fitsTcl 2.0 User's Guide
FITS (Flexible Image Transport System) file format is widely used in the
astronomical community. But its complexity makes it very difficult for
general users to interact with data in FITS format.
Current FORTRAN and C interfaces to FITS -- FITSIO/CFITSIO, developed by
the HEASARC (High Energy Astrophysics Science Archive Research Center) at
the NASA Goddard Space Flight Center -- provide very powerful and robust
functions to interact with FITS files. However, utilizing FITSIO/CFITSIO
requires some basic FORTRAN/C programming skills, and this low level
library often is more powerful, and hence more complex, than the general
user needs.
To provide general users with a simple interface to read/write FITS
file, we developed fitsTcl, an extension to the simple but powerful
script language TCL/TK,
which uses the CFITSIO library. fitsTcl was developed for the
fv project and is distributed as part of fv, which can
be obtained by visiting the fv
website. fitsTcl can also be downloaded separately from the fitsTcl
web page. fitsTcl runs under Unix, Windows, and MacOS.
fitsTcl is compiled as a dynamic library which can be loaded as a TCL
extension. To load fitsTcl, type the following command in a Tcl shell
(running on UNIX)
% load libfitstcl.so
(Under MacOS and Windows, the library is instead named
fitstcl.dll.) If the library is not found you may need to
set the environment variable LD_LIBRARY_PATH (Unix only) to the
directory containing libfitstcl.so or include an explicit path to the
file.
In fitsTcl, every FITS file is treated as a FitsFile object. The
following sections describe how to create FitsFile objects and all
the methods in a FitsFile object.
Handling FitsFile objects
- fits open fileName mode ?objName?
- Create a FitsFile object, open the file and read the
header of the current HDU. (Read the CFITSIO web pages
or documentation to learn about its "Extended File Name Syntax".)
mode can be 0 (read only) or 1 (read and write). If no
objName is given, a name will be automatically assigned. A
new empty FITS file can be created by setting mode to 2. In
this case, if the named file exists, then it will be
overwritten. Returns objName which will be used in most of
the methods as the handle to the object.
- fits info ?objName ...?
- Return a list of all the existing objects, or the named objects
(supports regular expressions), with the following information:
{objName fileName mode CHDU hduType}
where CHDU is the current HDU (1-primary array, 2-the first extension
...) and hduType is the type of the HDU (0-image, 1-ASCII table,
2-Binary table).
- objName move ?+/-?n
- If n has ``+/-'' sign, then move n units relative to the
CHDU, otherwise move to the nth HDU. Return the extension type of
the new HDU: 0-image, 1-ASCII table, 2-Binary Table.
- objName close
- Delete object objName.
- fits close
- Delete all the existing objects.
Getting information about the CHDU
- objName info chdu
- Return the CHDU. 1=primary, 2=first extension, etc.
- objName info filesize
- Return the size of the file (in units of 2880 bytes).
- objName info hdutype
- Return ``Primary array'', ``Image extension'', ``Binary Table'',
or ``ASCII Table''.
- objName info imgdim
- Return the dimensions of the image if the CHDU is an image extension.
- objName info ncols
- Return the number of columns in the table if the CHDU is a table
extension.
- objName info nrows
- Return the number of rows in the table if the CHDU is a table extension.
- objName info nkwds
- Return the number of keywords in the header of the CHDU.
- objName info column ?-exact? ?colList?
- Without any arguments, return a list of all the column names
in the table if the CHDU is a table extension. If colList
is present, detailed information about the listed columns, using
regular expression name comparisons, is returned in the form of:
{Name Type Unit DisplayFormat DefaultFormat ColumnWidth isOffset
isScaled defaultNull}
where
|
Name |
name of the column (TTYPE keyword) |
Type |
type of the column (TFORM keyword) |
Unit |
unit of the column (TUNIT keyword) |
DisplayFormat |
format for display (TDISP keyword) |
DefaultFormat |
default format for display (if TDISP keyword is absent) |
ColumnWidth |
the width of the column (in units of characters). |
isOffset |
0 = no offset (no TZERO keyword), 1 = offset. |
isScaled |
0 = not scaled (no TSCALE keyword), 1 = scaled. |
defaultNull |
default NULL value (TNULL keyword) |
The -exact option turns off the regular expression matching.
Reading data from a FITS file
- objName dump ?-s/-e/-l?
- Return all the keyword records in the current header. The
following options control how the information is formatted:
|
none |
return list of all cards in header |
-s |
return three lists: {keywords} {values}
{comments} |
-e |
return a single string containing a newline-separated list
of all header records |
-l |
return all the keyword names |
- objName get keyword ?keyList?
- Return a list of {keyword value comment} for all the
keywords in keyList (supports regular expressions). If no
keyList is given, return a list of entries for all the keywords
in the CHDU. The order in which keywords are listed is
undefined. (HISTORY and COMMENT keywords are never returned.)
- objName get keyword -num n
- Return a list of {keyword value comment} of the nth
keyword.
- objName get wcs ?RAcol DecCol?
- Return a list of the WCS parameters -- {xrval yrval xrpix yrpix xinc
yinc rot ctype} -- for the CHDU. If the current HDU is a table,
supply the RAcol and DecCol parameters specifying the two
columns (as names or column numbers) containing WCS data. Use defaults for
any missing WCS keywords: 1.0 for xinc/yinc; ``none'' for
ctype; and 0.0 or all others. If the RA/Dec identity of the
columns (or image axes) are not known, an extra WCS parameter --
wcsSwap --
can be obtained which, when on, indicates that the X parameters are
for Dec instead of RA. Turn this option on with
fits option wcsSwap 1.
- objName get image ?firstElem? ?numElem?
- Read data elements from the current IMAGE extension and return
them in a list. Without any parameters, the entire image will be returned.
If only firstElem is given, return just one element.
- objName get table ?-c? ?colList? ?rows?
- Read a block of a table and return list(s) of data. The block range is
set in
|
colList |
a list of all the columns one wants to read. Read all
columns if missing or is ``*''. |
rows |
a comma-separated list of row ranges of the form
start-end (eg, ``3-5,7,9-'') |
-c |
return data from each column as a separate list |
- objName get vtable colName n
- Return a list of data in the nth vector element in column
colName.
Loading data into memory or to a TCL array
When loading data into memory, fitsTcl will normally return
address dataType numElements where these are:
|
address |
Memory address of the data. Can be recovered in C by
sscanf(address, "%p", &databuff);
with void *databuff |
dataType |
|
|
|
0 - BYTE_DATA | 1 byte data |
1 - SHORTINT_DATA | 2 byte integer |
2 - INT_DATA | 4 byte integer |
3 - FLOAT_DATA | 4 byte floating point |
4 - DOUBLE_DATA | 8 byte floating point |
|
numElement |
Number of data elements in the array |
- objName load image ?slice? ?rotate?
- Load a full 2D image in the CHDU into memory and return the address
address (see above). slice indicates which frame to use
if image is 3D. rotate indicates how many 90-degree
counter-clockwise rotations to perform on the image (0-3).
- objName load iblock arrayName firstRow numRows
firstCol numCols ?slice?
- Load a block (start from firstCol, firstRow with size
numCols x numRows) of image data to a 2-D TCL array
arrayName or to memory if arrayName is ``--''. The
indices of the array variable are (firstCol-1 ...
firstCol+numCols-1, firstRow-1 ... firstRow+numRows-1).
A 1D image will be treated as either a single row or column, depending
on the row/column parameters supplied. For a 3D image, slice
indicates which frame to use.
If arrayName is ``--'', read the data block into memory and
return the pointer information: address dataType numElements.
- objName load irows firstRow lastRow ?slice?
- objName load icols firstCol lastCol ?slice?
- Read and average together a range of rows or columns of an image.
Returns address dataType numElements. dataType will be
3 for all image data types except for double for which dataType
is 4.
- objName load column colName ?defaultNull?
?firstElement?
- Load a column of a table into memory and return address dataType
numElements. Use the value of defaultNull for NULL values, or
internal defaults (normally the data type's maximum value) if ``NULL''
or absent. If colName is a vector column, read the
firstElement-th element. One can only load numerical columns.
- objName load vtable colName
- Load all elements of the vector column colName and return
address dataType numElements.
- objName load tblock arrayName colList firstRow numRows
colIndex ?firstElem?
- Load a block of table data to a 2-D TCL array arrayName.
colIndex is (almost) the column index of the array to use for
the first column read. The first data index is actually (as for images)
(colIndex-1,firstRow-1). For
scaler columns, firstElem = 1.
- objName load expr expression ?defaultNull?
- Evaluate the arithmetic expression on each row of the CHDU's
table and return address dataType numElements, using
defaultNull as the null value of any NULL results.
- objName load keyword
- Load keywords into fitsTcl's internal hash table. One usually will
not need to call this routine.
- objName load chdu
- Reload the information about the current HDU. One usually will not
need to call this routine.
Writing to a FITS file
- objName insert image bitpix naxis naxesList
- objName insert image -p ?bitpix naxis naxesList?
- Insert an empty image HDU (primary header with -p) after the
current HDU (or, at start of file for primary array). The image parameters
-- bitpix naxis naxesList -- can be left off if
creating a zero-length primary array.
- objName insert table numRows {colNames}
{colFormats} ?{colUnits} extensionName?
- objName insert table -ascii numRows {colNames}
{colFormats} ?{colUnits} {colPosition} extensionName rowWidth?
- Insert an empty BINARY (default) or ASCII (with -ascii
flag) table HDU after the current HDU. The colNames and
colFormats lists must be of equal length (possibly both
empty). The optional parameters can each be empty if one wants to
use default/empty values.
- objName insert keyword index record ?formatFlag?
- Insert a new record at the index-th keyword. If
formatFlag is 0, write the record exactly as passed, otherwise
parse it as a free form keyname value comment and reformat it into a
standardized KEYNAME = VALUE / COMMENT (default).
- objName put keyword ?-num index? record
?formatFlag?
- Write a keyword record either at position index; the position
of a pre-existing keyword with the same name; or append it if no such
keyword exists. See above for format/meaning of record and
formatFlag.
- objName put history historyStr
- Write a HISTORY keyword with content historyStr
- objName insert column index colName colFormat
- Insert an empty column with format colFormat before the
index-th column. colFormat specifies the column format
as, for example:
ASCII Table: A15, I10, E12.5, D20.10, F14.6 ...
BINARY Table: 15A, 1I, 1J, 1E, 1D, 1L, 1X, 1B, 1C, 1M
- objName add column colName colFormat ?expression?
- Without an expression, append an empty column to the
CHDU table; return nothing. Given an arithmetic
expression, though, the indicated column (new or old) will
be filled in with the results of the expression evaluated for each
row of the table. If colName does not exist in the current
table extension, a new one will be created with format
colFormat (see above). If colFormat is ``default'',
the column will be created based on the data type of the result.
Return 1 or 0 to indicate whether a new column was created (1) or
not (0). expression can be in either C or Fortran format
with table columns and keywords referenced by their names. Here
are some expression samples:
|
Expression |
Result |
|
17.2 |
17.2 for every row |
17 + 4*(PHI > 32) |
17 or 21, depending on
whether that row of the PHI column is greater than
32 |
sqrt( (X-X0)^2 + (Y-Y0)^2 ) |
Distance of the (X,Y) column coordinates from the (X0,Y0)
keyword values |
See the CFITSIO
web pages and documentation for more details.
- objName insert row index numRows
- Insert number of numRows rows after the index-th row.
- objName add row numRows
- Add numRows rows at the end of the CHDU table.
- objName put image firstElem numElem listOfData
- Write a block of data to the CHDU image. The first pixel of an
image has firstElem = 1, not zero.
- objName put table colName firstElem
rowSpan listOfData
- Write a list of data to the firstElem-th element of column
colName in the CHDU table. (For scalar columns firstElem
is 1.)
rowSpan is a single row range of form start-end with ``-''
indicating all rows.
Deleting data from a FITS file
- objName delete keyword keyList
- Delete listed keywords, where the keyList can be the mix of
keyword names and index numbers. Keywords are deleted individually
in sequence, causing keyword positions to change after each deletion,
so be careful when deleting multiple keywords by index.
- objName delete cols colList
- Delete listed columns in a table extension.
- objName delete rows firstRow numRows
- Delete a block of rows.
- objName delete rows -expr expression
- Delete rows using expression which must evaluate to a boolean
value. Rows for which expression evaluates to TRUE get deleted.
- objName delete chdu
- Delete the current HDU. The HDU immediately following the one deleted
(or the preceeding one if the current HDU is the last one of the file) will
become the new current HDU. Returns extension type of new HDU:
0-image, 1-ASCII table, 2-Binary Table.
Analyzing FITS data
- objName sort ?-merge? colList ?ascendFlags?
- Sort table rows using columns in colList. When -merge
is present, if multiple rows have identical sort keys all but one of the
rows will be deleted. If present, ascendFlags contains a list of
1s and 0s indicating whether that column will be sorted in ascending
(1, the default) or descending (0) order.
- objName column -stat colName ?firstElem? ?rows?
- Return statistics on the firstElem-th
element (1 for scalar columns) of column colName in the order
min firstMinRow max firstMaxRow mean stdDev numData.
- objName column -minmax colName ?firstElem?
?rows?
- Returns minimum and maximum values of the firstElem-th
element (1 for scalar columns) of column colName.
- objName create 1dhisto filename {colList} {binParams}
?rowSpan?
- objName create 2dhisto filename {colList} {xbinParams}
{ybinParams} ?rowSpan?
- Create a 1D or 2D histogram from data in columns
colList. An optional 2nd or 3rd column, respectively, in
colList can be used as a weight. The limits and binning
size are given by the 3-element list binParams which holds
the min max binSize parameters. If min or
max is ``-'', the true min/max values of the column(s)
will be used. A single range of rows can be indicated via the
optional rowSpan parameter. The resulting histogram will
be written into the primary array of a new file filename.
WCS keywords will be written in the output file to record the
binning scales and offsets.
List/Pointer Manipulation
- lst2ptr dataList ?dataType? ?naxesList?
- Convert a TCL list into a memory-resident array. Returns
address dataType naxesList (see the load
commands above). The TCL list will be cast as double values
in the absence of the dataType paramter. The parameter
naxesList gives the vector dimensions of the list. It
can be a single number indicating the length of the list (the
default value when absent) or a list of numbers which must
multiply together to be the list length. If a list entry
contains the string "NULL", the maximum value of the given
dataType will be inserted.
- ptr2lst address dataType naxesList
- Convert a memory pointer into a TCL list. Returns dataList
dataType naxesList (see lst2ptr). If an array element
contains the maximum value for the given dataType, the
string "NULL" will be inserted into the list.
- vexpr ?-ptr? ?-use getDataCallback? expression
- Perform a vector calculation using lists or arrays where
expression is a C-style arithmetic expression.
(Ie, do not use $var notation unless you explicitly want
variable substitution before passing expression to
vexpr.) Without any of the options, variable
references within expression will be interpretted as
local TCL lists (or scalars) containing double data and the
result will itself be a simple TCL list. With the -ptr option
set, the answer instead will be returned as a pointer to a
memory-resident array. With either option set, two additional
parameters will be returned in a list of the form dataList
dataType naxesList or address dataType naxesList.
The -use option provides a "callback" routine which is
to be used for supplying the variable data. It gets called
when a variable is encountered in the expression and
will be passed a single argument containing the name of the
variable. The callback routine should return either
"dataList dataType naxesList" or "-ptr address
dataType naxesList". If the callback routine doesn't
recognize the variable, it should return an empty string. The
parser will then try to find the variable in the local
namespace, as when no callback is supplied. If this fallback behavior
is not desired, raise an error instead. The callback function
can itself check for local variables using the TCL command
upvar 1 $varName localName. The following code
snippet provides a sample callback function which assumes the
expression contains variables storing the results of a load
or lst2ptr command (ie, address dataType naxesList):
proc getPtrData { varName } {
upvar $varName myName
if [info exists myName] {
return [eval list -ptr $myName]
} else {
error "Variable doesn't exist"
}
}
then the following code would calculate the average value of a table
column:
set myCol [objName load column X]
vexpr -use getPtrData "sum( (double)myCol ) / nelem( myCol )"
By default, a vector with a single-valued naxesList is
interpretted as a single vector containing n elements such
that the above "sum" function sums all of the elements. In the
context of loading a column, though, one actually has an
array of n rows with each row containing 1 (or perhaps more)
elements. To handle this case, one must use the callback
function to provide an naxisList of "1 n" (or "2 m",
etc) instead of merely "n". In that case each row will be
calculated separately and have its own result. When
naxesList has multiple elements, the final element
specifies the number of rows.
See the CFITSIO web
pages and documentation for details on expression
syntax. One can also read the fv online help file on expression
syntax, noting that vexpr can only handle numerical
input and output.
- fits free addressList
- Free the memory occupied by the data generated by one of the
load, lst2ptr, and vexpr commands.
Other commands
- objName copy filename
- Copy the CHDU to a new FITS file. If the CHDU is not an image array,
then an empty primary array is inserted.
- objName sascii table filename fileMode firstRow
numRows colList widthList
- Write the listed columns into an ASCII file. In the output file, each
column has the width listed in widthList. fileMode indicates
how to write the data to filename: 0-create new file and write
column headings; 1-append to old file and write column headings;
2-append to old file, but don't write any headings.
- objName sascii image filename fileMode firstRow
numRows firstCol numCols cellSize ?slice?
- Write a block of the image data to an ASCII file. In the output file,
the data has width of cellSize. Use frame slice for 3D
image data.
- objName flush ?clear?
- Flush any dirty buffers to the disk. With clear, free the
buffers, too.
- objName append filename
- Append current HDU to another FITS file.
- objName checksum update|verify
- Update or Verify the checksum keywords in the current HDU. If
none exist, update will create them. For verify,
the return values are 1 for valid values, 0 if either keyword
(CHECKSUM or DATASUM) is missing, and -1 for invalid values.
- objName info expr expression
- Get information on the result of the supplied arithmetic
expression. Return value is cDataType numElements
{naxesList} where cDataType is the CFITSIO datatype (not
fitsTcl's): TDOUBLE, TSTRING, etc; numElements is the number
of elements
in the result; and naxesList contains the vector dimensions
of the result ({1} for scalar result). An expression which
evaluates to a constant (no dependency on the contents of the
table) will be indicated by a negative number of elements.
- fits option ?option? ?value?
- Modify fitsTcl's default behavior. With no options, this returns
a list of all the current fitsTcl options and their values. With
the option parameter, return the value of that parameter. When
both option and value are supplied, set the parameter
to the given value. Current options: wcsSwap (0).
- fits version
- Get version numbers for fitsTcl and underlying cfitsio.
Lawrence
Brown and Peter
Wilson
Last modified: Thu Jul 22 09:42:41 EDT 1999
|