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 @chapter System File Format
14 An SPSS system file holds a set of cases and dictionary information
15 that describes how they may be interpreted. The system file format
16 dates back 40+ years and has evolved greatly over that time to support
17 new features, but in a way to facilitate interchange between even the
18 oldest and newest versions of software. This chapter describes the
21 System files use four data types: 8-bit characters, 32-bit integers,
23 and 64-bit floating points, called here @code{char}, @code{int32},
25 @code{flt64}, respectively. Data is not necessarily aligned on a word
26 or double-word boundary: the long variable name record (@pxref{Long
27 Variable Names Record}) and very long string records (@pxref{Very Long
28 String Record}) have arbitrary byte length and can therefore cause all
29 data coming after them in the file to be misaligned.
31 Integer data in system files may be big-endian or little-endian. A
32 reader may detect the endianness of a system file by examining
33 @code{layout_code} in the file header record
34 (@pxref{layout_code,,@code{layout_code}}).
36 Floating-point data in system files may nominally be in IEEE 754, IBM,
37 or VAX formats. A reader may detect the floating-point format in use
38 by examining @code{bias} in the file header record
39 (@pxref{bias,,@code{bias}}).
41 PSPP detects big-endian and little-endian integer formats in system
42 files and translates as necessary. PSPP also detects the
43 floating-point format in use, as well as the endianness of IEEE 754
44 floating-point numbers, and translates as needed. However, only IEEE
45 754 numbers with the same endianness as integer data in the same file
46 have actually been observed in system files, and it is likely that
47 other formats are obsolete or were never used.
49 System files use a few floating point values for special purposes:
53 The system-missing value is represented by the largest possible
54 negative number in the floating point format (@code{-DBL_MAX}).
57 HIGHEST is used as the high end of a missing value range with an
58 unbounded maximum. It is represented by the largest possible positive
59 number (@code{DBL_MAX}).
62 LOWEST is used as the low end of a missing value range with an
63 unbounded minimum. It was originally represented by the
64 second-largest negative number (in IEEE 754 format,
65 @code{0xffeffffffffffffe}). System files written by SPSS 21 and later
66 instead use the largest negative number (@code{-DBL_MAX}), the same
67 value as SYSMIS. This does not lead to ambiguity because LOWEST
68 appears in system files only in missing value ranges, which never
72 System files may use most character encodings based on an 8-bit unit.
73 UTF-16 and UTF-32, based on wider units, appear to be unacceptable.
74 @code{rec_type} in the file header record is sufficient to distinguish
75 between ASCII and EBCDIC based encodings. The best way to determine
76 the specific encoding in use is to consult the character encoding
77 record (@pxref{Character Encoding Record}), if present, and failing
78 that the @code{character_code} in the machine integer info record
79 (@pxref{Machine Integer Info Record}). The same encoding should be
80 used for the dictionary and the data in the file, although it is
81 possible to artificially synthesize files that use different encodings
82 (@pxref{Character Encoding Record}).
85 * System File Record Structure::
86 * File Header Record::
88 * Value Labels Records::
90 * Machine Integer Info Record::
91 * Machine Floating-Point Info Record::
92 * Multiple Response Sets Records::
93 * Extra Product Info Record::
94 * Variable Display Parameter Record::
95 * Long Variable Names Record::
96 * Very Long String Record::
97 * Character Encoding Record::
98 * Long String Value Labels Record::
99 * Long String Missing Values Record::
100 * Data File and Variable Attributes Records::
101 * Extended Number of Cases Record::
102 * Other Informational Records::
103 * Dictionary Termination Record::
107 @node System File Record Structure
108 @section System File Record Structure
110 System files are divided into records with the following format:
117 This header does not identify the length of the @code{data} or any
118 information about what it contains, so the system file reader must
119 understand the format of @code{data} based on @code{type}. However,
120 records with type 7, called @dfn{extension records}, have a stricter
128 char data[size * count];
132 @item int32 rec_type;
133 Record type. Always set to 7.
136 Record subtype. This value identifies a particular kind of extension
140 The size of each piece of data that follows the header, in bytes.
141 Known extension records use 1, 4, or 8, for @code{char}, @code{int32},
142 and @code{flt64} format data, respectively.
145 The number of pieces of data that follow the header.
147 @item char data[size * count];
148 Data, whose format and interpretation depend on the subtype.
151 An extension record contains exactly @code{size * count} bytes of
152 data, which allows a reader that does not understand an extension
153 record to skip it. Extension records provide only nonessential
154 information, so this allows for files written by newer software to
155 preserve backward compatibility with older or less capable readers.
157 Records in a system file must appear in the following order:
167 All pairs of value labels records and value label variables records,
171 Document record, if present.
174 Extension (type 7) records, in ascending numerical order of their
177 System files written by SPSS include at most one of each kind of
178 extension record. This is generally true of system files written by
179 other software as well, with known exceptions noted below in the
180 individual sections about each type of record.
183 Dictionary termination record.
189 We advise authors of programs that read system files to tolerate
190 format variations. Various kinds of misformatting and corruption have
191 been observed in system files written by SPSS and other software
192 alike. In particular, because extension records provide nonessential
193 information, it is generally better to ignore an extension record
194 entirely than to refuse to read a system file.
196 The following sections describe the known kinds of records.
198 @node File Header Record
199 @section File Header Record
201 A system file begins with the file header, with the following format:
207 int32 nominal_case_size;
212 char creation_date[9];
213 char creation_time[8];
219 @item char rec_type[4];
220 Record type code, either @samp{$FL2} for system files with
221 uncompressed data or data compressed with simple bytecode compression,
222 or @samp{$FL3} for system files with ZLIB compressed data.
224 This is truly a character field that uses the character encoding as
225 other strings. Thus, in a file with an ASCII-based character encoding
226 this field contains @code{24 46 4c 32} or @code{24 46 4c 33}, and in a
227 file with an EBCDIC-based encoding this field contains @code{5b c6 d3
228 f2}. (No EBCDIC-based ZLIB-compressed files have been observed.)
230 @item char prod_name[60];
231 Product identification string. This always begins with the characters
232 @samp{@@(#) SPSS DATA FILE}. PSPP uses the remaining characters to
233 give its version and the operating system name; for example, @samp{GNU
234 pspp 0.1.4 - sparc-sun-solaris2.5.2}. The string is truncated if it
235 would be longer than 60 characters; otherwise it is padded on the right
238 The product name field allow readers to behave differently based on
239 quirks in the way that particular software writes system files.
240 @xref{Value Labels Records}, for the detail of the quirk that the PSPP
241 system file reader tolerates in files written by ReadStat, which has
242 @code{https://github.com/WizardMac/ReadStat} in @code{prod_name}.
245 @item int32 layout_code;
246 Normally set to 2, although a few system files have been spotted in
247 the wild with a value of 3 here. PSPP use this value to determine the
248 file's integer endianness (@pxref{System File Format}).
250 @item int32 nominal_case_size;
251 Number of data elements per case. This is the number of variables,
252 except that long string variables add extra data elements (one for every
253 8 characters after the first 8). However, string variables do not
254 contribute to this value beyond the first 255 bytes. Further, some
255 software always writes -1 or 0 in this field. In general, it is
256 unsafe for systems reading system files to rely upon this value.
258 @item int32 compression;
259 Set to 0 if the data in the file is not compressed, 1 if the data is
260 compressed with simple bytecode compression, 2 if the data is ZLIB
261 compressed. This field has value 2 if and only if @code{rec_type} is
264 @item int32 weight_index;
265 If one of the variables in the data set is used as a weighting
266 variable, set to the dictionary index of that variable, plus 1
267 (@pxref{Dictionary Index}). Otherwise, set to 0.
270 Set to the number of cases in the file if it is known, or -1 otherwise.
272 In the general case it is not possible to determine the number of cases
273 that will be output to a system file at the time that the header is
274 written. The way that this is dealt with is by writing the entire
275 system file, including the header, then seeking back to the beginning of
276 the file and writing just the @code{ncases} field. For files in which
277 this is not valid, the seek operation fails. In this case,
278 @code{ncases} remains -1.
282 Compression bias, ordinarily set to 100. Only integers between
283 @code{1 - bias} and @code{251 - bias} can be compressed.
285 By assuming that its value is 100, PSPP uses @code{bias} to determine
286 the file's floating-point format and endianness (@pxref{System File
287 Format}). If the compression bias is not 100, PSPP cannot auto-detect
288 the floating-point format and assumes that it is IEEE 754 format with
289 the same endianness as the system file's integers, which is correct
290 for all known system files.
292 @item char creation_date[9];
293 Date of creation of the system file, in @samp{dd mmm yy}
294 format, with the month as standard English abbreviations, using an
295 initial capital letter and following with lowercase. If the date is not
296 available then this field is arbitrarily set to @samp{01 Jan 70}.
298 @item char creation_time[8];
299 Time of creation of the system file, in @samp{hh:mm:ss}
300 format and using 24-hour time. If the time is not available then this
301 field is arbitrarily set to @samp{00:00:00}.
303 @item char file_label[64];
304 File label declared by the user, if any (@pxref{FILE LABEL,,,pspp,
305 PSPP Users Guide}). Padded on the right with spaces.
307 A product that identifies itself as @code{VOXCO INTERVIEWER 4.3} uses
308 CR-only line ends in this field, rather than the more usual LF-only or
311 @item char padding[3];
312 Ignored padding bytes to make the structure a multiple of 32 bits in
313 length. Set to zeros.
316 @node Variable Record
317 @section Variable Record
319 There must be one variable record for each numeric variable and each
320 string variable with width 8 bytes or less. String variables wider
321 than 8 bytes have one variable record for each 8 bytes, rounding up.
322 The first variable record for a long string specifies the variable's
323 correct dictionary information. Subsequent variable records for a
324 long string are filled with dummy information: a type of -1, no
325 variable label or missing values, print and write formats that are
326 ignored, and an empty string as name. A few system files have been
327 encountered that include a variable label on dummy variable records,
328 so readers should take care to parse dummy variable records in the
329 same way as other variable records.
331 @anchor{Dictionary Index}
332 The @dfn{dictionary index} of a variable is a 1-based offset in the set of
333 variable records, including dummy variable records for long string
334 variables. The first variable record has a dictionary index of 1, the
335 second has a dictionary index of 2, and so on.
337 The system file format does not directly support string variables
338 wider than 255 bytes. Such very long string variables are represented
339 by a number of narrower string variables. @xref{Very Long String
340 Record}, for details.
342 A system file should contain at least one variable and thus at least
343 one variable record, but system files have been observed in the wild
344 without any variables (thus, no data either).
350 int32 n_missing_values;
355 /* @r{Present only if @code{has_var_label} is 1.} */
359 /* @r{Present only if @code{n_missing_values} is nonzero}. */
360 flt64 missing_values[];
364 @item int32 rec_type;
365 Record type code. Always set to 2.
368 Variable type code. Set to 0 for a numeric variable. For a short
369 string variable or the first part of a long string variable, this is set
370 to the width of the string. For the second and subsequent parts of a
371 long string variable, set to -1, and the remaining fields in the
372 structure are ignored.
374 @item int32 has_var_label;
375 If this variable has a variable label, set to 1; otherwise, set to 0.
377 @item int32 n_missing_values;
378 If the variable has no missing values, set to 0. If the variable has
379 one, two, or three discrete missing values, set to 1, 2, or 3,
380 respectively. If the variable has a range for missing variables, set to
381 -2; if the variable has a range for missing variables plus a single
382 discrete value, set to -3.
384 A long string variable always has the value 0 here. A separate record
385 indicates missing values for long string variables (@pxref{Long String
386 Missing Values Record}).
389 Print format for this variable. See below.
392 Write format for this variable. See below.
395 Variable name. The variable name must begin with a capital letter or
396 the at-sign (@samp{@@}). Subsequent characters may also be digits, octothorpes
397 (@samp{#}), dollar signs (@samp{$}), underscores (@samp{_}), or full
398 stops (@samp{.}). The variable name is padded on the right with spaces.
400 The @samp{name} fields should be unique within a system file. System
401 files written by SPSS that contain very long string variables with
402 similar names sometimes contain duplicate names that are later
403 eliminated by resolving the very long string names (@pxref{Very Long
404 String Record}). PSPP handles duplicates by assigning them new,
407 @item int32 label_len;
408 This field is present only if @code{has_var_label} is set to 1. It is
409 set to the length, in characters, of the variable label. The
410 documented maximum length varies from 120 to 255 based on SPSS
411 version, but some files have been seen with longer labels. PSPP
412 accepts labels of any length.
415 This field is present only if @code{has_var_label} is set to 1. It has
416 length @code{label_len}, rounded up to the nearest multiple of 32 bits.
417 The first @code{label_len} characters are the variable's variable label.
419 @item flt64 missing_values[];
420 This field is present only if @code{n_missing_values} is nonzero. It
421 has the same number of 8-byte elements as the absolute value of
422 @code{n_missing_values}. Each element is interpreted as a number for
423 numeric variables (with HIGHEST and LOWEST indicated as described in
424 the chapter introduction). For string variables of width less than 8
425 bytes, elements are right-padded with spaces; for string variables
426 wider than 8 bytes, only the first 8 bytes of each missing value are
427 specified, with the remainder implicitly all spaces.
429 For discrete missing values, each element represents one missing
430 value. When a range is present, the first element denotes the minimum
431 value in the range, and the second element denotes the maximum value
432 in the range. When a range plus a value are present, the third
433 element denotes the additional discrete missing value.
436 @anchor{System File Output Formats}
437 The @code{print} and @code{write} members of sysfile_variable are output
438 formats coded into @code{int32} types. The least-significant byte
439 of the @code{int32} represents the number of decimal places, and the
440 next two bytes in order of increasing significance represent field width
441 and format type, respectively. The most-significant byte is not
442 used and should be set to zero.
444 Format types are defined as follows:
447 @multitable {Value} {@code{DATETIME}}
537 A few system files have been observed in the wild with invalid
538 @code{write} fields, in particular with value 0. Readers should
539 probably treat invalid @code{print} or @code{write} fields as some
542 @node Value Labels Records
543 @section Value Labels Records
545 The value label records documented in this section are used for
546 numeric and short string variables only. Long string variables may
547 have value labels, but their value labels are recorded using a
548 different record type (@pxref{Long String Value Labels Record}).
550 ReadStat (@pxref{File Header Record}) writes value labels that label a
551 single value more than once. In more detail, it emits value labels
552 whose values are longer than string variables' widths, that are
553 identical in the actual width of the variable, e.g.@: labels for
554 values @code{ABC123} and @code{ABC456} for a string variable with
555 width 3. For files written by this software, PSPP ignores such
558 The value label record has the following format:
564 /* @r{Repeated @code{n_label} times}. */
571 @item int32 rec_type;
572 Record type. Always set to 3.
574 @item int32 label_count;
575 Number of value labels present in this record.
578 The remaining fields are repeated @code{count} times. Each
579 repetition specifies one value label.
583 A numeric value or a short string value padded as necessary to 8 bytes
584 in length. Its type and width cannot be determined until the
585 following value label variables record (see below) is read.
587 @item char label_len;
588 The label's length, in bytes. The documented maximum length varies
589 from 60 to 120 based on SPSS version. PSPP supports value labels up
593 @code{label_len} bytes of the actual label, followed by up to 7 bytes
594 of padding to bring @code{label} and @code{label_len} together to a
595 multiple of 8 bytes in length.
598 The value label record is always immediately followed by a value label
599 variables record with the following format:
608 @item int32 rec_type;
609 Record type. Always set to 4.
611 @item int32 var_count;
612 Number of variables that the associated value labels from the value
613 label record are to be applied.
616 A list of 1-based dictionary indexes of variables to which to apply the value
617 labels (@pxref{Dictionary Index}). There are @code{var_count}
620 String variables wider than 8 bytes may not be specified in this list.
623 @node Document Record
624 @section Document Record
626 The document record, if present, has the following format:
635 @item int32 rec_type;
636 Record type. Always set to 6.
639 Number of lines of documents present. This should be greater than
640 zero, but ReadStats writes system files with zero @code{n_lines}.
642 @item char lines[][80];
643 Document lines. The number of elements is defined by @code{n_lines}.
644 Lines shorter than 80 characters are padded on the right with spaces.
647 @node Machine Integer Info Record
648 @section Machine Integer Info Record
650 The integer info record, if present, has the following format:
662 int32 version_revision;
664 int32 floating_point_rep;
665 int32 compression_code;
667 int32 character_code;
671 @item int32 rec_type;
672 Record type. Always set to 7.
675 Record subtype. Always set to 3.
678 Size of each piece of data in the data part, in bytes. Always set to 4.
681 Number of pieces of data in the data part. Always set to 8.
683 @item int32 version_major;
684 PSPP major version number. In version @var{x}.@var{y}.@var{z}, this
687 @item int32 version_minor;
688 PSPP minor version number. In version @var{x}.@var{y}.@var{z}, this
691 @item int32 version_revision;
692 PSPP version revision number. In version @var{x}.@var{y}.@var{z},
695 @item int32 machine_code;
696 Machine code. PSPP always set this field to value to -1, but other
699 @item int32 floating_point_rep;
700 Floating point representation code. For IEEE 754 systems this is 1.
701 IBM 370 sets this to 2, and DEC VAX E to 3.
703 @item int32 compression_code;
704 Compression code. Always set to 1, regardless of whether or how the
707 @item int32 endianness;
708 Machine endianness. 1 indicates big-endian, 2 indicates little-endian.
710 @item int32 character_code;
711 @anchor{character-code} Character code. The following values have
712 been actually observed in system files:
722 The @code{windows-1250} code page for Central European and Eastern
726 The @code{windows-1252} code page for Western European languages.
735 The following additional values are known to be defined:
745 Other Windows code page numbers are known to be generally valid.
747 Old versions of SPSS for Unix and Windows always wrote value 2 in this
748 field, regardless of the encoding in use. Newer versions also write
749 the character encoding as a string (see @ref{Character Encoding
753 @node Machine Floating-Point Info Record
754 @section Machine Floating-Point Info Record
756 The floating-point info record, if present, has the following format:
772 @item int32 rec_type;
773 Record type. Always set to 7.
776 Record subtype. Always set to 4.
779 Size of each piece of data in the data part, in bytes. Always set to 8.
782 Number of pieces of data in the data part. Always set to 3.
785 @itemx flt64 highest;
787 The system missing value, the value used for HIGHEST in missing
788 values, and the value used for LOWEST in missing values, respectively.
789 @xref{System File Format}, for more information.
791 The SPSSWriter library in PHP, which identifies itself as @code{FOM
792 SPSS 1.0.0} in the file header record @code{prod_name} field, writes
793 unexpected values to these fields, but it uses the same values
794 consistently throughout the rest of the file.
797 @node Multiple Response Sets Records
798 @section Multiple Response Sets Records
800 The system file format has two different types of records that
801 represent multiple response sets (@pxref{MRSETS,,,pspp, PSPP Users
802 Guide}). The first type of record describes multiple response sets
803 that can be understood by SPSS before version 14. The second type of
804 record, with a closely related format, is used for multiple dichotomy
805 sets that use the CATEGORYLABELS=COUNTEDVALUES feature added in
815 /* @r{Exactly @code{count} bytes of data.} */
820 @item int32 rec_type;
821 Record type. Always set to 7.
824 Record subtype. Set to 7 for records that describe multiple response
825 sets understood by SPSS before version 14, or to 19 for records that
826 describe dichotomy sets that use the CATEGORYLABELS=COUNTEDVALUES
827 feature added in version 14.
830 The size of each element in the @code{mrsets} member. Always set to 1.
833 The total number of bytes in @code{mrsets}.
836 Zero or more line feeds (byte 0x0a), followed by a series of multiple
837 response sets, each of which consists of the following:
841 The set's name (an identifier that begins with @samp{$}), in mixed
842 upper and lower case.
845 An equals sign (@samp{=}).
848 @samp{C} for a multiple category set, @samp{D} for a multiple
849 dichotomy set with CATEGORYLABELS=VARLABELS, or @samp{E} for a
850 multiple dichotomy set with CATEGORYLABELS=COUNTEDVALUES.
853 For a multiple dichotomy set with CATEGORYLABELS=COUNTEDVALUES, a
854 space, followed by a number expressed as decimal digits, followed by a
855 space. If LABELSOURCE=VARLABEL was specified on MRSETS, then the
856 number is 11; otherwise it is 1.@footnote{This part of the format may
857 not be fully understood, because only a single example of each
858 possibility has been examined.}
861 For either kind of multiple dichotomy set, the counted value, as a
862 positive integer count specified as decimal digits, followed by a
863 space, followed by as many string bytes as specified in the count. If
864 the set contains numeric variables, the string consists of the counted
865 integer value expressed as decimal digits. If the set contains string
866 variables, the string contains the counted string value. Either way,
867 the string may be padded on the right with spaces (older versions of
868 SPSS seem to always pad to a width of 8 bytes; newer versions don't).
874 The multiple response set's label, using the same format as for the
875 counted value for multiple dichotomy sets. A string of length 0 means
876 that the set does not have a label. A string of length 0 is also
877 written if LABELSOURCE=VARLABEL was specified.
883 The short names of the variables in the set, converted to lowercase,
884 each separated from the previous by a single space.
886 Even though a multiple response set must have at least two variables,
887 some system files contain multiple response sets with no variables or
888 one variable. The source and meaning of these multiple response sets is
889 unknown. (Perhaps they arise from creating a multiple response set
890 then deleting all the variables that it contains?)
893 One line feed (byte 0x0a). Sometimes multiple, even hundreds, of line
898 Example: Given appropriate variable definitions, consider the
899 following MRSETS command:
902 MRSETS /MCGROUP NAME=$a LABEL='my mcgroup' VARIABLES=a b c
903 /MDGROUP NAME=$b VARIABLES=g e f d VALUE=55
904 /MDGROUP NAME=$c LABEL='mdgroup #2' VARIABLES=h i j VALUE='Yes'
905 /MDGROUP NAME=$d LABEL='third mdgroup' CATEGORYLABELS=COUNTEDVALUES
906 VARIABLES=k l m VALUE=34
907 /MDGROUP NAME=$e CATEGORYLABELS=COUNTEDVALUES LABELSOURCE=VARLABEL
908 VARIABLES=n o p VALUE='choice'.
911 The above would generate the following multiple response set record of
915 $a=C 10 my mcgroup a b c
917 $c=D3 Yes 10 mdgroup #2 h i j
920 It would also generate the following multiple response set record with
924 $d=E 1 2 34 13 third mdgroup k l m
925 $e=E 11 6 choice 0 n o p
928 @node Extra Product Info Record
929 @section Extra Product Info Record
931 This optional record appears to contain a text string that describes
932 the program that wrote the file and the source of the data. (This is
933 redundant with the file label and product info found in the file
943 /* @r{Exactly @code{count} bytes of data.} */
948 @item int32 rec_type;
949 Record type. Always set to 7.
952 Record subtype. Always set to 10.
955 The size of each element in the @code{info} member. Always set to 1.
958 The total number of bytes in @code{info}.
961 A text string. A product that identifies itself as @code{VOXCO
962 INTERVIEWER 4.3} uses CR-only line ends in this field, rather than the
963 more usual LF-only or CR LF line ends.
966 @node Variable Display Parameter Record
967 @section Variable Display Parameter Record
969 The variable display parameter record, if present, has the following
979 /* @r{Repeated @code{count} times}. */
981 int32 width; /* @r{Not always present.} */
986 @item int32 rec_type;
987 Record type. Always set to 7.
990 Record subtype. Always set to 11.
993 The size of @code{int32}. Always set to 4.
996 The number of sets of variable display parameters (ordinarily the
997 number of variables in the dictionary), times 2 or 3.
1000 The remaining members are repeated @code{count} times, in the same
1001 order as the variable records. No element corresponds to variable
1002 records that continue long string variables. The meanings of these
1003 members are as follows:
1006 @item int32 measure;
1007 The measurement level of the variable:
1019 An ``unknown'' @code{measure} of 0 means that the variable was created
1020 in some way that doesn't make the measurement level clear, e.g.@: with
1021 a @code{COMPUTE} transformation. PSPP sets the measurement level the
1022 first time it reads the data using the rules documented in
1023 @ref{Measurement Level,,,pspp, PSPP Users Guide}, so this should
1027 The width of the display column for the variable in characters.
1029 This field is present if @var{count} is 3 times the number of
1030 variables in the dictionary. It is omitted if @var{count} is 2 times
1031 the number of variables.
1033 @item int32 alignment;
1034 The alignment of the variable for display purposes:
1046 @node Long Variable Names Record
1047 @section Long Variable Names Record
1049 If present, the long variable names record has the following format:
1058 /* @r{Exactly @code{count} bytes of data.} */
1059 char var_name_pairs[];
1063 @item int32 rec_type;
1064 Record type. Always set to 7.
1066 @item int32 subtype;
1067 Record subtype. Always set to 13.
1070 The size of each element in the @code{var_name_pairs} member. Always set to 1.
1073 The total number of bytes in @code{var_name_pairs}.
1075 @item char var_name_pairs[];
1076 A list of @var{key}--@var{value} tuples, where @var{key} is the name
1077 of a variable, and @var{value} is its long variable name.
1078 The @var{key} field is at most 8 bytes long and must match the
1079 name of a variable which appears in the variable record (@pxref{Variable
1081 The @var{value} field is at most 64 bytes long.
1082 The @var{key} and @var{value} fields are separated by a @samp{=} byte.
1083 Each tuple is separated by a byte whose value is 09. There is no
1084 trailing separator following the last tuple.
1085 The total length is @code{count} bytes.
1088 @node Very Long String Record
1089 @section Very Long String Record
1091 Old versions of SPSS limited string variables to a width of 255 bytes.
1092 For backward compatibility with these older versions, the system file
1093 format represents a string longer than 255 bytes, called a @dfn{very
1094 long string}, as a collection of strings no longer than 255 bytes
1095 each. The strings concatenated to make a very long string are called
1096 its @dfn{segments}; for consistency, variables other than very long
1097 strings are considered to have a single segment.
1099 A very long string with a width of @var{w} has @var{n} =
1100 (@var{w} + 251) / 252 segments, that is, one segment for every
1101 252 bytes of width, rounding up. It would be logical, then, for each
1102 of the segments except the last to have a width of 252 and the last
1103 segment to have the remainder, but this is not the case. In fact,
1104 each segment except the last has a width of 255 bytes. The last
1105 segment has width @var{w} - (@var{n} - 1) * 252; some versions
1106 of SPSS make it slightly wider, but not wide enough to make the last
1107 segment require another 8 bytes of data.
1109 Data is packed tightly into segments of a very long string, 255 bytes
1110 per segment. Because 255 bytes of segment data are allocated for
1111 every 252 bytes of the very long string's width (approximately), some
1112 unused space is left over at the end of the allocated segments. Data
1113 in unused space is ignored.
1115 Example: Consider a very long string of width 20,000. Such a very
1116 long string has 20,000 / 252 = 80 (rounding up) segments. The first
1117 79 segments have width 255; the last segment has width 20,000 - 79 *
1118 252 = 92 or slightly wider (up to 96 bytes, the next multiple of 8).
1119 The very long string's data is actually stored in the 19,890 bytes in
1120 the first 78 segments, plus the first 110 bytes of the 79th segment
1121 (19,890 + 110 = 20,000). The remaining 145 bytes of the 79th segment
1122 and all 92 bytes of the 80th segment are unused.
1124 The very long string record explains how to stitch together segments
1125 to obtain very long string data. For each of the very long string
1126 variables in the dictionary, it specifies the name of its first
1127 segment's variable and the very long string variable's actual width.
1128 The remaining segments immediately follow the named variable in the
1129 system file's dictionary.
1131 The very long string record, which is present only if the system file
1132 contains very long string variables, has the following format:
1141 /* @r{Exactly @code{count} bytes of data.} */
1142 char string_lengths[];
1146 @item int32 rec_type;
1147 Record type. Always set to 7.
1149 @item int32 subtype;
1150 Record subtype. Always set to 14.
1153 The size of each element in the @code{string_lengths} member. Always set to 1.
1156 The total number of bytes in @code{string_lengths}.
1158 @item char string_lengths[];
1159 A list of @var{key}--@var{value} tuples, where @var{key} is the name
1160 of a variable, and @var{value} is its length.
1161 The @var{key} field is at most 8 bytes long and must match the
1162 name of a variable which appears in the variable record (@pxref{Variable
1164 The @var{value} field is exactly 5 bytes long. It is a zero-padded,
1165 ASCII-encoded string that is the length of the variable.
1166 The @var{key} and @var{value} fields are separated by a @samp{=} byte.
1167 Tuples are delimited by a two-byte sequence @{00, 09@}.
1168 After the last tuple, there may be a single byte 00, or @{00, 09@}.
1169 The total length is @code{count} bytes.
1172 @node Character Encoding Record
1173 @section Character Encoding Record
1175 This record, if present, indicates the character encoding for string data,
1176 long variable names, variable labels, value labels and other strings in the
1186 /* @r{Exactly @code{count} bytes of data.} */
1191 @item int32 rec_type;
1192 Record type. Always set to 7.
1194 @item int32 subtype;
1195 Record subtype. Always set to 20.
1198 The size of each element in the @code{encoding} member. Always set to 1.
1201 The total number of bytes in @code{encoding}.
1203 @item char encoding[];
1204 The name of the character encoding. Normally this will be an official
1205 IANA character set name or alias.
1206 See @url{http://www.iana.org/assignments/character-sets}.
1207 Character set names are not case-sensitive, but SPSS appears to write
1208 them in all-uppercase.
1211 This record is not present in files generated by older software. See
1212 also the @code{character_code} field in the machine integer info
1213 record (@pxref{character-code}).
1215 When the character encoding record and the machine integer info record
1216 are both present, all system files observed in practice indicate the
1217 same character encoding, e.g.@: 1252 as @code{character_code} and
1218 @code{windows-1252} as @code{encoding}, 65001 and @code{UTF-8}, etc.
1220 If, for testing purposes, a file is crafted with different
1221 @code{character_code} and @code{encoding}, it seems that
1222 @code{character_code} controls the encoding for all strings in the
1223 system file before the dictionary termination record, including
1224 strings in data (e.g.@: string missing values), and @code{encoding}
1225 controls the encoding for strings following the dictionary termination
1228 @node Long String Value Labels Record
1229 @section Long String Value Labels Record
1231 This record, if present, specifies value labels for long string
1241 /* @r{Repeated up to exactly @code{count} bytes.} */
1246 long_string_label labels[];
1250 @item int32 rec_type;
1251 Record type. Always set to 7.
1253 @item int32 subtype;
1254 Record subtype. Always set to 21.
1260 The number of bytes following the header until the next header.
1262 @item int32 var_name_len;
1263 @itemx char var_name[];
1264 The number of bytes in the name of the variable that has long string
1265 value labels, plus the variable name itself, which consists of exactly
1266 @code{var_name_len} bytes. The variable name is not padded to any
1267 particular boundary, nor is it null-terminated.
1269 @item int32 var_width;
1270 The width of the variable, in bytes, which will be between 9 and
1273 @item int32 n_labels;
1274 @itemx long_string_label labels[];
1275 The long string labels themselves. The @code{labels} array contains
1276 exactly @code{n_labels} elements, each of which has the following
1287 @item int32 value_len;
1288 @itemx char value[];
1289 The string value being labeled. @code{value_len} is the number of
1290 bytes in @code{value}; it is equal to @code{var_width}. The
1291 @code{value} array is not padded or null-terminated.
1293 @item int32 label_len;
1294 @itemx char label[];
1295 The label for the string value. @code{label_len}, which must be
1296 between 0 and 120, is the number of bytes in @code{label}. The
1297 @code{label} array is not padded or null-terminated.
1301 @node Long String Missing Values Record
1302 @section Long String Missing Values Record
1304 This record, if present, specifies missing values for long string
1314 /* @r{Repeated up to exactly @code{count} bytes.} */
1317 char n_missing_values;
1319 char values[values_len * n_missing_values];
1323 @item int32 rec_type;
1324 Record type. Always set to 7.
1326 @item int32 subtype;
1327 Record subtype. Always set to 22.
1333 The number of bytes following the header until the next header.
1335 @item int32 var_name_len;
1336 @itemx char var_name[];
1337 The number of bytes in the name of the long string variable that has
1338 missing values, plus the variable name itself, which consists of
1339 exactly @code{var_name_len} bytes. The variable name is not padded to
1340 any particular boundary, nor is it null-terminated.
1342 @item char n_missing_values;
1343 The number of missing values, either 1, 2, or 3. (This is, unusually,
1344 a single byte instead of a 32-bit number.)
1346 @item int32 value_len;
1347 The length of each missing value string, in bytes. This value should
1348 be 8, because long string variables are at least 8 bytes wide (by
1349 definition), only the first 8 bytes of a long string variable's
1350 missing values are allowed to be non-spaces, and any spaces within the
1351 first 8 bytes are included in the missing value here.
1353 @item char values[values_len * n_missing_values]
1354 The missing values themselves, without any padding or null
1358 An earlier version of this document stated that @code{value_len} was
1359 repeated before each of the missing values, so that there was an extra
1360 @code{int32} value of 8 before each missing value after the first.
1361 Old versions of PSPP wrote data files in this format. Readers can
1362 tolerate this mistake, if they wish, by noticing and skipping the
1363 extra @code{int32} values, which wouldn't ordinarily occur in strings.
1365 @node Data File and Variable Attributes Records
1366 @section Data File and Variable Attributes Records
1368 The data file and variable attributes records represent custom
1369 attributes for the system file or for individual variables in the
1370 system file, as defined on the DATAFILE ATTRIBUTE (@pxref{DATAFILE
1371 ATTRIBUTE,,,pspp, PSPP Users Guide}) and VARIABLE ATTRIBUTE commands
1372 (@pxref{VARIABLE ATTRIBUTE,,,pspp, PSPP Users Guide}), respectively.
1381 /* @r{Exactly @code{count} bytes of data.} */
1386 @item int32 rec_type;
1387 Record type. Always set to 7.
1389 @item int32 subtype;
1390 Record subtype. Always set to 17 for a data file attribute record or
1391 to 18 for a variable attributes record.
1394 The size of each element in the @code{attributes} member. Always set to 1.
1397 The total number of bytes in @code{attributes}.
1399 @item char attributes[];
1400 The attributes, in a text-based format.
1402 In record subtype 17, this field contains a single attribute set. An
1403 attribute set is a sequence of one or more attributes concatenated
1404 together. Each attribute consists of a name, which has the same
1405 syntax as a variable name, followed by, inside parentheses, a sequence
1406 of one or more values. Each value consists of a string enclosed in
1407 single quotes (@code{'}) followed by a line feed (byte 0x0a). A value
1408 may contain single quote characters, which are not themselves escaped
1409 or quoted or required to be present in pairs. There is no apparent
1410 way to embed a line feed in a value. There is no distinction between
1411 an attribute with a single value and an attribute array with one
1414 In record subtype 18, this field contains a sequence of one or more
1415 variable attribute sets. If more than one variable attribute set is
1416 present, each one after the first is delimited from the previous by
1417 @code{/}. Each variable attribute set consists of a long
1419 followed by @code{:}, followed by an attribute set with the same
1420 syntax as on record subtype 17.
1422 System files written by @code{Stata 14.1/-savespss- 1.77 by
1423 S.Radyakin} may include multiple records with subtype 18, one per
1424 variable that has variable attributes.
1426 The total length is @code{count} bytes.
1431 A system file produced with the following VARIABLE ATTRIBUTE commands
1435 VARIABLE ATTRIBUTE VARIABLES=dummy ATTRIBUTE=fred[1]('23') fred[2]('34').
1436 VARIABLE ATTRIBUTE VARIABLES=dummy ATTRIBUTE=bert('123').
1440 will contain a variable attribute record with the following contents:
1443 0000 07 00 00 00 12 00 00 00 01 00 00 00 22 00 00 00 |............"...|
1444 0010 64 75 6d 6d 79 3a 66 72 65 64 28 27 32 33 27 0a |dummy:fred('23'.|
1445 0020 27 33 34 27 0a 29 62 65 72 74 28 27 31 32 33 27 |'34'.)bert('123'|
1453 @node Variable Roles
1454 @subsection Variable Roles
1456 A variable's role is represented as an attribute named @code{$@@Role}.
1457 This attribute has a single element whose values and their meanings
1462 Input. This, the default, is the most common role.
1475 @node Extended Number of Cases Record
1476 @section Extended Number of Cases Record
1478 The file header record expresses the number of cases in the system
1479 file as an int32 (@pxref{File Header Record}). This record allows the
1480 number of cases in the system file to be expressed as a 64-bit number.
1492 @item int32 rec_type;
1493 Record type. Always set to 7.
1495 @item int32 subtype;
1496 Record subtype. Always set to 16.
1499 Size of each element. Always set to 8.
1502 Number of pieces of data in the data part. Alway set to 2.
1504 @item int64 unknown;
1505 Meaning unknown. Always set to 1.
1507 @item int64 ncases64;
1508 Number of cases in the file as a 64-bit integer. Presumably this
1509 could be -1 to indicate that the number of cases is unknown, for the
1510 same reason as @code{ncases} in the file header record, but this has
1511 not been observed in the wild.
1514 @node Other Informational Records
1515 @section Other Informational Records
1517 This chapter documents many specific types of extension records are
1518 documented here, but others are known to exist. PSPP ignores unknown
1519 extension records when reading system files.
1521 The following extension record subtypes have also been observed, with
1522 the following believed meanings:
1526 A named variable set for use in the GUI (according to Aapi
1530 Date info, probably related to USE (according to Aapi H@"am@"al@"ainen).
1533 A UUID in the format described in RFC 4122. Only two examples
1534 observed, both written by SPSS 13, and in each case the UUID contained
1535 both upper and lower case.
1538 XML that describes how data in the file should be displayed on-screen.
1541 @node Dictionary Termination Record
1542 @section Dictionary Termination Record
1544 The dictionary termination record separates all other records from the
1553 @item int32 rec_type;
1554 Record type. Always set to 999.
1557 Ignored padding. Should be set to 0.
1561 @section Data Record
1563 The data record must follow all other records in the system file.
1564 Every system file must have a data record that specifies data for at
1565 least one case. The format of the data record varies depending on the
1566 value of @code{compression} in the file header record:
1569 @item 0: no compression
1570 Data is arranged as a series of 8-byte elements.
1571 Each element corresponds to
1572 the variable declared in the respective variable record (@pxref{Variable
1573 Record}). Numeric values are given in @code{flt64} format; string
1574 values are literal characters string, padded on the right when
1575 necessary to fill out 8-byte units.
1577 @item 1: bytecode compression
1579 of the data record is divided into a series of 1-byte command
1580 codes. These codes have meanings as described below:
1584 Ignored. If the program writing the system file accumulates compressed
1585 data in blocks of fixed length, 0 bytes can be used to pad out extra
1586 bytes remaining at the end of a fixed-size block.
1590 value @var{code} - @var{bias}, where
1591 @var{code} is the value of the compression code and @var{bias} is the
1592 variable @code{bias} from the file header. For example,
1593 code 105 with bias 100.0 (the normal value) indicates a numeric variable
1596 A code of 0 (after subtracting the bias) in a string field encodes
1597 null bytes. This is unusual, since a string field normally encodes
1598 text data, but it exists in real system files.
1601 End of file. This code may or may not appear at the end of the data
1602 stream. PSPP always outputs this code but its use is not required.
1605 A numeric or string value that is not
1606 compressible. The value is stored in the 8 bytes following the
1607 current block of command bytes. If this value appears twice in a block
1608 of command bytes, then it indicates the second group of 8 bytes following the
1609 command bytes, and so on.
1612 An 8-byte string value that is all spaces.
1615 The system-missing value.
1618 The end of the 8-byte group of bytecodes is followed by any 8-byte
1619 blocks of non-compressible values indicated by code 253. After that
1620 follows another 8-byte group of bytecodes, then those bytecodes'
1621 non-compressible values. The pattern repeats to the end of the file
1622 or a code with value 252.
1624 @item 2: ZLIB compression
1625 The data record consists of the following, in order:
1629 ZLIB data header, 24 bytes long.
1632 One or more variable-length blocks of ZLIB compressed data.
1635 ZLIB data trailer, with a 24-byte fixed header plus an additional 24
1636 bytes for each preceding ZLIB compressed data block.
1639 The ZLIB data header has the following format:
1648 @item int64 zheader_ofs;
1649 The offset, in bytes, of the beginning of this structure within the
1652 @item int64 ztrailer_ofs;
1653 The offset, in bytes, of the first byte of the ZLIB data trailer.
1655 @item int64 ztrailer_len;
1656 The number of bytes in the ZLIB data trailer. This and the previous
1657 field sum to the size of the system file in bytes.
1660 The data header is followed by @code{(ztrailer_len - 24) / 24} ZLIB
1661 compressed data blocks. Each ZLIB compressed data block begins with a
1662 ZLIB header as specified in RFC@tie{}1950, e.g.@: hex bytes @code{78
1663 01} (the only header yet observed in practice). Each block
1664 decompresses to a fixed number of bytes (in practice only
1665 @code{0x3ff000}-byte blocks have been observed), except that the last
1666 block of data may be shorter. The last ZLIB compressed data block
1667 gends just before offset @code{ztrailer_ofs}.
1669 The result of ZLIB decompression is bytecode compressed data as
1670 described above for compression format 1.
1672 The ZLIB data trailer begins with the following 24-byte fixed header:
1682 @item int64 int_bias;
1683 The compression bias as a negative integer, e.g.@: if @code{bias} in
1684 the file header record is 100.0, then @code{int_bias} is @minus{}100
1685 (this is the only value yet observed in practice).
1688 Always observed to be zero.
1690 @item int32 block_size;
1691 The number of bytes in each ZLIB compressed data block, except
1692 possibly the last, following decompression. Only @code{0x3ff000} has
1693 been observed so far.
1695 @item int32 n_blocks;
1696 The number of ZLIB compressed data blocks, always exactly
1697 @code{(ztrailer_len - 24) / 24}.
1700 The fixed header is followed by @code{n_blocks} 24-byte ZLIB data
1701 block descriptors, each of which describes the compressed data block
1702 corresponding to its offset. Each block descriptor has the following
1706 int64 uncompressed_ofs;
1707 int64 compressed_ofs;
1708 int32 uncompressed_size;
1709 int32 compressed_size;
1713 @item int64 uncompressed_ofs;
1714 The offset, in bytes, that this block of data would have in a similar
1715 system file that uses compression format 1. This is
1716 @code{zheader_ofs} in the first block descriptor, and in each
1717 succeeding block descriptor it is the sum of the previous desciptor's
1718 @code{uncompressed_ofs} and @code{uncompressed_size}.
1720 @item int64 compressed_ofs;
1721 The offset, in bytes, of the actual beginning of this compressed data
1722 block. This is @code{zheader_ofs + 24} in the first block descriptor,
1723 and in each succeeding block descriptor it is the sum of the previous
1724 descriptor's @code{compressed_ofs} and @code{compressed_size}. The
1725 final block descriptor's @code{compressed_ofs} and
1726 @code{compressed_size} sum to @code{ztrailer_ofs}.
1728 @item int32 uncompressed_size;
1729 The number of bytes in this data block, after decompression. This is
1730 @code{block_size} in every data block except the last, which may be
1733 @item int32 compressed_size;
1734 The number of bytes in this data block, as stored compressed in this
1739 @setfilename ignored