script to check problem also output sentences without any problem

scripts/conlluCheckProblems.py

@@ -24,11 +24,11 @@ def checkMCD(mcd) :
 
 ################################################################################
 def logError(message, sentence) :
-  print(message)
+  print(message, file=sys.stderr)
   for line in sentence :
     for col in line :
-      print(col,end="\t")
-    print("")
+      print(col,end="\t", file=sys.stderr)
+    print("", file=sys.stderr)
 ################################################################################
 
 ################################################################################
@@ -43,17 +43,19 @@ def checkSentence(fileLineIndex, sentence, conllMCD, conllMCDr) :
   multiWordEmptyNodes = set()
   id2index = {}
 
+  hadErr = False
 # Verifying IDS
   for i in range(len(sentence)) :
 
     for col in sentence[i] :
       if len(col) == 0 :
         logError("Empty column on line %s"%(fileLineIndex+i), sentence)
-        return
+        return False
 
     idStr = sentence[i][idIndex]
     if idStr in id2index :
       logError("ERROR in IDs : line %s '%s' already seen"%(fileLineIndex+i,idStr), sentence)
+      hadErr = True
 
     id2index[idStr] = i
 
@@ -63,11 +65,13 @@ def checkSentence(fileLineIndex, sentence, conllMCD, conllMCDr) :
       multiWordEmptyNodes.add(i)
       if int(splited[0]) != curId+1 or int(splited[0]) >= int(splited[1]) :
         logError("ERROR in line %s, IDs : %s"%(fileLineIndex+i,idStr), sentence)
+        hadErr = True
     elif len(idStr.split('.')) == 2 :
       multiWordEmptyNodes.add(i)
       splited = idStr.split('.')
       if int(splited[0]) != curId or int(splited[1]) != curDigit :
         logError("ERROR in line %s, IDs : %s"%(fileLineIndex+i,idStr), sentence)
+        hadErr = True
       curDigit += 1
     elif idStr.isdigit() :
       curId += 1
@@ -75,8 +79,10 @@ def checkSentence(fileLineIndex, sentence, conllMCD, conllMCDr) :
       maxId = max(maxId, int(idStr))
       if int(idStr) != curId :
         logError("ERROR in line %s, IDs : %s"%(fileLineIndex+i,idStr), sentence)
+        hadErr = True
     else :
       logError("ERROR in line %s, IDs : %s"%(fileLineIndex+i,idStr), sentence)
+      hadErr = True
 
   nbRoot = 0
 # Verifying root
@@ -87,6 +93,7 @@ def checkSentence(fileLineIndex, sentence, conllMCD, conllMCDr) :
 
   if nbRoot != 1 :
     logError("ERROR %d root in sentence on line %s:"%(nbRoot,fileLineIndex), sentence)
+    hadErr = True
 
 # Verifying govs
   for i in range(len(sentence)) :
@@ -95,8 +102,11 @@ def checkSentence(fileLineIndex, sentence, conllMCD, conllMCDr) :
     govStr = sentence[i][govIndex]
     if not govStr.isdigit() :
       logError("ERROR line %d gov \'%s\' is not positive integer :"%(fileLineIndex+i,govStr), sentence)
+      hadErr = True
+
     if int(govStr) > maxId :
       logError("ERROR line %d gov \'%s\' is out of sentence :"%(fileLineIndex+i,govStr), sentence)
+      hadErr = True
 
 # Verifying cycles
   alreadyReported = {}
@@ -114,8 +124,11 @@ def checkSentence(fileLineIndex, sentence, conllMCD, conllMCDr) :
       if currentNode in alreadySeen :
         if currentNode not in alreadyReported :
           logError("ERROR line %d (id=%s) loop in governors :"%(fileLineIndex+currentNode, sentence[currentNode][idIndex]), sentence)
+          hadErr = True
           alreadyReported[currentNode] = True
         break
+
+  return not hadErr
 ################################################################################
 
 
@@ -139,13 +152,17 @@ if __name__ == "__main__" :
     if len(clean) < 3 :
       if sentFirstLine == -1 :
         exit(1)
-      checkSentence(sentFirstLine, sentence, conllMCDr, conllMCD)
+      if checkSentence(sentFirstLine, sentence, conllMCDr, conllMCD) :
+        for line in sentence :
+          print("\t".join(line))
+        print("")
       sentence = []
       sentFirstLine = -1
       continue
     if clean[0] == '#' :
       splited = line.split("global.columns =")
       if len(splited) > 1 :
+        print(line.strip())
         conllMCD, conllMCDr = readMCD(splited[-1].strip())
         checkMCD(conllMCD)
       continue