1 @node System File Format
2 @appendix System File Format
4 A system file encapsulates a set of cases and dictionary information
5 that describes how they may be interpreted. This chapter describes
6 the format of a system file.
8 System files use three data types: 8-bit characters, 32-bit integers,
9 and 64-bit floating points, called here @code{char}, @code{int32}, and
10 @code{flt64}, respectively. Data is not necessarily aligned on a word
11 or double-word boundary: the long variable name record (@pxref{Long
12 Variable Names Record}) and very long string records (@pxref{Very Long
13 String Record}) have arbitrary byte length and can therefore cause all
14 data coming after them in the file to be misaligned.
16 Integer data in system files may be big-endian or little-endian. A
17 reader may detect the endianness of a system file by examining
18 @code{layout_code} in the file header record
19 (@pxref{layout_code,,@code{layout_code}}).
21 Floating-point data in system files may nominally be in IEEE 754, IBM,
22 or VAX formats. A reader may detect the floating-point format in use
23 by examining @code{bias} in the file header record
24 (@pxref{bias,,@code{bias}}).
26 PSPP detects big-endian and little-endian integer formats in system
27 files and translates as necessary. PSPP also detects the
28 floating-point format in use, as well as the endianness of IEEE 754
29 floating-point numbers, and translates as needed. However, only IEEE
30 754 numbers with the same endianness as integer data in the same file
31 has actually been observed in system files, and it is likely that
32 other formats are obsolete or were never used.
34 The PSPP system-missing value is represented by the largest possible
35 negative number in the floating point format (@code{-DBL_MAX}). Two
36 other values are important for use as missing values: @code{HIGHEST},
37 represented by the largest possible positive number (@code{DBL_MAX}),
38 and @code{LOWEST}, represented by the second-largest negative number
39 (in IEEE 754 format, @code{0xffeffffffffffffe}).
41 System files are divided into records, each of which begins with a
42 4-byte record type, usually regarded as an @code{int32}.
44 The records must appear in the following order:
54 All pairs of value labels records and value label variables records,
58 Document record, if present.
61 Any of the following records, if present, in any order:
65 Machine integer info record.
68 Machine floating-point info record.
71 Variable display parameter record.
74 Long variable names record.
77 Miscellaneous informational records.
81 Dictionary termination record.
87 Each type of record is described separately below.
90 * File Header Record::
92 * Value Labels Records::
94 * Machine Integer Info Record::
95 * Machine Floating-Point Info Record::
96 * Variable Display Parameter Record::
97 * Long Variable Names Record::
98 * Very Long String Record::
99 * Data File and Variable Attributes Records::
100 * Miscellaneous Informational Records::
101 * Dictionary Termination Record::
105 @node File Header Record
106 @section File Header Record
108 The file header is always the first record in the file. It has the
115 int32 nominal_case_size;
120 char creation_date[9];
121 char creation_time[8];
127 @item char rec_type[4];
128 Record type code, set to @samp{$FL2}.
130 @item char prod_name[60];
131 Product identification string. This always begins with the characters
132 @samp{@@(#) SPSS DATA FILE}. PSPP uses the remaining characters to
133 give its version and the operating system name; for example, @samp{GNU
134 pspp 0.1.4 - sparc-sun-solaris2.5.2}. The string is truncated if it
135 would be longer than 60 characters; otherwise it is padded on the right
139 @item int32 layout_code;
140 Normally set to 2, although a few system files have been spotted in
141 the wild with a value of 3 here. PSPP use this value to determine the
142 file's integer endianness (@pxref{System File Format}).
144 @item int32 nominal_case_size;
145 Number of data elements per case. This is the number of variables,
146 except that long string variables add extra data elements (one for every
147 8 characters after the first 8). However, string variables do not
148 contribute to this value beyond the first 255 bytes. Further, system
149 files written by some systems set this value to -1. In general, it is
150 unsafe for systems reading system files to rely upon this value.
152 @item int32 compressed;
153 Set to 1 if the data in the file is compressed, 0 otherwise.
155 @item int32 weight_index;
156 If one of the variables in the data set is used as a weighting
157 variable, set to the dictionary index of that variable, plus 1
158 (@pxref{Dictionary Index}). Otherwise, set to 0.
161 Set to the number of cases in the file if it is known, or -1 otherwise.
163 In the general case it is not possible to determine the number of cases
164 that will be output to a system file at the time that the header is
165 written. The way that this is dealt with is by writing the entire
166 system file, including the header, then seeking back to the beginning of
167 the file and writing just the @code{ncases} field. For `files' in which
168 this is not valid, the seek operation fails. In this case,
169 @code{ncases} remains -1.
173 Compression bias, ordinarily set to 100. Only integers between
174 @code{1 - bias} and @code{251 - bias} can be compressed.
176 By assuming that its value is 100, PSPP uses @code{bias} to determine
177 the file's floating-point format and endianness (@pxref{System File
178 Format}). If the compression bias is not 100, PSPP cannot auto-detect
179 the floating-point format and assumes that it is IEEE 754 format with
180 the same endianness as the system file's integers, which is correct
181 for all known system files.
183 @item char creation_date[9];
184 Date of creation of the system file, in @samp{dd mmm yy}
185 format, with the month as standard English abbreviations, using an
186 initial capital letter and following with lowercase. If the date is not
187 available then this field is arbitrarily set to @samp{01 Jan 70}.
189 @item char creation_time[8];
190 Time of creation of the system file, in @samp{hh:mm:ss}
191 format and using 24-hour time. If the time is not available then this
192 field is arbitrarily set to @samp{00:00:00}.
194 @item char file_label[64];
195 File label declared by the user, if any (@pxref{FILE LABEL,,,pspp,
196 PSPP Users Guide}). Padded on the right with spaces.
198 @item char padding[3];
199 Ignored padding bytes to make the structure a multiple of 32 bits in
200 length. Set to zeros.
203 @node Variable Record
204 @section Variable Record
206 There must be one variable record for each numeric variable and each
207 string variable with width 8 bytes or less. String variables wider
208 than 8 bytes have one variable record for each 8 bytes, rounding up.
209 The first variable record for a long string specifies the variable's
210 correct dictionary information. Subsequent variable records for a
211 long string are filled with dummy information: a type of -1, no
212 variable label or missing values, print and write formats that are
213 ignored, and an empty string as name. A few system files have been
214 encountered that include a variable label on dummy variable records,
215 so readers should take care to parse dummy variable records in the
216 same way as other variable records.
218 @anchor{Dictionary Index}
219 The @dfn{dictionary index} of a variable is its offset in the set of
220 variable records, including dummy variable records for long string
221 variables. The first variable record has a dictionary index of 0, the
222 second has a dictionary index of 1, and so on.
224 The system file format does not directly support string variables
225 wider than 255 bytes. Such very long string variables are represented
226 by a number of narrower string variables. @xref{Very Long String
227 Record}, for details.
233 int32 n_missing_values;
238 /* @r{Present only if @code{has_var_label} is 1.} */
242 /* @r{Present only if @code{n_missing_values} is nonzero}. */
243 flt64 missing_values[];
247 @item int32 rec_type;
248 Record type code. Always set to 2.
251 Variable type code. Set to 0 for a numeric variable. For a short
252 string variable or the first part of a long string variable, this is set
253 to the width of the string. For the second and subsequent parts of a
254 long string variable, set to -1, and the remaining fields in the
255 structure are ignored.
257 @item int32 has_var_label;
258 If this variable has a variable label, set to 1; otherwise, set to 0.
260 @item int32 n_missing_values;
261 If the variable has no missing values, set to 0. If the variable has
262 one, two, or three discrete missing values, set to 1, 2, or 3,
263 respectively. If the variable has a range for missing variables, set to
264 -2; if the variable has a range for missing variables plus a single
265 discrete value, set to -3.
268 Print format for this variable. See below.
271 Write format for this variable. See below.
274 Variable name. The variable name must begin with a capital letter or
275 the at-sign (@samp{@@}). Subsequent characters may also be digits, octothorpes
276 (@samp{#}), dollar signs (@samp{$}), underscores (@samp{_}), or full
277 stops (@samp{.}). The variable name is padded on the right with spaces.
279 @item int32 label_len;
280 This field is present only if @code{has_var_label} is set to 1. It is
281 set to the length, in characters, of the variable label, which must be a
282 number between 0 and 120.
285 This field is present only if @code{has_var_label} is set to 1. It has
286 length @code{label_len}, rounded up to the nearest multiple of 32 bits.
287 The first @code{label_len} characters are the variable's variable label.
289 @item flt64 missing_values[];
290 This field is present only if @code{n_missing_values} is not 0. It has
291 the same number of elements as the absolute value of
292 @code{n_missing_values}. For discrete missing values, each element
293 represents one missing value. When a range is present, the first
294 element denotes the minimum value in the range, and the second element
295 denotes the maximum value in the range. When a range plus a value are
296 present, the third element denotes the additional discrete missing
297 value. HIGHEST and LOWEST are indicated as described in the chapter
301 The @code{print} and @code{write} members of sysfile_variable are output
302 formats coded into @code{int32} types. The least-significant byte
303 of the @code{int32} represents the number of decimal places, and the
304 next two bytes in order of increasing significance represent field width
305 and format type, respectively. The most-significant byte is not
306 used and should be set to zero.
308 Format types are defined as follows:
311 @multitable {Value} {@code{DATETIME}}
397 @node Value Labels Records
398 @section Value Labels Records
400 The value label record has the following format:
406 /* @r{Repeated @code{label_cnt} times}. */
413 @item int32 rec_type;
414 Record type. Always set to 3.
416 @item int32 label_count;
417 Number of value labels present in this record.
420 The remaining fields are repeated @code{count} times. Each
421 repetition specifies one value label.
425 A numeric value or a short string value padded as necessary to 8 bytes
426 in length. Its type and width cannot be determined until the
427 following value label variables record (see below) is read.
429 @item char label_len;
430 The label's length, in bytes.
433 @code{label_len} bytes of the actual label, followed by up to 7 bytes
434 of padding to bring @code{label} and @code{label_len} together to a
435 multiple of 8 bytes in length.
438 The value label record is always immediately followed by a value label
439 variables record with the following format:
448 @item int32 rec_type;
449 Record type. Always set to 4.
451 @item int32 var_count;
452 Number of variables that the associated value labels from the value
453 label record are to be applied.
456 A list of dictionary indexes of variables to which to apply the value
457 labels (@pxref{Dictionary Index}). There are @code{var_count}
460 String variables wider than 8 bytes may not have value labels.
463 @node Document Record
464 @section Document Record
466 The document record, if present, has the following format:
475 @item int32 rec_type;
476 Record type. Always set to 6.
479 Number of lines of documents present.
481 @item char lines[][80];
482 Document lines. The number of elements is defined by @code{n_lines}.
483 Lines shorter than 80 characters are padded on the right with spaces.
486 @node Machine Integer Info Record
487 @section Machine Integer Info Record
489 The integer info record, if present, has the following format:
501 int32 version_revision;
503 int32 floating_point_rep;
504 int32 compression_code;
506 int32 character_code;
510 @item int32 rec_type;
511 Record type. Always set to 7.
514 Record subtype. Always set to 3.
517 Size of each piece of data in the data part, in bytes. Always set to 4.
520 Number of pieces of data in the data part. Always set to 8.
522 @item int32 version_major;
523 PSPP major version number. In version @var{x}.@var{y}.@var{z}, this
526 @item int32 version_minor;
527 PSPP minor version number. In version @var{x}.@var{y}.@var{z}, this
530 @item int32 version_revision;
531 PSPP version revision number. In version @var{x}.@var{y}.@var{z},
534 @item int32 machine_code;
535 Machine code. PSPP always set this field to value to -1, but other
538 @item int32 floating_point_rep;
539 Floating point representation code. For IEEE 754 systems this is 1.
540 IBM 370 sets this to 2, and DEC VAX E to 3.
542 @item int32 compression_code;
543 Compression code. Always set to 1.
545 @item int32 endianness;
546 Machine endianness. 1 indicates big-endian, 2 indicates little-endian.
548 @item int32 character_code;
549 Character code. 1 indicates EBCDIC, 2 indicates 7-bit ASCII, 3
550 indicates 8-bit ASCII, 4 indicates DEC Kanji.
551 Windows code page numbers are also valid.
554 @node Machine Floating-Point Info Record
555 @section Machine Floating-Point Info Record
557 The floating-point info record, if present, has the following format:
573 @item int32 rec_type;
574 Record type. Always set to 7.
577 Record subtype. Always set to 4.
580 Size of each piece of data in the data part, in bytes. Always set to 8.
583 Number of pieces of data in the data part. Always set to 3.
586 The system missing value.
589 The value used for HIGHEST in missing values.
592 The value used for LOWEST in missing values.
595 @node Variable Display Parameter Record
596 @section Variable Display Parameter Record
598 The variable display parameter record, if present, has the following
608 /* @r{Repeated @code{count} times}. */
610 int32 width; /* @r{Not always present.} */
615 @item int32 rec_type;
616 Record type. Always set to 7.
619 Record subtype. Always set to 11.
622 The size of @code{int32}. Always set to 4.
625 The number of sets of variable display parameters (ordinarily the
626 number of variables in the dictionary), times 2 or 3.
629 The remaining members are repeated @code{count} times, in the same
630 order as the variable records. No element corresponds to variable
631 records that continue long string variables. The meanings of these
632 members are as follows:
636 The measurement type of the variable:
646 SPSS 14 sometimes writes a @code{measure} of 0. PSPP interprets this
650 The width of the display column for the variable in characters.
652 This field is present if @var{count} is 3 times the number of
653 variables in the dictionary. It is omitted if @var{count} is 2 times
654 the number of variables.
656 @item int32 alignment;
657 The alignment of the variable for display purposes:
669 @node Long Variable Names Record
670 @section Long Variable Names Record
672 If present, the long variable names record has the following format:
681 /* @r{Exactly @code{count} bytes of data.} */
682 char var_name_pairs[];
686 @item int32 rec_type;
687 Record type. Always set to 7.
690 Record subtype. Always set to 13.
693 The size of each element in the @code{var_name_pairs} member. Always set to 1.
696 The total number of bytes in @code{var_name_pairs}.
698 @item char var_name_pairs[];
699 A list of @var{key}--@var{value} tuples, where @var{key} is the name
700 of a variable, and @var{value} is its long variable name.
701 The @var{key} field is at most 8 bytes long and must match the
702 name of a variable which appears in the variable record (@pxref{Variable
704 The @var{value} field is at most 64 bytes long.
705 The @var{key} and @var{value} fields are separated by a @samp{=} byte.
706 Each tuple is separated by a byte whose value is 09. There is no
707 trailing separator following the last tuple.
708 The total length is @code{count} bytes.
711 @node Very Long String Record
712 @section Very Long String Record
714 Old versions of SPSS limited string variables to a width of 255 bytes.
715 For backward compatibility with these older versions, the system file
716 format represents a string longer than 255 bytes, called a @dfn{very
717 long string}, as a collection of strings no longer than 255 bytes
718 each. The strings concatenated to make a very long string are called
719 its @dfn{segments}; for consistency, variables other than very long
720 strings are considered to have a single segment.
722 A very long string with a width of @var{w} has @var{n} =
723 (@var{w} + 251) / 252 segments, that is, one segment for every
724 252 bytes of width, rounding up. It would be logical, then, for each
725 of the segments except the last to have a width of 252 and the last
726 segment to have the remainder, but this is not the case. In fact,
727 each segment except the last has a width of 255 bytes. The last
728 segment has width @var{w} - (@var{n} - 1) * 252; some versions
729 of SPSS make it slightly wider, but not wide enough to make the last
730 segment require another 8 bytes of data.
732 Data is packed tightly into segments of a very long string, 255 bytes
733 per segment. Because 255 bytes of segment data are allocated for
734 every 252 bytes of the very long string's width (approximately), some
735 unused space is left over at the end of the allocated segments. Data
736 in unused space is ignored.
738 Example: Consider a very long string of width 20,000. Such a very
739 long string has 20,000 / 252 = 80 (rounding up) segments. The first
740 79 segments have width 255; the last segment has width 20,000 - 79 *
741 252 = 92 or slightly wider (up to 96 bytes, the next multiple of 8).
742 The very long string's data is actually stored in the 19,890 bytes in
743 the first 78 segments, plus the first 110 bytes of the 79th segment
744 (19,890 + 110 = 20,000). The remaining 145 bytes of the 79th segment
745 and all 92 bytes of the 80th segment are unused.
747 The very long string record explains how to stitch together segments
748 to obtain very long string data. For each of the very long string
749 variables in the dictionary, it specifies the name of its first
750 segment's variable and the very long string variable's actual width.
751 The remaining segments immediately follow the named variable in the
752 system file's dictionary.
754 The very long string record, which is present only if the system file
755 contains very long string variables, has the following format:
764 /* @r{Exactly @code{count} bytes of data.} */
765 char string_lengths[];
769 @item int32 rec_type;
770 Record type. Always set to 7.
773 Record subtype. Always set to 14.
776 The size of each element in the @code{string_lengths} member. Always set to 1.
779 The total number of bytes in @code{string_lengths}.
781 @item char string_lengths[];
782 A list of @var{key}--@var{value} tuples, where @var{key} is the name
783 of a variable, and @var{value} is its length.
784 The @var{key} field is at most 8 bytes long and must match the
785 name of a variable which appears in the variable record (@pxref{Variable
787 The @var{value} field is exactly 5 bytes long. It is a zero-padded,
788 ASCII-encoded string that is the length of the variable.
789 The @var{key} and @var{value} fields are separated by a @samp{=} byte.
790 Tuples are delimited by a two-byte sequence @{00, 09@}.
791 After the last tuple, there may be a single byte 00, or @{00, 09@}.
792 The total length is @code{count} bytes.
795 @node Data File and Variable Attributes Records
796 @section Data File and Variable Attributes Records
798 The data file and variable attributes records represent custom
799 attributes for the system file or for individual variables in the
800 system file, as defined on the DATAFILE ATTRIBUTE (@pxref{DATAFILE
801 ATTRIBUTE,,,pspp, PSPP Users Guide}) and VARIABLE ATTRIBUTE commands
802 (@pxref{VARIABLE ATTRIBUTE,,,pspp, PSPP Users Guide}), respectively.
811 /* @r{Exactly @code{count} bytes of data.} */
816 @item int32 rec_type;
817 Record type. Always set to 7.
820 Record subtype. Always set to 17 for a data file attribute record or
821 to 18 for a variable attributes record.
824 The size of each element in the @code{attributes} member. Always set to 1.
827 The total number of bytes in @code{attributes}.
829 @item char attributes[];
830 The attributes, in a text-based format.
832 In record type 17, this field contains a single attribute set. An
833 attribute set is a sequence of one or more attributes concatenated
834 together. Each attribute consists of a name, which has the same
835 syntax as a variable name, followed by, inside parentheses, a sequence
836 of one or more values. Each value consists of a string enclosed in
837 single quotes (@code{'}) followed by a line feed (byte 0x0a). A value
838 may contain single quote characters, which are not themselves escaped
839 or quoted or required to be present in pairs. There is no apparent
840 way to embed a line feed in a value. There is no distinction between
841 an attribute with a single value and an attribute array with one
844 In record type 18, this field contains a sequence of one or more
845 variable attribute sets. If more than one variable attribute set is
846 present, each one after the first is delimited from the previous by
847 @code{/}. Each variable attribute set consists of a variable name,
848 followed by @code{:}, followed by an attribute set with the same
849 syntax as on record type 17.
851 The total length is @code{count} bytes.
856 A system file produced with the following VARIABLE ATTRIBUTE commands
860 VARIABLE ATTRIBUTE VARIABLES=dummy ATTRIBUTE=fred[1]('23') fred[2]('34').
861 VARIABLE ATTRIBUTE VARIABLES=dummy ATTRIBUTE=bert('123').
865 will contain a variable attribute record with the following contents:
868 00000000 07 00 00 00 12 00 00 00 01 00 00 00 22 00 00 00 |............"...|
869 00000010 64 75 6d 6d 79 3a 66 72 65 64 28 27 32 33 27 0a |dummy:fred('23'.|
870 00000020 27 33 34 27 0a 29 62 65 72 74 28 27 31 32 33 27 |'34'.)bert('123'|
874 @node Miscellaneous Informational Records
875 @section Miscellaneous Informational Records
877 Some specific types of miscellaneous informational records are
878 documented here, but others are known to exist. PSPP ignores unknown
879 miscellaneous informational records when reading system files.
888 /* @r{Exactly @code{size * count} bytes of data.} */
893 @item int32 rec_type;
894 Record type. Always set to 7.
897 Record subtype. May take any value. According to Aapi
898 H@"am@"al@"ainen, value 5 indicates a set of grouped variables and 6
899 indicates date info (probably related to USE).
902 Size of each piece of data in the data part. Should have the value 1,
903 4, or 8, for @code{char}, @code{int32}, and @code{flt64} format data,
907 Number of pieces of data in the data part.
910 Arbitrary data. There must be @code{size} times @code{count} bytes of
914 @node Dictionary Termination Record
915 @section Dictionary Termination Record
917 The dictionary termination record separates all other records from the
926 @item int32 rec_type;
927 Record type. Always set to 999.
930 Ignored padding. Should be set to 0.
936 Data records must follow all other records in the system file. There must
937 be at least one data record in every system file.
939 The format of data records varies depending on whether the data is
940 compressed. Regardless, the data is arranged in a series of 8-byte
943 When data is not compressed,
944 each element corresponds to
945 the variable declared in the respective variable record (@pxref{Variable
946 Record}). Numeric values are given in @code{flt64} format; string
947 values are literal characters string, padded on the right when
948 necessary to fill out 8-byte units.
950 Compressed data is arranged in the following manner: the first 8 bytes
951 in the data section is divided into a series of 1-byte command
952 codes. These codes have meanings as described below:
956 Ignored. If the program writing the system file accumulates compressed
957 data in blocks of fixed length, 0 bytes can be used to pad out extra
958 bytes remaining at the end of a fixed-size block.
962 value @var{code} - @var{bias}, where
963 @var{code} is the value of the compression code and @var{bias} is the
964 variable @code{bias} from the file header. For example,
965 code 105 with bias 100.0 (the normal value) indicates a numeric variable
969 End of file. This code may or may not appear at the end of the data
970 stream. PSPP always outputs this code but its use is not required.
973 A numeric or string value that is not
974 compressible. The value is stored in the 8 bytes following the
975 current block of command bytes. If this value appears twice in a block
976 of command bytes, then it indicates the second group of 8 bytes following the
977 command bytes, and so on.
980 An 8-byte string value that is all spaces.
983 The system-missing value.
986 When the end of the an 8-byte group of command bytes is reached, any
987 blocks of non-compressible values indicated by code 253 are skipped,
988 and the next element of command bytes is read and interpreted, until
989 the end of the file or a code with value 252 is reached.