From 40e5a5811c6ffce9b0974e93cdd927cbcf60c157 Mon Sep 17 00:00:00 2001 From: Joe Hunkeler Date: Tue, 11 Aug 2015 16:51:37 -0400 Subject: Repatch (from linux) of OSX IRAF --- pkg/tbtables/doc/fileformat.doc | 91 +++++++++++++++++++++++++++++++++++++++++ 1 file changed, 91 insertions(+) create mode 100644 pkg/tbtables/doc/fileformat.doc (limited to 'pkg/tbtables/doc/fileformat.doc') diff --git a/pkg/tbtables/doc/fileformat.doc b/pkg/tbtables/doc/fileformat.doc new file mode 100644 index 00000000..3c9d3ef6 --- /dev/null +++ b/pkg/tbtables/doc/fileformat.doc @@ -0,0 +1,91 @@ + 1999 August 17 + + This note describes the file format for STSDAS tables. See +text_tables.doc for information on support for text tables in the +TABLES package. + +There are four sections to an STSDAS table: + +(1) a size-information record which gives the number of rows, etc +(2) optional header-parameter records +(3) a record for each column that describes the column +(4) the table data + + +1. The size-information record is 12 integers in length. Only the first +ten are used at the moment, and these have the following meanings: + + 1 The number of header parameters that have been written to the table. + 2 The maximum number of header parameters for which space has been allocated. + 3 The number of rows that have been written to the table. + 4 The allocated number of rows (relevant only for a column-ordered table). + 5 The number of columns that have been defined. + 6 The maximum number of column descriptors for which space has been + allocated. + 7 The length of the portion of the row that is used by columns that have + been defined; unit = SZ_CHAR (= two bytes). + 8 The allocated row length; unit = SZ_CHAR. This is relevant only for + a row-ordered table. + 9 Table type: 11 implies row-ordered; 12 implies column-ordered. + 10 Table software version number. Originally this was zero, and the + current value is three. The differences between versions are + described in the file "versions.doc". + + +2. The header parameters are FITS-like records for storing information such +as comments or numerical values. They are not used by the table routines +at all. There need not be any header parameters, and there need not be any +space allocated for them; that is, words one and two of the size-information +record may be zero. Each header-parameter record is 80 bytes in length and +contains the following: + +bytes description +----- ----------- + 1-8 Keyword; from one to eight characters padded on the right with blanks. + All letters will be in upper case. + 9 Data type; a single lower-case letter: t, b, i, r, d indicating a + type of text, boolean, integer, real, or double-precision respectively. + The data type has little real meaning since the value is ASCII. +10-80 The value, a left-justified ASCII string terminated by an ASCII null + and followed by garbage. Boolean true and false are represented by + one and zero respectively. + + +3. There is a column descriptor for each column that has been defined. Each +column-descriptor record has a length of 16 integer words and contains the +following: + + word description + ---- ----------- + 1 The column number. + 2 The offset from the start of the row; unit = SZ_CHAR (= two bytes). + This is the sum of the widths of all previous columns, or zero for + the first column. + 3 The amount of space (unit = SZ_CHAR) required to store one element. + 4 The data type: + 6 = single-precision real + 7 = double-precision real + 4 = integer + 3 = short integer + 1 = boolean + -n = character string containing up to n characters + 5-9 The column name, up to 20 characters, left justified. If the name + is shorter than 20 characters, it will be terminated by an ASCII + NULL; otherwise, the NULL will be omitted. Upper and lower case + are allowed, but they are not distinguished. + 10-14 The units, up to 20 characters, left justified. If the units string + is shorter than 20 characters, it will be terminated by an ASCII + NULL; otherwise, the NULL will be omitted. Upper and lower case + are allowed. + 15-16 The format for printing the column, up to eight characters. If the + string is shorter than eight characters, it will be terminated by + an ASCII NULL; otherwise, the NULL will be omitted. This format + string is an SPP-style format without the leading %. + + +4. The table data may be either row-ordered or column-ordered. Integer, +single- and double-precision reals, and boolean (logical) are stored in the +Fortran binary format of the host machine. For character data type, each +element will be terminated with an ASCII null if there is room for the null; +if an element fills the entire column width (a multiple of two bytes) then +the null will not be present. -- cgit