u8-istream: Fix handling of partial reads.

The u8-istream code did not retry upon a partial read, assuming that that
was the end of the file.  When the partial read was shorter than
ENCODING_GUESS_MIN, this could cause the encoding guesser, in turn, to
guess the wrong encoding (especially if the encoding was really UTF-16 and
the partial read was an odd number of bytes).

Reported at https://bugs.debian.org/760841.
Reported by Friedrich Beckmann and Steven Chamberlain.

diff --git a/src/libpspp/u8-istream.c b/src/libpspp/u8-istream.c
index 2486ca1ada31c53966e556dfc9932f4ace47a3be..6347a6701a9b86c2147f5d86b3b9a82823045682 100644 (file)
--- a/src/libpspp/u8-istream.c
+++ b/src/libpspp/u8-istream.c
@@ -1,5 +1,5 @@
 /* PSPP - a program for statistical analysis.
-   Copyright (C) 2010, 2011, 2012, 2013 Free Software Foundation, Inc.
+   Copyright (C) 2010, 2011, 2012, 2013, 2014 Free Software Foundation, Inc.
 
    This program is free software: you can redistribute it and/or modify
    it under the terms of the GNU General Public License as published by
@@ -203,14 +203,22 @@ fill_buffer (struct u8_istream *is)
   is->head = is->buffer;
 
   /* Read more input. */
+  n = 0;
   do
     {
-      n = read (is->fd, is->buffer + is->length,
-                U8_ISTREAM_BUFFER_SIZE - is->length);
+      ssize_t retval = read (is->fd, is->buffer + is->length,
+                             U8_ISTREAM_BUFFER_SIZE - is->length);
+      if (retval > 0)
+        {
+          n += retval;
+          is->length += retval;
+        }
+      else if (retval == 0)
+        return n;
+      else if (errno != EINTR)
+        return n > 0 ? n : -1;
     }
-  while (n < 0 && errno == EINTR);
-  if (n > 0)
-    is->length += n;
+  while (is->length < U8_ISTREAM_BUFFER_SIZE);
   return n;
 }
 

diff --git a/tests/libpspp/u8-istream.at b/tests/libpspp/u8-istream.at
index 24af08cc7c847f9a0332ef1421b3ee2f86a82412..5b5b4e006e64e81760680a39f868f854b1c8c480 100644 (file)
--- a/tests/libpspp/u8-istream.at
+++ b/tests/libpspp/u8-istream.at
@@ -140,7 +140,9 @@ AT_CLEANUP
 AT_SETUP([read UTF-16 as Auto])
 AT_KEYWORDS([u8_istream])
 AT_CHECK([i18n-test supports_encodings UTF-16 UTF-16BE UTF-16LE])
-AT_CHECK([printf '\0e\0n\0t\0r\0\351\0e\0\n' | u8-istream-test read - Auto],
+dnl The "sleep 1" checks for a bug in which u8-istream did not properly
+dnl handle receiving data in multiple chunks.
+AT_CHECK([{ printf '\0e\0n\0t\0'; sleep 1; printf 'r\0\351\0e\0\n'; } | u8-istream-test read - Auto],
   [0], [entrée
 ])
 AT_CHECK([printf 'e\0n\0t\0r\0\351\0e\0\n\0' | u8-istream-test read - Auto],