sys-file-reader: Accept document records with no document lines.

This allows PSPP to accept system files that appear to be produced by the
ReadStat software at https://github.com/WizardMac/ReadStat.

Thanks to ftr for reporting this bug via pspp-users at:
http://lists.gnu.org/archive/html/pspp-users/2017-07/msg00002.html

diff --git a/doc/dev/system-file-format.texi b/doc/dev/system-file-format.texi
index 7cc0342c4cd81e966f45abbeafb9c65a3d2cce60..fb131f5274f0d61e90f9eefdf0c994c413b7e468 100644 (file)
--- a/doc/dev/system-file-format.texi
+++ b/doc/dev/system-file-format.texi
@@ -605,7 +605,10 @@ char                lines[][80];
 Record type.  Always set to 6.
 
 @item int32 n_lines;
-Number of lines of documents present.
+Number of lines of documents present.  This should be greater than
+zero, but the system file writer that identifies itself as
+@url{https://github.com/WizardMac/ReadStat} writes document records
+with zero @code{n_lines}.
 
 @item char lines[][80];
 Document lines.  The number of elements is defined by @code{n_lines}.

diff --git a/src/data/sys-file-reader.c b/src/data/sys-file-reader.c
index d1676564d8aa965f831465418a10ad22cfd13400..54788deb6e217bb918d8873077e2cb305ffbea77 100644 (file)
--- a/src/data/sys-file-reader.c
+++ b/src/data/sys-file-reader.c
@@ -282,7 +282,7 @@ static bool read_variable_record (struct sfm_reader *,
                                   struct sfm_var_record *);
 static bool read_value_label_record (struct sfm_reader *,
                                      struct sfm_value_label_record *);
-static struct sfm_document_record *read_document_record (struct sfm_reader *);
+static bool read_document_record (struct sfm_reader *);
 static bool read_extension_record (struct sfm_reader *, int subtype,
                                    struct sfm_extension_record **);
 static bool skip_extension_record (struct sfm_reader *, int subtype);
@@ -500,8 +500,7 @@ read_record (struct sfm_reader *r, int type,
           sys_error (r, r->pos, _("Duplicate type 6 (document) record."));
           return false;
         }
-      r->document = read_document_record (r);
-      return r->document != NULL;
+      return read_document_record (r);
 
     case 7:
       if (!read_int (r, &subtype))
@@ -1229,33 +1228,35 @@ read_value_label_record (struct sfm_reader *r,
   return true;
 }
 
-/* Reads a document record from R and returns it. */
-static struct sfm_document_record *
+/* Reads a document record from R.  Returns true if successful, false on
+   error. */
+static bool
 read_document_record (struct sfm_reader *r)
 {
-  struct sfm_document_record *record;
   int n_lines;
-
-  record = pool_malloc (r->pool, sizeof *record);
-  record->pos = r->pos;
-
   if (!read_int (r, &n_lines))
-    return NULL;
-  if (n_lines <= 0 || n_lines >= INT_MAX / DOC_LINE_LENGTH)
+    return false;
+  else if (n_lines == 0)
+    return true;
+  else if (n_lines < 0 || n_lines >= INT_MAX / DOC_LINE_LENGTH)
     {
-      sys_error (r, record->pos,
+      sys_error (r, r->pos,
                  _("Number of document lines (%d) "
                    "must be greater than 0 and less than %d."),
                  n_lines, INT_MAX / DOC_LINE_LENGTH);
-      return NULL;
+      return false;
     }
 
+  struct sfm_document_record *record;
+  record = pool_malloc (r->pool, sizeof *record);
+  record->pos = r->pos;
   record->n_lines = n_lines;
   record->documents = pool_malloc (r->pool, DOC_LINE_LENGTH * n_lines);
   if (!read_bytes (r, record->documents, DOC_LINE_LENGTH * n_lines))
-    return NULL;
+    return false;
 
-  return record;
+  r->document = record;
+  return true;
 }
 
 static bool

diff --git a/tests/data/sys-file-reader.at b/tests/data/sys-file-reader.at
index 669ed5b9696c65a4c4c963ef8284f92995f4585b..be4782284c0ef8c0a87d0994088c0fd7fe00e34c 100644 (file)
--- a/tests/data/sys-file-reader.at
+++ b/tests/data/sys-file-reader.at
@@ -584,6 +584,53 @@ num1
 done
 AT_CLEANUP
 
+AT_SETUP([empty document record])
+AT_KEYWORDS([sack synthetic system file positive])
+AT_DATA([sys-file.sack], [dnl
+dnl File header.
+"$FL2"; s60 "$(#) SPSS DATA FILE PSPP synthetic test file";
+2; dnl Layout code
+1; dnl Nominal case size
+0; dnl Not compressed
+0; dnl Not weighted 
+1; dnl 1 case.
+100.0; dnl Bias.
+"01 Jan 11"; "20:53:52"; s64 "PSPP synthetic test file";
+i8 0 *3;
+
+dnl Numeric variable, no label or missing values.
+2; 0; 0; 0; 0x050800 *2; s8 "NUM1";
+
+dnl Machine integer info record.
+7; 3; 4; 8; 1; 2; 3; -1; 1; 1; ENDIAN; 1252;
+
+dnl Document record.
+6; 0;
+
+dnl Character encoding record.
+7; 20; 1; 12; "windows-1252";
+
+dnl Dictionary termination record.
+999; 0;
+
+dnl Data.
+1.0;
+])
+for variant in be le; do
+  AT_CHECK([sack --$variant sys-file.sack > sys-file.sav])
+  AT_DATA([sys-file.sps], [dnl
+GET FILE='sys-file.sav'.
+LIST.
+])
+  AT_CHECK([pspp -o pspp.csv sys-file.sps])
+  AT_CHECK([cat pspp.csv], [0], [dnl
+Table: Data List
+num1
+1
+])
+done
+AT_CLEANUP
+
 AT_SETUP([multiple response sets])
 AT_KEYWORDS([sack synthetic system file positive])
 AT_DATA([sys-file.sack], [dnl
@@ -2213,36 +2260,6 @@ done
 AT_CLEANUP
 
 
-AT_SETUP([empty document record])
-AT_KEYWORDS([sack synthetic system file negative])
-AT_DATA([sys-file.sack], [dnl
-dnl File header.
-"$FL2"; s60 "$(#) SPSS DATA FILE PSPP synthetic test file";
-2; 2; 1; 0; -1; 100.0; "01 Jan 11"; "20:53:52"; s64 ""; i8 0 *3;
-
-dnl Numeric variable, no label or missing values.
-2; 0; 0; 0; 0x050800 *2; s8 "NUM1";
-
-dnl Empty document record.
-6; >>0<<;
-
-dnl Dictionary termination record.
-999; 0;
-
-dnl Data.
-1.0;
-])
-for variant in be le; do
-  AT_CHECK([sack --$variant sys-file.sack > sys-file.sav])
-  AT_DATA([sys-file.sps], [dnl
-GET FILE='sys-file.sav'.
-])
-  AT_CHECK([pspp -O format=csv sys-file.sps], [1], [dnl
-error: `sys-file.sav' near offset 0xd4: Number of document lines (0) must be greater than 0 and less than 26843545.
-])
-done
-AT_CLEANUP
-
 AT_SETUP([extension record too large])
 AT_KEYWORDS([sack synthetic system file negative])
 AT_DATA([sys-file.sack], [dnl