Improved conllu2latex, now alow line breaks

scripts/conllu2latex.py

@@ -90,29 +90,44 @@ if __name__ == "__main__" :
     i = j
 
   for i in range(len(ranges)-1) :
-    ranges[i][1] = ranges[i+1][0]-1
+    if ranges[i][1] != ranges[i+1][0]-1 :
+      if ranges[i][1]-ranges[i][0] <= ranges[i+1][1]-ranges[i+1][0] :
+        ranges[i][1] = ranges[i+1][0]-1
+      else :
+        ranges[i+1][0] = ranges[i][1]+1
 
-  print([text[r[0]:r[1]+1] for r in ranges])
+  maxNbLetters = 45
 
-  print("\\newcolumntype{x}[0]{>{\centering\\arraybackslash}m{2.2mm}}") 
+  parts = [[]]
+  first = 0
+  for i in range(len(ranges)) :
+    if ranges[i][1]-first > maxNbLetters :
+      parts.append([])
+      first = ranges[i][0]
+    parts[-1].append(i)
+
+  partSizes = [-ranges[parts[partId][0]][0]+ranges[parts[partId][-1]][1]+1 for partId in range(len(parts))]
+
+  #print("\\newcolumntype{x}[0]{>{\centering\\arraybackslash}m{2.2mm}}") 
   print("\\begin{table}[t]")
   print("\centering")
   print("\\footnotesize")
-  print("\\tabcolsep=0.45mm")
-  print("\\begin{tabular}{|l|%s|}"%("|".join('x'*len(text))))
-  print("\hline\n")
+  print("\\tabcolsep=0.40mm")
+  print("\\begin{tabular}{|l|%s|}"%("|".join(["c"]*max(partSizes))))
+  for partId in range(len(parts)) :
+    print("\cline{1-%d}\n"%(partSizes[partId]+1))
+    for i in range(len(columns))[::-1] :
+      print("\\texttt{\\textbf{%s}}"%columns[i].lower(), end=" &\n")
+      for j in parts[partId] :
+        value = sentence[j][i]
+        if columns[i] not in ["FORM","LEMMA"] :
+          value = "\\texttt{%s}"%(value.lower())
+        else :
+          value = "\\texttt{%s}"%(value)
+        print("\multicolumn{%d}{c|}{%s}"%(ranges[j][1]-ranges[j][0]+1, value), end=" &\n" if j != parts[partId][-1] else "")
+      print("\\\\ \cline{1-%d}\n"%(partSizes[partId]+1))
   
-  for i in range(len(columns))[::-1] :
-    print("\\texttt{\\textsc{%s}}"%columns[i].lower(), end=" &\n")
-    for j in range(len(sentence)) :
-      value = sentence[j][i]
-      if columns[i] not in ["FORM","LEMMA"] :
-        value = "\\textsc{%s}"%(value.lower())
-      print("\multicolumn{%d}{c|}{%s}"%(ranges[j][1]-ranges[j][0]+1, value), end=" &\n" if j != len(sentence)-1 else "")
-    print("\\\\ \hline\n")
-
-  print("\\texttt{\\textsc{input}} & %s\\\\ \hline"%" & ".join(text))
-
+    print("\\texttt{\\textbf{input}} & %s\\\\ \cline{1-%d}"%(" & ".join(["\\texttt{%s}"%c for c in text[ranges[parts[partId][0]][0]:ranges[parts[partId][-1]][1]+1]]), partSizes[partId]+1))
   print("\end{tabular}")
   print("\end{table}")
 ################################################################################