src/data/por-file-reader.c

   1 /* PSPP - a program for statistical analysis.
   2    Copyright (C) 1997-9, 2000, 2006, 2009, 2010, 2011, 2012, 2013, 2014, 2015 Free Software Foundation, Inc.
   3
   4    This program is free software: you can redistribute it and/or modify
   5    it under the terms of the GNU General Public License as published by
   6    the Free Software Foundation, either version 3 of the License, or
   7    (at your option) any later version.
   8
   9    This program is distributed in the hope that it will be useful,
  10    but WITHOUT ANY WARRANTY; without even the implied warranty of
  11    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  12    GNU General Public License for more details.
  13
  14    You should have received a copy of the GNU General Public License
  15    along with this program.  If not, see <http://www.gnu.org/licenses/>. */
  16
  17 #include <config.h>
  18
  19 #include <ctype.h>
  20 #include <errno.h>
  21 #include <math.h>
  22 #include <setjmp.h>
  23 #include <stdarg.h>
  24 #include <stdbool.h>
  25 #include <stdio.h>
  26 #include <stdlib.h>
  27
  28 #include "data/any-reader.h"
  29 #include "data/casereader-provider.h"
  30 #include "data/casereader.h"
  31 #include "data/dictionary.h"
  32 #include "data/file-handle-def.h"
  33 #include "data/file-name.h"
  34 #include "data/format.h"
  35 #include "data/missing-values.h"
  36 #include "data/short-names.h"
  37 #include "data/value-labels.h"
  38 #include "data/variable.h"
  39 #include "libpspp/compiler.h"
  40 #include "libpspp/i18n.h"
  41 #include "libpspp/message.h"
  42 #include "libpspp/misc.h"
  43 #include "libpspp/pool.h"
  44 #include "libpspp/str.h"
  45
  46 #include "gl/minmax.h"
  47 #include "gl/xalloc.h"
  48 #include "gl/xmemdup0.h"
  49
  50 #include "gettext.h"
  51 #define _(msgid) gettext (msgid)
  52 #define N_(msgid) (msgid)
  53
  54 /* portable_to_local[PORTABLE] translates the given portable
  55    character into the local character set. */
  56 static const char portable_to_local[256] =
  57   {
  58     "                                                                "
  59     "0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz ."
  60     "<(+|&[]!$*);^-/|,%_>?`:$@'=\"      ~-   0123456789   -() {}\\     "
  61     "                                                                "
  62   };
  63
  64 /* Portable file reader. */
  65 struct pfm_reader
  66   {
  67     struct any_reader any_reader;
  68     struct pool *pool;          /* All the portable file state. */
  69
  70     jmp_buf bail_out;           /* longjmp() target for error handling. */
  71
  72     struct dictionary *dict;
  73     struct any_read_info info;
  74     struct file_handle *fh;     /* File handle. */
  75     struct fh_lock *lock;       /* Read lock for file. */
  76     FILE *file;                 /* File stream. */
  77     int line_length;            /* Number of characters so far on this line. */
  78     char cc;                    /* Current character. */
  79     char *trans;                /* 256-byte character set translation table. */
  80     int n_vars;                 /* Number of variables. */
  81     int weight_index;           /* 0-based index of weight variable, or -1. */
  82     struct caseproto *proto;    /* Format of output cases. */
  83     bool ok;                    /* Set false on I/O error. */
  84   };
  85
  86 static const struct casereader_class por_file_casereader_class;
  87
  88 static struct pfm_reader *
  89 pfm_reader_cast (const struct any_reader *r_)
  90 {
  91   assert (r_->klass == &por_file_reader_class);
  92   return UP_CAST (r_, struct pfm_reader, any_reader);
  93 }
  94
  95 static void
  96 error (struct pfm_reader *r, const char *msg,...)
  97      PRINTF_FORMAT (2, 3)
  98      NO_RETURN;
  99
 100 /* Displays MSG as an error message and aborts reading the
 101    portable file via longjmp(). */
 102 static void
 103 error (struct pfm_reader *r, const char *msg, ...)
 104 {
 105   struct string text;
 106   va_list args;
 107
 108   ds_init_empty (&text);
 109   ds_put_format (&text, _("portable file %s corrupt at offset 0x%llx: "),
 110                  fh_get_file_name (r->fh), (long long int) ftello (r->file));
 111   va_start (args, msg);
 112   ds_put_vformat (&text, msg, args);
 113   va_end (args);
 114
 115   struct msg *m = xmalloc (sizeof *m);
 116   *m = (struct msg) {
 117     .category = MSG_C_GENERAL,
 118     .severity = MSG_S_ERROR,
 119     .text = ds_steal_cstr (&text),
 120   };
 121   msg_emit (m);
 122
 123   r->ok = false;
 124
 125   longjmp (r->bail_out, 1);
 126 }
 127
 128 /* Displays MSG as an warning for the current position in
 129    portable file reader R. */
 130 static void
 131 warning (struct pfm_reader *r, const char *msg, ...)
 132 {
 133   struct string text;
 134   va_list args;
 135
 136   ds_init_empty (&text);
 137   ds_put_format (&text, _("reading portable file %s at offset 0x%llx: "),
 138                  fh_get_file_name (r->fh), (long long int) ftello (r->file));
 139   va_start (args, msg);
 140   ds_put_vformat (&text, msg, args);
 141   va_end (args);
 142
 143   struct msg *m = xmalloc (sizeof *m);
 144   *m = (struct msg) {
 145     .category = MSG_C_GENERAL,
 146     .severity = MSG_S_WARNING,
 147     .text = ds_steal_cstr (&text),
 148   };
 149   msg_emit (m);
 150 }
 151
 152 /* Close and destroy R.
 153    Returns false if an error was detected on R, true otherwise. */
 154 static bool
 155 pfm_close (struct any_reader *r_)
 156 {
 157   struct pfm_reader *r = pfm_reader_cast (r_);
 158   bool ok;
 159
 160   dict_unref (r->dict);
 161   any_read_info_destroy (&r->info);
 162   if (r->file)
 163     {
 164       if (fn_close (r->fh, r->file) == EOF)
 165         {
 166           msg (ME, _("Error closing portable file `%s': %s."),
 167                fh_get_file_name (r->fh), strerror (errno));
 168           r->ok = false;
 169         }
 170       r->file = NULL;
 171     }
 172
 173   fh_unlock (r->lock);
 174   fh_unref (r->fh);
 175
 176   ok = r->ok;
 177   pool_destroy (r->pool);
 178
 179   return ok;
 180 }
 181
 182 /* Closes portable file reader R, after we're done with it. */
 183 static void
 184 por_file_casereader_destroy (struct casereader *reader, void *r_)
 185 {
 186   struct pfm_reader *r = r_;
 187   if (!pfm_close (&r->any_reader))
 188     casereader_force_error (reader);
 189 }
 190
 191 /* Read a single character into cur_char.  */
 192 static void
 193 advance (struct pfm_reader *r)
 194 {
 195   int c;
 196
 197   /* Read the next character from the file.
 198      Ignore carriage returns entirely.
 199      Mostly ignore new-lines, but if a new-line occurs before the
 200      line has reached 80 bytes in length, then treat the
 201      "missing" bytes as spaces. */
 202   for (;;)
 203     {
 204       while ((c = getc (r->file)) == '\r')
 205         continue;
 206       if (c != '\n')
 207         break;
 208
 209       if (r->line_length < 80)
 210         {
 211           c = ' ';
 212           ungetc ('\n', r->file);
 213           break;
 214         }
 215       r->line_length = 0;
 216     }
 217   if (c == EOF)
 218     error (r, _("unexpected end of file"));
 219
 220   if (r->trans != NULL)
 221     c = r->trans[c];
 222   r->cc = c;
 223   r->line_length++;
 224 }
 225
 226 /* Skip a single character if present, and return whether it was
 227    skipped. */
 228 static inline bool
 229 match (struct pfm_reader *r, int c)
 230 {
 231   if (r->cc == c)
 232     {
 233       advance (r);
 234       return true;
 235     }
 236   else
 237     return false;
 238 }
 239
 240 static void read_header (struct pfm_reader *);
 241 static void read_version_data (struct pfm_reader *, struct any_read_info *);
 242 static void read_variables (struct pfm_reader *, struct dictionary *);
 243 static void read_value_label (struct pfm_reader *, struct dictionary *);
 244 static void read_documents (struct pfm_reader *, struct dictionary *);
 245
 246 /* Reads the dictionary from file with handle H, and returns it in a
 247    dictionary structure.  This dictionary may be modified in order to
 248    rename, reorder, and delete variables, etc. */
 249 static struct any_reader *
 250 pfm_open (struct file_handle *fh)
 251 {
 252   struct pool *volatile pool = NULL;
 253   struct pfm_reader *volatile r = NULL;
 254
 255   /* Create and initialize reader. */
 256   pool = pool_create ();
 257   r = pool_alloc (pool, sizeof *r);
 258   r->any_reader.klass = &por_file_reader_class;
 259   r->dict = dict_create (get_default_encoding ());
 260   memset (&r->info, 0, sizeof r->info);
 261   r->pool = pool;
 262   r->fh = fh_ref (fh);
 263   r->lock = NULL;
 264   r->file = NULL;
 265   r->line_length = 0;
 266   r->weight_index = -1;
 267   r->trans = NULL;
 268   r->n_vars = 0;
 269   r->proto = NULL;
 270   r->ok = true;
 271   if (setjmp (r->bail_out))
 272     goto error;
 273
 274   /* Lock file. */
 275   /* TRANSLATORS: this fragment will be interpolated into
 276      messages in fh_lock() that identify types of files. */
 277   r->lock = fh_lock (fh, FH_REF_FILE, N_("portable file"), FH_ACC_READ, false);
 278   if (r->lock == NULL)
 279     goto error;
 280
 281   /* Open file. */
 282   r->file = fn_open (r->fh, "rb");
 283   if (r->file == NULL)
 284     {
 285       msg (ME, _("An error occurred while opening `%s' for reading "
 286                  "as a portable file: %s."),
 287            fh_get_file_name (r->fh), strerror (errno));
 288       goto error;
 289     }
 290
 291   /* Read header, version, date info, product id, variables. */
 292   read_header (r);
 293   read_version_data (r, &r->info);
 294   read_variables (r, r->dict);
 295
 296   /* Read value labels. */
 297   while (match (r, 'D'))
 298     read_value_label (r, r->dict);
 299
 300   /* Read documents. */
 301   if (match (r, 'E'))
 302     read_documents (r, r->dict);
 303
 304   /* Check that we've made it to the data. */
 305   if (!match (r, 'F'))
 306     error (r, _("Data record expected."));
 307
 308   r->proto = caseproto_ref_pool (dict_get_proto (r->dict), r->pool);
 309   return &r->any_reader;
 310
 311  error:
 312   pfm_close (&r->any_reader);
 313   return NULL;
 314 }
 315
 316 static struct casereader *
 317 pfm_decode (struct any_reader *r_, const char *encoding UNUSED,
 318             struct dictionary **dictp, struct any_read_info *info)
 319 {
 320   struct pfm_reader *r = pfm_reader_cast (r_);
 321
 322   *dictp = r->dict;
 323   r->dict = NULL;
 324
 325   if (info)
 326     {
 327       *info = r->info;
 328       memset (&r->info, 0, sizeof r->info);
 329     }
 330
 331   return casereader_create_sequential (NULL, r->proto, CASENUMBER_MAX,
 332                                        &por_file_casereader_class, r);
 333 }
 334 \f
 335 /* Returns the value of base-30 digit C,
 336    or -1 if C is not a base-30 digit. */
 337 static int
 338 base_30_value (unsigned char c)
 339 {
 340   static const char base_30_digits[] = "0123456789ABCDEFGHIJKLMNOPQRST";
 341   const char *p = strchr (base_30_digits, c);
 342   return p != NULL ? p - base_30_digits : -1;
 343 }
 344
 345 /* Read a floating point value and return its value. */
 346 static double
 347 read_float (struct pfm_reader *r)
 348 {
 349   double num = 0.;
 350   int exponent = 0;
 351   bool got_dot = false;         /* Seen a decimal point? */
 352   bool got_digit = false;       /* Seen any digits? */
 353   bool negative = false;        /* Number is negative? */
 354
 355   /* Skip leading spaces. */
 356   while (match (r, ' '))
 357     continue;
 358
 359   /* `*' indicates system-missing. */
 360   if (match (r, '*'))
 361     {
 362       advance (r);      /* Probably a dot (.) but doesn't appear to matter. */
 363       return SYSMIS;
 364     }
 365
 366   negative = match (r, '-');
 367   for (;;)
 368     {
 369       int digit = base_30_value (r->cc);
 370       if (digit != -1)
 371         {
 372           got_digit = true;
 373
 374           /* Make sure that multiplication by 30 will not overflow.  */
 375           if (num > DBL_MAX * (1. / 30.))
 376             /* The value of the digit doesn't matter, since we have already
 377                gotten as many digits as can be represented in a `double'.
 378                This doesn't necessarily mean the result will overflow.
 379                The exponent may reduce it to within range.
 380
 381                We just need to record that there was another
 382                digit so that we can multiply by 10 later.  */
 383             ++exponent;
 384           else
 385             num = (num * 30.0) + digit;
 386
 387           /* Keep track of the number of digits after the decimal point.
 388              If we just divided by 30 here, we would lose precision.  */
 389           if (got_dot)
 390             --exponent;
 391         }
 392       else if (!got_dot && r->cc == '.')
 393         /* Record that we have found the decimal point.  */
 394         got_dot = 1;
 395       else
 396         /* Any other character terminates the number.  */
 397         break;
 398
 399       advance (r);
 400     }
 401
 402   /* Check that we had some digits. */
 403   if (!got_digit)
 404     error (r, _("Number expected."));
 405
 406   /* Get exponent if any. */
 407   if (r->cc == '+' || r->cc == '-')
 408     {
 409       long int exp = 0;
 410       bool negative_exponent = r->cc == '-';
 411       int digit;
 412
 413       for (advance (r); (digit = base_30_value (r->cc)) != -1; advance (r))
 414         {
 415           if (exp > LONG_MAX / 30)
 416             {
 417               exp = LONG_MAX;
 418               break;
 419             }
 420           exp = exp * 30 + digit;
 421         }
 422
 423       /* We don't check whether there were actually any digits, but we
 424          probably should. */
 425       if (negative_exponent)
 426         exp = -exp;
 427       exponent += exp;
 428     }
 429
 430   /* Numbers must end with `/'. */
 431   if (!match (r, '/'))
 432     error (r, _("Missing numeric terminator."));
 433
 434   /* Multiply `num' by 30 to the `exponent' power, checking for
 435      overflow.  */
 436   if (exponent < 0)
 437     num *= pow (30.0, (double) exponent);
 438   else if (exponent > 0)
 439     {
 440       if (num > DBL_MAX * pow (30.0, (double) -exponent))
 441         num = DBL_MAX;
 442       else
 443         num *= pow (30.0, (double) exponent);
 444     }
 445
 446   return negative ? -num : num;
 447 }
 448
 449 /* Read an integer and return its value. */
 450 static int
 451 read_int (struct pfm_reader *r)
 452 {
 453   double f = read_float (r);
 454   if (floor (f) != f || f >= INT_MAX || f <= INT_MIN)
 455     error (r, _("Invalid integer."));
 456   return f;
 457 }
 458
 459 /* Reads a string into BUF, which must have room for 256
 460    characters. */
 461 static void
 462 read_string (struct pfm_reader *r, char *buf)
 463 {
 464   int n = read_int (r);
 465   if (n < 0 || n > 255)
 466     error (r, _("Bad string length %d."), n);
 467
 468   while (n-- > 0)
 469     {
 470       *buf++ = r->cc;
 471       advance (r);
 472     }
 473   *buf = '\0';
 474 }
 475
 476
 477 /* Reads a string into BUF, which must have room for 256
 478    characters.
 479    Returns the number of bytes read.
 480 */
 481 static size_t
 482 read_bytes (struct pfm_reader *r, uint8_t *buf)
 483 {
 484   int n = read_int (r);
 485   if (n < 0 || n > 255)
 486     error (r, _("Bad string length %d."), n);
 487
 488   while (n-- > 0)
 489     {
 490       *buf++ = r->cc;
 491       advance (r);
 492     }
 493   return n;
 494 }
 495
 496
 497
 498 /* Reads a string and returns a copy of it allocated from R's
 499    pool. */
 500 static char *
 501 read_pool_string (struct pfm_reader *r)
 502 {
 503   char string[256];
 504   read_string (r, string);
 505   return pool_strdup (r->pool, string);
 506 }
 507 \f
 508 /* Reads the 464-byte file header. */
 509 static void
 510 read_header (struct pfm_reader *r)
 511 {
 512   char *trans;
 513   int i;
 514
 515   /* Read and ignore vanity splash strings. */
 516   for (i = 0; i < 200; i++)
 517     advance (r);
 518
 519   /* Skip the first 64 characters of the translation table.
 520      We don't care about these.  They are probably all set to
 521      '0', marking them as untranslatable, and that would screw
 522      up our actual translation of the real '0'. */
 523   for (i = 0; i < 64; i++)
 524     advance (r);
 525
 526   /* Read the rest of the translation table. */
 527   trans = pool_malloc (r->pool, 256);
 528   memset (trans, 0, 256);
 529   for (; i < 256; i++)
 530     {
 531       unsigned char c;
 532
 533       advance (r);
 534
 535       c = r->cc;
 536       if (trans[c] == 0)
 537         trans[c] = portable_to_local[i];
 538     }
 539
 540   /* Set up the translation table, then read the first
 541      translated character. */
 542   r->trans = trans;
 543   advance (r);
 544
 545   /* Skip and verify signature. */
 546   for (i = 0; i < 8; i++)
 547     if (!match (r, "SPSSPORT"[i]))
 548       {
 549         msg (SE, _("%s: Not a portable file."), fh_get_file_name (r->fh));
 550         longjmp (r->bail_out, 1);
 551       }
 552 }
 553
 554 /* Reads the version and date info record, as well as product and
 555    subproduct identification records if present. */
 556 static void
 557 read_version_data (struct pfm_reader *r, struct any_read_info *info)
 558 {
 559   static const char empty_string[] = "";
 560   char *date, *time;
 561   const char *product, *subproduct;
 562   int i;
 563
 564   /* Read file. */
 565   if (!match (r, 'A'))
 566     error (r, _("Unrecognized version code `%c'."), r->cc);
 567   date = read_pool_string (r);
 568   time = read_pool_string (r);
 569   product = match (r, '1') ? read_pool_string (r) : empty_string;
 570   if (match (r, '2'))
 571     {
 572       /* Skip "author" field. */
 573       read_pool_string (r);
 574     }
 575   subproduct = match (r, '3') ? read_pool_string (r) : empty_string;
 576
 577   /* Validate file. */
 578   if (strlen (date) != 8)
 579     error (r, _("Bad date string length %zu."), strlen (date));
 580   if (strlen (time) != 6)
 581     error (r, _("Bad time string length %zu."), strlen (time));
 582
 583   /* Save file info. */
 584   if (info != NULL)
 585     {
 586       memset (info, 0, sizeof *info);
 587
 588       info->float_format = FLOAT_NATIVE_DOUBLE;
 589       info->integer_format = INTEGER_NATIVE;
 590       info->compression = ANY_COMP_NONE;
 591       info->n_cases = -1;
 592
 593       /* Date. */
 594       info->creation_date = xmalloc (11);
 595       for (i = 0; i < 8; i++)
 596         {
 597           static const int map[] = {6, 7, 8, 9, 3, 4, 0, 1};
 598           info->creation_date[map[i]] = date[i];
 599         }
 600       info->creation_date[2] = info->creation_date[5] = ' ';
 601       info->creation_date[10] = '\0';
 602
 603       /* Time. */
 604       info->creation_time = xmalloc (9);
 605       for (i = 0; i < 6; i++)
 606         {
 607           static const int map[] = {0, 1, 3, 4, 6, 7};
 608           info->creation_time[map[i]] = time[i];
 609         }
 610       info->creation_time[2] = info->creation_time[5] = ' ';
 611       info->creation_time[8] = 0;
 612
 613       /* Product. */
 614       info->product = xstrdup (product);
 615       info->product_ext = xstrdup (subproduct);
 616     }
 617 }
 618
 619 /* Translates a format specification read from portable file R as
 620    the three integers INTS into a normal format specifier FORMAT,
 621    checking that the format is appropriate for variable V. */
 622 static struct fmt_spec
 623 convert_format (struct pfm_reader *r, const int portable_format[3],
 624                 struct variable *v, bool *report_error)
 625 {
 626   struct fmt_spec format;
 627   bool ok;
 628
 629   if (!fmt_from_io (portable_format[0], &format.type))
 630     {
 631       if (*report_error)
 632         warning (r, _("%s: Bad format specifier byte (%d).  Variable "
 633                       "will be assigned a default format."),
 634                  var_get_name (v), portable_format[0]);
 635       goto assign_default;
 636     }
 637
 638   format.w = portable_format[1];
 639   format.d = portable_format[2];
 640
 641   msg_disable ();
 642   ok = (fmt_check_output (&format)
 643         && fmt_check_width_compat (&format, var_get_width (v)));
 644   msg_enable ();
 645
 646   if (!ok)
 647     {
 648       if (*report_error)
 649         {
 650           char fmt_string[FMT_STRING_LEN_MAX + 1];
 651           fmt_to_string (&format, fmt_string);
 652           if (var_is_numeric (v))
 653             warning (r, _("Numeric variable %s has invalid format "
 654                           "specifier %s."),
 655                      var_get_name (v), fmt_string);
 656           else
 657             warning (r, _("String variable %s with width %d has "
 658                           "invalid format specifier %s."),
 659                      var_get_name (v), var_get_width (v), fmt_string);
 660         }
 661       goto assign_default;
 662     }
 663
 664   return format;
 665
 666 assign_default:
 667   *report_error = false;
 668   return fmt_default_for_width (var_get_width (v));
 669 }
 670
 671 static void parse_value (struct pfm_reader *, int width, union value *);
 672
 673 /* Read information on all the variables.  */
 674 static void
 675 read_variables (struct pfm_reader *r, struct dictionary *dict)
 676 {
 677   char *weight_name = NULL;
 678   int i;
 679
 680   if (!match (r, '4'))
 681     error (r, _("Expected variable count record."));
 682
 683   r->n_vars = read_int (r);
 684   if (r->n_vars <= 0)
 685     error (r, _("Invalid number of variables %d."), r->n_vars);
 686
 687   if (match (r, '5'))
 688     read_int (r);
 689
 690   if (match (r, '6'))
 691     {
 692       weight_name = read_pool_string (r);
 693       if (strlen (weight_name) > SHORT_NAME_LEN)
 694         error (r, _("Weight variable name (%s) truncated."), weight_name);
 695     }
 696
 697   for (i = 0; i < r->n_vars; i++)
 698     {
 699       int width;
 700       char name[256];
 701       int fmt[6];
 702       struct variable *v;
 703       struct missing_values miss;
 704       struct fmt_spec print, write;
 705       bool report_error = true;
 706       int j;
 707
 708       if (!match (r, '7'))
 709         error (r, _("Expected variable record."));
 710
 711       width = read_int (r);
 712       if (width < 0)
 713         error (r, _("Invalid variable width %d."), width);
 714
 715       read_string (r, name);
 716       for (j = 0; j < 6; j++)
 717         fmt[j] = read_int (r);
 718
 719       if (!dict_id_is_valid (dict, name, false)
 720           || *name == '#' || *name == '$')
 721         error (r, _("Invalid variable name `%s' in position %d."), name, i);
 722       str_uppercase (name);
 723
 724       if (width < 0 || width > 255)
 725         error (r, _("Bad width %d for variable %s."), width, name);
 726
 727       v = dict_create_var (dict, name, width);
 728       if (v == NULL)
 729         {
 730           unsigned long int i;
 731           for (i = 1; ; i++)
 732             {
 733               char *try_name = xasprintf ("%s_%lu", name, i);
 734               v = dict_create_var (dict, try_name, width);
 735               free (try_name);
 736               if (v != NULL)
 737                 break;
 738             }
 739           warning (r, _("Duplicate variable name %s in position %d renamed "
 740                         "to %s."), name, i, var_get_name (v));
 741         }
 742
 743       print = convert_format (r, &fmt[0], v, &report_error);
 744       write = convert_format (r, &fmt[3], v, &report_error);
 745       var_set_print_format (v, &print);
 746       var_set_write_format (v, &write);
 747
 748       /* Range missing values. */
 749       mv_init (&miss, width);
 750       if (match (r, 'B'))
 751         {
 752           double x = read_float (r);
 753           double y = read_float (r);
 754           mv_add_range (&miss, x, y);
 755         }
 756       else if (match (r, 'A'))
 757         mv_add_range (&miss, read_float (r), HIGHEST);
 758       else if (match (r, '9'))
 759         mv_add_range (&miss, LOWEST, read_float (r));
 760
 761       /* Single missing values. */
 762       while (match (r, '8'))
 763         {
 764           int mv_width = MIN (width, 8);
 765           union value value;
 766
 767           parse_value (r, mv_width, &value);
 768           value_resize (&value, mv_width, width);
 769           mv_add_value (&miss, &value);
 770           value_destroy (&value, width);
 771         }
 772
 773       var_set_missing_values (v, &miss);
 774       mv_destroy (&miss);
 775
 776       if (match (r, 'C'))
 777         {
 778           char label[256];
 779           read_string (r, label);
 780           var_set_label (v, label); /* XXX */
 781         }
 782     }
 783
 784   if (weight_name != NULL)
 785     {
 786       struct variable *weight_var = dict_lookup_var (dict, weight_name);
 787       if (weight_var == NULL)
 788         error (r, _("Weighting variable %s not present in dictionary."),
 789                weight_name);
 790
 791       dict_set_weight (dict, weight_var);
 792     }
 793 }
 794
 795 /* Parse a value of with WIDTH into value V. */
 796 static void
 797 parse_value (struct pfm_reader *r, int width, union value *v)
 798 {
 799   value_init (v, width);
 800   if (width > 0)
 801     {
 802       uint8_t buf[256];
 803       size_t n_bytes = read_bytes (r, buf);
 804       value_copy_buf_rpad (v, width, buf, n_bytes, ' ');
 805     }
 806   else
 807     v->f = read_float (r);
 808 }
 809
 810 /* Parse a value label record and return success. */
 811 static void
 812 read_value_label (struct pfm_reader *r, struct dictionary *dict)
 813 {
 814   /* Variables. */
 815   int nv;
 816   struct variable **v;
 817
 818   /* Labels. */
 819   int n_labels;
 820
 821   int i;
 822
 823   nv = read_int (r);
 824   v = pool_nalloc (r->pool, nv, sizeof *v);
 825   for (i = 0; i < nv; i++)
 826     {
 827       char name[256];
 828       read_string (r, name);
 829
 830       v[i] = dict_lookup_var (dict, name);
 831       if (v[i] == NULL)
 832         error (r, _("Unknown variable %s while parsing value labels."), name);
 833
 834       if (var_get_type (v[0]) != var_get_type (v[i]))
 835         error (r, _("Cannot assign value labels to %s and %s, which "
 836                     "have different variable types."),
 837                var_get_name (v[0]), var_get_name (v[i]));
 838     }
 839
 840   n_labels = read_int (r);
 841   for (i = 0; i < n_labels; i++)
 842     {
 843       union value val;
 844       char label[256];
 845       int j;
 846
 847       parse_value (r, var_get_width (v[0]), &val);
 848       read_string (r, label);
 849
 850       /* Assign the value label to each variable. */
 851       for (j = 0; j < nv; j++)
 852         var_replace_value_label (v[j], &val, label);
 853
 854       value_destroy (&val, var_get_width (v[0]));
 855     }
 856 }
 857
 858 /* Reads a set of documents from portable file R into DICT. */
 859 static void
 860 read_documents (struct pfm_reader *r, struct dictionary *dict)
 861 {
 862   int n_lines = read_int (r);
 863   for (int i = 0; i < n_lines; i++)
 864     {
 865       char line[256];
 866       read_string (r, line);
 867       dict_add_document_line (dict, line, false);
 868     }
 869 }
 870
 871 /* Reads and returns one case from portable file R.  Returns a
 872    null pointer on failure. */
 873 static struct ccase *
 874 por_file_casereader_read (struct casereader *reader, void *r_)
 875 {
 876   struct pfm_reader *r = r_;
 877   struct ccase *volatile c;
 878   size_t i;
 879
 880   c = case_create (r->proto);
 881   setjmp (r->bail_out);
 882   if (!r->ok)
 883     {
 884       casereader_force_error (reader);
 885       case_unref (c);
 886       return NULL;
 887     }
 888
 889   /* Check for end of file. */
 890   if (r->cc == 'Z')
 891     {
 892       case_unref (c);
 893       return NULL;
 894     }
 895
 896   for (i = 0; i < r->n_vars; i++)
 897     {
 898       int width = caseproto_get_width (r->proto, i);
 899
 900       if (width == 0)
 901         *case_num_rw_idx (c, i) = read_float (r);
 902       else
 903         {
 904           uint8_t buf[256];
 905           size_t n_bytes = read_bytes (r, buf);
 906           u8_buf_copy_rpad (case_str_rw_idx (c, i), width, buf, n_bytes, ' ');
 907         }
 908     }
 909
 910   return c;
 911 }
 912
 913 /* Detects whether FILE is an SPSS portable file.  Returns 1 if so, 0 if not,
 914    and a negative errno value if there is an error reading FILE. */
 915 static int
 916 pfm_detect (FILE *file)
 917 {
 918   unsigned char header[464];
 919   char trans[256];
 920   int n_cooked, n_raws, line_len;
 921   int i;
 922
 923   n_cooked = n_raws = 0;
 924   line_len = 0;
 925   while (n_cooked < sizeof header)
 926     {
 927       int c = getc (file);
 928       if (c == EOF || n_raws++ > 512)
 929         return ferror (file) ? -errno : 0;
 930       else if (c == '\n')
 931         {
 932           while (line_len < 80 && n_cooked < sizeof header)
 933             {
 934               header[n_cooked++] = ' ';
 935               line_len++;
 936             }
 937           line_len = 0;
 938         }
 939       else if (c != '\r')
 940         {
 941           header[n_cooked++] = c;
 942           line_len++;
 943         }
 944     }
 945
 946   memset (trans, 0, 256);
 947   for (i = 64; i < 256; i++)
 948     {
 949       unsigned char c = header[i + 200];
 950       if (trans[c] == 0)
 951         trans[c] = portable_to_local[i];
 952     }
 953
 954   for (i = 0; i < 8; i++)
 955     if (trans[header[i + 456]] != "SPSSPORT"[i])
 956       return 0;
 957
 958   return 1;
 959 }
 960
 961 static const struct casereader_class por_file_casereader_class =
 962   {
 963     por_file_casereader_read,
 964     por_file_casereader_destroy,
 965     NULL,
 966     NULL,
 967   };
 968
 969 const struct any_reader_class por_file_reader_class =
 970   {
 971     N_("SPSS Portable File"),
 972     pfm_detect,
 973     pfm_open,
 974     pfm_close,
 975     pfm_decode,
 976     NULL,                       /* get_strings */
 977   };