1 @c PSPP - a program for statistical analysis.
2 @c Copyright (C) 2019 Free Software Foundation, Inc.
3 @c Permission is granted to copy, distribute and/or modify this document
4 @c under the terms of the GNU Free Documentation License, Version 1.3
5 @c or any later version published by the Free Software Foundation;
6 @c with no Invariant Sections, no Front-Cover Texts, and no Back-Cover Texts.
7 @c A copy of the license is included in the section entitled "GNU
8 @c Free Documentation License".
11 @node System File Format
12 @appendix System File Format
14 A system file encapsulates a set of cases and dictionary information
15 that describes how they may be interpreted. This chapter describes
16 the format of a system file.
18 System files use four data types: 8-bit characters, 32-bit integers,
20 and 64-bit floating points, called here @code{char}, @code{int32},
22 @code{flt64}, respectively. Data is not necessarily aligned on a word
23 or double-word boundary: the long variable name record (@pxref{Long
24 Variable Names Record}) and very long string records (@pxref{Very Long
25 String Record}) have arbitrary byte length and can therefore cause all
26 data coming after them in the file to be misaligned.
28 Integer data in system files may be big-endian or little-endian. A
29 reader may detect the endianness of a system file by examining
30 @code{layout_code} in the file header record
31 (@pxref{layout_code,,@code{layout_code}}).
33 Floating-point data in system files may nominally be in IEEE 754, IBM,
34 or VAX formats. A reader may detect the floating-point format in use
35 by examining @code{bias} in the file header record
36 (@pxref{bias,,@code{bias}}).
38 PSPP detects big-endian and little-endian integer formats in system
39 files and translates as necessary. PSPP also detects the
40 floating-point format in use, as well as the endianness of IEEE 754
41 floating-point numbers, and translates as needed. However, only IEEE
42 754 numbers with the same endianness as integer data in the same file
43 have actually been observed in system files, and it is likely that
44 other formats are obsolete or were never used.
46 System files use a few floating point values for special purposes:
50 The system-missing value is represented by the largest possible
51 negative number in the floating point format (@code{-DBL_MAX}).
54 HIGHEST is used as the high end of a missing value range with an
55 unbounded maximum. It is represented by the largest possible positive
56 number (@code{DBL_MAX}).
59 LOWEST is used as the low end of a missing value range with an
60 unbounded minimum. It was originally represented by the
61 second-largest negative number (in IEEE 754 format,
62 @code{0xffeffffffffffffe}). System files written by SPSS 21 and later
63 instead use the largest negative number (@code{-DBL_MAX}), the same
64 value as SYSMIS. This does not lead to ambiguity because LOWEST
65 appears in system files only in missing value ranges, which never
69 System files may use most character encodings based on an 8-bit unit.
70 UTF-16 and UTF-32, based on wider units, appear to be unacceptable.
71 @code{rec_type} in the file header record is sufficient to distinguish
72 between ASCII and EBCDIC based encodings. The best way to determine
73 the specific encoding in use is to consult the character encoding
74 record (@pxref{Character Encoding Record}), if present, and failing
75 that the @code{character_code} in the machine integer info record
76 (@pxref{Machine Integer Info Record}). The same encoding should be
77 used for the dictionary and the data in the file, although it is
78 possible to artificially synthesize files that use different encodings
79 (@pxref{Character Encoding Record}).
82 * System File Record Structure::
83 * File Header Record::
85 * Value Labels Records::
87 * Machine Integer Info Record::
88 * Machine Floating-Point Info Record::
89 * Multiple Response Sets Records::
90 * Extra Product Info Record::
91 * Variable Display Parameter Record::
92 * Long Variable Names Record::
93 * Very Long String Record::
94 * Character Encoding Record::
95 * Long String Value Labels Record::
96 * Long String Missing Values Record::
97 * Data File and Variable Attributes Records::
98 * Extended Number of Cases Record::
99 * Other Informational Records::
100 * Dictionary Termination Record::
104 @node System File Record Structure
105 @section System File Record Structure
107 System files are divided into records with the following format:
114 This header does not identify the length of the @code{data} or any
115 information about what it contains, so the system file reader must
116 understand the format of @code{data} based on @code{type}. However,
117 records with type 7, called @dfn{extension records}, have a stricter
125 char data[size * count];
129 @item int32 rec_type;
130 Record type. Always set to 7.
133 Record subtype. This value identifies a particular kind of extension
137 The size of each piece of data that follows the header, in bytes.
138 Known extension records use 1, 4, or 8, for @code{char}, @code{int32},
139 and @code{flt64} format data, respectively.
142 The number of pieces of data that follow the header.
144 @item char data[size * count];
145 Data, whose format and interpretation depend on the subtype.
148 An extension record contains exactly @code{size * count} bytes of
149 data, which allows a reader that does not understand an extension
150 record to skip it. Extension records provide only nonessential
151 information, so this allows for files written by newer software to
152 preserve backward compatibility with older or less capable readers.
154 Records in a system file must appear in the following order:
164 All pairs of value labels records and value label variables records,
168 Document record, if present.
171 Extension (type 7) records, in ascending numerical order of their
174 System files written by SPSS include at most one of each kind of
175 extension record. This is generally true of system files written by
176 other software as well, with known exceptions noted below in the
177 individual sections about each type of record.
180 Dictionary termination record.
186 We advise authors of programs that read system files to tolerate
187 format variations. Various kinds of misformatting and corruption have
188 been observed in system files written by SPSS and other software
189 alike. In particular, because extension records provide nonessential
190 information, it is generally better to ignore an extension record
191 entirely than to refuse to read a system file.
193 The following sections describe the known kinds of records.
195 @node File Header Record
196 @section File Header Record
198 A system file begins with the file header, with the following format:
204 int32 nominal_case_size;
209 char creation_date[9];
210 char creation_time[8];
216 @item char rec_type[4];
217 Record type code, either @samp{$FL2} for system files with
218 uncompressed data or data compressed with simple bytecode compression,
219 or @samp{$FL3} for system files with ZLIB compressed data.
221 This is truly a character field that uses the character encoding as
222 other strings. Thus, in a file with an ASCII-based character encoding
223 this field contains @code{24 46 4c 32} or @code{24 46 4c 33}, and in a
224 file with an EBCDIC-based encoding this field contains @code{5b c6 d3
225 f2}. (No EBCDIC-based ZLIB-compressed files have been observed.)
227 @item char prod_name[60];
228 Product identification string. This always begins with the characters
229 @samp{@@(#) SPSS DATA FILE}. PSPP uses the remaining characters to
230 give its version and the operating system name; for example, @samp{GNU
231 pspp 0.1.4 - sparc-sun-solaris2.5.2}. The string is truncated if it
232 would be longer than 60 characters; otherwise it is padded on the right
235 The product name field allow readers to behave differently based on
236 quirks in the way that particular software writes system files.
237 @xref{Value Labels Records}, for the detail of the quirk that the PSPP
238 system file reader tolerates in files written by ReadStat, which has
239 @code{https://github.com/WizardMac/ReadStat} in @code{prod_name}.
242 @item int32 layout_code;
243 Normally set to 2, although a few system files have been spotted in
244 the wild with a value of 3 here. PSPP use this value to determine the
245 file's integer endianness (@pxref{System File Format}).
247 @item int32 nominal_case_size;
248 Number of data elements per case. This is the number of variables,
249 except that long string variables add extra data elements (one for every
250 8 characters after the first 8). However, string variables do not
251 contribute to this value beyond the first 255 bytes. Further, some
252 software always writes -1 or 0 in this field. In general, it is
253 unsafe for systems reading system files to rely upon this value.
255 @item int32 compression;
256 Set to 0 if the data in the file is not compressed, 1 if the data is
257 compressed with simple bytecode compression, 2 if the data is ZLIB
258 compressed. This field has value 2 if and only if @code{rec_type} is
261 @item int32 weight_index;
262 If one of the variables in the data set is used as a weighting
263 variable, set to the dictionary index of that variable, plus 1
264 (@pxref{Dictionary Index}). Otherwise, set to 0.
267 Set to the number of cases in the file if it is known, or -1 otherwise.
269 In the general case it is not possible to determine the number of cases
270 that will be output to a system file at the time that the header is
271 written. The way that this is dealt with is by writing the entire
272 system file, including the header, then seeking back to the beginning of
273 the file and writing just the @code{ncases} field. For files in which
274 this is not valid, the seek operation fails. In this case,
275 @code{ncases} remains -1.
279 Compression bias, ordinarily set to 100. Only integers between
280 @code{1 - bias} and @code{251 - bias} can be compressed.
282 By assuming that its value is 100, PSPP uses @code{bias} to determine
283 the file's floating-point format and endianness (@pxref{System File
284 Format}). If the compression bias is not 100, PSPP cannot auto-detect
285 the floating-point format and assumes that it is IEEE 754 format with
286 the same endianness as the system file's integers, which is correct
287 for all known system files.
289 @item char creation_date[9];
290 Date of creation of the system file, in @samp{dd mmm yy}
291 format, with the month as standard English abbreviations, using an
292 initial capital letter and following with lowercase. If the date is not
293 available then this field is arbitrarily set to @samp{01 Jan 70}.
295 @item char creation_time[8];
296 Time of creation of the system file, in @samp{hh:mm:ss}
297 format and using 24-hour time. If the time is not available then this
298 field is arbitrarily set to @samp{00:00:00}.
300 @item char file_label[64];
301 File label declared by the user, if any (@pxref{FILE LABEL,,,pspp,
302 PSPP Users Guide}). Padded on the right with spaces.
304 A product that identifies itself as @code{VOXCO INTERVIEWER 4.3} uses
305 CR-only line ends in this field, rather than the more usual LF-only or
308 @item char padding[3];
309 Ignored padding bytes to make the structure a multiple of 32 bits in
310 length. Set to zeros.
313 @node Variable Record
314 @section Variable Record
316 There must be one variable record for each numeric variable and each
317 string variable with width 8 bytes or less. String variables wider
318 than 8 bytes have one variable record for each 8 bytes, rounding up.
319 The first variable record for a long string specifies the variable's
320 correct dictionary information. Subsequent variable records for a
321 long string are filled with dummy information: a type of -1, no
322 variable label or missing values, print and write formats that are
323 ignored, and an empty string as name. A few system files have been
324 encountered that include a variable label on dummy variable records,
325 so readers should take care to parse dummy variable records in the
326 same way as other variable records.
328 @anchor{Dictionary Index}
329 The @dfn{dictionary index} of a variable is a 1-based offset in the set of
330 variable records, including dummy variable records for long string
331 variables. The first variable record has a dictionary index of 1, the
332 second has a dictionary index of 2, and so on.
334 The system file format does not directly support string variables
335 wider than 255 bytes. Such very long string variables are represented
336 by a number of narrower string variables. @xref{Very Long String
337 Record}, for details.
339 A system file should contain at least one variable and thus at least
340 one variable record, but system files have been observed in the wild
341 without any variables (thus, no data either).
347 int32 n_missing_values;
352 /* @r{Present only if @code{has_var_label} is 1.} */
356 /* @r{Present only if @code{n_missing_values} is nonzero}. */
357 flt64 missing_values[];
361 @item int32 rec_type;
362 Record type code. Always set to 2.
365 Variable type code. Set to 0 for a numeric variable. For a short
366 string variable or the first part of a long string variable, this is set
367 to the width of the string. For the second and subsequent parts of a
368 long string variable, set to -1, and the remaining fields in the
369 structure are ignored.
371 @item int32 has_var_label;
372 If this variable has a variable label, set to 1; otherwise, set to 0.
374 @item int32 n_missing_values;
375 If the variable has no missing values, set to 0. If the variable has
376 one, two, or three discrete missing values, set to 1, 2, or 3,
377 respectively. If the variable has a range for missing variables, set to
378 -2; if the variable has a range for missing variables plus a single
379 discrete value, set to -3.
381 A long string variable always has the value 0 here. A separate record
382 indicates missing values for long string variables (@pxref{Long String
383 Missing Values Record}).
386 Print format for this variable. See below.
389 Write format for this variable. See below.
392 Variable name. The variable name must begin with a capital letter or
393 the at-sign (@samp{@@}). Subsequent characters may also be digits, octothorpes
394 (@samp{#}), dollar signs (@samp{$}), underscores (@samp{_}), or full
395 stops (@samp{.}). The variable name is padded on the right with spaces.
397 The @samp{name} fields should be unique within a system file. System
398 files written by SPSS that contain very long string variables with
399 similar names sometimes contain duplicate names that are later
400 eliminated by resolving the very long string names (@pxref{Very Long
401 String Record}). PSPP handles duplicates by assigning them new,
404 @item int32 label_len;
405 This field is present only if @code{has_var_label} is set to 1. It is
406 set to the length, in characters, of the variable label. The
407 documented maximum length varies from 120 to 255 based on SPSS
408 version, but some files have been seen with longer labels. PSPP
409 accepts labels of any length.
412 This field is present only if @code{has_var_label} is set to 1. It has
413 length @code{label_len}, rounded up to the nearest multiple of 32 bits.
414 The first @code{label_len} characters are the variable's variable label.
416 @item flt64 missing_values[];
417 This field is present only if @code{n_missing_values} is nonzero. It
418 has the same number of 8-byte elements as the absolute value of
419 @code{n_missing_values}. Each element is interpreted as a number for
420 numeric variables (with HIGHEST and LOWEST indicated as described in
421 the chapter introduction). For string variables of width less than 8
422 bytes, elements are right-padded with spaces; for string variables
423 wider than 8 bytes, only the first 8 bytes of each missing value are
424 specified, with the remainder implicitly all spaces.
426 For discrete missing values, each element represents one missing
427 value. When a range is present, the first element denotes the minimum
428 value in the range, and the second element denotes the maximum value
429 in the range. When a range plus a value are present, the third
430 element denotes the additional discrete missing value.
433 @anchor{System File Output Formats}
434 The @code{print} and @code{write} members of sysfile_variable are output
435 formats coded into @code{int32} types. The least-significant byte
436 of the @code{int32} represents the number of decimal places, and the
437 next two bytes in order of increasing significance represent field width
438 and format type, respectively. The most-significant byte is not
439 used and should be set to zero.
441 Format types are defined as follows:
444 @multitable {Value} {@code{DATETIME}}
534 A few system files have been observed in the wild with invalid
535 @code{write} fields, in particular with value 0. Readers should
536 probably treat invalid @code{print} or @code{write} fields as some
539 @node Value Labels Records
540 @section Value Labels Records
542 The value label records documented in this section are used for
543 numeric and short string variables only. Long string variables may
544 have value labels, but their value labels are recorded using a
545 different record type (@pxref{Long String Value Labels Record}).
547 ReadStat (@pxref{File Header Record}) writes value labels that label a
548 single value more than once. In more detail, it emits value labels
549 whose values are longer than string variables' widths, that are
550 identical in the actual width of the variable, e.g.@: labels for
551 values @code{ABC123} and @code{ABC456} for a string variable with
552 width 3. For files written by this software, PSPP ignores such
555 The value label record has the following format:
561 /* @r{Repeated @code{n_label} times}. */
568 @item int32 rec_type;
569 Record type. Always set to 3.
571 @item int32 label_count;
572 Number of value labels present in this record.
575 The remaining fields are repeated @code{count} times. Each
576 repetition specifies one value label.
580 A numeric value or a short string value padded as necessary to 8 bytes
581 in length. Its type and width cannot be determined until the
582 following value label variables record (see below) is read.
584 @item char label_len;
585 The label's length, in bytes. The documented maximum length varies
586 from 60 to 120 based on SPSS version. PSPP supports value labels up
590 @code{label_len} bytes of the actual label, followed by up to 7 bytes
591 of padding to bring @code{label} and @code{label_len} together to a
592 multiple of 8 bytes in length.
595 The value label record is always immediately followed by a value label
596 variables record with the following format:
605 @item int32 rec_type;
606 Record type. Always set to 4.
608 @item int32 var_count;
609 Number of variables that the associated value labels from the value
610 label record are to be applied.
613 A list of 1-based dictionary indexes of variables to which to apply the value
614 labels (@pxref{Dictionary Index}). There are @code{var_count}
617 String variables wider than 8 bytes may not be specified in this list.
620 @node Document Record
621 @section Document Record
623 The document record, if present, has the following format:
632 @item int32 rec_type;
633 Record type. Always set to 6.
636 Number of lines of documents present. This should be greater than
637 zero, but ReadStats writes system files with zero @code{n_lines}.
639 @item char lines[][80];
640 Document lines. The number of elements is defined by @code{n_lines}.
641 Lines shorter than 80 characters are padded on the right with spaces.
644 @node Machine Integer Info Record
645 @section Machine Integer Info Record
647 The integer info record, if present, has the following format:
659 int32 version_revision;
661 int32 floating_point_rep;
662 int32 compression_code;
664 int32 character_code;
668 @item int32 rec_type;
669 Record type. Always set to 7.
672 Record subtype. Always set to 3.
675 Size of each piece of data in the data part, in bytes. Always set to 4.
678 Number of pieces of data in the data part. Always set to 8.
680 @item int32 version_major;
681 PSPP major version number. In version @var{x}.@var{y}.@var{z}, this
684 @item int32 version_minor;
685 PSPP minor version number. In version @var{x}.@var{y}.@var{z}, this
688 @item int32 version_revision;
689 PSPP version revision number. In version @var{x}.@var{y}.@var{z},
692 @item int32 machine_code;
693 Machine code. PSPP always set this field to value to -1, but other
696 @item int32 floating_point_rep;
697 Floating point representation code. For IEEE 754 systems this is 1.
698 IBM 370 sets this to 2, and DEC VAX E to 3.
700 @item int32 compression_code;
701 Compression code. Always set to 1, regardless of whether or how the
704 @item int32 endianness;
705 Machine endianness. 1 indicates big-endian, 2 indicates little-endian.
707 @item int32 character_code;
708 @anchor{character-code} Character code. The following values have
709 been actually observed in system files:
719 The @code{windows-1250} code page for Central European and Eastern
723 The @code{windows-1252} code page for Western European languages.
732 The following additional values are known to be defined:
742 Other Windows code page numbers are known to be generally valid.
744 Old versions of SPSS for Unix and Windows always wrote value 2 in this
745 field, regardless of the encoding in use. Newer versions also write
746 the character encoding as a string (see @ref{Character Encoding
750 @node Machine Floating-Point Info Record
751 @section Machine Floating-Point Info Record
753 The floating-point info record, if present, has the following format:
769 @item int32 rec_type;
770 Record type. Always set to 7.
773 Record subtype. Always set to 4.
776 Size of each piece of data in the data part, in bytes. Always set to 8.
779 Number of pieces of data in the data part. Always set to 3.
782 @itemx flt64 highest;
784 The system missing value, the value used for HIGHEST in missing
785 values, and the value used for LOWEST in missing values, respectively.
786 @xref{System File Format}, for more information.
788 The SPSSWriter library in PHP, which identifies itself as @code{FOM
789 SPSS 1.0.0} in the file header record @code{prod_name} field, writes
790 unexpected values to these fields, but it uses the same values
791 consistently throughout the rest of the file.
794 @node Multiple Response Sets Records
795 @section Multiple Response Sets Records
797 The system file format has two different types of records that
798 represent multiple response sets (@pxref{MRSETS,,,pspp, PSPP Users
799 Guide}). The first type of record describes multiple response sets
800 that can be understood by SPSS before version 14. The second type of
801 record, with a closely related format, is used for multiple dichotomy
802 sets that use the CATEGORYLABELS=COUNTEDVALUES feature added in
812 /* @r{Exactly @code{count} bytes of data.} */
817 @item int32 rec_type;
818 Record type. Always set to 7.
821 Record subtype. Set to 7 for records that describe multiple response
822 sets understood by SPSS before version 14, or to 19 for records that
823 describe dichotomy sets that use the CATEGORYLABELS=COUNTEDVALUES
824 feature added in version 14.
827 The size of each element in the @code{mrsets} member. Always set to 1.
830 The total number of bytes in @code{mrsets}.
833 Zero or more line feeds (byte 0x0a), followed by a series of multiple
834 response sets, each of which consists of the following:
838 The set's name (an identifier that begins with @samp{$}), in mixed
839 upper and lower case.
842 An equals sign (@samp{=}).
845 @samp{C} for a multiple category set, @samp{D} for a multiple
846 dichotomy set with CATEGORYLABELS=VARLABELS, or @samp{E} for a
847 multiple dichotomy set with CATEGORYLABELS=COUNTEDVALUES.
850 For a multiple dichotomy set with CATEGORYLABELS=COUNTEDVALUES, a
851 space, followed by a number expressed as decimal digits, followed by a
852 space. If LABELSOURCE=VARLABEL was specified on MRSETS, then the
853 number is 11; otherwise it is 1.@footnote{This part of the format may
854 not be fully understood, because only a single example of each
855 possibility has been examined.}
858 For either kind of multiple dichotomy set, the counted value, as a
859 positive integer count specified as decimal digits, followed by a
860 space, followed by as many string bytes as specified in the count. If
861 the set contains numeric variables, the string consists of the counted
862 integer value expressed as decimal digits. If the set contains string
863 variables, the string contains the counted string value. Either way,
864 the string may be padded on the right with spaces (older versions of
865 SPSS seem to always pad to a width of 8 bytes; newer versions don't).
871 The multiple response set's label, using the same format as for the
872 counted value for multiple dichotomy sets. A string of length 0 means
873 that the set does not have a label. A string of length 0 is also
874 written if LABELSOURCE=VARLABEL was specified.
880 The short names of the variables in the set, converted to lowercase,
881 each separated from the previous by a single space.
883 Even though a multiple response set must have at least two variables,
884 some system files contain multiple response sets with no variables or
885 one variable. The source and meaning of these multiple response sets is
886 unknown. (Perhaps they arise from creating a multiple response set
887 then deleting all the variables that it contains?)
890 One line feed (byte 0x0a). Sometimes multiple, even hundreds, of line
895 Example: Given appropriate variable definitions, consider the
896 following MRSETS command:
899 MRSETS /MCGROUP NAME=$a LABEL='my mcgroup' VARIABLES=a b c
900 /MDGROUP NAME=$b VARIABLES=g e f d VALUE=55
901 /MDGROUP NAME=$c LABEL='mdgroup #2' VARIABLES=h i j VALUE='Yes'
902 /MDGROUP NAME=$d LABEL='third mdgroup' CATEGORYLABELS=COUNTEDVALUES
903 VARIABLES=k l m VALUE=34
904 /MDGROUP NAME=$e CATEGORYLABELS=COUNTEDVALUES LABELSOURCE=VARLABEL
905 VARIABLES=n o p VALUE='choice'.
908 The above would generate the following multiple response set record of
912 $a=C 10 my mcgroup a b c
914 $c=D3 Yes 10 mdgroup #2 h i j
917 It would also generate the following multiple response set record with
921 $d=E 1 2 34 13 third mdgroup k l m
922 $e=E 11 6 choice 0 n o p
925 @node Extra Product Info Record
926 @section Extra Product Info Record
928 This optional record appears to contain a text string that describes
929 the program that wrote the file and the source of the data. (This is
930 redundant with the file label and product info found in the file
940 /* @r{Exactly @code{count} bytes of data.} */
945 @item int32 rec_type;
946 Record type. Always set to 7.
949 Record subtype. Always set to 10.
952 The size of each element in the @code{info} member. Always set to 1.
955 The total number of bytes in @code{info}.
958 A text string. A product that identifies itself as @code{VOXCO
959 INTERVIEWER 4.3} uses CR-only line ends in this field, rather than the
960 more usual LF-only or CR LF line ends.
963 @node Variable Display Parameter Record
964 @section Variable Display Parameter Record
966 The variable display parameter record, if present, has the following
976 /* @r{Repeated @code{count} times}. */
978 int32 width; /* @r{Not always present.} */
983 @item int32 rec_type;
984 Record type. Always set to 7.
987 Record subtype. Always set to 11.
990 The size of @code{int32}. Always set to 4.
993 The number of sets of variable display parameters (ordinarily the
994 number of variables in the dictionary), times 2 or 3.
997 The remaining members are repeated @code{count} times, in the same
998 order as the variable records. No element corresponds to variable
999 records that continue long string variables. The meanings of these
1000 members are as follows:
1003 @item int32 measure;
1004 The measurement level of the variable:
1016 An ``unknown'' @code{measure} of 0 means that the variable was created
1017 in some way that doesn't make the measurement level clear, e.g.@: with
1018 a @code{COMPUTE} transformation. PSPP sets the measurement level the
1019 first time it reads the data using the rules documented in
1020 @ref{Measurement Level,,,pspp, PSPP Users Guide}, so this should
1024 The width of the display column for the variable in characters.
1026 This field is present if @var{count} is 3 times the number of
1027 variables in the dictionary. It is omitted if @var{count} is 2 times
1028 the number of variables.
1030 @item int32 alignment;
1031 The alignment of the variable for display purposes:
1043 @node Long Variable Names Record
1044 @section Long Variable Names Record
1046 If present, the long variable names record has the following format:
1055 /* @r{Exactly @code{count} bytes of data.} */
1056 char var_name_pairs[];
1060 @item int32 rec_type;
1061 Record type. Always set to 7.
1063 @item int32 subtype;
1064 Record subtype. Always set to 13.
1067 The size of each element in the @code{var_name_pairs} member. Always set to 1.
1070 The total number of bytes in @code{var_name_pairs}.
1072 @item char var_name_pairs[];
1073 A list of @var{key}--@var{value} tuples, where @var{key} is the name
1074 of a variable, and @var{value} is its long variable name.
1075 The @var{key} field is at most 8 bytes long and must match the
1076 name of a variable which appears in the variable record (@pxref{Variable
1078 The @var{value} field is at most 64 bytes long.
1079 The @var{key} and @var{value} fields are separated by a @samp{=} byte.
1080 Each tuple is separated by a byte whose value is 09. There is no
1081 trailing separator following the last tuple.
1082 The total length is @code{count} bytes.
1085 @node Very Long String Record
1086 @section Very Long String Record
1088 Old versions of SPSS limited string variables to a width of 255 bytes.
1089 For backward compatibility with these older versions, the system file
1090 format represents a string longer than 255 bytes, called a @dfn{very
1091 long string}, as a collection of strings no longer than 255 bytes
1092 each. The strings concatenated to make a very long string are called
1093 its @dfn{segments}; for consistency, variables other than very long
1094 strings are considered to have a single segment.
1096 A very long string with a width of @var{w} has @var{n} =
1097 (@var{w} + 251) / 252 segments, that is, one segment for every
1098 252 bytes of width, rounding up. It would be logical, then, for each
1099 of the segments except the last to have a width of 252 and the last
1100 segment to have the remainder, but this is not the case. In fact,
1101 each segment except the last has a width of 255 bytes. The last
1102 segment has width @var{w} - (@var{n} - 1) * 252; some versions
1103 of SPSS make it slightly wider, but not wide enough to make the last
1104 segment require another 8 bytes of data.
1106 Data is packed tightly into segments of a very long string, 255 bytes
1107 per segment. Because 255 bytes of segment data are allocated for
1108 every 252 bytes of the very long string's width (approximately), some
1109 unused space is left over at the end of the allocated segments. Data
1110 in unused space is ignored.
1112 Example: Consider a very long string of width 20,000. Such a very
1113 long string has 20,000 / 252 = 80 (rounding up) segments. The first
1114 79 segments have width 255; the last segment has width 20,000 - 79 *
1115 252 = 92 or slightly wider (up to 96 bytes, the next multiple of 8).
1116 The very long string's data is actually stored in the 19,890 bytes in
1117 the first 78 segments, plus the first 110 bytes of the 79th segment
1118 (19,890 + 110 = 20,000). The remaining 145 bytes of the 79th segment
1119 and all 92 bytes of the 80th segment are unused.
1121 The very long string record explains how to stitch together segments
1122 to obtain very long string data. For each of the very long string
1123 variables in the dictionary, it specifies the name of its first
1124 segment's variable and the very long string variable's actual width.
1125 The remaining segments immediately follow the named variable in the
1126 system file's dictionary.
1128 The very long string record, which is present only if the system file
1129 contains very long string variables, has the following format:
1138 /* @r{Exactly @code{count} bytes of data.} */
1139 char string_lengths[];
1143 @item int32 rec_type;
1144 Record type. Always set to 7.
1146 @item int32 subtype;
1147 Record subtype. Always set to 14.
1150 The size of each element in the @code{string_lengths} member. Always set to 1.
1153 The total number of bytes in @code{string_lengths}.
1155 @item char string_lengths[];
1156 A list of @var{key}--@var{value} tuples, where @var{key} is the name
1157 of a variable, and @var{value} is its length.
1158 The @var{key} field is at most 8 bytes long and must match the
1159 name of a variable which appears in the variable record (@pxref{Variable
1161 The @var{value} field is exactly 5 bytes long. It is a zero-padded,
1162 ASCII-encoded string that is the length of the variable.
1163 The @var{key} and @var{value} fields are separated by a @samp{=} byte.
1164 Tuples are delimited by a two-byte sequence @{00, 09@}.
1165 After the last tuple, there may be a single byte 00, or @{00, 09@}.
1166 The total length is @code{count} bytes.
1169 @node Character Encoding Record
1170 @section Character Encoding Record
1172 This record, if present, indicates the character encoding for string data,
1173 long variable names, variable labels, value labels and other strings in the
1183 /* @r{Exactly @code{count} bytes of data.} */
1188 @item int32 rec_type;
1189 Record type. Always set to 7.
1191 @item int32 subtype;
1192 Record subtype. Always set to 20.
1195 The size of each element in the @code{encoding} member. Always set to 1.
1198 The total number of bytes in @code{encoding}.
1200 @item char encoding[];
1201 The name of the character encoding. Normally this will be an official
1202 IANA character set name or alias.
1203 See @url{http://www.iana.org/assignments/character-sets}.
1204 Character set names are not case-sensitive, but SPSS appears to write
1205 them in all-uppercase.
1208 This record is not present in files generated by older software. See
1209 also the @code{character_code} field in the machine integer info
1210 record (@pxref{character-code}).
1212 When the character encoding record and the machine integer info record
1213 are both present, all system files observed in practice indicate the
1214 same character encoding, e.g.@: 1252 as @code{character_code} and
1215 @code{windows-1252} as @code{encoding}, 65001 and @code{UTF-8}, etc.
1217 If, for testing purposes, a file is crafted with different
1218 @code{character_code} and @code{encoding}, it seems that
1219 @code{character_code} controls the encoding for all strings in the
1220 system file before the dictionary termination record, including
1221 strings in data (e.g.@: string missing values), and @code{encoding}
1222 controls the encoding for strings following the dictionary termination
1225 @node Long String Value Labels Record
1226 @section Long String Value Labels Record
1228 This record, if present, specifies value labels for long string
1238 /* @r{Repeated up to exactly @code{count} bytes.} */
1243 long_string_label labels[];
1247 @item int32 rec_type;
1248 Record type. Always set to 7.
1250 @item int32 subtype;
1251 Record subtype. Always set to 21.
1257 The number of bytes following the header until the next header.
1259 @item int32 var_name_len;
1260 @itemx char var_name[];
1261 The number of bytes in the name of the variable that has long string
1262 value labels, plus the variable name itself, which consists of exactly
1263 @code{var_name_len} bytes. The variable name is not padded to any
1264 particular boundary, nor is it null-terminated.
1266 @item int32 var_width;
1267 The width of the variable, in bytes, which will be between 9 and
1270 @item int32 n_labels;
1271 @itemx long_string_label labels[];
1272 The long string labels themselves. The @code{labels} array contains
1273 exactly @code{n_labels} elements, each of which has the following
1284 @item int32 value_len;
1285 @itemx char value[];
1286 The string value being labeled. @code{value_len} is the number of
1287 bytes in @code{value}; it is equal to @code{var_width}. The
1288 @code{value} array is not padded or null-terminated.
1290 @item int32 label_len;
1291 @itemx char label[];
1292 The label for the string value. @code{label_len}, which must be
1293 between 0 and 120, is the number of bytes in @code{label}. The
1294 @code{label} array is not padded or null-terminated.
1298 @node Long String Missing Values Record
1299 @section Long String Missing Values Record
1301 This record, if present, specifies missing values for long string
1311 /* @r{Repeated up to exactly @code{count} bytes.} */
1314 char n_missing_values;
1315 long_string_missing_value values[];
1319 @item int32 rec_type;
1320 Record type. Always set to 7.
1322 @item int32 subtype;
1323 Record subtype. Always set to 22.
1329 The number of bytes following the header until the next header.
1331 @item int32 var_name_len;
1332 @itemx char var_name[];
1333 The number of bytes in the name of the long string variable that has
1334 missing values, plus the variable name itself, which consists of
1335 exactly @code{var_name_len} bytes. The variable name is not padded to
1336 any particular boundary, nor is it null-terminated.
1338 @item char n_missing_values;
1339 The number of missing values, either 1, 2, or 3. (This is, unusually,
1340 a single byte instead of a 32-bit number.)
1342 @item long_string_missing_value values[];
1343 The missing values themselves. This array contains exactly
1344 @code{n_missing_values} elements, each of which has the following
1353 @item int32 value_len;
1354 The length of the missing value string, in bytes. This value should
1355 be 8, because long string variables are at least 8 bytes wide (by
1356 definition), only the first 8 bytes of a long string variable's
1357 missing values are allowed to be non-spaces, and any spaces within the
1358 first 8 bytes are included in the missing value here.
1361 The missing value string, exactly @code{value_len} bytes, without
1362 any padding or null terminator.
1366 @node Data File and Variable Attributes Records
1367 @section Data File and Variable Attributes Records
1369 The data file and variable attributes records represent custom
1370 attributes for the system file or for individual variables in the
1371 system file, as defined on the DATAFILE ATTRIBUTE (@pxref{DATAFILE
1372 ATTRIBUTE,,,pspp, PSPP Users Guide}) and VARIABLE ATTRIBUTE commands
1373 (@pxref{VARIABLE ATTRIBUTE,,,pspp, PSPP Users Guide}), respectively.
1382 /* @r{Exactly @code{count} bytes of data.} */
1387 @item int32 rec_type;
1388 Record type. Always set to 7.
1390 @item int32 subtype;
1391 Record subtype. Always set to 17 for a data file attribute record or
1392 to 18 for a variable attributes record.
1395 The size of each element in the @code{attributes} member. Always set to 1.
1398 The total number of bytes in @code{attributes}.
1400 @item char attributes[];
1401 The attributes, in a text-based format.
1403 In record subtype 17, this field contains a single attribute set. An
1404 attribute set is a sequence of one or more attributes concatenated
1405 together. Each attribute consists of a name, which has the same
1406 syntax as a variable name, followed by, inside parentheses, a sequence
1407 of one or more values. Each value consists of a string enclosed in
1408 single quotes (@code{'}) followed by a line feed (byte 0x0a). A value
1409 may contain single quote characters, which are not themselves escaped
1410 or quoted or required to be present in pairs. There is no apparent
1411 way to embed a line feed in a value. There is no distinction between
1412 an attribute with a single value and an attribute array with one
1415 In record subtype 18, this field contains a sequence of one or more
1416 variable attribute sets. If more than one variable attribute set is
1417 present, each one after the first is delimited from the previous by
1418 @code{/}. Each variable attribute set consists of a long
1420 followed by @code{:}, followed by an attribute set with the same
1421 syntax as on record subtype 17.
1423 System files written by @code{Stata 14.1/-savespss- 1.77 by
1424 S.Radyakin} may include multiple records with subtype 18, one per
1425 variable that has variable attributes.
1427 The total length is @code{count} bytes.
1432 A system file produced with the following VARIABLE ATTRIBUTE commands
1436 VARIABLE ATTRIBUTE VARIABLES=dummy ATTRIBUTE=fred[1]('23') fred[2]('34').
1437 VARIABLE ATTRIBUTE VARIABLES=dummy ATTRIBUTE=bert('123').
1441 will contain a variable attribute record with the following contents:
1444 0000 07 00 00 00 12 00 00 00 01 00 00 00 22 00 00 00 |............"...|
1445 0010 64 75 6d 6d 79 3a 66 72 65 64 28 27 32 33 27 0a |dummy:fred('23'.|
1446 0020 27 33 34 27 0a 29 62 65 72 74 28 27 31 32 33 27 |'34'.)bert('123'|
1454 @node Variable Roles
1455 @subsection Variable Roles
1457 A variable's role is represented as an attribute named @code{$@@Role}.
1458 This attribute has a single element whose values and their meanings
1463 Input. This, the default, is the most common role.
1476 @node Extended Number of Cases Record
1477 @section Extended Number of Cases Record
1479 The file header record expresses the number of cases in the system
1480 file as an int32 (@pxref{File Header Record}). This record allows the
1481 number of cases in the system file to be expressed as a 64-bit number.
1493 @item int32 rec_type;
1494 Record type. Always set to 7.
1496 @item int32 subtype;
1497 Record subtype. Always set to 16.
1500 Size of each element. Always set to 8.
1503 Number of pieces of data in the data part. Alway set to 2.
1505 @item int64 unknown;
1506 Meaning unknown. Always set to 1.
1508 @item int64 ncases64;
1509 Number of cases in the file as a 64-bit integer. Presumably this
1510 could be -1 to indicate that the number of cases is unknown, for the
1511 same reason as @code{ncases} in the file header record, but this has
1512 not been observed in the wild.
1515 @node Other Informational Records
1516 @section Other Informational Records
1518 This chapter documents many specific types of extension records are
1519 documented here, but others are known to exist. PSPP ignores unknown
1520 extension records when reading system files.
1522 The following extension record subtypes have also been observed, with
1523 the following believed meanings:
1527 A named variable set for use in the GUI (according to Aapi
1531 Date info, probably related to USE (according to Aapi H@"am@"al@"ainen).
1534 A UUID in the format described in RFC 4122. Only two examples
1535 observed, both written by SPSS 13, and in each case the UUID contained
1536 both upper and lower case.
1539 XML that describes how data in the file should be displayed on-screen.
1542 @node Dictionary Termination Record
1543 @section Dictionary Termination Record
1545 The dictionary termination record separates all other records from the
1554 @item int32 rec_type;
1555 Record type. Always set to 999.
1558 Ignored padding. Should be set to 0.
1562 @section Data Record
1564 The data record must follow all other records in the system file.
1565 Every system file must have a data record that specifies data for at
1566 least one case. The format of the data record varies depending on the
1567 value of @code{compression} in the file header record:
1570 @item 0: no compression
1571 Data is arranged as a series of 8-byte elements.
1572 Each element corresponds to
1573 the variable declared in the respective variable record (@pxref{Variable
1574 Record}). Numeric values are given in @code{flt64} format; string
1575 values are literal characters string, padded on the right when
1576 necessary to fill out 8-byte units.
1578 @item 1: bytecode compression
1580 of the data record is divided into a series of 1-byte command
1581 codes. These codes have meanings as described below:
1585 Ignored. If the program writing the system file accumulates compressed
1586 data in blocks of fixed length, 0 bytes can be used to pad out extra
1587 bytes remaining at the end of a fixed-size block.
1591 value @var{code} - @var{bias}, where
1592 @var{code} is the value of the compression code and @var{bias} is the
1593 variable @code{bias} from the file header. For example,
1594 code 105 with bias 100.0 (the normal value) indicates a numeric variable
1597 A code of 0 (after subtracting the bias) in a string field encodes
1598 null bytes. This is unusual, since a string field normally encodes
1599 text data, but it exists in real system files.
1602 End of file. This code may or may not appear at the end of the data
1603 stream. PSPP always outputs this code but its use is not required.
1606 A numeric or string value that is not
1607 compressible. The value is stored in the 8 bytes following the
1608 current block of command bytes. If this value appears twice in a block
1609 of command bytes, then it indicates the second group of 8 bytes following the
1610 command bytes, and so on.
1613 An 8-byte string value that is all spaces.
1616 The system-missing value.
1619 The end of the 8-byte group of bytecodes is followed by any 8-byte
1620 blocks of non-compressible values indicated by code 253. After that
1621 follows another 8-byte group of bytecodes, then those bytecodes'
1622 non-compressible values. The pattern repeats to the end of the file
1623 or a code with value 252.
1625 @item 2: ZLIB compression
1626 The data record consists of the following, in order:
1630 ZLIB data header, 24 bytes long.
1633 One or more variable-length blocks of ZLIB compressed data.
1636 ZLIB data trailer, with a 24-byte fixed header plus an additional 24
1637 bytes for each preceding ZLIB compressed data block.
1640 The ZLIB data header has the following format:
1649 @item int64 zheader_ofs;
1650 The offset, in bytes, of the beginning of this structure within the
1653 @item int64 ztrailer_ofs;
1654 The offset, in bytes, of the first byte of the ZLIB data trailer.
1656 @item int64 ztrailer_len;
1657 The number of bytes in the ZLIB data trailer. This and the previous
1658 field sum to the size of the system file in bytes.
1661 The data header is followed by @code{(ztrailer_len - 24) / 24} ZLIB
1662 compressed data blocks. Each ZLIB compressed data block begins with a
1663 ZLIB header as specified in RFC@tie{}1950, e.g.@: hex bytes @code{78
1664 01} (the only header yet observed in practice). Each block
1665 decompresses to a fixed number of bytes (in practice only
1666 @code{0x3ff000}-byte blocks have been observed), except that the last
1667 block of data may be shorter. The last ZLIB compressed data block
1668 gends just before offset @code{ztrailer_ofs}.
1670 The result of ZLIB decompression is bytecode compressed data as
1671 described above for compression format 1.
1673 The ZLIB data trailer begins with the following 24-byte fixed header:
1683 @item int64 int_bias;
1684 The compression bias as a negative integer, e.g.@: if @code{bias} in
1685 the file header record is 100.0, then @code{int_bias} is @minus{}100
1686 (this is the only value yet observed in practice).
1689 Always observed to be zero.
1691 @item int32 block_size;
1692 The number of bytes in each ZLIB compressed data block, except
1693 possibly the last, following decompression. Only @code{0x3ff000} has
1694 been observed so far.
1696 @item int32 n_blocks;
1697 The number of ZLIB compressed data blocks, always exactly
1698 @code{(ztrailer_len - 24) / 24}.
1701 The fixed header is followed by @code{n_blocks} 24-byte ZLIB data
1702 block descriptors, each of which describes the compressed data block
1703 corresponding to its offset. Each block descriptor has the following
1707 int64 uncompressed_ofs;
1708 int64 compressed_ofs;
1709 int32 uncompressed_size;
1710 int32 compressed_size;
1714 @item int64 uncompressed_ofs;
1715 The offset, in bytes, that this block of data would have in a similar
1716 system file that uses compression format 1. This is
1717 @code{zheader_ofs} in the first block descriptor, and in each
1718 succeeding block descriptor it is the sum of the previous desciptor's
1719 @code{uncompressed_ofs} and @code{uncompressed_size}.
1721 @item int64 compressed_ofs;
1722 The offset, in bytes, of the actual beginning of this compressed data
1723 block. This is @code{zheader_ofs + 24} in the first block descriptor,
1724 and in each succeeding block descriptor it is the sum of the previous
1725 descriptor's @code{compressed_ofs} and @code{compressed_size}. The
1726 final block descriptor's @code{compressed_ofs} and
1727 @code{compressed_size} sum to @code{ztrailer_ofs}.
1729 @item int32 uncompressed_size;
1730 The number of bytes in this data block, after decompression. This is
1731 @code{block_size} in every data block except the last, which may be
1734 @item int32 compressed_size;
1735 The number of bytes in this data block, as stored compressed in this
1740 @setfilename ignored