From 58521d1c5461ab98bc63415ad26e91d54612e43e Mon Sep 17 00:00:00 2001
From: Nick White <git@njw.me.uk>
Date: Wed, 16 Nov 2011 21:46:59 +0000
Subject: Vastly improve ocrpdf script, and create ocrtxt script

---
 extras/mkocrtxt.sh | 26 ++++++++++++++++++++++++++
 1 file changed, 26 insertions(+)
 create mode 100644 extras/mkocrtxt.sh

(limited to 'extras/mkocrtxt.sh')

diff --git a/extras/mkocrtxt.sh b/extras/mkocrtxt.sh
new file mode 100644
index 0000000..02f7146
--- /dev/null
+++ b/extras/mkocrtxt.sh
@@ -0,0 +1,26 @@
+#!/bin/sh
+# See COPYING file for copyright and license details.
+#
+# Makes a text file with text extracted by tesseract
+#
+# Note: Unfortunately tesseract works much better if one first
+#       makes the image to be OCRed significantly larger. This
+#       script therefore temporarily creates a larger file to
+#       feed to tesseract.
+
+for i in `ls *png`
+do
+	# create a much bigger version of the page image
+	width=`identify "$i" | awk '{print $3}' | sed 's/x.*//'`
+	bigwidth=`expr $width \* 4`
+	convert "$i" -geometry ${bigwidth}x "$i.big.png"
+
+	# scan the page image
+	tesseract "$i.big.png" "$i" 2>&1 | sed '/Tesseract Open Source OCR Engine/d'
+
+	# combine the page text with the rest of the book
+	cat "$i.txt" >> book.txt
+
+	# remove working files
+	rm -f "$i.big.png" "$i.txt"
+done
-- 
cgit v1.2.3