]> git.notmuchmail.org Git - notmuch/blobdiff - notmuch
notmuch: refactor stuff into a Notmuch class
[notmuch] / notmuch
diff --git a/notmuch b/notmuch
index 2f65ae49e631d38b2e6eb069812788d44a762e25..46f65791cc1618f316446e651294e24115eb239c 100755 (executable)
--- a/notmuch
+++ b/notmuch
@@ -1,8 +1,346 @@
 #!/usr/bin/env python
-"""This is a notmuch implementation in python. It's goal is to allow running the test suite on the cnotmuch python bindings."""
-import sys
+"""This is a notmuch implementation in python. 
+It's goal is to allow running the test suite on the cnotmuch python bindings.
+
+This "binary" honors the NOTMUCH_CONFIG environmen variable for reading a user's
+notmuch configuration (e.g. the database path).
+
+   (c) 2010 by Sebastian Spaeth <Sebastian@SSpaeth.de>
+               Jesse Rosenthal <jrosenthal@jhu.edu>
+   This code is licensed under the GNU GPL v3+.
+"""
+from __future__ import with_statement # This isn't required in Python 2.6
+import sys, os, re, stat
+from cnotmuch.notmuch import Database, Query, NotmuchError, STATUS
+PREFIX=re.compile('(\w+):(.*$)')
 #-------------------------------------------------------------------------
-HELPTEXT="""The notmuch mail system.
+def quote_query_line(argv):
+   #mangle arguments wrapping terms with spaces in quotes
+   for i in xrange(0,len(argv)):
+      if argv[i].find(' ') >= 0:
+         #if we use prefix:termWithSpaces, put quotes around term
+         m = PREFIX.match(argv[i])
+         if m:
+            argv[i] = '%s:"%s"' % (m.group(1), m.group(2))
+         else:
+            argv[i] = '"'+argv[i]+'"'
+   return ' '.join(argv)
+
+#-------------------------------------------------------------------------
+class Notmuch:
+
+    def __init__(self):
+        self._config = None
+
+    def cmd_usage(self):
+       """Print the usage text and exits"""
+       data={}
+       names = self.get_user_email_addresses()
+       data['fullname']   =names[0] if names[0] else 'My Name'
+       data['mailaddress']=names[1] if names[1] else 'My@email.address'
+       print (Notmuch.USAGE % data)
+
+    def cmd_new(self):
+        """Run 'notmuch new'"""
+        #get the database directory
+        db = Database(mode=Database.MODE.READ_WRITE)
+        path = db.get_path()
+
+        (added, moved, removed) = self._add_new_files_recursively(path, db)
+        print (added, moved, removed)
+
+    def cmd_help(self, subcmd=None):
+       """Print help text for 'notmuch help'"""
+       if len(subcmd) > 1:
+          print "Help for specific commands not implemented"
+          return
+
+       print (Notmuch.HELPTEXT)
+
+    def _get_user_notmuch_config(self):
+        """Returns the ConfigParser of the user's notmuch-config"""
+       # return the cached config parser if we read it already
+       if self._config is not None:
+           return self._config
+
+       from ConfigParser import SafeConfigParser
+       config = SafeConfigParser()
+       conf_f = os.getenv('NOTMUCH_CONFIG',
+                          os.path.expanduser('~/.notmuch-config'))
+       config.read(conf_f)
+       self._config = config
+       return config
+
+    def _add_new_files_recursively(self, path, db):
+        """:returns: (added, moved, removed)"""
+        print "Enter add new files with path %s" % path
+        (added, moved, removed) = (0,)*3
+
+        try:
+           #get the Directory() object for this path
+           db_dir = db.get_directory(path)
+        except NotmuchError:
+           #Occurs if we have wrong absolute paths in the db, for example
+           return (0,0,0)
+
+
+        #for folder in subdirs:
+        #   (new_added, new_moved, new_removed) = \
+        #           self._add_new_files_recursively(
+        #                  os.path.join(db_dir.path, folder), db)
+        #   added += new_added
+        #   moved += new_moved
+        #   removed += new_removed
+
+        #TODO, retrieve dir mtime here and store it later
+        #as long as Filenames() does not allow multiple iteration, we need to
+        #use this kludgy way to get a sorted list of filenames
+        #db_files is a list of subdirectories and filenames in this folder
+        db_files = set()
+        db_folders = set()
+        for subdir in db_dir.get_child_directories():
+           db_folders.add(os.path.normpath(subdir))
+        for file in db_dir.get_child_files():
+           db_files.add(file)
+
+        fs_files = set(os.listdir(db_dir.path))
+
+        #list of folders in both db and fs. Just descend into dirs
+        for fs_file in (fs_files | db_folders):
+           absfile = os.path.normpath(os.path.join(db_dir.path, fs_file))
+           if os.path.isdir(absfile):
+              #This is a directory
+              if fs_file in ['.notmuch','tmp','.']:
+                 continue
+              self._add_new_files_recursively(absfile, db)
+           # we are not interested in anything but directories here
+
+        #list of files and folders in the fs, but not the db
+        for fs_file in (fs_files - db_files):
+           absfile = os.path.normpath(os.path.join(db_dir.path, fs_file))
+           statinfo = os.stat(absfile)
+
+           if stat.S_ISDIR(statinfo.st_mode):
+              #This is a directory
+              if fs_file in ['.notmuch','.']:
+                 continue
+              print "descending into %s" % absfile
+              #self._add_new_files_recursively(absfile, db)
+           elif stat.S_ISLNK(statinfo.st_mode):
+              print ("%s is a symbolic link (%d)" % (absfile, statinfo.st_mode))
+           else:
+              print "This file needs to be added %s" % (absfile)
+              #TODO
+              #(msg, status) = db.add_message(os.path.join(db_dir.path, db_file))
+              #if status == STATUS.DUPLICATE_MESSAGE_ID:
+           #   #This message was already in the database, continue with next one
+           #   continue
+
+        #list of files and folders in the database, but not the filesystem
+        for db_file in (db_files - fs_files):
+           absfile = os.path.normpath(os.path.join(db_dir.path, db_file))
+           statinfo = os.stat(absfile)
+
+           if stat.S_ISDIR(statinfo.st_mode):
+              #This is a directory
+              if db_file in ['.notmuch', '.']:
+                 continue
+              print "descending into %s" % absfile
+              self._add_new_files_recursively(absfile, db)
+              #TODO, is there no way to REMOVE a directory entry from the db?
+           else:
+              #remove a mail message from the db
+              print ("%s is not on the fs anymore. Delete" % absfile)
+              status = db.remove_message(absfile)
+              if status == STATUS.SUCCESS:
+                 # we just deleted the last reference, so this was a remove
+                 removed += 1
+              elif status == STATUS.DUPLICATE_MESSAGE_ID:
+                 # The filename exists already somewhere else, so this is a move
+                 moved += 1
+              else: 
+                 print "This must not happen. %s " % (absfile)
+                 sys.exit(1)
+
+        return (added, moved, removed)
+        #Read the mtime of a directory from the filesystem
+        #
+        #* Call :meth:`Database.add_message` for all mail files in
+        #  the directory
+
+        #* Call notmuch_directory_set_mtime with the mtime read from the 
+        #  filesystem.  Then, when wanting to check for updates to the
+        #  directory in the future, the client can call :meth:`get_mtime`
+        #  and know that it only needs to add files if the mtime of the 
+        #  directory and files are newer than the stored timestamp.
+
+    def get_user_email_addresses(self):
+        """ Reads a user's notmuch config and returns his email addresses as
+       list (name, primary_address, other_address1,...)"""
+        import email.utils
+
+       #read the config file
+       config = self._get_user_notmuch_config()
+
+       if not config.has_option('user','name'): name = ""
+       else:name = config.get('user','name')
+
+       if not config.has_option('user','primary_email'): mail = ""
+       else:mail = config.get('user','primary_email')
+
+       if not config.has_option('user','other_email'): other = []
+       else:other = config.get('user','other_email').rstrip(';').split(';')
+
+       other.insert(0, mail)
+       other.insert(0, name)
+       return other
+
+    def quote_msg_body(self, oldbody ,date, from_address):
+        """Transform a mail body into a quoted text, 
+        starting with On blah, x wrote:
+
+        :param body: a str with a mail body
+        :returns: The new payload of the email.message()
+        """
+        from cStringIO import StringIO
+
+        #we get handed a string, wrap it in a file-like object
+        oldbody = StringIO(oldbody)
+        newbody = StringIO()
+
+        newbody.write("On %s, %s wrote:\n" % (date, from_address))
+
+        for line in oldbody:
+           newbody.write("> " + line)
+
+        return newbody.getvalue()
+
+    def format_reply(self, msgs):
+        """Gets handed Messages() and displays the reply to them
+
+        This is pretty ugly and hacky. It tries to mimic the "real"
+        notmuch output as much as it can to pass the test suite. It
+        could deserve a healthy bit of love.  It is also buggy because
+        it returns after the first message it has handled."""
+        import email
+
+        for msg in msgs:
+           f = open(msg.get_filename(),"r")
+           reply = email.message_from_file(f)
+
+           #handle the easy non-multipart case:
+           if not reply.is_multipart():
+              reply.set_payload(self.quote_msg_body(reply.get_payload(),
+                                                   reply['date'],reply['from']))
+           else:
+              #handle the tricky multipart case
+              deleted = ""
+              """A string describing which nontext attachements that
+                 have been deleted"""
+              delpayloads = []
+              """A list of payload indices to be deleted"""
+
+              payloads = reply.get_payload()
+
+              for i, part in enumerate(payloads):
+
+                 mime_main = part.get_content_maintype()
+                 if mime_main not in ['multipart', 'message', 'text']:
+                    deleted += "Non-text part: %s\n" % (part.get_content_type())
+                    payloads[i].set_payload("Non-text part: %s" % (part.get_content_type()))
+                    payloads[i].set_type('text/plain')
+                    delpayloads.append(i)
+                 elif mime_main == 'text':
+                    payloads[i].set_payload(self.quote_msg_body(payloads[i].get_payload(),reply['date'],reply['from']))
+                 else:
+                    #TODO handle deeply nested multipart messages
+                    sys.stderr.write ("FIXME: Ignoring multipart part. Handle me\n")
+
+              # Delete those payloads that we don't need anymore
+              for i in reversed(sorted(delpayloads)):
+                 del payloads[i]
+
+        #Back to single- and multipart handling
+        my_addresses = self.get_user_email_addresses()
+        used_address = None
+        # filter our email addresses from all to: cc: and bcc: fields
+        # if we find one of "my" addresses being used, 
+        # it is stored in used_address
+        for header in ['To', 'CC', 'Bcc']:
+           if not header in reply:
+              #only handle fields that exist
+              continue
+           addresses = email.utils.getaddresses(reply.get_all(header,[]))
+           purged_addr = []
+           for name, mail in addresses:
+              if mail in my_addresses[1:]:
+                 used_address = email.utils.formataddr((my_addresses[0],mail))
+              else:
+                 purged_addr.append(email.utils.formataddr((name,mail)))
+
+           if len(purged_addr):
+              reply.replace_header(header, ", ".join(purged_addr))
+           else: 
+              #we deleted all addresses, delete the header
+              del reply[header]
+
+        # Use our primary email address to the From
+        # (save original from line, we still need it)
+        orig_from = reply['From']
+        del reply['From']
+        reply['From'] = used_address if used_address \
+            else email.utils.formataddr((my_addresses[0],my_addresses[1]))
+      
+        #reinsert the Subject after the From
+        orig_subject = reply['Subject']
+        del reply['Subject']
+        reply['Subject'] = 'Re: ' + orig_subject
+
+        # Calculate our new To: field
+        new_to = orig_from
+        # add all remaining original 'To' addresses
+        if 'To' in reply:
+           new_to += ", " + reply['To']
+        del reply['To']
+        reply.add_header('To', new_to)
+
+        # Add our primary email address to the BCC
+        new_bcc = my_addresses[1]
+        if reply.has_key('Bcc'):
+           new_bcc += ', '  + reply['Bcc']
+           del reply['Bcc']
+        reply['Bcc'] = new_bcc
+
+        # Set replies 'In-Reply-To' header to original's Message-ID
+        if reply.has_key('Message-ID') :
+           del reply['In-Reply-To']
+           reply['In-Reply-To'] = reply['Message-ID']
+
+        #Add original's Message-ID to replies 'References' header.
+        if reply.has_key('References'):
+           ref = reply['References'] + ' ' +reply['Message-ID']
+        else:
+           ref = reply['Message-ID']
+        del reply['References']
+        reply['References'] = ref
+      
+        # Delete the original Message-ID.
+        del(reply['Message-ID'])
+
+        # filter all existing headers but a few and delete them from 'reply'
+        delheaders = filter(lambda x: x not in ['From','To','Subject','CC',
+                                                'Bcc','In-Reply-To',
+                                                'References','Content-Type'],
+                            reply.keys())
+        map(reply.__delitem__, delheaders)
+
+        # TODO: OUCH, we return after the first msg we have handled rather than
+        # handle all of them
+        #return resulting message without Unixfrom
+        return reply.as_string(False)
+
+
+    HELPTEXT="""The notmuch mail system.
 
 Usage: notmuch <command> [args...]
 
@@ -54,17 +392,16 @@ Use "notmuch help <command>" for more details on each command.
 And "notmuch help search-terms" for the common search-terms syntax.
 """
 
-#TODO: replace the dynamic pieces
-USAGE="""Notmuch is configured and appears to have a database. Excellent!
+    USAGE="""Notmuch is configured and appears to have a database. Excellent!
 
 At this point you can start exploring the functionality of notmuch by
 using commands such as:
 
        notmuch search tag:inbox
 
-       notmuch search to:"Sebastian Spaeth"
+       notmuch search to:"%(fullname)s"
 
-       notmuch search from:"Sebastian@SSpaeth.de"
+       notmuch search from:"%(mailaddress)s"
 
        notmuch search subject:"my favorite things"
 
@@ -79,59 +416,261 @@ And don't forget to run "notmuch new" whenever new mail arrives.
 
 Have fun, and may your inbox never have much mail.
 """
+
+# MAIN
 #-------------------------------------------------------------------------
 if __name__ == '__main__':
 
    # Handle command line options
-   # No option 
+   #-------------------------------------
+   # No option given, print USAGE and exit
    if len(sys.argv) == 1:
-      print USAGE
-
+      Notmuch().cmd_usage()
+   #-------------------------------------
    elif sys.argv[1] == 'setup':
-       """ Interactively setup notmuch for first use. """
+       """Interactively setup notmuch for first use."""
        print "Not implemented."
-
+   #-------------------------------------
+   elif sys.argv[1] == 'new':
+       """Check for new and removed messages."""
+       Notmuch().cmd_new()
+   #-------------------------------------
    elif sys.argv[1] == 'help':
-       if len(sys.argv) == 2: print HELPTEXT
-       else: print "Not implemented"
-
+      """Print the help text"""
+      Notmuch().cmd_help(sys.argv[1:])
+   #-------------------------------------
+   elif sys.argv[1] == 'part':
+      db = Database()
+      query_string = ''
+      part_num=0
+      first_search_term = None
+      for (i, arg) in enumerate(sys.argv[1:]):
+         if arg.startswith('--part='):
+            part_num_str=arg.split("=")[1]
+            try:
+               part_num = int(part_num_str)
+            except ValueError:
+               # just emulating behavior
+               sys.exit()
+         elif not arg.startswith('--'):
+              #save the position of the first sys.argv that is a search term
+              first_search_term = i+1
+
+      if first_search_term:
+          #mangle arguments wrapping terms with spaces in quotes
+          querystr = quote_query_line(sys.argv[first_search_term:])
+      
+      qry = Query(db,querystr)
+      msgs = qry.search_messages()
+      msg_list = []
+      for m in msgs:
+        msg_list.append(m)
+        
+      if len(msg_list) == 0:
+        sys.exit()
+      elif len(msg_list) > 1:
+        raise Exception("search term did not match precisely one message")
+      else:
+        msg = msg_list[0]
+        print(msg.get_part(part_num))
+   #-------------------------------------
+   elif sys.argv[1] == 'search':
+      db = Database()
+      query_string = ''
+      sort_order="newest-first"
+      first_search_term = None
+      for (i, arg) in enumerate(sys.argv[1:]):
+         if arg.startswith('--sort='):
+            sort_order=arg.split("=")[1]
+            if not sort_order in ("oldest-first", "newest-first"):
+               raise Exception("unknown sort order")
+         elif not arg.startswith('--'):
+              #save the position of the first sys.argv that is a search term
+              first_search_term = i+1
+
+      if first_search_term:
+          #mangle arguments wrapping terms with spaces in quotes
+          querystr = quote_query_line(sys.argv[first_search_term:])
+
+      qry = Query(db,querystr)
+      if sort_order == "oldest-first":
+        qry.set_sort(Query.SORT.OLDEST_FIRST)
+      else:
+        qry.set_sort(Query.SORT.NEWEST_FIRST)
+      t = qry.search_threads()
+
+      for thread in t:
+        print(str(thread))
+
+   #-------------------------------------
+   elif sys.argv[1] == 'show':
+      entire_thread = False
+      db = Database()
+      out_format="text"
+      querystr=''
+      first_search_term = None
+
+      #ugly homegrown option parsing
+      #TODO: use OptionParser
+      for (i, arg) in enumerate(sys.argv[1:]):
+         if arg == '--entire-thread':
+              entire_thread = True
+         elif arg.startswith("--format="):
+              out_format = arg.split("=")[1]
+             if out_format == 'json':
+                  #for compatibility use --entire-thread for json
+                  entire_thread = True
+              if not out_format in ("json", "text"):
+                  raise Exception("unknown format")
+         elif not arg.startswith('--'):
+              #save the position of the first sys.argv that is a search term
+              first_search_term = i+1
+
+      if first_search_term:
+          #mangle arguments wrapping terms with spaces in quotes
+          querystr = quote_query_line(sys.argv[first_search_term:])
+
+      t = Query(db,querystr).search_threads()
+
+      first_toplevel=True
+      if out_format.lower()=="json":
+         sys.stdout.write("[")
+
+      for thrd in t:
+         msgs = thrd.get_toplevel_messages()
+
+         if not first_toplevel:
+            if out_format.lower()=="json":
+               sys.stdout.write(", ")
+
+         first_toplevel = False
+
+         msgs.print_messages(out_format, 0, entire_thread)
+
+      if out_format.lower() == "json":
+         sys.stdout.write("]")
+      sys.stdout.write("\n")
+
+   #-------------------------------------
+   elif sys.argv[1] == 'reply':
+      db = Database()
+      if len(sys.argv) == 2:
+         #no search term. abort
+         print("Error: notmuch reply requires at least one search term.")
+         sys.exit()
+
+      #mangle arguments wrapping terms with spaces in quotes
+      querystr = quote_query_line(sys.argv[2:])
+      msgs = Query(db,querystr).search_messages()
+      print (Notmuch().format_reply(msgs))
+
+   #-------------------------------------
+   elif sys.argv[1] == 'count':
+      if len(sys.argv) == 2:
+         #no further search term, count all
+         querystr=''
+      else:
+         #mangle arguments wrapping terms with spaces in quotes
+         querystr = quote_query_line(sys.argv[2:])
+      print(Database().create_query(querystr).count_messages())
+      
+   #-------------------------------------
+   elif sys.argv[1] == 'tag':
+      #build lists of tags to be added and removed
+      add, remove = [], []
+      while not sys.argv[2]=='--' and \
+            (sys.argv[2].startswith('+') or sys.argv[2].startswith('-')):
+         if sys.argv[2].startswith('+'):
+            #append to add list without initial +
+            add.append(sys.argv.pop(2)[1:])
+         else:
+            #append to remove list without initial -
+            remove.append(sys.argv.pop(2)[1:])
+      #skip eventual '--'
+      if sys.argv[2]=='--': sys.argv.pop(2)
+      #the rest is search terms
+      querystr = quote_query_line(sys.argv[2:])
+      db = Database(mode=Database.MODE.READ_WRITE)
+      m  = Query(db,querystr).search_messages()
+      for msg in m:
+         #actually add and remove all tags
+         map(msg.add_tag, add)
+         map(msg.remove_tag, remove)
+   #-------------------------------------
+   elif sys.argv[1] == 'search-tags':
+      if len(sys.argv) == 2:
+         #no further search term
+         print("\n".join(Database().get_all_tags()))
+      else:
+         #mangle arguments wrapping terms with spaces in quotes
+         querystr = quote_query_line(sys.argv[2:])
+         db = Database()
+         m  = Query(db,querystr).search_messages()
+         print("\n".join([t for t in m.collect_tags()]))
+   #-------------------------------------
+   elif sys.argv[1] == 'dump':
+      #TODO: implement "dump <filename>"
+      if len(sys.argv) == 2:
+         f = sys.stdout
+      else:
+         f = open(sys.argv[2],"w")
+      db = Database()
+      q = Query(db,'')
+      q.set_sort(Query.SORT.MESSAGE_ID)
+      m = q.search_messages()
+      for msg in m:
+         f.write("%s (%s)\n" % (msg.get_message_id(), msg.get_tags()))
+   #-------------------------------------
+   elif sys.argv[1] == 'restore':
+      import re
+      if len(sys.argv) == 2:
+         print("No filename given. Reading dump from stdin.")
+         f = sys.stdin
+      else:
+         f = open(sys.argv[2],"r")
+      #split the msg id and the tags
+      MSGID_TAGS = re.compile("(\S+)\s\((.*)\)$")
+      db = Database(mode=Database.MODE.READ_WRITE)
+
+      #read each line of the dump file
+      for line in f:
+         m = MSGID_TAGS.match(line)
+         if not m:
+            sys.stderr.write("Warning: Ignoring invalid input line: %s" % 
+                             line)
+            continue
+         # split line in components and fetch message
+         msg_id = m.group(1)
+         new_tags= set(m.group(2).split())
+         msg    = db.find_message(msg_id)
+
+         if msg == None:
+            sys.stderr.write(
+               "Warning: Cannot apply tags to missing message: %s\n" % id)
+            continue
+
+         #do nothing if the old set of tags is the same as the new one
+         old_tags = set(msg.get_tags())
+         if old_tags == new_tags: continue
+
+         #set the new tags
+         msg.freeze()
+         #only remove tags if the new ones are not a superset anyway
+         if not (new_tags > old_tags): msg.remove_all_tags()
+         for tag in new_tags: msg.add_tag(tag)
+         msg.thaw()
+            
+   #-------------------------------------
    else:
        # unknown command
        print "Error: Unknown command '%s' (see \"notmuch help\")" % sys.argv[1]
 
 
    #TODO: implement
-   """ 
-new    [--verbose]
-
-       Find and import new messages to the notmuch database.
-
-search [options...] <search-terms> [...]
-
-       Search for messages matching the given search terms.
-
+   """
+setup
+new
 show   <search-terms> [...]
-
-       Show all messages matching the search terms.
-
-count  <search-terms> [...]
-
-       Count messages matching the search terms.
-
 reply  [options...] <search-terms> [...]
-
-       Construct a reply template for a set of messages.
-
-tag    +<tag>|-<tag> [...] [--] <search-terms> [...]
-
-       Add/remove tags for all messages matching the search terms.
-
-dump   [<filename>]
-
-       Create a plain-text dump of the tags for each message.
-
 restore        <filename>
-       search-tags     [<search-terms> [...] ]
-
-               List all tags found in the database or matching messages.
    """