From ef82770cbefb4ceaf894d0f37cd8468c9054b86a Mon Sep 17 00:00:00 2001
From: Nick White <git@njw.name>
Date: Wed, 5 Jun 2019 17:37:23 +0100
Subject: Rename bookgraphv2.sh to the canonical bookgraph

Add word count to the graph. Use a scaled figure so it's easy to compare with the confidence.
---
 bookgraph      | 45 +++++++++++++++++++++++++++++++++++++++++++++
 bookgraphv2.sh | 40 ----------------------------------------
 2 files changed, 45 insertions(+), 40 deletions(-)
 create mode 100755 bookgraph
 delete mode 100644 bookgraphv2.sh

diff --git a/bookgraph b/bookgraph
new file mode 100755
index 0000000..8010bdc
--- /dev/null
+++ b/bookgraph
@@ -0,0 +1,45 @@
+#!/bin/sh
+usage="$0 bookdir
+
+Creates a graph showing the average confidence and a (scaled) word
+count of each page in a book. The word count is scaled (divided by
+10 and plus 50) to make it easy to compare to the confidence, by
+generally occupying a similar scale."
+
+test $# -ne 1 && echo "$usage" && exit 1
+
+if ! command -v pgconf > /dev/null ; then
+	echo "Error: no pgconf tool found"
+	exit 1
+fi
+
+t=`mktemp`
+
+find "$1/best" -maxdepth 1 -type f -name '*hocr' | while read i; do
+	c=`pgconf "$i"`
+	n=`basename "$i" .hocr | sed 's/_.*//'`
+	w=`grep ocrx_word "$i" | wc -l`
+	printf '%s\t%d\t%d\n' "$n" "$c" "$w" >> "$t"
+done
+
+r=`readlink -f "$1"`
+b=`basename "$r"`
+
+gp=`mktemp`
+printf '
+
+set style data lines
+set title "Book Confidence for %s"
+set xlabel "Page Number"
+set ylabel "Page Confidence %%"
+set mxtics
+set terminal png truecolor size 4600,1700
+set output "%s"
+plot "%s" using 1:2 with lines title "Confidence",\\
+     "" using 1:2:(sprintf("%%d", $1)) with labels point pt 2 notitle,\\
+     "" using 1:($3/10+50) with lines title "Number of words (scaled)"
+quit
+' "$b" "$1/${b}_bookgraph.png" "$t" >> "$gp"
+gnuplot "$gp"
+
+rm -f "$t" "$gp"
diff --git a/bookgraphv2.sh b/bookgraphv2.sh
deleted file mode 100644
index 3f6cfaa..0000000
--- a/bookgraphv2.sh
+++ /dev/null
@@ -1,40 +0,0 @@
-#!/bin/sh
-usage="$0 bookdir
-
-Creates a graph showing the average confidence of each page in a book."
-
-test $# -ne 1 && echo "$usage" && exit 1
-
-if ! command -v pgconf > /dev/null ; then
-	echo "Error: no pgconf tool found"
-	exit 1
-fi
-
-t=`mktemp`
-
-find "$1/best" -maxdepth 1 -type f -name '*hocr' | while read i; do
-	c=`pgconf "$i"`
-	n=`basename "$i" .hocr | sed 's/_.*//'`
-	printf '%s\t%d\n' "$n" "$c" >> "$t"
-done
-
-r=`readlink -f "$1"`
-b=`basename "$r"`
-
-gp=`mktemp`
-printf '
-
-set style data lines
-set title "Book Confidence for %s"
-set xlabel "Page Number"
-set ylabel "Page Confidence %%"
-set mxtics
-set terminal png truecolor size 4600,1700
-set output "%s"
-plot "%s" using 1:2 with lines title "Confidence",\\
-     "" using 1:2:(sprintf("%%d", $1)) with labels point pt 2 notitle
-quit
-' "$b" "$1/${b}_bookgraph.png" "$t" >> "$gp"
-gnuplot "$gp"
-
-rm -f "$t" "$gp"
-- 
cgit v1.2.1-24-ge1ad