MATRIX DATA: Cope with long variable names

diff --git a/src/language/data-io/matrix-data.c b/src/language/data-io/matrix-data.c
index db7c9d05bdd29941640027151b89daabc1965828..02673f37d20639c107717e26e7118c1e3e76b020 100644 (file)
--- a/src/language/data-io/matrix-data.c
+++ b/src/language/data-io/matrix-data.c
@@ -34,6 +34,7 @@
 #include "language/lexer/variable-parser.h"
 #include "libpspp/i18n.h"
 #include "libpspp/message.h"
+#include "libpspp/misc.h"
 
 #include "gl/xsize.h"
 #include "gl/xalloc.h"
@@ -65,6 +66,8 @@ enum triangle
     FULL
   };
 
+static const int ROWTYPE_WIDTH = 8;
+
 struct matrix_format
 {
   enum triangle triangle;
@@ -94,19 +97,27 @@ valid rowtype_ values:
   PROX.
 */
 
-/* Sets the value of OUTCASE which corresponds to MFORMAT's varname variable
-   to the string STR. VAR must be of type string.
+/* Sets the value of OUTCASE which corresponds to VNAME
+   to the value STR.  VNAME must be of type string.
  */
 static void
-set_varname_column (struct ccase *outcase, const struct matrix_format *mformat,
-     const char *str, int len)
+set_varname_column (struct ccase *outcase, const struct variable *vname,
+     const char *str)
 {
-  const struct variable *var = mformat->varname;
-  uint8_t *s = value_str_rw (case_data_rw (outcase, var), len);
+  int len = var_get_width (vname);
+  uint8_t *s = value_str_rw (case_data_rw (outcase, vname), len);
 
   strncpy ((char *) s, str, len);
 }
 
+static void
+blank_varname_column (struct ccase *outcase, const struct variable *vname)
+{
+  int len = var_get_width (vname);
+  uint8_t *s = value_str_rw (case_data_rw (outcase, vname), len);
+
+  memset (s, ' ', len);
+}
 
 static struct casereader *
 preprocess (struct casereader *casereader0, const struct dictionary *dict, void *aux)
@@ -155,9 +166,9 @@ preprocess (struct casereader *casereader0, const struct dictionary *dict, void
       if (mformat->triangle == UPPER && mformat->diagonal == NO_DIAGONAL)
        c_offset++;
       const union value *v = case_data (c, mformat->rowtype);
-      const char *val = (const char *) value_str (v, 8);
-      if (0 == strncasecmp (val, "corr    ", 8) ||
-         0 == strncasecmp (val, "cov     ", 8))
+      const char *val = (const char *) value_str (v, ROWTYPE_WIDTH);
+      if (0 == strncasecmp (val, "corr    ", ROWTYPE_WIDTH) ||
+         0 == strncasecmp (val, "cov     ", ROWTYPE_WIDTH))
        {
          int col;
          for (col = c_offset; col < mformat->n_continuous_vars; ++col)
@@ -209,14 +220,14 @@ preprocess (struct casereader *casereader0, const struct dictionary *dict, void
       struct ccase *outcase = case_create (proto);
       case_copy (outcase, 0, c, 0, caseproto_get_n_widths (proto));
       const union value *v = case_data (c, mformat->rowtype);
-      const char *val = (const char *) value_str (v, 8);
-      if (0 == strncasecmp (val, "corr    ", 8) ||
-         0 == strncasecmp (val, "cov     ", 8))
+      const char *val = (const char *) value_str (v, ROWTYPE_WIDTH);
+      if (0 == strncasecmp (val, "corr    ", ROWTYPE_WIDTH) ||
+         0 == strncasecmp (val, "cov     ", ROWTYPE_WIDTH))
        {
          int col;
          const struct variable *var = dict_get_var (dict, idx + 1 + row);
-         set_varname_column (outcase, mformat, var_get_name (var), 8);
-         value_copy (case_data_rw (outcase, mformat->rowtype), v, 8);
+         set_varname_column (outcase, mformat->varname, var_get_name (var));
+         value_copy (case_data_rw (outcase, mformat->rowtype), v, ROWTYPE_WIDTH);
 
          for (col = 0; col < mformat->n_continuous_vars; ++col)
            {
@@ -231,18 +242,18 @@ preprocess (struct casereader *casereader0, const struct dictionary *dict, void
        }
       else
        {
-         set_varname_column (outcase, mformat, "        ", 8);
+         blank_varname_column (outcase, mformat->varname);
        }
 
       /* Special case for SD and N_VECTOR: Rewrite as STDDEV and N respectively */
-      if (0 == strncasecmp (val, "sd      ", 8))
+      if (0 == strncasecmp (val, "sd      ", ROWTYPE_WIDTH))
        {
-         value_copy_buf_rpad (case_data_rw (outcase, mformat->rowtype), 8,
+         value_copy_buf_rpad (case_data_rw (outcase, mformat->rowtype), ROWTYPE_WIDTH,
                               (uint8_t *) "STDDEV", 6, ' ');
        }
-      else if (0 == strncasecmp (val, "n_vector", 8))
+      else if (0 == strncasecmp (val, "n_vector", ROWTYPE_WIDTH))
        {
-         value_copy_buf_rpad (case_data_rw (outcase, mformat->rowtype), 8,
+         value_copy_buf_rpad (case_data_rw (outcase, mformat->rowtype), ROWTYPE_WIDTH,
                               (uint8_t *) "N", 1, ' ');
        }
 
@@ -260,7 +271,7 @@ preprocess (struct casereader *casereader0, const struct dictionary *dict, void
 
 
       const struct variable *var = dict_get_var (dict, idx + 1 + row);
-      set_varname_column (outcase, mformat, var_get_name (var), 8);
+      set_varname_column (outcase, mformat->varname, var_get_name (var));
 
       for (col = 0; col < mformat->n_continuous_vars; ++col)
        {
@@ -315,8 +326,7 @@ cmd_matrix (struct lexer *lexer, struct dataset *ds)
   data_parser_set_warn_missing_fields (parser, false);
   data_parser_set_span (parser, false);
 
-  mformat.rowtype = dict_create_var (dict, "ROWTYPE_", 8);
-  mformat.varname = dict_create_var (dict, "VARNAME_", 8);
+  mformat.rowtype = dict_create_var (dict, "ROWTYPE_", ROWTYPE_WIDTH);
 
   mformat.n_continuous_vars = 0;
   mformat.n_split_vars = 0;
@@ -335,6 +345,15 @@ cmd_matrix (struct lexer *lexer, struct dataset *ds)
       goto error;
     }
 
+  int longest_name = 0;
+  for (i = 0; i < n_names; ++i)
+    {
+      maximize_int (&longest_name, strlen (names[i]));
+    }
+
+  mformat.varname = dict_create_var (dict, "VARNAME_",
+                                    8 * DIV_RND_UP (longest_name, 8));
+
   for (i = 0; i < n_names; ++i)
     {
       if (0 == strcasecmp (names[i], "ROWTYPE_"))

diff --git a/tests/language/data-io/matrix-data.at b/tests/language/data-io/matrix-data.at
index 727470cefee6319e97893770b5b1dbe6dee073d7..c200f4284c3d411cfe8a28483a95da9f54c7ea13 100644 (file)
--- a/tests/language/data-io/matrix-data.at
+++ b/tests/language/data-io/matrix-data.at
@@ -275,3 +275,39 @@ matrix-data.pspp:6: error: Stopping syntax file processing here to avoid a casca
 ])
 
 AT_CLEANUP
+
+
+
+AT_SETUP([Matrix data - long variable names])
+
+AT_DATA([matrix-data.pspp], [dnl
+matrix data
+    variables = rowtype_  var01 var_two variable_number_three variableFour
+    /format = upper diagonal.
+
+begin data
+mean         34  35  36  37
+sd           22  11  55  66
+n_vector    100 101 102 103
+corr          1   9   8   7
+corr              1   6   5
+corr                  1   4
+corr                      1
+end data.
+
+list.
+])
+
+AT_CHECK([pspp -O format=csv matrix-data.pspp], [0], [dnl
+Table: Data List
+ROWTYPE_,VARNAME_,var01,var_two,variable_number_three,variableFour
+mean    ,,34.0000,35.0000,36.0000,37.0000
+STDDEV  ,,22.0000,11.0000,55.0000,66.0000
+N       ,,100.0000,101.0000,102.0000,103.0000
+corr    ,var01,1.0000,9.0000,8.0000,7.0000
+corr    ,var_two,9.0000,1.0000,6.0000,5.0000
+corr    ,variable_number_three,8.0000,6.0000,1.0000,4.0000
+corr    ,variableFour,7.0000,5.0000,4.0000,1.0000
+])
+
+AT_CLEANUP