Parse UNRAR output with pyparsing

test_parse.py

#!/bin/env python3
import json

from pyparsing import (
    ParserElement, Literal, Group, restOfLine,
    ZeroOrMore, LineEnd, CharsNotIn
)

TEXT = """UNRAR 5.71 freeware      Copyright (c) 1993-2019 Alexander Roshal

Archive: test_advarchs_direct_link.rar
Details: RAR 5

        Name: PyCharmHotKeys.pdf
        Type: File
        Size: 350573
 Packed size: 301085
       Ratio: 85%
       mtime: 2017-05-30 06:26:53,000000000
  Attributes: ..A....
       CRC32: 538D9621
     Host OS: Windows
 Compression: RAR 5.0(v50) -m3 -md=16M

        Name: sublime_text_shortcuts.pdf
        Type: File
        Size: 67750
 Packed size: 63559
       Ratio: 93%
       mtime: 2017-12-06 01:54:03,000000000
  Attributes: ..A....
       CRC32: F08BA6D1
     Host OS: Windows
 Compression: RAR 5.0(v50) -m3 -md=16M

        Name: TextInMotion-VideoSample.mp4
        Type: File
        Size: 10454044
 Packed size: 9955216
       Ratio: 95%
       mtime: 2019-08-06 13:34:27,207165800
  Attributes: ..A....
       CRC32: C799547F
     Host OS: Windows
 Compression: RAR 5.0(v50) -m3 -md=16M

        Name: bensound-summer.mp3
        Type: File
        Size: 4349086
 Packed size: 4329351
       Ratio: 99%
       mtime: 2019-08-06 13:32:47,950623500
  Attributes: ..A....
       CRC32: A6203CE2
     Host OS: Windows
 Compression: RAR 5.0(v50) -m3 -md=16M

        Name: Financial Sample.xlsx
        Type: File
        Size: 83418
 Packed size: 77547
       Ratio: 92%
       mtime: 2019-08-06 13:23:29,329002900
  Attributes: ..A....
       CRC32: 84D57D24
     Host OS: Windows
 Compression: RAR 5.0(v50) -m3 -md=16M

        Name: QO
        Type: Service
        Size: 351
 Packed size: 351
       Ratio: 100%
  Attributes: .B
     Host OS: Windows
 Compression: RAR 5.0(v50) -m0 -md=128K


Service: EOF"""


def parse(value: str):
    ParserElement.defaultWhitespaceChars = " \t"
    SEMICOLON = Literal(': ').suppress()
    EOL = LineEnd().suppress()

    header = Literal('UNRAR') + restOfLine

    row = Group(
        CharsNotIn('\n:')   # Key
        + SEMICOLON
        + restOfLine        # Value
        + EOL
    )

    # Sections are separated from each other by empty lines
    section = Group(
        EOL[1, 2]
        + ZeroOrMore(row)
    )

    expression = (
        header.suppress()      # Software info
        + section.suppress()   # Archive info
        + ZeroOrMore(section)  # Files info
    )

    sections = expression.parseString(value).asList()
    return sections


def main():
    # TODO we have to do `.strip()` here because we grab everything
    return [
        dict((key.strip(), value) for key, value in section)
        for section in parse(value=TEXT)
    ]


if __name__ == '__main__':
    print(json.dumps(main(), indent=2))

'''
[
  {
    "Name": "PyCharmHotKeys.pdf",
    "Type": "File",
    "Size": "350573",
    "Packed size": "301085",
    "Ratio": "85%",
    "mtime": "2017-05-30 06:26:53,000000000",
    "Attributes": "..A....",
    "CRC32": "538D9621",
    "Host OS": "Windows",
    "Compression": "RAR 5.0(v50) -m3 -md=16M"
  },
  {
    "Name": "sublime_text_shortcuts.pdf",
    "Type": "File",
    "Size": "67750",
    "Packed size": "63559",
    "Ratio": "93%",
    "mtime": "2017-12-06 01:54:03,000000000",
    "Attributes": "..A....",
    "CRC32": "F08BA6D1",
    "Host OS": "Windows",
    "Compression": "RAR 5.0(v50) -m3 -md=16M"
  },
  {
    "Name": "TextInMotion-VideoSample.mp4",
    "Type": "File",
    "Size": "10454044",
    "Packed size": "9955216",
    "Ratio": "95%",
    "mtime": "2019-08-06 13:34:27,207165800",
    "Attributes": "..A....",
    "CRC32": "C799547F",
    "Host OS": "Windows",
    "Compression": "RAR 5.0(v50) -m3 -md=16M"
  },
  {
    "Name": "bensound-summer.mp3",
    "Type": "File",
    "Size": "4349086",
    "Packed size": "4329351",
    "Ratio": "99%",
    "mtime": "2019-08-06 13:32:47,950623500",
    "Attributes": "..A....",
    "CRC32": "A6203CE2",
    "Host OS": "Windows",
    "Compression": "RAR 5.0(v50) -m3 -md=16M"
  },
  {
    "Name": "Financial Sample.xlsx",
    "Type": "File",
    "Size": "83418",
    "Packed size": "77547",
    "Ratio": "92%",
    "mtime": "2019-08-06 13:23:29,329002900",
    "Attributes": "..A....",
    "CRC32": "84D57D24",
    "Host OS": "Windows",
    "Compression": "RAR 5.0(v50) -m3 -md=16M"
  },
  {
    "Name": "QO",
    "Type": "Service",
    "Size": "351",
    "Packed size": "351",
    "Ratio": "100%",
    "Attributes": ".B",
    "Host OS": "Windows",
    "Compression": "RAR 5.0(v50) -m0 -md=128K"
  },
  {
    "Service": "EOF"
  }
]
'''