2023年6月12日 23:48:35go评论66阅读模式

英文:

YAML representation of a dictionary from multi-level Ctype Structure gets a strange object

问题

TL;DR: 我有一个包含另一个结构的ctype结构，我将其（看似正确）转换为Python字典，然后尝试将其转储到YAML中。但内部结构的值显示错误。

Python版本：3.10
Pyyaml版本：6.0

背景：我试图使我们的内部配置处理更加用户友好。我们的配置文件是从C结构的序列化副本中的数据生成的，直到现在，它是通过十六进制编辑器手动更改的。我的计划是使这个过程更加人类可读，短期内通过读取和写入YAML文件来实现。

我使用Ctypes Structure来序列化和反序列化数据，主要是因为我必须支持位字段。Ctype Structure可以用作模板，没有实际的值。

我通过删除不相关的函数并缩短结构来简化了代码。

我创建了一个My_Structure类，它继承自Ctypes的Structure，并允许以不同的方式表示数据，包括字典。这似乎效果不错。

但是，当我尝试将这个字典转储到YAML文件时，ee_sruct内部的substruct的值被错误地转储，就像一个Python对象一样。我不明白为什么会这样，因为类型检查显示它仍然是一个字典。

如果我使用具有与get_as_dict返回的内容完全相同的硬编码字典，一切正常。

显然，转储函数没有获得与从get_as_dict打印的数据相同的数据。为什么会这样，我该如何修复它？

我尝试的方法：
我的第一个想法是实现一个递归函数，以返回内部结构的dict（类似于我为数组所做的）。但是我不确定从何处开始，因为substruct已经被报告为dict，并且使用字符串（硬编码）表示法可行。

How to export a Pydantic model instance as YAML with URL type as string 看起来是一个不错的方法，但是将Structure和YAMLObject结合使用导致了元类冲突，我无法解决。

我尝试将数据转储为Json或使用ruamel.yml，但两者都引发了异常，抱怨substruct_t。

Combining Dumper class with string representer to get exact required YAML output 可能是正确的方法，但看起来相当复杂，我希望有一个更简单的解决方案，我可能只是忽略了它。

我只是找到了一个“脏”解决方法，步骤如下：

将从get_as_dict()得到的字典转换为字符串
将所有'字符替换为"
使用json.loads()在字符串上创建一个新字典，并使用该字典
它可以工作，但只是强调了我的问题，为什么两个字典与转储器不同？

（以上内容是对您提供的问题和背景信息的总结，不包含具体的回答或解决方案。）

英文:

TL;DR: I have a ctype Structure with another Structure inside, which I convert (seemingly correct) to a python dictionary, then I attempt to dump it into YAML. However the value of the internal Structure is shown wrong.

Python version: 3.10
Pyyaml version: 6.0

Background: I am trying to make our internal configuration handling more user-friendly. Our configuration files are the serialized copy of data from a C structure, and until now it was changed manually via hexeditor. My plan is to make this process more human readable, in a short term reading and writing YAML files.

I am using Ctypes Structure to serialize and deserialize the data, mainly because I have to support bitfields. The Ctype Structure could be used as a template, without actual values in it.

I have simplified the code by removing irrelevant functions and shortening the structures.

class substruct_t(My_Structure):
    _pack_ = 1
    _fields_ = [
        (&quot;app&quot;, c_uint8, 4),
        (&quot;showScale&quot;, c_uint8, 2),
        (&quot;showIdleTemp&quot;, c_uint8, 2),
        (&quot;type&quot;, c_uint8),
    ]

class ee_struct(My_Structure):
    _pack_ = 1
    _fields_ = [
        (&quot;txStatusDelay&quot;, c_uint8, 5),
        (&quot;overrideHours&quot;, c_uint8, 3),
        (&quot;manualSet&quot;, c_int16),
        (&quot;tempOffset&quot;, c_int8),

        (&quot;substruct&quot;, substruct_t),

        (&quot;LogArr&quot;, (c_uint8*6)*3),
        (&quot;frostTemp&quot;, c_int8),
        (&quot;fileVer&quot;, c_uint8*4),
    ]

class eeprom_t(Union):
    _fields_ = [(&quot;as_struct&quot;, ee_struct), (&quot;as_bytes&quot;, c_uint8*29)]

    def __str__(self) -&gt; str:
        return str(self.as_struct)
    
    def get_as_dict(self):
        return self.as_struct.as_dict()
    
    def get_as_bytes(self):
        return np.ndarray((29, ), &#39;b&#39;, self.as_bytes, order=&#39;C&#39;)

I have created a My_Structure class, which inherits from Ctypes Structure, and allows different representation of the data, including dict. This seems to work well.

# Child class of Structure with string and dictionary representation functions, unwrapping arrays
class My_Structure(Structure):
    def __recursive_carray_get(self, value):
        # Necessary recursive function, if value is ctype array
        if hasattr(value, &#39;__len__&#39;):
            rtn = list()
            for i in range(value.__len__()):
                rtn.append(self.__recursive_carray_get(value.__getitem__(i)))
        else:
            rtn = value
        return rtn

    def __handle_array_type__(self, type):
        # example unformatted type: &lt;class &#39;__main__.c_ubyte_Array_6_Array_3&#39;&gt;
        return StringBetween(&quot;&#39;&quot;, &quot;&#39;&quot;, str(type)).split(&quot;.&quot;)[1]
    
    def __repr__(self) -&gt; str:
        return str(self.as_dict())

    def __str__(self) -&gt; str:
        values = &quot;,\n&quot;.join(f&quot;{name}={value[&#39;value&#39;]}&quot; for name, value in self.as_dict().items())
        return f&quot;&lt;{self.__class__.__name__}: {values}&gt;&quot;
    
    def as_dict(self) -&gt; dict:
        return {field[0]: {&#39;value&#39;: self.__recursive_carray_get(getattr(self, field[0])), &#39;type&#39;: self.__handle_array_type__(field[1])}
                for field in self._fields_}

However when I want to dump this dict into a YAML file, the value of substruct within ee_sruct is dumped badly, like a python object. I do not understand why, as a typecheck shows it is still a dict.

 ### Dict representation of substructure:
{&#39;value&#39;: {&#39;app&#39;: {&#39;value&#39;: 2, &#39;type&#39;: &#39;c_ubyte&#39;}, &#39;showScale&#39;: {&#39;value&#39;: 0, &#39;type&#39;: &#39;c_ubyte&#39;}, &#39;showIdleTemp&#39;: {&#39;value&#39;: 1, &#39;type&#39;: &#39;c_ubyte&#39;}, &#39;type&#39;: {&#39;value&#39;: 2, &#39;type&#39;: &#39;c_ubyte&#39;}}, &#39;type&#39;: &#39;substruct_t&#39;}

 ### pyyaml dump of entire structure:
txStatusDelay: {value: 8, type: c_ubyte}
overrideHours: {value: 1, type: c_ubyte}
manualSet: {value: 100, type: c_short}
tempOffset: {value: 0, type: c_byte}
substruct:
  value: !!python/object/apply:_ctypes._unpickle
  - !!python/name:__main__.substruct_t &#39;&#39;
  - !!python/tuple
    - {}
    - !!binary |
      QgI=
  type: substruct_t
LogArr:
  value:
  - [1, 0, 0, 0, 0, 0]
  - [2, 0, 0, 0, 0, 0]
  - [3, 0, 0, 0, 0, 0]
  type: c_ubyte_Array_6_Array_3
frostTemp: {value: 16, type: c_byte}
fileVer:
  value: [65, 66, 104, 0]
  type: c_ubyte_Array_4

If I use a hardcoded dict with the exact same contents I get from get_as_dict, everything works.

Apparently, the dump functions don't get the same data as what gets printed from get_as_dict. Why is that, and how can I fix it?

What I tried:

My first idea was to implement a recursive function to return dict for internal structures (similarly what I did for arrays), but I was not sure where to start, as substruct is already reported as dict, and using the string (hardcoded) representation works.

How to export a Pydantic model instance as YAML with URL type as string seemed like a good approach, but combining Structure and YAMLObject resulted in a metaclass conflict, which I was unable to resolve.

I tried to dump into Json or using ruamel.yml, both throw an exception, complaining about substruct_t.

Combining Dumper class with string representer to get exact required YAML output could be the right approach, however it looks quite complicated, and I am hoping, there is a more simple solution that I just overlooked.

I just found a dirty fix, following the steps:

convert the dict from get_as_dict() to a string
replace all ' characters to "
use json.loads() on the string to create a new dict, and use that instead
It works, but it just underlines my question, why are the two dicts different to the dumpers?

答案1

得分: 0

以下是您提供的内容的翻译：

Listing [Python.Docs]: ctypes - 用于Python的外部函数库。

问题在于*__recursive_carray_get*，它像其名称所示（处理数组，这是一致的）。

但是，当涉及到（子）结构时，它不处理它们（或者将它们处理为任何基本类型）：

因此，当将ee_struct实例序列化为字典（通过调用其as_dict方法）时，与substruct键对应的值实际上是substruct_t实例
由于重写了*__repr__（也使用as_dict*），在打印字典时，** substruct_t实例也显示为字典，掩盖了以前的错误**

我修复了您的代码中的错误，并进行了一些其他改进（更改最少的情况下）。

code00.py：

#!/usr/bin/env python

import ctypes as cts
import sys
from pprint import pprint as pp

import yaml

# 省略了代码的其余部分

输出：

>    ```lang-bat
>    [cfati@CFATI-5510-0:e:\Work\Dev\StackExchange\StackOverflow\q076458298]&gt; &quot;e:\Work\Dev\VEnvs\py_pc064_03.10_test0\Scripts\python.exe&quot; ./code00.py
>    Python 3.10.9 (tags/v3.10.9:1dd9be6, Dec  6 2022, 20:01:21) [MSC v.1934 64 bit (AMD64)] 064bit on win32
>    
>    ---------- Original object ----------
>    &lt;EEStruct: txStatusDelay=0,
>    overrideHours=0,
>    manualSet=0,
>    tempOffset=0,
>    substruct={&#39;app&#39;: {&#39;value&#39;: 0, &#39;type&#39;: &#39;c_ubyte&#39;}, &#39;showScale&#39;: {&#39;value&#39;: 0, &#39;type&#39;: &#39;c_ubyte&#39;}, &#39;showIdleTemp&#39;: {&#39;value&#39;: 0, &#39;type&#39;: &#39;c_ubyte&#39;}, &#39;type&#39;: {&#39;value&#39;: 0, &#39;type&#39;: &#39;c_ubyte&#39;}},
>    LogArr=[[0, 0, 0, 0, 0, 0], [0, 0, 0, 0, 0, 0], [0, 0, 0, 0, 0, 0]],
>    frostTemp=0,
>    fileVer=[0, 0, 0, 0]&gt;
>    
>    ---------- Type: &lt;class &#39;dict&#39;&gt; ----------
>    
>    ---------- Dictionary representation ----------
>    {&#39;LogArr&#39;: {&#39;type&#39;: &#39;c_ubyte_Array_6_Array_3&#39;,
>                &#39;value&#39;: [[0, 0, 0, 0, 0, 0],
>                          [0, 0, 0, 0, 0, 0],
>                          [0, 0, 0, 0, 0, 0]]},
>     &#39;fileVer&#39;: {&#39;type&#39;: &#39;c_ubyte_Array_4&#39;, &#39;value&#39;: [0, 0, 0, 0]},
>     &#39;frostTemp&#39;: {&#39;type&#39;: &#39;c_byte&#39;, &#39;value&#39;: 0},
>     &#39;manualSet&#39;: {&#39;type&#39;: &#39;c_short&#39;, &#39;value&#39;: 0},
>     &#39;overrideHours&#39;: {&#39;type&#39;: &#39;c_ubyte&#39;, &#39;value&#39;: 0},
>     &#39;substruct&#39;: {&#39;type&#39;: &#39;Substruct_t&#39;,
>                   &#39;value&#39;: {&#39;app&#39;: {&#39;type&#39;: &#39;c_ubyte&#39;, &#39;value&#39;: 0},
>                             &#39;showIdleTemp&#39;: {&#39;type&#39;: &#39;c_ubyte&#39;, &#39;value&#39;: 0},
>                             &#39;showScale&#39;: {&#39;type&#39;: &#39;c_ubyte&#39;, &#39;value&#39;: 0},
>                             &#39;type&#39;: {&#39;type&#39;: &#39;c_ubyte&#39;, &#39;value&#39;: 0}}},
>     &#39;tempOffset&#39;: {&#39;type&#39;: &#39;c_byte&#39;, &#39;value&#39;: 0},
>     &#39;txStatusDelay&#39;: {&#39;type&#39;: &#39;c_ubyte&#39;, &#39;value&#39;: 0}}
>    
>    ---------- YAML representation ----------
>    LogArr:
>      type: c_ubyte_Array_6_Array_3
>      value:
>      - - 0
>        - 0
>        - 0
>        - 0
>        - 0
>        - 0
>      - - 0
>        - 0
>        - 0
>        - 0
>        - 0
>        - 0
>      - - 0
>        - 0
>        - 0
>        - 0
>        - 0
>        - 0
>    fileVer:
>      type: c_ubyte_Array_4
>      value:
>      - 0
>      - 0
>      - 0
>      - 0
>    frostTemp:
>      type: c_byte
>      value: 0
>    manualSet:
>      type: c_short
>     

<details>
<summary>英文:</summary>

Listing [\[Python.Docs\]: ctypes - A foreign function library for Python](https://docs.python.org/library/ctypes.html#module-ctypes).

The problem is that *\_\_recursive\_carray\_get*, does what it name suggests (handles arrays, which is consistent).

**But**, when it comes to (sub) structures, it doesn&#39;t handle them (or it handles them as any basic type):

1. So, when the *ee\_struct* instance is serialized to a dictionary (by calling its *as\_dict* method), **the value corresponding to the *substruct* key is actually a *substruct\_t* instance**

2. Due to the fact that *\_\_repr\_\_* is overridden (to also use *as\_dict*), when printing the dictionary, **the *substruct\_t* instance is also displayed as a dictionary, masking the previous error**

I fixed the errors in your code, and added some other improvements (with minimum changes).

*code00.py*:

```python
#!/usr/bin/env python

import ctypes as cts
import sys
from pprint import pprint as pp

import yaml


class SerializableStructure(cts.Structure):
    @classmethod
    def _as_dict(cls, value):
        if isinstance(value, cts.Array):
            ret = [cls._as_dict(e) for e in value]
        elif hasattr(value, &quot;as_dict&quot;):
            ret = value.as_dict()
        else:
            ret = value
        return ret

    def __repr__(self) -&gt; str:
        return str(self.as_dict())

    def __str__(self) -&gt; str:
        values = &quot;,\n&quot;.join(f&quot;{name}={value[&#39;value&#39;]}&quot; for name, value in self.as_dict().items())
        return f&quot;&lt;{self.__class__.__name__}: {values}&gt;&quot;

    def as_dict(self) -&gt; dict:
        return {f[0]: {&quot;value&quot;: self._as_dict(getattr(self, f[0])), &quot;type&quot;: f[1].__name__}
                for f in self._fields_}


class Substruct_t(SerializableStructure):
    _pack_ = 1
    _fields_ = (
        (&quot;app&quot;, cts.c_uint8, 4),
        (&quot;showScale&quot;, cts.c_uint8, 2),
        (&quot;showIdleTemp&quot;, cts.c_uint8, 2),
        (&quot;type&quot;, cts.c_uint8),
    )


class EEStruct(SerializableStructure):
    _pack_ = 1
    _fields_ = (
        (&quot;txStatusDelay&quot;, cts.c_uint8, 5),
        (&quot;overrideHours&quot;, cts.c_uint8, 3),
        (&quot;manualSet&quot;, cts.c_int16),
        (&quot;tempOffset&quot;, cts.c_int8),
        (&quot;substruct&quot;, Substruct_t),
        (&quot;LogArr&quot;, (cts.c_uint8 * 6) * 3),
        (&quot;frostTemp&quot;, cts.c_int8),
        (&quot;fileVer&quot;, cts.c_uint8 * 4),
    )


def main(*argv):
    ees = EEStruct()
    marker = &quot;----------&quot;
    print(&quot;{:s} Original object {:s}&quot;.format(marker, marker))
    print(ees)
    d = ees.as_dict()
    print(&quot;\n{:s} Type: {:} {:s}&quot;.format(marker, type(d[&quot;substruct&quot;][&quot;value&quot;]), marker))  # @TODO - cfati: Check for the old implementation
    print(&quot;\n{:s} Dictionary representation {:s}&quot;.format(marker, marker))
    pp(d)
    print(&quot;\n{:s} YAML representation {:s}&quot;.format(marker, marker))
    print(yaml.dump(d))


if __name__ == &quot;__main__&quot;:
    print(&quot;Python {:s} {:03d}bit on {:s}\n&quot;.format(&quot; &quot;.join(elem.strip() for elem in sys.version.split(&quot;\n&quot;)),
                                                   64 if sys.maxsize &gt; 0x100000000 else 32, sys.platform))
    rc = main(*sys.argv[1:])
    print(&quot;\nDone.\n&quot;)
    sys.exit(rc)

Output:

> lang-bat > [cfati@CFATI-5510-0:e:\Work\Dev\StackExchange\StackOverflow\q076458298]> "e:\Work\Dev\VEnvs\py_pc064_03.10_test0\Scripts\python.exe" ./code00.py > Python 3.10.9 (tags/v3.10.9:1dd9be6, Dec 6 2022, 20:01:21) [MSC v.1934 64 bit (AMD64)] 064bit on win32 > > ---------- Original object ---------- > <EEStruct: txStatusDelay=0, > overrideHours=0, > manualSet=0, > tempOffset=0, > substruct={'app': {'value': 0, 'type': 'c_ubyte'}, 'showScale': {'value': 0, 'type': 'c_ubyte'}, 'showIdleTemp': {'value': 0, 'type': 'c_ubyte'}, 'type': {'value': 0, 'type': 'c_ubyte'}}, > LogArr=[[0, 0, 0, 0, 0, 0], [0, 0, 0, 0, 0, 0], [0, 0, 0, 0, 0, 0]], > frostTemp=0, > fileVer=[0, 0, 0, 0]> > > ---------- Type: <class 'dict'> ---------- > > ---------- Dictionary representation ---------- > {'LogArr': {'type': 'c_ubyte_Array_6_Array_3', > 'value': [[0, 0, 0, 0, 0, 0], > [0, 0, 0, 0, 0, 0], > [0, 0, 0, 0, 0, 0]]}, > 'fileVer': {'type': 'c_ubyte_Array_4', 'value': [0, 0, 0, 0]}, > 'frostTemp': {'type': 'c_byte', 'value': 0}, > 'manualSet': {'type': 'c_short', 'value': 0}, > 'overrideHours': {'type': 'c_ubyte', 'value': 0}, > 'substruct': {'type': 'Substruct_t', > 'value': {'app': {'type': 'c_ubyte', 'value': 0}, > 'showIdleTemp': {'type': 'c_ubyte', 'value': 0}, > 'showScale': {'type': 'c_ubyte', 'value': 0}, > 'type': {'type': 'c_ubyte', 'value': 0}}}, > 'tempOffset': {'type': 'c_byte', 'value': 0}, > 'txStatusDelay': {'type': 'c_ubyte', 'value': 0}} > > ---------- YAML representation ---------- > LogArr: > type: c_ubyte_Array_6_Array_3 > value: > - - 0 > - 0 > - 0 > - 0 > - 0 > - 0 > - - 0 > - 0 > - 0 > - 0 > - 0 > - 0 > - - 0 > - 0 > - 0 > - 0 > - 0 > - 0 > fileVer: > type: c_ubyte_Array_4 > value: > - 0 > - 0 > - 0 > - 0 > frostTemp: > type: c_byte > value: 0 > manualSet: > type: c_short > value: 0 > overrideHours: > type: c_ubyte > value: 0 > substruct: > type: Substruct_t > value: > app: > type: c_ubyte > value: 0 > showIdleTemp: > type: c_ubyte > value: 0 > showScale: > type: c_ubyte > value: 0 > type: > type: c_ubyte > value: 0 > tempOffset: > type: c_byte > value: 0 > txStatusDelay: > type: c_ubyte > value: 0 > > > Done. >

For more details on the same (or similar) topic, check:

通过集体智慧和协作来改善编程学习和解决问题的方式。致力于成为全球开发者共同参与的知识库，让每个人都能够通过互相帮助和分享经验来进步。

多层C类型结构的字典的YAML表示得到了一个奇怪的对象。

问题

答案1

相反的是互斥的 – 两个论点必须共存。

无预训练权重的迁移学习

Pyqt 应用在不同的监视器上使用 matplotlib 绘图时出现奇怪行为。

ValueError: 输出操作数的形状 (1,64) 与广播形状 (2,64) 不匹配。

What's the correct way to type hint an empty list as a literal in python?

如何在Highcharts Gantt中更改本地化的星期名称

如何在同一个流中使用多个过滤器和映射函数？

如何使用Map/Set来将代码优化到O(n)？

.NET MAUI Android在GitHub Actions上构建失败，错误代码为1。

如何在Playwright视觉比较中屏蔽多个定位器？

在C++中，可以使用可变模板参数来检索类型的内部类型。

selenium.common.exceptions.StaleElementReferenceException: Message: stale element reference: stale element not found

Creating and opening a URL to log in to Website via Basic Auth with Robot Framework/Selenium (Python)

AG Grid 在上下文菜单中以大文本形式打开

发表评论