(#53) merge Leonel's work

tools/bootstrap/prepare-delaf.sh

@@ -5,6 +5,8 @@
 # tr -d '\r' < Delaf2015v04.dic > delaf.dic
 
 function splitW31 {
+    # split entries what end in W31 but should be two entries, one
+    # with W3s and the other with W1s
     sed "s/\\(.*W\\)31$/\\13s\\n\\11s/"
 }
 
@@ -28,6 +30,7 @@ grep -F ".N:" "$1" | # select nouns
 
 # 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/:-/-/" |
-    splitW31 > delaf.clitics
+    sed "s/:-/-/" | # rm spurious colon like in abstinhas:-lhe,abster.V+PRO:I2s
+    splitW31 |
+    # add hyphen where it should exist amarnos -> amar-nos
+    SeparateHyphen.py > delaf.clitics

tools/convert-format/MorphoMbr.gf

@@ -70,7 +70,7 @@ concrete MorphoMbr of Morpho = open Prelude, Predef in {
     mkEntry fo l fs = ss (fo.s ++ "&t" ++ l.s ++ fs.s) ;
 
     clitic : Str ;
-    clitic = personstr ++ mode ++ optStr personnum ++ optStr gender ++ optStr number ;`
+    clitic = personstr ++ mode ++ optStr personnum ++ optStr gender ++ optStr number ;
 
     personstr, mode, personnum, gender, number : Str ;
     personstr = "." ++ ("ele" | "vós" | "nós" | "eu" | "tu") ;

tools/python-tools/AnnotateClitics.py

@@ -0,0 +1,42 @@
+#! /usr/bin/env python2.7
+# -*- coding: utf-8 -*-
+
+# Author: Leonel Figueiredo de Alencar - Federal University of Ceará
+# [email protected]
+# Date: July 2, 2018
+"""
+This module annotates enclitic or mesoclitic pronouns in entries in the MBR format
+
+Usage: cat infile.mbr | AnnotateClitics.py > outfile.mbr
+
+It reads entries in the MBR format from standard input and substitutes
+the +PRO tag for tags representing the lemma and morpho-syntactic features
+of the pronouns. The result is written to standard output.
+For example, an entry like
+
+degustares-lhe	degustar+V+PRO+INF+2+SG
+
+is converted to
+
+degustares-lhe	degustar+V.ele.DAT.3.SG+SBJF+2+SG
+
+Tag conversion is performed by the AnnotateClitic function from
+module ConvertDELAF.py. Ambiguity of clitic "nos" is also handled. 
+For more details, see the respective module documentation.
+"""
+import sys
+from ConvertDELAF import *
+
+
+def main():
+    entries=ExtractEntries(sys.stdin)
+    for entry in entries:
+        if HasClitic(entry):
+            parts=ParseEntry(entry,r"\t|\+")
+            word,lemma,cat,feats=parts[0],parts[1],parts[2],parts[4:]
+            sys.stdout.write("%s\n" % AnnotateClitic(word,lemma,cat,feats).encode("utf-8"))
+        else:
+            sys.stdout.write("%s\n" % entry.encode("utf-8"))
+
+if __name__ == '__main__':
+	main()