![\"A](\"\/user\/pages\/02.blog\/04.til\/049.edit-multiple-file-locations-in-vscode\/_editing_multiple_locations_in_vscode.gif?decoding=auto&fetchpriority=auto\")
<\/p>\nToday I learned how to edit multiple lines in the VS Code code editor.<\/p>\n\n
<\/p>\n
Inside VS Code, if you hold Ctrl<\/kbd> while clicking different lines with your mouse left button,\nyou will place multiple cursors.\nThen, you can start typing and those changes will affect all the locations you clicked.<\/p>\n With this keyboard shortcut, you can hold Ctrl<\/kbd> and click the end of multiple lines to edit them all at once.<\/p>\n That's it for now! Stay tuned<\/a> and I'll see you around!<\/p>","summary":"Today I learned how to edit multiple lines in the VS Code code editor.","date_modified":"2025-07-23T16:49:02+02:00","tags":["json","programming","python","vscode"],"image":"\/user\/pages\/02.blog\/04.til\/049.edit-multiple-file-locations-in-vscode\/thumbnail.webp"},{"title":"Custom JSON encoder and decoder","date_published":"2022-06-20T17:00:00+02:00","id":"https:\/\/mathspp.com\/blog\/custom-json-encoder-and-decoder","url":"https:\/\/mathspp.com\/blog\/custom-json-encoder-and-decoder","content_html":" This article explains how to extend the JSON format by using a custom encoder and a custom decoder to turn arbitrary Python objects into JSON and back.<\/p>\n\n The Python module In this article I want to define a system that makes it easy to extend the JSON format,\nso that we can encode more Python objects into JSON and back.<\/p>\n My goal is to define a mechanism through which you can easily define small, atomic encoders and decoders,\nand to have them all operate together.<\/p>\n I think it will be easier to understand what I want to achieve if I show you how I want the end product to look like.<\/p>\n Suppose that you want to extend the JSON format so that you can also encode and decode complex numbers and Python When you are done with the article, you will be able to define something like this:<\/p>\n Then, you will be able to use these two classes as the The point, here, is that I want to make it as easy as possible to extend the JSON standard,\nsimply by providing the encoders and the decoders for each new type you want to be able to handle.<\/p>\n The main issue I have to struggle with is in defining the mechanism that will allow the custom JSON decoder to recognise that certain JSON objects should actually be parsed into something else.\nFor example, in the previous article about custom JSON decoding<\/a>,\nI showed how to convert the following JSON:<\/p>\n into the Python complex number \\(1 + 2i\\)<\/span>:<\/p>\n However, suppose that we actually have the following Python dictionary:<\/p>\n If we convert this dictionary to JSON, we get a string:<\/p>\n Now, if we use our custom decoder, we will get a complex number back instead of the original dictionary!\nWhy?\nBecause complex numbers and some dictionaries have the same JSON representations.<\/p>\n In mathematical terms, we say that the JSON encoding is not injective.\nAfter all, I can find two objects We have two options here:<\/p>\n Today I finished learning how to use a custom JSON decoder to convert JSON to arbitrary Python objects.<\/p>\n\n In the two previous TIL articles,\nI wrote briefly about how to use a custom JSON encoder<\/a>\nand how to use a custom JSON decoder<\/a>.\nLooking at the two articles linked,\nthere is a noticeable asymmetry in the strategies used to do the encoding and the decoding.<\/p>\n We subclassed Reading the two previous sentences, even if I didn't know the module As it turns out, the former is true:\nthe module The documentation doesn't tell you much about how to use The functions “To use a custom Then, if you go and read the documentation on Thus, the obvious path forward seems to be to subclass Something along the lines of this blueprint:<\/p>\n With the blueprint above in mind, let us go back to the example of decoding complex numbers.<\/p>\n Let us assume that a complex number is represented as a JSON dictionary with the following structure:<\/p>\n Taking that into account, we can implement the object hook as follows:<\/p>\n Today I learned how to do custom JSON decoding in Python with the module The module By using The method You can read more about If you want to take a look at how to encode<\/strong> Python as JSON,\ntake a look at the TIL about custom JSON encoding of Python objects<\/a>.<\/p>\n Because we cannot encode arbitrary Python objects as JSON<\/strong> by default,\nwe implement custom JSON encoders<\/a> to work around that.<\/p>\n The reverse side of that coin is that we must be able to decode those objects again.\nHowever, the module But first, let us take a look at what the module According to the How to edit multiple lines in VS Code?<\/a><\/h2>\n
![\"There](\"\/images\/8\/4\/c\/f\/f\/84cffcf37c06d63dda8f033d02de61ef5d3408f5-thumbnail.webp\")
<\/p>\nCustom JSON encoder and decoder<\/a><\/h2>\n
json<\/code> allows you to work with the JSON data format.\nIn previous articles,\nI've written about doing custom JSON encoding<\/a> of arbitrary Python objects\nand custom JSON decoding<\/a> into<\/em> arbitrary Python objects.<\/p>\nExtending the JSON format<\/a><\/h2>\n
range<\/code> objects.\nThis is what you want to do.\nHow do you achieve that?<\/p>\nclass ComplexAndRangeEncoder(...):\n def encode_complex(self, c):\n return {\"real\": c.real, \"imag\": c.imag}\n\n def encode_range(self, r):\n return {\"start\": r.start, \"stop\": r.stop, \"step\": r.step}\n\nclass ComplexAndRangeDecoder(...):\n def decode_complex(self, obj):\n return complex(obj[\"real\"], obj[\"imag\"])\n\n def decode_range(self, obj):\n return range(obj[\"start\"], obj[\"stop\"], obj[\"step\"])<\/code><\/pre>\ncls<\/code> argument to the json<\/code> methods,\nenabling you to encode complex numbers and ranges to JSON, and then decoding them back.<\/p>\nAutomatically recognising non-standard JSON<\/a><\/h2>\n
{\n \"real\": 1.0,\n \"imag\": 2.0\n}<\/code><\/pre>\n(1+2j)<\/code><\/pre>\ndict_looks_like_complex = {\n \"real\": 1.0,\n \"imag\": 2.0,\n}<\/code><\/pre>\n'{\"real\": 1.0, \"imag\": 2.0}'<\/code><\/pre>\n\nobj1<\/code> and obj2<\/code> such that obj1 != obj2<\/code> and yet\njson.dumps(obj1) == json.dumps(obj2)<\/code>.<\/p>\n<\/div>\n![\"A](\"\/images\/3\/e\/0\/4\/0\/3e0406998371d120d1868896120b6c55823beb54-thumbnail.webp\")
<\/p>\nAsymmetry suggests other alternatives<\/a><\/h2>\n
json.JSONEncoder<\/code> in order to implement custom JSON encoding for our Python objects<\/a>,\nand then we passed that class to the methods json.dump<\/code> and json.dumps<\/code>.\nHowever, in order to implement custom JSON decoding into arbitrary Python objects<\/a>,\nwe just implemented a function that was then passed into json.load<\/code> and json.loads<\/code> as the parameter object_hook<\/code>.<\/p>\njson<\/code>,\nI would expect one of the two to be true:<\/p>\njson.JSONDecoder<\/code> that can be subclassed and then passed in to json.load<\/code> and json.loads<\/code>,\nwhich will also allow for custom JSON decoding; or<\/li>\njson.dump<\/code> and json.dumps<\/code> accept a parameter which should have a name similar to encode_hook<\/code> or encoder_hook<\/code>,\nwhich would allow us to implement just a function that would take care of the custom JSON encoding.<\/li>\n<\/ul>json<\/code> has a class JSONDecoder<\/code> that we can subclass – and then feed to json.load<\/code> and json.loads<\/code> –\nwhich allows us to decode JSON into arbitrary Python objects.\nLet us see how to do that.<\/p>\njson.JSONDecoder<\/code><\/a><\/h2>\njson.JSONDecoder<\/code><\/a>,\nso I will share with you what I could piece together.<\/p>\njson.load<\/code> and json.loads<\/code> accept a keyword argument cls<\/code>,\nwhich the documentation describes as behaving in the following way:<\/p>\n\n
JSONDecoder<\/code> subclass, specify it with the cls<\/code> kwarg; otherwise JSONDecoder<\/code> is used. Additional keyword arguments will be passed to the constructor of the class.”<\/p>\n<\/blockquote>\nJSONDecoder<\/code>,\nyou see that it accepts an object_hook<\/code> upon initialisation,\nwhich is what we have used to implement custom JSON decoding<\/a> in the previous article.<\/p>\njson.JSONDecoder<\/code> and make it so that,\nby default, it uses a custom object_hook<\/code> argument.<\/p>\nimport json\n\nclass MyJSONDecoder(json.JSONDecoder):\n def __init__(self, **kwargs):\n kwargs.setdefault(\"object_hook\", my_custom_object_hook)\n # Just need to implement this ---^^^^^^^^^^^^^^^^^^^^^\n super().__init__(**kwargs)\n\nmy_object = json.loads(json_data, cls=MyJSONDecoder)<\/code><\/pre>\nDefining a custom decoder class<\/a><\/h2>\n
{\n \"real\": 1.0,\n \"imag\": 2.0\n}<\/code><\/pre>\nimport json\n\nclass ComplexDecoder(json.JSONDecoder):\n def __init__(self, **kwargs):\n kwargs.setdefault(\"object_hook\", self.object_hook)\n super().__init__(**kwargs)\n\n def object_hook(self, dict_):\n \"\"\"Try to decode a complex number.\"\"\"\n try:\n return complex(dict_[\"real\"], dict_[\"imag\"])...<\/code><\/pre>","summary":"Today I finished learning how to use a custom JSON decoder to convert JSON to arbitrary Python objects.","date_modified":"2025-07-23T16:49:02+02:00","tags":["json","programming","python"],"image":"\/user\/pages\/02.blog\/04.til\/048.custom-json-decoder-part-2\/thumbnail.webp"},{"title":"TIL #047 \u2013 custom JSON decoder","date_published":"2022-06-04T00:00:00+02:00","id":"https:\/\/mathspp.com\/blog\/til\/custom-json-decoder","url":"https:\/\/mathspp.com\/blog\/til\/custom-json-decoder","content_html":"json<\/code>.<\/p>\n\n![\"A](\"\/images\/6\/0\/1\/c\/2\/601c27006d0b78c0aebe54c6778d1931bac6b93f-thumbnail.webp\")
<\/p>\nHow to decode Python objects as JSON<\/a><\/h2>\n
json<\/code> is your go-to tool whenever you need to work with JSON data in Python.\nMuch like you have the methods json.dump<\/code> and json.dumps<\/code> to turn Python objects into JSON,\nyou can use the methods json.load<\/code> and json.loads<\/code> to turn JSON into Python.<\/p>\njson.loads<\/code>,\nyou can take a JSON string and turn it into a valid Python object:<\/p>\nimport json\n\nstring = '[42, {\"key\": null, \"other_key\": false}]'\n\nprint(json.loads(string))<\/code><\/pre>\n[42, {'key': None, 'other_key': False}]<\/code><\/pre>\njson.load<\/code> comes in handy if you have a file (or something similar) to read the data from.<\/p>\njson<\/code>, json.load<\/code>, and json.loads<\/code> in the json<\/code> documentation<\/a>.<\/p>\njson<\/code> cannot guess what custom encodings we may have done,\nso we need to specify our custom decodings<\/strong> when appropriate.<\/p>\njson<\/code> does in its default decoding.<\/p>\nDefault JSON decoding into Python objects<\/a><\/h2>\n
json<\/code> docs<\/a>,\nthis is what the default decoding does:<\/p>\n
![\"A](\"\/user\/pages\/02.blog\/04.til\/046.custom-json-encoder\/thumbnail.webp\")
<\/p>\n