Fix roam files exported to org

org-roam-rescue.py

#!/usr/bin/env python

import urllib as ul
import re
import sys
import os.path
#Pandoc turns this: [foo]([[My Foo Page]])
#Into this: [[file:%5B%5BMy%20Foo%20Page%5B%5B][foo]]
#We want this: [[file:My Foo Page.org][foo]]

#So we process and "un-escape the escaped characters".
#Then we add .org to links and remove the superfluous brackets
#This may break some links with %s in them - sorry.
#Also breaks file links to non-org files.

slink_re = r"\[\[([^\[]*?)\]\]" #unaliased link
alink_re = r"\[\[file\:(.*?)\]\[(.*?)\]\]"

def slinkfixr(matchobj):
    if matchobj.group(1)[:4] == "http":
        return matchobj.group(0) #the entire thing
    return "[[file:" + matchobj.group(1) + "][" + matchobj.group(1) + "]]"
#No need to "unquote" these.
#We will add the ".org" in the next step, so don't do it here.

def alinkfixr(matchobj):
    return "[[file:" + ul.unquote(matchobj.group(1)).strip("[]") + ".org][" + matchobj.group(2) + "]]"

def fix_links(text):
    t1 = re.sub(slink_re, slinkfixr, text)
    t2 = re.sub(alink_re, alinkfixr, t1)
    return t2


def run(filename):
    f = open(filename, "r+")
    old = f.read()
    new = fix_links(old)
    print(new)
    f.seek(0)
    f.write(new)

def dryrun(filename):
    f = open(filename, "r")
    old = f.read()
    new = fix_links(old)
    print(new)

for filename in sys.argv[1:]:
    if os.path.isfile(filename):
        run(filename)