Allow non-ascii strings. Handle substitutions in dimension titles.

diff --git a/dump.c b/dump.c
index 60056f55b951f8450badf7b9e3001b50ab2acff8..d42580895e34e552d6a762c7650201484e74dad1 100644 (file)
--- a/dump.c
+++ b/dump.c
@@ -128,7 +128,7 @@ static char *
 get_string(void)
 {
   if (data[pos + 1] == 0 && data[pos + 2] == 0 && data[pos + 3] == 0
-      && all_ascii(&data[pos + 4], data[pos]))
+      /*&& all_ascii(&data[pos + 4], data[pos])*/)
     {
       int len = data[pos];
       char *s = malloc(len + 1);
@@ -249,6 +249,105 @@ dump_value(int level)
     }
 }
 
+static void
+dump_dim_value(int level)
+{
+  for (int i = 0; i <= level; i++)
+    printf ("    ");
+
+  if (match_byte (3))
+    {
+      get_string();
+      if (match_byte (0x31))
+        {
+          match_u32 (1);
+          printf("(footnote %d) ", get_u32());
+          match_byte_assert (0);
+          match_byte_assert (0);
+          int subn = get_u32 ();
+          printf ("nested %d bytes", subn);
+          pos += subn;
+        }
+      else
+        match_byte_assert (0x58);
+      get_string();
+      printf("string \"%s\"", get_string());
+      match_byte (0);
+      match_byte_assert (1);
+      match_byte (0);
+      match_byte (0);
+      match_byte (0);
+      match_byte (1);
+    }
+  else if (match_byte (5))
+    {
+      match_byte_assert (0x58);
+      printf ("variable \"%s\"", get_string());
+      get_string();
+      match_byte_assert (2);
+    }
+  else if (match_byte (2))
+    {
+      unsigned int format;
+      char *var, *vallab;
+      double value;
+
+      match_byte_assert (0x58);
+      format = get_u32 ();
+      value = get_double ();
+      var = get_string ();
+      vallab = get_string ();
+      printf ("value %g format %d(%d.%d) var \"%s\" vallab \"%s\"",
+              value, format >> 16, (format >> 8) & 0xff, format & 0xff, var, vallab);
+      if (!match_u32 (3))
+        match_u32_assert (2);
+      match_byte (0);
+    }
+  else if (match_byte (1))
+    {
+      unsigned int format;
+      double value;
+
+      match_byte_assert (0x58);
+      format = get_u32 ();
+      value = get_double ();
+      printf ("value %g format %d(%d.%d)", value, format >> 16, (format >> 8) & 0xff, format & 0xff);
+      match_byte (1);
+      match_byte (0);
+      match_byte (0);
+      match_byte (0);
+      match_byte (1);
+    }
+  else
+    {
+      int subn;
+      int total_subs = 1;
+
+      match_byte (0);
+      match_byte_assert (0x31);
+      match_u32_assert (0);
+      match_u32_assert (0);
+      subn = get_u32 ();
+      printf ("nested %d bytes", subn);
+      pos += subn;
+      printf ("; \"%s\", substitutions:", get_string());
+      for (;;)
+        {
+          int n_subst = get_u32();
+          if (!n_subst)
+            break;
+          printf (" %d", n_subst);
+          total_subs *= n_subst;
+        }
+
+      for (int i = 0; i < total_subs; i++)
+        {
+          putc ('\n', stdout);
+          dump_value (level + 1);
+        }
+    }
+}
+
 static void
 dump_category(int level)
 {
@@ -285,6 +384,8 @@ static void
 dump_dim(void)
 {
   int n_categories;
+  printf("next dim\n");
+  match_byte(0);
   if (match_byte(3))
     {
       get_string();
@@ -293,7 +394,7 @@ dump_dim(void)
       printf("string \"%s\": ", get_string());
       match_byte_assert(1);
     }
-  else if (match_byte(5))
+  else if (match_byte(5)) 
     {
       match_byte_assert(0x58);
       printf("variable \"%s\": ", get_string());
@@ -303,8 +404,30 @@ dump_dim(void)
     }
   else
     {
-      fprintf(stderr, "%08x: unexpected byte\n", pos);
-      exit(1);
+      int subn;
+      int total_subs = 1;
+
+      match_byte_assert(0x31);
+      match_u32_assert (0);
+      match_u32_assert (0);
+      subn = get_u32 ();
+      printf ("nested %d bytes", subn);
+      pos += subn;
+      printf ("; \"%s\", substitutions:", get_string());
+      for (;;)
+        {
+          int n_subst = get_u32();
+          if (!n_subst)
+            break;
+          printf (" %d", n_subst);
+          total_subs *= n_subst;
+        }
+
+      for (int i = 0; i < total_subs; i++)
+        {
+          putc ('\n', stdout);
+          dump_dim_value (0);
+        }
     }
 
   match_byte_assert(0);

diff --git a/notes b/notes
index f5b8cf3a6f15c4e13c2262795c5ca4ac35f3f6a8..1a8c1234bbbd909ee375aaf332ca8da9f51d1bcb 100644 (file)
--- a/notes
+++ b/notes
@@ -817,6 +817,50 @@ tdump25 ./germano/Crosstabs/00000000013_lightTableData.bin
                05 58 "post1.3" 00 00 00 00 i2 i2 i5 00 00 00 00 00
         ...
 
+web/1ae63a04381624dac939ac62eca63fec/00000000053_lightTableData.bin (descriptives)
+
+    i2
+
+    03 "Variables" 58 "variable_s_1" "Variables" 01 00 00 i2 01 00 i1 00 i8
+       05 58 "GODINEPACIJENTA" "GODINE PACIJENTA" i2 i2 00 00 00 00 00 00 00 00
+       05 58 "HLOCORPUS" "HLO CORPUS" i2 i2 i1 00 00 00 00
+       05 58 "INFLCORPUS" "INFL CORPUS" i2 i2 i2 00 00 00 00
+       05 58 "AKT.CORPUS" "AKT. CORPUS" i2 i2 i3 00 00 00 00
+       05 58 "VgmCORPUS" "Vgm CORPUS" i2 i2 i4 00 00 00 00
+       05 58 "VgfCORPUS" "Vgf CORPUS" i2 i2 i5 00 00 00 00
+       05 58 i9 56 67 9e 43 4f 52 50 55 53 i10 56 67 9e 20 43 4f 52 50 55 53 i2 i2 i6 00 00 00 00
+       05 58 "VglpCORPUS" "Vglp CORPUS" i2 i2 i7 00 00 00 00
+
+    03 "Statistics" 58 "statistics_10" "Statistics" 01 00 01 i2 01 00 01 i1 i3
+       03 "Mean" 58 "mean" "Mean" i1 i2 00 00 00 00 00 00 00 00
+       03 "Std. Deviation" 58 "gbar11" "Std. Deviation" i1 i2 i1 00 00 00 00
+       03 "N" 58 "n" "N" i1 i2 i2 00 00 00 00 00 00 00 00
+
+./web/76e9b53e9f775fb690a88919ccdbb03c/000000007251_lightTableData.bin:
+
+    i6
+
+    05 58 "Gender_R" "" 02 00 02 i2 00 00 i1 00 i2
+       02 58 F40.0(0) "Gender_R" "Male" i2 i2 00 00 00 00 00 00 00 00
+       02 58 F40.0(1) "Gender_R" "Female" i2 i2 i1 00 00 00 00
+
+    03 "Statistics" 58 "statistics_10" "Statistics" 01 00 00 i2 01 00 01 i1 i4
+    03 "Mean Difference (I-J)" 58 "mean_difference_i-j" "Mean Difference (I-J)" i1 i2 00 00 00 00 00 00 00 00
+    03 "Std. Error" 58 "standard_error" "Std. Error" i1 i2 i1 00 00 00 00
+    03 "Sig." 58 "significance_level" "Significance" i1 i2 i2 00 00 00 00
+    00 31 i0 i0 i38: (i32: (i0 31 "^1% Confidence Interval") 58 58) "^1% Confidence Interval" i1 00 00 00 00
+       01 58 00 03 i5 00 00 00 c0 57 40 00 00 01                       i2 i-1 i2
+           03 "Lower Bound" 58 "lower_bound" "Lower Bound" i1 i2 i3 00 00 00 00
+           03 "Upper Bound" 58 "upper_bound" "Upper Bound" i1 i2 i4 00 00 00 00
+
+    00 31 i0 i0 i21: (i15: (i0 31 "(J) ^1") 58 58) "(J) ^1" i1 00 00 00 00
+       05 58 "sexorient_r" "" 02 00 00       i2 00 00 01 i2 i3
+           02 58 F40.0(0) "sexorient_r" "Straight" i2 i2 00 00 00 00 00 00 00 00
+           02 58 F40.0(1)"sexorient_r" "Bisexual" i2 i2 i1 00 00 00 00
+           02 58 F40.0(2) "sexorient_r" "Gay/Lesbian/Homosexual" i2 i2 i2 00 00 00 00 00
+
+    ...
+
 
 Data: tdump21 (germano/Crosstabs.pdf)
 -------------
@@ -834,7 +878,7 @@ i32 i0 01 58 F40.3(0.026)
 i52 i0 01 58 F40.3(0.030)
 i36 i0 01 58 F40.3(0.196)
 i56 i0 01 58 F40.3(0.077)
-
+-
 i1 i0 01 58 F40.3(0.055)
 i21 i0 01 58 F40.3(0.077)
 i41 i0 01 58 F40.3(0.056)