多行宏的正则 记录基本单元

[2439905184]

匹配长度为4的int数组
意思是匹配=号右边的这些字符：" [数字,数字,数字,数字] " 且中间没有空格

(?<=\=)\[\d+,\d+,\d+,\d+]

测试用例

[2439905184]

作用：匹配=号右边的字符串
解释：匹配=号右边的 双引号并紧跟 范围内(所有小写字母 大写字母 小数点 0-9数字 空格 )（多次匹配）并紧跟双引号 的所有字符

(\"[a-zA-Z\._0-9\s]+\")

测试用例

[2439905184]

匹配宏名称
意思是匹配所有[号后面紧跟的字母（多次匹配） (分组)

((?<=\[)\w+)

测试用例

[2439905184]

完整正则备份

((?<=\[)\w+)|(\w+(?=\=))|(?<=\=)\[\d,\d,\d,\d]|(\"[a-zA-Z\._0-9\s]+\")

"((?<=\[)\w+)|(\w+(?=\=))|(?<=\=)\[\d,\d,\d,\d]|(\"[a-zA-Z\._0-9\s]+\")"gm

"((?<=\[)\w+)|(\w+(?=\=))|((?<=\=)\d+)|(?<=\=)\[\d,\d,\d,\d]|(\"[a-zA-Z\._0-9\s]+\")"gm

//注意去除开头结尾的" 结尾的gm表示全局并且多行匹配 （修改器）
"((?<=\[)\w+)|(\w+(?=\=))|((?<=\=)\d+)|((?<=\=)\[\d,\d\])|((?<=\=)\[\d,\d,\d,\d])|(\"[a-zA-Z\._0-9\s]+\")|(\'[a-zA-Z\._0-9\s]+\')"gm

[2439905184]

作用：匹配=号前面的参数名称
解释：匹配=号前面的所有字母 （多次匹配）（分组模式）

(\w+(?=\=))

测试用例

[2439905184]

作用：匹配=号后面的参数值
解释：匹配=号后面的所有数字 （多次匹配）

(?<=\=)\d+

测试用例

[2439905184]

作用：匹配坐标型 pos_t指长度为2的整数型数组，分别为横坐标x、纵坐标y

(?<=\=)\[\d,\d\]

测试用例

[2439905184]

作用：匹配使用单引号的字符串
'[a-zA-Z._0-9\s]+'

[2439905184]

公开正则库：https://regex101.com/r/T2wUDF/1

说明备份：用于解析bkengine的宏代码
如果代码的开头是形如// 这种注释，你需要先通过编程代码手动判断此行的开头是否为注释，如果是，则不处理后面的代码
如果注释在宏代码的后面 此正则不会处理在尾部的单行注释
注：此宏仅支持当行宏匹配（多行宏匹配可能会出现小问题）

[2439905184]

处理多行宏词法分析的步骤原则 （基于正则）

1. 如果发现注释，去除注释的那些字符串
2. 去除数组参数数字左右的空格，并替换为""
3. 使用正则进行分词

[2439905184]

预处理数组流程 string.replace()

\s+(?=\d)|(?<=\d)\s+(?#匹配数字前后多个空格，并替换为"")

变成这种形式以后

file=[0,0,800,600]

使用下面的方法取数组的值

((?<=\=)\[\d+,\d+\])(?#取长度为2的int数组值)|((?<=\=)\[\d+,\d+,\d+,\d+\])(?#取长度为4的int数组值)

[2439905184]

最终成品 (最后一步)

((?<=\[)\w+)(?#匹配宏名称)|(\w+(?=\=))(?#匹配参数名称)|((?<=\=)\d+)(?#匹配int参数)|((?<=\=)\[\d+,\d+\])(?#匹配长度为2的int数组)|((?<=\=)\[\d+,\d+,\d+,\d+])(?#匹配长度为4的int数组)|(\"[a-zA-Z\._0-9\s]+\")(?#匹配使用双引号的string参数)|(\'[a-zA-Z\._0-9\s]+\')(?#匹配使用单引号的string参数)

[2439905184]

@ktabata
I think regex is not easy to write .
I was wanted to make all parse function in just one regex , but I cant do that.
I think it is may be too hard. It has been the 极限 of regex 。
I think I need to use at least 2 or more regex to parser macro code .
Just resolve things into smaller chunks , so that may be easy.
I like this method to 解决各种问题。

[ktabata]

@2439905184
In this example, we could use regex.
But from computer science's point of view, it is processed in two steps: lexer and parser.
Please search for "词法分析" and "语法分析".
You can generate lexer and parser from some small expressions.

[2439905184]

@2439905184
In this example, we could use regex.
But from computer science's point of view, it is processed in two steps: lexer and parser.
Please search for "词法分析" and "语法分析".
You can generate lexer and parser from some small expressions.

但是bkengine的多行宏语法格式存在歧义lexer可以解决这个问题吗?

[sprite index=1 rect=[0,0,0,0]] [addto index=1 target=basic_layer]

'[ ]'can bu used as keyword of array
also can be uses for macro define.
this macro format is like krkr kag system

'[sprite' is the macro format , word must behind '[' and cant have space.
for different params , there must be only for one 'space' ,more spaces is not the right script format规范
and mulit macro must be close by ']'

[ktabata]

[sprite index=1 rect=[0,0,0,0]] [addto index=1 target=basic_layer]

You need a parser. Parser can parse multiple statements.

You can use lexer/parser generator like:
https://github.com/loloicci/nimly

But before using it, you have to be familiar with how lexer/parser works.