A library that parses text plists from OpenStep and GNUstep as well as .strings
files. Has a plistlib-style API.
Uses a Packrat parser from tatsu, because I am too lazy to manually write a recursive descent thing.
Text plists are Property lists written in a
human-readable, textual form. They roamed the Earth in the times of OpenStep, and is
generally better for humans than the new XML format. This form, NSPropertyListOpenStepFormat
, only supports strings, hexdata, arrays, and dictionaries.
The GNUstep project created an extension based on the old ASCII format called NSPropertyListGNUstepFormat
. This format adds support for integers, floats, and dates, making it as expressive as the XML and binary formats.
The format of textual plists are quite easy to grasp:
{
/* Strings can be unquoted, or quoted for C-style escapes */
"loremIpsum" = "A story about the good, \n the bad \u0000, and the ugly\x2E";
foo = bar;
hexdata = <deadbeef>;
int = <*I3>;
date = <*D2006-01-02 15:04:05 -0700>;
// collection objects
array = (1, 2, 3);
object = {
a = (1, 2, 3);
d = ();
e = {};
f = <>;
};
}
.strings
files are similar to OpenStep plist dictionaries, except that the braces are omitted. By convention all values are strings, and the = value
part can be omitted for a null or empty value.
This library accepts the following extensions to textual plists:
- Custom encoding: historically, plists may be written in a wide variety of encodings. We default to UTF-8, but this is configurable.
- Trailing commas for arrays: everyone loves trailing commas. Seems to be in GNUstep and Apple.
- Values in collections are nullable. This means that all dictionaries can use the
.strings
extension ofkey;
and null elements of arrays can be denoted by commas. In Python they translate toNone
.- On a second thought, it does make more sense to use something like
<*N>
instead of the JavaScript-like handling of array gaps.
- On a second thought, it does make more sense to use something like
- UID. Unsigned values of the
plistlib.UID
classes are serialized in the dialect as<*U12345>
. In other dialects, they are encoded as{ "CF$UID" = <*I12345> }
. - The top level of dictionary is assumed as in
.strings
files, no matter how complex the inner data structures are. This makes for better config files.
The generation of these extension elements can be turned off by a dialect control.
MIT/Expat license or Python Software Foundation License.
- python3 -m text_plistlib.impl ./tests/oneval.plist