A parser for the KDL language.
Project description
KDL-py
A handwritten Python 3.7+ implemenation of a parser for the KDL Document Language, fully compliant with KDL 1.0.0.
KDL is, as the name suggests, a document language, filling approximately the same niche as JSON/YAML/XML/etc to be a simple but powerful language for config files. It combines the best of several of these languages, while avoiding their pitfalls: more general than JSON and more powerful than XML, while avoiding the verbosity of XML or the explosive complexity of YAML.
kdl-py, in particular, is focused on ease-of-use, supporting things like date/times, ip addresses, urls, uuids, regexes, and binary data directly in your KDL document (via powerful but simple tagged values), and parsing them into native Python types automatically, or doing the reverse and letting you build KDL document trees with these values directly, and automatically and safely serializing them into KDL text for later parsing!
You can customize parsing and serialization further for your application very easily,
by providing node or value "converters"
to turn plain KDL values or nodes
into application-specific types,
and then turning them back into KDL text automatically
merely by adding a .to_kdl()
method to your classes.
Installing
pipx install kdl-py
When installed, a kdlreformat
command-line program is also made available,
which can canonicalize a KDL document. See below for options.
Using
The kdl.parse(str, parseConfig|None)
function parses, you guessed it, a string of KDL into a KDL document object:
import kdl
>>> import kdl
>>> doc = kdl.parse('''
... node_name "arg" {
... child_node foo=1 bar=true
... }
... ''')
>>>
>>> doc
Document(
nodes=[
Node(
name='node',
args=['arg'],
nodes=[
Node(
name='child',
props=OrderedDict([
('foo', 1.0),
('bar', True)
])
)
]
)
]
)
You can also create a kdl.Parser()
object and call its .parse()
method; Parser
objects can set up parsing and printing options that'll apply by default. See below for how to configure parsing options.
Either way, you'll get back a kdl.Document
object, which is fully mutable. By default, untagged KDL values are represented with native Python objects.
>>> doc.nodes[0].nodes[0].props["foo"] = 2
>>>
>>> print(doc)
node_name "arg" {
child_node foo=2 bar=true
}
Stringifying a kdl.Document
object will produce a valid KDL document back. You can also call doc.print(printConfig|None)
to customize the printing with a PrintConfig
object, described below. See below for how to configure printing options.
Inserting Native Values
kdl-py allows a number of native Python objects to be used directly in KDL documents by default, and allows you to customize your own objects for use.
kdl-py automatically recognizes and correctly serializes the following objects:
bool
: as untaggedtrue
orfalse
None
: as untaggednull
int
,float
: as untagged decimal numberstr
: as untagged stringbytes
: as(base64)
-tagged stringdecimal.Decimal
: as(decimal)
-tagged stringdatetime
,date
, andtime
: as(date-time)
,(date)
, or(time)
-tagged stringsipaddress.IPv4Address
andipaddress.IPv6Address
: as(ipv4)
or(ipv6)
-tagged stringsurllib.parse.ParseResult
(the result of callingurllib.parse.urlparse()
): as(url)
-tagged stringuuid.UUID
: as(uuid)
-tagged stringre.Pattern
(the result of callingre.compile()
): as(regex)
-tagged raw string
All of the tags used above are reserved and predefined by the KDL specification.
In addition, any value with a .to_kdl()
method
can be used in a kdl-py document.
The method will be called when the document is stringified,
and must return one of the kdl-py types,
or any of the native types defined above.
(For parsing KDL into these native types,
or your own types,
see the ParseConfig
section, below.)
Customizing Parsing
Parsing can be controlled via a kdl.ParseConfig
object,
which can be provided in three ways.
In order of importance:
- Passing a
ParseConfig
object tokdl.parse(str, ParseConfig|None)
orparser.parse(str, ParseConfig|None)
(if you've constructed akdl.Parser
). - Creating a
kdl.Parser(parseConfig|None, printConfig|None)
, which automatically applies it to its.parse()
method if not overriden. - Fiddling with the
kdl.parsing.defaults
object, which is used if nothing else provides a config.
A ParseConfig
object has the following properties:
-
nativeUntaggedValues: bool = True
Controls whether the parser produces native Python objects (
str
,int
,float
,bool
,None
) when parsing untagged values (those without a(foo)
prefix), or always produces kdl-py objects (such askdl.String
,kdl.Decimal
, etc). -
nativeTaggedValues: bool = True
Controls whether the parser produces native Python objects when parsing tagged values, for some of KDL's predefined tags:
i8
,i16
,i32
,i64
,u8
,u16
,u32
,u64
on numbers: Checks that the value is in the specified range, then converts it to anint
. (It will serialize back out as an ordinary untagged number.)f32
,f64
on numbers: Converts it to afloat
. (It will serialize back out as an ordinary untagged number.)decimal64
,decimal128
on numbers, anddecimal
on strings: Converts it to adecimal.Decimal
object. (Always reserializes to a(decimal)
-tagged string.)date-time
,date
,time
on strings: Converts to adatetime
,time
, ordate
object.ipv4
,ipv6
on strings: Converts it to anipaddress.IPv4Address
oripaddress.IPv6Address
object.url
on strings: Converts it to aurllib.parse.ParseResult
tuple.uuid
on strings: Converts it to auuid.UUID
object.regex
on strings: Converts it to are.Pattern
object. (It will serialize back out as a raw string.)base64
on strings: Converts it to abytes
object.
-
valueConverters: Dict[ValueKey, Callable] = {}
A dictionary of ValueKey->converter functions, letting you parse values (like
(date)"2021-01-01"
) into whatever types you'd like.Whenever the parser encounters a value, it will attempt to find an entry in this dict whose
ValueKey
matches the value. If it succeeds, it will call the associated converter function with two arguments: the fully-constructed kdl-py object, and aParseFragment
object giving you access to the precise characters parsed from the document (for the value; the tag, if any, won't be included, as you can get it from the object itself). Whatever you return will be inserted into the document instead of the originally-parsed value.You can produce KDL values (such as parsing
(hex)"0x12.e5"
into akdl.Decimal
, since KDL doesn't support fractional hex values), or produce any other Python type. If you return a non-KDL type, you probably want to ensure it has a.to_kdl()
method (or is one of the supported built-in types), so it can be serialized back into a KDL document.Note that only one conversion can happen to a given value. Your converters are checked in the dictionary's iteration order, then if none of them succeeded it will attempt to run the
nativeUntaggedValues
ornativeTaggedValues
behaviors, if they're turned on. If a converter returnsNotImplemented
, it will continue looking for a matching converter. -
nodeConverters: Dict[NodeKey, Callable] = {}
Similar to
valueConverters
, except it usesNodeKey
s, and it convertskdl.Node
s instead.There is no native conversion of nodes; if none of your converters successfully run, the node will be inserted as-is.
ParseFragment
kdl.ParseFragment
is passed to your custom converters,
specified in kdl.ParseConfig.tags
,
giving you direct access to the input characters
before any additional processing was done on them.
This is useful, for example,
to handle numeric types
that might have lost precision in the normal parse.
It exposes a .fragment
property,
containing the raw text of the value
(after the tag, if any).
It also exposes a .error(str)
method,
which takes a custom error message
and returns a kdl.ParseError
with the ParseFragment
's location already built in,
ready for you to raise
.
This should be used if your conversion fails for any reason,
so your errors look the same as native parse errors.
Customizing Printing
Like parsing, printing a kdl-py Document
back to a KDL string can be controlled by a kdl.PrintConfig
object,
which can be provided in three ways.
In order of importance:
- Passing a
PrintConfig
object todoc.print(PrintConfig|None)
. - Setting
doc.printConfig
to aPrintConfig
. (This is done automatically for any documents produced by aParser
, if you pass theprintConfig
option to the constructor.) - Fiddling with the
kdl.printing.defaults
object, which is used if nothing else provides a config.
A PrintConfig
object has the following properties:
-
indent: str = "\t"
The string used for each indent level. Defaults to tabs, but can be set to a sequence of spaces if desired (or anything else).
-
semicolons: bool = False
Whether or not nodes are ended with semicolons. (The printer always ends nodes with a newline anyway, so this is purely a stylistic choice.)
-
printNullArgs: bool = True
When
False
, automatically skips over any "null"/None
arguments. This will corrupt documents that use the "null" keyword intentionally, but can be useful if you'd prefer to use aNone
value as a signal that the argument has been removed. -
printNullProps: bool = True
Identical to
printNullArgs
, but applies to properties rather than arguments. -
respectStringType: bool = True
When
True
, the printer will output strings as the same type they were in the input, either raw (r#"foo"#
) or normal ("foo"
). WhenFalse
, the printer always outputs normal strings.Note that this only has an effect on
kdl.String
andkdl.RawString
objects; if the document contains Pythonstr
objects, they will always output as normal strings. -
respectRadix: bool = True
Similar to
respectStringType
, whenTrue
the printer will output numbers as the radix they were in the input, like0x1b
for hex numbers. WhenFalse
, the printer always outputs decimal numbers.Again, this only has an effect on kdl-py objects; native Python numbers are printed as normal for Python.
-
exponent: str = "e"
What character to use for the exponent part of decimal numbers, when printed with scientific notation. Should only be set to "e" or "E".
Like the previous options, this only has an effect on kdl-py objects; native Python numbers are printed as normal for Python.
Full API Reference
-
kdl.parse(str, config: kdl.ParseConfig|None) -> kdl.Document
-
kdl.Parser(parseConfig: kdl.ParseConfig|None, printConfig: kdl.PrintConfig|None)
parser.parse(str, config: kdl.ParseConfig|None) -> kdl.Document
parser.print(config: kdl.PrintConfig|None) -> str
-
kdl.Document(nodes: list[kdl.Node]?, printConfig: kdl.PrintConfig|None)
doc.print(PrintConfig|None) -> str
doc[NodeKey] -> Node
returns the first child node matching theNodeKey
. Raises aKeyError
if nothing matches theNodeKey
, similar to adict
.doc.get(NodeKey, default: T = None) -> kdl.Node | T
returns the first child node matching theNodeKey
. Returns the default value if nothing matches.doc.getAll(NodeKey) -> Iterable[kdl.Node]
returns all child nodes matching theNodeKey
-
kdl.Node(name: str, tag: str|None, args: list[Any]?, props: dict[str, Any]?, nodes: list[kdl.Node]?)
node[NodeKey] -> Node
returns the first child node matching theNodeKey
. Raises aKeyError
if nothing matches theNodeKey
, similar to adict
.node.get(NodeKey, default: T = None) -> kdl.Node | T
returns the first child node matching theNodeKey
. Returns the default value if nothing matches.node.getAll(NodeKey) -> Iterable[kdl.Node]
returns all child nodes matching theNodeKey
node.getProps(ValueKey) -> Iterable[tuple[str, Any]]
returns an iterator of(name, value)
pairs for the properties whose value matches theValueKey
(Note: for this purpose, non-KDL values, such asint
, have no tag, and can only meaningfully be tested via aTypeKey
.)node.getArgs(ValueKey) -> Iterable[Any]
returns an iterator of the arguments that match theValueKey
(Same disclaimer as.getProps()
.)node.matchesKey(NodeKey) -> bool
returns whether the node matches theNodeKey
-
kdl.Value
‡val.matchesKey(ValueKey) -> bool
returns whether the value matches theValueKey
-
kdl.Binary(value: int, tag: str|None)
-
kdl.Octal(value: int, tag: str|None)
-
kdl.Decimal(mantissa: int|float, exponent: int|None, tag: str|None)
dec.value
: readonly,mantissa * (10**exponent)
-
kdl.Hex(value: int, tag: str|None)
-
kdl.Bool(value: bool, tag: str|None)
-
kdl.Null(tag: str|None)
null.value
: readonly, alwaysNone
-
kdl.RawString(value: str, tag: str|None)
-
kdl.String(value: str, tag: str|None)
-
kdl.ExactValue(chars: str, tag: str|None)
† -
kdl.Numberish
,kdl.Stringish
‡ -
kdl.ParseConfig(...)
see above for optionskdl.parsing.defaults
: defaultParseConfig
-
kdl.PrintConfig(...)
see above for optionskdl.printing.defaults
: defaultPrintConfig
-
kdl.ParseError
: thrown for all parsing errorserror.msg: str
: hopefully informativeerror.line: int
: 1-indexederror.col: int
: 1-indexed
-
kdl.ParseFragment
: passed to converter functionspf.fragment
: slice from the source stringpf.error(msg: str)
returns akdl.ParseError
with error location set properly already
-
kdl.nodeMatchesKey(val: Any, key: kdl.NodeKey) -> bool
-
kdl.valueMatchesKey(val: Any, key: kdl.ValueKey) -> bool
-
kdl.tagMatchesKey(val: str|None, key: kdl.TagKey) -> bool
-
kdl.nameMatchesKey(val: str|None, key: kdl.NameKey) -> bool
-
kdl.typeMatchesKey(val: str|None, key: kdl.TypeKey) -> bool
- Functions implementing the tag/name/type matching
used by the
node.matchesKey()
andvalue.matchesKey()
methods, in case you want to use the same filtering yourself.
- Functions implementing the tag/name/type matching
used by the
† Not produced by the parser.
Can be returned by a user's .to_kdl()
method
if they want to produce a value precisely in a particular syntax,
in a way that the built-in kdl-py classes don't.
‡ Not produced by the parser.
These are abstract base classes to help in type testing:
Value
matches all eight value classes,
Numberish
matches all four numeric value classes,
and Stringish
matches both string value classes.
A few type aliases also exist, used by the module in a few places and potentially useful for your code:
kdl.KDLAny
: aDocument
,Node
, or any of theValue
subtypeskdl.KDLValue
: any of theValue
subtypeskdl.KDLishValue
: aKDLValue
or one of the supported Python native types see "Inserting Native Types"kdl.ValueKey
,kdl.NodeKey
,kdl.TagKey
,kdl.NameKey
,kdl.TypeKey
: the types forValueKey
s andNodeKey
s (and their individual pieces)
These aliases only exist when typing.TYPE_CHECKING
is true,
so they're only useful for writing types;
they won't be visible at runtime.
NodeKey
A few data structures and functions take a NodeKey
to match against a node,
based on its name and/or tag.
A NodeKey
is either a NameKey
or a (TagKey, NameKey)
tuple,
matching against the node's name and/or tag.
NameKey
s and TagKey
s can be:
None
: When matched against a tag, only succeeds against aNone
tag (aka a tagless node likefoo
). When matched against a name, automatically succeeds, since nodes are guaranteed to have a name....
: Always succeeds. Use this when, say, you want to match against a particular tag, regardless of the nodename, like("my-tag", ...)
.- a
str
: Succeeds if the tag/name is that exact string. - a
re.Pattern
(a regex, such asre.compile(r".*foo")
: Succeeds if the regex matches the tag/name. Note that this uses.match()
semantics, automatically anchoring against the start of the string; prepend your regex with.*?
if you want it to match anywhere. - a function, taking a
str | None
and returning abool
: Succeeds if the function, when called with the tag/name, returns True.
So, for example,
doc.getAll("foo")
would return all nodes whose name is "foo",
regardless of tag.
doc.getAll(("my-tag", ...))
would return all nodes whose tag is "my-tag",
regardless of the node name.
doc.getAll(re.compile(r"-"))
would return all nodes whose name starts with "-".
Etc.
ValueKey
Similarly to NodeKey
, a few things take a ValueKey
to match against values,
based on its tag and/or type.
A ValueKey
is either a TagKey
or a (TagKey, TypeKey)
tuple,
matching against the value's tag and/or type.
TagKey
works identically to how it appears in NodeKey
.
TypeKey
is either ...
, which matches any type,
or the same sort of argument you'd pass as the second argument to isinstance()
.
For example,
a "foo"
key would match any values with the tag "foo", like (foo)1
or (foo)"bar"
.
A (..., kdl.Numberish)
will match any value that's a "number" type: a Hex, Octal, Binary, or Decimal, regardless of its tag.
Etc.
kdlreformat
The kdlreformat
command-line program is installed by default
when you install this module from pypi.
It can also be run manually from the kdlreformat.py
file
at the root of this repository
(or from the kdl.cli.cli()
function)
usage: kdlreformat [-h] [--indent INDENT] [--semicolons] [--radix]
[--no-radix] [--raw-strings] [--no-raw-strings]
[--exponent EXPONENT]
[infile] [outfile]
KDL parser/printer, letting you easily reformat KDL files into a canonical
representation.
positional arguments:
infile
outfile
optional arguments:
-h, --help show this help message and exit
--indent INDENT How many spaces for each level of indent. -1 indicates
to indent with tabs.
--semicolons Whether to end nodes with semicolons or not.
--radix Output numeric values in the radix used by the input.
(0x1a outputs as 0x1a)
--no-radix Convert all numeric arguments to decimal. (0x1a outputs
as 26)
--raw-strings Output string values in the string type used by the
input.
--no-raw-strings Convert all string arguments into plain strings.
--exponent EXPONENT What character to use ('e' or 'E') for indicating
exponents on scinot numbers.
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
File details
Details for the file kdl-py-1.2.0.tar.gz
.
File metadata
- Download URL: kdl-py-1.2.0.tar.gz
- Upload date:
- Size: 29.2 kB
- Tags: Source
- Uploaded using Trusted Publishing? No
- Uploaded via: twine/4.0.2 CPython/3.11.6
File hashes
Algorithm | Hash digest | |
---|---|---|
SHA256 | 63f3f46c6277dedadce44dcfc94672bf6e6bf3330b90122e0134c22857e29973 |
|
MD5 | f00cf488691078b565b78134847c5450 |
|
BLAKE2b-256 | a435caa43d276dd46efe73926e25024429fab25effca24ba0af3875ea5cdc362 |
File details
Details for the file kdl_py-1.2.0-py3-none-any.whl
.
File metadata
- Download URL: kdl_py-1.2.0-py3-none-any.whl
- Upload date:
- Size: 23.7 kB
- Tags: Python 3
- Uploaded using Trusted Publishing? No
- Uploaded via: twine/4.0.2 CPython/3.11.6
File hashes
Algorithm | Hash digest | |
---|---|---|
SHA256 | d11e556dbb226cccf666bb4ffe17c73e9d5bb08d29976649fa9292bac973d3e2 |
|
MD5 | 9908ee328c276afcafc424231c68ceb6 |
|
BLAKE2b-256 | b704766286ba77621edf37225c66ce0ef2bceead987c69439b572e85ad27980c |