(#53) bootstrap scripts

- add splitting of `W31` in delaf
- add script for freeling
- update upstream-problems.org

tools/prepare-delaf.sh

@@ -1,7 +1,13 @@
 #!/bin/bash
+# first argument is path to delaf dictionary
+
 # don't forget to change CRLF to LF!
 # tr -d '\r' < Delaf2015v04.dic > delaf.dic
 
+function splitW31 {
+    sed "s/\(.*W\)31$/\13s\n\11s/"
+}
+
 # adjectives
 grep -F ".A:" $1 > delaf.adj
 
@@ -13,13 +19,16 @@ grep -F ".N:" $1 | # select nouns
 #remove uppercase lemmas (Gloria)
     grep -v ",[A-Z]" > delaf.nouns
 
-# simple verbs
-grep -F ".V:" $1 | # select simple verbs
-    sed "s/:,/,/" | # rm entries like mantinhas:,manter.V:I2s
-# sorri,sorrir.V:Y2S -> sorri,sorrir.V:Y2s
-    sed "s/2S$/2s/" > delaf.verbs
+# # simple verbs
+ grep -F ".V:" $1 | # select simple verbs
+     sed "s/:,/,/" | # rm entries like mantinhas:,manter.V:I2s
+# # sorri,sorrir.V:Y2S -> sorri,sorrir.V:Y2s
+     sed "s/2S$/2s/" | # split entries like abstrair,abstrair.V:W31
+     splitW31 > delaf.verbs
 
 # verbs with clitics
 grep -F ".V+PRO:" $1 | # select verbs with clitics
 # rm spurious colon like in abstinhas:-lhe,abster.V+PRO:I2s
-    sed "s/:-/-/" > delaf.clitics
+    sed "s/:-/-/" |
+    splitW31 > delaf.clitics
+

tools/prepare-freeling.sh

@@ -0,0 +1,17 @@
+#!/bin/bash
+# first argument if path to directory where freeling dictionary files
+# are, with their original names
+
+# adjectives
+# change C as diminutive tag to D (as is in nouns)
+sed "s/ AQC/ AQD/" adjs > fl.adjectives
+
+# adverbs
+mv adv fl.adverbs
+
+# nouns
+# correct nouns with wrong C tag such as habeas-corpus
+sed "s/ NCMC/ NCMN/" nouns > fl.nouns
+
+# verbs
+mv verbs fl.verb

tools/upstream-problems.org

@@ -25,7 +25,8 @@
   subjuntive instead of singular number:
   : sorri,sorrir.V:Y2S
 - several hundred forms of infinitive had been marked as having two
-  persons, where these should have been in two lines instead:
+  persons, where these should have been in two lines instead, and
+  should have a number tag:
   : abstrair,abstrair.V:W31
 - several hundred forms missing hifen:
   : protrairnos	protrair+V.None+SBJF+1+SG