Added a new get_cycles(recipe_xml_filepath) function in

gaworkflow.pipeline.recipe_parser.

diff --git a/gaworkflow/automation/spoolwatcher.py b/gaworkflow/automation/spoolwatcher.py
index d50eefbd6ea9b3cf11b2575ce7181d642b282b8b..abd970982263975cd23b0144024be88711f81f06 100644 (file)
--- a/gaworkflow/automation/spoolwatcher.py
+++ b/gaworkflow/automation/spoolwatcher.py
@@ -4,9 +4,9 @@ import os
 import re
 import sys
 import time
-import glob
-
+#import glob
 
+#from gaworkflow.pipeline.recipe_parser import get_cycles
 
 # this uses pyinotify
 import pyinotify
@@ -15,41 +15,6 @@ from pyinotify import EventsCodes
 from benderjab import rpc
 
 
-s_cycles = re.compile('No. Cycles: (?P<cycles>[0-9]+)')
-
-def get_cycles(run_dir="."):
-  """
-  Find the number of cycles from the Recipe*.xml file found in run_dir.
-  """
-  
-  file_path_list = glob.glob(os.path.join(run_dir, "Recipe*.xml"))
-
-  # Error handling
-  if len(file_path_list) == 0:
-    msg = "Recipe xml file not found."
-    raise IOError, msg
-
-  elif len(file_path_list) > 1:
-    msg = "%s Recipe files found, expected 1." % (len(file_path_list))
-    raise ValueError, msg
-
-  f = open(file_path_list[0], 'r')
-
-  #Find the line of the file with the cycle number in it.
-  for line in f:
-    mo = s_cycles.search(line)
-    if mo:
-      break
-
-  f.close()
-
-  # Process the line with the cycle number in it.
-  cycle_num = int(mo.group('cycles'))
-
-  return cycle_num
-
-
-
 class WatcherEvents(object):
     # two events need to be tracked
     # one to send startCopy

diff --git a/gaworkflow/pipeline/recipe_parser.py b/gaworkflow/pipeline/recipe_parser.py
new file mode 100644 (file)
index 0000000..7f5ced6
--- /dev/null
+++ b/gaworkflow/pipeline/recipe_parser.py
@@ -0,0 +1,48 @@
+from xml import sax
+
+
+def get_cycles(recipe_xml_filepath):
+  """
+  returns the number of cycles found in Recipe*.xml
+  """
+  handler = CycleXmlHandler()
+  sax.parse(recipe_xml_filepath, handler)
+  return handler.cycle_count
+
+
+
+class CycleXmlHandler(sax.ContentHandler):
+
+  def __init__(self):
+    self.cycle_count = 0
+    self.in_protocol = False
+    sax.ContentHandler.__init__(self)
+
+
+  def startDocument(self):
+    self.cycle_count = 0
+    self.in_protocol = False
+
+
+  def startElement(self, name, attrs):
+
+    #Only count Incorporations as cycles if within
+    # the protocol section of the xml document.
+    if name == "Incorporation" and self.in_protocol:
+      #print 'Found a cycle!'
+      self.cycle_count += 1
+      return
+    
+    elif name == 'Protocol':
+      #print 'In protocol'
+      self.in_protocol = True
+      return
+
+    #print 'Skipping: %s' % (name)
+    
+
+  def endElement(self, name):
+    
+    if name == 'Protocol':
+      #print 'End protocol'
+      self.in_protocol = False