Break out tables from a tex or latex file into separate tables .tex files..

latex_replace_text_by_input_file.py

#!python
"""
Break out tables from a tex or latex file into separate tables .tex files.. 

To see help
	>> python latex_replace_text_by_input_file.py -h
To run user either of:
	>> python latex_replace_text_by_input_file.py -f epr-main
	>> python latex_replace_text_by_input_file.py -f epr-main -d _SUBMIT_

@license: APACHE 2.0
@author:  Zlatko K. Minev (2021) 
"""


import argparse
import re
from pathlib import Path


def latex_replace_text_by_input_file(text:str, FOLDER:str,
                                     start  = r'\begin{tabular}',
                                     end    = r'\end{tabular}' ,
                                     prefix = r'table') -> str:
    """Replaces text, such as tables of the file by input files
    and saves the input files.
    """

    # without r, Use the syntax "\\" within the string literal to represent a single backslash.
    start2= start.replace('\\', r"\\") 
    end2  = end.replace(  '\\', r"\\")
    pattern = f"{start2}(.*?){end2}"

    substrings = re.findall(pattern, text, re.DOTALL) # -> List[Str]
    print(f'Found {len(substrings)} substrings')

    text2 = text
    for n, substring in enumerate(substrings):
        full_substring = f"{start}{substring}{end}"
        fn = f"{prefix}{n}.tex"
        (Path(FOLDER) / fn).write_text(full_substring)
        text2 = text2.replace(full_substring, "\input{"+fn+"}")
    
    return text2

my_args = None 
# Can use the following to test 
# my_args= ["-h"] 
# my_args= ["-f", "epr-main"] 
# my_args= ["-f", "epr-main", "-d", "_SUBMIT_"] 
if __name__ == '__main__':
    parser = argparse.ArgumentParser(description = """
    Zlatko's code to break out tables from a tex or latex file into separate tables .tex files.. 

    """)
    parser.add_argument("--file", "-f", type=str, required=True, help="filename to process, without extension .tex")
    parser.add_argument("--dir",  "-d", type=str, required=False, help="folder for file", default='')
    args = parser.parse_args(my_args)
    print(args)
    
    ### INPUTS ################################ 
    FOLDER=args.dir
    filename=f'{args.file}.tex'# if FOLDER else f'{args.file}.tex'

    ### CODE ##################################
    path = Path(FOLDER) / filename
    print("")
    print('path.resolve()  = ', path.resolve())
    print('path.is_file()  = ', path.is_file())
    text  = path.read_text()
    text2 = latex_replace_text_by_input_file(text, FOLDER) 
    path.write_text(text2)
    print("")