tabulate-data

#!/bin/bash

git log --no-abbrev --oneline --no-decorate | \
  while read commit other_stuff; do
	[[ $other_stuff =~ ^Update ]] || continue
	printf '%s:\t%s\n' commit "$commit" date "$other_stuff"
	git cat-file -p "$commit":coronavirus-cases.html 2> /dev/null | \
	  awk '
		/<td>Negative/ || \
		/<td>Currently under investigation/ || \
		/<td>Confirmed positive/ || \
		/<td>Resolved/ || \
		/<td>Deceased/ || \
		/Total number of patients approved for COVID-19 testing/ {
			key = $0
			sub(/.*<td>(<strong>)?/, "", key)
			sub(/<.*/, "", key)
			getline
			val = $0
			sub(/.*<td>(<b>)?/, "", val)
			sub(/<.*/, "", val)
			print key ":\t" val
	}'
	echo '%%'
done | \
  awk '
	BEGIN {
		fmt = "%40s %24s %7d %7d %7d %7d %7d %7d\n"
		fmt2 = fmt
		gsub(/%7d/, "%7s", fmt2)
		gsub(/%/, "%-", fmt2)
		printf(fmt2, "Commit", "Date", "Neg.", "Investi.", "Positv", \
		  "Resolv.", "Dec'"'"'d", "Tst.Ap.")
		  # These column header/footer names are trash, I am sorry
		  # about that.  Just trying to fit them into 7 characters.
	}

	function getval(  str) {
		str = $0
		sub(/.*:\t/, "", str)
		return str
	}

	/^commit:/ {
		commit = getval()
		next
	};
	/^date:/ {
		date = getval()
		sub(/^Update as of /, "", date)
		next
	}
	/^Negative:/ {
		neg = getval()
		next
	}
	/^Currently under investigation:/ {
		inv = getval()
		next
	}
	/^Confirmed positive/ {
		pos = getval()
		next
	}
	/^Resolved/ {
		res = getval()
		next
	}
	/^Deceased/ {
		dec = getval()
		next
	}
	/^Total number/ {
		tot = getval()
		next
	}

	/^%%$/ {
		printf(fmt, commit, date, neg, inv, pos, res, dec, tot)
	}' | \
  tac