#! /bin/sh # Convert Twitter Archive Parser markdown into reasonable per-month HTML # # Designed to process the output of: # # https://github.com/timhutton/twitter-archive-parser # # over a 2022 Twitter Archive download # # TODO: convert 2/3/4 sets of images into something formatted more like # Twitter selector boxes # # Written by Ewen McNeill , 2023-09-25 # Updated by Ewen McNeill , 2023-09-26 # --------------------------------------------------------------------------- FILENAME="${1}" if [ -z "${FILENAME}" ]; then echo "Usage: $0 FILENAME.md" >&2 exit 1 fi BASEFILENAME="$(basename ${FILENAME})" case $(echo "${BASEFILENAME}" | cut -f 2 -d -) in 01) MONTH=January; ;; 02) MONTH=February; ;; 03) MONTH=March; ;; 04) MONTH=April; ;; 05) MONTH=May; ;; 06) MONTH=June; ;; 07) MONTH=July; ;; 08) MONTH=August; ;; 09) MONTH=September; ;; 10) MONTH=October; ;; 11) MONTH=November; ;; 12) MONTH=December; ;; esac YEAR="$(echo ${BASEFILENAME} | cut -f 1 -d -)" cat "${FILENAME}" | # Special case hash tag at start of a Tweet / Tweet line, as Markdown # will incorrectly turn those into a heading; we insert a zero width # space, and then remove it again after conversion to HTML. # sed 's/> #/> \&zwsp;#/;' | markdown_py -x fenced_code | sed 's/\&zwsp;#/#/g;' | ( # HTML Header cat < Twitter: @ewenmcneill -- ${MONTH} ${YEAR}

Twitter: @ewenmcneill -- ${MONTH} ${YEAR}

EOF # Generated content from above, with the blockquote (ie, Tweet) # moved *below* the next paragraph (ie, the tweet date). If # the Tweet is a reply, then there's a stand alone "Replying to..." # paragraph, which we also want to move the tweet date before that. # (cat; echo "") | perl -ne ' if (/

/ or /
Replying to/) { while (! /<\/blockquote>/) { $_ .= <>; } $blockquote = $_; $_ = <>; print $_; print $blockquote; } else { print; }' | # Rewrite the Twitter Icon / Link HTML to inject an anchor tag # for the tweet number, and a link to that anchor tag # sed 's/^
\(]*media\/tweet.ico[^>]*>\)\(.*status\/\)\([0-9][0-9]*\)"\(.*\)<\/p>/
\1<\/a>\2\3"\4 (#<\/a>)<\/p>/;' | # Rewrite Tweeted images to have a class we can select on # sed 's/^
\(\1/;'; # HTML Footer cat < EOF )