A python library to read from and write to FITS files.
[](https://travis-ci.com/esheldon/fitsio)
[](https://github.com/esheldon/fitsio/actions?query=workflow%3Atests)
## Description
This is a python extension written in c and python. Data are read into
numerical python arrays.
A version of cfitsio is bundled with this package, there is no need to install
your own, nor will this conflict with a version you have installed.
## Some Features
- Read from and write to image, binary, and ascii table extensions.
- Read arbitrary subsets of table columns and rows without loading all the data
to memory.
- Read image subsets without reading the whole image. Write subsets to existing images.
- Write and read variable length table columns.
- Read images and tables using slice notation similar to numpy arrays. This is like a more
powerful memmap, since it is column-aware for tables.
- Append rows to an existing table. Delete row sets and row ranges. Resize tables,
or insert rows.
- Query the columns and rows in a table.
- Read and write header keywords.
- Read and write images in tile-compressed format (RICE,GZIP,PLIO,HCOMPRESS).
- Read/write gzip files directly. Read unix compress (.Z,.zip) and bzip2 (.bz2) files.
- TDIM information is used to return array columns in the correct shape.
- Write and read string table columns, including array columns of arbitrary
shape.
- Read and write complex, bool (logical), unsigned integer, signed bytes types.
- Write checksums into the header and verify them.
- Insert new columns into tables in-place.
- Iterate over rows in a table. Data are buffered for efficiency.
- python 3 support, including python 3 strings
## Examples
```python
import fitsio
from fitsio import FITS,FITSHDR
# Often you just want to quickly read or write data without bothering to
# create a FITS object. In that case, you can use the read and write
# convienience functions.
# read all data from the first hdu that has data
filename='data.fits'
data = fitsio.read(filename)
# read a subset of rows and columns from a table
data = fitsio.read(filename, rows=[35,1001], columns=['x','y'], ext=2)
# read the header
h = fitsio.read_header(filename)
# read both data and header
data,h = fitsio.read(filename, header=True)
# open the file and write a new binary table extension with the data
# array, which is a numpy array with fields, or "recarray".
data = np.zeros(10, dtype=[('id','i8'),('ra','f8'),('dec','f8')])
fitsio.write(filename, data)
# Write an image to the same file. By default a new extension is
# added to the file. use clobber=True to overwrite an existing file
# instead. To append rows to an existing table, see below.
fitsio.write(filename, image)
#
# the FITS class gives the you the ability to explore the data, and gives
# more control
#
# open a FITS file for reading and explore
fits=fitsio.FITS('data.fits')
# see what is in here; the FITS object prints itself
print(fits)
file: data.fits
mode: READONLY
extnum hdutype hduname
0 IMAGE_HDU
1 BINARY_TBL mytable
# at the python or ipython prompt the fits object will
# print itself
>>> fits
file: data.fits
... etc
# explore the extensions, either by extension number or
# extension name if available
>>> fits[0]
file: data.fits
extension: 0
type: IMAGE_HDU
image info:
data type: f8
dims: [4096,2048]
# by name; can also use fits[1]
>>> fits['mytable']
file: data.fits
extension: 1
type: BINARY_TBL
extname: mytable
rows: 4328342
column info:
i1scalar u1
f f4
fvec f4 array[2]
darr f8 array[3,2]
dvarr f8 varray[10]
s S5
svec S6 array[3]
svar S0 vstring[8]
sarr S2 array[4,3]
# See bottom for how to get more information for an extension
# [-1] to refers the last HDU
>>> fits[-1]
...
# if there are multiple HDUs with the same name, and an EXTVER
# is set, you can use it. Here extver=2
# fits['mytable',2]
# read the image from extension zero
img = fits[0].read()
img = fits[0][:,:]
# read a subset of the image without reading the whole image
img = fits[0][25:35, 45:55]
# read all rows and columns from a binary table extension
data = fits[1].read()
data = fits['mytable'].read()
data = fits[1][:]
# read a subset of rows and columns. By default uses a case-insensitive
# match. The result retains the names with original case. If columns is a
# sequence, a numpy array with fields, or recarray is returned
data = fits[1].read(rows=[1,5], columns=['index','x','y'])
# Similar but using slice notation
# row subsets
data = fits[1][10:20]
data = fits[1][10:20:2]
data = fits[1][[1,5,18]]
# Using EXTNAME and EXTVER values
data = fits['SCI',2][10:20]
# Slicing with reverse (flipped) striding
data = fits[1][40:25]
data = fits[1][40:25:-5]
# all rows of column 'x'
data = fits[1]['x'][:]
# Read a few columns at once. This is more efficient than separate read for
# each column
data = fits[1]['x','y'][:]
# General column and row subsets.
columns=['index','x','y']
rows = [1, 5]
data = fits[1][columns][rows]
# data are returned in the order requested by the user
# and duplicates are preserved
rows = [2, 2, 5]
data = fits[1][columns][rows]
# iterate over rows in a table hdu
# faster if we buffer some rows, let's buffer 1000 at a time
fits=fitsio.FITS(filename,iter_row_buffer=1000)
for row in fits[1]:
print(row)
# iterate over HDUs in a FITS object
for hdu in fits:
data=hdu.read()
# Note dvarr shows type varray[10] and svar shows type vstring[8]. These
# are variable length columns and the number specified is the maximum size.
# By default they are read into fixed-length fields in the output array.
# You can over-ride this by constructing the FITS object with the vstorage
# keyword or specifying vstorage when reading. Sending vstorage='object'
# will store the data in variable size object fields to save memory; the
# default is vstorage='fixed'. Object fields can also be written out to a
# new FITS file as variable length to save disk space.
fits = fitsio.FITS(filename,vstorage='object')
# OR
data = fits[1].read(vstorage='object')
print(data['dvarr'].dtype)
dtype('object')
# you can grab a FITS HDU object to simplify notation
hdu1 = fits[1]
data = hdu1['x','y'][35:50]
# get rows that satisfy the input expression. See "Row Filtering
# Specification" in the cfitsio manual (note no temporary table is
# created in this case, contrary to the cfitsio docs)
w=fits[1].where("x > 0.25 && y < 35.0")
data = fits[1][w]
# read the header
h = fits[0].read_header()
print(h['BITPIX'])
-64
fits.close()
# now write some data
fits = FITS('test.fits','rw')
# create a rec array. Note vstr
# is a variable length string
nrows=35
data = np.zeros(nrows, dtype=[('index','i4'),('vstr','O'),('x','f8'),
('arr','f4',(3,4))])
data['index'] = np.arange(nrows,dtype='i4')
data['x'] = np.random.random(nrows)
data['vstr'] = [str(i) for i in xrange(nrows)]
data['arr'] = np.arange(nrows*3*4,dtype='f4').reshape(nrows,3,4)
# create a new table extension and write the data
fits.write(data)
# can also be a list of ordinary arrays if you send the names
array_list=[xarray,yarray,namearray]
names=['x','y','name']
fits.write(array_list, names=names)
# similarly a dict of arrays
fits.write(dict_of_arrays)
fits.write(dict_of_arrays, names=names) # control name order
# append more rows to the table. The fields in data2 should match columns
# in the table. missing columns will be filled with zeros
fits[-1].append(data2)
# insert a new column into a table
fits[-1].insert_column('newcol', data)
# insert with a specific colnum
fits[-1].insert_column('newcol', data, colnum=2)
# overwrite rows
fits[-1].write(data)
# overwrite starting at a particular row. The table will grow if needed
fits[-1].write(data, firstrow=350)
# create an image
img=np.arange(2*3,dtype='i4').reshape(2,3)
# write an image in a new HDU (if this is a new file, the primary HDU)
fits.write(img)
# write an image with rice compression
fits.write(img, compress='rice')
# control the compression
fimg=np.random.normal(size=2*3).reshape(2, 3)
fits.write(img, compress='rice', qlevel=16, qmethod='SUBTRACTIVE_DITHER_2')
# lossless gzip compression for integers or floating point
fits.write(img, compress='gzip', qlevel=None)
fits.write(fimg, compress='gzip', qlevel=None)
# overwrite the image
fits[ext].write(img2)
# write into an existing image, starting at the location [300,400]
# the image will be expanded if needed
fits[ext].write(img3, start=[300,400])
# change the shape of the image on disk
fits[ext].reshape([250,100])
# add checksums for the data
fits[-1].write_checksum()
# can later verify data integridy
fits[-1].verify_checksum()
# you can also write a header at the same time. The header can be
# - a simple dict (no comments)
# - a list of dicts with 'name','value','comment' fields
# - a FITSHDR object
hdict = {'somekey': 35, 'location': 'kitt peak'}
fits.write(data, header=hdict)
hlist = [{'name':'observer', 'value':'ES', 'comment':'who'},
{'name':'location','value':'CTIO'},
{'name':'photometric','value':True}]
fits.write(data, header=hlist)
hdr=FITSHDR(hlist)
fits.write(data, header=hdr)
# you can add individual keys to an existing HDU
fits[1].write_key(name, value, comment="my comment")
# Write multiple header keys to an existing HDU. Here records
# is the same as sent with header= above
fits[1].write_keys(records)
# write special COMMENT fields
fits[1].write_comment("observer JS")
fits[1].write_comment("we had good weather")
# write special history fields
fits[1].write_history("processed with software X")
fits[1].write_history("re-processed with software Y")
fits.close()
# using a context, the file is closed automatically after leaving the block
with FITS('path/to/file') as fits:
data = fits[ext].read()
# you can check if a header exists using "in":
if 'blah' in fits:
data=fits['blah'].read()
if 2 in f:
data=fits[2].read()
# methods to get more information about extension. For extension 1:
f[1].get_info() # lots of info about the extension
f[1].has_data() # returns True if data is present in extension
f[1].get_extname()
f[1].get_extver()
f[1].get_extnum() # return zero-offset extension number
f[1].get_exttype() # 'BINARY_TBL' or 'ASCII_TBL' or 'IMAGE_HDU'
f[1].get_offsets() # byte offsets (header_start, data_start, data_end)
f[1].is_compressed() # for images. True if tile-compressed
f[1].get_colnames() # for tables
f[1].get_colname(colnum) # for tables find the name from column number
f[1].get_nrows() # for tables
f[1].get_rec_dtype() # for tables
f[1].get_rec_column_descr() # for tables
f[1].get_vstorage() # for tables, storage mechanism for variable
# length columns
# public attributes you can feel free to change as needed
f[1].lower # If True, lower case colnames on output
f[1].upper # If True, upper case colnames on output
f[1].case_sensitive # if True, names are matched case sensitive
```
## Installation
The easiest way is using pip or conda. To get the latest release
pip install fitsio
# update fitsio (and everything else)
pip install fitsio --upgrade
# if pip refuses to update to a newer version
pip install fitsio --upgrade --ignore-installed
# if you only want to upgrade fitsio
pip install fitsio --no-deps --upgrade --ignore-installed
# for conda, use conda-forge
conda install -c conda-forge fitsio
You can also get the latest source tarball release from
https://pypi.python.org/pypi/fitsio
or the bleeding edge source from github or use git. To check out
the code for the first time
git clone https://github.com/esheldon/fitsio.git
Or at a later time to update to the latest
cd fitsio
git update
Use tar xvfz to untar the file, enter the fitsio directory and type
python setup.py install
optionally with a prefix
python setup.py install --prefix=/some/path
## Requirements
- python 2 or python 3
- a C compiler and build tools like `make`, `patch`, etc.
- numpy (See the note below. Generally, numpy 1.11 or later is better.)
### Do not use numpy 1.10.0 or 1.10.1
There is a serious performance regression in numpy 1.10 that results
in fitsio running tens to hundreds of times slower. A fix may be
forthcoming in a later release. Please comment here if this
has already impacted your work https://github.com/numpy/numpy/issues/6467
## Tests
The unit tests should all pass for full support.
```bash
pytest fitsio
```
Some tests may fail if certain libraries are not available, such
as bzip2. This failure only implies that bzipped files cannot
be read, without affecting other functionality.
## Notes on Usage and Features
### cfitsio bundling
We bundle cfitsio partly because many deployed versions of cfitsio in the
wild do not have support for interesting features like tiled image compression.
Bundling a version that meets our needs is a safe alternative.
### array ordering
Since numpy uses C order, FITS uses fortran order, we have to write the TDIM
and image dimensions in reverse order, but write the data as is. Then we need
to also reverse the dims as read from the header when creating the numpy dtype,
but read as is.
### `distutils` vs `setuptools`
As of version `1.0.0`, `fitsio` has been transitioned to `setuptools` for packaging
and installation. There are many reasons to do this (and to not do this). However,
at a practical level, what this means for you is that you may have trouble uninstalling
older versions with `pip` via `pip uninstall fitsio`. If you do, the best thing to do is
to manually remove the files manually. See this [stackoverflow question](https://stackoverflow.com/questions/402359/how-do-you-uninstall-a-python-package-that-was-installed-using-distutils)
for example.
### python 3 strings
As of version `1.0.0`, fitsio now supports Python 3 strings natively. This support
means that for Python 3, native strings are read from and written correctly to
FITS files. All byte string columns are treated as ASCII-encoded unicode strings
as well. For FITS files written with a previous version of fitsio, the data
in Python 3 will now come back as a string and not a byte string. Note that this
support is not the same as full unicode support. Internally, fitsio only supports
the ASCII character set.
## TODO
- HDU groups: does anyone use these? If so open an issue!
Raw data
{
"_id": null,
"home_page": "https://github.com/esheldon/fitsio",
"name": "fitsio",
"maintainer": null,
"docs_url": null,
"requires_python": null,
"maintainer_email": null,
"keywords": null,
"author": "Erin Scott Sheldon",
"author_email": "erin.sheldon@gmail.com",
"download_url": "https://files.pythonhosted.org/packages/ae/f0/36c78d904385548db6c3eb050241cc8773f7ba4c0e168ace2a347fbd986a/fitsio-1.2.3.tar.gz",
"platform": null,
"description": "A python library to read from and write to FITS files.\n\n[](https://travis-ci.com/esheldon/fitsio)\n[](https://github.com/esheldon/fitsio/actions?query=workflow%3Atests)\n\n## Description\n\nThis is a python extension written in c and python. Data are read into\nnumerical python arrays.\n\nA version of cfitsio is bundled with this package, there is no need to install\nyour own, nor will this conflict with a version you have installed.\n\n\n## Some Features\n\n- Read from and write to image, binary, and ascii table extensions.\n- Read arbitrary subsets of table columns and rows without loading all the data\n to memory.\n- Read image subsets without reading the whole image. Write subsets to existing images.\n- Write and read variable length table columns.\n- Read images and tables using slice notation similar to numpy arrays. This is like a more\n powerful memmap, since it is column-aware for tables.\n- Append rows to an existing table. Delete row sets and row ranges. Resize tables,\n or insert rows.\n- Query the columns and rows in a table.\n- Read and write header keywords.\n- Read and write images in tile-compressed format (RICE,GZIP,PLIO,HCOMPRESS).\n- Read/write gzip files directly. Read unix compress (.Z,.zip) and bzip2 (.bz2) files.\n- TDIM information is used to return array columns in the correct shape.\n- Write and read string table columns, including array columns of arbitrary\n shape.\n- Read and write complex, bool (logical), unsigned integer, signed bytes types.\n- Write checksums into the header and verify them.\n- Insert new columns into tables in-place.\n- Iterate over rows in a table. Data are buffered for efficiency.\n- python 3 support, including python 3 strings\n\n\n## Examples\n\n```python\nimport fitsio\nfrom fitsio import FITS,FITSHDR\n\n# Often you just want to quickly read or write data without bothering to\n# create a FITS object. In that case, you can use the read and write\n# convienience functions.\n\n# read all data from the first hdu that has data\nfilename='data.fits'\ndata = fitsio.read(filename)\n\n# read a subset of rows and columns from a table\ndata = fitsio.read(filename, rows=[35,1001], columns=['x','y'], ext=2)\n\n# read the header\nh = fitsio.read_header(filename)\n# read both data and header\ndata,h = fitsio.read(filename, header=True)\n\n# open the file and write a new binary table extension with the data\n# array, which is a numpy array with fields, or \"recarray\".\n\ndata = np.zeros(10, dtype=[('id','i8'),('ra','f8'),('dec','f8')])\nfitsio.write(filename, data)\n\n# Write an image to the same file. By default a new extension is\n# added to the file. use clobber=True to overwrite an existing file\n# instead. To append rows to an existing table, see below.\n\nfitsio.write(filename, image)\n\n#\n# the FITS class gives the you the ability to explore the data, and gives\n# more control\n#\n\n# open a FITS file for reading and explore\nfits=fitsio.FITS('data.fits')\n\n# see what is in here; the FITS object prints itself\nprint(fits)\n\nfile: data.fits\nmode: READONLY\nextnum hdutype hduname\n0 IMAGE_HDU\n1 BINARY_TBL mytable\n\n# at the python or ipython prompt the fits object will\n# print itself\n>>> fits\nfile: data.fits\n... etc\n\n# explore the extensions, either by extension number or\n# extension name if available\n>>> fits[0]\n\nfile: data.fits\nextension: 0\ntype: IMAGE_HDU\nimage info:\n data type: f8\n dims: [4096,2048]\n\n# by name; can also use fits[1]\n>>> fits['mytable']\n\nfile: data.fits\nextension: 1\ntype: BINARY_TBL\nextname: mytable\nrows: 4328342\ncolumn info:\n i1scalar u1\n f f4\n fvec f4 array[2]\n darr f8 array[3,2]\n dvarr f8 varray[10]\n s S5\n svec S6 array[3]\n svar S0 vstring[8]\n sarr S2 array[4,3]\n\n# See bottom for how to get more information for an extension\n\n# [-1] to refers the last HDU\n>>> fits[-1]\n...\n\n# if there are multiple HDUs with the same name, and an EXTVER\n# is set, you can use it. Here extver=2\n# fits['mytable',2]\n\n\n# read the image from extension zero\nimg = fits[0].read()\nimg = fits[0][:,:]\n\n# read a subset of the image without reading the whole image\nimg = fits[0][25:35, 45:55]\n\n\n# read all rows and columns from a binary table extension\ndata = fits[1].read()\ndata = fits['mytable'].read()\ndata = fits[1][:]\n\n# read a subset of rows and columns. By default uses a case-insensitive\n# match. The result retains the names with original case. If columns is a\n# sequence, a numpy array with fields, or recarray is returned\ndata = fits[1].read(rows=[1,5], columns=['index','x','y'])\n\n# Similar but using slice notation\n# row subsets\ndata = fits[1][10:20]\ndata = fits[1][10:20:2]\ndata = fits[1][[1,5,18]]\n\n# Using EXTNAME and EXTVER values\ndata = fits['SCI',2][10:20]\n\n# Slicing with reverse (flipped) striding\ndata = fits[1][40:25]\ndata = fits[1][40:25:-5]\n\n# all rows of column 'x'\ndata = fits[1]['x'][:]\n\n# Read a few columns at once. This is more efficient than separate read for\n# each column\ndata = fits[1]['x','y'][:]\n\n# General column and row subsets.\ncolumns=['index','x','y']\nrows = [1, 5]\ndata = fits[1][columns][rows]\n\n# data are returned in the order requested by the user\n# and duplicates are preserved\nrows = [2, 2, 5]\ndata = fits[1][columns][rows]\n\n# iterate over rows in a table hdu\n# faster if we buffer some rows, let's buffer 1000 at a time\nfits=fitsio.FITS(filename,iter_row_buffer=1000)\nfor row in fits[1]:\n print(row)\n\n# iterate over HDUs in a FITS object\nfor hdu in fits:\n data=hdu.read()\n\n# Note dvarr shows type varray[10] and svar shows type vstring[8]. These\n# are variable length columns and the number specified is the maximum size.\n# By default they are read into fixed-length fields in the output array.\n# You can over-ride this by constructing the FITS object with the vstorage\n# keyword or specifying vstorage when reading. Sending vstorage='object'\n# will store the data in variable size object fields to save memory; the\n# default is vstorage='fixed'. Object fields can also be written out to a\n# new FITS file as variable length to save disk space.\n\nfits = fitsio.FITS(filename,vstorage='object')\n# OR\ndata = fits[1].read(vstorage='object')\nprint(data['dvarr'].dtype)\n dtype('object')\n\n\n# you can grab a FITS HDU object to simplify notation\nhdu1 = fits[1]\ndata = hdu1['x','y'][35:50]\n\n# get rows that satisfy the input expression. See \"Row Filtering\n# Specification\" in the cfitsio manual (note no temporary table is\n# created in this case, contrary to the cfitsio docs)\nw=fits[1].where(\"x > 0.25 && y < 35.0\")\ndata = fits[1][w]\n\n# read the header\nh = fits[0].read_header()\nprint(h['BITPIX'])\n -64\n\nfits.close()\n\n\n# now write some data\nfits = FITS('test.fits','rw')\n\n\n# create a rec array. Note vstr\n# is a variable length string\nnrows=35\ndata = np.zeros(nrows, dtype=[('index','i4'),('vstr','O'),('x','f8'),\n ('arr','f4',(3,4))])\ndata['index'] = np.arange(nrows,dtype='i4')\ndata['x'] = np.random.random(nrows)\ndata['vstr'] = [str(i) for i in xrange(nrows)]\ndata['arr'] = np.arange(nrows*3*4,dtype='f4').reshape(nrows,3,4)\n\n# create a new table extension and write the data\nfits.write(data)\n\n# can also be a list of ordinary arrays if you send the names\narray_list=[xarray,yarray,namearray]\nnames=['x','y','name']\nfits.write(array_list, names=names)\n\n# similarly a dict of arrays\nfits.write(dict_of_arrays)\nfits.write(dict_of_arrays, names=names) # control name order\n\n# append more rows to the table. The fields in data2 should match columns\n# in the table. missing columns will be filled with zeros\nfits[-1].append(data2)\n\n# insert a new column into a table\nfits[-1].insert_column('newcol', data)\n\n# insert with a specific colnum\nfits[-1].insert_column('newcol', data, colnum=2)\n\n# overwrite rows\nfits[-1].write(data)\n\n# overwrite starting at a particular row. The table will grow if needed\nfits[-1].write(data, firstrow=350)\n\n\n# create an image\nimg=np.arange(2*3,dtype='i4').reshape(2,3)\n\n# write an image in a new HDU (if this is a new file, the primary HDU)\nfits.write(img)\n\n# write an image with rice compression\nfits.write(img, compress='rice')\n\n# control the compression\nfimg=np.random.normal(size=2*3).reshape(2, 3)\nfits.write(img, compress='rice', qlevel=16, qmethod='SUBTRACTIVE_DITHER_2')\n\n# lossless gzip compression for integers or floating point\nfits.write(img, compress='gzip', qlevel=None)\nfits.write(fimg, compress='gzip', qlevel=None)\n\n# overwrite the image\nfits[ext].write(img2)\n\n# write into an existing image, starting at the location [300,400]\n# the image will be expanded if needed\nfits[ext].write(img3, start=[300,400])\n\n# change the shape of the image on disk\nfits[ext].reshape([250,100])\n\n# add checksums for the data\nfits[-1].write_checksum()\n\n# can later verify data integridy\nfits[-1].verify_checksum()\n\n# you can also write a header at the same time. The header can be\n# - a simple dict (no comments)\n# - a list of dicts with 'name','value','comment' fields\n# - a FITSHDR object\n\nhdict = {'somekey': 35, 'location': 'kitt peak'}\nfits.write(data, header=hdict)\nhlist = [{'name':'observer', 'value':'ES', 'comment':'who'},\n {'name':'location','value':'CTIO'},\n {'name':'photometric','value':True}]\nfits.write(data, header=hlist)\nhdr=FITSHDR(hlist)\nfits.write(data, header=hdr)\n\n# you can add individual keys to an existing HDU\nfits[1].write_key(name, value, comment=\"my comment\")\n\n# Write multiple header keys to an existing HDU. Here records\n# is the same as sent with header= above\nfits[1].write_keys(records)\n\n# write special COMMENT fields\nfits[1].write_comment(\"observer JS\")\nfits[1].write_comment(\"we had good weather\")\n\n# write special history fields\nfits[1].write_history(\"processed with software X\")\nfits[1].write_history(\"re-processed with software Y\")\n\nfits.close()\n\n# using a context, the file is closed automatically after leaving the block\nwith FITS('path/to/file') as fits:\n data = fits[ext].read()\n\n # you can check if a header exists using \"in\":\n if 'blah' in fits:\n data=fits['blah'].read()\n if 2 in f:\n data=fits[2].read()\n\n# methods to get more information about extension. For extension 1:\nf[1].get_info() # lots of info about the extension\nf[1].has_data() # returns True if data is present in extension\nf[1].get_extname()\nf[1].get_extver()\nf[1].get_extnum() # return zero-offset extension number\nf[1].get_exttype() # 'BINARY_TBL' or 'ASCII_TBL' or 'IMAGE_HDU'\nf[1].get_offsets() # byte offsets (header_start, data_start, data_end)\nf[1].is_compressed() # for images. True if tile-compressed\nf[1].get_colnames() # for tables\nf[1].get_colname(colnum) # for tables find the name from column number\nf[1].get_nrows() # for tables\nf[1].get_rec_dtype() # for tables\nf[1].get_rec_column_descr() # for tables\nf[1].get_vstorage() # for tables, storage mechanism for variable\n # length columns\n\n# public attributes you can feel free to change as needed\nf[1].lower # If True, lower case colnames on output\nf[1].upper # If True, upper case colnames on output\nf[1].case_sensitive # if True, names are matched case sensitive\n```\n\n\n## Installation\n\nThe easiest way is using pip or conda. To get the latest release\n\n pip install fitsio\n\n # update fitsio (and everything else)\n pip install fitsio --upgrade\n\n # if pip refuses to update to a newer version\n pip install fitsio --upgrade --ignore-installed\n\n # if you only want to upgrade fitsio\n pip install fitsio --no-deps --upgrade --ignore-installed\n\n # for conda, use conda-forge\n conda install -c conda-forge fitsio\n\nYou can also get the latest source tarball release from\n\n https://pypi.python.org/pypi/fitsio\n\nor the bleeding edge source from github or use git. To check out\nthe code for the first time\n\n git clone https://github.com/esheldon/fitsio.git\n\nOr at a later time to update to the latest\n\n cd fitsio\n git update\n\nUse tar xvfz to untar the file, enter the fitsio directory and type\n\n python setup.py install\n\noptionally with a prefix\n\n python setup.py install --prefix=/some/path\n\n## Requirements\n\n- python 2 or python 3\n- a C compiler and build tools like `make`, `patch`, etc.\n- numpy (See the note below. Generally, numpy 1.11 or later is better.)\n\n\n### Do not use numpy 1.10.0 or 1.10.1\n\nThere is a serious performance regression in numpy 1.10 that results\nin fitsio running tens to hundreds of times slower. A fix may be\nforthcoming in a later release. Please comment here if this\nhas already impacted your work https://github.com/numpy/numpy/issues/6467\n\n\n## Tests\n\nThe unit tests should all pass for full support.\n\n```bash\npytest fitsio\n```\n\nSome tests may fail if certain libraries are not available, such\nas bzip2. This failure only implies that bzipped files cannot\nbe read, without affecting other functionality.\n\n## Notes on Usage and Features\n\n### cfitsio bundling\n\nWe bundle cfitsio partly because many deployed versions of cfitsio in the\nwild do not have support for interesting features like tiled image compression.\nBundling a version that meets our needs is a safe alternative.\n\n### array ordering\n\nSince numpy uses C order, FITS uses fortran order, we have to write the TDIM\nand image dimensions in reverse order, but write the data as is. Then we need\nto also reverse the dims as read from the header when creating the numpy dtype,\nbut read as is.\n\n### `distutils` vs `setuptools`\n\nAs of version `1.0.0`, `fitsio` has been transitioned to `setuptools` for packaging\nand installation. There are many reasons to do this (and to not do this). However,\nat a practical level, what this means for you is that you may have trouble uninstalling\nolder versions with `pip` via `pip uninstall fitsio`. If you do, the best thing to do is\nto manually remove the files manually. See this [stackoverflow question](https://stackoverflow.com/questions/402359/how-do-you-uninstall-a-python-package-that-was-installed-using-distutils)\nfor example.\n\n### python 3 strings\n\nAs of version `1.0.0`, fitsio now supports Python 3 strings natively. This support\nmeans that for Python 3, native strings are read from and written correctly to\nFITS files. All byte string columns are treated as ASCII-encoded unicode strings\nas well. For FITS files written with a previous version of fitsio, the data\nin Python 3 will now come back as a string and not a byte string. Note that this\nsupport is not the same as full unicode support. Internally, fitsio only supports\nthe ASCII character set.\n\n## TODO\n\n- HDU groups: does anyone use these? If so open an issue!\n",
"bugtrack_url": null,
"license": "GPL",
"summary": "A full featured python library to read from and write to FITS files.",
"version": "1.2.3",
"project_urls": {
"Homepage": "https://github.com/esheldon/fitsio"
},
"split_keywords": [],
"urls": [
{
"comment_text": "",
"digests": {
"blake2b_256": "aef036c78d904385548db6c3eb050241cc8773f7ba4c0e168ace2a347fbd986a",
"md5": "7209da362575976eb5dab25e7a76ed75",
"sha256": "7f1c6f9c516cc9a0343a0e86eaff96af8939ae95ec005c816cc9cd202f800067"
},
"downloads": -1,
"filename": "fitsio-1.2.3.tar.gz",
"has_sig": false,
"md5_digest": "7209da362575976eb5dab25e7a76ed75",
"packagetype": "sdist",
"python_version": "source",
"requires_python": null,
"size": 4501227,
"upload_time": "2024-06-12T21:29:48",
"upload_time_iso_8601": "2024-06-12T21:29:48.781938Z",
"url": "https://files.pythonhosted.org/packages/ae/f0/36c78d904385548db6c3eb050241cc8773f7ba4c0e168ace2a347fbd986a/fitsio-1.2.3.tar.gz",
"yanked": false,
"yanked_reason": null
}
],
"upload_time": "2024-06-12 21:29:48",
"github": true,
"gitlab": false,
"bitbucket": false,
"codeberg": false,
"github_user": "esheldon",
"github_project": "fitsio",
"travis_ci": true,
"coveralls": false,
"github_actions": true,
"lcname": "fitsio"
}
JSON
