Check SAM file problems

dark/alignments.py

@@ -1,6 +1,6 @@
 import re
 
-from dark.taxonomy import LineageFetcher
+# from dark.taxonomy import LineageFetcher
 from dark.filter import TitleFilter
 from dark.score import HigherIsBetterScore
 

dark/sam/alignments.py

@@ -24,7 +24,8 @@ class SamReadsAlignments(ReadsAlignments):
         files.
     """
     def __init__(self, reads, samFilename):
-        self.samFilename = samFilename
+        if checkSAMfile(samFilename):
+            self.samFilename = samFilename
         self.reads = reads
         # Prepare application parameters in order to initialize self.
         self.head = self._convertSamHeaderToDict()
@@ -87,9 +88,8 @@ def _getReader(self, filename, applicationParams, scoreClass):
         @param filename: The C{str} file name holding the SAM records.
         @param scoreClass: A class to hold and compare scores (see scores.py).
         """
-        if checkSAMfile(filename):
-            return SAMRecordsReader(self.samFilename, self.head,
-                                    self.scoreClass)
+        return SAMRecordsReader(self.samFilename, self.head,
+                                self.scoreClass)
 
     def iter(self):
         """

dark/sam/conversion.py

@@ -247,7 +247,7 @@ def _lineToHSP(self, line):
                 if 'read reverse strand' in explain_sam_flags(flag):
                     # What it says on the tin...
                     pass
-
+                # TODO add in readStartinsbjct etc
                 hsp = HSP(score, readStart=readStart, readEnd=readEnd,
                           subjectStart=subjStart, subjectEnd=subjEnd)
                 return hsp

dark/sam/hacks.py

@@ -4,6 +4,63 @@
 import subprocess as sp
 
 
+def checkSAMfile(samFilename):
+    """
+    Checks that the file inputted as a SAM file is indeed a SAM file.
+    @param samFilename: a C{str} of a SAM file.
+    """
+    headerLines = 0
+    with open(samFilename) as samFile:
+        for line in samFile:
+            if not str(line):
+                raise ValueError('SAM file %s was empty.' % samFilename)
+            elif line[0] == '@':
+                headerLines += 1
+            else:
+                elements = line.strip().split()
+                assert len(elements) > 10, ('SAM file %s does not contain '
+                                            'at least 11 fields.'
+                                            % samFilename)
+    assert headerLines > 0, ('SAM file %s does not contain header.'
+                             % samFilename)
+    return True
+
+
+def checkFASTAfile(fastaFilename):
+    """
+    Checks that the file inputted as a FASTA file is indeed a FASTA file.
+    @param fastaFilename: a C{str} of a FASTA file.
+    """
+    with open(fastaFilename) as fastaFile:
+        # Make into an iterable so can compare two lines.
+        fastaFile = iter(fastaFile)
+        for line in fastaFile:
+            assert line[0] == '>', 'FASTA file does not begin with a title'
+            line = next(fastaFile)
+            # Invalid if two lines begin with >
+            assert line[0] != '>' and line, 'Invalid FASTA file format'
+
+
+def checkFASTQfile(fastqFilename):
+    """
+    TODO: Finish
+    Checks that the file inputted as a FASTQ file is indeed a FASTQ file.
+    @param fastqFilename: a C{str} of a FASTQ file.
+    """
+    with open(fastqFilename) as fastqFile:
+        fastqFile = iter(fastqFile)
+        for line in fastqFile:
+            header = line[1:]
+            assert line[0] == '@', 'Invalid header of entry: %s' % header
+            line = next(fastqFile)
+            assert line, 'Empty raw sequence in entry: %s' % header
+            line = next(fastqFile)
+            assert line[0] == '+', 'Invalid third line of entry: %s' % header
+            line = next(fastqFile)
+            assert line, 'Empty quality score in entry: %s' % header
+    return True
+
+
 def samSubtract(samFile, outFile):
     """
     Takes a SAM file, makes a set of the seqids of unaligned sequences.
@@ -95,56 +152,3 @@ def findMD(samFile, fastaFile):
         sp.Popen(['samtools', 'fillmd', '-S', samFile, fastaFile, '>',
                  samFileNew], stderr=sp.PIPE)
         return samFileNew
-
-
-def checkSAMfile(samFilename):
-    """
-    Checks that the file inputted as a SAM file is indeed a SAM file.
-    @param samFilename: a C{str} of a SAM file.
-    """
-    headerLines = 0
-    with open(samFilename) as samFile:
-        for line in samFile:
-            if line[0] == '@':
-                headerLines += 1
-            else:
-                elements = line.strip().split()
-                assert len(elements) > 10, ('SAM file does not contain at '
-                                            'least 11 fields.')
-    assert headerLines > 0, 'SAM file does not contain header.'
-    return True
-
-
-def checkFASTAfile(fastaFilename):
-    """
-    Checks that the file inputted as a FASTA file is indeed a FASTA file.
-    @param fastaFilename: a C{str} of a FASTA file.
-    """
-    with open(fastaFilename) as fastaFile:
-        # Make into an iterable so can compare two lines.
-        fastaFile = iter(fastaFile)
-        for line in fastaFile:
-            assert line[0] == '>', 'FASTA file does not begin with a title'
-            line = next(fastaFile)
-            # Invalid if two lines begin with >
-            assert line[0] != '>' and line, 'Invalid FASTA file format'
-
-
-def checkFASTQfile(fastqFilename):
-    """
-    TODO: Finish
-    Checks that the file inputted as a FASTQ file is indeed a FASTQ file.
-    @param fastqFilename: a C{str} of a FASTQ file.
-    """
-    with open(fastqFilename) as fastqFile:
-        fastqFile = iter(fastqFile)
-        for line in fastqFile:
-            header = line[1:]
-            assert line[0] == '@', 'Invalid header of entry: %s' % header
-            line = next(fastqFile)
-            assert line, 'Empty raw sequence in entry: %s' % header
-            line = next(fastqFile)
-            assert line[0] == '+', 'Invalid third line of entry: %s' % header
-            line = next(fastqFile)
-            assert line, 'Empty quality score in entry: %s' % header
-    return True

test/sam/test_alignments.py

@@ -47,7 +47,7 @@ def testEmptySAMInput(self):
         mockOpener = mockOpen()
         with patch('__builtin__.open', mockOpener, create=True):
             reads = Reads()
-            error = "SAM file 'file.SAM' was empty."
+            error = "SAM file file.SAM was empty."
             self.assertRaisesRegexp(
                 ValueError, error, SamReadsAlignments, reads, 'file.SAM')
 
@@ -60,9 +60,9 @@ def testNonSAMInput(self):
         mockOpener = mockOpen(read_data='not SAM\n')
         with patch('__builtin__.open', mockOpener, create=True):
             reads = Reads()
-            error = ("No header lines in file.SAM")
+            error = "SAM file file.SAM does not contain at least 11 fields."
             self.assertRaisesRegexp(
-                ValueError, error, SamReadsAlignments, reads, 'file.SAM')
+                AssertionError, error, SamReadsAlignments, reads, 'file.SAM')
 
     def testApplicationParams(self):
         """