markdown.sh

#!/bin/bash

# uncomment the following two lines if you have both BSD sed and GNU sed installed
# this script is only tested to work with GNU sed which may have the command gsed
# shopt -s expand_aliases
# alias sed=gsed

# move the original text to a temp file that can be progressively modified
temp_file="/tmp/markdown.$$"
cat "$@" > "$temp_file"

# All of this below business is for reference-style links and images
# We need to loop across newlines and not spaces
IFS='
'
refs=$(sed -nr "/^\[.+\]: +/p" "$@")
for ref in $refs
do
    ref_id=$(echo -n "$ref" | sed -nr "s/^\[(.+)\]: .*/\1/p" | tr -d '\n')
    ref_url=$(echo -n "$ref" | sed -nr "s/^\[.+\]: (.+)/\1/p" | cut -d' ' -f1 | tr -d '\n')
    ref_title=$(echo -n "$ref" | sed -nr "s/^\[.+\]: (.+) \"(.+)\"/\2/p" | sed 's@|@!@g' | tr -d '\n')

    # reference-style image using the label
    sed -ri "s|!\[([^]]+)\]\[($ref_id)\]|<img src=\"$ref_url\" title=\"$ref_title\" alt=\"\1\" />|gI" "$temp_file"
    # reference-style link using the label
    sed -ri "s|\[([^]]+)\]\[($ref_id)\]|<a href=\"$ref_url\" title=\"$ref_title\">\1</a>|gI" "$temp_file"

    # implicit reference-style
    sed -ri "s|!\[($ref_id)\]\[\]|<img src=\"$ref_url\" title=\"$ref_title\" alt=\"\1\" />|gI" "$temp_file"
    # implicit reference-style
    sed -ri "s|\[($ref_id)\]\[\]|<a href=\"$ref_url\" title=\"$ref_title\">\1</a>|gI" "$temp_file"
done

# delete the reference lines
sed -ri "/^\[.+\]: +/d" "$temp_file"

# blockquotes
# use grep to find all the nested blockquotes
while grep '^> ' "$temp_file" >/dev/null
do
    sed -nri '
/^$/b blockquote

H
$ b blockquote
b

:blockquote
x
s/(\n+)(> .*)/\1<blockquote>\n\2\n<\/blockquote>/ # wrap the tags in a blockquote
p
' "$temp_file"

    sed -i '1 d' "$temp_file" # cleanup superfluous first line

    # cleanup blank lines and remove subsequent blockquote characters
    sed -ri '
/^> /s/^> (.*)/\1/
' "$temp_file"
done

# Setext-style headers
sed -nri '
# Setext-style headers need to be wrapped around newlines
/^$/ b print

# else, append to holding area
H
$ b print
b

:print
x
/=+$/{
s/\n(.*)\n=+$/\n<h1>\1<\/h1>/
p
b
}
/\-+$/{
s/\n(.*)\n\-+$/\n<h2>\1<\/h2>/
p
b
}
p
' "$temp_file"

sed -i '1 d' "$temp_file" # cleanup superfluous first line

# atx-style headers and other block styles
sed -ri '
/^#+ /s/ #+$// # kill all ending header characters
/^# /s/# (.*)/<h1>\1<\/h1>/g # H1
/^#{2} /s/#{2} (.*)/<h2>\1<\/h2>/g # H2
/^#{3} /s/#{3} (.*)/<h3>\1<\/h3>/g # H3
/^#{4} /s/#{4} (.*)/<h4>\1<\/h4>/g # H4
/^#{5} /s/#{5} (.*)/<h5>\1<\/h5>/g # H5
/^#{6} /s/#{6} (.*)/<h6>\1<\/h6>/g # H6

/^\*\*\*+$/s/\*\*\*+/<hr \/>/ # hr with *
/^---+$/s/---+/<hr \/>/ # hr with -
/^___+$/s/___+/<hr \/>/ # hr with _

' "$temp_file"

# unordered lists
# use grep to find all the nested lists
while grep '^[\*\+\-] ' "$temp_file" >/dev/null
do
sed -nri '
# wrap the list
/^$/b list

# wrap the li tags then add to the hold buffer
# use uli instead of li to avoid collisions when processing nested lists
/^[\*\+\-] /s/[\*\+\-] (.*)/<\/uli>\n<uli>\n\1/

H
$ b list # if at end of file, check for the end of a list
b # else, branch to the end of the script

# this is where a list is checked for the pattern
:list
# exchange the hold space into the pattern space
x
# look for the list items, if there wrap the ul tags
/<uli>/{
s/(.*)/\n<ul>\1\n<\/uli>\n<\/ul>/ # close the ul tags
s/\n<\/uli>// # kill the first superfluous closing tag
p
b
}
p
' "$temp_file"

sed -i '1 d' "$temp_file" # cleanup superfluous first line

# convert to the proper li to avoid collisions with nested lists
sed -i 's/uli>/li>/g' "$temp_file"

# prepare any nested lists
sed -ri '/^[\*\+\-] /s/(.*)/\n\1\n/' "$temp_file"
done

# ordered lists
# use grep to find all the nested lists
while grep -E '^[1-9]+\. ' "$temp_file" >/dev/null
do
sed -nri '
# wrap the list
/^$/b list

# wrap the li tags then add to the hold buffer
# use oli instead of li to avoid collisions when processing nested lists
/^[1-9]+\. /s/[1-9]+\. (.*)/<\/oli>\n<oli>\n\1/

H
$ b list # if at end of file, check for the end of a list
b # else, branch to the end of the script

:list
# exchange the hold space into the pattern space
x
# look for the list items, if there wrap the ol tags
/<oli>/{
s/(.*)/\n<ol>\1\n<\/oli>\n<\/ol>/ # close the ol tags
s/\n<\/oli>// # kill the first superfluous closing tag
p
b
}
p
' "$temp_file"

sed -i '1 d' "$temp_file" # cleanup superfluous first line

# convert list items into proper list items to avoid collisions with nested lists
sed -i 's/oli>/li>/g' "$temp_file"

# prepare any nested lists
sed -ri '/^[1-9]+\. /s/(.*)/\n\1\n/' "$temp_file"
done

# make escaped periods literal
sed -ri '/^[1-9]+\\. /s/([1-9]+)\\. /\1\. /' "$temp_file"


# code blocks
sed -nri '
# if at end of file, append the current line to the hold buffer and print it
${
H
b code
}

# wrap the code block on any non code block lines
/^\t| {4}/!b code

# else, append to the holding buffer and do nothing
H
b # else, branch to the end of the script

:code
# exchange the hold space with the pattern space
x
# look for the code items, if there wrap the pre-code tags
/\t| {4}/{
s/(\t| {4})(.*)/<pre><code>\n\1\2\n<\/code><\/pre>/ # wrap the ending tags
p
b
}
p
' "$temp_file"

sed -i '1 d' "$temp_file" # cleanup superfluous first line

# convert html characters inside pre-code tags into printable representations
sed -ri '
# get inside pre-code tags
/^<pre><code>/{
:inside
n
# if you found the end tags, branch out
/^<\/code><\/pre>/!{
s/&/\&amp;/g # ampersand
s/</\&lt;/g # less than
s/>/\&gt;/g # greater than
b inside
}
}
' "$temp_file"

# remove the first tab (or 4 spaces) from the code lines
sed -ri 's/^\t| {4}(.*)/\1/' "$temp_file"

# br tags
sed -ri '
# if an empty line, append it to the next line, then check on whether there is two in a row
/^$/ {
N
N
/^\n{2}/s/(.*)/\n<br \/>\1/
}
' "$temp_file"

# emphasis and strong emphasis
sed -nri '
# batch up the entire stream of text until a line break in the action
/^$/b emphasis

H
$ b emphasis
b

:emphasis
x
s/\*\*(.+)\*\*/<strong>\1<\/strong>/g
s/__([^_]+)__/<strong>\1<\/strong>/g
s/\*([^\*]+)\*/<em>\1<\/em>/g
s/([^\\])_([^_]+)_/\1<em>\2<\/em>/g
p
' "$temp_file"

sed -i '1 d' "$temp_file" # cleanup superfluous first line

# paragraphs
sed -nri '
# if an empty line, check the paragraph
/^$/ b para
# else append it to the hold buffer
H
# at end of file, check paragraph
$ b para
# now branch to end of script
b
# this is where a paragraph is checked for the pattern
:para
# return the entire paragraph into the pattern space
x
# look for non block-level elements, if there - print the p tags
/\n<(div|table|pre|p|[ou]l|h[1-6]|[bh]r|blockquote|li)/!{
s/(\n+)(.*)/\1<p>\n\2\n<\/p>/
p
b
}
p
' "$temp_file"

sed -i '1 d' "$temp_file" # cleanup superfluous first line

# cleanup area where P tags have broken nesting
sed -nri '
# if the line looks like like an end tag
/^<\/(div|table|pre|p|[ou]l|h[1-6]|[bh]r|blockquote)>/{
h
# if EOF, print the line
$ {
x
b done
}
# fetch the next line and check on whether or not it is a P tag
n
/^<\/p>/{
G
b done
}
# else, append the line to the previous line and print them both
H
x
}
:done
p
' "$temp_file"

# inline styles and special characters
sed -ri '
s/<(http[s]?:\/\/.*)>/<a href=\"\1\">\1<\/a>/g # automatic links
s/<(.*@.*\..*)>/<a href=\"mailto:\1\">\1<\/a>/g # automatic email address links

# inline code
s/([^\\])``+ *([^ ]*) *``+/\1<code>\2<\/code>/g
s/([^\\])`(.*)`/\1<code>\2<\/code>/g

s/!\[(.*)\]\((.*) \"(.*)\"\)/<img alt=\"\1\" src=\"\2\" title=\"\3\" \/>/g # inline image with title
s/!\[(.*)\]\((.*)\)/<img alt=\"\1\" src=\"\2\" \/>/g # inline image without title

s/\[(.*)]\((.*) "(.*)"\)/<a href=\"\2\" title=\"\3\">\1<\/a>/g # inline link with title
s/\[(.*)]\((.*)\)/<a href=\"\2\">\1<\/a>/g # inline link

# special characters
/&.+;/!s/&/\&amp;/g # ampersand
/<[\/a-zA-Z]/!s/</\&lt;/g# less than bracket

# backslash escapes for literal characters
s/\\\*/\*/g # asterisk
s/\\_/_/g # underscore
s/\\`/`/g # underscore
s/\\#/#/g # pound or hash
s/\\\+/\+/g # plus
s/\\\-/\-/g # minus
s/\\\\/\\/g # backslash
' "$temp_file"

# display and cleanup
cat "$temp_file" && rm "$temp_file"