#!/usr/bin/env python3

# ox-hugo handles links that are like [[Text][foo]] - i.e. a link to
# 'foo' which displays as 'Text'.  Org also has links like [[foo]]
# which behave identically to [[foo][foo]], i.e. they use the link
# itself as the text to show.  ox-hugo doesn't handle these, however -
# they simply disappear in the generated Markdown.
#
# This script just uses a regex to turn [[foo]] to [[foo][foo]].

import re
import sys

# This is a bit dense, but: it matches any text wrapped in [[ and ]]
# provided that text contains no square brackets.  That text itself is
# then group 1.
link_re = re.compile(r"\[\[([^][]+)\]\]")

bare_link_re = re.compile(r"[^][](http\S*(jpg|jpeg|JPG|JPEG|gif|GIF|png|PNG))")

proper_link_re = re.compile(r"\[\[([^][]+)\]\[([^][]+)\]\]")

target_re = re.compile(r"<<([^\<\>]+)>>")

seen = set()
targets = set()
seen_line = {}

def fix_link(m):
    t = m.group(1)
    seen.add(t)
    return "[[" + t + "][" + t + "]]"

def fix_image_link(m):
    t = m.group(1)
    seen.add(t)
    descr = "image"
    return "[[" + t + "][" + descr + "]]"

for n,line in enumerate(sys.stdin):
    for g1,_ in proper_link_re.findall(line):
        #print("FOUND: {}".format(g1))
        seen.add(g1)
        seen_line[g1] = n+1
    for g1 in link_re.findall(line):
        #print("FOUND: {}".format(g1))
        seen.add(g1)
        seen_line[g1] = n-1
    for g1 in target_re.findall(line):
        targets.add(g1)
    f1 = link_re.sub(fix_link, line)
    f2 = bare_link_re.sub(fix_image_link, f1)
    sys.stdout.write(f2)

sys.stdout.write("* Missing links\n")
for link in seen.difference(targets):
    if not link.startswith("http"):
        if link in seen_line:
            line = seen_line[link]
            sys.stdout.write("- line {}: <<{}>>\n".format(line, link))