From edf799ed0b9b1a4858b0f841d08145b273945643 Mon Sep 17 00:00:00 2001
From: Simon Will <will@cl.uni-heidelberg.de>
Date: Wed, 26 Sep 2018 20:18:39 +0200
Subject: [PATCH] Add script for meter-based Hypotactic extraction
---
scripts/extract_verses_by_meters.py | 41 +++++++++++++++++++++++++++++
1 file changed, 41 insertions(+)
create mode 100644 scripts/extract_verses_by_meters.py
diff --git a/scripts/extract_verses_by_meters.py b/scripts/extract_verses_by_meters.py
new file mode 100644
index 0000000..a87a267
--- /dev/null
+++ b/scripts/extract_verses_by_meters.py
@@ -0,0 +1,41 @@
+#!/usr/bin/python3
+# -*- coding: utf-8 -*-
+
+import argparse
+import json
+import os
+
+import allzweckmesser as azm
+
+
+def main(hypotactic_dir, top_out_dir, meters=['hexameter']):
+ corpus = azm.corpus.HypotacticCorpus.from_directory(hypotactic_dir)
+ for document in corpus.documents:
+ doc_out_dir = os.path.join(top_out_dir, document.title)
+ os.makedirs(doc_out_dir, exist_ok=True)
+ for meter in meters:
+ verses_for_meter = [
+ azm.corpus.HypotacticLine(line).verse.to_dict()
+ for line in document.get_lines_with_meter([meter])
+ ]
+ if verses_for_meter:
+ with open(os.path.join(doc_out_dir, '{}.json'.format(meter)),
+ 'w') as f:
+ json.dump(verses_for_meter, f)
+
+
+def parse_args_and_main():
+ d = 'Extract lines from a Hypotactic corpus'
+ parser = argparse.ArgumentParser(description=d)
+ parser.add_argument('--meters', '-m', nargs='+',
+ help='Meters of the lines to extract')
+ parser.add_argument('hypotactic_dir',
+ help='Top level directory of the Hypotactic corpus')
+ parser.add_argument('top_out_dir',
+ help='Top level directory of the created JSON files')
+ args = parser.parse_args()
+ main(**vars(args))
+
+
+if __name__ == '__main__':
+ parse_args_and_main()
--
GitLab