>
|
4777
|
+
<afterbreak/>
|
4778
|
+
</rule>
|
4779
|
+
<rule break="no">
|
4751
4780
|
<beforebreak>\b(dc|(?iu)(n|[Aa]yto|Mr|C|Dr|Dra|E|Emm|Emma|Excm|Excma|Hble|I|Il·lm|Il·lma|Il·ltre|Im|Ima|Mgfc|Mgfca|Mn|R|Rev|Sr|Sra|Sres|Sras|Srs|St|Sta|a|abr|abs|acad|add|adj|adm|admdor|admdora|admtiu|admtiva|adv|ag|agl|agr|agron|agròn|aj|ajud|al|alim|amb|ampl|ant|ap|apmt|apnt|apr|aprox|apt|arm|arq|arqueol|arquit|assign|assoc|atm|aut|aux|av|b|batx|bda|bibl|bl|bnc|butll|bxs|c|calef|cartogr|cat|catedr|catol|cf|cia|cin|cint|circul|cit|climat|col|col·l|compt|cons|constr|cont|contr|conv|corp|corr|cpl|cpt|cró|ct|cte|ctra|cts|d|dept|derog|des|desp|dg|dip|disp|distr|div|dj|dl|doc|drec|ds|dt|dta|dte|dupl|dv|e|econ|ed|ef|entl|esc|esp|espf|esq|ex|exc|exp|exped|ext|f|fac|fca|febr|fig|figs|fra|gen|gov|gral|i|imp|impr|impt|inc|insp|inst|int|inv|j|jul|jur|jurispr|leg|llic|loc|ltda|làm|merc|mil·l|màx|mín|neg|nov|nre|núm|o|oct|op|p|pàg|pàgs|paq|par|pda|pg|pl|pobl|pol|ppda|ppt|pral|prev|prof|progr|prov|pta|ptes|ptge|pvt|pàg|quadr|quint|r|rbla|ref|reg|rev|secr|serv|sgt|sotsp|subsp|supl|supt|t|tel|telegr|tit|trad|trans|transcr|transf|trav|tripl|trv|tt|tèc|univ|urb|v|var|veg|venc|vid|vig|vocab|vs|x|àt|íd))\.[\s\u00A0]</beforebreak>
|
4752
4781
|
<afterbreak/>
|
4753
4782
|
</rule>
|
@@ -4861,11 +4890,15 @@
|
|
4861
4890
|
</rule>
|
4862
4891
|
<rule break="no"><!-- https://de.wikipedia.org/wiki/VW_ID.3 -->
|
4863
4892
|
<beforebreak>ID.</beforebreak>
|
4864
|
-
<afterbreak>3|Buzz|Crozz</afterbreak>
|
4893
|
+
<afterbreak>3|4|Buzz|Crozz</afterbreak>
|
4865
4894
|
</rule>
|
4866
4895
|
<rule break="no">
|
4867
4896
|
<beforebreak>[1-3]\.[\u00A0\s]</beforebreak>
|
4868
|
-
<afterbreak>Liga|Bundesliga|Fußball(-B|b)undesliga</afterbreak>
|
4897
|
+
<afterbreak>Liga|Bundesliga|(Fußball|Handball|Basketball)(-B|b)undesliga</afterbreak>
|
4898
|
+
</rule>
|
4899
|
+
<rule break="no">
|
4900
|
+
<beforebreak>\d+\.[\u00A0\s]</beforebreak>
|
4901
|
+
<afterbreak>Klässler[sn]?</afterbreak>
|
4869
4902
|
</rule>
|
4870
4903
|
<rule break="no">
|
4871
4904
|
<beforebreak>\bP[Hh]\.</beforebreak>
|
@@ -4889,13 +4922,13 @@
|
|
4889
4922
|
</rule>
|
4890
4923
|
<!-- Don't split [.?!] when they're quoted -->
|
4891
4924
|
<rule break="no">
|
4892
|
-
<beforebreak>['"„][\.!?…]['"
|
4925
|
+
<beforebreak>['"„][\.!?…]['"“«»][\u00A0\s]</beforebreak>
|
4893
4926
|
<afterbreak></afterbreak>
|
4894
4927
|
</rule>
|
4895
4928
|
<!-- Don't break after quote unless there's a capital letter
|
4896
4929
|
e.g.: "That's right!" he said. -->
|
4897
4930
|
<rule break="no">
|
4898
|
-
<beforebreak>["'
|
4931
|
+
<beforebreak>["'“«»][\u00A0\s]</beforebreak>
|
4899
4932
|
<afterbreak>\p{Ll}</afterbreak>
|
4900
4933
|
</rule>
|
4901
4934
|
<!-- e.g. "Das ist . so." - assume one sentence. -->
|
@@ -4955,7 +4988,7 @@
|
|
4955
4988
|
</rule>
|
4956
4989
|
<!-- English abbreviations - but these work globally for all languages -->
|
4957
4990
|
<rule break="no">
|
4958
|
-
<beforebreak>\b(Mrs?|No|pp|St|no|Sr|Jr|Bros|etc|[Bb]tw|vs|esp|[Ff]ig|Jan|Feb|Mar|Apr|Ju[nl]|Aug|Sept?|O[ck]t|Nov|Dec|PhD|BSc|BEng|BComp|BArch|al|cf|Inc|Ms|MEng|MSc|MComp|Gen|Sen|Prof|Corp|Co|co|Ltd)\.[\u00A0\s]</beforebreak>
|
4991
|
+
<beforebreak>\b(Mrs?|No|pp|St|no|Sr|Jr|[Ss]ek|Bros|etc|[Bb]tw|vs|esp|[Ff]ig|Jan|Feb|Mar|Apr|Ju[nl]|Aug|Sept?|O[ck]t|Nov|Dec|PhD|BSc|BEng|BComp|BArch|al|cf|Inc|Ms|MEng|MSc|MComp|Gen|Sen|Prof|Corp|Co|co|Ltd|Buchst)\.[\u00A0\s]</beforebreak>
|
4959
4992
|
<afterbreak></afterbreak>
|
4960
4993
|
</rule>
|
4961
4994
|
<!-- Latin abbreviations - but these work globally for all languages -->
|
@@ -4965,11 +4998,11 @@
|
|
4965
4998
|
</rule>
|
4966
4999
|
<!-- German abbreviations -->
|
4967
5000
|
<rule break="no">
|
4968
|
-
<beforebreak>\b(ggü|Mag|mtl|versch|d|Übers|usw|Bzw|bzw|Ab[hkst]|abzgl|bezgl|Abzw|ahd|Akk|aktual|allg|alltagsspr|altdt|alttest|amerikan|Anh|Ank|Anm|Art|autom|Auftragsnr|Az|Bat|bayr|Bde?|bearb|Bed|Bem|bes|bez|Bez|Bhf|bspw|btto|bw)\.[\u00A0\s]</beforebreak>
|
5001
|
+
<beforebreak>\b(betr|Geb|Stk|ggü|Mag|mtl|versch|d|Übers|usw|Bzw|bzw|Ab[hkst]|abzgl|[Ll]tda|[Ee]inschl|[Vv]mtl|Ev|bezgl|Abzw|[Vv]sl|ahd|Akk|aktual|allg|alltagsspr|altdt|alttest|amerikan|Anh|Ank|Anm|Art|autom|Auftragsnr|Az|Bat|bayr|Bde?|bearb|Bed|Bem|bes|bez|Bez|Bhf|bspw|btto|bw|Dtl|Dez)\.[\u00A0\s]</beforebreak>
|
4969
5002
|
<afterbreak></afterbreak>
|
4970
5003
|
</rule>
|
4971
5004
|
<rule break="no">
|
4972
|
-
<beforebreak>\b(cts?|Ca|ca|chem|chin|Chr|cresc|dat|Dat|desgl|ders|dgl|Dipl|Dir?|Doz?|durchg|durchges|Dr|dt|ebd|Ed|eigtl|Eigtl|eigl|Eigl|akt|Engl|engl|Erg|al|et[cw]|Etw|ev(tl)?|Evtl|exkl|Expl|Exz)\.[\u00A0\s]</beforebreak>
|
5005
|
+
<beforebreak>\b(cts?|Ca|ca|chem|chin|Chr|cresc|dat|Dat|desgl|ders|dgl|Dipl|Dir?|Doz?|durchg|durchges|Dr|dt|ebd|Ed|eigtl|Eigtl|eigl|Eigl|akt|Engl|engl|Erg|al|et[cw]|Etw|ev(tl)?|Evtl|Evt|evt|exkl|Expl|Exz)\.[\u00A0\s]</beforebreak>
|
4973
5006
|
<afterbreak></afterbreak>
|
4974
5007
|
</rule>
|
4975
5008
|
<rule break="no">
|
@@ -4977,7 +5010,11 @@
|
|
4977
5010
|
<afterbreak></afterbreak>
|
4978
5011
|
</rule>
|
4979
5012
|
<rule break="no">
|
4980
|
-
<beforebreak>\b
|
5013
|
+
<beforebreak>\b[BM]\.[\u00A0\s]Sc\.[\u00A0\s]</beforebreak>
|
5014
|
+
<afterbreak>\p{Ll}</afterbreak>
|
5015
|
+
</rule>
|
5016
|
+
<rule break="no">
|
5017
|
+
<beforebreak>\b(ff|Fa|fachspr|fam|fem|Fem|Fr|franz|frz?|[Aa]ltfranz|frdl|Frl|Fut|Gd|gebr?|Gebr|geh|geleg|gen|Gen|germ|gesch|ges|get|ggf|Ggf|Ggs|ggT|Gr|[Gg]rds|griech)\.[\u00A0\s]</beforebreak>
|
4981
5018
|
<afterbreak></afterbreak>
|
4982
5019
|
</rule>
|
4983
5020
|
<rule break="no">
|
@@ -4985,7 +5022,7 @@
|
|
4985
5022
|
<afterbreak></afterbreak>
|
4986
5023
|
</rule>
|
4987
5024
|
<rule break="no">
|
4988
|
-
<beforebreak>\b(lat|lfd|Lit|lt|Lz|Mask|mask|max|Mrd|mdal|me[dt]|phil|mhd|Mio?|mind?|Mo|mod|nachm|nördlBr|neutr|Nhd|Nom|Nrn?|Num|Obj|od|dgl|offz)\.[\u00A0\s]</beforebreak>
|
5025
|
+
<beforebreak>\b([A-ZÖÄÜ][a-zöäüß]+nr|tel|gem|Pat|prov|Betr|lat|lfd|Lit|lt|Lz|Mask|mask|max|Mrd|mdal|me[dt]|phil|mhd|Mio?|mind?|Mo|mod|nachm|nördlBr|neutr|Nhd|Nom|Nrn?|Num|Obj|od|dgl|offz)\.[\u00A0\s]</beforebreak>
|
4989
5026
|
<afterbreak></afterbreak>
|
4990
5027
|
</rule>
|
4991
5028
|
<rule break="no">
|
@@ -4993,7 +5030,7 @@
|
|
4993
5030
|
<afterbreak></afterbreak>
|
4994
5031
|
</rule>
|
4995
5032
|
<rule break="no">
|
4996
|
-
<beforebreak>\b(Tel|teilw|Temp|trans|Tsd|übertr|übl|ff|überarb|ugs|univ|unveränd|urspr|USt|UST|USt\-IdNr|sw|vgl|vll|Vll|vlt|Vlt|vllt|Vllt|Vgl|Vol|vollst|vorm|Vp|Vs|vs|wesentl|wg|Whg|Hd|Ztr|zus|Zus|zzt?|zzgl|zB|zb|Zz|Zt|zw|Min|Bzgl|bzgl|bezügl|Frhr|ggfs|insb|autom|Mw[sS]t)\.[\u00A0\s]</beforebreak>
|
5033
|
+
<beforebreak>\b(Tel|teilw|Temp|trans|Tsd|übertr|übl|ff|überarb|ugs|univ|unveränd|urspr|USt|UST|USt\-IdNr|[Aa][bn]schl|sw|kl|[Gg]r|vgl|vll|Vll|vlt|Vlt|vllt|Vllt|Vgl|Vol|vollst|vorm|Vp|Vs|vs|wesentl|wg|Whg|Hd|Ztr|zus|Zus|zzt?|zzgl|zB|zb|Zz|Zt|zw|Min|Bzgl|bzgl|bezügl|Frhr|ggfs|insb|autom|Mw[sS]t)\.[\u00A0\s]</beforebreak>
|
4997
5034
|
<afterbreak></afterbreak>
|
4998
5035
|
</rule>
|
4999
5036
|
<!-- Break rules -->
|
@@ -5233,6 +5270,11 @@
|
|
5233
5270
|
<beforebreak>Yahoo![\s\u00A0]</beforebreak>
|
5234
5271
|
<afterbreak>\p{Ll}</afterbreak>
|
5235
5272
|
</rule>
|
5273
|
+
<!-- !? + lowercase -->
|
5274
|
+
<rule break="no">
|
5275
|
+
<beforebreak>(\!|\?)[\s\u00A0]</beforebreak>
|
5276
|
+
<afterbreak>\p{Ll}</afterbreak>
|
5277
|
+
</rule>
|
5236
5278
|
<rule break="yes">
|
5237
5279
|
<beforebreak>\.\[\d+\][\s\u00A0]</beforebreak>
|
5238
5280
|
<afterbreak></afterbreak>
|
@@ -5386,7 +5428,7 @@
|
|
5386
5428
|
</rule>
|
5387
5429
|
<rule break="no"><!-- https://de.wikipedia.org/wiki/VW_ID.3 -->
|
5388
5430
|
<beforebreak>ID.</beforebreak>
|
5389
|
-
<afterbreak>3|Buzz|Crozz</afterbreak>
|
5431
|
+
<afterbreak>3|4|Buzz|Crozz</afterbreak>
|
5390
5432
|
</rule>
|
5391
5433
|
<rule break="no"><!-- Ph.D. (see rule PH_D) -->
|
5392
5434
|
<beforebreak>\bP[Hh]\.?[\s\u00A0]?[Dd]\.[\s\u00A0]</beforebreak>
|
@@ -5444,6 +5486,10 @@
|
|
5444
5486
|
<beforebreak>\bCo\.[\s\u00A0]</beforebreak>
|
5445
5487
|
<afterbreak></afterbreak>
|
5446
5488
|
</rule>
|
5489
|
+
<rule break="no">
|
5490
|
+
<beforebreak>\bE\.[\s\u00A0]</beforebreak>
|
5491
|
+
<afterbreak>\b[Cc]oli\b</afterbreak>
|
5492
|
+
</rule>
|
5447
5493
|
<!-- Break rules -->
|
5448
5494
|
<rule break="yes">
|
5449
5495
|
<beforebreak>[\.!?…][\u0002|'|"|«|\)|\]|\}¹²³]?[\s\u00A0]+</beforebreak>
|
@@ -5516,7 +5562,7 @@
|
|
5516
5562
|
<!-- І. Коваль -->
|
5517
5563
|
<rule break="no">
|
5518
5564
|
<beforebreak>[\h\v.]([А-ЯІЇЄҐACEIHOPX]\.-)?(?<!°)[А-ЯІЇЄҐABCEIHOPX](?<!(Куан[\h]+Ю|(Петр|Олександр)([аоу]|ові|ом)?[\h]+[IІ]+))\.[\h\v]*</beforebreak>
|
5519
|
-
<afterbreak>
|
5565
|
+
<afterbreak>[А-ЯІЇЄҐ][а-яіїєґА-ЯІЇЄҐ'’ʼ]{3}</afterbreak>
|
5520
5566
|
</rule>
|
5521
5567
|
<!-- Ів. Франко (але Ів Бутільє) -->
|
5522
5568
|
<rule break="no">
|
@@ -5570,7 +5616,7 @@
|
|
5570
5616
|
</rule>
|
5571
5617
|
<rule break="no">
|
5572
5618
|
<!-- unfortunately \b ignores \u0301 -->
|
5573
|
-
<beforebreak>\b
|
5619
|
+
<beforebreak>\b[сС]т\.[\h\v]</beforebreak>
|
5574
5620
|
<afterbreak>[\h]*(?!([АВУОІЄ]|На|Але|Так?)[\h\v])</afterbreak>
|
5575
5621
|
</rule>
|
5576
5622
|
<rule break="no">
|
@@ -5578,9 +5624,9 @@
|
|
5578
5624
|
<beforebreak>\bдол\.[\h\v]*</beforebreak>
|
5579
5625
|
<afterbreak>США</afterbreak>
|
5580
5626
|
</rule>
|
5581
|
-
<!-- п. 10 від 11.10.1933 -->
|
5627
|
+
<!-- п. 10 від 11.10.1933, д. Василь -->
|
5582
5628
|
<rule break="no">
|
5583
|
-
<beforebreak>(?<!т\.[\h\v]?)\b
|
5629
|
+
<beforebreak>(?<!т\.[\h\v]?)\b[пд]\.[\h\v]*</beforebreak>
|
5584
5630
|
<afterbreak></afterbreak>
|
5585
5631
|
</rule>
|
5586
5632
|
<!-- усталені скорочення, що збігаються з нескороченими словами -->
|
data/srx-languagetool.gemspec
CHANGED
@@ -16,6 +16,7 @@ Gem::Specification.new do |spec|
|
|
16
16
|
spec.metadata['homepage_uri'] = spec.homepage
|
17
17
|
spec.metadata['source_code_uri'] = 'https://github.com/amake/srx-languagetool-ruby.git'
|
18
18
|
spec.metadata['changelog_uri'] = 'https://github.com/amake/srx-languagetool-ruby/blob/master/CHANGELOG.md'
|
19
|
+
spec.metadata['rubygems_mfa_required'] = 'true'
|
19
20
|
|
20
21
|
# Specify which files should be added to the gem when it is released.
|
21
22
|
# The `git ls-files -z` loads the files in the RubyGem that have been added into git.
|
metadata
CHANGED
@@ -1,14 +1,14 @@
|
|
1
1
|
--- !ruby/object:Gem::Specification
|
2
2
|
name: srx-languagetool
|
3
3
|
version: !ruby/object:Gem::Version
|
4
|
-
version: 0.
|
4
|
+
version: 0.7.0
|
5
5
|
platform: ruby
|
6
6
|
authors:
|
7
7
|
- Aaron Madlon-Kay
|
8
8
|
autorequire:
|
9
9
|
bindir: exe
|
10
10
|
cert_chain: []
|
11
|
-
date:
|
11
|
+
date: 2022-07-04 00:00:00.000000000 Z
|
12
12
|
dependencies:
|
13
13
|
- !ruby/object:Gem::Dependency
|
14
14
|
name: srx
|
@@ -141,6 +141,7 @@ metadata:
|
|
141
141
|
homepage_uri: https://github.com/amake/srx-languagetool-ruby
|
142
142
|
source_code_uri: https://github.com/amake/srx-languagetool-ruby.git
|
143
143
|
changelog_uri: https://github.com/amake/srx-languagetool-ruby/blob/master/CHANGELOG.md
|
144
|
+
rubygems_mfa_required: 'true'
|
144
145
|
post_install_message:
|
145
146
|
rdoc_options: []
|
146
147
|
require_paths:
|