From: Diane Trout <diane@caltech.edu>
Date: Fri, 27 Mar 2015 18:03:45 +0000 (-0700)
Subject: record file size information for a file in the triple store
X-Git-Url: http://woldlab.caltech.edu/gitweb/?p=htsworkflow.git;a=commitdiff_plain;h=a1833e98a7d806c3c4e29e9cfc6b6283c62c2ab8

record file size information for a file in the triple store
---

diff --git a/htsworkflow/submission/submission.py b/htsworkflow/submission/submission.py
index 897053f..ed4d898 100644
--- a/htsworkflow/submission/submission.py
+++ b/htsworkflow/submission/submission.py
@@ -123,6 +123,7 @@ class Submission(object):
         # add file specific information
         fileNode = self.make_file_node(pathname, an_analysis)
         self.add_md5s(filename, fileNode, analysis_dir)
+        self.add_file_size(filename, fileNode, analysis_dir)
         self.add_fastq_metadata(filename, fileNode)
         self.add_label(file_type, fileNode, libNode)
         self.model.add_statement(
@@ -168,6 +169,13 @@ class Submission(object):
             self.model.add_statement(
                 RDF.Statement(fileNode, dafTermOntology['md5sum'], md5))
 
+    def add_file_size(self, filename, fileNode, analysis_dir):
+        LOGGER.debug("Updating file size")
+        submission_pathname = os.path.join(analysis_dir, filename)
+        file_size = os.stat(submission_pathname).st_size
+        self.model.add_statement(
+            RDF.Statement(fileNode, dafTermOntology['file_size'], toTypedNode(file_size)))
+
     def add_fastq_metadata(self, filename, fileNode):
         # How should I detect if this is actually a fastq file?
         try: