Extend handling of GMail labels header

Format headers X-Label and Keywords as a space separated list and all
other ones as comma-separated entities.  This makes OfflineIMAP label
handling to be compatible with some user agents that recognise these
headers.

Signed-off-by: Eygene Ryabinkin <rea@codelabs.ru>

offlineimap.conf

@@ -277,7 +277,14 @@ remoterepository = RemoteExample
 #
 #synclabels = no
 
-# Name of the header to use for label storage.
+# Name of the header to use for label storage.  Format for the header
+# value differs for different headers, because there are some de-facto
+# standards set by popular clients:
+# - X-Label or Keywords keep values separated with spaces; for these
+#   you, obviously, should not have label values that contain spaces;
+# - X-Keywords use comma (',') as the separator.
+# To be consistent with the usual To-like headers, for the rest of header
+# types we use comma as the separator.
 #
 #labelsheader = X-Keywords
 

offlineimap/folder/Gmail.py

@@ -92,8 +92,8 @@ class GmailFolder(IMAPFolder):
             else:
                 labels = set()
             labels = labels - self.ignorelabels
-            labels = ', '.join(sorted(labels))
-            body = self.addmessageheader(body, self.labelsheader, labels)
+            labels_str = imaputil.format_labels_string(self.labelsheader, sorted(labels))
+            body = self.addmessageheader(body, self.labelsheader, labels_str)
 
         if len(body)>200:
             dbg_output = "%s...%s" % (str(body)[:150], str(body)[-50:])
@@ -183,11 +183,8 @@ class GmailFolder(IMAPFolder):
         if not self.synclabels:
             return super(GmailFolder, self).savemessage(uid, content, flags, rtime)
 
-        labels = self.getmessageheader(content, self.labelsheader)
-        if labels:
-            labels = set([lb.strip() for lb in labels.split(',') if len(lb.strip()) > 0])
-        else:
-            labels = set()
+        labels = imaputil.labels_from_header(self.labelsheader,
+          self.getmessageheader(content, self.labelsheader))
 
         ret = super(GmailFolder, self).savemessage(uid, content, flags, rtime)
         self.savemessagelabels(ret, labels)

offlineimap/folder/GmailMaildir.py

@@ -20,6 +20,7 @@ import os
 from .Maildir import MaildirFolder
 from offlineimap import OfflineImapError
 import offlineimap.accounts
+from offlineimap import imaputil
 
 class GmailMaildirFolder(MaildirFolder):
     """Folder implementation to support adding labels to messages in a Maildir.
@@ -78,12 +79,10 @@ class GmailMaildirFolder(MaildirFolder):
             content = file.read()
             file.close()
 
-            labels = self.getmessageheader(content, self.labelsheader)
-            if labels:
-                labels = set([lb.strip() for lb in labels.split(',') if len(lb.strip()) > 0])
-            else:
-                labels = set()
-            self.messagelist[uid]['labels'] = labels
+            self.messagelist[uid]['labels'] = \
+              imaputil.labels_from_header(self.labelsheader,
+              self.getmessageheader(content, self.labelsheader))
+
 
         return self.messagelist[uid]['labels']
 
@@ -103,11 +102,8 @@ class GmailMaildirFolder(MaildirFolder):
         if not self.synclabels:
             return super(GmailMaildirFolder, self).savemessage(uid, content, flags, rtime)
 
-        labels = self.getmessageheader(content, self.labelsheader)
-        if labels:
-            labels = set([lb.strip() for lb in labels.split(',') if len(lb.strip()) > 0])
-        else:
-            labels = set()
+        labels = imaputil.labels_from_header(self.labelsheader,
+          self.getmessageheader(content, self.labelsheader))
         ret = super(GmailMaildirFolder, self).savemessage(uid, content, flags, rtime)
 
         # Update the mtime and labels
@@ -130,12 +126,9 @@ class GmailMaildirFolder(MaildirFolder):
         content = file.read()
         file.close()
 
-        oldlabels = self.getmessageheader(content, self.labelsheader)
+        oldlabels = imaputil.labels_from_header(self.labelsheader,
+          self.getmessageheader(content, self.labelsheader))
 
-        if oldlabels:
-            oldlabels = set([lb.strip() for lb in oldlabels.split(',') if len(lb.strip()) > 0])
-        else:
-            oldlabels = set()
 
         labels = labels - ignorelabels
         ignoredlabels = oldlabels & ignorelabels
@@ -146,13 +139,14 @@ class GmailMaildirFolder(MaildirFolder):
             return
 
         # Change labels into content
-        labels_str = ', '.join(sorted(labels | ignoredlabels))
+        labels_str = imaputil.format_labels_string(self.labelsheader,
+          sorted(labels | ignoredlabels))
         content = self.addmessageheader(content, self.labelsheader, labels_str)
         rtime = self.messagelist[uid].get('rtime', None)
 
         # write file with new labels to a unique file in tmp
         messagename = self.new_message_filename(uid, set())
-        tmpname = self.save_tmp_file(messagename, content)
+        tmpname = self.save_to_tmp_file(messagename, content)
         tmppath = os.path.join(self.getfullname(), tmpname)
 
         # move to actual location

offlineimap/imaputil.py

@@ -21,6 +21,12 @@ import string
 from offlineimap.ui import getglobalui
 
 
+## Globals
+
+# Message headers that use space as the separator (for label storage)
+SPACE_SEPARATED_LABEL_HEADERS = ('X-Label', 'Keywords')
+
+
 def __debug(*args):
     msg = []
     for arg in args:
@@ -260,3 +266,69 @@ def __split_quoted(string):
 		rest = rest[next_q + 1:]
 		if not is_escaped:
 			return (quoted, rest.lstrip())
+
+
+def format_labels_string(header, labels):
+    """
+    Formats labels for embedding into a message,
+    with format according to header name.
+    
+    Headers from SPACE_SEPARATED_LABEL_HEADERS keep space-separated list
+    of labels, the rest uses comma (',') as the separator.
+
+    Also see parse_labels_string() and modify it accordingly
+    if logics here gets changed.
+
+    """
+    if header in SPACE_SEPARATED_LABEL_HEADERS:
+        sep = ' '
+    else:
+        sep = ','
+
+    return sep.join(labels)
+
+
+def parse_labels_string(header, labels_str):
+    """
+    Parses a string into a set of labels, with a format according to
+    the name of the header.
+
+    See __format_labels_string() for explanation on header handling
+    and keep these two functions synced with each other.
+
+    TODO: add test to ensure that
+      format_labels_string * parse_labels_string is unity
+    and
+      parse_labels_string * format_labels_string is unity
+
+    """
+
+    if header in SPACE_SEPARATED_LABEL_HEADERS:
+        sep = ' '
+    else:
+        sep = ','
+
+    labels = labels_str.strip().split(sep)
+
+    return set([l.strip() for l in labels if l.strip()])
+
+
+def labels_from_header(header_name, header_value):
+    """
+    Helper that builds label set from the corresponding header value.
+
+    Arguments:
+    - header_name: name of the header that keeps labels;
+    - header_value: value of the said header, can be None
+
+    Returns: set of labels parsed from the header (or empty set).
+
+    """
+
+    if header_value:
+        labels = parse_labels_string(header_name, header_value)
+    else:
+        labels = set()
+
+    return labels
+