Differs

Facilities for diffing two FITS files. Includes objects for diffing entire FITS files, individual HDUs, FITS headers, or just FITS data.

Used to implement the fitsdiff program.

FITSDiff

class astropy.io.fits.FITSDiff(a, b, ignore_hdus=[], ignore_keywords=[], ignore_comments=[], ignore_fields=[], numdiffs=10, rtol=0.0, atol=0.0, ignore_blanks=True, ignore_blank_cards=True, tolerance=None)[source] [edit on github]

Bases: astropy.io.fits.diff._BaseDiff

Diff two FITS files by filename, or two HDUList objects.

FITSDiff objects have the following diff attributes:

  • diff_hdu_count: If the FITS files being compared have different numbers of HDUs, this contains a 2-tuple of the number of HDUs in each file.
  • diff_hdus: If any HDUs with the same index are different, this contains a list of 2-tuples of the HDU index and the HDUDiff object representing the differences between the two HDUs.
Parameters:
a : str or HDUList

The filename of a FITS file on disk, or an HDUList object.

b : str or HDUList

The filename of a FITS file on disk, or an HDUList object to compare to the first file.

ignore_hdus : sequence, optional

HDU names to ignore when comparing two FITS files or HDU lists; the presence of these HDUs and their contents are ignored. Wildcard strings may also be included in the list.

ignore_keywords : sequence, optional

Header keywords to ignore when comparing two headers; the presence of these keywords and their values are ignored. Wildcard strings may also be included in the list.

ignore_comments : sequence, optional

A list of header keywords whose comments should be ignored in the comparison. May contain wildcard strings as with ignore_keywords.

ignore_fields : sequence, optional

The (case-insensitive) names of any table columns to ignore if any table data is to be compared.

numdiffs : int, optional

The number of pixel/table values to output when reporting HDU data differences. Though the count of differences is the same either way, this allows controlling the number of different values that are kept in memory or output. If a negative value is given, then numdiffs is treated as unlimited (default: 10).

rtol : float, optional

The relative difference to allow when comparing two float values either in header values, image arrays, or table columns (default: 0.0). Values which satisfy the expression

\[\left| a - b \right| > \text{atol} + \text{rtol} \cdot \left| b \right|\]

are considered to be different. The underlying function used for comparison is numpy.allclose.

Changed in version 2.0: rtol replaces the deprecated tolerance argument.

atol : float, optional

The allowed absolute difference. See also rtol parameter.

New in version 2.0.

ignore_blanks : bool, optional

Ignore extra whitespace at the end of string values either in headers or data. Extra leading whitespace is not ignored (default: True).

ignore_blank_cards : bool, optional

Ignore all cards that are blank, i.e. they only contain whitespace (default: True).

fromdiff(other, a, b) [edit on github]

Returns a new Diff object of a specific subclass from an existing diff object, passing on the values for any arguments they share in common (such as ignore_keywords).

For example:

>>> from astropy.io import fits
>>> hdul1, hdul2 = fits.HDUList(), fits.HDUList()
>>> headera, headerb = fits.Header(), fits.Header()
>>> fd = fits.FITSDiff(hdul1, hdul2, ignore_keywords=['*'])
>>> hd = fits.HeaderDiff.fromdiff(fd, headera, headerb)
>>> list(hd.ignore_keywords)
['*']
identical

True if all the .diff_* attributes on this diff instance are empty, implying that no differences were found.

Any subclass of _BaseDiff must have at least one .diff_* attribute, which contains a non-empty value if and only if some difference was found between the two objects being compared.

report(fileobj=None, indent=0, overwrite=False) [edit on github]

Generates a text report on the differences (if any) between two objects, and either returns it as a string or writes it to a file-like object.

Parameters:
fileobj : file-like object, string, or None (optional)

If None, this method returns the report as a string. Otherwise it returns None and writes the report to the given file-like object (which must have a .write() method at a minimum), or to a new file at the path specified.

indent : int

The number of 4 space tabs to indent the report.

overwrite : bool, optional

If True, overwrite the output file if it exists. Raises an OSError if False and the output file exists. Default is False.

Changed in version 1.3: overwrite replaces the deprecated clobber argument.

Returns:
report : str or None

HDUDiff

class astropy.io.fits.HDUDiff(a, b, ignore_keywords=[], ignore_comments=[], ignore_fields=[], numdiffs=10, rtol=0.0, atol=0.0, ignore_blanks=True, ignore_blank_cards=True, tolerance=None)[source] [edit on github]

Bases: astropy.io.fits.diff._BaseDiff

Diff two HDU objects, including their headers and their data (but only if both HDUs contain the same type of data (image, table, or unknown).

HDUDiff objects have the following diff attributes:

  • diff_extnames: If the two HDUs have different EXTNAME values, this contains a 2-tuple of the different extension names.
  • diff_extvers: If the two HDUS have different EXTVER values, this contains a 2-tuple of the different extension versions.
  • diff_extlevels: If the two HDUs have different EXTLEVEL values, this contains a 2-tuple of the different extension levels.
  • diff_extension_types: If the two HDUs have different XTENSION values, this contains a 2-tuple of the different extension types.
  • diff_headers: Contains a HeaderDiff object for the headers of the two HDUs. This will always contain an object–it may be determined whether the headers are different through diff_headers.identical.
  • diff_data: Contains either a ImageDataDiff, TableDataDiff, or RawDataDiff as appropriate for the data in the HDUs, and only if the two HDUs have non-empty data of the same type (RawDataDiff is used for HDUs containing non-empty data of an indeterminate type).
Parameters:
a : HDUList

An HDUList object.

b : str or HDUList

An HDUList object to compare to the first HDUList object.

ignore_keywords : sequence, optional

Header keywords to ignore when comparing two headers; the presence of these keywords and their values are ignored. Wildcard strings may also be included in the list.

ignore_comments : sequence, optional

A list of header keywords whose comments should be ignored in the comparison. May contain wildcard strings as with ignore_keywords.

ignore_fields : sequence, optional

The (case-insensitive) names of any table columns to ignore if any table data is to be compared.

numdiffs : int, optional

The number of pixel/table values to output when reporting HDU data differences. Though the count of differences is the same either way, this allows controlling the number of different values that are kept in memory or output. If a negative value is given, then numdiffs is treated as unlimited (default: 10).

rtol : float, optional

The relative difference to allow when comparing two float values either in header values, image arrays, or table columns (default: 0.0). Values which satisfy the expression

\[\left| a - b \right| > \text{atol} + \text{rtol} \cdot \left| b \right|\]

are considered to be different. The underlying function used for comparison is numpy.allclose.

Changed in version 2.0: rtol replaces the deprecated tolerance argument.

atol : float, optional

The allowed absolute difference. See also rtol parameter.

New in version 2.0.

ignore_blanks : bool, optional

Ignore extra whitespace at the end of string values either in headers or data. Extra leading whitespace is not ignored (default: True).

ignore_blank_cards : bool, optional

Ignore all cards that are blank, i.e. they only contain whitespace (default: True).

fromdiff(other, a, b) [edit on github]

Returns a new Diff object of a specific subclass from an existing diff object, passing on the values for any arguments they share in common (such as ignore_keywords).

For example:

>>> from astropy.io import fits
>>> hdul1, hdul2 = fits.HDUList(), fits.HDUList()
>>> headera, headerb = fits.Header(), fits.Header()
>>> fd = fits.FITSDiff(hdul1, hdul2, ignore_keywords=['*'])
>>> hd = fits.HeaderDiff.fromdiff(fd, headera, headerb)
>>> list(hd.ignore_keywords)
['*']
identical

True if all the .diff_* attributes on this diff instance are empty, implying that no differences were found.

Any subclass of _BaseDiff must have at least one .diff_* attribute, which contains a non-empty value if and only if some difference was found between the two objects being compared.

report(fileobj=None, indent=0, overwrite=False) [edit on github]

Generates a text report on the differences (if any) between two objects, and either returns it as a string or writes it to a file-like object.

Parameters:
fileobj : file-like object, string, or None (optional)

If None, this method returns the report as a string. Otherwise it returns None and writes the report to the given file-like object (which must have a .write() method at a minimum), or to a new file at the path specified.

indent : int

The number of 4 space tabs to indent the report.

overwrite : bool, optional

If True, overwrite the output file if it exists. Raises an OSError if False and the output file exists. Default is False.

Changed in version 1.3: overwrite replaces the deprecated clobber argument.

Returns:
report : str or None

HeaderDiff

class astropy.io.fits.HeaderDiff(a, b, ignore_keywords=[], ignore_comments=[], rtol=0.0, atol=0.0, ignore_blanks=True, ignore_blank_cards=True, tolerance=None)[source] [edit on github]

Bases: astropy.io.fits.diff._BaseDiff

Diff two Header objects.

HeaderDiff objects have the following diff attributes:

  • diff_keyword_count: If the two headers contain a different number of keywords, this contains a 2-tuple of the keyword count for each header.

  • diff_keywords: If either header contains one or more keywords that don’t appear at all in the other header, this contains a 2-tuple consisting of a list of the keywords only appearing in header a, and a list of the keywords only appearing in header b.

  • diff_duplicate_keywords: If a keyword appears in both headers at least once, but contains a different number of duplicates (for example, a different number of HISTORY cards in each header), an item is added to this dict with the keyword as the key, and a 2-tuple of the different counts of that keyword as the value. For example:

    {'HISTORY': (20, 19)}
    

    means that header a contains 20 HISTORY cards, while header b contains only 19 HISTORY cards.

  • diff_keyword_values: If any of the common keyword between the two headers have different values, they appear in this dict. It has a structure similar to diff_duplicate_keywords, with the keyword as the key, and a 2-tuple of the different values as the value. For example:

    {'NAXIS': (2, 3)}
    

    means that the NAXIS keyword has a value of 2 in header a, and a value of 3 in header b. This excludes any keywords matched by the ignore_keywords list.

  • diff_keyword_comments: Like diff_keyword_values, but contains differences between keyword comments.

HeaderDiff objects also have a common_keywords attribute that lists all keywords that appear in both headers.

Parameters:
a : HDUList

An HDUList object.

b : HDUList

An HDUList object to compare to the first HDUList object.

ignore_keywords : sequence, optional

Header keywords to ignore when comparing two headers; the presence of these keywords and their values are ignored. Wildcard strings may also be included in the list.

ignore_comments : sequence, optional

A list of header keywords whose comments should be ignored in the comparison. May contain wildcard strings as with ignore_keywords.

numdiffs : int, optional

The number of pixel/table values to output when reporting HDU data differences. Though the count of differences is the same either way, this allows controlling the number of different values that are kept in memory or output. If a negative value is given, then numdiffs is treated as unlimited (default: 10).

rtol : float, optional

The relative difference to allow when comparing two float values either in header values, image arrays, or table columns (default: 0.0). Values which satisfy the expression

\[\left| a - b \right| > \text{atol} + \text{rtol} \cdot \left| b \right|\]

are considered to be different. The underlying function used for comparison is numpy.allclose.

Changed in version 2.0: rtol replaces the deprecated tolerance argument.

atol : float, optional

The allowed absolute difference. See also rtol parameter.

New in version 2.0.

ignore_blanks : bool, optional

Ignore extra whitespace at the end of string values either in headers or data. Extra leading whitespace is not ignored (default: True).

ignore_blank_cards : bool, optional

Ignore all cards that are blank, i.e. they only contain whitespace (default: True).

fromdiff(other, a, b) [edit on github]

Returns a new Diff object of a specific subclass from an existing diff object, passing on the values for any arguments they share in common (such as ignore_keywords).

For example:

>>> from astropy.io import fits
>>> hdul1, hdul2 = fits.HDUList(), fits.HDUList()
>>> headera, headerb = fits.Header(), fits.Header()
>>> fd = fits.FITSDiff(hdul1, hdul2, ignore_keywords=['*'])
>>> hd = fits.HeaderDiff.fromdiff(fd, headera, headerb)
>>> list(hd.ignore_keywords)
['*']
identical

True if all the .diff_* attributes on this diff instance are empty, implying that no differences were found.

Any subclass of _BaseDiff must have at least one .diff_* attribute, which contains a non-empty value if and only if some difference was found between the two objects being compared.

report(fileobj=None, indent=0, overwrite=False) [edit on github]

Generates a text report on the differences (if any) between two objects, and either returns it as a string or writes it to a file-like object.

Parameters:
fileobj : file-like object, string, or None (optional)

If None, this method returns the report as a string. Otherwise it returns None and writes the report to the given file-like object (which must have a .write() method at a minimum), or to a new file at the path specified.

indent : int

The number of 4 space tabs to indent the report.

overwrite : bool, optional

If True, overwrite the output file if it exists. Raises an OSError if False and the output file exists. Default is False.

Changed in version 1.3: overwrite replaces the deprecated clobber argument.

Returns:
report : str or None

ImageDataDiff

class astropy.io.fits.ImageDataDiff(a, b, numdiffs=10, rtol=0.0, atol=0.0, tolerance=None)[source] [edit on github]

Bases: astropy.io.fits.diff._BaseDiff

Diff two image data arrays (really any array from a PRIMARY HDU or an IMAGE extension HDU, though the data unit is assumed to be “pixels”).

ImageDataDiff objects have the following diff attributes:

  • diff_dimensions: If the two arrays contain either a different number of dimensions or different sizes in any dimension, this contains a 2-tuple of the shapes of each array. Currently no further comparison is performed on images that don’t have the exact same dimensions.

  • diff_pixels: If the two images contain any different pixels, this contains a list of 2-tuples of the array index where the difference was found, and another 2-tuple containing the different values. For example, if the pixel at (0, 0) contains different values this would look like:

    [(0, 0), (1.1, 2.2)]
    

    where 1.1 and 2.2 are the values of that pixel in each array. This array only contains up to self.numdiffs differences, for storage efficiency.

  • diff_total: The total number of different pixels found between the arrays. Although diff_pixels does not necessarily contain all the different pixel values, this can be used to get a count of the total number of differences found.

  • diff_ratio: Contains the ratio of diff_total to the total number of pixels in the arrays.

Parameters:
a : HDUList

An HDUList object.

b : HDUList

An HDUList object to compare to the first HDUList object.

numdiffs : int, optional

The number of pixel/table values to output when reporting HDU data differences. Though the count of differences is the same either way, this allows controlling the number of different values that are kept in memory or output. If a negative value is given, then numdiffs is treated as unlimited (default: 10).

rtol : float, optional

The relative difference to allow when comparing two float values either in header values, image arrays, or table columns (default: 0.0). Values which satisfy the expression

\[\left| a - b \right| > \text{atol} + \text{rtol} \cdot \left| b \right|\]

are considered to be different. The underlying function used for comparison is numpy.allclose.

Changed in version 2.0: rtol replaces the deprecated tolerance argument.

atol : float, optional

The allowed absolute difference. See also rtol parameter.

New in version 2.0.

fromdiff(other, a, b) [edit on github]

Returns a new Diff object of a specific subclass from an existing diff object, passing on the values for any arguments they share in common (such as ignore_keywords).

For example:

>>> from astropy.io import fits
>>> hdul1, hdul2 = fits.HDUList(), fits.HDUList()
>>> headera, headerb = fits.Header(), fits.Header()
>>> fd = fits.FITSDiff(hdul1, hdul2, ignore_keywords=['*'])
>>> hd = fits.HeaderDiff.fromdiff(fd, headera, headerb)
>>> list(hd.ignore_keywords)
['*']
identical

True if all the .diff_* attributes on this diff instance are empty, implying that no differences were found.

Any subclass of _BaseDiff must have at least one .diff_* attribute, which contains a non-empty value if and only if some difference was found between the two objects being compared.

report(fileobj=None, indent=0, overwrite=False) [edit on github]

Generates a text report on the differences (if any) between two objects, and either returns it as a string or writes it to a file-like object.

Parameters:
fileobj : file-like object, string, or None (optional)

If None, this method returns the report as a string. Otherwise it returns None and writes the report to the given file-like object (which must have a .write() method at a minimum), or to a new file at the path specified.

indent : int

The number of 4 space tabs to indent the report.

overwrite : bool, optional

If True, overwrite the output file if it exists. Raises an OSError if False and the output file exists. Default is False.

Changed in version 1.3: overwrite replaces the deprecated clobber argument.

Returns:
report : str or None

RawDataDiff

class astropy.io.fits.RawDataDiff(a, b, numdiffs=10)[source] [edit on github]

Bases: astropy.io.fits.ImageDataDiff

RawDataDiff is just a special case of ImageDataDiff where the images are one-dimensional, and the data is treated as a 1-dimensional array of bytes instead of pixel values. This is used to compare the data of two non-standard extension HDUs that were not recognized as containing image or table data.

ImageDataDiff objects have the following diff attributes:

  • diff_dimensions: Same as the diff_dimensions attribute of ImageDataDiff objects. Though the “dimension” of each array is just an integer representing the number of bytes in the data.
  • diff_bytes: Like the diff_pixels attribute of ImageDataDiff objects, but renamed to reflect the minor semantic difference that these are raw bytes and not pixel values. Also the indices are integers instead of tuples.
  • diff_total and diff_ratio: Same as ImageDataDiff.
Parameters:
a : HDUList

An HDUList object.

b : HDUList

An HDUList object to compare to the first HDUList object.

numdiffs : int, optional

The number of pixel/table values to output when reporting HDU data differences. Though the count of differences is the same either way, this allows controlling the number of different values that are kept in memory or output. If a negative value is given, then numdiffs is treated as unlimited (default: 10).

fromdiff(other, a, b) [edit on github]

Returns a new Diff object of a specific subclass from an existing diff object, passing on the values for any arguments they share in common (such as ignore_keywords).

For example:

>>> from astropy.io import fits
>>> hdul1, hdul2 = fits.HDUList(), fits.HDUList()
>>> headera, headerb = fits.Header(), fits.Header()
>>> fd = fits.FITSDiff(hdul1, hdul2, ignore_keywords=['*'])
>>> hd = fits.HeaderDiff.fromdiff(fd, headera, headerb)
>>> list(hd.ignore_keywords)
['*']
identical

True if all the .diff_* attributes on this diff instance are empty, implying that no differences were found.

Any subclass of _BaseDiff must have at least one .diff_* attribute, which contains a non-empty value if and only if some difference was found between the two objects being compared.

report(fileobj=None, indent=0, overwrite=False) [edit on github]

Generates a text report on the differences (if any) between two objects, and either returns it as a string or writes it to a file-like object.

Parameters:
fileobj : file-like object, string, or None (optional)

If None, this method returns the report as a string. Otherwise it returns None and writes the report to the given file-like object (which must have a .write() method at a minimum), or to a new file at the path specified.

indent : int

The number of 4 space tabs to indent the report.

overwrite : bool, optional

If True, overwrite the output file if it exists. Raises an OSError if False and the output file exists. Default is False.

Changed in version 1.3: overwrite replaces the deprecated clobber argument.

Returns:
report : str or None

TableDataDiff

class astropy.io.fits.TableDataDiff(a, b, ignore_fields=[], numdiffs=10, rtol=0.0, atol=0.0, tolerance=None)[source] [edit on github]

Bases: astropy.io.fits.diff._BaseDiff

Diff two table data arrays. It doesn’t matter whether the data originally came from a binary or ASCII table–the data should be passed in as a recarray.

TableDataDiff objects have the following diff attributes:

  • diff_column_count: If the tables being compared have different numbers of columns, this contains a 2-tuple of the column count in each table. Even if the tables have different column counts, an attempt is still made to compare any columns they have in common.

  • diff_columns: If either table contains columns unique to that table, either in name or format, this contains a 2-tuple of lists. The first element is a list of columns (these are full Column objects) that appear only in table a. The second element is a list of tables that appear only in table b. This only lists columns with different column definitions, and has nothing to do with the data in those columns.

  • diff_column_names: This is like diff_columns, but lists only the names of columns unique to either table, rather than the full Column objects.

  • diff_column_attributes: Lists columns that are in both tables but have different secondary attributes, such as TUNIT or TDISP. The format is a list of 2-tuples: The first a tuple of the column name and the attribute, the second a tuple of the different values.

  • diff_values: TableDataDiff compares the data in each table on a column-by-column basis. If any different data is found, it is added to this list. The format of this list is similar to the diff_pixels attribute on ImageDataDiff objects, though the “index” consists of a (column_name, row) tuple. For example:

    [('TARGET', 0), ('NGC1001', 'NGC1002')]
    

    shows that the tables contain different values in the 0-th row of the ‘TARGET’ column.

  • diff_total and diff_ratio: Same as ImageDataDiff.

TableDataDiff objects also have a common_columns attribute that lists the Column objects for columns that are identical in both tables, and a common_column_names attribute which contains a set of the names of those columns.

Parameters:
a : HDUList

An HDUList object.

b : HDUList

An HDUList object to compare to the first HDUList object.

ignore_fields : sequence, optional

The (case-insensitive) names of any table columns to ignore if any table data is to be compared.

numdiffs : int, optional

The number of pixel/table values to output when reporting HDU data differences. Though the count of differences is the same either way, this allows controlling the number of different values that are kept in memory or output. If a negative value is given, then numdiffs is treated as unlimited (default: 10).

rtol : float, optional

The relative difference to allow when comparing two float values either in header values, image arrays, or table columns (default: 0.0). Values which satisfy the expression

\[\left| a - b \right| > \text{atol} + \text{rtol} \cdot \left| b \right|\]

are considered to be different. The underlying function used for comparison is numpy.allclose.

Changed in version 2.0: rtol replaces the deprecated tolerance argument.

atol : float, optional

The allowed absolute difference. See also rtol parameter.

New in version 2.0.

fromdiff(other, a, b) [edit on github]

Returns a new Diff object of a specific subclass from an existing diff object, passing on the values for any arguments they share in common (such as ignore_keywords).

For example:

>>> from astropy.io import fits
>>> hdul1, hdul2 = fits.HDUList(), fits.HDUList()
>>> headera, headerb = fits.Header(), fits.Header()
>>> fd = fits.FITSDiff(hdul1, hdul2, ignore_keywords=['*'])
>>> hd = fits.HeaderDiff.fromdiff(fd, headera, headerb)
>>> list(hd.ignore_keywords)
['*']
identical

True if all the .diff_* attributes on this diff instance are empty, implying that no differences were found.

Any subclass of _BaseDiff must have at least one .diff_* attribute, which contains a non-empty value if and only if some difference was found between the two objects being compared.

report(fileobj=None, indent=0, overwrite=False) [edit on github]

Generates a text report on the differences (if any) between two objects, and either returns it as a string or writes it to a file-like object.

Parameters:
fileobj : file-like object, string, or None (optional)

If None, this method returns the report as a string. Otherwise it returns None and writes the report to the given file-like object (which must have a .write() method at a minimum), or to a new file at the path specified.

indent : int

The number of 4 space tabs to indent the report.

overwrite : bool, optional

If True, overwrite the output file if it exists. Raises an OSError if False and the output file exists. Default is False.

Changed in version 1.3: overwrite replaces the deprecated clobber argument.

Returns:
report : str or None