erange version 4.0a dev release

[erange.git] / makerdsfromeland2.py

diff --git a/makerdsfromeland2.py b/makerdsfromeland2.py
index 317ceda90359c6cda17541af71b39e400eee8b80..66209eef973f36072ba964aac73f8579685aed99 100755 (executable)
--- a/makerdsfromeland2.py
+++ b/makerdsfromeland2.py
@@ -8,38 +8,23 @@ try:
 except:
     pass
 
-import sys, string, optparse
-from commoncode import readDataset
+import sys
+import string
+import optparse
+import ReadDataset
+from commoncode import getConfigParser, getConfigOption, getConfigIntOption, getConfigBoolOption
 
 def main(argv=None):
     if not argv:
         argv = sys.argv
 
-    verstring = "%prog: version 3.4"
+    verstring = "makerdsfromeland2: version 3.5"
     print verstring
 
     usage = "usage:  %prog label infilename outrdsfile [propertyName::propertyValue] [options]\
             \ninput reads must be sorted to properly record multireads"
 
-    parser = optparse.OptionParser(usage=usage)
-    parser.add_option("--append", action="store_false", dest="init",
-                      help="append to existing rds file [default: create new]")
-    parser.add_option("--RNA", dest="geneDataFileName",
-                      help="set data type to RNA [default: DNA]")
-    parser.add_option("--index", action="store_true", dest="doIndex",
-                      help="index the output rds file")
-    parser.add_option("--cache", type="int", dest="cachePages",
-                      help="number of cache pages to use [default: 100000")
-    parser.add_option("--olddelimiter", action="store_true", dest="useOldDelimiter",
-                      help="use : as the delimiter")
-    parser.add_option("--paired", dest="pairID",
-                      help="pairID value")
-    parser.add_option("--extended", action="store_true", dest="extended",
-                      help="use eland_extended input")
-    parser.add_option("--verbose", action="store_true", dest="verbose")
-    parser.add_option("--maxlines", type="int", dest="maxLines",
-                      help="[default: 1000000000")
-    parser.set_defaults(init=True, doIndex=False, cachePages=100000, geneDataFileName=None, useOldDelimiter=False, pairID=None, maxLines=1000000000, extended=False, verbose=False)
+    parser = getParser(usage)
     (options, args) = parser.parse_args(argv[1:])
 
     if len(args) < 3:
@@ -68,10 +53,54 @@ def main(argv=None):
     if options.geneDataFileName is not None:
         dataType = 'RNA'
 
-    makeRDSFromEland2(label, filename, outdbname, options.doIndex, delimiter, paired, options.init, options.pairID, dataType, options.geneDataFileName, options.cachePages, options.maxLines, options.extended, options.verbose)
+    makeRDSFromEland2(label, filename, outdbname, options.doIndex, delimiter, paired, options.init,
+                      options.pairID, dataType, options.geneDataFileName, options.cachePages,
+                      options.maxLines, options.extended, options.verbose)
+
+
+def getParser(usage):
+    parser = optparse.OptionParser(usage=usage)
+    parser.add_option("--append", action="store_false", dest="init",
+                      help="append to existing rds file [default: create new]")
+    parser.add_option("--RNA", dest="geneDataFileName",
+                      help="set data type to RNA [default: DNA]")
+    parser.add_option("--index", action="store_true", dest="doIndex",
+                      help="index the output rds file")
+    parser.add_option("--cache", type="int", dest="cachePages",
+                      help="number of cache pages to use [default: 100000")
+    parser.add_option("--olddelimiter", action="store_true", dest="useOldDelimiter",
+                      help="use : as the delimiter")
+    parser.add_option("--paired", dest="pairID",
+                      help="pairID value")
+    parser.add_option("--extended", action="store_true", dest="extended",
+                      help="use eland_extended input")
+    parser.add_option("--verbose", action="store_true", dest="verbose")
+    parser.add_option("--maxlines", type="int", dest="maxLines",
+                      help="[default: 1000000000")
+
+    configParser = getConfigParser()
+    section = "makerdsfromeland2"
+    init = getConfigBoolOption(configParser, section, "init", True)
+    doIndex = getConfigBoolOption(configParser, section, "doIndex", False)
+    cachePages = getConfigIntOption(configParser, section, "cachePages", 100000)
+    geneDataFileName = getConfigOption(configParser, section, "geneDataFileName", None)
+    useOldDelimiter = getConfigBoolOption(configParser, section, "useOldDelimiter", False)
+    pairID = getConfigOption(configParser, section, "pairID", None)
+    maxLines = getConfigIntOption(configParser, section, "maxLines", 1000000000)
+    extended = getConfigBoolOption(configParser, section, "extended", False)
+    verbose = getConfigBoolOption(configParser, section, "verbose", False)
+
+    parser.set_defaults(init=init, doIndex=doIndex, cachePages=cachePages,
+                        geneDataFileName=geneDataFileName, useOldDelimiter=useOldDelimiter,
+                        pairID=pairID, maxLines=maxLines, extended=extended, verbose=verbose)
+
+    return parser
+
 
+def makeRDSFromEland2(label, filename, outdbname, doIndex=False, delimiter="|", paired=False,
+                      init=True, pairID="1", dataType="DNA", geneDataFileName=None,
+                      cachePages=100000, maxLines=1000000000, extended=False, verbose=False):
 
-def makeRDSFromEland2(label, filename, outdbname, doIndex=False, delimiter="|", paired=False, init=True, pairID="1", dataType="DNA", geneDataFileName=None, cachePages=100000, maxLines=1000000000, extended=False, verbose=False):
     maxBorder = 0
     index = 0
     insertSize = 100000
@@ -104,7 +133,7 @@ def makeRDSFromEland2(label, filename, outdbname, doIndex=False, delimiter="|",
             mapDict[uname] = []
         genedatafile.close()
 
-    rds = readDataset(outdbname, init, dataType, verbose=True)
+    rds = ReadDataset.ReadDataset(outdbname, init, dataType, verbose=True)
 
     if cachePages > rds.getDefaultCacheSize():
         if init: