How to write a simple JSONParser using Python

JSON Tokenizer

For lexical analysis of JSON, I mainly referred to the method in the screenshot above and wrote a simple example myself. It is relatively simple to write, and it should be said that it can only support a simple subset of JSON.

For the types of TOKEN here, refer to https://json.org, but its JSON syntax format is with whitespace. I am not used to dealing with this, so I did not refer to its syntax. After lexical analysis, spaces, newlines, and tabs are filtered out. I simply discard them without processing them.

json_tokenizer.py

Use regular expressions to perform lexical analysis of JSON.

import json
import re

from typing import Dict, List, Union


# TOKEN 的种类
LEFT_BRACE = "LEFT_BRACE"        # {
RIGHT_BRACE = "RIGHT_BRACE"      # }
LEFT_BRACKET = "LEFT_BRACKET"    # ]
RIGHT_BRACKET = "RIGHT_BRACKET"  # [
COLON = "COLON"                  # :
COMMA = "COMMA"                  # ,
NUMBER = "NUMBER"                # ".*?"
STRING = "STRING"                # [1-9]\d*
BOOL = "BOOL"                    # true/false
NULL = "NULL"                    # null
NEWLINE = "NEWLINE"            # \n
SKIP = "SKIP"                    # ' ', '\t'
MISMATCH = "MISMATCH"            # mismatch

# 处理 token 的正则
token_specification = [
    ('LEFT_BRACE', r'[{]'),
    ('RIGHT_BRACE', r'[}]'),
    ('LEFT_BRACKET', r'[\[]'),
    ('RIGHT_BRACKET', r'[\]]'),
    ('COLON', r'[:]'),
    ('COMMA', r'[,]'),
    ('NUMBER', r'-?[1-9]+[0-9]*'),
    ('STRING', r'".*?"'),
    ('BOOL', r'(true)|(false)'),
    ('NULL', r'null'),
    ('NEWLINE', r'\n'),
    ('SKIP', r'[ \t]'),
    ('MISMATCH', r'.')
]

tok_regex = &#39;|&#39;.join(&#39;(?P<%s>%s)&#39; % pair for pair in token_specification)
print("Debug: ", tok_regex)


def process(kind: str, value: str) -> Dict[str, Union[str, bool, int, None]]:
    """
    处理输入的 kind 和 value，并生成 Dict 对象，简单表示 token 对象
    """
    if kind == STRING:
        # 去掉外层的双引号，暂时没有比较好的方式
        return {"kind": kind, "value": value[1:-1]}
    if kind == NUMBER:
        return {"kind": kind, "value": int(value)}
    if kind == BOOL:
        if value == "true":
            return {"kind": kind, "value": True}
        else:
            return {"kind": kind, "value": False}
    if kind == NULL:
        return {"kind": kind, "value": None}
    return {"kind": kind, "value": value}


def tokenizer(json_str: str) -> List[Dict[str, Union[str, bool, int, None]]]:
    """
    tokenizer
    """
    tokens = []
    for m in re.finditer(tok_regex, json_str):
        # 获取 token 的类型
        kind = m.lastgroup
        # 获取 token 的值
        value = m.group()
        if kind == MISMATCH:
            raise Exception("json format is error")
        if kind == NEWLINE:
            continue
        if kind == SKIP:
            continue
        token = process(kind=kind, value=value)
        tokens.append(token)

    return tokens


if __name__ == "__main__":
    json_doc = open("./demo.json", "r", encoding="utf-8").read()
    tokens = tokenizer(json_doc)
    if tokens:
        json.dump(tokens, open("./json_tokens.json", "w",
                               encoding="utf-8"), ensure_ascii=False)

I have put all the input and output data in the document. I will post my input data and part of the output data below.

demo.json

{
    "name": "小黑子",
    "age": 3,
    "gender": false,
    "other_info": {
        "friends": [
            "嘎子",
            "潘叔",
            "狗"
        ],
        "declaration": "练习时长两年半",
        "hobbies": [
            "唱",
            "跳",
            "rap",
            "篮球????"
        ]
    }
}

json_token.json Part of the data. I formatted the data, so it is relatively long. Here is only a part.

JSON Parser

json_parser.py

Parser the token sequence generated in the previous step to generate a Dict object corresponding to JSON. The implementation of parser refers to the json syntax file of antlr4, which removes the whitespace and is simpler to process.

import json
from typing import Dict, Union

# TOKEN 的种类
LEFT_BRACE = "LEFT_BRACE"        # {
RIGHT_BRACE = "RIGHT_BRACE"      # }
LEFT_BRACKET = "LEFT_BRACKET"    # ]
RIGHT_BRACKET = "RIGHT_BRACKET"  # [
COLON = "COLON"                  # :
COMMA = "COMMA"                  # ,
NUMBER = "NUMBER"                # ".*?"
STRING = "STRING"                # [1-9]\d*
BOOL = "BOOL"                    # true/false
NULL = "NULL"                    # null


class Token(object):
    """为了简单，就不创建这个了"""


class JSON_Parser(object):
    """
    JSON_Parser the class aims parse input token sequence into a python object or array.
    """

    def __init__(self, tokens) -> None:
        self.index = 0
        self.tokens = tokens

    def get_token(self) -> Dict[str, Union[str, int, bool, None]]:
        """
        get current's token
        """
        if self.index < len(self.tokens):
            return self.tokens[self.index]
        else:
            raise Exception("index out of range.")

    def move_token(self) -> Dict[str, Union[str, int, bool, None]]:
        """
        move to next token and return it
        """
        if self.index + 1 < len(self.tokens):
            self.index = self.index + 1
            return self.tokens[self.index]
        else:
            raise Exception("index out of range.")

    def parse(self):
        """
        parse whole json
        """
        token = self.get_token()
        if token.get("kind") == LEFT_BRACE:
            return self.parse_obj()
        elif token.get("kind") == LEFT_BRACKET:
            return self.parse_arr()
        else:
            raise Exception("error json, neither object or array.")

    def parse_obj(self):
        """
        parse object
        """
        obj = {}
        token = self.move_token()
        kind = token.get("kind")
        # &#39;{&#39; &#39;}&#39;
        if kind == RIGHT_BRACE:
            return obj
        # &#39;{&#39; pair (&#39;,&#39; pair)* &#39;}&#39;
        name, val = self.parse_pair()
        obj[name] = val

        while self.index < len(self.tokens):
            token = self.move_token()
            kind = token.get("kind")
            if kind == COMMA:
                self.move_token()
                name, val = self.parse_pair()
                obj[name] = val
            elif kind == RIGHT_BRACE:
                return obj
            else:
                raise Exception("parse object encounter error")

    def parse_arr(self):
        """
        parse array
        """
        arr = []
        token = self.move_token()
        kind = token.get("kind")
        # &#39;[&#39; &#39;]&#39;
        if kind == RIGHT_BRACE:
            return arr
        # &#39;[&#39; value (&#39;,&#39; value)* &#39;]&#39;
        val = self.parse_value()
        arr.append(val)

        while self.index < len(self.tokens):
            token = self.move_token()
            kind = token.get("kind")
            if kind == COMMA:
                self.move_token()
                val = self.parse_value()
                arr.append(val)
            elif kind == RIGHT_BRACKET:
                return arr
            else:
                raise Exception("parse array encounter error")

    def parse_value(self):
        """
        parse value
        """
        token = self.get_token()
        kind = token.get("kind")
        if kind == LEFT_BRACE:
            return self.parse_obj()
        elif kind == LEFT_BRACKET:
            return self.parse_arr()
        elif kind == STRING or kind == NUMBER or kind == BOOL:
            return token.get("value")
        elif kind == NULL:
            return
        else:
            raise Exception("encounter unexcepted token")

    def parse_pair(self):
        """
        parse pair
        """
        token = self.get_token()
        kind = token.get("kind")
        name = token.get("value")
        # STRING &#39;:&#39; value
        if kind == STRING:
            token = self.move_token()
            kind = token.get("kind")
            if kind == COLON:
                token = self.move_token()
                return name, self.parse_value()

        raise Exception("parse pair encounter error")


if __name__ == "__main__":
    # json token 文件路径
    TOKEN_PATH = "./json_tokens.json"
    # 读取 token 序列
    input_tokens = [token for token in json.load(
        open(TOKEN_PATH, "r", encoding="utf-8"))]
    if not input_tokens:
        raise Exception("input token sequence is empty")

    # 调试的时候，用来查表的，很方便定位到 index 走到哪一个 token 了

    for i, tok in enumerate(input_tokens):
        print(f"debug {i:2d} --> {tok}")
    print("\n===========================================\n")
    parser = JSON_Parser(tokens=input_tokens)
    json_obj = parser.parse()
    # 再将 object 转成 json 并格式化后输出
    print(json.dumps(json_obj, ensure_ascii=False, indent=4))

Output result:

The above is the detailed content of How to write a simple JSONParser using Python. For more information, please follow other related articles on the PHP Chinese website!