Some observations

[uliw]

This looks pretty great!

running it on typical file I noticed the following

1. section titles sometimes miss a newline, so that the whole paragraph is rendered as a heading
2. complex tables fail to convert but produce no warning. Would it be possible to simply wrap them in a latex block, rather than converting them to an org-table? I.e.,

   #+BEGIN_EXPORT latex
   ....
   .....
   #+END_EXPORT

   This could be extended to any non-standard environment

3. complex cite commands do not convert (e.g., \cite[e.g.,][]{hansen2012}). Would it be possible that any latex command that does not parse defaults to a newline with #+latex: foo....

Would it be possible that rather than parsing all the pre-amble commands to simply collect them in a l_headers.tex file, and include this with #+latex_header: \input{l_headers.tex}?

[uliw]

this will parse the latex headers. Writing to a separate file would be cleaner but works in a pinch.

def read_header(lines, ofl) -> tuple([str, bool]):
    """header lines are all lines between \documentclass and \begin{document}
    precede all read lines with #+latexheader:

    parameters:
    lines: io_wrapper
        ofl: iowrapper

    return values:
        header: bool value
    """

    nl: str = next(lines)
    header: bool = True

    if "documentclass" in nl:
        header = False
        while "begin{document}" not in nl:
            if nl.strip() != "":
                ofl.write(f"#+latexheader: {nl}")
            nl: str = next(lines)

    return nl, header

call just before you read the next line

if header:
                 header = read_header(lines, ofl)
line = next(lines)

[uliw]

and here is a snippet that deals with tables etc:

ef special_environments(nl, lines, ofl):
    """Some environments a better handled by a latex export block
    parameters:
    nl: current line
    lines: io_wrapper
    ofl: iowrapper
    """

    se: list = ["table", "tabular", "tabularx", "longtable"]

    for e in se:
        if f"begin{{{e}}}" in nl:
            ofl.write("#+BEGIN_EXPORT latex\n")
            while f"end{{{e}}}" not in nl:
                if nl.strip() != "":
                    ofl.write(nl)
                    nl: str = next(lines)
            ofl.write("#+END_EXPORT\n")
        break

call as:

line = next(lines)
                special_environments(line, lines, ofl)
                line = line_latex_to_orgmode(line)

[MarcvdSluys]

Hi @uliw,

Thank you for your feedback. I must admit I haven't spent a great deal of time on these - rather hacked something together until my needs were satisfied for the (sometimes single) case at hand. However, that doesn't mean I'm not open to improvements :-)

Is it much work to post these updates as Pull requests?

[uliw]

will do later. Do you recall why you match section commands, including the newline character? The following will remove the newline at the end of the section command line = re.sub(r"^ *\\section[* ]*\{([^}]*)\} *\n", r"\n* \1", line) whereas this line = re.sub(r"^ *\\section[* ]*\{([^}]*)\}", r"\n* \1", line) results in the correct translation.

[uliw]

Not much of the original code survived, and my purpose (round-trip editing) differs from yours. I made my own version. Feel free to link to https://github.com/uliw/l2org