从 Python 3 中的给定字符串解析测量值(多维) [英] Parse measurements (multiple dimensions) from a given string in Python 3

问题描述

我知道这篇文章和这个库 但他们在下面的这些特定情况下没有帮助我.我如何解析如下测量值:

我有如下字符串；

方形 10 x 3 x 5 毫米"第 23/22 轮；24,9 x 12,2 x 12,3"正方形 10x2"直10x2mm"

我正在寻找一个 Python 包或某种方式来获得如下结果；

<预><代码>>>>a =amazing_parser.parse("正方形 10 x 3 x 5 毫米")>>>打印(一)10 x 3 x 5 毫米

同样；

<预><代码>>>>a =amazing_parser.parse("Round 23/22; 24,9x12,2")>>>打印(一)24,9 x 12,2

我也尝试使用命名实体识别" 使用 "ner_ontonotes_bert_mult" 模型.但结果如下:

<预><代码>>>>从 deeppavlov 导入配置，build_model>>>ner_model = build_model(configs.ner.ner_ontonotes_bert_mult，下载=真)>>>打印(ner_model([第 23/22 轮；24,9 x 12,2 x 12,3"]))<class 'list'>: [[['Round', '23', '/', '22', ';', '24', ',', '9', 'x', '12', ',', '2', 'x', '12', ',', '3']], [['O', 'B-CARDINAL', 'O', 'B-CARDINAL','O', 'B-基数', 'O', 'B-基数', 'O', 'B-基数', 'O', 'B-基数', 'O', 'B-基数','O', 'B-CARDINAL']]]]

我不知道如何正确地从这个列表中提取这些测量值.

我还发现了这个正则表达式:

>>>re.findall("(\d+(?:,\d+)?) x (\d+(?:,\d+)?)(?: x (\d+(?:,\d+)?))?", "直 10 x 2 毫米")<class 'list'>: [('10', '2', '')]

但如果输入包含 2 个维度，它会在结果列表中留下一个空值，如果数字和x"之间没有空格，它就不起作用.我不擅长正则表达式...

解决方案

对于给定的示例，您可以使用:

(?

部分

• (?<!\S) 负向后视，断言左边的不是非空白字符
• \d+(?:,\d+)? 匹配 1+ 个数字和可选的 , 和 1+ 个数字
• ?x ? 在可选空格之间匹配x
• \d+(?:,\d+)? 匹配 1+ 个数字和可选的 , 和 1+ 个数字
• (?: 非捕获组
  • ?x ?\d+匹配x` 可选空格和 1+ 位数字
  • (?:,\d+)? 可选择匹配一个 , 和 1+ 个数字
• )* 关闭非捕获组并重复 0+ 次

正则表达式演示 |Python 演示

例如

导入重新正则表达式 = r"(?

输出

['10 x 3 x 5', '24,9 x 12,2 x 12,3', '10x2', '10x2', '24,9x12,2']

I'm aware of this post and this library but they didn't help me with these specific cases below. How can I parse measurements like below:

I have strings like below;

"Square 10 x 3 x 5 mm"
"Round 23/22; 24,9 x 12,2 x 12,3"
"Square 10x2"
"Straight 10x2mm"

I'm looking for a Python package or some way to get results like below;

>>> a = amazing_parser.parse("Square 10 x 3 x 5 mm")
>>> print(a)
10 x 3 x 5 mm

Likewise;

>>> a = amazing_parser.parse("Round 23/22; 24,9x12,2")
>>> print(a)
24,9 x 12,2

I also tried to use "named entity recognition" using "ner_ontonotes_bert_mult" model. But the results were like below:

>>> from deeppavlov import configs, build_model
>>> ner_model = build_model(configs.ner.ner_ontonotes_bert_mult, download=True)
>>> print(ner_model(["Round 23/22; 24,9 x 12,2 x 12,3"]))
<class 'list'>: [[['Round', '23', '/', '22', ';', '24', ',', '9', 'x', '12', ',', '2', 'x', '12', ',', '3']], [['O', 'B-CARDINAL', 'O', 'B-CARDINAL', 'O', 'B-CARDINAL', 'O', 'B-CARDINAL', 'O', 'B-CARDINAL', 'O', 'B-CARDINAL', 'O', 'B-CARDINAL', 'O', 'B-CARDINAL']]]

I have no idea how to extract those measurements from this list properly.

I also found this regex:

>>>re.findall("(\d+(?:,\d+)?) x (\d+(?:,\d+)?)(?: x (\d+(?:,\d+)?))?", "Straight 10 x 2 mm")
<class 'list'>: [('10', '2', '')]

But it does leave an empty value in the resulting list if the input contains 2 dimensions and it doesn't work if there is no whitespace between numbers and "x"s. I'm not good with regex...

解决方案

For the given examples, you might use:

(?<!\S)\d+(?:,\d+)? ?x ?\d+(?:,\d+)?(?: ?x ?\d+(?:,\d+)?)*

In parts

• (?<!\S) Negative lookbehind, assert what is on the left is not a non whitespace char
• \d+(?:,\d+)? Match 1+ digits and optionally a , and 1+ digits
• ?x ? Match x between optional spaces
• \d+(?:,\d+)? Match 1+ digits and optionally a , and 1+ digits
• (?: Non capturing group
  • ?x ?\d+Matchx` between optional spaces and 1+ digits
  • (?:,\d+)? Optionally match a , and 1+ digits
• )* Close non capturing group and repeat 0+ times

Regex demo | Python demo

For example

import re

regex = r"(?<!\S)\d+(?:,\d+)? ?x ?\d+(?:,\d+)?(?: ?x ?\d+(?:,\d+)?)*"
test_str = ("Square 10 x 3 x 5 mm\n"
    "Round 23/22; 24,9 x 12,2 x 12,3\n"
    "Square 10x2\n"
    "Straight 10x2mm\n"
    "Round 23/22; 24,9x12,2")
result = re.findall(regex, test_str)
print(result)

Output

['10 x 3 x 5', '24,9 x 12,2 x 12,3', '10x2', '10x2', '24,9x12,2']

这篇关于从 Python 3 中的给定字符串解析测量值(多维)的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！