Re: Balsa can't download a mail with a long line, gets stuck in endless dup-creating loop.

[Pawel Salek <pawsa0 gmail com>]

Rob,

thanks for reporting the issue. I believe the attached patch should  
address the issue you see.

BTW, have you considered accessing gmail via IMAP and copying messages  
this way?

/Pawel

On 10/22/2012 07:33:01 PM, Rob Landley wrote:
> I'm attempting to switch over to balsa (the version built into current
> xubuntu lts, 2.4.11-1), and download a year or so of archives off
> gmail via pop3. Since gmail only gives ~500 messages per pop session,
> I told it to check mail once a minute and left it running overnight.
>
> When I get up, it's stuck in an endless loop appending duplicate
> messages to my inbox, because one of the messages won't download and
> in that case it drops the connection rather than calling quit, meaning
> none of the messages are deleted off the server, so it re-downloads
> them (and hits the same error again) next time.
>
> (Pause to write quick python program to detect/kill duplicates in  
> inbox.)
>
> A couple complicating factors:
>
> 1) The error message pops up in two xfce bubble that go away again in
> about 3 seconds.
>
> 2) The message number it's complaining about is only identifiable via
> the pop interface, not via gmail's web view.
>
> (Pause to write quick python program to download a batch of pop
> messages and write them to numbered files.)
>
> I use said python program to delete the batch of mail it couldn't
> download (but which I copied to separate files; it's from september
> 2011 anyway), and set it downloading again. It gets up to february of
> this year, and does it again.
>
> This time, I caught the line of text it was complaining about, which
> starts with ">> Ethtool", and the word "Ethtool" is not in message 129
> (as it said) but message 126 in the batch, which is attached.
>
> I can't control what people on linux-kernel send me. Is there any way
> to fix balsa?
>
> Rob

diff --git a/libbalsa/imap/pop3.c b/libbalsa/imap/pop3.c
index cf5d76c..2387daa 100644
--- a/libbalsa/imap/pop3.c
+++ b/libbalsa/imap/pop3.c
@@ -650,10 +650,14 @@ pop_fetch_message_s(PopHandle *pop, unsigned msgno,
   resp = pop_check_status(pop, err);
   /* do here the fetch */
   if(resp) {
+    /* whether next line will be a continuation line. */
+    gboolean continuation_line = FALSE;
     while( sio_gets(pop->sio, line, sizeof(line)) &&
-           strcmp(line, ".\r\n") ) {
+           (continuation_line || strcmp(line, ".\r\n")) ) {
       char *arg = line[0] == '.' ? line+1 : line;
       unsigned len = strlen(arg);
+      continuation_line = (len >= POP_LINE_LEN-1);
+      printf("Line '%s'\n", line);
       if(pop->filter_cr && len>=2 && arg[len-2] == '\r') 
         arg[(--len)-1] = '\n';
       if(resp) 
@@ -663,6 +667,7 @@ pop_fetch_message_s(PopHandle *pop, unsigned msgno,
           resp = FALSE;
         }
     }
+    printf("LAST Line is '%s'\n", line);
   }
   return resp;
 }
@@ -743,10 +748,13 @@ pop_complete_retr(PopHandle *pop, PopAsyncCb cb, void *arg)
     cb(rc, arg, &err);
   if(resp) { /* same code as in fetch_message() */
     char * str;
+    /* whether next line will be a continuation line. */
+    gboolean continuation_line = FALSE;
     while( (str = sio_gets(pop->sio, line, sizeof(line))) &&
-           strcmp(line, ".\r\n") ) {
+           (continuation_line || strcmp(line, ".\r\n")) ) {
       char *buf = line[0] == '.' ? line+1 : line;
       unsigned len = strlen(buf);
+      continuation_line = (len >= POP_LINE_LEN-1);
       if(pop->filter_cr && len>=2 && buf[len-2] == '\r') 
         buf[(--len)-1] = '\n';
       if(cb)