1 /* PSPP - a program for statistical analysis.
2 Copyright (C) 1997-2000, 2006-2007, 2009-2016 Free Software Foundation, Inc.
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.
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.
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/>. */
19 #include "data/sys-file-private.h"
28 #include "data/any-reader.h"
29 #include "data/attributes.h"
30 #include "data/case.h"
31 #include "data/casereader-provider.h"
32 #include "data/casereader.h"
33 #include "data/dictionary.h"
34 #include "data/file-handle-def.h"
35 #include "data/file-name.h"
36 #include "data/format.h"
37 #include "data/identifier.h"
38 #include "data/missing-values.h"
39 #include "data/mrset.h"
40 #include "data/short-names.h"
41 #include "data/value-labels.h"
42 #include "data/value.h"
43 #include "data/variable.h"
44 #include "libpspp/array.h"
45 #include "libpspp/assertion.h"
46 #include "libpspp/compiler.h"
47 #include "libpspp/i18n.h"
48 #include "libpspp/ll.h"
49 #include "libpspp/message.h"
50 #include "libpspp/misc.h"
51 #include "libpspp/pool.h"
52 #include "libpspp/str.h"
53 #include "libpspp/stringi-set.h"
55 #include "gl/c-strtod.h"
56 #include "gl/c-ctype.h"
57 #include "gl/inttostr.h"
58 #include "gl/localcharset.h"
59 #include "gl/minmax.h"
60 #include "gl/unlocked-io.h"
61 #include "gl/xalloc.h"
62 #include "gl/xalloc-oversized.h"
66 #define _(msgid) gettext (msgid)
67 #define N_(msgid) (msgid)
71 /* subtypes 0-2 unknown */
72 EXT_INTEGER = 3, /* Machine integer info. */
73 EXT_FLOAT = 4, /* Machine floating-point info. */
74 EXT_VAR_SETS = 5, /* Variable sets. */
75 EXT_DATE = 6, /* DATE. */
76 EXT_MRSETS = 7, /* Multiple response sets. */
77 EXT_DATA_ENTRY = 8, /* SPSS Data Entry. */
78 /* subtype 9 unknown */
79 EXT_PRODUCT_INFO = 10, /* Extra product info text. */
80 EXT_DISPLAY = 11, /* Variable display parameters. */
81 /* subtype 12 unknown */
82 EXT_LONG_NAMES = 13, /* Long variable names. */
83 EXT_LONG_STRINGS = 14, /* Long strings. */
84 /* subtype 15 unknown */
85 EXT_NCASES = 16, /* Extended number of cases. */
86 EXT_FILE_ATTRS = 17, /* Data file attributes. */
87 EXT_VAR_ATTRS = 18, /* Variable attributes. */
88 EXT_MRSETS2 = 19, /* Multiple response sets (extended). */
89 EXT_ENCODING = 20, /* Character encoding. */
90 EXT_LONG_LABELS = 21, /* Value labels for long strings. */
91 EXT_LONG_MISSING = 22, /* Missing values for long strings. */
92 EXT_DATAVIEW = 24 /* "Format properties in dataview table". */
95 /* Fields from the top-level header record. */
96 struct sfm_header_record
98 char magic[5]; /* First 4 bytes of file, then null. */
99 int weight_idx; /* 0 if unweighted, otherwise a var index. */
100 int nominal_case_size; /* Number of var positions. */
102 /* These correspond to the members of struct any_file_info or a dictionary
103 but in the system file's encoding rather than ASCII. */
104 char creation_date[10]; /* "dd mmm yy". */
105 char creation_time[9]; /* "hh:mm:ss". */
106 char eye_catcher[61]; /* Eye-catcher string, then product name. */
107 char file_label[65]; /* File label. */
110 struct sfm_var_record
117 int missing_value_code;
120 struct variable *var;
123 struct sfm_value_label
129 struct sfm_value_label_record
132 struct sfm_value_label *labels;
133 unsigned int n_labels;
139 struct sfm_document_record
148 const char *name; /* Name. */
149 const char *label; /* Human-readable label for group. */
150 enum mrset_type type; /* Group type. */
151 const char **vars; /* Constituent variables' names. */
152 size_t n_vars; /* Number of constituent variables. */
155 enum mrset_md_cat_source cat_source; /* Source of category labels. */
156 bool label_from_var_label; /* 'label' taken from variable label? */
157 const char *counted; /* Counted value, as string. */
160 struct sfm_extension_record
162 struct ll ll; /* In struct sfm_reader 'var_attrs' list. */
163 int subtype; /* Record subtype. */
164 off_t pos; /* Starting offset in file. */
165 unsigned int size; /* Size of data elements. */
166 unsigned int count; /* Number of data elements. */
167 void *data; /* Contents. */
170 /* System file reader. */
173 struct any_reader any_reader;
175 /* Resource tracking. */
176 struct pool *pool; /* All system file state. */
179 struct any_read_info info;
180 struct sfm_header_record header;
181 struct sfm_var_record *vars;
183 struct sfm_value_label_record *labels;
185 struct sfm_document_record *document;
186 struct sfm_mrset *mrsets;
188 struct sfm_extension_record *extensions[32];
189 struct ll_list var_attrs; /* Contains "struct sfm_extension_record"s. */
192 struct file_handle *fh; /* File handle. */
193 struct fh_lock *lock; /* Mutual exclusion for file handle. */
194 FILE *file; /* File stream. */
195 off_t pos; /* Position in file. */
196 bool error; /* I/O or corruption error? */
197 struct caseproto *proto; /* Format of output cases. */
200 enum integer_format integer_format; /* On-disk integer format. */
201 enum float_format float_format; /* On-disk floating point format. */
202 struct sfm_var *sfm_vars; /* Variables. */
203 size_t sfm_var_cnt; /* Number of variables. */
204 int case_cnt; /* Number of cases */
205 const char *encoding; /* String encoding. */
206 bool written_by_readstat; /* From https://github.com/WizardMac/ReadStat? */
209 enum any_compression compression;
210 double bias; /* Compression bias, usually 100.0. */
211 uint8_t opcodes[8]; /* Current block of opcodes. */
212 size_t opcode_idx; /* Next opcode to interpret, 8 if none left. */
213 bool corruption_warning; /* Warned about possible corruption? */
215 /* ZLIB decompression. */
216 long long int ztrailer_ofs; /* Offset of ZLIB trailer at end of file. */
217 #define ZIN_BUF_SIZE 4096
218 uint8_t *zin_buf; /* Inflation input buffer. */
219 #define ZOUT_BUF_SIZE 16384
220 uint8_t *zout_buf; /* Inflation output buffer. */
221 unsigned int zout_end; /* Number of bytes of data in zout_buf. */
222 unsigned int zout_pos; /* First unconsumed byte in zout_buf. */
223 z_stream zstream; /* ZLIB inflater. */
226 static const struct casereader_class sys_file_casereader_class;
228 static struct sfm_reader *
229 sfm_reader_cast (const struct any_reader *r_)
231 assert (r_->klass == &sys_file_reader_class);
232 return UP_CAST (r_, struct sfm_reader, any_reader);
235 static bool sfm_close (struct any_reader *);
237 static void sys_msg (struct sfm_reader *r, off_t, int class,
238 const char *format, va_list args)
239 PRINTF_FORMAT (4, 0);
240 static void sys_warn (struct sfm_reader *, off_t, const char *, ...)
241 PRINTF_FORMAT (3, 4);
242 static void sys_error (struct sfm_reader *, off_t, const char *, ...)
243 PRINTF_FORMAT (3, 4);
245 static bool read_bytes (struct sfm_reader *, void *, size_t)
247 static int try_read_bytes (struct sfm_reader *, void *, size_t)
249 static bool read_int (struct sfm_reader *, int *) WARN_UNUSED_RESULT;
250 static bool read_uint (struct sfm_reader *, unsigned int *) WARN_UNUSED_RESULT;
251 static bool read_int64 (struct sfm_reader *, long long int *)
253 static bool read_uint64 (struct sfm_reader *, unsigned long long int *)
255 static bool read_string (struct sfm_reader *, char *, size_t)
257 static bool skip_bytes (struct sfm_reader *, size_t) WARN_UNUSED_RESULT;
259 /* ZLIB compressed data handling. */
260 static bool read_zheader (struct sfm_reader *) WARN_UNUSED_RESULT;
261 static bool open_zstream (struct sfm_reader *) WARN_UNUSED_RESULT;
262 static bool close_zstream (struct sfm_reader *) WARN_UNUSED_RESULT;
263 static int read_bytes_zlib (struct sfm_reader *, void *, size_t)
265 static int read_compressed_bytes (struct sfm_reader *, void *, size_t)
267 static int try_read_compressed_bytes (struct sfm_reader *, void *, size_t)
269 static bool read_compressed_float (struct sfm_reader *, double *)
272 static char *fix_line_ends (const char *);
274 static int parse_int (const struct sfm_reader *, const void *data, size_t ofs);
275 static double parse_float (const struct sfm_reader *,
276 const void *data, size_t ofs);
278 static bool read_variable_record (struct sfm_reader *,
279 struct sfm_var_record *);
280 static bool read_value_label_record (struct sfm_reader *,
281 struct sfm_value_label_record *);
282 static bool read_document_record (struct sfm_reader *);
283 static bool read_extension_record (struct sfm_reader *, int subtype,
284 struct sfm_extension_record **);
285 static bool skip_extension_record (struct sfm_reader *, int subtype);
287 static struct text_record *open_text_record (
288 struct sfm_reader *, const struct sfm_extension_record *,
289 bool recode_to_utf8);
290 static void close_text_record (struct sfm_reader *,
291 struct text_record *);
292 static bool read_variable_to_value_pair (struct sfm_reader *,
294 struct text_record *,
295 struct variable **var, char **value);
296 static void text_warn (struct sfm_reader *r, struct text_record *text,
297 const char *format, ...) PRINTF_FORMAT (3, 4);
298 static char *text_get_token (struct text_record *,
299 struct substring delimiters, char *delimiter);
300 static bool text_match (struct text_record *, char c);
301 static bool text_read_variable_name (struct sfm_reader *, struct dictionary *,
302 struct text_record *,
303 struct substring delimiters,
305 static bool text_read_short_name (struct sfm_reader *, struct dictionary *,
306 struct text_record *,
307 struct substring delimiters,
309 static const char *text_parse_counted_string (struct sfm_reader *,
310 struct text_record *);
311 static size_t text_pos (const struct text_record *);
312 static const char *text_get_all (const struct text_record *);
314 /* Dictionary reader. */
322 static bool read_dictionary (struct sfm_reader *);
323 static bool read_record (struct sfm_reader *, int type,
324 size_t *allocated_vars, size_t *allocated_labels);
325 static bool read_header (struct sfm_reader *, struct any_read_info *,
326 struct sfm_header_record *);
327 static void parse_header (struct sfm_reader *,
328 const struct sfm_header_record *,
329 struct any_read_info *, struct dictionary *);
330 static bool parse_variable_records (struct sfm_reader *, struct dictionary *,
331 struct sfm_var_record *, size_t n);
332 static void parse_format_spec (struct sfm_reader *, off_t pos,
333 unsigned int format, enum which_format,
334 struct variable *, int *format_warning_cnt);
335 static void parse_document (struct dictionary *, struct sfm_document_record *);
336 static void parse_display_parameters (struct sfm_reader *,
337 const struct sfm_extension_record *,
338 struct dictionary *);
339 static bool parse_machine_integer_info (struct sfm_reader *,
340 const struct sfm_extension_record *,
341 struct any_read_info *);
342 static void parse_machine_float_info (struct sfm_reader *,
343 const struct sfm_extension_record *);
344 static void parse_extra_product_info (struct sfm_reader *,
345 const struct sfm_extension_record *,
346 struct any_read_info *);
347 static void parse_mrsets (struct sfm_reader *,
348 const struct sfm_extension_record *,
349 size_t *allocated_mrsets);
350 static void decode_mrsets (struct sfm_reader *, struct dictionary *);
351 static void parse_long_var_name_map (struct sfm_reader *,
352 const struct sfm_extension_record *,
353 struct dictionary *);
354 static bool parse_long_string_map (struct sfm_reader *,
355 const struct sfm_extension_record *,
356 struct dictionary *);
357 static void parse_value_labels (struct sfm_reader *, struct dictionary *);
358 static struct variable *parse_weight_var (struct sfm_reader *,
359 const struct sfm_var_record *, size_t n_var_recs,
361 static void parse_data_file_attributes (struct sfm_reader *,
362 const struct sfm_extension_record *,
363 struct dictionary *);
364 static void parse_variable_attributes (struct sfm_reader *,
365 const struct sfm_extension_record *,
366 struct dictionary *);
367 static void assign_variable_roles (struct sfm_reader *, struct dictionary *);
368 static void parse_long_string_value_labels (struct sfm_reader *,
369 const struct sfm_extension_record *,
370 struct dictionary *);
371 static void parse_long_string_missing_values (
372 struct sfm_reader *, const struct sfm_extension_record *,
373 struct dictionary *);
375 /* Frees the strings inside INFO. */
377 any_read_info_destroy (struct any_read_info *info)
381 free (info->creation_date);
382 free (info->creation_time);
383 free (info->product);
384 free (info->product_ext);
388 /* Tries to open FH for reading as a system file. Returns an sfm_reader if
389 successful, otherwise NULL. */
390 static struct any_reader *
391 sfm_open (struct file_handle *fh)
393 size_t allocated_mrsets = 0;
394 struct sfm_reader *r;
396 /* Create and initialize reader. */
397 r = xzalloc (sizeof *r);
398 r->any_reader.klass = &sys_file_reader_class;
399 r->pool = pool_create ();
400 pool_register (r->pool, free, r);
402 r->opcode_idx = sizeof r->opcodes;
403 ll_init (&r->var_attrs);
405 /* TRANSLATORS: this fragment will be interpolated into
406 messages in fh_lock() that identify types of files. */
407 r->lock = fh_lock (fh, FH_REF_FILE, N_("system file"), FH_ACC_READ, false);
411 r->file = fn_open (fh, "rb");
414 msg (ME, _("Error opening `%s' for reading as a system file: %s."),
415 fh_get_file_name (r->fh), strerror (errno));
419 if (!read_dictionary (r))
422 if (r->extensions[EXT_MRSETS] != NULL)
423 parse_mrsets (r, r->extensions[EXT_MRSETS], &allocated_mrsets);
425 if (r->extensions[EXT_MRSETS2] != NULL)
426 parse_mrsets (r, r->extensions[EXT_MRSETS2], &allocated_mrsets);
428 return &r->any_reader;
432 sfm_close (&r->any_reader);
437 read_dictionary (struct sfm_reader *r)
439 size_t allocated_vars;
440 size_t allocated_labels;
442 if (!read_header (r, &r->info, &r->header))
446 allocated_labels = 0;
451 if (!read_int (r, &type))
455 if (!read_record (r, type, &allocated_vars, &allocated_labels))
459 if (!skip_bytes (r, 4))
462 if (r->compression == ANY_COMP_ZLIB && !read_zheader (r))
469 read_record (struct sfm_reader *r, int type,
470 size_t *allocated_vars, size_t *allocated_labels)
477 if (r->n_vars >= *allocated_vars)
478 r->vars = pool_2nrealloc (r->pool, r->vars, allocated_vars,
480 return read_variable_record (r, &r->vars[r->n_vars++]);
483 if (r->n_labels >= *allocated_labels)
484 r->labels = pool_2nrealloc (r->pool, r->labels, allocated_labels,
486 return read_value_label_record (r, &r->labels[r->n_labels++]);
489 /* A Type 4 record is always immediately after a type 3 record,
490 so the code for type 3 records reads the type 4 record too. */
491 sys_error (r, r->pos, _("Misplaced type 4 record."));
495 if (r->document != NULL)
496 sys_warn (r, r->pos, _("Duplicate type 6 (document) record."));
497 return read_document_record (r);
500 if (!read_int (r, &subtype))
503 || subtype >= sizeof r->extensions / sizeof *r->extensions)
506 _("Unrecognized record type 7, subtype %d. For help, "
507 "please send this file to %s and mention that you were "
509 subtype, PACKAGE_BUGREPORT, PACKAGE_STRING);
510 return skip_extension_record (r, subtype);
512 else if (subtype == 18)
514 /* System files written by "Stata 14.1/-savespss- 1.77 by S.Radyakin"
515 put each variable attribute into a separate record with subtype
516 18. I'm surprised that SPSS puts up with this. */
517 struct sfm_extension_record *ext;
518 bool ok = read_extension_record (r, subtype, &ext);
520 ll_push_tail (&r->var_attrs, &ext->ll);
523 else if (r->extensions[subtype] != NULL)
526 _("Record type 7, subtype %d found here has the same "
527 "type as the record found near offset 0x%llx. For "
528 "help, please send this file to %s and mention that "
529 "you were using %s."),
530 subtype, (long long int) r->extensions[subtype]->pos,
531 PACKAGE_BUGREPORT, PACKAGE_STRING);
532 return skip_extension_record (r, subtype);
535 return read_extension_record (r, subtype, &r->extensions[subtype]);
538 sys_error (r, r->pos, _("Unrecognized record type %d."), type);
545 /* Returns the character encoding obtained from R, or a null pointer if R
546 doesn't have an indication of its character encoding. */
548 sfm_get_encoding (const struct sfm_reader *r)
550 /* The EXT_ENCODING record is the best way to determine dictionary
552 if (r->extensions[EXT_ENCODING])
553 return r->extensions[EXT_ENCODING]->data;
555 /* But EXT_INTEGER is better than nothing as a fallback. */
556 if (r->extensions[EXT_INTEGER])
558 int codepage = parse_int (r, r->extensions[EXT_INTEGER]->data, 7 * 4);
559 const char *encoding;
568 /* These ostensibly mean "7-bit ASCII" and "8-bit ASCII"[sic]
569 respectively. However, many files have character code 2 but data
570 which are clearly not ASCII. Therefore, ignore these values. */
577 encoding = sys_get_encoding_from_codepage (codepage);
578 if (encoding != NULL)
584 /* If the file magic number is EBCDIC then its character data is too. */
585 if (!strcmp (r->header.magic, EBCDIC_MAGIC))
591 struct get_strings_aux
602 add_string__ (struct get_strings_aux *aux,
603 const char *string, bool id, char *title)
605 if (aux->n >= aux->allocated)
607 aux->allocated = 2 * (aux->allocated + 1);
608 aux->titles = pool_realloc (aux->pool, aux->titles,
609 aux->allocated * sizeof *aux->titles);
610 aux->strings = pool_realloc (aux->pool, aux->strings,
611 aux->allocated * sizeof *aux->strings);
612 aux->ids = pool_realloc (aux->pool, aux->ids,
613 aux->allocated * sizeof *aux->ids);
616 aux->titles[aux->n] = title;
617 aux->strings[aux->n] = pool_strdup (aux->pool, string);
618 aux->ids[aux->n] = id;
622 static void PRINTF_FORMAT (3, 4)
623 add_string (struct get_strings_aux *aux,
624 const char *string, const char *title, ...)
628 va_start (args, title);
629 add_string__ (aux, string, false, pool_vasprintf (aux->pool, title, args));
633 static void PRINTF_FORMAT (3, 4)
634 add_id (struct get_strings_aux *aux, const char *id, const char *title, ...)
638 va_start (args, title);
639 add_string__ (aux, id, true, pool_vasprintf (aux->pool, title, args));
643 /* Retrieves significant string data from R in its raw format, to allow the
644 caller to try to detect the encoding in use.
646 Returns the number of strings retrieved N. Sets each of *TITLESP, *IDSP,
647 and *STRINGSP to an array of N elements allocated from POOL. For each I in
648 0...N-1, UTF-8 string *TITLESP[I] describes *STRINGSP[I], which is in
649 whatever encoding system file R uses. *IDS[I] is true if *STRINGSP[I] must
650 be a valid PSPP language identifier, false if *STRINGSP[I] is free-form
653 sfm_get_strings (const struct any_reader *r_, struct pool *pool,
654 char ***titlesp, bool **idsp, char ***stringsp)
656 struct sfm_reader *r = sfm_reader_cast (r_);
657 const struct sfm_mrset *mrset;
658 struct get_strings_aux aux;
670 for (i = 0; i < r->n_vars; i++)
671 if (r->vars[i].width != -1)
672 add_id (&aux, r->vars[i].name, _("Variable %zu"), ++var_idx);
675 for (i = 0; i < r->n_vars; i++)
676 if (r->vars[i].width != -1)
679 if (r->vars[i].label)
680 add_string (&aux, r->vars[i].label, _("Variable %zu Label"),
685 for (i = 0; i < r->n_labels; i++)
686 for (j = 0; j < r->labels[i].n_labels; j++)
687 add_string (&aux, r->labels[i].labels[j].label,
688 _("Value Label %zu"), k++);
690 add_string (&aux, r->header.creation_date, _("Creation Date"));
691 add_string (&aux, r->header.creation_time, _("Creation Time"));
692 add_string (&aux, r->header.eye_catcher, _("Product"));
693 add_string (&aux, r->header.file_label, _("File Label"));
695 if (r->extensions[EXT_PRODUCT_INFO])
696 add_string (&aux, r->extensions[EXT_PRODUCT_INFO]->data,
697 _("Extra Product Info"));
703 for (i = 0; i < r->document->n_lines; i++)
707 memcpy (line, r->document->documents + i * 80, 80);
710 add_string (&aux, line, _("Document Line %zu"), i + 1);
714 for (mrset = r->mrsets; mrset < &r->mrsets[r->n_mrsets]; mrset++)
716 size_t mrset_idx = mrset - r->mrsets + 1;
718 add_id (&aux, mrset->name, _("MRSET %zu"), mrset_idx);
720 add_string (&aux, mrset->label, _("MRSET %zu Label"), mrset_idx);
722 /* Skip the variables because they ought to be duplicates. */
725 add_string (&aux, mrset->counted, _("MRSET %zu Counted Value"),
729 /* data file attributes */
730 /* variable attributes */
732 /* long string value labels */
733 /* long string missing values */
735 *titlesp = aux.titles;
737 *stringsp = aux.strings;
741 /* Decodes the dictionary read from R, saving it into into *DICT. Character
742 strings in R are decoded using ENCODING, or an encoding obtained from R if
743 ENCODING is null, or the locale encoding if R specifies no encoding.
745 If INFOP is non-null, then it receives additional info about the system
746 file, which the caller must eventually free with any_read_info_destroy()
747 when it is no longer needed.
749 This function consumes R. The caller must use it again later, even to
750 destroy it with sfm_close(). */
751 static struct casereader *
752 sfm_decode (struct any_reader *r_, const char *encoding,
753 struct dictionary **dictp, struct any_read_info *infop)
755 struct sfm_reader *r = sfm_reader_cast (r_);
756 struct dictionary *dict;
758 if (encoding == NULL)
760 encoding = sfm_get_encoding (r);
761 if (encoding == NULL)
763 sys_warn (r, -1, _("This system file does not indicate its own "
764 "character encoding. Using default encoding "
765 "%s. For best results, specify an encoding "
766 "explicitly. Use SYSFILE INFO with "
767 "ENCODING=\"DETECT\" to analyze the possible "
770 encoding = locale_charset ();
774 dict = dict_create (encoding);
775 r->encoding = dict_get_encoding (dict);
777 /* These records don't use variables at all. */
778 if (r->document != NULL)
779 parse_document (dict, r->document);
781 if (r->extensions[EXT_INTEGER] != NULL
782 && !parse_machine_integer_info (r, r->extensions[EXT_INTEGER], &r->info))
785 if (r->extensions[EXT_FLOAT] != NULL)
786 parse_machine_float_info (r, r->extensions[EXT_FLOAT]);
788 if (r->extensions[EXT_PRODUCT_INFO] != NULL)
789 parse_extra_product_info (r, r->extensions[EXT_PRODUCT_INFO], &r->info);
791 if (r->extensions[EXT_FILE_ATTRS] != NULL)
792 parse_data_file_attributes (r, r->extensions[EXT_FILE_ATTRS], dict);
794 parse_header (r, &r->header, &r->info, dict);
796 /* Parse the variable records, the basis of almost everything else. */
797 if (!parse_variable_records (r, dict, r->vars, r->n_vars))
800 /* Parse value labels and the weight variable immediately after the variable
801 records. These records use indexes into var_recs[], so we must parse them
802 before those indexes become invalidated by very long string variables. */
803 parse_value_labels (r, dict);
804 if (r->header.weight_idx != 0)
805 dict_set_weight (dict, parse_weight_var (r, r->vars, r->n_vars,
806 r->header.weight_idx));
808 if (r->extensions[EXT_DISPLAY] != NULL)
809 parse_display_parameters (r, r->extensions[EXT_DISPLAY], dict);
811 /* The following records use short names, so they need to be parsed before
812 parse_long_var_name_map() changes short names to long names. */
813 decode_mrsets (r, dict);
815 if (r->extensions[EXT_LONG_STRINGS] != NULL
816 && !parse_long_string_map (r, r->extensions[EXT_LONG_STRINGS], dict))
819 /* Now rename variables to their long names. */
820 parse_long_var_name_map (r, r->extensions[EXT_LONG_NAMES], dict);
822 /* The following records use long names, so they need to follow renaming. */
823 if (!ll_is_empty (&r->var_attrs))
825 struct sfm_extension_record *ext;
826 ll_for_each (ext, struct sfm_extension_record, ll, &r->var_attrs)
827 parse_variable_attributes (r, ext, dict);
829 /* Roles use the $@Role attribute. */
830 assign_variable_roles (r, dict);
832 if (r->extensions[EXT_LONG_LABELS] != NULL)
833 parse_long_string_value_labels (r, r->extensions[EXT_LONG_LABELS], dict);
834 if (r->extensions[EXT_LONG_MISSING] != NULL)
835 parse_long_string_missing_values (r, r->extensions[EXT_LONG_MISSING],
838 /* Warn if the actual amount of data per case differs from the
839 amount that the header claims. SPSS version 13 gets this
840 wrong when very long strings are involved, so don't warn in
842 if (r->header.nominal_case_size > 0
843 && r->header.nominal_case_size != r->n_vars
844 && r->info.version_major != 13)
845 sys_warn (r, -1, _("File header claims %d variable positions but "
846 "%zu were read from file."),
847 r->header.nominal_case_size, r->n_vars);
849 /* Create an index of dictionary variable widths for
850 sfm_read_case to use. We cannot use the `struct variable's
851 from the dictionary we created, because the caller owns the
852 dictionary and may destroy or modify its variables. */
853 sfm_dictionary_to_sfm_vars (dict, &r->sfm_vars, &r->sfm_var_cnt);
854 pool_register (r->pool, free, r->sfm_vars);
855 r->proto = caseproto_ref_pool (dict_get_proto (dict), r->pool);
861 memset (&r->info, 0, sizeof r->info);
864 return casereader_create_sequential
866 r->case_cnt == -1 ? CASENUMBER_MAX: r->case_cnt,
867 &sys_file_casereader_class, r);
876 /* Closes R, which should have been returned by sfm_open() but not already
877 closed with sfm_decode() or this function.
878 Returns true if an I/O error has occurred on READER, false
881 sfm_close (struct any_reader *r_)
883 struct sfm_reader *r = sfm_reader_cast (r_);
888 if (fn_close (r->fh, r->file) == EOF)
890 msg (ME, _("Error closing system file `%s': %s."),
891 fh_get_file_name (r->fh), strerror (errno));
897 any_read_info_destroy (&r->info);
902 pool_destroy (r->pool);
907 /* Destroys READER. */
909 sys_file_casereader_destroy (struct casereader *reader UNUSED, void *r_)
911 struct sfm_reader *r = r_;
912 sfm_close (&r->any_reader);
915 /* Detects whether FILE is an SPSS system file. Returns 1 if so, 0 if not, and
916 a negative errno value if there is an error reading FILE. */
918 sfm_detect (FILE *file)
922 if (fseek (file, 0, SEEK_SET) != 0)
924 if (fread (magic, 4, 1, file) != 1)
925 return ferror (file) ? -errno : 0;
928 return (!strcmp (ASCII_MAGIC, magic)
929 || !strcmp (ASCII_ZMAGIC, magic)
930 || !strcmp (EBCDIC_MAGIC, magic));
933 /* Reads the global header of the system file. Initializes *HEADER and *INFO,
934 except for the string fields in *INFO, which parse_header() will initialize
935 later once the file's encoding is known. */
937 read_header (struct sfm_reader *r, struct any_read_info *info,
938 struct sfm_header_record *header)
940 uint8_t raw_layout_code[4];
945 if (!read_string (r, header->magic, sizeof header->magic)
946 || !read_string (r, header->eye_catcher, sizeof header->eye_catcher))
948 r->written_by_readstat = strstr (header->eye_catcher,
949 "https://github.com/WizardMac/ReadStat");
951 if (!strcmp (ASCII_MAGIC, header->magic)
952 || !strcmp (EBCDIC_MAGIC, header->magic))
954 else if (!strcmp (ASCII_ZMAGIC, header->magic))
958 sys_error (r, 0, _("This is not an SPSS system file."));
962 /* Identify integer format. */
963 if (!read_bytes (r, raw_layout_code, sizeof raw_layout_code))
965 if ((!integer_identify (2, raw_layout_code, sizeof raw_layout_code,
967 && !integer_identify (3, raw_layout_code, sizeof raw_layout_code,
969 || (r->integer_format != INTEGER_MSB_FIRST
970 && r->integer_format != INTEGER_LSB_FIRST))
972 sys_error (r, 64, _("This is not an SPSS system file."));
976 if (!read_int (r, &header->nominal_case_size))
979 if (header->nominal_case_size < 0
980 || header->nominal_case_size > INT_MAX / 16)
981 header->nominal_case_size = -1;
983 if (!read_int (r, &compressed))
988 r->compression = ANY_COMP_NONE;
989 else if (compressed == 1)
990 r->compression = ANY_COMP_SIMPLE;
991 else if (compressed != 0)
993 sys_error (r, 0, "System file header has invalid compression "
994 "value %d.", compressed);
1000 if (compressed == 2)
1001 r->compression = ANY_COMP_ZLIB;
1004 sys_error (r, 0, "ZLIB-compressed system file header has invalid "
1005 "compression value %d.", compressed);
1010 if (!read_int (r, &header->weight_idx))
1013 if (!read_int (r, &r->case_cnt))
1015 if (r->case_cnt > INT_MAX / 2)
1018 /* Identify floating-point format and obtain compression bias. */
1019 if (!read_bytes (r, raw_bias, sizeof raw_bias))
1021 if (float_identify (100.0, raw_bias, sizeof raw_bias, &r->float_format) == 0)
1023 uint8_t zero_bias[8] = { 0, 0, 0, 0, 0, 0, 0, 0 };
1025 if (memcmp (raw_bias, zero_bias, 8))
1026 sys_warn (r, r->pos - 8,
1027 _("Compression bias is not the usual "
1028 "value of 100, or system file uses unrecognized "
1029 "floating-point format."));
1032 /* Some software is known to write all-zeros to this
1033 field. Such software also writes floating-point
1034 numbers in the format that we expect by default
1035 (it seems that all software most likely does, in
1036 reality), so don't warn in this case. */
1039 if (r->integer_format == INTEGER_MSB_FIRST)
1040 r->float_format = FLOAT_IEEE_DOUBLE_BE;
1042 r->float_format = FLOAT_IEEE_DOUBLE_LE;
1044 float_convert (r->float_format, raw_bias, FLOAT_NATIVE_DOUBLE, &r->bias);
1046 if (!read_string (r, header->creation_date, sizeof header->creation_date)
1047 || !read_string (r, header->creation_time, sizeof header->creation_time)
1048 || !read_string (r, header->file_label, sizeof header->file_label)
1049 || !skip_bytes (r, 3))
1052 info->integer_format = r->integer_format;
1053 info->float_format = r->float_format;
1054 info->compression = r->compression;
1055 info->case_cnt = r->case_cnt;
1060 /* Reads a variable (type 2) record from R into RECORD. */
1062 read_variable_record (struct sfm_reader *r, struct sfm_var_record *record)
1064 int has_variable_label;
1066 memset (record, 0, sizeof *record);
1068 record->pos = r->pos;
1069 if (!read_int (r, &record->width)
1070 || !read_int (r, &has_variable_label)
1071 || !read_int (r, &record->missing_value_code)
1072 || !read_int (r, &record->print_format)
1073 || !read_int (r, &record->write_format)
1074 || !read_string (r, record->name, sizeof record->name))
1077 if (has_variable_label == 1)
1079 enum { MAX_LABEL_LEN = 65536 };
1080 unsigned int len, read_len;
1082 if (!read_uint (r, &len))
1085 /* Read up to MAX_LABEL_LEN bytes of label. */
1086 read_len = MIN (MAX_LABEL_LEN, len);
1087 record->label = pool_malloc (r->pool, read_len + 1);
1088 if (!read_string (r, record->label, read_len + 1))
1091 /* Skip unread label bytes. */
1092 if (!skip_bytes (r, len - read_len))
1095 /* Skip label padding up to multiple of 4 bytes. */
1096 if (!skip_bytes (r, ROUND_UP (len, 4) - len))
1099 else if (has_variable_label != 0)
1101 sys_error (r, record->pos,
1102 _("Variable label indicator field is not 0 or 1."));
1106 /* Set missing values. */
1107 if (record->missing_value_code != 0)
1109 int code = record->missing_value_code;
1110 if (record->width == 0)
1112 if (code < -3 || code > 3 || code == -1)
1114 sys_error (r, record->pos,
1115 _("Numeric missing value indicator field is not "
1116 "-3, -2, 0, 1, 2, or 3."));
1122 if (code < 1 || code > 3)
1124 sys_error (r, record->pos,
1125 _("String missing value indicator field is not "
1131 if (!read_bytes (r, record->missing, 8 * abs (code)))
1138 /* Reads value labels from R into RECORD. */
1140 read_value_label_record (struct sfm_reader *r,
1141 struct sfm_value_label_record *record)
1146 /* Read type 3 record. */
1147 record->pos = r->pos;
1148 if (!read_uint (r, &record->n_labels))
1150 if (record->n_labels > UINT_MAX / sizeof *record->labels)
1152 sys_error (r, r->pos - 4, _("Invalid number of labels %u."),
1156 record->labels = pool_nmalloc (r->pool, record->n_labels,
1157 sizeof *record->labels);
1158 for (i = 0; i < record->n_labels; i++)
1160 struct sfm_value_label *label = &record->labels[i];
1161 unsigned char label_len;
1164 if (!read_bytes (r, label->value, sizeof label->value))
1167 /* Read label length. */
1168 if (!read_bytes (r, &label_len, sizeof label_len))
1170 padded_len = ROUND_UP (label_len + 1, 8);
1172 /* Read label, padding. */
1173 label->label = pool_malloc (r->pool, padded_len + 1);
1174 if (!read_bytes (r, label->label, padded_len - 1))
1176 label->label[label_len] = '\0';
1179 /* Read record type of type 4 record. */
1180 if (!read_int (r, &type))
1184 sys_error (r, r->pos - 4,
1185 _("Variable index record (type 4) does not immediately "
1186 "follow value label record (type 3) as it should."));
1190 /* Read number of variables associated with value label from type 4
1192 if (!read_uint (r, &record->n_vars))
1194 if (record->n_vars < 1 || record->n_vars > r->n_vars)
1196 sys_error (r, r->pos - 4,
1197 _("Number of variables associated with a value label (%u) "
1198 "is not between 1 and the number of variables (%zu)."),
1199 record->n_vars, r->n_vars);
1203 record->vars = pool_nmalloc (r->pool, record->n_vars, sizeof *record->vars);
1204 for (i = 0; i < record->n_vars; i++)
1205 if (!read_int (r, &record->vars[i]))
1211 /* Reads a document record from R. Returns true if successful, false on
1214 read_document_record (struct sfm_reader *r)
1217 if (!read_int (r, &n_lines))
1219 else if (n_lines == 0)
1221 else if (n_lines < 0 || n_lines >= INT_MAX / DOC_LINE_LENGTH)
1223 sys_error (r, r->pos,
1224 _("Number of document lines (%d) "
1225 "must be greater than 0 and less than %d."),
1226 n_lines, INT_MAX / DOC_LINE_LENGTH);
1230 struct sfm_document_record *record;
1231 record = pool_malloc (r->pool, sizeof *record);
1232 record->pos = r->pos;
1233 record->n_lines = n_lines;
1234 record->documents = pool_malloc (r->pool, DOC_LINE_LENGTH * n_lines);
1235 if (!read_bytes (r, record->documents, DOC_LINE_LENGTH * n_lines))
1238 r->document = record;
1243 read_extension_record_header (struct sfm_reader *r, int subtype,
1244 struct sfm_extension_record *record)
1246 record->subtype = subtype;
1247 record->pos = r->pos;
1248 if (!read_uint (r, &record->size) || !read_uint (r, &record->count))
1251 /* Check that SIZE * COUNT + 1 doesn't overflow. Adding 1
1252 allows an extra byte for a null terminator, used by some
1253 extension processing routines. */
1254 if (record->size != 0
1255 && xsum (1, xtimes (record->count, record->size)) >= UINT_MAX)
1257 sys_error (r, record->pos, "Record type 7 subtype %d too large.",
1265 /* Reads an extension record from R into RECORD. */
1267 read_extension_record (struct sfm_reader *r, int subtype,
1268 struct sfm_extension_record **recordp)
1270 struct extension_record_type
1277 static const struct extension_record_type types[] =
1279 /* Implemented record types. */
1280 { EXT_INTEGER, 4, 8 },
1281 { EXT_FLOAT, 8, 3 },
1282 { EXT_MRSETS, 1, 0 },
1283 { EXT_PRODUCT_INFO, 1, 0 },
1284 { EXT_DISPLAY, 4, 0 },
1285 { EXT_LONG_NAMES, 1, 0 },
1286 { EXT_LONG_STRINGS, 1, 0 },
1287 { EXT_NCASES, 8, 2 },
1288 { EXT_FILE_ATTRS, 1, 0 },
1289 { EXT_VAR_ATTRS, 1, 0 },
1290 { EXT_MRSETS2, 1, 0 },
1291 { EXT_ENCODING, 1, 0 },
1292 { EXT_LONG_LABELS, 1, 0 },
1293 { EXT_LONG_MISSING, 1, 0 },
1295 /* Ignored record types. */
1296 { EXT_VAR_SETS, 0, 0 },
1298 { EXT_DATA_ENTRY, 0, 0 },
1299 { EXT_DATAVIEW, 0, 0 },
1302 const struct extension_record_type *type;
1303 struct sfm_extension_record *record;
1307 record = pool_malloc (r->pool, sizeof *record);
1308 if (!read_extension_record_header (r, subtype, record))
1310 n_bytes = record->count * record->size;
1312 for (type = types; type < &types[sizeof types / sizeof *types]; type++)
1313 if (subtype == type->subtype)
1315 if (type->size > 0 && record->size != type->size)
1316 sys_warn (r, record->pos,
1317 _("Record type 7, subtype %d has bad size %u "
1318 "(expected %d)."), subtype, record->size, type->size);
1319 else if (type->count > 0 && record->count != type->count)
1320 sys_warn (r, record->pos,
1321 _("Record type 7, subtype %d has bad count %u "
1322 "(expected %d)."), subtype, record->count, type->count);
1323 else if (type->count == 0 && type->size == 0)
1325 /* Ignore this record. */
1329 char *data = pool_malloc (r->pool, n_bytes + 1);
1330 data[n_bytes] = '\0';
1332 record->data = data;
1333 if (!read_bytes (r, record->data, n_bytes))
1342 sys_warn (r, record->pos,
1343 _("Unrecognized record type 7, subtype %d. For help, please "
1344 "send this file to %s and mention that you were using %s."),
1345 subtype, PACKAGE_BUGREPORT, PACKAGE_STRING);
1348 return skip_bytes (r, n_bytes);
1352 skip_extension_record (struct sfm_reader *r, int subtype)
1354 struct sfm_extension_record record;
1356 return (read_extension_record_header (r, subtype, &record)
1357 && skip_bytes (r, record.count * record.size));
1361 parse_header (struct sfm_reader *r, const struct sfm_header_record *header,
1362 struct any_read_info *info, struct dictionary *dict)
1364 const char *dict_encoding = dict_get_encoding (dict);
1365 struct substring product;
1366 struct substring label;
1369 /* Convert file label to UTF-8 and put it into DICT. */
1370 label = recode_substring_pool ("UTF-8", dict_encoding,
1371 ss_cstr (header->file_label), r->pool);
1372 ss_trim (&label, ss_cstr (" "));
1373 label.string[label.length] = '\0';
1374 fixed_label = fix_line_ends (label.string);
1375 dict_set_label (dict, fixed_label);
1378 /* Put creation date and time in UTF-8 into INFO. */
1379 info->creation_date = recode_string ("UTF-8", dict_encoding,
1380 header->creation_date, -1);
1381 info->creation_time = recode_string ("UTF-8", dict_encoding,
1382 header->creation_time, -1);
1384 /* Put product name into INFO, dropping eye-catcher string if present. */
1385 product = recode_substring_pool ("UTF-8", dict_encoding,
1386 ss_cstr (header->eye_catcher), r->pool);
1387 ss_match_string (&product, ss_cstr ("@(#) SPSS DATA FILE"));
1388 ss_trim (&product, ss_cstr (" "));
1389 info->product = ss_xstrdup (product);
1392 static struct variable *
1393 add_var_with_generated_name (struct dictionary *dict, int width)
1395 char *name = dict_make_unique_var_name (dict, NULL, NULL);
1396 struct variable *var = dict_create_var_assert (dict, name, width);
1401 /* Reads a variable (type 2) record from R and adds the
1402 corresponding variable to DICT.
1403 Also skips past additional variable records for long string
1406 parse_variable_records (struct sfm_reader *r, struct dictionary *dict,
1407 struct sfm_var_record *var_recs, size_t n_var_recs)
1409 const char *dict_encoding = dict_get_encoding (dict);
1410 struct sfm_var_record *rec;
1413 for (rec = var_recs; rec < &var_recs[n_var_recs];)
1419 name = recode_string_pool ("UTF-8", dict_encoding,
1420 rec->name, -1, r->pool);
1421 name[strcspn (name, " ")] = '\0';
1423 if (rec->width < 0 || rec->width > 255)
1425 sys_error (r, rec->pos,
1426 _("Bad width %d for variable %s."), rec->width, name);
1430 struct variable *var;
1431 if (!dict_id_is_valid (dict, name, false)
1432 || name[0] == '$' || name[0] == '#')
1434 var = add_var_with_generated_name (dict, rec->width);
1435 sys_warn (r, rec->pos, _("Renaming variable with invalid name "
1436 "`%s' to `%s'."), name, var_get_name (var));
1440 var = dict_create_var (dict, name, rec->width);
1443 var = add_var_with_generated_name (dict, rec->width);
1444 sys_warn (r, rec->pos, _("Renaming variable with duplicate name "
1446 name, var_get_name (var));
1451 /* Set the short name the same as the long name (even if we renamed
1453 var_set_short_name (var, 0, var_get_name (var));
1455 /* Get variable label, if any. */
1460 utf8_label = recode_string_pool ("UTF-8", dict_encoding,
1461 rec->label, -1, r->pool);
1462 var_set_label (var, utf8_label);
1465 /* Set missing values. */
1466 if (rec->missing_value_code != 0)
1468 int width = var_get_width (var);
1469 struct missing_values mv;
1471 mv_init_pool (r->pool, &mv, width);
1472 if (var_is_numeric (var))
1474 bool has_range = rec->missing_value_code < 0;
1475 int n_discrete = (has_range
1476 ? rec->missing_value_code == -3
1477 : rec->missing_value_code);
1482 double low = parse_float (r, rec->missing, 0);
1483 double high = parse_float (r, rec->missing, 8);
1485 /* Deal with SPSS 21 change in representation. */
1489 mv_add_range (&mv, low, high);
1493 for (i = 0; i < n_discrete; i++)
1495 mv_add_num (&mv, parse_float (r, rec->missing, ofs));
1500 for (i = 0; i < rec->missing_value_code; i++)
1501 mv_add_str (&mv, rec->missing + 8 * i, MIN (width, 8));
1502 var_set_missing_values (var, &mv);
1506 parse_format_spec (r, rec->pos + 12, rec->print_format,
1507 PRINT_FORMAT, var, &n_warnings);
1508 parse_format_spec (r, rec->pos + 16, rec->write_format,
1509 WRITE_FORMAT, var, &n_warnings);
1511 /* Account for values.
1512 Skip long string continuation records, if any. */
1513 n_values = rec->width == 0 ? 1 : DIV_RND_UP (rec->width, 8);
1514 for (i = 1; i < n_values; i++)
1515 if (i + (rec - var_recs) >= n_var_recs || rec[i].width != -1)
1517 sys_error (r, rec->pos, _("Missing string continuation record."));
1526 /* Translates the format spec from sysfile format to internal
1529 parse_format_spec (struct sfm_reader *r, off_t pos, unsigned int format,
1530 enum which_format which, struct variable *v,
1533 const int max_warnings = 8;
1536 if (fmt_from_u32 (format, var_get_width (v), false, &f))
1538 if (which == PRINT_FORMAT)
1539 var_set_print_format (v, &f);
1541 var_set_write_format (v, &f);
1543 else if (format == 0)
1545 /* Actually observed in the wild. No point in warning about it. */
1547 else if (++*n_warnings <= max_warnings)
1549 if (which == PRINT_FORMAT)
1550 sys_warn (r, pos, _("Variable %s with width %d has invalid print "
1552 var_get_name (v), var_get_width (v), format);
1554 sys_warn (r, pos, _("Variable %s with width %d has invalid write "
1556 var_get_name (v), var_get_width (v), format);
1558 if (*n_warnings == max_warnings)
1559 sys_warn (r, -1, _("Suppressing further invalid format warnings."));
1564 parse_document (struct dictionary *dict, struct sfm_document_record *record)
1568 for (p = record->documents;
1569 p < record->documents + DOC_LINE_LENGTH * record->n_lines;
1570 p += DOC_LINE_LENGTH)
1572 struct substring line;
1574 line = recode_substring_pool ("UTF-8", dict_get_encoding (dict),
1575 ss_buffer (p, DOC_LINE_LENGTH), NULL);
1576 ss_rtrim (&line, ss_cstr (" "));
1577 line.string[line.length] = '\0';
1579 dict_add_document_line (dict, line.string, false);
1585 /* Parses record type 7, subtype 3. */
1587 parse_machine_integer_info (struct sfm_reader *r,
1588 const struct sfm_extension_record *record,
1589 struct any_read_info *info)
1591 int float_representation, expected_float_format;
1592 int integer_representation, expected_integer_format;
1594 /* Save version info. */
1595 info->version_major = parse_int (r, record->data, 0);
1596 info->version_minor = parse_int (r, record->data, 4);
1597 info->version_revision = parse_int (r, record->data, 8);
1599 /* Check floating point format. */
1600 float_representation = parse_int (r, record->data, 16);
1601 if (r->float_format == FLOAT_IEEE_DOUBLE_BE
1602 || r->float_format == FLOAT_IEEE_DOUBLE_LE)
1603 expected_float_format = 1;
1604 else if (r->float_format == FLOAT_Z_LONG)
1605 expected_float_format = 2;
1606 else if (r->float_format == FLOAT_VAX_G || r->float_format == FLOAT_VAX_D)
1607 expected_float_format = 3;
1610 if (float_representation != expected_float_format)
1612 sys_error (r, record->pos,
1613 _("Floating-point representation indicated by "
1614 "system file (%d) differs from expected (%d)."),
1615 float_representation, expected_float_format);
1619 /* Check integer format. */
1620 integer_representation = parse_int (r, record->data, 24);
1621 if (r->integer_format == INTEGER_MSB_FIRST)
1622 expected_integer_format = 1;
1623 else if (r->integer_format == INTEGER_LSB_FIRST)
1624 expected_integer_format = 2;
1627 if (integer_representation != expected_integer_format)
1628 sys_warn (r, record->pos,
1629 _("Integer format indicated by system file (%d) "
1630 "differs from expected (%d)."),
1631 integer_representation, expected_integer_format);
1636 /* Parses record type 7, subtype 4. */
1638 parse_machine_float_info (struct sfm_reader *r,
1639 const struct sfm_extension_record *record)
1641 double sysmis = parse_float (r, record->data, 0);
1642 double highest = parse_float (r, record->data, 8);
1643 double lowest = parse_float (r, record->data, 16);
1645 if (sysmis != SYSMIS)
1646 sys_warn (r, record->pos,
1647 _("File specifies unexpected value %g (%a) as %s, "
1648 "instead of %g (%a)."),
1649 sysmis, sysmis, "SYSMIS", SYSMIS, SYSMIS);
1651 if (highest != HIGHEST)
1652 sys_warn (r, record->pos,
1653 _("File specifies unexpected value %g (%a) as %s, "
1654 "instead of %g (%a)."),
1655 highest, highest, "HIGHEST", HIGHEST, HIGHEST);
1657 /* SPSS before version 21 used a unique value just bigger than SYSMIS as
1658 LOWEST. SPSS 21 uses SYSMIS for LOWEST, which is OK because LOWEST only
1659 appears in a context (missing values) where SYSMIS cannot. */
1660 if (lowest != LOWEST && lowest != SYSMIS)
1661 sys_warn (r, record->pos,
1662 _("File specifies unexpected value %g (%a) as %s, "
1663 "instead of %g (%a) or %g (%a)."),
1664 lowest, lowest, "LOWEST", LOWEST, LOWEST, SYSMIS, SYSMIS);
1667 /* Parses record type 7, subtype 10. */
1669 parse_extra_product_info (struct sfm_reader *r,
1670 const struct sfm_extension_record *record,
1671 struct any_read_info *info)
1673 struct text_record *text;
1675 text = open_text_record (r, record, true);
1676 info->product_ext = fix_line_ends (text_get_all (text));
1677 close_text_record (r, text);
1680 /* Parses record type 7, subtype 7 or 19. */
1682 parse_mrsets (struct sfm_reader *r, const struct sfm_extension_record *record,
1683 size_t *allocated_mrsets)
1685 struct text_record *text;
1687 text = open_text_record (r, record, false);
1690 struct sfm_mrset *mrset = NULL;
1691 size_t allocated_vars = 0;
1692 char delimiter = '4';
1694 /* Skip extra line feeds if present. */
1695 while (text_match (text, '\n'))
1698 if (r->n_mrsets >= *allocated_mrsets)
1699 r->mrsets = pool_2nrealloc (r->pool, r->mrsets, allocated_mrsets,
1701 mrset = &r->mrsets[r->n_mrsets];
1702 memset(mrset, 0, sizeof *mrset);
1704 mrset->name = text_get_token (text, ss_cstr ("="), NULL);
1705 if (mrset->name == NULL)
1708 if (text_match (text, 'C'))
1710 mrset->type = MRSET_MC;
1711 if (!text_match (text, ' '))
1713 sys_warn (r, record->pos,
1714 _("Missing space following `%c' at offset %zu "
1715 "in MRSETS record."), 'C', text_pos (text));
1719 else if (text_match (text, 'D'))
1721 mrset->type = MRSET_MD;
1722 mrset->cat_source = MRSET_VARLABELS;
1724 else if (text_match (text, 'E'))
1728 mrset->type = MRSET_MD;
1729 mrset->cat_source = MRSET_COUNTEDVALUES;
1730 if (!text_match (text, ' '))
1732 sys_warn (r, record->pos,
1733 _("Missing space following `%c' at offset %zu "
1734 "in MRSETS record."), 'E', text_pos (text));
1738 number = text_get_token (text, ss_cstr (" "), NULL);
1740 sys_warn (r, record->pos,
1741 _("Missing label source value "
1742 "following `E' at offset %zu in MRSETS record."),
1744 else if (!strcmp (number, "11"))
1745 mrset->label_from_var_label = true;
1746 else if (strcmp (number, "1"))
1747 sys_warn (r, record->pos,
1748 _("Unexpected label source value following `E' "
1749 "at offset %zu in MRSETS record."),
1754 sys_warn (r, record->pos,
1755 _("Missing `C', `D', or `E' at offset %zu "
1756 "in MRSETS record."),
1761 if (mrset->type == MRSET_MD)
1763 mrset->counted = text_parse_counted_string (r, text);
1764 if (mrset->counted == NULL)
1768 mrset->label = text_parse_counted_string (r, text);
1769 if (mrset->label == NULL)
1777 var = text_get_token (text, ss_cstr (" \n"), &delimiter);
1780 if (delimiter != '\n')
1781 sys_warn (r, record->pos,
1782 _("Missing new-line parsing variable names "
1783 "at offset %zu in MRSETS record."),
1788 if (mrset->n_vars >= allocated_vars)
1789 mrset->vars = pool_2nrealloc (r->pool, mrset->vars,
1791 sizeof *mrset->vars);
1792 mrset->vars[mrset->n_vars++] = var;
1794 while (delimiter != '\n');
1798 close_text_record (r, text);
1802 decode_mrsets (struct sfm_reader *r, struct dictionary *dict)
1804 const struct sfm_mrset *s;
1806 for (s = r->mrsets; s < &r->mrsets[r->n_mrsets]; s++)
1808 struct stringi_set var_names;
1809 struct mrset *mrset;
1814 name = recode_string ("UTF-8", r->encoding, s->name, -1);
1815 if (!mrset_is_valid_name (name, dict_get_encoding (dict), false))
1817 sys_warn (r, -1, _("Invalid multiple response set name `%s'."),
1823 mrset = xzalloc (sizeof *mrset);
1825 mrset->type = s->type;
1826 mrset->cat_source = s->cat_source;
1827 mrset->label_from_var_label = s->label_from_var_label;
1828 if (s->label[0] != '\0')
1829 mrset->label = recode_string ("UTF-8", r->encoding, s->label, -1);
1831 stringi_set_init (&var_names);
1832 mrset->vars = xmalloc (s->n_vars * sizeof *mrset->vars);
1834 for (i = 0; i < s->n_vars; i++)
1836 struct variable *var;
1839 var_name = recode_string ("UTF-8", r->encoding, s->vars[i], -1);
1841 var = dict_lookup_var (dict, var_name);
1847 if (!stringi_set_insert (&var_names, var_name))
1850 _("MRSET %s contains duplicate variable name %s."),
1851 mrset->name, var_name);
1857 if (mrset->label == NULL && mrset->label_from_var_label
1858 && var_has_label (var))
1859 mrset->label = xstrdup (var_get_label (var));
1862 && var_get_type (var) != var_get_type (mrset->vars[0]))
1865 _("MRSET %s contains both string and "
1866 "numeric variables."), mrset->name);
1869 width = MIN (width, var_get_width (var));
1871 mrset->vars[mrset->n_vars++] = var;
1874 if (mrset->n_vars < 2)
1876 if (mrset->n_vars == 0)
1877 sys_warn (r, -1, _("MRSET %s has no variables."), mrset->name);
1879 sys_warn (r, -1, _("MRSET %s has only one variable."),
1881 mrset_destroy (mrset);
1882 stringi_set_destroy (&var_names);
1886 if (mrset->type == MRSET_MD)
1888 mrset->width = width;
1889 value_init (&mrset->counted, width);
1891 mrset->counted.f = c_strtod (s->counted, NULL);
1893 value_copy_str_rpad (&mrset->counted, width,
1894 (const uint8_t *) s->counted, ' ');
1897 dict_add_mrset (dict, mrset);
1898 stringi_set_destroy (&var_names);
1902 /* Read record type 7, subtype 11, which specifies how variables
1903 should be displayed in GUI environments. */
1905 parse_display_parameters (struct sfm_reader *r,
1906 const struct sfm_extension_record *record,
1907 struct dictionary *dict)
1909 bool includes_width;
1910 bool warned = false;
1915 n_vars = dict_get_var_cnt (dict);
1916 if (record->count == 3 * n_vars)
1917 includes_width = true;
1918 else if (record->count == 2 * n_vars)
1919 includes_width = false;
1922 sys_warn (r, record->pos,
1923 _("Extension 11 has bad count %u (for %zu variables)."),
1924 record->count, n_vars);
1929 for (i = 0; i < n_vars; ++i)
1931 struct variable *v = dict_get_var (dict, i);
1932 int measure, width, align;
1934 measure = parse_int (r, record->data, ofs);
1939 width = parse_int (r, record->data, ofs);
1945 align = parse_int (r, record->data, ofs);
1948 /* SPSS sometimes seems to set variables' measure to zero. */
1952 if (measure < 1 || measure > 3 || align < 0 || align > 2)
1955 sys_warn (r, record->pos,
1956 _("Invalid variable display parameters for variable "
1957 "%zu (%s). Default parameters substituted."),
1958 i, var_get_name (v));
1963 var_set_measure (v, (measure == 1 ? MEASURE_NOMINAL
1964 : measure == 2 ? MEASURE_ORDINAL
1966 var_set_alignment (v, (align == 0 ? ALIGN_LEFT
1967 : align == 1 ? ALIGN_RIGHT
1970 /* Older versions (SPSS 9.0) sometimes set the display
1971 width to zero. This causes confusion in the GUI, so
1972 only set the width if it is nonzero. */
1974 var_set_display_width (v, width);
1979 rename_var_and_save_short_names (struct sfm_reader *r, off_t pos,
1980 struct dictionary *dict,
1981 struct variable *var, const char *new_name)
1983 size_t n_short_names;
1987 /* Renaming a variable may clear its short names, but we
1988 want to retain them, so we save them and re-set them
1990 n_short_names = var_get_short_name_cnt (var);
1991 short_names = xnmalloc (n_short_names, sizeof *short_names);
1992 for (i = 0; i < n_short_names; i++)
1994 const char *s = var_get_short_name (var, i);
1995 short_names[i] = s != NULL ? xstrdup (s) : NULL;
1998 /* Set long name. */
1999 if (!dict_try_rename_var (dict, var, new_name))
2000 sys_warn (r, pos, _("Duplicate long variable name `%s'."), new_name);
2002 /* Restore short names. */
2003 for (i = 0; i < n_short_names; i++)
2005 var_set_short_name (var, i, short_names[i]);
2006 free (short_names[i]);
2011 /* Parses record type 7, subtype 13, which gives the long name that corresponds
2012 to each short name. Modifies variable names in DICT accordingly. */
2014 parse_long_var_name_map (struct sfm_reader *r,
2015 const struct sfm_extension_record *record,
2016 struct dictionary *dict)
2018 struct text_record *text;
2019 struct variable *var;
2024 /* There are no long variable names. Use the short variable names,
2025 converted to lowercase, as the long variable names. */
2028 for (i = 0; i < dict_get_var_cnt (dict); i++)
2030 struct variable *var = dict_get_var (dict, i);
2033 new_name = utf8_to_lower (var_get_name (var));
2034 rename_var_and_save_short_names (r, -1, dict, var, new_name);
2041 /* Rename each of the variables, one by one. (In a correctly constructed
2042 system file, this cannot create any intermediate duplicate variable names,
2043 because all of the new variable names are longer than any of the old
2044 variable names and thus there cannot be any overlaps.) */
2045 text = open_text_record (r, record, true);
2046 while (read_variable_to_value_pair (r, dict, text, &var, &long_name))
2048 /* Validate long name. */
2049 if (!dict_id_is_valid (dict, long_name, false)
2050 || long_name[0] == '$' || long_name[0] == '#')
2052 sys_warn (r, record->pos,
2053 _("Long variable mapping from %s to invalid "
2054 "variable name `%s'."),
2055 var_get_name (var), long_name);
2059 rename_var_and_save_short_names (r, record->pos, dict, var, long_name);
2061 close_text_record (r, text);
2064 /* Reads record type 7, subtype 14, which gives the real length
2065 of each very long string. Rearranges DICT accordingly. */
2067 parse_long_string_map (struct sfm_reader *r,
2068 const struct sfm_extension_record *record,
2069 struct dictionary *dict)
2071 struct text_record *text;
2072 struct variable *var;
2075 text = open_text_record (r, record, true);
2076 while (read_variable_to_value_pair (r, dict, text, &var, &length_s))
2078 size_t idx = var_get_dict_index (var);
2084 length = strtol (length_s, NULL, 10);
2085 if (length < 1 || length > MAX_STRING)
2087 sys_warn (r, record->pos,
2088 _("%s listed as string of invalid length %s "
2089 "in very long string record."),
2090 var_get_name (var), length_s);
2094 /* Check segments. */
2095 segment_cnt = sfm_width_to_segments (length);
2096 if (segment_cnt == 1)
2098 sys_warn (r, record->pos,
2099 _("%s listed in very long string record with width %s, "
2100 "which requires only one segment."),
2101 var_get_name (var), length_s);
2104 if (idx + segment_cnt > dict_get_var_cnt (dict))
2106 sys_error (r, record->pos,
2107 _("Very long string %s overflows dictionary."),
2108 var_get_name (var));
2112 /* Get the short names from the segments and check their
2114 for (i = 0; i < segment_cnt; i++)
2116 struct variable *seg = dict_get_var (dict, idx + i);
2117 int alloc_width = sfm_segment_alloc_width (length, i);
2118 int width = var_get_width (seg);
2121 var_set_short_name (var, i, var_get_short_name (seg, 0));
2122 if (ROUND_UP (width, 8) != ROUND_UP (alloc_width, 8))
2124 sys_error (r, record->pos,
2125 _("Very long string with width %ld has segment %d "
2126 "of width %d (expected %d)."),
2127 length, i, width, alloc_width);
2131 dict_delete_consecutive_vars (dict, idx + 1, segment_cnt - 1);
2132 var_set_width (var, length);
2134 close_text_record (r, text);
2135 dict_compact_values (dict);
2140 #define MAX_LABEL_WARNINGS 5
2142 /* Displays a warning for offset OFFSET in the file. */
2144 value_label_warning (struct sfm_reader *r, off_t offset, int *n_label_warnings,
2145 const char *format, ...)
2147 if (++*n_label_warnings > MAX_LABEL_WARNINGS)
2152 va_start (args, format);
2153 sys_msg (r, offset, MW, format, args);
2157 #define MAX_LABEL_WARNINGS 5
2160 parse_one_value_label_set (struct sfm_reader *r, struct dictionary *dict,
2161 const struct sfm_var_record *var_recs,
2163 const struct sfm_value_label_record *record,
2164 int *n_label_warnings)
2167 = pool_nmalloc (r->pool, record->n_labels, sizeof *utf8_labels);
2168 for (size_t i = 0; i < record->n_labels; i++)
2169 utf8_labels[i] = recode_string_pool ("UTF-8", dict_get_encoding (dict),
2170 record->labels[i].label, -1,
2173 struct variable **vars = pool_nmalloc (r->pool,
2174 record->n_vars, sizeof *vars);
2175 unsigned int n_vars = 0;
2176 for (size_t i = 0; i < record->n_vars; i++)
2178 int idx = record->vars[i];
2179 if (idx < 1 || idx > n_var_recs)
2181 value_label_warning (
2182 r, record->pos, n_label_warnings,
2183 _("Value label variable index %d not in valid range 1...%zu."),
2188 const struct sfm_var_record *rec = &var_recs[idx - 1];
2189 if (rec->var == NULL)
2191 value_label_warning (
2192 r, record->pos, n_label_warnings,
2193 _("Value label variable index %d "
2194 "refers to long string continuation."), idx);
2198 vars[n_vars++] = rec->var;
2203 for (size_t i = 1; i < n_vars; i++)
2204 if (var_get_type (vars[i]) != var_get_type (vars[0]))
2206 value_label_warning (
2207 r, record->pos, n_label_warnings,
2208 _("Variables associated with value label are not all of "
2209 "identical type. Variable %s is %s, but variable "
2211 var_get_name (vars[0]),
2212 var_is_numeric (vars[0]) ? _("numeric") : _("string"),
2213 var_get_name (vars[i]),
2214 var_is_numeric (vars[i]) ? _("numeric") : _("string"));
2218 for (size_t i = 0; i < n_vars; i++)
2220 struct variable *var = vars[i];
2221 int width = var_get_width (var);
2224 value_label_warning (
2225 r, record->pos, n_label_warnings,
2226 _("Value labels may not be added to long string "
2227 "variables (e.g. %s) using records types 3 and 4."),
2228 var_get_name (var));
2232 for (size_t j = 0; j < record->n_labels; j++)
2234 struct sfm_value_label *label = &record->labels[j];
2237 value_init (&value, width);
2239 value.f = parse_float (r, label->value, 0);
2241 memcpy (value.s, label->value, width);
2243 if (!var_add_value_label (var, &value, utf8_labels[j]))
2245 if (r->written_by_readstat)
2247 /* Ignore the problem. ReadStat is buggy and emits value
2248 labels whose values are longer than string variables'
2249 widths, that are identical in the actual width of the
2250 variable, e.g. both values "ABC123" and "ABC456" for a
2251 string variable with width 3. */
2253 else if (var_is_numeric (var))
2254 value_label_warning (r, record->pos, n_label_warnings,
2255 _("Duplicate value label for %g on %s."),
2256 value.f, var_get_name (var));
2258 value_label_warning (
2259 r, record->pos, n_label_warnings,
2260 _("Duplicate value label for `%.*s' on %s."),
2261 width, value.s, var_get_name (var));
2264 value_destroy (&value, width);
2268 pool_free (r->pool, vars);
2269 for (size_t i = 0; i < record->n_labels; i++)
2270 pool_free (r->pool, utf8_labels[i]);
2271 pool_free (r->pool, utf8_labels);
2275 parse_value_labels (struct sfm_reader *r, struct dictionary *dict)
2277 int n_label_warnings = 0;
2278 for (size_t i = 0; i < r->n_labels; i++)
2279 parse_one_value_label_set (r, dict, r->vars, r->n_vars, &r->labels[i],
2281 if (n_label_warnings > MAX_LABEL_WARNINGS)
2283 _("Suppressed %d additional warnings for value labels."),
2284 n_label_warnings - MAX_LABEL_WARNINGS);
2287 static struct variable *
2288 parse_weight_var (struct sfm_reader *r,
2289 const struct sfm_var_record *var_recs, size_t n_var_recs,
2292 off_t offset = 76; /* Offset to variable index in header. */
2294 if (idx < 1 || idx > n_var_recs)
2296 sys_warn (r, offset,
2297 _("Weight variable index %d not in valid range 1...%zu. "
2298 "Treating file as unweighted."),
2303 const struct sfm_var_record *rec = &var_recs[idx - 1];
2304 if (rec->var == NULL)
2306 sys_warn (r, offset,
2307 _("Weight variable index %d refers to long string "
2308 "continuation. Treating file as unweighted."), idx);
2312 struct variable *weight_var = rec->var;
2313 if (!var_is_numeric (weight_var))
2315 sys_warn (r, offset, _("Ignoring string variable `%s' set "
2316 "as weighting variable."),
2317 var_get_name (weight_var));
2324 /* Parses a set of custom attributes from TEXT into ATTRS.
2325 ATTRS may be a null pointer, in which case the attributes are
2326 read but discarded. */
2328 parse_attributes (struct sfm_reader *r, struct text_record *text,
2329 struct attrset *attrs)
2333 struct attribute *attr;
2337 /* Parse the key. */
2338 key = text_get_token (text, ss_cstr ("("), NULL);
2342 attr = attribute_create (key);
2343 for (index = 1; ; index++)
2345 /* Parse the value. */
2349 value = text_get_token (text, ss_cstr ("\n"), NULL);
2352 text_warn (r, text, _("Error parsing attribute value %s[%d]."),
2357 length = strlen (value);
2358 if (length >= 2 && value[0] == '\'' && value[length - 1] == '\'')
2360 value[length - 1] = '\0';
2361 attribute_add_value (attr, value + 1);
2366 _("Attribute value %s[%d] is not quoted: %s."),
2368 attribute_add_value (attr, value);
2371 /* Was this the last value for this attribute? */
2372 if (text_match (text, ')'))
2375 if (attrs != NULL && attribute_get_n_values (attr) > 0)
2377 if (!attrset_try_add (attrs, attr))
2379 text_warn (r, text, _("Duplicate attribute %s."),
2380 attribute_get_name (attr));
2381 attribute_destroy (attr);
2385 attribute_destroy (attr);
2387 while (!text_match (text, '/'));
2390 /* Reads record type 7, subtype 17, which lists custom
2391 attributes on the data file. */
2393 parse_data_file_attributes (struct sfm_reader *r,
2394 const struct sfm_extension_record *record,
2395 struct dictionary *dict)
2397 struct text_record *text = open_text_record (r, record, true);
2398 parse_attributes (r, text, dict_get_attributes (dict));
2399 close_text_record (r, text);
2402 /* Parses record type 7, subtype 18, which lists custom
2403 attributes on individual variables. */
2405 parse_variable_attributes (struct sfm_reader *r,
2406 const struct sfm_extension_record *record,
2407 struct dictionary *dict)
2409 struct text_record *text;
2410 struct variable *var;
2412 text = open_text_record (r, record, true);
2413 while (text_read_variable_name (r, dict, text, ss_cstr (":"), &var))
2414 parse_attributes (r, text, var != NULL ? var_get_attributes (var) : NULL);
2415 close_text_record (r, text);
2419 assign_variable_roles (struct sfm_reader *r, struct dictionary *dict)
2421 size_t n_warnings = 0;
2424 for (i = 0; i < dict_get_var_cnt (dict); i++)
2426 struct variable *var = dict_get_var (dict, i);
2427 struct attrset *attrs = var_get_attributes (var);
2428 const struct attribute *attr = attrset_lookup (attrs, "$@Role");
2429 if (attr != NULL && attribute_get_n_values (attr) > 0)
2431 int value = atoi (attribute_get_value (attr, 0));
2453 role = ROLE_PARTITION;
2462 if (n_warnings++ == 0)
2463 sys_warn (r, -1, _("Invalid role for variable %s."),
2464 var_get_name (var));
2467 var_set_role (var, role);
2472 sys_warn (r, -1, _("%zu other variables had invalid roles."),
2477 check_overflow (struct sfm_reader *r,
2478 const struct sfm_extension_record *record,
2479 size_t ofs, size_t length)
2481 size_t end = record->size * record->count;
2482 if (length >= end || ofs + length > end)
2484 sys_warn (r, record->pos + end,
2485 _("Extension record subtype %d ends unexpectedly."),
2493 parse_long_string_value_labels (struct sfm_reader *r,
2494 const struct sfm_extension_record *record,
2495 struct dictionary *dict)
2497 const char *dict_encoding = dict_get_encoding (dict);
2498 size_t end = record->size * record->count;
2505 struct variable *var;
2510 /* Parse variable name length. */
2511 if (!check_overflow (r, record, ofs, 4))
2513 var_name_len = parse_int (r, record->data, ofs);
2516 /* Parse variable name, width, and number of labels. */
2517 if (!check_overflow (r, record, ofs, var_name_len)
2518 || !check_overflow (r, record, ofs, var_name_len + 8))
2520 var_name = recode_string_pool ("UTF-8", dict_encoding,
2521 (const char *) record->data + ofs,
2522 var_name_len, r->pool);
2523 width = parse_int (r, record->data, ofs + var_name_len);
2524 n_labels = parse_int (r, record->data, ofs + var_name_len + 4);
2525 ofs += var_name_len + 8;
2527 /* Look up 'var' and validate. */
2528 var = dict_lookup_var (dict, var_name);
2530 sys_warn (r, record->pos + ofs,
2531 _("Ignoring long string value label record for "
2532 "unknown variable %s."), var_name);
2533 else if (var_is_numeric (var))
2535 sys_warn (r, record->pos + ofs,
2536 _("Ignoring long string value label record for "
2537 "numeric variable %s."), var_name);
2540 else if (width != var_get_width (var))
2542 sys_warn (r, record->pos + ofs,
2543 _("Ignoring long string value label record for variable "
2544 "%s because the record's width (%d) does not match the "
2545 "variable's width (%d)."),
2546 var_name, width, var_get_width (var));
2551 value_init_pool (r->pool, &value, width);
2552 for (i = 0; i < n_labels; i++)
2554 size_t value_length, label_length;
2555 bool skip = var == NULL;
2557 /* Parse value length. */
2558 if (!check_overflow (r, record, ofs, 4))
2560 value_length = parse_int (r, record->data, ofs);
2564 if (!check_overflow (r, record, ofs, value_length))
2568 if (value_length == width)
2569 memcpy (value.s, (const uint8_t *) record->data + ofs, width);
2572 sys_warn (r, record->pos + ofs,
2573 _("Ignoring long string value label %zu for "
2574 "variable %s, with width %d, that has bad value "
2576 i, var_get_name (var), width, value_length);
2580 ofs += value_length;
2582 /* Parse label length. */
2583 if (!check_overflow (r, record, ofs, 4))
2585 label_length = parse_int (r, record->data, ofs);
2589 if (!check_overflow (r, record, ofs, label_length))
2595 label = recode_string_pool ("UTF-8", dict_encoding,
2596 (const char *) record->data + ofs,
2597 label_length, r->pool);
2598 if (!var_add_value_label (var, &value, label))
2599 sys_warn (r, record->pos + ofs,
2600 _("Duplicate value label for `%.*s' on %s."),
2601 width, value.s, var_get_name (var));
2602 pool_free (r->pool, label);
2604 ofs += label_length;
2610 parse_long_string_missing_values (struct sfm_reader *r,
2611 const struct sfm_extension_record *record,
2612 struct dictionary *dict)
2614 const char *dict_encoding = dict_get_encoding (dict);
2615 size_t end = record->size * record->count;
2620 struct missing_values mv;
2622 struct variable *var;
2623 int n_missing_values;
2627 /* Parse variable name length. */
2628 if (!check_overflow (r, record, ofs, 4))
2630 var_name_len = parse_int (r, record->data, ofs);
2633 /* Parse variable name. */
2634 if (!check_overflow (r, record, ofs, var_name_len)
2635 || !check_overflow (r, record, ofs, var_name_len + 1))
2637 var_name = recode_string_pool ("UTF-8", dict_encoding,
2638 (const char *) record->data + ofs,
2639 var_name_len, r->pool);
2640 ofs += var_name_len;
2642 /* Parse number of missing values. */
2643 n_missing_values = ((const uint8_t *) record->data)[ofs];
2644 if (n_missing_values < 1 || n_missing_values > 3)
2645 sys_warn (r, record->pos + ofs,
2646 _("Long string missing values record says variable %s "
2647 "has %d missing values, but only 1 to 3 missing values "
2649 var_name, n_missing_values);
2652 /* Look up 'var' and validate. */
2653 var = dict_lookup_var (dict, var_name);
2655 sys_warn (r, record->pos + ofs,
2656 _("Ignoring long string missing value record for "
2657 "unknown variable %s."), var_name);
2658 else if (var_is_numeric (var))
2660 sys_warn (r, record->pos + ofs,
2661 _("Ignoring long string missing value record for "
2662 "numeric variable %s."), var_name);
2667 mv_init_pool (r->pool, &mv, var ? var_get_width (var) : 8);
2668 for (i = 0; i < n_missing_values; i++)
2670 size_t value_length;
2672 /* Parse value length. */
2673 if (!check_overflow (r, record, ofs, 4))
2675 value_length = parse_int (r, record->data, ofs);
2679 if (!check_overflow (r, record, ofs, value_length))
2683 && !mv_add_str (&mv, (const uint8_t *) record->data + ofs,
2685 sys_warn (r, record->pos + ofs,
2686 _("Ignoring long string missing value %zu for variable "
2687 "%s, with width %d, that has bad value width %zu."),
2688 i, var_get_name (var), var_get_width (var),
2690 ofs += value_length;
2693 var_set_missing_values (var, &mv);
2699 static void partial_record (struct sfm_reader *);
2701 static void read_error (struct casereader *, const struct sfm_reader *);
2703 static bool read_case_number (struct sfm_reader *, double *);
2704 static int read_case_string (struct sfm_reader *, uint8_t *, size_t);
2705 static int read_opcode (struct sfm_reader *);
2706 static bool read_compressed_number (struct sfm_reader *, double *);
2707 static int read_compressed_string (struct sfm_reader *, uint8_t *);
2708 static int read_whole_strings (struct sfm_reader *, uint8_t *, size_t);
2709 static bool skip_whole_strings (struct sfm_reader *, size_t);
2711 /* Reads and returns one case from READER's file. Returns a null
2712 pointer if not successful. */
2713 static struct ccase *
2714 sys_file_casereader_read (struct casereader *reader, void *r_)
2716 struct sfm_reader *r = r_;
2721 if (r->error || !r->sfm_var_cnt)
2724 c = case_create (r->proto);
2726 for (i = 0; i < r->sfm_var_cnt; i++)
2728 struct sfm_var *sv = &r->sfm_vars[i];
2729 union value *v = case_data_rw_idx (c, sv->case_index);
2731 if (sv->var_width == 0)
2732 retval = read_case_number (r, &v->f);
2735 retval = read_case_string (r, v->s + sv->offset, sv->segment_width);
2738 retval = skip_whole_strings (r, ROUND_DOWN (sv->padding, 8));
2740 sys_error (r, r->pos, _("File ends in partial string value."));
2752 if (r->case_cnt != -1)
2753 read_error (reader, r);
2758 /* Issues an error that R ends in a partial record. */
2760 partial_record (struct sfm_reader *r)
2762 sys_error (r, r->pos, _("File ends in partial case."));
2765 /* Issues an error that an unspecified error occurred SFM, and
2768 read_error (struct casereader *r, const struct sfm_reader *sfm)
2770 msg (ME, _("Error reading case from file %s."), fh_get_name (sfm->fh));
2771 casereader_force_error (r);
2774 /* Reads a number from R and stores its value in *D.
2775 If R is compressed, reads a compressed number;
2776 otherwise, reads a number in the regular way.
2777 Returns true if successful, false if end of file is
2778 reached immediately. */
2780 read_case_number (struct sfm_reader *r, double *d)
2782 if (r->compression == ANY_COMP_NONE)
2785 if (!try_read_bytes (r, number, sizeof number))
2787 float_convert (r->float_format, number, FLOAT_NATIVE_DOUBLE, d);
2791 return read_compressed_number (r, d);
2794 /* Reads LENGTH string bytes from R into S. Always reads a multiple of 8
2795 bytes; if LENGTH is not a multiple of 8, then extra bytes are read and
2796 discarded without being written to S. Reads compressed strings if S is
2797 compressed. Returns 1 if successful, 0 if end of file is reached
2798 immediately, or -1 for some kind of error. */
2800 read_case_string (struct sfm_reader *r, uint8_t *s, size_t length)
2802 size_t whole = ROUND_DOWN (length, 8);
2803 size_t partial = length % 8;
2807 int retval = read_whole_strings (r, s, whole);
2815 int retval = read_whole_strings (r, bounce, sizeof bounce);
2827 memcpy (s + whole, bounce, partial);
2833 /* Reads and returns the next compression opcode from R. */
2835 read_opcode (struct sfm_reader *r)
2837 assert (r->compression != ANY_COMP_NONE);
2841 if (r->opcode_idx >= sizeof r->opcodes)
2844 int retval = try_read_compressed_bytes (r, r->opcodes,
2850 opcode = r->opcodes[r->opcode_idx++];
2857 /* Reads a compressed number from R and stores its value in D.
2858 Returns true if successful, false if end of file is
2859 reached immediately. */
2861 read_compressed_number (struct sfm_reader *r, double *d)
2863 int opcode = read_opcode (r);
2871 return read_compressed_float (r, d);
2874 float_convert (r->float_format, " ", FLOAT_NATIVE_DOUBLE, d);
2875 if (!r->corruption_warning)
2877 r->corruption_warning = true;
2878 sys_warn (r, r->pos,
2879 _("Possible compressed data corruption: "
2880 "compressed spaces appear in numeric field."));
2889 *d = opcode - r->bias;
2896 /* Reads a compressed 8-byte string segment from R and stores it in DST. */
2898 read_compressed_string (struct sfm_reader *r, uint8_t *dst)
2903 opcode = read_opcode (r);
2911 retval = read_compressed_bytes (r, dst, 8);
2912 return retval == 1 ? 1 : -1;
2915 memset (dst, ' ', 8);
2920 double value = opcode - r->bias;
2921 float_convert (FLOAT_NATIVE_DOUBLE, &value, r->float_format, dst);
2924 /* This has actually been seen "in the wild". The submitter of the
2925 file that showed that the contents decoded as spaces, but they
2926 were at the end of the field so it's possible that the null
2927 bytes just acted as null terminators. */
2929 else if (!r->corruption_warning)
2931 r->corruption_warning = true;
2932 sys_warn (r, r->pos,
2933 _("Possible compressed data corruption: "
2934 "string contains compressed integer (opcode %d)."),
2942 /* Reads LENGTH string bytes from R into S. LENGTH must be a multiple of 8.
2943 Reads compressed strings if S is compressed. Returns 1 if successful, 0 if
2944 end of file is reached immediately, or -1 for some kind of error. */
2946 read_whole_strings (struct sfm_reader *r, uint8_t *s, size_t length)
2948 assert (length % 8 == 0);
2949 if (r->compression == ANY_COMP_NONE)
2950 return try_read_bytes (r, s, length);
2955 for (ofs = 0; ofs < length; ofs += 8)
2957 int retval = read_compressed_string (r, s + ofs);
2972 /* Skips LENGTH string bytes from R.
2973 LENGTH must be a multiple of 8.
2974 (LENGTH is also limited to 1024, but that's only because the
2975 current caller never needs more than that many bytes.)
2976 Returns true if successful, false if end of file is
2977 reached immediately. */
2979 skip_whole_strings (struct sfm_reader *r, size_t length)
2981 uint8_t buffer[1024];
2982 assert (length < sizeof buffer);
2983 return read_whole_strings (r, buffer, length);
2986 /* Helpers for reading records that contain structured text
2989 /* Maximum number of warnings to issue for a single text
2991 #define MAX_TEXT_WARNINGS 5
2996 struct substring buffer; /* Record contents. */
2997 off_t start; /* Starting offset in file. */
2998 size_t pos; /* Current position in buffer. */
2999 int n_warnings; /* Number of warnings issued or suppressed. */
3000 bool recoded; /* Recoded into UTF-8? */
3003 static struct text_record *
3004 open_text_record (struct sfm_reader *r,
3005 const struct sfm_extension_record *record,
3006 bool recode_to_utf8)
3008 struct text_record *text;
3009 struct substring raw;
3011 text = pool_alloc (r->pool, sizeof *text);
3012 raw = ss_buffer (record->data, record->size * record->count);
3013 text->start = record->pos;
3014 text->buffer = (recode_to_utf8
3015 ? recode_substring_pool ("UTF-8", r->encoding, raw, r->pool)
3018 text->n_warnings = 0;
3019 text->recoded = recode_to_utf8;
3024 /* Closes TEXT, frees its storage, and issues a final warning
3025 about suppressed warnings if necessary. */
3027 close_text_record (struct sfm_reader *r, struct text_record *text)
3029 if (text->n_warnings > MAX_TEXT_WARNINGS)
3030 sys_warn (r, -1, _("Suppressed %d additional related warnings."),
3031 text->n_warnings - MAX_TEXT_WARNINGS);
3033 pool_free (r->pool, ss_data (text->buffer));
3036 /* Reads a variable=value pair from TEXT.
3037 Looks up the variable in DICT and stores it into *VAR.
3038 Stores a null-terminated value into *VALUE. */
3040 read_variable_to_value_pair (struct sfm_reader *r, struct dictionary *dict,
3041 struct text_record *text,
3042 struct variable **var, char **value)
3046 if (!text_read_short_name (r, dict, text, ss_cstr ("="), var))
3049 *value = text_get_token (text, ss_buffer ("\t\0", 2), NULL);
3053 text->pos += ss_span (ss_substr (text->buffer, text->pos, SIZE_MAX),
3054 ss_buffer ("\t\0", 2));
3062 text_read_variable_name (struct sfm_reader *r, struct dictionary *dict,
3063 struct text_record *text, struct substring delimiters,
3064 struct variable **var)
3068 name = text_get_token (text, delimiters, NULL);
3072 *var = dict_lookup_var (dict, name);
3076 text_warn (r, text, _("Dictionary record refers to unknown variable %s."),
3083 text_read_short_name (struct sfm_reader *r, struct dictionary *dict,
3084 struct text_record *text, struct substring delimiters,
3085 struct variable **var)
3087 char *short_name = text_get_token (text, delimiters, NULL);
3088 if (short_name == NULL)
3091 *var = dict_lookup_var (dict, short_name);
3093 text_warn (r, text, _("Dictionary record refers to unknown variable %s."),
3098 /* Displays a warning for the current file position, limiting the
3099 number to MAX_TEXT_WARNINGS for TEXT. */
3101 text_warn (struct sfm_reader *r, struct text_record *text,
3102 const char *format, ...)
3104 if (text->n_warnings++ < MAX_TEXT_WARNINGS)
3108 va_start (args, format);
3109 sys_msg (r, text->start + text->pos, MW, format, args);
3115 text_get_token (struct text_record *text, struct substring delimiters,
3118 struct substring token;
3121 if (!ss_tokenize (text->buffer, delimiters, &text->pos, &token))
3123 if (delimiter != NULL)
3124 *delimiter = ss_data (text->buffer)[text->pos-1];
3128 end = &ss_data (token)[ss_length (token)];
3129 if (delimiter != NULL)
3132 return ss_data (token);
3135 /* Reads a integer value expressed in decimal, then a space, then a string that
3136 consists of exactly as many bytes as specified by the integer, then a space,
3137 from TEXT. Returns the string, null-terminated, as a subset of TEXT's
3138 buffer (so the caller should not free the string). */
3140 text_parse_counted_string (struct sfm_reader *r, struct text_record *text)
3148 while (text->pos < text->buffer.length)
3150 int c = text->buffer.string[text->pos];
3151 if (c < '0' || c > '9')
3153 n = (n * 10) + (c - '0');
3156 if (text->pos >= text->buffer.length || start == text->pos)
3158 sys_warn (r, text->start,
3159 _("Expecting digit at offset %zu in MRSETS record."),
3164 if (!text_match (text, ' '))
3166 sys_warn (r, text->start,
3167 _("Expecting space at offset %zu in MRSETS record."),
3172 if (text->pos + n > text->buffer.length)
3174 sys_warn (r, text->start,
3175 _("%zu-byte string starting at offset %zu "
3176 "exceeds record length %zu."),
3177 n, text->pos, text->buffer.length);
3181 s = &text->buffer.string[text->pos];
3184 sys_warn (r, text->start,
3185 _("Expecting space at offset %zu following %zu-byte string."),
3195 text_match (struct text_record *text, char c)
3197 if (text->pos >= text->buffer.length)
3200 if (text->buffer.string[text->pos] == c)
3209 /* Returns the current byte offset (as converted to UTF-8, if it was converted)
3210 inside the TEXT's string. */
3212 text_pos (const struct text_record *text)
3218 text_get_all (const struct text_record *text)
3220 return text->buffer.string;
3225 /* Displays a corruption message. */
3227 sys_msg (struct sfm_reader *r, off_t offset,
3228 int class, const char *format, va_list args)
3232 ds_init_empty (&text);
3234 ds_put_format (&text, _("`%s' near offset 0x%llx: "),
3235 fh_get_file_name (r->fh), (long long int) offset);
3237 ds_put_format (&text, _("`%s': "), fh_get_file_name (r->fh));
3238 ds_put_vformat (&text, format, args);
3241 .category = msg_class_to_category (class),
3242 .severity = msg_class_to_severity (class),
3243 .text = ds_cstr (&text),
3248 /* Displays a warning for offset OFFSET in the file. */
3250 sys_warn (struct sfm_reader *r, off_t offset, const char *format, ...)
3254 va_start (args, format);
3255 sys_msg (r, offset, MW, format, args);
3259 /* Displays an error for the current file position and marks it as in an error
3262 sys_error (struct sfm_reader *r, off_t offset, const char *format, ...)
3266 va_start (args, format);
3267 sys_msg (r, offset, ME, format, args);
3273 /* Reads BYTE_CNT bytes into BUF.
3274 Returns 1 if exactly BYTE_CNT bytes are successfully read.
3275 Returns -1 if an I/O error or a partial read occurs.
3276 Returns 0 for an immediate end-of-file and, if EOF_IS_OK is false, reports
3279 read_bytes_internal (struct sfm_reader *r, bool eof_is_ok,
3280 void *buf, size_t byte_cnt)
3282 size_t bytes_read = fread (buf, 1, byte_cnt, r->file);
3283 r->pos += bytes_read;
3284 if (bytes_read == byte_cnt)
3286 else if (ferror (r->file))
3288 sys_error (r, r->pos, _("System error: %s."), strerror (errno));
3291 else if (!eof_is_ok || bytes_read != 0)
3293 sys_error (r, r->pos, _("Unexpected end of file."));
3300 /* Reads BYTE_CNT into BUF.
3301 Returns true if successful.
3302 Returns false upon I/O error or if end-of-file is encountered. */
3304 read_bytes (struct sfm_reader *r, void *buf, size_t byte_cnt)
3306 return read_bytes_internal (r, false, buf, byte_cnt) == 1;
3309 /* Reads BYTE_CNT bytes into BUF.
3310 Returns 1 if exactly BYTE_CNT bytes are successfully read.
3311 Returns 0 if an immediate end-of-file is encountered.
3312 Returns -1 if an I/O error or a partial read occurs. */
3314 try_read_bytes (struct sfm_reader *r, void *buf, size_t byte_cnt)
3316 return read_bytes_internal (r, true, buf, byte_cnt);
3319 /* Reads a 32-bit signed integer from R and stores its value in host format in
3320 *X. Returns true if successful, otherwise false. */
3322 read_int (struct sfm_reader *r, int *x)
3325 if (read_bytes (r, integer, sizeof integer) != 1)
3327 *x = integer_get (r->integer_format, integer, sizeof integer);
3332 read_uint (struct sfm_reader *r, unsigned int *x)
3337 ok = read_int (r, &y);
3342 /* Reads a 64-bit signed integer from R and returns its value in
3345 read_int64 (struct sfm_reader *r, long long int *x)
3348 if (read_bytes (r, integer, sizeof integer) != 1)
3350 *x = integer_get (r->integer_format, integer, sizeof integer);
3354 /* Reads a 64-bit signed integer from R and returns its value in
3357 read_uint64 (struct sfm_reader *r, unsigned long long int *x)
3362 ok = read_int64 (r, &y);
3368 parse_int (const struct sfm_reader *r, const void *data, size_t ofs)
3370 return integer_get (r->integer_format, (const uint8_t *) data + ofs, 4);
3374 parse_float (const struct sfm_reader *r, const void *data, size_t ofs)
3376 return float_get_double (r->float_format, (const uint8_t *) data + ofs);
3379 /* Reads exactly SIZE - 1 bytes into BUFFER
3380 and stores a null byte into BUFFER[SIZE - 1]. */
3382 read_string (struct sfm_reader *r, char *buffer, size_t size)
3387 ok = read_bytes (r, buffer, size - 1);
3389 buffer[size - 1] = '\0';
3393 /* Skips BYTES bytes forward in R. */
3395 skip_bytes (struct sfm_reader *r, size_t bytes)
3400 size_t chunk = MIN (sizeof buffer, bytes);
3401 if (!read_bytes (r, buffer, chunk))
3409 /* Returns a malloc()'d copy of S in which all lone CRs and CR LF pairs have
3410 been replaced by LFs.
3412 (A product that identifies itself as VOXCO INTERVIEWER 4.3 produces system
3413 files that use CR-only line ends in the file label and extra product
3416 fix_line_ends (const char *s)
3420 d = dst = xmalloc (strlen (s) + 1);
3439 read_ztrailer (struct sfm_reader *r,
3440 long long int zheader_ofs,
3441 long long int ztrailer_len);
3444 zalloc (voidpf pool_, uInt items, uInt size)
3446 struct pool *pool = pool_;
3448 return (!size || xalloc_oversized (items, size)
3450 : pool_malloc (pool, items * size));
3454 zfree (voidpf pool_, voidpf address)
3456 struct pool *pool = pool_;
3458 pool_free (pool, address);
3462 read_zheader (struct sfm_reader *r)
3465 long long int zheader_ofs;
3466 long long int ztrailer_ofs;
3467 long long int ztrailer_len;
3469 if (!read_int64 (r, &zheader_ofs)
3470 || !read_int64 (r, &ztrailer_ofs)
3471 || !read_int64 (r, &ztrailer_len))
3474 if (zheader_ofs != pos)
3476 sys_error (r, pos, _("Wrong ZLIB data header offset %#llx "
3477 "(expected %#llx)."),
3478 zheader_ofs, (long long int) pos);
3482 if (ztrailer_ofs < r->pos)
3484 sys_error (r, pos, _("Impossible ZLIB trailer offset 0x%llx."),
3489 if (ztrailer_len < 24 || ztrailer_len % 24)
3491 sys_error (r, pos, _("Invalid ZLIB trailer length %lld."), ztrailer_len);
3495 r->ztrailer_ofs = ztrailer_ofs;
3496 if (!read_ztrailer (r, zheader_ofs, ztrailer_len))
3499 if (r->zin_buf == NULL)
3501 r->zin_buf = pool_malloc (r->pool, ZIN_BUF_SIZE);
3502 r->zout_buf = pool_malloc (r->pool, ZOUT_BUF_SIZE);
3503 r->zstream.next_in = NULL;
3504 r->zstream.avail_in = 0;
3507 r->zstream.zalloc = zalloc;
3508 r->zstream.zfree = zfree;
3509 r->zstream.opaque = r->pool;
3511 return open_zstream (r);
3515 seek (struct sfm_reader *r, off_t offset)
3517 if (fseeko (r->file, offset, SEEK_SET))
3518 sys_error (r, 0, _("%s: seek failed (%s)."),
3519 fh_get_file_name (r->fh), strerror (errno));
3523 /* Performs some additional consistency checks on the ZLIB compressed data
3526 read_ztrailer (struct sfm_reader *r,
3527 long long int zheader_ofs,
3528 long long int ztrailer_len)
3530 long long int expected_uncmp_ofs;
3531 long long int expected_cmp_ofs;
3534 unsigned int block_size;
3535 unsigned int n_blocks;
3539 if (fstat (fileno (r->file), &s))
3541 sys_error (r, 0, _("%s: stat failed (%s)."),
3542 fh_get_file_name (r->fh), strerror (errno));
3546 if (!S_ISREG (s.st_mode))
3548 /* We can't seek to the trailer and then back to the data in this file,
3549 so skip doing extra checks. */
3553 if (r->ztrailer_ofs + ztrailer_len != s.st_size)
3554 sys_warn (r, r->pos,
3555 _("End of ZLIB trailer (0x%llx) is not file size (0x%llx)."),
3556 r->ztrailer_ofs + ztrailer_len, (long long int) s.st_size);
3558 seek (r, r->ztrailer_ofs);
3560 /* Read fixed header from ZLIB data trailer. */
3561 if (!read_int64 (r, &bias))
3563 if (-bias != r->bias)
3565 sys_error (r, r->pos, _("ZLIB trailer bias (%lld) differs from "
3566 "file header bias (%.2f)."),
3571 if (!read_int64 (r, &zero))
3574 sys_warn (r, r->pos,
3575 _("ZLIB trailer \"zero\" field has nonzero value %lld."), zero);
3577 if (!read_uint (r, &block_size))
3579 if (block_size != ZBLOCK_SIZE)
3580 sys_warn (r, r->pos,
3581 _("ZLIB trailer specifies unexpected %u-byte block size."),
3584 if (!read_uint (r, &n_blocks))
3586 if (n_blocks != (ztrailer_len - 24) / 24)
3588 sys_error (r, r->pos,
3589 _("%lld-byte ZLIB trailer specifies %u data blocks (expected "
3591 ztrailer_len, n_blocks, (ztrailer_len - 24) / 24);
3595 expected_uncmp_ofs = zheader_ofs;
3596 expected_cmp_ofs = zheader_ofs + 24;
3597 for (i = 0; i < n_blocks; i++)
3599 off_t desc_ofs = r->pos;
3600 unsigned long long int uncompressed_ofs;
3601 unsigned long long int compressed_ofs;
3602 unsigned int uncompressed_size;
3603 unsigned int compressed_size;
3605 if (!read_uint64 (r, &uncompressed_ofs)
3606 || !read_uint64 (r, &compressed_ofs)
3607 || !read_uint (r, &uncompressed_size)
3608 || !read_uint (r, &compressed_size))
3611 if (uncompressed_ofs != expected_uncmp_ofs)
3613 sys_error (r, desc_ofs,
3614 _("ZLIB block descriptor %u reported uncompressed data "
3615 "offset %#llx, when %#llx was expected."),
3616 i, uncompressed_ofs, expected_uncmp_ofs);
3620 if (compressed_ofs != expected_cmp_ofs)
3622 sys_error (r, desc_ofs,
3623 _("ZLIB block descriptor %u reported compressed data "
3624 "offset %#llx, when %#llx was expected."),
3625 i, compressed_ofs, expected_cmp_ofs);
3629 if (i < n_blocks - 1)
3631 if (uncompressed_size != block_size)
3632 sys_warn (r, desc_ofs,
3633 _("ZLIB block descriptor %u reported block size %#x, "
3634 "when %#x was expected."),
3635 i, uncompressed_size, block_size);
3639 if (uncompressed_size > block_size)
3640 sys_warn (r, desc_ofs,
3641 _("ZLIB block descriptor %u reported block size %#x, "
3642 "when at most %#x was expected."),
3643 i, uncompressed_size, block_size);
3646 /* http://www.zlib.net/zlib_tech.html says that the maximum expansion
3647 from compression, with worst-case parameters, is 13.5% plus 11 bytes.
3648 This code checks for an expansion of more than 14.3% plus 11
3650 if (compressed_size > uncompressed_size + uncompressed_size / 7 + 11)
3652 sys_error (r, desc_ofs,
3653 _("ZLIB block descriptor %u reports compressed size %u "
3654 "and uncompressed size %u."),
3655 i, compressed_size, uncompressed_size);
3659 expected_uncmp_ofs += uncompressed_size;
3660 expected_cmp_ofs += compressed_size;
3663 if (expected_cmp_ofs != r->ztrailer_ofs)
3665 sys_error (r, r->pos, _("ZLIB trailer is at offset %#llx but %#llx "
3666 "would be expected from block descriptors."),
3667 r->ztrailer_ofs, expected_cmp_ofs);
3671 seek (r, zheader_ofs + 24);
3676 open_zstream (struct sfm_reader *r)
3680 r->zout_pos = r->zout_end = 0;
3681 error = inflateInit (&r->zstream);
3684 sys_error (r, r->pos, _("ZLIB initialization failed (%s)."),
3692 close_zstream (struct sfm_reader *r)
3696 error = inflateEnd (&r->zstream);
3699 sys_error (r, r->pos, _("Inconsistency at end of ZLIB stream (%s)."),
3707 read_bytes_zlib (struct sfm_reader *r, void *buf_, size_t byte_cnt)
3709 uint8_t *buf = buf_;
3718 /* Use already inflated data if there is any. */
3719 if (r->zout_pos < r->zout_end)
3721 unsigned int n = MIN (byte_cnt, r->zout_end - r->zout_pos);
3722 memcpy (buf, &r->zout_buf[r->zout_pos], n);
3731 /* We need to inflate some more data.
3732 Get some more input data if we don't have any. */
3733 if (r->zstream.avail_in == 0)
3735 unsigned int n = MIN (ZIN_BUF_SIZE, r->ztrailer_ofs - r->pos);
3740 int retval = try_read_bytes (r, r->zin_buf, n);
3743 r->zstream.avail_in = n;
3744 r->zstream.next_in = r->zin_buf;
3748 /* Inflate the (remaining) input data. */
3749 r->zstream.avail_out = ZOUT_BUF_SIZE;
3750 r->zstream.next_out = r->zout_buf;
3751 error = inflate (&r->zstream, Z_SYNC_FLUSH);
3753 r->zout_end = r->zstream.next_out - r->zout_buf;
3754 if (r->zout_end == 0)
3756 if (error != Z_STREAM_END)
3758 sys_error (r, r->pos, _("ZLIB stream inconsistency (%s)."),
3762 else if (!close_zstream (r) || !open_zstream (r))
3767 /* Process the output data and ignore 'error' for now. ZLIB will
3768 present it to us again on the next inflate() call. */
3774 read_compressed_bytes (struct sfm_reader *r, void *buf, size_t byte_cnt)
3776 if (r->compression == ANY_COMP_SIMPLE)
3777 return read_bytes (r, buf, byte_cnt);
3780 int retval = read_bytes_zlib (r, buf, byte_cnt);
3782 sys_error (r, r->pos, _("Unexpected end of ZLIB compressed data."));
3788 try_read_compressed_bytes (struct sfm_reader *r, void *buf, size_t byte_cnt)
3790 if (r->compression == ANY_COMP_SIMPLE)
3791 return try_read_bytes (r, buf, byte_cnt);
3793 return read_bytes_zlib (r, buf, byte_cnt);
3796 /* Reads a 64-bit floating-point number from R and returns its
3797 value in host format. */
3799 read_compressed_float (struct sfm_reader *r, double *d)
3803 if (!read_compressed_bytes (r, number, sizeof number))
3806 *d = float_get_double (r->float_format, number);
3810 static const struct casereader_class sys_file_casereader_class =
3812 sys_file_casereader_read,
3813 sys_file_casereader_destroy,
3818 const struct any_reader_class sys_file_reader_class =
3820 N_("SPSS System File"),