initial support for icalendar splitting

src/filters/rclexecm.py

@@ -1,3 +1,5 @@
+#!/usr/bin/env python
+
 ###########################################
 ## Generic recoll multifilter communication code
 import sys
@@ -6,12 +8,15 @@ import os
 class RclExecM:
     def __init__(self):
         self.myname = os.path.basename(sys.argv[0])
+        self.mimetype = ""
 
     def rclog(self, s, doexit = 0, exitvalue = 1):
         print >> sys.stderr, "RCLMFILT:", self.myname, ":", s
         if doexit:
             exit(exitvalue)
-
+    # Our worker sometimes knows the mime types of the data it sends
+    def setmimetype(self, mt):
+        self.mimetype = mt
     def readparam(self):
         s = sys.stdin.readline()
         if s == '':
@@ -49,6 +54,10 @@ class RclExecM:
             print "Ipath:", len(ipath)
             sys.stdout.write(ipath)
 
+        if len(self.mimetype):
+            print "Mimetype:", len(self.mimetype)
+            sys.stdout.write(self.mimetype)
+
         # If we're at the end of the contents, say so
         if iseof:
             print "Eof: 0"

src/filters/rclics

@@ -0,0 +1,50 @@
+#!/usr/bin/env python
+
+import rclexecm
+from icalendar import Calendar, Event
+
+
+class IcalExtractor:
+    def __init__(self, em):
+        self.file = ""
+        self.em = em
+        em.setmimetype("text/plain")
+
+    def extractone(self, index):
+        if index >= len(self.contents):
+            return(False, "", "", True)
+        docdata = self.contents[index].as_string()
+	#self.em.rclog(docdata)
+        eof = (self.currentindex >= len(self.contents) -1)
+        return (True, docdata, str(index), eof)
+
+    ###### File type handler api, used by rclexecm ---------->
+    def openfile(self, params):
+        self.file = params["filename:"]
+        try:
+            self.cal = Calendar.from_string(open(self.file,'rb').read())
+        except:
+            return False
+        # Skip the top level object
+        self.currentindex = 1
+        self.contents = self.cal.walk()
+        return True
+
+    def getipath(self, params):
+        try:
+            index = int(params["ipath:"])
+        except:
+            return False
+        return self.extractone(index)
+        
+    def getnext(self, params):
+        if self.currentindex >= len(self.contents):
+            #em.rclog("getnext: EOF hit")
+            return (False, "", "", 1)
+        else:
+            ret= self.extractone(self.currentindex)
+            self.currentindex += 1
+            return ret
+
+e = rclexecm.RclExecM()
+e.mainloop(IcalExtractor(e))