Changeset 27

Timestamp:

Aug 11, 2010, 4:05:09 PM (10 years ago)

Author:

j@…

bzr:base-revision:

j@dannynavarro.net-20100806120031-x0349ul02lytyb8h

bzr:committer:

Danny Navarro <j@dannynavarro.net>

bzr:file-ids:

mzcms/parsers.py parsers.py-20100806092910-g1sxvv1o5b9umkof-1

bzr:mapping-version:

v4

bzr:repository-uuid:

724254b2-fbe6-419d-9466-c04ef4c9d29d

bzr:revision-id:

j@dannynavarro.net-20100806125108-r2yigtsr1ksh7h7e

bzr:revno:

27

bzr:revprop:branch-nick:

trunk

bzr:root:

trunk

bzr:timestamp:

2010-08-06 14:51:08.160000086 +0200

bzr:user-agent:

bzr2.1.2+bzr-svn1.0.3

svn:original-date:

2010-08-06T12:51:08.160000Z

Message:

specified default mgf regex to parse

File:

• trunk/mzcms/parsers.py (modified) (5 diffs)

trunk/mzcms/parsers.py

@@ -6 +6 @@
 import re
 import csv
+import urllib
 from optparse import OptionParser
 from string import Template
@@ -30 +31 @@
                         )
 #fraction_regex = re.compile(r'fraction%3a%20(\d+)')
-RAWFN_REGEX = re.compile(r'raw%3a%20(.+raw)')
 
 class DatParser(object):
     """Mascot Dat Parser
     """
-    def __init__(self, decoy_str, scans_in_title=False):
+    def __init__(self,
+                 scan_str=r'FinneganScanNumber%3a%20(\d+)',
+                 rawfn_str=r'RawFile%3a%20(.+raw)',
+                 decoy_str=r'^IPI:REV_:IPI',
+                 contaminant_str=r'^IPI:CON_:IPI'
+                 ):
+        self.scan_regex = re.compile(scan_str)
         self.decoy_regex = re.compile(decoy_str)
-        self.scans_in_title = scans_in_title
-        if scans_in_title:
-            self.scan_regex = re.compile(r'scan%3a%20(\d+)')
+        self.rawfn_regex = re.compile(rawfn_str)
+        self.contaminant_regex = re.compile(contaminant_str)
 
     def _parse_spectra_fn(self, datfile):
@@ -115 +120 @@
                 #current_id_mgf["Fraction"] = fraction
                 try:
-                    rawfn = re.search(RAWFN_REGEX, line).group(1)
+                    quoted_rawfn = re.search(rawfn_regex, line).group(1)
                 except AttributeError:
                     sys.exit("It seems there is no raw: field in TITLE")
-                rawfn = rawfn.replace('%2e', '.')
+                rawfn = urllib.unquote(quoted_rawfn)
                 current_id_mgf["RawFile"] = rawfn
                 if self.scans_in_title:
@@ -196 +201 @@
     spectra = spectra_factory()
     psms = psms_factory()
-    dat_parser = DatParser(decoy_str, scans_in_title)
+    dat_parser = DatParser()
     for dat_path in dat_paths:
         with open(dat_path) as dat_file:
@@ -226 +231 @@
                       "--decoy-str",
                       dest="decoy_str",
-                      default="^IPI:REV_:IPI",
+                      default=r'^IPI:REV_:IPI',
                       help="1 to parse the scan number from the TITLE field, "
                            "0 to parse the scan number from the SCAN field "