commit 7767f5c1c1466e27b4b01bb9f36e41541fd65882
Author: aboelhamd <aboelhamd.abotreka@gmail.com>
Date:   Sat Jun 1 03:26:37 2019 +0200

    Fix some bugs in scripts and add option -n to RulesApplier

diff --git a/choose-best-sents.py b/choose-best-sents.py
index 28b22c6..e9b93ba 100644
--- a/choose-best-sents.py
+++ b/choose-best-sents.py
@@ -8,10 +8,7 @@ minwerFile = open(sys.argv[3], 'w+')
 minperFile = open(sys.argv[4], 'w+')
 minwerperFile = open(sys.argv[5], 'w+')
 
-sents = []
-wers = []
-pers = []
-werspers = []
+sents, wers, pers, werspers = [], [], [], []
 minwer, minper, minwerper, minwerI, minperI, minwerperI = 10000.,10000.,10000.,0,0,0
 
 with open(sys.argv[1]) as scoresFile, open(sys.argv[2]) as combFile: 
@@ -46,6 +43,7 @@ with open(sys.argv[1]) as scoresFile, open(sys.argv[2]) as combFile:
       minwerperFile.write(sents[minwerperI])
       
       minwer, minper, minwerper, minwerI, minperI, minwerperI = 10000.,10000.,10000.,0,0,0
+      sents, wers, pers, werspers = [], [], [], []
       
 
 scoresFile.close()
diff --git a/rem-bad-sents.py b/rem-bad-sents.py
index ad2e263..e7aec1d 100644
--- a/rem-bad-sents.py
+++ b/rem-bad-sents.py
@@ -1,7 +1,7 @@
 import sys
 
 if (len(sys.argv) != 4) :
-	print('\nUsage: python rem-bad-sents.py source-file ambig-target-file new-source-file');
+	print('\nUsage: python rem-bad-sents.py source-file ambig-target-file(with new lines) new-source-file');
 	sys.exit()
 
 srcFile = open(sys.argv[1], 'r')
@@ -15,10 +15,14 @@ for sent in ambigTarFile:
     sents.append(sent)
 
   else :
-    src = scrFile.readline()
+    src = srcFile.readline()
+    bad = False    
     for sent in sents :
-      if (line.find("#") == -1 and line.find("@") == -1) :
-        newSrcFile.write()
+      if (sent.find("*") > -1 or sent.find("#") > -1 or sent.find("@") > -1) :
+        bad = True
+        break
+    if (not bad) : 
+      newSrcFile.write(src)
     sents.clear()
 
 srcFile.close()
diff --git a/src/RulesApplier.cpp b/src/RulesApplier.cpp
index ea36284..692ddfb 100644
--- a/src/RulesApplier.cpp
+++ b/src/RulesApplier.cpp
@@ -28,12 +28,16 @@ int main(int argc, char **argv) {
 	string localeId, transferFilePath, lextorFilePath, chunkerFilePath,
 			newLextorFilePath;
 
+	bool newline = false;
 	int opt;
-	while ((opt = getopt(argc, argv, ":u:")) != -1) {
+	while ((opt = getopt(argc, argv, ":u:n")) != -1) {
 		switch (opt) {
 		case 'u':
 			newLextorFilePath = optarg;
 			break;
+		case 'n':
+			newline = true;
+			break;
 		case ':':
 			printf("option %c needs a value\n", optopt);
 			return -1;
@@ -71,7 +75,7 @@ int main(int argc, char **argv) {
 
 		cout << "Error in parameters !" << endl;
 		cout << "Parameters are : localeId transferFilePath"
-				<< " lextorFilePath chunkerFilePath [-u newlextorFilePath]"
+				<< " lextorFilePath chunkerFilePath [-u newlextorFilePath] [-n]"
 				<< endl;
 		cout
 				<< "localeId : ICU locale ID for the source language. For Kazakh => kk_KZ"
@@ -89,6 +93,9 @@ int main(int argc, char **argv) {
 		cout
 				<< "newlextorFilePath : write the new sentences lextor in this lextor file."
 				<< endl;
+		cout
+				<< "-n : put newline after each sentence ambiguous chunker (to use it removing bad sentences)."
+				<< endl;
 		return -1;
 	}
 
@@ -186,7 +193,8 @@ int main(int argc, char **argv) {
 			for (unsigned j = 0; j < outs.size(); j++) {
 				chunkerFile << outs[j] << endl;
 			}
-//	  interInFile << endl;
+			if (newline)
+				chunkerFile << endl;
 
 			// delete AmbigInfo pointers
 			for (unsigned j = 0; j < ambigInfo.size(); j++) {