The tread function reads tables from an external location
into JyStilts. Its arguments are as follows:
tread(location, fmt='(auto)', random=False)and its return value is a table object, which can be interrogated directly, or used in other JyStilts commands. Usually, the location argument should be a string which gives the filename or URL at which a table can be found. You can alternatively use a readable python file (or file-like) object for the location, but be aware that this may be less efficient on memory. As with command-line STILTS, the
fmt argument
is one of the options in Section 5.1.1, but may be
left as the default if the format auto-detectable,
which currently means if the file is in
VOTable, FITS, CDF, ECSV, PDS4, Parquet, Feather or GBIN format.
The random argument can be used to ensure that the returned file
has random (i.e. not sequential-only) access;
for some table formats the default way of reading them in means that
their rows can only be accessed in sequence.
Depending on what processing you are doing, that may or may not be
satisfactory.
Examples of reading a table are:
>>> import stilts
>>> t1 = stilts.tread('cat.fits')
>>> t2 = stilts.tread(open('cat.fits', 'rb')) # less efficient
>>> t3 = stilts.tread('data.csv', fmt='csv', random=True)
The most straightforward way to write a table
(presumably the result of one or a sequence of JyStilts commands)
is using the write table method:
write(self, location=None, fmt='(auto)')The
location gives either a string which is a filename,
or a writable python file (or file-like) object.
Again, use of a filename is preferred as it may(?) be more efficient.
If no location is supplied, the table will be written to standard output
(useful for inspection, but a bad idea for binary formats or very large tables).
The fmt argument is one of the output formats in
Section 5.1.2, but may be left as the default if the
format can be guessed from the filename.
Examples of writing a table are:
>>> table.write('out.fits')
>>> table.write(open('out.fits', 'wb')) # less efficient?
>>> table.write('catalogue.dat', fmt='csv')
>>> table.write() # display to stdout
Often it's convenient to combine examining the table with filtering steps, for instance:
>>> table.every(100).write()would write only every hundredth row, and
>>> (table.cmd_sorthead(10, 'BMAG')
... .cmd_select('!NULL_VMAG')
... .cmd_keepcols('BMAG VMAG')
... .write())
would write only the BMAG and VMAG columns
for the ten rows in which VMAG is non-null with the lowest BMAG values.
You can also read and write multiple tables, if you use a table
format for which that is appropriate.
This generally means FITS (which can store tables in multiple extensions)
or VOTable (which can store multiple TABLE elements in one document).
This is done using the treads and twrites functions.
The functions look like this:
treads(location, fmt='(auto)', random=False) twrites(tables, location=None, fmt='(auto)')These are similar to the
tread and twrite functions,
except that treads returns a list of tables rather than
a single table, and twrites's tables argument is
an iterable over tables rather than a single table.
Here is an example of reading multiple tables from a multi-extension FITS
file, counting the rows in each, and then writing them out to a multi-TABLE
VOTable file:
import stilts
tables = stilts.treads('multi.fits')
print([t.getRowCount() for t in tables])
stilts.twrites(tables, 'multi.vot', fmt='votable')