在 PyParsing 中，如何指定一个 Word 不等于给定的文字？

Question

我正在尝试解析来自 http://www.apkmirror.com such as http://www.apkmirror.com/apk/google-inc/gmail/gmail-7-3-26-152772569-release-release/gmail-7-3-26-152772569-release-android-apk-download/ 的 APK 下载页面。通常，"APK details" 部分具有以下结构：

我想将“17329196”解析为 version_code，"arm" 解析为 architecture，"com.skype.m2" 解析为 package。但是，有时 architecture 行会丢失，如下所示：

到目前为止，使用 Scrapy 选择器

apk_details = response.xpath('//*[@title="APK details"]/following-sibling::*[@class="appspec-value"]//text()').extract()

我已经能够提取包含上面显示的 'lines' 的列表。我正在尝试编写一个函数 parse_apk_details 以便通过以下测试：

import pytest

def test_parse_apk_details_with_architecture():
    apk_details = [u'Version: 3.0.38_ww (4030038)',
                   u'arm ',
                   u'Package: com.lenovo.anyshare.gps',
                   u'\n',
                   u'2,239 downloads ']

    version_code, architecture, package = parse_apk_details(apk_details)

    assert version_code == 4030038
    assert architecture == "arm"
    assert package == "com.lenovo.anyshare.gps"

@pytest.mark.skip(reason="This does not work yet, because 'Package:' is interpreted by the parser as the architecture.")
def test_parse_apk_details_without_architecture():
    apk_details = [u'Version: 3.0.38_ww (4030038)',
                   u'Package: com.lenovo.anyshare.gps',
                   u'\n',
                   u'2,239 downloads ']

    version_code, architecture, package = parse_apk_details(apk_details)

    assert version_code == 4030038
    assert package == "com.lenovo.anyshare.gps"


if __name__ == "__main__":
    pytest.main([__file__])

但是，如上所述，第二个测试还没有通过。这是到目前为止的功能：

from pyparsing import Word, printables, nums, Optional

def parse_apk_details(apk_details):
    apk_details = "\n".join(apk_details)    # The newline character is ignored by PyParsing (by default)
    version_name = Word(printables)         # The version name can consist of all printable, non-whitespace characters
    version_code = Word(nums)               # The version code is expected to be an integer
    architecture = Word(printables)
    package = Word(printables)

    expression = "Version:" + version_name + "(" + version_code("version_code") + ")" + Optional(architecture("architecture")) + "Package:" + package("package")
    result = expression.parseString(apk_details)

    return int(result.get("version_code")), result.get("architecture"), result.get("package")

我尝试运行第二个测试时遇到的错误是：

ParseException: Expected "Package:" (at char 38), (line:2, col:10)

我相信正在发生的事情是 "Package:" 正在 'consumed' 作为 architecture。解决此问题的一种方法是将 architecture = Word(printables) 行更改为类似（伪代码）architecture = Word(printables) + ~"Package:" 的内容，以指示它可以是任何由可打印字符组成的内容，但单词 "Package:" 除外.

如何确保 architecture 仅在不是特定词 "Package:" 时才被解析？（我也对原始问题的基于 scrapy 的替代解决方案感兴趣）。

Answer 1

我最终使用了包含体系结构的行的不同特征（例如 "arm"）：体系结构后跟一个换行符（如果存在）这一事实。我将方法 parse_apk_details 修改为以下内容：

from pyparsing import Word, printables, nums, Optional, LineEnd, FollowedBy, Suppress

def parse_apk_details(apk_details):
    apk_details = "\n".join(apk_details)    # The newline character is ignored by PyParsing (by default)
    version_name = Word(printables).setResultsName("version")       # The version name can consist of all printable, non-whitespace characters
    version_code = Word(nums).setResultsName("version_code")                # The version code is expected to be an integer
    architecture = Word(printables).setResultsName("architecture") + Suppress(FollowedBy(LineEnd()))
    package = Word(printables).setResultsName("package")

    expression = "Version:" + version_name + "(" + version_code + ")" + Optional(architecture) + "Package:" + package
    result = expression.parseString(apk_details)

    return int(result.get("version_code")), result.get("architecture"), result.get("package")

这两个测试都通过了。

Answer 2

你和 architecture = Word(printables) + ~Literal("Package:") 真的很亲密。要进行否定前瞻，从否定开始，然后是匹配：

architecture = ~Literal("Package:") + Word(printables)

在 PyParsing 中，如何指定一个 Word 不等于给定的文字？

In PyParsing, how to specify that a Word is not equal to a given literal?

python

pyparsing

scrapy