prettyparser 1.0.15

Installation

$ git clone https://github.com/leandroroser/prettyparser
$ cd prettyparser
$ pip install -e .

or

$ pip install prettyparser

Example: processing a folder with multiple PDF files

import regex as re
from prettyparser import PrettyParser

directory = "./BOOKS/PDF"
output = "./BOOKS/TXT"
parser = PrettyParser(directory, output, mode = 'pdf',
                      args = [[r"(\n\s*\d+\s*\n)|(\n\s*\d+\s*$)", r'\n\n'],
                            [r"\n\s*-\d-\s*\n", r'\n\n'],
                            [r"\n\s*(\* *)+\s*\n", r'\n\n'],
                            [r"__some_header_text", r'\n\n', re.IGNORECASE]],
                            remove_whitelines = True,
                            paragraphs_spacing = 1,
                            remove_hyphen_eol = True)
parser.run()

Example: processing a folder with multiple TXT files

Let’s assume that the previous output isn’t good enough and needs additional corrections. A quicker way for testing additional corrections can be implemented by using the previous TXT output:

directory = "./BOOKS/TXT"
output = "./BOOKS/TXT_REPARSED"
parser = PrettyParser(directory, output,  mode = 'txt',
                        args=[[r"some other header.*\d+", r''],
                            [r"^\d+.*", r'', re.MULTILINE],
                            [r"([A-Z]+)( *\n)([A-Z]+)", r'\1\3'],
                            remove_whitelines = True,
                            paragraphs_spacing = 1,
                            remove_hyphen_eol = True)
parser.run()

Example: processing a Python str for a quick test of the app

import regex as re
from prettyparser import PrettyParser


txt = """
header to remove

This is a text with multiple problems. For exam-
ple the latter word can be joined.
The portions of this line can be
joined
in a single line.
HERE ALSO IS SOME
UPPERCASE TEXT
TO JOIN
Some Other Ugly Stuff To Remove IGNORING Case.

Remove the line below:

* * *

Remove empty lines and finally separate paragraphs with a blank line.


Below is the page number->.
99
"""
parser = PrettyParser(txt, mode = "pyobj", args = [[r"\s*header to remove\s*\n",r""],
                                                    [r"(\n\s*\d+\s*\n)", r'\n\n'],
                                                    [r"\n\s*(\* *)+\s*\n", r'\n\n'],
                                                    [r"\n.*some other ugly stuff.*",
                                                    r'\n\n', re.IGNORECASE]],
                                                    remove_whitelines = True,
                                                    paragraphs_spacing = 1,
                                                    remove_hyphen_eol = True)
output = parser.run()
print(output[0])

This is a text with multiple problems. For example the latter word can be joined.

The portions of this line can be joined in a single line.

HERE ALSO IS SOME UPPERCASE CASE TEXT TO JOIN

Remove the line below:

Remove empty lines and finally separate each line with a blank line.

Below is the page number->.

Arguments

• files (list or str): Path to parse for pdf/txt operations. If a string is passed, it will be treated as a directory when mode is ‘pdf’ or ‘txt’. If a str or list is passed when mode is ‘pyobj’, it will be treated as a str/list of text files already loaded in memory in the corresponding object

• output (str): output directory

• args (list): list of tuples of the form (regex, replacement, flags). The flag can be absent.

• mode (str): ‘pdf’, ‘txt’ or ‘pyobj’ (the latter for Python lists and strings)

• default (bool): if True, perform several default cleanup operations (default)

• remove_whitelines (bool): if True, remove whitespaces

• paragraphs_spacing (int): number of newlines between paragraphs

• page_spacing (str): string to insert between pages

• remove_hyphen_eol (bool): if True, remove end of line hyphens and merge subwords

• custom_pdf_fun (Callable): custom function to parse pdf files It must accept a pdfplumber page as argument and return a text to be joined with previous pages

Current language support for the default parser

English, Spanish, German, French, Portuguese

License

© Leandro Roser, 2021. Licensed under an Apache-2 license.