From 254389803dedc7ecb41ddc159a850439e04371e6 Mon Sep 17 00:00:00 2001
From: Rinke Hoekstra <rjhoekstr@gmail.com>
Date: Wed, 12 Dec 2012 14:13:47 +0100
Subject: [PATCH] Graph file writer for Virtuoso

---
 src/util/graph_mappings.csv      | 14 +++++++
 src/util/virtuoso_graph_files.py | 71 ++++++++++++++++++++++++++++++++
 2 files changed, 85 insertions(+)
 create mode 100644 src/util/graph_mappings.csv
 create mode 100644 src/util/virtuoso_graph_files.py

diff --git a/src/util/graph_mappings.csv b/src/util/graph_mappings.csv
new file mode 100644
index 0000000..591651d
--- /dev/null
+++ b/src/util/graph_mappings.csv
@@ -0,0 +1,14 @@
+annotations-anita.ttl;http://aers.data2semantics.org/resource/graph/annotations
+annotations-richard.ttl;http://aers.data2semantics.org/resource/graph/annotations
+annotations-first-list.n3;http://aers.data2semantics.org/resource/graph/annotations
+annotations-second-list.n3;http://aers.data2semantics.org/resource/graph/annotations
+patients.nt;http://aers.data2semantics.org/resource/graph/patients
+dbpedia_3.7.owl;http://dbpedia.org
+instance_types_en.nt;http://dbpedia.org
+labels_en.nt;http://dbpedia.org
+redirects_en.nt;http://dbpedia.org
+nci_thesaurus.owl;http://ncicb.nci.nih.gov/xml/owl/EVS/Thesaurus.owl
+CTCAE_4.03_2010-06-14_v4.03.owl;http://ncicb.nci.nih.gov/xml/owl/EVS/ctcae.owl
+diseasome_dump.nt;http://www4.wiwiss.fu-berlin.de/diseasome
+drugbank_dump.nt;http://www4.wiwiss.fu-berlin.de/drugbank
+sider_dump.nt;http://www4.wiwiss.fu-berlin.de/sider
diff --git a/src/util/virtuoso_graph_files.py b/src/util/virtuoso_graph_files.py
new file mode 100644
index 0000000..51bc7c4
--- /dev/null
+++ b/src/util/virtuoso_graph_files.py
@@ -0,0 +1,71 @@
+'''
+Created on Dec 7, 2012
+
+@author: hoekstra
+'''
+from csv import reader
+from glob import glob
+import argparse
+import os.path
+import logging
+
+TRANSLATION_TABLE = 'graph_mappings.csv'
+GRAPH_BASE = 'http://aers.data2semantics.org/resource/graph/'
+
+## GLOBAL SETTINGS
+
+log = logging.getLogger(__name__)
+log.setLevel(logging.DEBUG)
+
+logHandler = logging.StreamHandler()
+logHandler.setLevel(logging.DEBUG)
+
+logFormatter = logging.Formatter('%(asctime)s - %(name)s - %(levelname)s - %(message)s')
+logHandler.setFormatter(logFormatter)
+
+log.addHandler(logHandler)
+
+if __name__ == '__main__':
+    parser = argparse.ArgumentParser()
+    
+    parser.add_argument('path',help='Path to prepare')
+    
+    args = parser.parse_args()
+    
+    path = args.path
+    if not path.endswith('/') :
+        path += '/'
+    
+    files = []
+    
+    files.extend(glob("{}*.nt".format(path)))
+    files.extend(glob("{}*.n3".format(path)))
+    files.extend(glob("{}*.ttl".format(path)))
+    files.extend(glob("{}*.owl".format(path)))
+    
+    r = reader(open(TRANSLATION_TABLE,"r"),delimiter=';')
+    
+    table = {}
+    for row in r:
+        table.setdefault(row[0], row[1])
+    
+    for p in files :
+        (dir,f) = os.path.split(p)
+        
+        if f in table:
+            graph_uri = table[f]
+            log.info("Found graph URI <{}> for '{}'".format(graph_uri,f))
+        else :
+            graph_uri = "{}{}".format(GRAPH_BASE,f)
+            log.info("Generated graph URI <{}> for '{}'".format(graph_uri,f))
+
+        
+        graph_filename = "{}.graph".format(p)
+        
+        # Open a file with the graph_filename name, if it already exists: overwrite
+        graph_file = open(graph_filename,"w")
+        
+        graph_file.write("{}\n".format(graph_uri))
+        log.info("Wrote <{}> to '{}'".format(graph_uri,graph_filename))
+    
+    
\ No newline at end of file