Replace longtable with table in LaTeX generated by Quarto and/or Pandoc

postprocess_latex.py

#!/usr/bin/env python

"""
This script patches the .tex generated by Quarto to replace usage of \longtable with vanilla tables, and then compiles that.

This is necessary because longtable doesn't support two-column layouts, and even after applying the hacky solution, the PDF output is mangled in such a way that entire pages are skipped.
"""

import os
import shutil
import subprocess
from pathlib import Path


def is_big_table(h):
    # I only had one big table, but otherwise you'll want some more sophisticated logic.
    return h == "\\hypertarget{tbl-big}{}\n"


def patch_table(in_filename, out_filename):
    with open(in_filename, "r") as f:
        lines = f.readlines()

    out_lines = []
    inside_table = False
    inside_caption = False
    inside_duplicate_header = False
    caption_count_opens = 0
    table_definition = ""
    in_big_table = False

    for line in lines:
        if inside_table:
            if inside_duplicate_header:
                out_lines.append(f"% {line}")

                if line == "\endhead\n":
                    pass
                elif line == "\endlastfoot\n":
                    inside_duplicate_header = False

            elif inside_caption:
                out_lines.append(line)
                caption_count_opens += line.count("{")
                caption_count_opens -= line.count("}")

                if caption_count_opens <= 0:
                    inside_caption = False
                    out_lines.append(table_definition)

            else:
                if line.startswith("\\caption{"):
                    inside_caption = True
                    out_lines.append(line)
                    caption_count_opens = line.count("{")
                    caption_count_opens -= line.count("}")

                elif line == "\endfirsthead\n":
                    out_lines.append(f"% {line}")
                    inside_duplicate_header = True

                elif line.endswith("\\noalign{}\n"):
                    out_lines.append(line.replace("\\noalign{}", "%\\noalign"))

                elif line.startswith("\\end{longtable}"):
                    inside_table = False

                    out_lines.append(line.replace("{longtable}", "{tabular}"))

                    if in_big_table:
                        out_lines.append("\\end{table*}\n")
                    else:
                        out_lines.append("\\end{table}\n")
                else:
                    out_lines.append(line)

        elif line.startswith("\\begin{longtable}"):
            inside_table = True

            if in_big_table:
                out_lines.append("\\begin{table*}\n")
            else:
                out_lines.append("\\begin{table}\n")

            table_definition = line.replace("{longtable}", "{tabular}")

        elif line.startswith("\\hypertarget{"):
            out_lines.append(line)

            in_big_table = is_big_table(line)

        else:
            out_lines.append(line)

    with open(out_filename, "w") as f:
        f.writelines(out_lines)


def run_latex(base_name):
    os.chdir("paper")
    subprocess.check_call(["pdflatex", base_name])
    subprocess.check_call(["bibtex", base_name])
    subprocess.check_call(["pdflatex", base_name])
    subprocess.check_call(["pdflatex", base_name])
    os.chdir("..")


def replace_current_output():
    current_pdf = Path("_output/paper/paper.pdf")
    backup_pdf = current_pdf.with_name("paper.bak.pdf")
    shutil.move(current_pdf, backup_pdf)
    replacement_pdf = Path("paper/main2.pdf")
    shutil.move(replacement_pdf, current_pdf)


def run():
    shutil.copytree(
        "_output/paper/main_files/figure-latex",
        "paper/main_files/figure-latex",
        dirs_exist_ok=True,
    )

    patch_table("paper/main.tex", "paper/main2.tex")

    run_latex("main2")

    replace_current_output()


if __name__ == "__main__":
    run()