From 2468672f8f2492f67c0efb3871e2759ee208b837 Mon Sep 17 00:00:00 2001
From: Martijn Vermaat <martijn@vermaat.name>
Date: Mon, 17 Feb 2014 10:06:34 +0100
Subject: [PATCH] Handle NCBI Entrez response validation errors
---
mutalyzer/Retriever.py | 24 ++++++++++++++++++++----
mutalyzer/parsers/genbank.py | 18 ++++++++++++++----
2 files changed, 34 insertions(+), 8 deletions(-)
diff --git a/mutalyzer/Retriever.py b/mutalyzer/Retriever.py
index 5ca0bb48..77da54e8 100644
--- a/mutalyzer/Retriever.py
+++ b/mutalyzer/Retriever.py
@@ -590,8 +590,16 @@ class GenBankRetriever(Retriever):
query = "%s[Gene] AND %s[Orgn]" % (gene, organism)
try:
handle = Entrez.esearch(db = "gene", term = query)
- searchresult = Entrez.read(handle)
- handle.close()
+ try:
+ searchresult = Entrez.read(handle)
+ except Entrez.Parser.ValidationError:
+ self._output.addMessage(__file__, -1, 'INFO',
+ 'Error reading Entrez esearch result.')
+ self._output.addMessage(__file__, 4, 'ERETR',
+ 'Could not search for gene %s.' % gene)
+ return None
+ finally:
+ handle.close()
except (IOError, urllib2.HTTPError, HTTPException) as e:
self._output.addMessage(__file__, -1, 'INFO',
'Error connecting to Entrez esearch: %s' % str(e))
@@ -604,8 +612,16 @@ class GenBankRetriever(Retriever):
for i in searchresult["IdList"] : # Inspect all results.
try:
handle = Entrez.esummary(db = "gene", id = i)
- summary = Entrez.read(handle)
- handle.close()
+ try:
+ summary = Entrez.read(handle)
+ except Entrez.Parser.ValidationError:
+ self._output.addMessage(__file__, -1, 'INFO',
+ 'Error reading Entrez esummary result.')
+ self._output.addMessage(__file__, 4, 'ERETR',
+ 'Could not get mapping information for gene %s.' % gene)
+ return None
+ finally:
+ handle.close()
except (IOError, urllib2.HTTPError, HTTPException) as e:
self._output.addMessage(__file__, -1, 'INFO',
'Error connecting to Entrez esummary: %s' % str(e))
diff --git a/mutalyzer/parsers/genbank.py b/mutalyzer/parsers/genbank.py
index e8283663..867fa78f 100644
--- a/mutalyzer/parsers/genbank.py
+++ b/mutalyzer/parsers/genbank.py
@@ -138,15 +138,25 @@ class GBparser():
return link.protein_accession
handle = Entrez.esearch(db = "nucleotide", term = transcriptAcc)
- result = Entrez.read(handle)
- handle.close()
+ try:
+ result = Entrez.read(handle)
+ except Entrez.Parser.ValidationError:
+ # Todo: Log this error.
+ return None
+ finally:
+ handle.close()
transcriptGI = result["IdList"][0]
handle = Entrez.elink(dbfrom = "nucleotide", db = "protein",
id = transcriptGI)
- result = Entrez.read(handle)
- handle.close()
+ try:
+ result = Entrez.read(handle)
+ except Entrez.Parser.ValidationError:
+ # Todo: Log this error.
+ return None
+ finally:
+ handle.close()
if not result[0]["LinkSetDb"] :
queries.update_transcript_protein_link(transcriptAcc)
--
GitLab