#!/bin/bash
#    Get a reasonably accurate word count for posts in the current month

usage="$0 [options] [month] print word counts for posts in a given month
  Options:
	-d | -C  -- change directory
	-h   	 -- help
	-j   	 -- use Jekyll-style dates (yyyy-mm-dd)
	-m month -- set month
	-n 	 -- just listed files, no monthly stats
	-y year	 -- set year
	-Y 	 -- count words in the entire year
	-r 	 -- remove (don't count) postng statistics
	-v	 -- verbose: print word count for each post
"

### Default values

m=$(date "+%_m"); this_month=$m	# has to be space-padded to keep it from looking like octal
y=$(date "+%Y"); this_year=$y
allYear=false

### get options

while :; do
    case $1 in
	-d | -C) # set directory
	    shift; POST_ARCHIVE=$1
	    shift
	    ;;
	-h) # help
	    echo "$usage"
	    exit
	    ;;
	-v) # -v lists all files with their word counts
	    verbose=-------
	    shift
	    ;;
	-r) # remove stats
	    remove_stats=1
	    shift
	    ;;
	-m) # set month
	    shift; m=$1
	    shift
	    ;;
	-n) # -n -- don't count words in the month
	    m=
	    shift
	    ;;
	-y) # set year
	    shift; y=$1
	    shift
	    ;;
	-Y) # count words in year to date
	    shift; allYear=true
	    ;;
	-j) # -j uses Jekyll-format filenames:  yyyy-mm-dd-SLuG <<TODO
	    sep="-"
	    shift
	    ;;
	*) break
	   ;;
    esac
done

sep=${sep:-/}

if [[ ! -z ${POST_ARCHIVE} ]]; then
    cd ${POST_ARCHIVE}; 
elif [[ ! -d $y ]]; then
   cd ~/vv/users/$USER/Private/Journals
fi
   
### Stats header:  special label for NaBloPoMo (November)

stats=Posting
if [[ -z $m ]]; then
    stats=Draft
elif (( $m == 11 )); then
    stats=NaBloPoMo
else
    stats=Posting
fi

echo "<pre>$stats stats:"

expand () {
    ls ${1}* 2>/dev/null
}

### count-words GLOB [sep]
count-words () {
    files="$@"

    total=0
    count=0
    missed=0
    last=0
    for f in $files ; do
	# the initial `sed` removes the email-style header from posts.
	#   (FIXME: removes thefirst paragraph if there isn't a header)
	# the final `tr` removes the dashes that pandoc uses as list bullets
	if [[ ! -z $remove_stats ]] && [[ "$f" != "${f%.html}" ]]; then
	    # optionally remove stats ("<pre>....stats:" - "</pre>") from html files
	    wc=$(sed -e '1,/^ *$/d' $f |
		     sed -e '/^<pre>[a-zA-Z]* stats\:$/,/^<\/pre>/d' | 
		     pandoc -f html -t plain | tr -- " -" "  " | wc -w)
	else
	    wc=$(sed -e '1,/^ *$/ d' $f |
		     pandoc -f html -t plain | tr -- " -" "  " | wc -w)
	fi
	total=$(( $total + $wc))
	count=$(( $count + 1))
	if [[ $f = [0-9][0-9][0-9][0-9]/[0-9][0-9]/[0-9][0-9]--* ]]; then
	    day=`echo $f | head -c10 | tail -c2 | sed s/^0//`
	    if (( $day > $last + 1 )); then
		missed=$(( $missed + 1 ))
	    fi
	    last=$day
	fi
	if [[ ! -z $verbose ]]; then
	    echo -e "`printf %7d $wc`"  $f
	fi
    done
    if (($count == 0)); then
	avg=0
    else
	avg=$(($total / $count))
    fi
    [[ -z $verbose ]] || echo $verbose
}

if $allYear; then
    echo all of $y by month:
    yearWords=0
    yearCount=0
    for month in `seq 1 12`; do
	(( $this_month < $month )) && (($this_year == $y)) && break
	(( $month == 1 )) || [[ -z $verbose ]] || echo # blank line between months
	mm=$(printf "%02d" $month)
	count-words $(expand $y$sep$mm$sep)
	yearWords=$(( $yearWords + $total ))
	yearCount=$(( $yearCount + $count ))
	echo -e  "`printf %7d $total`" words in "`printf %2d $count`" posts  \
	     in $y/$mm "(average $avg/post)"
    done
    avg=$(($yearWords / $yearCount))
    echo '---------------------------------'
    echo -e  "`printf %7d $yearWords`" words in $yearCount posts total \
	 in $y "(average $avg/post)"
    echo '</pre>'
    exit
fi

if [[ -z $m ]]; then		# no month stats
    count-words $*
    echo -e  "`printf %7d $total`" words total "(average $avg/file)"
    echo '</pre>'
    exit
fi

# Count words and posts so far this month

mm=$(printf "%02d" $m)
count-words $(expand $y$sep$mm$sep) $*

# Add missing days after the last date seen.
if (( $this_month == $m && $this_year == $y)); then    # We're doing this month
    end=$(date +%e)
    in="this month"
else
    end=$(date -d "$y/$m/1 +1month -1day" +%e)
    in=$(date -d "$y/$m/1" +"in %B")
    if (( $this_year != $y )); then in="$in of $y"; fi
fi
missed_days=$(( $missed + $end - $last ))

echo -e  "`printf %7d $total`" words in $count posts $in "(average $avg/post)"

if (( $m == $this_month && $y == $this_year)); then
    # we've already seen all the posts; clear $verbose so we don't see
    # today's listed a second time.
    verbose=
    count-words $(expand `date "+$y$sep$mm$sep%d"`)
    if [ $count == 1 ]; then s=" "; else s=s; fi
    echo -e  "`printf %7d $total`" words in $count post$s today
fi
if (( $missed_days != 0 )); then
    if [ $missed_days == 1 ]; then s=" "; else s=s; fi
    echo -e "`printf %7d $missed_days`" day$s with no posts
fi
echo '</pre>'