Merge commit 'origin/stable'

[pspp-builds.git] / src / data / por-file-reader.c

diff --git a/src/data/por-file-reader.c b/src/data/por-file-reader.c
index c8e492897f3c117bfbc86e6c6b38a5f68249928d..a463124274a20c6cd62e1071f15c51cd95edcb3a 100644 (file)
--- a/src/data/por-file-reader.c
+++ b/src/data/por-file-reader.c
@@ -43,6 +43,7 @@
 #include <libpspp/pool.h>
 #include <libpspp/str.h>
 
+#include "minmax.h"
 #include "xalloc.h"
 
 #include "gettext.h"
@@ -446,6 +447,28 @@ read_string (struct pfm_reader *r, char *buf)
   *buf = '\0';
 }
 
+
+/* Reads a string into BUF, which must have room for 256
+   characters.
+   Returns the number of bytes read.
+*/
+static size_t
+read_bytes (struct pfm_reader *r, uint8_t *buf)
+{
+  int n = read_int (r);
+  if (n < 0 || n > 255)
+    error (r, _("Bad string length %d."), n);
+
+  while (n-- > 0)
+    {
+      *buf++ = r->cc;
+      advance (r);
+    }
+  return n;
+}
+
+
+
 /* Reads a string and returns a copy of it allocated from R's
    pool. */
 static char *
@@ -606,8 +629,7 @@ assign_default:
   return fmt_default_for_width (var_get_width (v));
 }
 
-static void parse_value (struct pfm_reader *, struct variable *,
-                         union value *);
+static void parse_value (struct pfm_reader *, int width, union value *);
 
 /* Read information on all the variables.  */
 static void
@@ -686,7 +708,7 @@ read_variables (struct pfm_reader *r, struct dictionary *dict)
       var_set_write_format (v, &write);
 
       /* Range missing values. */
-      mv_init (&miss, var_get_width (v));
+      mv_init (&miss, width);
       if (match (r, 'B'))
         {
           double x = read_float (r);
@@ -701,13 +723,17 @@ read_variables (struct pfm_reader *r, struct dictionary *dict)
       /* Single missing values. */
       while (match (r, '8'))
         {
+          int mv_width = MIN (width, 8);
           union value value;
-          parse_value (r, v, &value);
+
+          parse_value (r, mv_width, &value);
+          value_resize (&value, mv_width, width);
           mv_add_value (&miss, &value);
-          value_destroy (&value, var_get_width (v));
+          value_destroy (&value, width);
         }
 
       var_set_missing_values (v, &miss);
+      mv_destroy (&miss);
 
       if (match (r, 'C'))
         {
@@ -728,16 +754,16 @@ read_variables (struct pfm_reader *r, struct dictionary *dict)
     }
 }
 
-/* Parse a value for variable VV into value V. */
+/* Parse a value of with WIDTH into value V. */
 static void
-parse_value (struct pfm_reader *r, struct variable *vv, union value *v)
+parse_value (struct pfm_reader *r, int width, union value *v)
 {
-  value_init (v, var_get_width (vv));
-  if (var_is_alpha (vv))
+  value_init (v, width);
+  if (width > 0)
     {
-      char string[256];
-      read_string (r, string);
-      buf_copy_str_rpad (value_str_rw (v, 8), 8, string, ' ');
+      uint8_t buf[256];
+      size_t n_bytes = read_bytes (r, buf);
+      value_copy_buf_rpad (v, width, buf, n_bytes, ' ');
     }
   else
     v->f = read_float (r);
@@ -780,7 +806,7 @@ read_value_label (struct pfm_reader *r, struct dictionary *dict)
       char label[256];
       int j;
 
-      parse_value (r, v[0], &val);
+      parse_value (r, var_get_width (v[0]), &val);
       read_string (r, label);
 
       /* Assign the value label to each variable. */
@@ -815,7 +841,6 @@ por_file_casereader_read (struct casereader *reader, void *r_)
   struct pfm_reader *r = r_;
   struct ccase *volatile c;
   size_t i;
-  size_t idx;
 
   c = case_create (r->proto);
   setjmp (r->bail_out);
@@ -833,22 +858,17 @@ por_file_casereader_read (struct casereader *reader, void *r_)
       return NULL;
     }
 
-  idx = 0;
   for (i = 0; i < r->var_cnt; i++)
     {
       int width = caseproto_get_width (r->proto, i);
 
       if (width == 0)
-        {
-          case_data_rw_idx (c, idx)->f = read_float (r);
-          idx++;
-        }
+        case_data_rw_idx (c, i)->f = read_float (r);
       else
         {
-          char string[256];
-          read_string (r, string);
-          buf_copy_str_rpad (case_str_rw_idx (c, idx), width, string, ' ');
-          idx += DIV_RND_UP (width, MAX_SHORT_STRING);
+          uint8_t buf[256];
+          size_t n_bytes = read_bytes (r, buf);
+          u8_buf_copy_rpad (case_str_rw_idx (c, i), width, buf, n_bytes, ' ');
         }
     }
 
@@ -862,17 +882,30 @@ pfm_detect (FILE *file)
 {
   unsigned char header[464];
   char trans[256];
-  int cooked_cnt, raw_cnt;
+  int cooked_cnt, raw_cnt, line_len;
   int i;
 
   cooked_cnt = raw_cnt = 0;
+  line_len = 0;
   while (cooked_cnt < sizeof header)
     {
       int c = getc (file);
       if (c == EOF || raw_cnt++ > 512)
         return false;
-      else if (c != '\n' && c != '\r')
-        header[cooked_cnt++] = c;
+      else if (c == '\n')
+        {
+          while (line_len < 80 && cooked_cnt < sizeof header)
+            {
+              header[cooked_cnt++] = ' ';
+              line_len++;
+            }
+          line_len = 0;
+        }
+      else if (c != '\r')
+        {
+          header[cooked_cnt++] = c;
+          line_len++;
+        }
     }
 
   memset (trans, 0, 256);