From d4199d0ddd1f9384f422dcfd28aadc2b72ce3f8d Mon Sep 17 00:00:00 2001
From: cneud <clemens.neudecker@gmail.com>
Date: Wed, 30 Oct 2019 15:02:29 +0100
Subject: [PATCH] add Python script for converting PAGE-XML to TSV

---
 tools/page2tsv.py | 12 ++++++++++++
 1 file changed, 12 insertions(+)
 create mode 100644 tools/page2tsv.py

diff --git a/tools/page2tsv.py b/tools/page2tsv.py
new file mode 100644
index 0000000..34e2ba3
--- /dev/null
+++ b/tools/page2tsv.py
@@ -0,0 +1,12 @@
+import sys
+import codecs
+import xml.etree.ElementTree as ET
+
+sys.stdout = codecs.getwriter('utf-8')(sys.stdout.buffer, 'strict')
+tree = ET.parse(sys.argv[1])
+xmlns = tree.getroot().tag.split('}')[0].strip('{')
+for words in tree.findall('.//{%s}Word' % xmlns):
+	for word in words.findall('.//{%s}Unicode' % xmlns):
+		text = word.text
+		for coords in words.findall('.//{%s}Coords' % xmlns):
+			sys.stdout.write('0\t'+text+'\tO\tO\t'+coords.attrib['points']+'\n')
\ No newline at end of file