Script to generate a PDF from Authorea markdown article

authorea_md_to_pdf.py

import os
import sys
import subprocess

def write_latex_preamble():
    """Write the preamble and ending lines for the tex document. 
    Abstract and title are read from the corresponding md files"""
    with open('title.md') as f:
        title = f.readline()
    with open('Abstract.md') as f:
        lines = f.readlines()
        abstract = "".join(lines)

    before = """
    \documentclass{{article}}

    \usepackage{{graphicx}}
    \usepackage[space]{{grffile}}
    \usepackage{{latexsym}}
    \usepackage{{amsfonts,amsmath,amssymb}}
    \usepackage{{url}}
    \usepackage[utf8]{{inputenc}}
    \usepackage{{fancyref}}
    \usepackage{{hyperref}}
    \hypersetup{{colorlinks=false,pdfborder={{0 0 0}},}}

    \\begin{{document}}
    \\title{{{title}}}
    \date{{\\today}}

    \\bibliographystyle{{plain}}
    \maketitle
    \\begin{{abstract}}
    {{{abstract}}}
    \end{{abstract}}
    """.format(title=title, abstract=abstract)

    after = """
    \\bibliography{bibliography/converted_to_latex.bib}

    \end{document}
    """
    with open('before.tex', 'w') as f:
        f.write(before)
    with open('after.tex', 'w') as f:
        f.write(after)

def remove_generated_files():
    """Removes files made during PDF generation"""
    cmd = "rm *.aux *.log *.bbl *.blg *.out *.tex"
    p = subprocess.call(cmd, shell=True, stderr=subprocess.PIPE)

def make_image_md(image_file):
    """Make tex file for each image"""
    caption = ""
    caption_dir = os.path.dirname(image_file)
    caption_file = os.path.join(caption_dir, "caption.md")
    caption_tex_file = os.path.join(caption_dir, "caption.tex")
    cmd = "pandoc {} -f markdown -t latex -o {}".format(caption_file, caption_tex_file)
    p = subprocess.call(cmd, shell=True, stderr=subprocess.PIPE)
    with open(caption_tex_file, 'r') as imf:
        lines = imf.readlines()
        caption += "".join(lines)

    content = """
\\begin{{figure}}[h!]
\\begin{{center}}
\includegraphics[width=0.98\columnwidth]{{{image_file}}}
\caption{{{caption}}}
\end{{center}}
\end{{figure}}
""".format(image_file=image_file, caption=caption)
    md_name = "{}.md".format(os.path.basename(image_file))
    with open(md_name,'w') as out:
        out.write(content)
    return md_name

def make_tex_output():
    """Generate pandoc command and run it to generate the tex file"""
    with open("layout.md", "r") as f:
        lines = f.readlines()
        files = []
        for l in lines:
            l = l.strip()
            if l=="Abstract.md":
                #skip Abstract.md as it's read in already
                #when generating the preabmle file
                continue
            elif "png" not in l:
                #Don't add image files in raw
                files.append(l)
            else:
                try:
                    files.append(make_image_md(l))
                except IOError:
                    continue
        cmd_files = " ".join(files)
        cmd = "pandoc {} -f markdown+tex_math_single_backslash -t latex -o out.tex".format(cmd_files)

    print cmd
    p = subprocess.call(cmd, shell=True)
    return p

def generate_pdf():
    """Concatenate generated tex files and run xelatex and bibtex"""
    cmd = "cat before.tex out.tex after.tex > final.tex && xelatex final\
        && bibtex final && xelatex final && xelatex final"
    p = subprocess.call(cmd, shell=True)
    return p

if __name__=="__main__":
    remove_generated_files()
    write_latex_preamble()
    make_tex_output()
    p = generate_pdf()
    sys.exit(p)