1
==================================
2
Reading and Writing Config Files
3
==================================
4
----------------------------------------
5
ConfigObj 4 Introduction and Reference
6
----------------------------------------
8
:Authors: Michael Foord, Nicola Larosa
9
:Version: ConfigObj 4.7.2
11
:Homepage: `ConfigObj Homepage`_
12
:PyPI Entry: `ConfigObj on PyPI <http://pypi.python.org/pypi/configobj/>`_
13
:Development: `Google Code Homepage`_
14
:License: `BSD License`_
15
:Support: `Mailing List`_
17
.. _Mailing List: http://lists.sourceforge.net/lists/listinfo/configobj-develop
18
.. _Google Code Homepage: http://code.google.com/p/configobj/
21
:description: ConfigObj - a Python module for easy reading and writing of
23
:keywords: python, script, module, config, configuration, data, persistence,
24
developer, configparser
27
.. contents:: ConfigObj Manual
32
The best introduction to working with ConfigObj, including the powerful configuration validation system,
35
* `An Introduction to ConfigObj <http://www.voidspace.org.uk/python/articles/configobj.shtml>`_
41
**ConfigObj** is a simple but powerful config file reader and writer: an *ini
42
file round tripper*. Its main feature is that it is very easy to use, with a
43
straightforward programmer's interface and a simple syntax for config files.
44
It has lots of other features though :
46
* Nested sections (subsections), to any level
48
* Multiple line values
49
* String interpolation (substitution)
50
* Integrated with a powerful validation system
52
- including automatic type checking/conversion
54
- and allowing default values
56
* When writing out config files, ConfigObj preserves all comments and the order of members and sections
57
* Many useful methods and options for working with configuration files (like the 'reload' method)
58
* Full Unicode support
61
For support and bug reports please use the ConfigObj `Mailing List`_ or the issue tracker on the
62
`Google Code Homepage`_.
68
The current version is **4.7.2**, dated 27th February 2010. ConfigObj 4 is
69
stable and mature. We still expect to pick up a few bugs along the way though [#]_.
71
You can get ConfigObj in the following ways :
73
* configobj.py_ from Voidspace
75
ConfigObj has no external dependencies. This file is sufficient to access
76
all the functionality except Validation_.
78
* configobj.zip_ from Voidspace
80
This also contains validate.py_ and `this document`_.
82
* validate.py_ from Voidspace
88
ConfigObj has a source distribution `on PyPI <http://pypi.python.org/pypi/configobj/>`_. If you unzip
89
the archive you can install it with::
93
Alternatively, you can install with easy install or pip::
95
easy_install configobj
101
*configobj.zip* also contains `this document`_.
103
* You can view `this document`_ online at the `ConfigObj Homepage`_.
109
It is sometimes possible to get the latest *development version* of ConfigObj
110
from the Subversion Repository maintained on the `Google Code Homepage`_.
112
.. _configobj.py: http://www.voidspace.org.uk/downloads/configobj.py
113
.. _configobj.zip: http://www.voidspace.org.uk/downloads/configobj-4.7.2.zip
114
.. _validate.py: http://www.voidspace.org.uk/downloads/validate.py
116
.. _configobj homepage: http://www.voidspace.org.uk/python/configobj.html
120
ConfigObj in the Real World
121
===========================
123
**ConfigObj** is widely used. Projects using it include:
125
* `Bazaar <http://bazaar-ng.org>`_.
127
Bazaar is a Python distributed {acro;VCS;Version Control System}.
128
ConfigObj is used to read ``bazaar.conf`` and ``branches.conf``.
130
* `Chandler <http://chandler.osafoundation.org/>`_
132
A Python and `wxPython <http://www.wxpython.org>`_
133
Personal Information Manager, being developed by the
134
`OSAFoundation <http://www.osafoundation.org/>`_.
136
* `matplotlib <http://matplotlib.sourceforge.net/>`_
138
A 2D plotting library.
140
* `IPython <http://ipython.scipy.org/moin/>`_
142
IPython is an enhanced interactive Python shell. IPython uses ConfigObj in a module called 'TConfig' that combines it with enthought `Traits <http://code.enthought.com/traits/>`_: `tconfig <http://ipython.scipy.org/ipython/ipython/browser/ipython/branches/saw/sandbox/tconfig>`_.
144
* `Elisa - the Fluendo Mediacenter <http://elisa.fluendo.com/>`_
146
Elisa is an open source cross-platform media center solution designed to be simple for people not particularly familiar with computers.
152
The outstanding feature of using ConfigObj is simplicity. Most functions can be
153
performed with single line commands.
156
Reading a Config File
157
---------------------
159
The normal way to read a config file, is to give ConfigObj the filename :
161
.. code-block:: python
163
from configobj import ConfigObj
164
config = ConfigObj(filename)
166
You can also pass the config file in as a list of lines, or a ``StringIO``
167
instance, so it doesn't matter where your config data comes from.
169
You can then access members of your config file as a dictionary. Subsections
170
will also be dictionaries.
172
.. code-block:: python
174
from configobj import ConfigObj
175
config = ConfigObj(filename)
177
value1 = config['keyword1']
178
value2 = config['keyword2']
180
section1 = config['section1']
181
value3 = section1['keyword3']
182
value4 = section1['keyword4']
184
# you could also write
185
value3 = config['section1']['keyword3']
186
value4 = config['section1']['keyword4']
189
Writing a Config File
190
---------------------
192
Creating a new config file is just as easy as reading one. You can specify a
193
filename when you create the ConfigObj, or do it later [#]_.
195
If you *don't* set a filename, then the ``write`` method will return a list of
196
lines instead of writing to file. See the write_ method for more details.
198
Here we show creating an empty ConfigObj, setting a filename and some values,
199
and then writing to file :
202
.. code-block:: python
204
from configobj import ConfigObj
206
config.filename = filename
208
config['keyword1'] = value1
209
config['keyword2'] = value2
211
config['section1'] = {}
212
config['section1']['keyword3'] = value3
213
config['section1']['keyword4'] = value4
222
config['section2'] = section2
224
config['section3'] = {}
225
config['section3']['keyword 8'] = [value8, value9, value10]
226
config['section3']['keyword 9'] = [value11, value12, value13]
233
Keywords and section names can only be strings [#]_. Attempting to set
234
anything else will raise a ``ValueError``.
236
See `String Interpolation and List Values`_ for an important note on
237
using lists in combination with `String Interpolation`_.
243
The config files that ConfigObj will read and write are based on the 'INI'
244
format. This means it will read and write files created for ``ConfigParser``
247
Keywords and values are separated by an ``'='``, and section markers are
248
between square brackets. Keywords, values, and section names can be surrounded
249
by single or double quotes. Indentation is not significant, but can be
252
Subsections are indicated by repeating the square brackets in the section
253
marker. You nest levels by using more brackets.
255
You can have list values by separating items with a comma, and values spanning
256
multiple lines by using triple quotes (single or double).
258
For full details on all these see `the config file format`_. Here's an example
261
# This is the 'initial_comment'
262
# Which may be several lines
264
'keyword 2' = 'value 2'
267
# This comment goes with keyword 3
269
'keyword 4' = value4, value 5, 'value 6'
271
[[ sub-section ]] # an inline comment
272
# sub-section is inside "section 1"
273
'keyword 5' = 'value 7'
274
'keyword 6' = '''A multiline value,
275
that spans more than one line :-)
276
The line breaks are included in the value.'''
278
[[[ sub-sub-section ]]]
279
# sub-sub-section is *in* 'sub-section'
280
# which is in 'section 1'
281
'keyword 7' = 'value 8'
283
[section 2] # an inline comment
285
keyword9 = value10 # an inline comment
286
# The 'final_comment'
287
# Which also may be several lines
290
ConfigObj specifications
291
========================
293
.. code-block:: python
295
config = ConfigObj(infile=None, options=None, configspec=None, encoding=None,
296
interpolation=True, raise_errors=False, list_values=True,
297
create_empty=False, file_error=False, stringify=True,
298
indent_type=None, default_encoding=None, unrepr=False,
299
write_empty_values=False, _inspec=False)
301
Many of the keyword arguments are available as attributes after the config file has been
306
New in ConfigObj 4.7.0: Instantiating ConfigObj with
307
an ``options`` dictionary is now deprecated. To modify code that used to
308
do this simply unpack the dictionary in the constructor call:
310
.. code-block:: python
312
config = ConfigObj(filename, **options)
314
ConfigObj takes the following arguments (with the default values shown) :
318
You don't need to specify an infile. If you omit it, an empty ConfigObj will be
319
created. ``infile`` *can* be :
321
* Nothing. In which case the ``filename`` attribute of your ConfigObj will be
322
``None``. You can set a filename at any time.
324
* A filename. What happens if the file doesn't already exist is determined by
325
the options ``file_error`` and ``create_empty``. The filename will be
326
preserved as the ``filename`` attribute. This can be changed at any time.
328
* A list of lines. Any trailing newlines will be removed from the lines. The
329
``filename`` attribute of your ConfigObj will be ``None``.
331
* A ``StringIO`` instance or file object, or any object with a ``read`` method.
332
The ``filename`` attribute of your ConfigObj will be ``None`` [#]_.
334
* A dictionary. You can initialise a ConfigObj from a dictionary [#]_. The
335
``filename`` attribute of your ConfigObj will be ``None``. All keys must be
336
strings. In this case, the order of values and sections is arbitrary.
338
* 'raise_errors': ``False``
340
When parsing, it is possible that the config file will be badly formed. The
341
default is to parse the whole file and raise a single error at the end. You
342
can set ``raise_errors = True`` to have errors raised immediately. See the
343
exceptions_ section for more details.
345
Altering this value after initial parsing has no effect.
347
* 'list_values': ``True``
349
If ``True`` (the default) then list values are possible. If ``False``, the
350
values are not parsed for lists.
352
If ``list_values = False`` then single line values are not quoted or
353
unquoted when reading and writing.
355
Changing this value affects whether single line values will be quoted or
358
* 'create_empty': ``False``
360
If this value is ``True`` and the file specified by ``infile`` doesn't
361
exist, ConfigObj will create an empty file. This can be a useful test that
362
the filename makes sense: an impossible filename will cause an error.
364
Altering this value after initial parsing has no effect.
366
* 'file_error': ``False``
368
If this value is ``True`` and the file specified by ``infile`` doesn't
369
exist, ConfigObj will raise an ``IOError``.
371
Altering this value after initial parsing has no effect.
373
* 'interpolation': ``True``
375
Whether string interpolation is switched on or not. It is on (``True``) by
378
You can set this attribute to change whether string interpolation is done
379
when values are fetched. See the `String Interpolation`_ section for more details.
381
New in ConfigObj 4.7.0: Interpolation will also be done in list values.
383
* 'configspec': ``None``
385
If you want to use the validation system, you supply a configspec. This is
386
effectively a type of config file that specifies a check for each member.
387
This check can be used to do type conversion as well as check that the
388
value is within your required parameters.
390
You provide a configspec in the same way as you do the initial file: a
391
filename, or list of lines, etc. See the validation_ section for full
392
details on how to use the system.
394
When parsed, every section has a ``configspec`` with a dictionary of
395
configspec checks for *that section*.
397
* 'stringify': ``True``
399
If you use the validation scheme, it can do type checking *and* conversion
400
for you. This means you may want to set members to integers, or other
403
If 'stringify' is set to ``True`` (default) then non-string values will
404
be converted to strings when you write the config file. The validation_
405
process converts values from strings to the required type.
407
If 'stringify' is set to ``False``, attempting to set a member to a
408
non-string value [#]_ will raise a ``TypeError`` (no type conversion is
411
* 'indent_type': ``' '``
413
Indentation is not significant; it can however be present in the input and
414
output config. Any combination of tabs and spaces may be used: the string
415
will be repeated for each level of indentation. Typical values are: ``''``
416
(no indentation), ``' '`` (indentation with four spaces, the default),
417
``'\t'`` (indentation with one tab).
419
If this option is not specified, and the ConfigObj is initialised with a
420
dictionary, the indentation used in the output is the default one, that is,
423
If this option is not specified, and the ConfigObj is initialised with a
424
list of lines or a file, the indentation used in the first indented line is
425
selected and used in all output lines. If no input line is indented, no
426
output line will be either.
428
If this option *is* specified, the option value is used in the output
429
config, overriding the type of indentation in the input config (if any).
431
* 'encoding': ``None``
433
By default **ConfigObj** does not decode the file/strings you pass it into
434
Unicode [#]_. If you want your config file as Unicode (keys and members)
435
you need to provide an encoding to decode the file with. This encoding will
436
also be used to encode the config file when writing.
438
You can change the encoding attribute at any time.
440
Any characters in your strings that can't be encoded with the specified
441
encoding will raise a ``UnicodeEncodeError``.
445
``UTF16`` encoded files will automatically be detected and decoded,
446
even if ``encoding`` is ``None``.
448
This is because it is a 16-bit encoding, and ConfigObj will mangle it
449
(split characters on byte boundaries) if it parses it without decoding.
451
* 'default_encoding': ``None``
453
When using the ``write`` method, **ConfigObj** uses the ``encoding``
454
attribute to encode the Unicode strings. If any members (or keys) have
455
been set as byte strings instead of Unicode, these must first be decoded
456
to Unicode before outputting in the specified encoding.
458
``default_encoding``, if specified, is the encoding used to decode byte
459
strings in the **ConfigObj** before writing. If this is ``None``, then
460
the Python default encoding (``sys.defaultencoding`` - usually ASCII) is
463
For most Western European users, a value of ``latin-1`` is sensible.
465
``default_encoding`` is *only* used if an ``encoding`` is specified.
467
Any characters in byte-strings that can't be decoded using the
468
``default_encoding`` will raise a ``UnicodeDecodeError``.
470
* 'unrepr': ``False``
472
The ``unrepr`` option reads and writes files in a different mode. This
473
allows you to store and retrieve the basic Python data-types using config
476
This uses Python syntax for lists and quoting. See `unrepr mode`_ for the
479
* 'write_empty_values': ``False``
481
If ``write_empty_values`` is ``True``, empty strings are written as
482
empty values. See `Empty Values`_ for more details.
484
* '_inspec': ``False``
486
Used internally by ConfigObj when parsing configspec files. If you are
487
creating a ConfigObj instance from a configspec file you must pass True
488
for this argument as well as ``list_values=False``.
494
The ConfigObj is a subclass of an object called ``Section``, which is itself a
495
subclass of ``dict``, the builtin dictionary type. This means it also has
496
**all** the normal dictionary methods.
498
In addition, the following `Section Methods`_ may be useful :
510
Read about Sections_ for details of all the methods.
514
The *merge* method of sections is a recursive update.
516
You can use this to merge sections, or even whole ConfigObjs, into each
519
You would typically use this to create a default ConfigObj and then merge
520
in user settings. This way users only need to specify values that are
521
different from the default. You can use configspecs and validation to
522
achieve the same thing of course.
525
The public methods available on ConfigObj are :
536
.. code-block:: python
538
write(file_object=None)
540
This method writes the current ConfigObj and takes a single, optional argument
543
If you pass in a file like object to the ``write`` method, the config file will
544
be written to this. (The only method of this object that is used is its
545
``write`` method, so a ``StringIO`` instance, or any other file like object
548
Otherwise, the behaviour of this method depends on the ``filename`` attribute
552
ConfigObj will write the configuration to the file specified.
555
``write`` returns a list of lines. (Not ``'\n'`` terminated)
557
First the 'initial_comment' is written, then the config file, followed by the
558
'final_comment'. Comment lines and inline comments are written with each
566
.. code-block:: python
568
validate(validator, preserve_errors=False, copy=False)
570
.. code-block:: python
572
# filename is the config file
573
# filename2 is the configspec
574
# (which could also be hardcoded into your program)
575
config = ConfigObj(filename, configspec=filename2)
577
from validate import Validator
579
test = config.validate(val)
583
The validate method uses the `validate
584
<http://www.voidspace.org.uk/python/validate.html>`__ module to do the
587
This method validates the ConfigObj against the configspec. By doing type
588
conversion as well it can abstract away the config file altogether and present
589
the config *data* to your application (in the types it expects it to be).
591
If the ``configspec`` attribute of the ConfigObj is ``None``, it raises a
594
If the stringify_ attribute is set, this process will convert values to the
595
type defined in the configspec.
597
The validate method uses checks specified in the configspec and defined in the
598
``Validator`` object. It is very easy to extend.
600
The configspec looks like the config file, but instead of the value, you
601
specify the check (and any default value). See the validation_ section for
606
The system of configspecs can seem confusing at first, but is actually
607
quite simple and powerful. The best guide to them is this article on
610
* `An Introduction to ConfigObj`_
612
The ``copy`` parameter fills in missing values from the configspec (default
613
values), *without* marking the values as defaults. It also causes comments to
614
be copied from the configspec into the config file. This allows you to use a
615
configspec to create default config files. (Normally default values aren't
616
written out by the ``write`` method.)
618
As of ConfigObj 4.3.0 you can also pass in a ConfigObj instance as your
619
configspec. This is especially useful if you need to specify the encoding of
620
your configspec file. When you read your configspec file, you *must* specify
621
``list_values=False``. If you need to support hashes inside the configspec
622
values then you must also pass in ``_inspec=True``. This is because configspec
623
files actually use a different syntax to config files and inline comment support
624
must be switched off to correctly read configspec files with hashes in the values.
627
.. code-block:: python
629
from configobj import ConfigObj
630
configspec = ConfigObj(configspecfilename, encoding='UTF8',
631
list_values=False, _inspec=True)
632
config = ConfigObj(filename, configspec=configspec)
638
By default, the validate method either returns ``True`` (everything passed)
639
or a dictionary of ``True`` / ``False`` representing pass/fail. The dictionary
640
follows the structure of the ConfigObj.
642
If a whole section passes then it is replaced with the value ``True``. If a
643
whole section fails, then it is replaced with the value ``False``.
645
If a value is missing, and there is no default in the check, then the check
648
The ``validate`` method takes an optional keyword argument ``preserve_errors``.
649
If you set this to ``True``, instead of getting ``False`` for failed checks you
650
get the actual error object from the **validate** module. This usually contains
651
useful information about why the check failed.
653
See the `flatten_errors`_ function for how to turn your results dictionary into
654
a useful list of error messages.
656
Even if ``preserve_errors`` is ``True``, missing keys or sections will still be
657
represented by a ``False`` in the results dictionary.
660
Mentioning Default Values
661
#########################
663
In the check in your configspec, you can specify a default to be used - by
664
using the ``default`` keyword. E.g. ::
666
key1 = integer(0, 30, default=15)
667
key2 = integer(default=15)
668
key3 = boolean(default=True)
669
key4 = option('Hello', 'Goodbye', 'Not Today', default='Not Today')
671
If the configspec check supplies a default and the value is missing in the
672
config, then the default will be set in your ConfigObj. (It is still passed to
673
the ``Validator`` so that type conversion can be done: this means the default
674
value must still pass the check.)
676
ConfigObj keeps a record of which values come from defaults, using the
677
``defaults`` attribute of sections_. Any key in this list isn't written out by
678
the ``write`` method. If a key is set from outside (even to the same value)
679
then it is removed from the ``defaults`` list.
683
Even if all the keys in a section are in the defaults list, the section
684
marker is still written out.
686
There is additionally a special case default value of ``None``. If you set the
687
default value to ``None`` and the value is missing, the value will always be
688
set to ``None``. As the other checks don't return ``None`` (unless you
689
implement your own that do), you can tell that this value came from a default
690
value (and was missing from the config file). It allows an easy way of
691
implementing optional values. Simply check (and ignore) members that are set
696
If stringify_ is ``False`` then ``default=None`` returns ``''`` instead of
697
``None``. This is because setting a value to a non-string raises an error
698
if stringify is unset.
700
The default value can be a list. See `List Values`_ for the way to do this.
702
Writing invalid default values is a *guaranteed* way of confusing your users.
703
Default values **must** pass the check.
706
Mentioning Repeated Sections and Values
707
#######################################
709
In the configspec it is possible to cause *every* sub-section in a section to
710
be validated using the same configspec. You do this with a section in the
711
configspec called ``__many__``. Every sub-section in that section has the
712
``__many__`` configspec applied to it (without you having to explicitly name
715
Your ``__many__`` section can have nested subsections, which can also include
716
``__many__`` type sections.
718
You can also specify that all values should be validated using the same configspec,
719
by having a member with the name ``__many__``. If you want to use repeated values
720
along with repeated sections then you can call one of them ``___many___`` (triple
723
Sections with repeated sections or values can also have specifically named sub-sections
724
or values. The ``__many__`` configspec will only be used to validate entries that don't
725
have an explicit configspec.
727
See `Repeated Sections`_ for examples.
733
If you just want to check if all members are present, then you can use the
734
``SimpleVal`` object that comes with ConfigObj. It only fails members if they
737
Write a configspec that has all the members you want to check for, but set
738
every section to ``''``.
740
.. code-block:: python
743
test = config.validate(val)
751
As discussed in `Mentioning Default Values`_, you can use a configspec to
752
supply default values. These are marked in the ConfigObj instance as defaults,
753
and *not* written out by the ``write`` mode. This means that your users only
754
need to supply values that are different from the defaults.
756
This can be inconvenient if you *do* want to write out the default values,
757
for example to write out a default config file.
759
If you set ``copy=True`` when you call validate, then no values are marked as
760
defaults. In addition, all comments from the configspec are copied into
761
your ConfigObj instance. You can then call ``write`` to create your config
764
There is a limitation with this. In order to allow `String Interpolation`_ to work
765
within configspecs, ``DEFAULT`` sections are not processed by
766
validation; even in copy mode.
772
If a ConfigObj instance was loaded from the filesystem, then this method will reload it. It
773
will also reuse any configspec you supplied at instantiation (including reloading it from
774
the filesystem if you passed it in as a filename).
776
If the ConfigObj does not have a filename attribute pointing to a file, then a ``ReloadError``
783
This method takes no arguments and doesn't return anything. It restores a ConfigObj
784
instance to a freshly created state.
790
A ConfigObj has the following attributes :
807
This doesn't include *comments*, *inline_comments*, *defaults*, or
808
*configspec*. These are actually attributes of Sections_.
810
It also has the following attributes as a result of parsing. They correspond to
811
options when the ConfigObj was created, but changing them has no effect.
821
ConfigObj can perform string interpolation in a *similar* way to
822
``ConfigParser``. See the `String Interpolation`_ section for full details.
824
If ``interpolation`` is set to ``False``, then interpolation is *not* done when
831
If this attribute is set (``True``) then the validate_ method changes the
832
values in the ConfigObj. These are turned back into strings when write_ is
835
If stringify is unset (``False``) then attempting to set a value to a non
836
string (or a list of strings) will raise a ``TypeError``.
842
If the initial config file *started* with the UTF8 Unicode signature (known
843
slightly incorrectly as the BOM - Byte Order Mark), or the UTF16 BOM, then
844
this attribute is set to ``True``. Otherwise it is ``False``.
846
If it is set to ``True`` when ``write`` is called then, if ``encoding`` is set
847
to ``None`` *or* to ``utf_8`` (and variants) a UTF BOM will be written.
849
For UTF16 encodings, a BOM is *always* written.
855
This is a list of lines. If the ConfigObj is created from an existing file, it
856
will contain any lines of comments before the start of the members.
858
If you create a new ConfigObj, this will be an empty list.
860
The write method puts these lines before it starts writing out the members.
866
This is a list of lines. If the ConfigObj is created from an existing file, it
867
will contain any lines of comments after the last member.
869
If you create a new ConfigObj, this will be an empty list.
871
The ``write`` method puts these lines after it finishes writing out the
878
This attribute is ``True`` or ``False``. If set to ``False`` then values are
879
not parsed for list values. In addition single line values are not unquoted.
881
This allows you to do your own parsing of values. It exists primarily to
882
support the reading of the configspec_ - but has other use cases.
884
For example you could use the ``LineParser`` from the
885
`listquote module <http://www.voidspace.org.uk/python/listquote.html#lineparser>`_
886
to read values for nested lists.
888
Single line values aren't quoted when writing - but multiline values are
893
Because values aren't quoted, leading or trailing whitespace can be lost. This behaviour was changed in version 4.0.1. Prior to this, single line values might have been quoted; even with ``list_values=False``. This means that files written by earlier versions of ConfigObj *could* now be incompatible and need the quotes removing by hand.
899
This is the encoding used to encode the output, when you call ``write``. It
900
must be a valid encoding `recognised by Python <http://docs.python.org/lib/standard-encodings.html>`_.
902
If this value is ``None`` then no encoding is done when ``write`` is called.
908
If encoding is set, any byte-strings in your ConfigObj instance (keys or
909
members) will first be decoded to Unicode using the encoding specified by the
910
``default_encoding`` attribute. This ensures that the output is in the encoding
913
If this value is ``None`` then ``sys.defaultencoding`` is used instead.
919
Another boolean value. If this is set, then ``repr(value)`` is used to write
920
values. This writes values in a slightly different way to the normal ConfigObj
923
This preserves basic Python data-types when read back in. See `unrepr mode`_
930
Also boolean. If set, values that are an empty string (``''``) are written as
931
empty values. See `Empty Values`_ for more details.
937
When a config file is read, ConfigObj records the type of newline separators in the
938
file and uses this separator when writing. It defaults to ``None``, and ConfigObj
939
uses the system default (``os.sep``) if write is called without newlines having
943
The Config File Format
944
======================
946
You saw an example config file in the `Config Files`_ section. Here is a fuller
947
specification of the config files used and created by ConfigObj.
949
The basic pattern for keywords is::
953
keyword = value # inline comment
955
Both keyword and value can optionally be surrounded in quotes. The equals sign
956
is the only valid divider.
958
Values can have comments on the lines above them, and an inline comment after
959
them. This, of course, is optional. See the comments_ section for details.
961
If a keyword or value starts or ends with whitespace, or contains a quote mark
962
or comma, then it should be surrounded by quotes. Quotes are not necessary if
963
whitespace is surrounded by non-whitespace.
965
Values can also be lists. Lists are comma separated. You indicate a single
966
member list by a trailing comma. An empty list is shown by a single comma::
968
keyword1 = value1, value2, value3
969
keyword2 = value1, # a single member list
970
keyword3 = , # an empty list
972
Values that contain line breaks (multi-line values) can be surrounded by triple
973
quotes. These can also be used if a value contains both types of quotes. List
974
members cannot be surrounded by triple quotes::
976
keyword1 = ''' A multi line value
978
lines''' # with a comment
979
keyword2 = '''I won't be "afraid".'''
981
keyword3 = """ A multi line value
983
lines""" # with a comment
984
keyword4 = """I won't be "afraid"."""
988
There is no way of safely quoting values that contain both types of triple
991
A line that starts with a '#', possibly preceded by whitespace, is a comment.
993
New sections are indicated by a section marker line. That is the section name
994
in square brackets. Whitespace around the section name is ignored. The name can
995
be quoted with single or double quotes. The marker can have comments before it
996
and an inline comment after it::
999
[ section name 1 ] # first section
1002
# The Second Section
1003
[ "section name 2" ] # second section
1006
Any subsections (sections that are *inside* the current section) are
1007
designated by repeating the square brackets before and after the section name.
1008
The number of square brackets represents the nesting level of the sub-section.
1009
Square brackets may be separated by whitespace; such whitespace, however, will
1010
not be present in the output config written by the ``write`` method.
1012
Indentation is not significant, but can be preserved. See the description of
1013
the ``indent_type`` option, in the `ConfigObj specifications`_ chapter, for the
1016
A *NestingError* will be raised if the number of the opening and the closing
1017
brackets in a section marker is not the same, or if a sub-section's nesting
1018
level is greater than the nesting level of it parent plus one.
1020
In the outer section, single values can only appear before any sub-section.
1021
Otherwise they will belong to the sub-section immediately before them::
1032
# this is in section 1
1036
[[[nested section]]]
1037
# this is in sub section
1042
# this is in section 1 again
1047
# this is also in section 1, indentation is misleading here
1053
When parsed, the above config file produces the following data structure:
1056
.. code-block:: python
1059
'keyword1': 'value1',
1060
'keyword2': 'value2',
1062
'keyword1': 'value1',
1063
'keyword2': 'value2',
1065
'keyword1': 'value1',
1066
'keyword2': 'value2',
1068
'keyword1': 'value1',
1069
'keyword2': 'value2',
1073
'keyword1': 'value1',
1074
'keyword2': 'value2',
1077
'keyword1': 'value1',
1078
'keyword2': 'value2',
1084
Sections are ordered: note how the structure of the resulting ConfigObj is in
1085
the same order as the original file.
1089
In ConfigObj 4.3.0 *empty values* became valid syntax. They are read as the
1090
empty string. There is also an option/attribute (``write_empty_values``) to
1091
allow the writing of these.
1093
This is mainly to support 'legacy' config files, written from other
1094
applications. This is documented under `Empty Values`_.
1096
`unrepr mode`_ introduces *another* syntax variation, used for storing
1097
basic Python datatypes in config files.
1103
Every section in a ConfigObj has certain properties. The ConfigObj itself also
1104
has these properties, because it too is a section (sometimes called the *root
1107
``Section`` is a subclass of the standard new-class dictionary, therefore it
1108
has **all** the methods of a normal dictionary. This means you can ``update``
1109
and ``clear`` sections.
1113
You create a new section by assigning a member to be a dictionary.
1115
The new ``Section`` is created *from* the dictionary, but isn't the same
1116
thing as the dictionary. (So references to the dictionary you use to create
1117
the section *aren't* references to the new section).
1121
.. code-block:: python
1123
config = ConfigObj()
1124
vals = {'key1': 'value 1',
1127
config['vals'] = vals
1128
config['vals'] == vals
1130
config['vals'] is vals
1133
If you now change ``vals``, the changes won't be reflected in ``config['vals']``.
1135
A section is ordered, following its ``scalars`` and ``sections``
1136
attributes documented below. This means that the following dictionary
1137
attributes return their results in order.
1141
More commonly known as ``for member in section:``.
1143
* '__repr__' and '__str__'
1145
Any time you print or display the ConfigObj.
1167
A reference to the main ConfigObj.
1171
A reference to the 'parent' section, the section that this section is a
1174
On the ConfigObj this attribute is a reference to itself. You can use this
1175
to walk up the sections, stopping when ``section.parent is section``.
1179
The nesting level of the current section.
1181
If you create a new ConfigObj and add sections, 1 will be added to the
1182
depth level between sections.
1186
This attribute is a list of scalars that came from default values. Values
1187
that came from defaults aren't written out by the ``write`` method.
1188
Setting any of these values in the section removes them from the defaults
1193
This attribute is a dictionary mapping keys to the default values for the
1194
keys. By default it is an empty dictionary and is populated when you
1195
validate the ConfigObj.
1199
These attributes are normal lists, representing the order that members,
1200
single values and subsections appear in the section. The order will either
1201
be the order of the original config file, *or* the order that you added
1204
The order of members in this lists is the order that ``write`` creates in
1205
the config file. The ``scalars`` list is output before the ``sections``
1208
Adding or removing members also alters these lists. You can manipulate the
1209
lists directly to alter the order of members.
1213
If you alter the ``scalars``, ``sections``, or ``defaults`` attributes
1214
so that they no longer reflect the contents of the section, you will
1215
break your ConfigObj.
1217
See also the ``rename`` method.
1221
This is a dictionary of comments associated with each member. Each entry is
1222
a list of lines. These lines are written out before the member.
1226
This is *another* dictionary of comments associated with each member. Each
1227
entry is a string that is put inline with the member.
1231
The configspec attribute is a dictionary mapping scalars to *checks*. A
1232
check defines the expected type and possibly the allowed values for a
1235
The configspec has the same format as a config file, but instead of values
1236
it has a specification for the value (which may include a default value).
1237
The validate_ method uses it to check the config file makes sense. If a
1238
configspec is passed in when the ConfigObj is created, then it is parsed
1239
and broken up to become the ``configspec`` attribute of each section.
1241
If you didn't pass in a configspec, this attribute will be ``None`` on the
1242
root section (the main ConfigObj).
1244
You can set the configspec attribute directly on a section.
1246
See the validation_ section for full details of how to write configspecs.
1250
By default an empty list. After validation_ this is populated with any members
1251
of the section that don't appear in the configspec (i.e. they are additional
1252
values). Rather than accessing this directly it may be more convenient to get
1253
all the extra values in a config file using the get_extra_values_ function.
1255
New in ConfigObj 4.7.0.
1263
This method takes no arguments. It returns a deep copy of the section as a
1264
dictionary. All subsections will also be dictionaries, and list values will
1265
be copies, rather than references to the original [#]_.
1269
``rename(oldkey, newkey)``
1271
This method renames a key, without affecting its position in the sequence.
1277
This method is a *recursive update* method. It allows you to merge two
1278
config files together.
1280
You would typically use this to create a default ConfigObj and then merge
1281
in user settings. This way users only need to specify values that are
1282
different from the default.
1286
.. code-block:: python
1288
# def_cfg contains your default config settings
1289
# user_cfg contains the user settings
1290
cfg = ConfigObj(def_cfg)
1291
usr = ConfigObj(user_cfg)
1296
cfg now contains a combination of the default settings and the user
1299
The user settings will have overwritten any of the default ones.
1304
This method can be used to transform values and names. See `walking a
1305
section`_ for examples and explanation.
1311
Returns ``True`` if the key contains a string that represents ``True``, or
1312
is the ``True`` object.
1314
Returns ``False`` if the key contains a string that represents ``False``,
1315
or is the ``False`` object.
1317
Raises a ``ValueError`` if the key contains anything else.
1319
Strings that represent ``True`` are (not case sensitive)::
1323
Strings that represent ``False`` are::
1332
This returns the value contained in the specified key as an integer.
1334
It raises a ``ValueError`` if the conversion can't be done.
1341
This returns the value contained in the specified key as a float.
1343
It raises a ``ValueError`` if the conversion can't be done.
1350
This returns the value contained in the specified key as a list.
1352
If it isn't a list it will be wrapped as a list so that you can
1353
guarantee the returned value will be a list.
1356
* **restore_default**
1358
``restore_default(key)``
1360
Restore (and return) the default value for the specified key.
1362
This method will only work for a ConfigObj that was created
1363
with a configspec and has been validated.
1365
If there is no default value for this key, ``KeyError`` is raised.
1367
* **restore_defaults**
1369
``restore_defaults()``
1371
Recursively restore default values to all members
1374
This method will only work for a ConfigObj that was created
1375
with a configspec and has been validated.
1377
It doesn't delete or modify entries without default values.
1385
The walk method allows you to call a function on every member/name.
1387
.. code-block:: python
1389
walk(function, raise_errors=True,
1390
call_on_sections=False, **keywargs)
1393
``walk`` is a method of the ``Section`` object. This means it is also a method
1396
It walks through every member and calls a function on the keyword and value. It
1397
walks recursively through subsections.
1399
It returns a dictionary of all the computed values.
1401
If the function raises an exception, the default is to propagate the error, and
1402
stop. If ``raise_errors=False`` then it sets the return value for that keyword
1403
to ``False`` instead, and continues. This is similar to the way validation_
1406
Your function receives the arguments ``(section, key)``. The current value is
1407
then ``section[key]`` [#]_. Any unrecognised keyword arguments you pass to
1408
walk, are passed on to the function.
1410
Normally ``walk`` just recurses into subsections. If you are transforming (or
1411
checking) names as well as values, then you want to be able to change the names
1412
of sections. In this case set ``call_on_sections`` to ``True``. Now, on
1413
encountering a sub-section, *first* the function is called for the *whole*
1414
sub-section, and *then* it recurses into it's members. This means your function
1415
must be able to handle receiving dictionaries as well as strings and lists.
1417
If you are using the return value from ``walk`` *and* ``call_on_sections``,
1418
note that walk discards the return value when it calls your function.
1422
You can use ``walk`` to transform the names of members of a section
1423
but you mustn't add or delete members.
1429
You can use this for transforming all values in your ConfigObj. For example
1430
you might like the nested lists from ConfigObj 3. This was provided by the
1431
listquote_ module. You could switch off the parsing for list values
1432
(``list_values=False``) and use listquote to parse every value.
1434
Another thing you might want to do is use the Python escape codes in your
1435
values. You might be *used* to using ``\n`` for line feed and ``\t`` for tab.
1436
Obviously we'd need to decode strings that come from the config file (using the
1437
escape codes). Before writing out we'll need to put the escape codes back in
1440
As an example we'll write a function to use with walk, that encodes or decodes
1441
values using the ``string-escape`` codec.
1443
The function has to take each value and set the new value. As a bonus we'll
1444
create one function that will do decode *or* encode depending on a keyword
1447
We don't want to work with section names, we're only transforming values, so
1448
we can leave ``call_on_sections`` as ``False``. This means the two datatypes we
1449
have to handle are strings and lists, we can ignore everything else. (We'll
1450
treat tuples as lists as well).
1452
We're not using the return values, so it doesn't need to return anything, just
1453
change the values if appropriate.
1456
.. code-block:: python
1458
def string_escape(section, key, encode=False):
1460
A function to encode or decode using the 'string-escape' codec.
1461
To be passed to the walk method of a ConfigObj.
1462
By default it decodes.
1463
To encode, pass in the keyword argument ``encode=True``.
1466
# is it a type we can work with
1467
# NOTE: for platforms where Python > 2.2
1468
# you can use basestring instead of (str, unicode)
1469
if not isinstance(val, (str, unicode, list, tuple)):
1472
elif isinstance(val, (str, unicode)):
1475
section[key] = val.decode('string-escape')
1477
section[key] = val.encode('string-escape')
1479
# it must be a list or tuple!
1480
# we'll be lazy and create a new list
1482
# we'll check every member of the list
1484
if isinstance(entry, (str, unicode)):
1486
newval.append(entry.decode('string-escape'))
1488
newval.append(entry.encode('string-escape'))
1490
newval.append(entry)
1492
section[key] = newval
1494
# assume we have a ConfigObj called ``config``
1497
config.walk(string_escape)
1500
# Because ``walk`` doesn't recognise the ``encode`` argument
1501
# it passes it to our function.
1502
config.walk(string_escape, encode=True)
1505
Here's a simple example of using ``walk`` to transform names and values. One
1506
usecase of this would be to create a *standard* config file with placeholders
1507
for section and keynames. You can then use walk to create new config files
1508
and change values and member names :
1510
.. code-block:: python
1512
# We use 'XXXX' as a placeholder
1514
XXXXkey1 = XXXXvalue1
1515
XXXXkey2 = XXXXvalue2
1516
XXXXkey3 = XXXXvalue3
1518
XXXXkey1 = XXXXvalue1
1519
XXXXkey2 = XXXXvalue2
1520
XXXXkey3 = XXXXvalue3
1522
XXXXkey1 = XXXXvalue1
1523
XXXXkey2 = XXXXvalue2
1524
XXXXkey3 = XXXXvalue3
1526
XXXXkey1 = XXXXvalue1
1527
XXXXkey2 = XXXXvalue2
1528
XXXXkey3 = XXXXvalue3
1530
cfg = ConfigObj(config)
1532
def transform(section, key):
1534
newkey = key.replace('XXXX', 'CLIENT1')
1535
section.rename(key, newkey)
1536
if isinstance(val, (tuple, list, dict)):
1539
val = val.replace('XXXX', 'CLIENT1')
1540
section[newkey] = val
1542
cfg.walk(transform, call_on_sections=True)
1544
ConfigObj({'CLIENT1key1': 'CLIENT1value1', 'CLIENT1key2': 'CLIENT1value2',
1545
'CLIENT1key3': 'CLIENT1value3',
1546
'CLIENT1section1': {'CLIENT1key1': 'CLIENT1value1',
1547
'CLIENT1key2': 'CLIENT1value2', 'CLIENT1key3': 'CLIENT1value3'},
1548
'CLIENT1section2': {'CLIENT1key1': 'CLIENT1value1',
1549
'CLIENT1key2': 'CLIENT1value2', 'CLIENT1key3': 'CLIENT1value3',
1550
'CLIENT1section1': {'CLIENT1key1': 'CLIENT1value1',
1551
'CLIENT1key2': 'CLIENT1value2', 'CLIENT1key3': 'CLIENT1value3'}}})
1557
There are several places where ConfigObj may raise exceptions (other than
1560
1) If a configspec filename you pass in doesn't exist, or a config file
1561
filename doesn't exist *and* ``file_error=True``, an ``IOError`` will be
1564
2) If you try to set a non-string key, or a non string value when
1565
``stringify=False``, a ``TypeError`` will be raised.
1567
3) A badly built config file will cause parsing errors.
1569
4) A parsing error can also occur when reading a configspec.
1571
5) In string interpolation you can specify a value that doesn't exist, or
1572
create circular references (recursion).
1574
Number 5 (which is actually two different types of exceptions) is documented
1575
in `String Interpolation`_.
1577
*This* section is about errors raised during parsing.
1579
The base error class is ``ConfigObjError``. This is a subclass of
1580
``SyntaxError``, so you can trap for ``SyntaxError`` without needing to
1581
directly import any of the ConfigObj exceptions.
1583
The following other exceptions are defined (all deriving from
1584
``ConfigObjError``) :
1588
This error indicates either a mismatch in the brackets in a section marker,
1589
or an excessive level of nesting.
1593
This error indicates that a line is badly written. It is neither a valid
1594
``key = value`` line, nor a valid section marker line, nor a comment line.
1596
* ``DuplicateError``
1598
The keyword or section specified already exists.
1600
* ``ConfigspecError``
1602
An error occurred whilst parsing a configspec.
1606
An error occurred when parsing a value in `unrepr mode`_.
1610
``reload`` was called on a ConfigObj instance that doesn't have a valid
1613
When parsing a configspec, ConfigObj will stop on the first error it
1614
encounters. It will raise a ``ConfigspecError``. This will have an ``error``
1615
attribute, which is the actual error that was raised.
1617
Behaviour when parsing a config file depends on the option ``raise_errors``.
1618
If ConfigObj encounters an error while parsing a config file:
1620
If ``raise_errors=True`` then ConfigObj will raise the appropriate error
1621
and parsing will stop.
1623
If ``raise_errors=False`` (the default) then parsing will continue to the
1624
end and *all* errors will be collected.
1626
If ``raise_errors`` is False and multiple errors are found a ``ConfigObjError``
1627
is raised. The error raised has a ``config`` attribute, which is the parts of
1628
the ConfigObj that parsed successfully. It also has an attribute ``errors``,
1629
which is a list of *all* the errors raised. Each entry in the list is an
1630
instance of the appropriate error type. Each one has the following attributes
1631
(useful for delivering a sensible error message to your user) :
1633
* ``line``: the original line that caused the error.
1635
* ``line_number``: its number in the config file.
1637
* ``message``: the error message that accompanied the error.
1639
If only one error is found, then that error is re-raised. The error still has
1640
the ``config`` and ``errors`` attributes. This means that your error handling
1641
code can be the same whether one error is raised in parsing , or several.
1643
It also means that in the most common case (a single error) a useful error
1644
message will be raised.
1648
One wrongly written line could break the basic structure of your config
1649
file. This could cause every line after it to flag an error, so having a
1650
list of all the lines that caused errors may not be as useful as it sounds.
1658
The system of configspecs can seem confusing at first, but is actually
1659
quite simple and powerful. The best reference is my article on ConfigObj:
1661
* `An Introduction to ConfigObj`_
1664
Validation is done through a combination of the configspec_ and a ``Validator``
1665
object. For this you need *validate.py* [#]_. See downloading_ if you don't
1668
Validation can perform two different operations :
1670
1) Check that a value meets a specification. For example, check that a value
1671
is an integer between one and six, or is a choice from a specific set of
1674
2) It can convert the value into the type required. For example, if one of
1675
your values is a port number, validation will turn it into an integer for
1678
So validation can act as a transparent layer between the datatypes of your
1679
application configuration (boolean, integers, floats, etc) and the text format
1680
of your config file.
1686
The ``validate`` method checks members against an entry in the configspec. Your
1687
configspec therefore resembles your config file, with a check for every member.
1689
In order to perform validation you need a ``Validator`` object. This has
1690
several useful built-in check functions. You can also create your own custom
1691
functions and register them with your Validator object.
1693
Each check is the name of one of these functions, including any parameters and
1694
keyword arguments. The configspecs look like function calls, and they map to
1697
The basic datatypes that an un-extended Validator can test for are :
1699
* boolean values (True and False)
1700
* integers (including minimum and maximum values)
1701
* floats (including min and max)
1702
* strings (including min and max length)
1703
* IP addresses (v4 only)
1705
It can also handle lists of these types and restrict a value to being one from
1708
An example configspec is going to look something like::
1710
port = integer(0, 100)
1711
user = string(max=25)
1712
mode = option('quiet', 'loud', 'silent')
1714
You can specify default values, and also have the same configspec applied to
1715
several sections. This is called `repeated sections`_.
1717
For full details on writing configspecs, please refer to the `validate.py
1722
Your configspec is read by ConfigObj in the same way as a config file.
1724
That means you can do interpolation *within* your configspec.
1726
In order to allow this, checks in the 'DEFAULT' section (of the root level
1727
of your configspec) are *not* used.
1729
If you want to use a configspec *without* interpolation being done in it
1730
you can create your configspec manually and switch off interpolation:
1732
.. code-block:: python
1734
from configobj import ConfigObj
1736
configspec = ConfigObj(spec_filename, interpolation=False, list_values=False,
1738
conf = ConfigObj(config_filename, configspec=configspec)
1740
If you need to specify the encoding of your configspec, then you can pass in a
1741
ConfigObj instance as your configspec. When you read your configspec file, you
1742
*must* specify ``list_values=False``. If you need to support hashes in
1743
configspec values then you must also pass in ``_inspec=True``.
1745
.. code-block:: python
1747
from configobj import ConfigObj
1748
configspec = ConfigObj(configspecfilename, encoding='UTF8',
1749
list_values=False, _inspec=True)
1750
config = ConfigObj(filename, configspec=configspec)
1752
.. _validate.py documentation: http://www.voidspace.org.uk/python/validate.html
1758
By default, validation does type conversion. This means that if you specify
1759
``integer`` as the check, then calling validate_ will actually change the value
1760
to an integer (so long as the check succeeds).
1762
It also means that when you call the write_ method, the value will be converted
1763
back into a string using the ``str`` function.
1765
To switch this off, and leave values as strings after validation, you need to
1766
set the stringify_ attribute to ``False``. If this is the case, attempting to
1767
set a value to a non-string will raise an error.
1773
You can set a default value in your check. If the value is missing from the
1774
config file then this value will be used instead. This means that your user
1775
only has to supply values that differ from the defaults.
1777
If you *don't* supply a default then for a value to be missing is an error,
1778
and this will show in the `return value`_ from validate.
1780
Additionally you can set the default to be ``None``. This means the value will
1781
be set to ``None`` (the object) *whichever check is used*. (It will be set to
1782
``''`` rather than ``None`` if stringify_ is ``False``). You can use this
1783
to easily implement optional values in your config files. ::
1785
port = integer(0, 100, default=80)
1786
user = string(max=25, default=0)
1787
mode = option('quiet', 'loud', 'silent', default='loud')
1788
nick = string(default=None)
1792
Because the default goes through type conversion, it also has to pass the
1795
Note that ``default=None`` is case sensitive.
1801
It's possible that you will want to specify a list as a default value. To avoid
1802
confusing syntax with commas and quotes you use a list constructor to specify
1803
that keyword arguments are lists. This includes the ``default`` value. This
1804
makes checks look something like::
1806
checkname(default=list('val1', 'val2', 'val3'))
1808
This works with all keyword arguments, but is most useful for default values.
1814
Repeated sections are a way of specifying a configspec for a section that
1815
should be applied to all unspecified subsections in the same section.
1817
The easiest way of explaining this is to give an example. Suppose you have a
1818
config file that describes a dog. That dog has various attributes, but it can
1819
also have many fleas. You don't know in advance how many fleas there will be,
1820
or what they will be called, but you want each flea validated against the same
1823
We can define a section called *fleas*. We want every flea in that section
1824
(every sub-section) to have the same configspec applied to it. We do this by
1825
defining a single section called ``__many__``. ::
1828
name = string(default=Rover)
1829
age = float(0, 99, default=0)
1834
bloodsucker = boolean(default=True)
1835
children = integer(default=10000)
1836
size = option(small, tiny, micro, default=tiny)
1838
Every flea on our dog will now be validated using the ``__many__`` configspec.
1840
``__many__`` sections can have sub-sections, including their own ``__many__``
1841
sub-sections. Defaults work in the normal way in repeated sections.
1847
As well as using ``__many__`` to validate unspecified sections you can use it to validate values. For
1848
example, to specify that all values in a section should be integers::
1853
If you want to use repeated values alongside repeated sections you can call one ``__many__`` and the
1854
other ``___many___`` (with three underscores).
1860
Because you can specify default values in your configspec, you can use
1861
ConfigObj to write out default config files for your application.
1863
However, normally values supplied from a default in a configspec are *not*
1864
written out by the ``write`` method.
1866
To do this, you need to specify ``copy=True`` when you call validate. As well
1867
as not marking values as default, all the comments in the configspec file
1868
will be copied into your ConfigObj instance.
1871
.. code-block:: python
1873
from configobj import ConfigObj
1874
from validate import Validator
1876
config = ConfigObj(configspec='default.ini')
1877
config.filename = 'new_default.ini'
1878
config.validate(vdt, copy=True)
1881
If you need to support hashes in the configspec values then you must create
1882
it with ``_inspec=True``. This has the side effect of switching off the parsing
1883
of inline comments, meaning that they won't be copied into the new config file.
1884
(ConfigObj syntax is slightly different from configspec syntax and the parser
1885
can't support both inline comments and hashes in configspec values.)
1888
Validation and Interpolation
1889
----------------------------
1891
String interpolation and validation don't play well together. When validation
1892
changes type it sets the value. If the value uses interpolation, then the
1893
interpolation reference would normally be overwritten. Calling ``write`` would
1894
then use the absolute value and the interpolation reference would be lost.
1896
As a compromise - if the value is unchanged by validation then it is not reset.
1897
This means strings that pass through validation unmodified will not be
1898
overwritten. If validation changes type - the value has to be overwritten, and
1899
any interpolation references are lost.
1905
After validation the ``extra_values`` member of every section that is listed in
1906
the configspec will be populated with the names of members that are in the
1907
config file but not in the configspec.
1909
If you are reporting configuration errors to your user this information can be
1910
useful, for example some missing entries may be due to misspelt entries that
1911
appear as extra values.
1913
See the get_extra_values_ function
1915
New in ConfigObj 4.7.0.
1922
You may not need a full validation process, but still want to check if all the
1923
expected values are present.
1925
Provided as part of the ConfigObj module is the ``SimpleVal`` object. This has
1926
a dummy ``test`` method that always passes.
1928
The only reason a test will fail is if the value is missing. The return value
1929
from ``validate`` will either be ``True``, meaning all present, or a dictionary
1930
with ``False`` for all missing values/sections.
1932
To use it, you still need to pass in a valid configspec when you create the
1933
ConfigObj, but just set all the values to ``''``. Then create an instance of
1934
``SimpleVal`` and pass it to the ``validate`` method.
1936
As a trivial example if you had the following config file::
1938
# config file for an application
1942
top_level_domain = org.uk
1944
You would write the following configspec::
1949
top_level_domain = ''
1952
.. code-block:: python
1954
config = Configobj(filename, configspec=configspec)
1956
test = config.validate(val)
1958
print 'All values present.'
1960
print 'No values present!'
1963
if test[entry] == False:
1964
print '"%s" missing.' % entry
1970
Many config files from other applications allow empty values. As of version
1971
4.3.0, ConfigObj will read these as an empty string.
1973
A new option/attribute has been added (``write_empty_values``) to allow
1974
ConfigObj to write empty strings as empty values.
1976
.. code-block:: python
1978
from configobj import ConfigObj
1983
config = ConfigObj(cfg)
1985
ConfigObj({'key': '', 'key2': ''})
1987
config.write_empty_values = True
1988
for line in config.write():
1998
The ``unrepr`` option allows you to store and retrieve the basic Python
1999
data-types using config files. It has to use a slightly different syntax to
2000
normal ConfigObj files. Unsurprisingly it uses Python syntax.
2002
This means that lists are different (they are surrounded by square brackets),
2003
and strings *must* be quoted.
2005
The types that ``unrepr`` can work with are :
2007
| strings, lists tuples
2009
| dictionaries, integers, floats
2010
| longs and complex numbers
2012
You can't store classes, types or instances.
2014
``unrepr`` uses ``repr(object)`` to write out values, so it currently *doesn't*
2015
check that you are writing valid objects. If you attempt to read an unsupported
2016
value, ConfigObj will raise a ``configobj.UnknownType`` exception.
2018
Values that are triple quoted cased. The triple quotes are removed *before*
2019
converting. This means that you can use triple quotes to write dictionaries
2020
over several lines in your config files. They won't be written like this
2023
If you are writing config files by hand, for use with ``unrepr``, you should
2024
be aware of the following differences from normal ConfigObj syntax :
2026
| List : ``['A List', 'With', 'Strings']``
2027
| Strings : ``"Must be quoted."``
2028
| Backslash : ``"The backslash must be escaped \\"``
2030
These all follow normal Python syntax.
2032
In unrepr mode *inline comments* are not saved. This is because lines are
2033
parsed using the `compiler package <http://docs.python.org/lib/compiler.html>`_
2034
which discards comments.
2037
String Interpolation
2038
====================
2042
String interpolation can slow down (slightly) the fetching of values
2043
from your config object. If you aren't using interpolation and it
2044
is performance critical then create your instance with
2045
``interpolation=False``.
2047
ConfigObj allows string interpolation *similar* to the way ``ConfigParser``
2048
or ``string.Template`` work. The value of the ``interpolation`` attribute
2049
determines which style of interpolation you want to use. Valid values are
2050
"ConfigParser" or "Template" (case-insensitive, so "configparser" and
2051
"template" will also work). For backwards compatibility reasons, the value
2052
``True`` is also a valid value for the ``interpolation`` attribute, and
2053
will select ``ConfigParser``-style interpolation. At some undetermined point
2054
in the future, that default *may* change to ``Template``-style interpolation.
2056
For ``ConfigParser``-style interpolation, you specify a value to be
2057
substituted by including ``%(name)s`` in the value.
2059
For ``Template``-style interpolation, you specify a value to be substituted
2060
by including ``${cl}name{cr}`` in the value. Alternately, if 'name' is a valid
2061
Python identifier (i.e., is composed of nothing but alphanumeric characters,
2062
plus the underscore character), then the braces are optional and the value
2063
can be written as ``$name``.
2065
Note that ``ConfigParser``-style interpolation and ``Template``-style
2066
interpolation are mutually exclusive; you cannot have a configuration file
2067
that's a mix of one or the other. Pick one and stick to it. ``Template``-style
2068
interpolation is simpler to read and write by hand, and is recommended if
2069
you don't have a particular reason to use ``ConfigParser``-style.
2071
Interpolation checks first the current section to see if ``name`` is the key
2072
to a value. ('name' is case sensitive).
2074
If it doesn't find it, next it checks the 'DEFAULT' sub-section of the current
2077
If it still doesn't find it, it moves on to check the parent section and the
2078
parent section's 'DEFAULT' subsection, and so on all the way up to the main
2081
If the value specified isn't found in any of these locations, then a
2082
``MissingInterpolationOption`` error is raised (a subclass of
2083
``ConfigObjError``).
2085
If it is found then the returned value is also checked for substitutions. This
2086
allows you to make up compound values (for example directory paths) that use
2087
more than one default value. It also means it's possible to create circular
2088
references. If there are any circular references which would cause an infinite
2089
interpolation loop, an ``InterpolationLoopError`` is raised.
2091
Both of these errors are subclasses of ``InterpolationError``, which is a
2092
subclass of ``ConfigObjError``.
2094
String interpolation and validation don't play well together. This is because
2095
validation overwrites values - and so may erase the interpolation references.
2096
See `Validation and Interpolation`_. (This can only happen if validation
2097
has to *change* the value).
2099
New in ConfigObj 4.7.0: String interpolation is now done in members of list
2103
String Interpolation and List Values
2104
------------------------------------
2106
Since version 4.7 string interpolation is done on string members of list values.
2107
If interpolation changes any members of the list then what you get back is a
2108
*copy* of the list rather than the original list.
2110
This makes fetching list values slightly slower when interpolation is on, it
2111
also means that if you mutate the list changes won't be reflected in the
2114
.. code-block:: python
2117
>>> c['foo'] = 'boo'
2118
>>> c['bar'] = ['%(foo)s']
2121
>>> c['bar'].append('fish')
2125
Instead of mutating the list you must create a new list and reassign it.
2131
Any line that starts with a '#', possibly preceded by whitespace, is a comment.
2133
If a config file starts with comments then these are preserved as the
2136
If a config file ends with comments then these are preserved as the
2139
Every key or section marker may have lines of comments immediately above it.
2140
These are saved as the ``comments`` attribute of the section. Each member is a
2143
You can also have a comment inline with a value. These are saved as the
2144
``inline_comments`` attribute of the section, with one entry per member of the
2147
Subsections (section markers in the config file) can also have comments.
2149
See `Section Attributes`_ for more on these attributes.
2151
These comments are all written back out by the ``write`` method.
2157
.. code-block:: python
2159
flatten_errors(cfg, res)
2161
Validation_ is a powerful way of checking that the values supplied by the user
2164
The validate_ method returns a results dictionary that represents pass or fail
2165
for each value. This doesn't give you any information about *why* the check
2168
``flatten_errors`` is an example function that turns a results dictionary into
2169
a flat list, that only contains values that *failed*.
2171
``cfg`` is the ConfigObj instance being checked, ``res`` is the results
2172
dictionary returned by ``validate``.
2174
It returns a list of keys that failed. Each member of the list is a tuple::
2176
([list of sections...], key, result)
2178
If ``validate`` was called with ``preserve_errors=False`` (the default)
2179
then ``result`` will always be ``False``.
2181
*list of sections* is a flattened list of sections that the key was found
2184
If the section was missing then key will be ``None``.
2186
If the value (or section) was missing then ``result`` will be ``False``.
2188
If ``validate`` was called with ``preserve_errors=True`` and a value
2189
was present, but failed the check, then ``result`` will be the exception
2190
object returned. You can use this as a string that describes the failure.
2194
*The value "3" is of the wrong type*.
2200
The output from ``flatten_errors`` is a list of tuples.
2202
Here is an example of how you could present this information to the user.
2205
.. code-block:: python
2207
vtor = validate.Validator()
2208
# ini is your config file - cs is the configspec
2209
cfg = ConfigObj(ini, configspec=cs)
2210
res = cfg.validate(vtor, preserve_errors=True)
2211
for entry in flatten_errors(cfg, res):
2212
# each entry is a tuple
2213
section_list, key, error = entry
2215
section_list.append(key)
2217
section_list.append('[missing section]')
2218
section_string = ', '.join(section_list)
2220
error = 'Missing value or section.'
2221
print section_string, ' = ', error
2228
.. code-block:: python
2230
get_extra_values(conf)
2232
New in ConfigObj 4.7.0.
2234
Find all the values and sections not in the configspec from a validated
2237
``get_extra_values`` returns a list of tuples where each tuple represents
2238
either an extra section, or an extra value.
2240
The tuples contain two values, a tuple representing the section the value
2241
is in and the name of the extra values. For extra values in the top level
2242
section the first member will be an empty tuple. For values in the 'foo'
2243
section the first member will be ``('foo',)``. For members in the 'bar'
2244
subsection of the 'foo' section the first member will be ``('foo', 'bar')``.
2246
Extra sections will only have one entry. Values and subsections inside
2247
an extra section aren't listed separately.
2249
NOTE: If you call ``get_extra_values`` on a ConfigObj instance that hasn't
2250
been validated it will return an empty list.
2256
The output from ``get_extra_values`` is a list of tuples.
2258
Here is an example of how you could present this information to the user.
2260
.. code-block:: python
2262
vtor = validate.Validator()
2263
# ini is your config file - cs is the configspec
2264
cfg = ConfigObj(ini, configspec=cs)
2265
cfg.validate(vtor, preserve_errors=True)
2267
for sections, name in get_extra_values(cfg):
2269
# this code gets the extra values themselves
2271
for section in sections:
2272
the_section = cfg[section]
2274
# the_value may be a section or a value
2275
the_value = the_section[name]
2277
section_or_value = 'value
2278
if isinstance(the_value, dict):
2279
# Sections are subclasses of dict
2280
section_or_value = 'section'
2282
section_string = ', '.join(sections) or "top level"
2283
print 'Extra entry in section: %s. Entry %r is a %s' % (section_string, name, section_or_value)
2290
ConfigObj 4 is written by (and copyright) `Michael Foord`_ and
2293
Particularly thanks to Nicola Larosa for help on the config file spec, the
2294
validation system and the doctests.
2296
*validate.py* was originally written by Michael Foord and Mark Andrews.
2298
Thanks to many others for input, patches and bugfixes.
2305
ConfigObj, and related files, are licensed under the BSD license. This is a
2306
very unrestrictive license, but it comes with the usual disclaimer. This is
2307
free software: test it, break it, just don't blame us if it eats your data !
2308
Of course if it does, let us know and we'll fix the problem so it doesn't
2309
happen to anyone else::
2311
Copyright (c) 2004 - 2010, Michael Foord & Nicola Larosa
2312
All rights reserved.
2314
Redistribution and use in source and binary forms, with or without
2315
modification, are permitted provided that the following conditions are
2319
* Redistributions of source code must retain the above copyright
2320
notice, this list of conditions and the following disclaimer.
2322
* Redistributions in binary form must reproduce the above
2323
copyright notice, this list of conditions and the following
2324
disclaimer in the documentation and/or other materials provided
2325
with the distribution.
2327
* Neither the name of Michael Foord nor Nicola Larosa
2328
may be used to endorse or promote products derived from this
2329
software without specific prior written permission.
2331
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
2332
"AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
2333
LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
2334
A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
2335
OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
2336
SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
2337
LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
2338
DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
2339
THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
2340
(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
2341
OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
2343
You should also be able to find a copy of this license at : `BSD License`_
2345
.. _BSD License: http://www.voidspace.org.uk/python/license.shtml
2351
Better support for configuration from multiple files, including tracking
2352
*where* the original file came from and writing changes to the correct
2355
Make ``newline`` a keyword argument (as well as an attribute) ?
2357
``UTF16`` encoded files, when returned as a list of lines, will have the
2358
BOM at the start of every line. Should this be removed from all but the
2361
Option to set warning type for unicode decode ? (Defaults to strict).
2363
A method to optionally remove uniform indentation from multiline values.
2364
(do as an example of using ``walk`` - along with string-escape)
2366
Should the results dictionary from validate be an ordered dictionary if
2367
`odict <http://www.voidspace.org.uk/python/odict.html>`_ is available ?
2369
Implement some of the sequence methods (which include slicing) from the
2372
Preserve line numbers of values (and possibly the original text of each value).
2380
Please file any bug reports to `Michael Foord`_ or the **ConfigObj**
2383
There is currently no way to specify the encoding of a configspec file.
2385
As a consequence of the changes to configspec handling in version 4.6.0, when
2386
you create a ConfigObj instance and provide a configspec, the configspec
2387
attribute is only set on the ConfigObj instance - it isn't set on the sections until you validate. You also can't set the configspec attribute to be a dictionary. This wasn't documented but did work previously.
2389
In order to fix the problem with hashes in configspecs I had to turn off the parsing of inline comments in configspecs. This will only affect you if you are using ``copy=True`` when validating and expecting inline comments to be copied from the configspec into the ConfigObj instance (all other comments will be copied as usual).
2391
If you *create* the configspec by passing in a ConfigObj instance (usual way is to pass in a filename or list of lines) then you should pass in ``_inspec=True`` to the constructor to allow hashes in values. This is the magic that switches off inline comment parsing.
2393
When using ``copy`` mode for validation, it won't copy ``DEFAULT``
2394
sections. This is so that you *can* use interpolation in configspec
2395
files. This is probably true even if interpolation is off in the
2398
You can't have a keyword with the same name as a section (in the same
2399
section). They are both dictionary keys - so they would overlap.
2401
ConfigObj doesn't quote and unquote values if ``list_values=False``.
2402
This means that leading or trailing whitespace in values will be lost when
2403
writing. (Unless you manually quote).
2405
Interpolation checks first the current section, then the 'DEFAULT' subsection
2406
of the current section, before moving on to the current section's parent and
2409
Does it matter that we don't support the ':' divider, which is supported
2410
by ``ConfigParser`` ?
2412
String interpolation and validation don't play well together. When
2413
validation changes type it sets the value. This will correctly fetch the
2414
value using interpolation - but then overwrite the interpolation reference.
2415
If the value is unchanged by validation (it's a string) - but other types
2422
This is an abbreviated changelog showing the major releases up to version 4.
2423
From version 4 it lists all releases and changes.
2426
2010/02/27 - Version 4.7.2
2427
--------------------------
2429
* BUGFIX: Restore Python 2.3 compatibility
2430
* BUGFIX: Members that were lists were being returned as copies due to interpolation
2431
introduced in 4.7. Lists are now only copies if interpolation changes a list
2433
* BUGFIX: ``pop`` now does interpolation in list values as well.
2434
* BUGFIX: where interpolation matches a section name rather than a value it is
2435
ignored instead of raising an exception on fetching the item.
2436
* BUGFIX: values that use interpolation to reference members that don't exist can
2438
* BUGFIX: Fix to avoid writing '\\r\\r\\n' on Windows when given a file opened in
2439
text write mode ('w').
2441
See `String Interpolation and List Values`_ for information about the problem with lists and interpolation.
2444
2010/02/06 - Version 4.7.1
2445
--------------------------
2447
* Fix bug in options deprecation warning added in 4.7.0
2450
2010/01/09 - Version 4.7.0
2451
--------------------------
2453
* Minimum supported version of Python is now 2.3
2454
* ~25% performance improvement thanks to Christian Heimes
2455
* String interpolation now works in list value members
2456
* After validation any additional entries not in the configspec are listed in
2457
the ``extra_values`` section member
2458
* Addition of the ``get_extra_values`` function for finding all extra values
2459
in a validated ConfigObj instance
2460
* Deprecated the use of the ``options`` dictionary in the ConfigObj constructor
2461
and added explicit keyword arguments instead. Use \*\*options if you want
2462
to initialise a ConfigObj instance from a dictionary
2463
* Constructing a ConfigObj from an existing ConfigObj instance now preserves
2464
the order of values and sections from the original instance in the new one
2465
* BUGFIX: Checks that failed validation would not populate ``default_values`` and
2466
``restore_default_value()`` wouldn't work for those entries
2467
* BUGFIX: clear() now clears 'defaults'
2468
* BUGFIX: empty values in list values were accidentally valid syntax. They now
2469
raise a ``ParseError``. e.g. "value = 1, , 2"
2470
* BUGFIX: Change to the result of a call to ``validate`` when ``preserve_errors``
2471
is True. Previously sections where *all* values failed validation would
2472
return False for the section rather than preserving the errors. False will
2473
now only be returned for a section if it is missing
2474
* Distribution includes version 1.0.1 of validate.py
2475
* Removed __revision__ and __docformat__
2478
2009/04/13 - Version 4.6.0
2479
--------------------------
2481
* Pickling of ConfigObj instances now supported (thanks to Christian Heimes)
2482
* Hashes in confgspecs are now allowed (see note below)
2483
* Replaced use of hasattr (which can swallow exceptions) with getattr
2484
* __many__ in configspecs can refer to scalars (ordinary values) as well as sections
2485
* You can use ___many___ (three underscores!) where you want to use __many__ as well
2486
* You can now have normal sections inside configspec sections that use __many__
2487
* You can now create an empty ConfigObj with a configspec, programmatically set values and then validate
2488
* A section that was supplied as a value (or vice-versa) in the actual config file would cause an exception during validation (the config file is still broken of course, but it is now handled gracefully)
2489
* Added ``as_list`` method
2490
* Removed the deprecated ``istrue``, ``encode`` and ``decode`` methods
2491
* Running test_configobj.py now also runs the doctests in the configobj module
2493
As a consequence of the changes to configspec handling, when you create a ConfigObj instance and provide
2494
a configspec, the configspec attribute is only set on the ConfigObj instance - it isn't set on the
2495
sections until you validate. You also can't set the configspec attribute to be a dictionary. This wasn't
2496
documented but did work previously.
2498
In order to fix the problem with hashes in configspecs I had to turn off the parsing of inline comments
2499
in configspecs. This will only affect you if you are using ``copy=True`` when validating and expecting
2500
inline comments to be copied from the configspec into the ConfigObj instance (all other comments will be
2503
If you *create* the configspec by passing in a ConfigObj instance (usual way is to pass in a filename or
2504
list of lines) then you should pass in ``_inspec=True`` to the constructor to allow hashes in values.
2505
This is the magic that switches off inline comment parsing.
2508
2008/06/27 - Version 4.5.3
2509
--------------------------
2511
BUGFIX: fixed a problem with ``copy=True`` when validating with configspecs that use
2512
``__many__`` sections.
2515
2008/02/05 - Version 4.5.2
2516
--------------------------
2518
Distribution updated to include version 0.3.2 of validate_. This means that
2519
``None`` as a default value in configspecs works.
2522
2008/02/05 - Version 4.5.1
2523
--------------------------
2525
Distribution updated to include version 0.3.1 of validate_. This means that
2526
Unicode configspecs now work.
2529
2008/02/05 - Version 4.5.0
2530
--------------------------
2532
ConfigObj will now guarantee that files will be written terminated with a
2535
ConfigObj will no longer attempt to import the ``validate`` module, until/unless
2536
you call ``ConfigObj.validate`` with ``preserve_errors=True``. This makes it
2539
New methods ``restore_default`` and ``restore_defaults``. ``restore_default``
2540
resets an entry to its default value (and returns that value). ``restore_defaults``
2541
resets all entries to their default value. It doesn't modify entries without a
2542
default value. You must have validated a ConfigObj (which populates the
2543
``default_values`` dictionary) before calling these methods.
2545
BUGFIX: Proper quoting of keys, values and list values that contain hashes
2546
(when writing). When ``list_values=False``, values containing hashes are
2549
Added the ``reload`` method. This reloads a ConfigObj from file. If the filename
2550
attribute is not set then a ``ReloadError`` (a new exception inheriting from
2551
``IOError``) is raised.
2553
BUGFIX: Files are read in with 'rb' mode, so that native/non-native line endings work!
2555
Minor efficiency improvement in ``unrepr`` mode.
2557
Added missing docstrings for some overidden dictionary methods.
2559
Added the ``reset`` method. This restores a ConfigObj to a freshly created state.
2561
Removed old CHANGELOG file.
2564
2007/02/04 - Version 4.4.0
2565
--------------------------
2567
Official release of 4.4.0
2570
2006/12/17 - Version 4.3.3-alpha4
2571
---------------------------------
2575
Allowed arbitrary indentation in the ``indent_type`` parameter, removed the
2576
``NUM_INDENT_SPACES`` and ``MAX_INTERPOL_DEPTH`` (a leftover) constants,
2577
added indentation tests (including another docutils workaround, sigh), updated
2582
Made the import of ``compiler`` conditional so that ``ConfigObj`` can be used
2583
with `IronPython <http://www.codeplex.com/IronPython>`_.
2586
2006/12/17 - Version 4.3.3-alpha3
2587
---------------------------------
2591
Added a missing ``self.`` in the _handle_comment method and a related test,
2592
per Sourceforge bug #1523975.
2595
2006/12/09 - Version 4.3.3-alpha2
2596
---------------------------------
2600
Changed interpolation search strategy, based on this patch by Robin Munn:
2601
http://sourceforge.net/mailarchive/message.php?msg_id=17125993
2604
2006/12/09 - Version 4.3.3-alpha1
2605
---------------------------------
2609
Added Template-style interpolation, with tests, based on this patch by
2610
Robin Munn: http://sourceforge.net/mailarchive/message.php?msg_id=17125991
2611
(awful archives, bad Sourceforge, bad).
2614
2006/06/04 - Version 4.3.2
2615
--------------------------
2617
Changed error handling, if parsing finds a single error then that error will
2618
be re-raised. That error will still have an ``errors`` and a ``config``
2621
Fixed bug where '\\n' terminated files could be truncated.
2623
Bugfix in ``unrepr`` mode, it couldn't handle '#' in values. (Thanks to
2624
Philippe Normand for the report.)
2626
As a consequence of this fix, ConfigObj doesn't now keep inline comments in
2627
``unrepr`` mode. This is because the parser in the `compiler package`_
2628
doesn't keep comments.
2630
Error messages are now more useful. They tell you the number of parsing errors
2631
and the line number of the first error. (In the case of multiple errors.)
2633
Line numbers in exceptions now start at 1, not 0.
2635
Errors in ``unrepr`` mode are now handled the same way as in the normal mode.
2636
The errors stored will be an ``UnreprError``.
2639
2006/04/29 - Version 4.3.1
2640
--------------------------
2642
Added ``validate.py`` back into ``configobj.zip``. (Thanks to Stewart
2645
Updated to `validate.py`_ 0.2.2.
2647
Preserve tuples when calling the ``dict`` method. (Thanks to Gustavo Niemeyer.)
2649
Changed ``__repr__`` to return a string that contains ``ConfigObj({ ... })``.
2651
Change so that an options dictionary isn't modified by passing it to ConfigObj.
2652
(Thanks to Artarious.)
2654
Added ability to handle negative integers in ``unrepr``. (Thanks to Kevin
2658
2006/03/24 - Version 4.3.0
2659
--------------------------
2661
Moved the tests and the CHANGELOG (etc) into a separate file. This has reduced
2662
the size of ``configobj.py`` by about 40%.
2664
Added the ``unrepr`` mode to reading and writing config files. Thanks to Kevin
2665
Dangoor for this suggestion.
2667
Empty values are now valid syntax. They are read as an empty string ``''``.
2668
(``key =``, or ``key = # comment``.)
2670
``validate`` now honours the order of the configspec.
2672
Added the ``copy`` mode to validate. Thanks to Louis Cordier for this
2675
Fixed bug where files written on windows could be given ``'\r\r\n'`` line
2678
Fixed bug where last occurring comment line could be interpreted as the
2679
final comment if the last line isn't terminated.
2681
Fixed bug where nested list values would be flattened when ``write`` is
2682
called. Now sub-lists have a string representation written instead.
2684
Deprecated ``encode`` and ``decode`` methods instead.
2686
You can now pass in a ConfigObj instance as a configspec (remember to read
2687
the configspec file using ``list_values=False``).
2689
Sorted footnotes in the docs.
2692
2006/02/16 - Version 4.2.0
2693
--------------------------
2695
Removed ``BOM_UTF8`` from ``__all__``.
2697
The ``BOM`` attribute has become a boolean. (Defaults to ``False``.) It is
2698
*only* ``True`` for the ``UTF16/UTF8`` encodings.
2700
File like objects no longer need a ``seek`` attribute.
2702
Full unicode support added. New options/attributes ``encoding``,
2703
``default_encoding``.
2705
ConfigObj no longer keeps a reference to file like objects. Instead the
2706
``write`` method takes a file like object as an optional argument. (Which
2707
will be used in preference of the ``filename`` attribute if that exists as
2710
utf16 files decoded to unicode.
2712
If ``BOM`` is ``True``, but no encoding specified, then the utf8 BOM is
2713
written out at the start of the file. (It will normally only be ``True`` if
2714
the utf8 BOM was found when the file was read.)
2716
Thanks to Aaron Bentley for help and testing on the unicode issues.
2718
File paths are *not* converted to absolute paths, relative paths will
2719
remain relative as the ``filename`` attribute.
2721
Fixed bug where ``final_comment`` wasn't returned if ``write`` is returning
2724
Deprecated ``istrue``, replaced it with ``as_bool``.
2726
Added ``as_int`` and ``as_float``.
2729
2005/12/14 - Version 4.1.0
2730
--------------------------
2732
Added ``merge``, a recursive update.
2734
Added ``preserve_errors`` to ``validate`` and the ``flatten_errors``
2737
Thanks to Matthew Brett for suggestions and helping me iron out bugs.
2739
Fixed bug where a config file is *all* comment, the comment will now be
2740
``initial_comment`` rather than ``final_comment``.
2742
Validation no longer done on the 'DEFAULT' section (only in the root level).
2743
This allows interpolation in configspecs.
2745
Also use the new list syntax in validate_ 0.2.1. (For configspecs).
2748
2005/12/02 - Version 4.0.2
2749
--------------------------
2751
Fixed bug in ``create_empty``. Thanks to Paul Jimenez for the report.
2754
2005/11/05 - Version 4.0.1
2755
--------------------------
2757
Fixed bug in ``Section.walk`` when transforming names as well as values.
2759
Added the ``istrue`` method. (Fetches the boolean equivalent of a string
2762
Fixed ``list_values=False`` - they are now only quoted/unquoted if they
2763
are multiline values.
2765
List values are written as ``item, item`` rather than ``item,item``.
2768
2005/10/17 - Version 4.0.0
2769
--------------------------
2771
**ConfigObj 4.0.0 Final**
2773
Fixed bug in ``setdefault``. When creating a new section with setdefault the
2774
reference returned would be to the dictionary passed in *not* to the new
2775
section. Bug fixed and behaviour documented.
2777
Obscure typo/bug fixed in ``write``. Wouldn't have affected anyone though.
2780
2005/09/09 - Version 4.0.0 beta 5
2781
---------------------------------
2783
Removed ``PositionError``.
2785
Allowed quotes around keys as documented.
2787
Fixed bug with commas in comments. (matched as a list value)
2790
2005/09/07 - Version 4.0.0 beta 4
2791
---------------------------------
2793
Fixed bug in ``__delitem__``. Deleting an item no longer deletes the
2794
``inline_comments`` attribute.
2796
Fixed bug in initialising ConfigObj from a ConfigObj.
2798
Changed the mailing list address.
2801
2005/08/28 - Version 4.0.0 beta 3
2802
---------------------------------
2804
Interpolation is switched off before writing out files.
2806
Fixed bug in handling ``StringIO`` instances. (Thanks to report from
2809
Moved the doctests from the ``__init__`` method to a separate function.
2810
(For the sake of IDE calltips).
2813
2005/08/25 - Version 4.0.0 beta 2
2814
---------------------------------
2816
Amendments to *validate.py*.
2818
First public release.
2821
2005/08/21 - Version 4.0.0 beta 1
2822
---------------------------------
2824
Reads nested subsections to any depth.
2828
Simplified options and methods.
2832
Faster, smaller, and better parser.
2834
Validation greatly improved. Includes:
2840
Improved error handling.
2842
Plus lots of other improvements.
2845
2004/05/24 - Version 3.0.0
2846
--------------------------
2848
Several incompatible changes: another major overhaul and change. (Lots of
2849
improvements though).
2851
Added support for standard config files with sections. This has an entirely
2852
new interface: each section is a dictionary of values.
2854
Changed the update method to be called writein: update clashes with a dict
2857
Made various attributes keyword arguments, added several.
2859
Configspecs and orderlists have changed a great deal.
2861
Removed support for adding dictionaries: use update instead.
2863
Now subclasses a new class called caselessDict. This should add various
2864
dictionary methods that could have caused errors before.
2866
It also preserves the original casing of keywords when writing them back out.
2868
Comments are also saved using a ``caselessDict``.
2870
Using a non-string key will now raise a ``TypeError`` rather than converting
2873
Added an exceptions keyword for *much* better handling of errors.
2875
Made ``creatempty=False`` the default.
2877
Now checks indict *and* any keyword args. Keyword args take precedence over
2880
``' ', ':', '=', ','`` and ``'\t'`` are now all valid dividers where the
2881
keyword is unquoted.
2883
ConfigObj now does no type checking against configspec when you set items.
2885
delete and add methods removed (they were unnecessary).
2887
Docs rewritten to include all this gumph and more; actually ConfigObj is
2888
*really* easy to use.
2890
Support for stdout was removed.
2892
A few new methods added.
2894
Charmap is now incorporated into ConfigObj.
2897
2004/03/14 - Version 2.0.0 beta
2898
-------------------------------
2900
Re-written it to subclass dict. My first forays into inheritance and operator
2903
The config object now behaves like a dictionary.
2905
I've completely broken the interface, but I don't think anyone was really
2908
This new version is much more 'classy'.
2910
It will also read straight from/to a filename and completely parse a config
2911
file without you *having* to supply a config spec.
2913
Uses listparse, so can handle nested list items as values.
2915
No longer has getval and setval methods: use normal dictionary methods, or add
2919
2004/01/29 - Version 1.0.5
2920
--------------------------
2922
Version 1.0.5 has a couple of bugfixes as well as a couple of useful additions
2923
over previous versions.
2925
Since 1.0.0 the buildconfig function has been moved into this distribution,
2926
and the methods reset, verify, getval and setval have been added.
2928
A couple of bugs have been fixed.
2934
ConfigObj originated in a set of functions for reading config files in the
2935
`atlantibots <http://www.voidspace.org.uk/atlantibots/>`_ project. The original
2936
functions were written by Rob McNeur.
2945
.. [#] And if you discover any bugs, let us know. We'll fix them quickly.
2947
.. [#] If you specify a filename that doesn't exist, ConfigObj will assume you
2948
are creating a new one. See the *create_empty* and *file_error* options.
2950
.. [#] They can be byte strings (*ordinary* strings) or Unicode.
2952
.. [#] Except we don't support the RFC822 style line continuations, nor ':' as
2955
.. [#] This is a change in ConfigObj 4.2.0. Note that ConfigObj doesn't call
2956
the seek method of any file like object you pass in. You may want to call
2957
``file_object.seek(0)`` yourself, first.
2959
.. [#] A side effect of this is that it enables you to copy a ConfigObj :
2961
.. code-block:: python
2963
# only copies members
2964
# not attributes/comments
2965
config2 = ConfigObj(config1)
2967
Since ConfigObj 4.7.0 the order of members and sections will be
2968
preserved when copying a ConfigObj instance.
2970
.. [#] Other than lists of strings.
2972
.. [#] The exception is if it detects a ``UTF16`` encoded file which it
2973
must decode before parsing.
2975
.. [#] The method signature shows that this method takes
2976
two arguments. The second is the section to be written. This is because the
2977
``write`` method is called recursively.
2979
.. [#] The dict method doesn't actually use the deepcopy mechanism. This means
2980
if you add nested lists (etc) to your ConfigObj, then the dictionary
2981
returned by dict may contain some references. For all *normal* ConfigObjs
2982
it will return a deepcopy.
2984
.. [#] Passing ``(section, key)`` rather than ``(value, key)`` allows you to
2985
change the value by setting ``section[key] = newval``. It also gives you
2986
access to the *rename* method of the section.
2988
.. [#] Minimum required version of *validate.py* 0.2.0 .
2990
.. _listquote: http://www.voidspace.org.uk/python/modules.shtml#listquote
2991
.. _Michael Foord: http://www.voidspace.org.uk/python/weblog/index.shtml
2992
.. _Nicola Larosa: http://www.teknico.net