nlsplit

commit 9446c742fa5d1c71c844922673fd1f61c879a040
parent 0f21ebaa834f97af49ad835df3093af9d843731a
Author: Antoine Amarilli <a3nm@a3nm.net>
Date:   Sun,  9 Oct 2011 21:41:17 +0200

fix eof detection

Diffstat:
nlsplit.c
 | 
16
++++++++++------

1 file changed, 10 insertions(+), 6 deletions(-)
diff --git a/nlsplit.c b/nlsplit.c
@@ -149,6 +149,8 @@ int split() {
 
   /* current and last character */
   char c, last = 0;
+  /* for eof detection */
+  int c_int;
 
   /* confidence for split at current position */
   float current = 0;
@@ -176,8 +178,9 @@ int split() {
   }
 
   /* read file, break after splitting at EOF */
+  /* do not break when reading EOF, because we must output last chunk */
   while (1) {
-    piece[(offset + (pos++)) % size] = (c = getchar());
+    piece[(offset + (pos++)) % size] = (c = (c_int = getchar()));
 
     current = 0;
 
@@ -265,9 +268,9 @@ int split() {
 
     assert(pos <= size);
     /* cut if we have to */
-    if (c == EOF || pos == size ||
-        (hd != tl && min_confidence > 0
-         && points[hd].confidence >= min_confidence)) {
+    if ((c_int == EOF && pos != 1) || (c_int != EOF && pos == size) ||
+        (hd != tl && min_confidence > 0 &&
+          points[hd].confidence >= min_confidence)) {
       /* pop old entries */
       while (hd != tl && points[hd].position <= offset)
         hd = (hd + 1) % size;
@@ -277,7 +280,7 @@ int split() {
         assert(points[hd].position > offset);
       }
       assert(points[hd].position > offset);
-      if (c == EOF) {
+      if (c_int == EOF) {
         /* special EOF cut */
         points[hd].confidence = EOF_SCORE;
         points[hd].position = offset + pos - 1;
@@ -300,7 +303,8 @@ int split() {
       npiece++;
     }
     last = c;
-    if (c == EOF) break;
+
+    if (c_int == EOF) break;
   }
 
   return 0;