New script

gather_pickledata.py

@@ -0,0 +1,45 @@
+#!/usr/bin/env python3
+# -*- coding: utf-8; mode: python -*-
+
+"""
+Gather some data for further conversion steps. This is originally part of fix_tei.
+"""
+
+__version__ = "1.0"
+__date__ = "20190718"
+__author__ = "kthoden@mpiwg-berlin.mpg.de"
+
+import argparse
+import logging
+import pickle
+import fix_tei
+from lxml import etree
+
+ns_tei = "http://www.tei-c.org/ns/1.0"
+NS_MAP = {"t" : ns_tei}
+
+logging.basicConfig(level=logging.DEBUG, format=' %(asctime)s - %(levelname)s - %(message)s')
+
+def main():
+    """The main bit"""
+
+    parser = argparse.ArgumentParser()
+    parser.add_argument("teifile", help="The XML file from which data is pickled.")
+    parser.add_argument("bibfile", help="The bibliography file for checking the references.")
+    args = parser.parse_args()
+
+    xml_tree = etree.parse(args.teifile)
+
+    bibdata = fix_tei.parse_bibtex(args.bibfile)
+
+    cited = xml_tree.xpath("//t:bibl/t:ref/@target", namespaces=NS_MAP)
+    used_citekeys = [fix_tei.unescape(c[1:]) for c in cited]
+    citekeys_not_in_bib = fix_tei.validate_citations(used_citekeys, bibdata)
+
+    picklefile = "output/imxml/tmp_files/data.pickle"
+    fix_tei.pickle_data(citekeys_not_in_bib, used_citekeys, picklefile)
+# def main ends here
+
+if __name__ == '__main__':
+    main()
+# finis