DPDK patches and discussions
 help / color / mirror / Atom feed
From: "Juraj Linkeš" <juraj.linkes@pantheon.tech>
To: Luca Vizzarro <luca.vizzarro@arm.com>, dev@dpdk.org
Cc: Jeremy Spewock <jspewock@iol.unh.edu>,
	Paul Szczepanek <paul.szczepanek@arm.com>
Subject: Re: [PATCH v2 3/5] dts: add parsing utility module
Date: Tue, 4 Jun 2024 17:13:07 +0200	[thread overview]
Message-ID: <685b9f60-7b5d-407f-be26-69115e23a9d2@pantheon.tech> (raw)
In-Reply-To: <20240509112635.1170557-4-luca.vizzarro@arm.com>

There are mostly documentation and basically inconsequential minor comments.

On 9. 5. 2024 13:26, Luca Vizzarro wrote:
> Adds parsing text into a custom dataclass. It provides a new
> `TextParser` dataclass to be inherited. This implements the `parse`
> method, which combined with the parser functions, it can automatically
> parse the value for each field.
> 
> This new utility will facilitate and simplify the parsing of complex
> command outputs, while ensuring that the codebase does not get bloated
> and stays flexible.
> 
> Signed-off-by: Luca Vizzarro <luca.vizzarro@arm.com>
> Reviewed-by: Paul Szczepanek <paul.szczepanek@arm.com>
> ---
>   dts/framework/parser.py | 199 ++++++++++++++++++++++++++++++++++++++++
>   1 file changed, 199 insertions(+)
>   create mode 100644 dts/framework/parser.py
> 
> diff --git a/dts/framework/parser.py b/dts/framework/parser.py
> new file mode 100644
> index 0000000000..5b4acddead
> --- /dev/null
> +++ b/dts/framework/parser.py
> @@ -0,0 +1,199 @@
> +# SPDX-License-Identifier: BSD-3-Clause
> +# Copyright(c) 2024 Arm Limited
> +
> +r"""Parsing utility module.
> +
> +This module provides :class:`~TextParser` which can be used to model any dataclass to a block of
> +text.
> +
> +Usage example::

I'd like to see a high level explanation of what the key pieces of the 
parsing are and how they're tied to the implementation:
1. The text we're about to parse, passed to the instance of a subclass
2. What we're parsing, the fields of the subclass
3. How we're parsing the fields, the functions of TextParser

This could be part of the example or mentioned before the example or in 
the class. I had to study the code to understand the API, so that should 
be better documented.

> +..code:: python
> +
> +    from dataclasses import dataclass, field
> +    from enum import Enum
> +    from framework.parser import TextParser
> +
> +    class Colour(Enum):
> +        BLACK = 1
> +        WHITE = 2
> +
> +        @classmethod
> +        def from_str(cls, text: str):
> +            match text:
> +                case "black":
> +                    return cls.BLACK
> +                case "white":
> +                    return cls.WHITE
> +                case _:
> +                    return None # unsupported colour
> +
> +        @classmethod
> +        def make_parser(cls):
> +            # make a parser function that finds a match and
> +            # then makes it a Colour object through Colour.from_str
> +            return TextParser.compose(cls.from_str, TextParser.find(r"is a (\w+)"))
> +
> +    @dataclass
> +    class Animal(TextParser):
> +        kind: str = field(metadata=TextParser.find(r"is a \w+ (\w+)"))
> +        name: str = field(metadata=TextParser.find(r"^(\w+)"))
> +        colour: Colour = field(metadata=Colour.make_parser())
> +        age: int = field(metadata=TextParser.find_int(r"aged (\d+)"))
> +
> +    steph = Animal.parse("Stephanie is a white cat aged 10")
> +    print(steph) # Animal(kind='cat', name='Stephanie', colour=<Colour.WHITE: 2>, age=10)
> +"""
> +
> +import re
> +from abc import ABC
> +from dataclasses import MISSING, dataclass, fields
> +from functools import partial
> +from typing import Any, Callable, TypedDict, cast
> +
> +from typing_extensions import Self
> +
> +
> +class ParserFn(TypedDict):
> +    """Parser function in a dict compatible with the :func:`dataclasses.field` metadata param."""
> +
> +    #:
> +    TextParser_fn: Callable[[str], Any]
> +
> +
> +@dataclass
> +class TextParser(ABC):
> +    """Helper abstract dataclass that parses a text according to the fields' rules.
> +
> +    This class provides a selection of parser functions and a function to compose generic functions
> +    with parser functions. Parser functions are designed to be passed to the fields' metadata param
> +    to enable parsing.
> +    """

We should add that this class should be subclassed; basically what I've 
laid out above so that devs know how to actually use the class. I'm not 
sure which is the better place to put it, but probably here.

> +
> +    """============ BEGIN PARSER FUNCTIONS ============"""
> +
> +    @staticmethod
> +    def compose(f: Callable, parser_fn: ParserFn) -> ParserFn:
> +        """Makes a composite parser function.
> +
> +        The parser function is run and if a non-None value was returned, f is called with it.

Let's use `parser_fn` instead of "The parser function". Let's also put f 
into backticks. It's also peculiar that we're passing the functions in 
the reverse order - first the second one is applied and then the first 
one is applied.

> +        Otherwise the function returns early with None.
> +
> +        Metadata modifier for :func:`dataclasses.field`.

This sentence is all alone here and I don't understand what it's saying 
without more context.

> +
> +        Args:
> +            f: the function to apply to the parser's result

This now refers to just parser's results instead of parser functions's 
result, which is confusing. But let's also use `parser_fn` here.

Also, the arg descriptions should end with a dot. And start with a 
capital letter.

https://www.sphinx-doc.org/en/master/usage/extensions/example_google.html#example-google

> +            parser_fn: the dictionary storing the parser function
> +        """

The docstring is missing a Returns: section.

> +        g = parser_fn["TextParser_fn"]

This is a matter of preference, but I like more descriptive variable 
names, something like initial_func. And then I'd change f to 
subsequent_func, which would make the function signature a bit more 
readable.

> +
> +        def _composite_parser_fn(text: str) -> Any:
> +            intermediate_value = g(text)
> +            if intermediate_value is None:
> +                return None
> +            return f(intermediate_value)
> +
> +        return ParserFn(TextParser_fn=_composite_parser_fn)
> +
> +    @staticmethod
> +    def find(
> +        pattern: str | re.Pattern[str],
> +        flags: re.RegexFlag = re.RegexFlag(0),
> +        named: bool = False,
> +    ) -> ParserFn:
> +        """Makes a parser function that finds a regular expression match in the text.
> +
> +        If the pattern has capturing groups, it returns None if no match was found. If the pattern
> +        has only one capturing group and a match was found, its value is returned. If the pattern
> +        has no capturing groups then either True or False is returned if the pattern had a match or
> +        not.
> +

This description would be a better fit in a Returns: section. The body 
could explain the various scenarios in which the function can be used. 
At least I imagine there are various scenarios based on the different 
things the function returns, but I don't really want to go around 
digging in the code to verify that. :-)

Also a note: what does the function return if the pattern has capturing 
groups and a match is found?

> +        Metadata modifier for :func:`dataclasses.field`.
> +
> +        Args:
> +            pattern: the regular expression pattern
> +            flags: the regular expression flags. Not used if the given pattern is already compiled
> +            named: if set to True only the named capture groups will be returned as a dictionary
> +        """

This is also missing the Returns: sections.

> +        if isinstance(pattern, str):
> +            pattern = re.compile(pattern, flags)
> +
> +        def _find(text: str) -> Any:
> +            m = pattern.search(text)
> +            if m is None:
> +                return None if pattern.groups > 0 else False
> +
> +            if pattern.groups == 0:
> +                return True
> +
> +            if named:
> +                return m.groupdict()
> +
> +            matches = m.groups()
> +            if len(matches) == 1:
> +                return matches[0]
> +
> +            return matches
> +
> +        return ParserFn(TextParser_fn=_find)
> +
> +    @classmethod

Is there a reason why find_int() is a classmethod while the rest are 
staticmethods? Looks like it could also be a staticmethod.

> +    def find_int(
> +        cls,
> +        pattern: str | re.Pattern[str],
> +        flags: re.RegexFlag = re.RegexFlag(0),
> +        int_base: int = 0,
> +    ) -> ParserFn:
> +        """Makes a parser function that converts the match of :meth:`~find` to int.
> +
> +        This function is compatible only with a pattern containing one capturing group.
> +
> +        Metadata modifier for :func:`dataclasses.field`.
> +
> +        Args:
> +            pattern: the regular expression pattern
> +            flags: the regular expression flags
> +            int_base: the base of the number to convert from
> +        Raises:
> +            RuntimeError: if the pattern does not have exactly one capturing group
> +        """
> +        if isinstance(pattern, str):
> +            pattern = re.compile(pattern, flags)
> +
> +        if pattern.groups != 1:
> +            raise RuntimeError("only one capturing group is allowed with this parser function")
> +

Have you considered using a subclass of DTSError so that we can add a 
severity to this error? The severity is there to "rank" DTS errors from 
worst to least worst.

> +        return cls.compose(partial(int, base=int_base), cls.find(pattern))
> +
> +    """============ END PARSER FUNCTIONS ============"""
> +
> +    @classmethod
> +    def parse(cls, text: str) -> Self:

Would converting this into __init__(self, text: str) work? Sounds like 
we could just use "for field in fields(self)" and then setattr() to 
populate the variables.

> +        """Creates a new instance of the class from the given text.
> +
> +        A new class instance is created with all the fields that have a parser function in their
> +        metadata. Fields without one are ignored and are expected to have a default value, otherwise
> +        the class initialization will fail.
> +
> +        A field is populated with the value returned by its corresponding parser function.
> +
> +        Args:
> +            text: the text to parse
> +        Raises:
> +            RuntimeError: if the parser did not find a match and the field does not have a default
> +                          value or default factory.
> +        """
> +        fields_values = {}
> +        for field in fields(cls):
> +            parse = cast(ParserFn, field.metadata).get("TextParser_fn")
> +            if parse is None:
> +                continue
> +
> +            value = parse(text)
> +            if value is not None:
> +                fields_values[field.name] = value

If we convert the method into a constructor, we would just use setattr() 
here.

> +            elif field.default is MISSING and field.default_factory is MISSING:
> +                raise RuntimeError(
> +                    f"parser for field {field.name} returned None, but the field has no default"
> +                )
> +
> +        return cls(**fields_values)

  parent reply	other threads:[~2024-06-04 15:13 UTC|newest]

Thread overview: 95+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2024-04-12 11:11 [PATCH 0/5] dts: testpmd show port info/stats Luca Vizzarro
2024-04-12 11:11 ` [PATCH 1/5] dts: fix InteractiveShell command prompt filtering Luca Vizzarro
2024-04-16  8:40   ` Juraj Linkeš
2024-04-16 12:12     ` Luca Vizzarro
2024-04-17 13:06       ` Juraj Linkeš
2024-04-17 14:17         ` Luca Vizzarro
2024-04-18  6:31           ` Juraj Linkeš
2024-04-29 16:16             ` Jeremy Spewock
2024-04-12 11:11 ` [PATCH 2/5] dts: skip first line of send_command output Luca Vizzarro
2024-04-16  8:48   ` Juraj Linkeš
2024-04-16 12:15     ` Luca Vizzarro
2024-04-17 13:18       ` Juraj Linkeš
2024-04-29 15:18         ` Jeremy Spewock
2024-06-14 18:58           ` Nicholas Pratte
2024-04-12 11:11 ` [PATCH 3/5] dts: add parsing utility module Luca Vizzarro
2024-04-16  8:59   ` Juraj Linkeš
2024-04-16 12:16     ` Luca Vizzarro
2024-04-29 16:15   ` Jeremy Spewock
2024-04-30 10:49     ` Luca Vizzarro
2024-04-30 20:03       ` Jeremy Spewock
2024-04-12 11:11 ` [PATCH 4/5] dts: add `show port info` command to TestPmdShell Luca Vizzarro
2024-04-16  9:03   ` Juraj Linkeš
2024-04-16 12:24     ` Luca Vizzarro
2024-04-17 13:22       ` Juraj Linkeš
2024-04-17 14:25         ` Luca Vizzarro
2024-04-17 15:29           ` Luca Vizzarro
2024-04-18  6:41             ` Juraj Linkeš
2024-04-18 10:52               ` Luca Vizzarro
2024-04-12 11:11 ` [PATCH 5/5] dts: add `show port stats` " Luca Vizzarro
2024-04-16  9:04   ` Juraj Linkeš
2024-04-29 15:54   ` Jeremy Spewock
2024-04-30 10:51     ` Luca Vizzarro
2024-05-09 11:26 ` [PATCH v2 0/5] dts: testpmd show port info/stats Luca Vizzarro
2024-05-09 11:26   ` [PATCH v2 1/5] dts: fix InteractiveShell command prompt filtering Luca Vizzarro
2024-05-31 21:06     ` Jeremy Spewock
2024-06-04 13:57     ` Juraj Linkeš
2024-05-09 11:26   ` [PATCH v2 2/5] dts: skip first line of send command output Luca Vizzarro
2024-05-31 21:06     ` Jeremy Spewock
2024-06-04 13:58     ` Juraj Linkeš
2024-05-09 11:26   ` [PATCH v2 3/5] dts: add parsing utility module Luca Vizzarro
2024-05-31 21:06     ` Jeremy Spewock
2024-06-04 15:13     ` Juraj Linkeš [this message]
2024-06-05 10:35       ` Luca Vizzarro
2024-06-05 12:19         ` Juraj Linkeš
2024-06-05 13:00           ` Luca Vizzarro
2024-05-09 11:26   ` [PATCH v2 4/5] dts: add `show port info` command to TestPmdShell Luca Vizzarro
2024-05-31 21:06     ` Jeremy Spewock
2024-06-03  9:11       ` Luca Vizzarro
2024-06-04 15:40     ` Juraj Linkeš
2024-06-05 11:03       ` Luca Vizzarro
2024-05-09 11:26   ` [PATCH v2 5/5] dts: add `show port stats` " Luca Vizzarro
2024-05-20 14:26     ` Nicholas Pratte
2024-05-21 15:00       ` Luca Vizzarro
2024-05-31 21:07     ` Jeremy Spewock
2024-06-04 15:44     ` Juraj Linkeš
2024-06-05 11:04       ` Luca Vizzarro
2024-06-05 14:44 ` [PATCH v3 0/5] dts: testpmd show port info/stats Luca Vizzarro
2024-06-05 14:45   ` [PATCH v3 1/5] dts: fix InteractiveShell command prompt filtering Luca Vizzarro
2024-06-05 14:45   ` [PATCH v3 2/5] dts: skip first line of send command output Luca Vizzarro
2024-06-05 14:45   ` [PATCH v3 3/5] dts: add parsing utility module Luca Vizzarro
2024-06-05 14:45   ` [PATCH v3 4/5] dts: add `show port info` command to TestPmdShell Luca Vizzarro
2024-06-05 14:45   ` [PATCH v3 5/5] dts: add `show port stats` " Luca Vizzarro
2024-06-06  9:17 ` [PATCH v4 0/5] dts: testpmd show port info/stats Luca Vizzarro
2024-06-06  9:17   ` [PATCH v4 1/5] dts: fix InteractiveShell command prompt filtering Luca Vizzarro
2024-06-06 18:52     ` Jeremy Spewock
2024-06-06  9:17   ` [PATCH v4 2/5] dts: skip first line of send command output Luca Vizzarro
2024-06-06 18:52     ` Jeremy Spewock
2024-06-06  9:17   ` [PATCH v4 3/5] dts: add parsing utility module Luca Vizzarro
2024-06-06 18:52     ` Jeremy Spewock
2024-06-06 21:14       ` Luca Vizzarro
2024-06-06  9:17   ` [PATCH v4 4/5] dts: add `show port info` command to TestPmdShell Luca Vizzarro
2024-06-06 18:52     ` Jeremy Spewock
2024-06-06  9:17   ` [PATCH v4 5/5] dts: add `show port stats` " Luca Vizzarro
2024-06-06 18:53     ` Jeremy Spewock
2024-06-06 21:34 ` [PATCH v5 0/5] dts: testpmd show port info/stats Luca Vizzarro
2024-06-06 21:34   ` [PATCH v5 1/5] dts: fix InteractiveShell command prompt filtering Luca Vizzarro
2024-06-07 11:15     ` Juraj Linkeš
2024-06-07 13:10     ` Jeremy Spewock
2024-06-14 20:05     ` Nicholas Pratte
2024-06-06 21:34   ` [PATCH v5 2/5] dts: skip first line of send command output Luca Vizzarro
2024-06-07 11:16     ` Juraj Linkeš
2024-06-07 13:10     ` Jeremy Spewock
2024-06-06 21:34   ` [PATCH v5 3/5] dts: add parsing utility module Luca Vizzarro
2024-06-07 11:16     ` Juraj Linkeš
2024-06-07 13:11     ` Jeremy Spewock
2024-06-14 17:39     ` Nicholas Pratte
2024-06-06 21:34   ` [PATCH v5 4/5] dts: add `show port info` command to TestPmdShell Luca Vizzarro
2024-06-07 11:16     ` Juraj Linkeš
2024-06-07 13:11     ` Jeremy Spewock
2024-06-14 17:36     ` Nicholas Pratte
2024-06-06 21:34   ` [PATCH v5 5/5] dts: add `show port stats` " Luca Vizzarro
2024-06-07 11:16     ` Juraj Linkeš
2024-06-07 13:11     ` Jeremy Spewock
2024-06-14 17:34     ` Nicholas Pratte
2024-06-20  3:04   ` [PATCH v5 0/5] dts: testpmd show port info/stats Thomas Monjalon

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=685b9f60-7b5d-407f-be26-69115e23a9d2@pantheon.tech \
    --to=juraj.linkes@pantheon.tech \
    --cc=dev@dpdk.org \
    --cc=jspewock@iol.unh.edu \
    --cc=luca.vizzarro@arm.com \
    --cc=paul.szczepanek@arm.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).