Does your favorite web novel use dashes to start dialouge instead of bounding it with qoutes? This program will fix it! Created to process this book/series on AO3: https://archiveofourown.org/works/51538063

dash_to_quote_dialouge.py

from bs4 import BeautifulSoup
from pathlib import Path

DASHES = ("—", "–", "-")
# Generate dashes_spaced
DASHES_SPACED = []
for dash in DASHES:
    # Handle all of the different space types
    DASHES_SPACED.append(f" {dash} ")
    DASHES_SPACED.append(f" {dash}\u00A0")
    DASHES_SPACED.append(f"\u00A0{dash} ")
    DASHES_SPACED.append(f"\u00A0{dash}\u00A0")
    DASHES_SPACED.append(f"\u202F{dash} ")
    DASHES_SPACED.append(f" {dash}\u202F")
    DASHES_SPACED.append(f"\u202F{dash}\u202F")
    DASHES_SPACED.append(f"\u2007{dash} ")
    DASHES_SPACED.append(f" {dash}\u2007")
    DASHES_SPACED.append(f"\u2007{dash}\u2007")

def main(book: Path, out: Path) -> None:
    with book.open('r') as f:
        soup = BeautifulSoup(f.read(), "html.parser")

    for p in soup.find_all("p"):
        # First dash isn't counted so assume 1 exists initially
        dash_count = 1
        if not p.text: 
            continue
        if not p.text.strip(): 
            continue
        # Perserve em 
        if p.em and not p.em.text:
            ps = p.em.text.strip()
            if ps:
                if ps[0] in DASHES:
                    # Count dashes to see if end quote is needed. Even num = no end quote
                    for dash in DASHES_SPACED:
                        dash_count += ps.count(dash)
                    p.em.string = '"' + ps[1:].strip() + '"'*(not dash_count % 2 == 0)
                    # Do in-paragraph dash subsitution
                    # This works becuase the in-paragraph ones are space coated
                    for dash in DASHES_SPACED:
                        p.em.string = p.em.string.replace(dash, '"')
        # No em with text
        else:
            ps = p.text.strip()
            if ps[0] in DASHES:
                # Count dashes to see if end quote is needed. Even num = no end quote
                for dash in DASHES_SPACED:
                    dash_count += ps.count(dash)
                p.string = '"' + ps[1:].strip() + '"'*(not dash_count % 2 == 0)
                # Do in-paragraph dash subsitution
                # This works becuase the in-paragraph ones are space coated
                for dash in DASHES_SPACED:
                    p.string = p.string.replace(dash, '"')

    with out.open('w') as f:
        f.write(str(soup))

if __name__ == "__main__":
    import sys, argparse 
    parser = argparse.ArgumentParser(description="Does your favorite web novel use dashes to start dialouge instead of bounding it with qoutes? \
                                     This program will fix it!")
    parser.add_argument("book", type=str, help="Path to input html book")
    parser.add_argument("out", type=str, help="Output location the program will write to")
    args = parser.parse_args()

    main(Path(args.book), Path(args.out))