find bash through /usr/bin/env

[cmccabe-bin] / snarf_mail.rb

diff --git a/snarf_mail.rb b/snarf_mail.rb
index a2490b1..76c831c 100755 (executable)
--- a/snarf_mail.rb
+++ b/snarf_mail.rb
@@ -9,6 +9,7 @@
 # http://ruby-doc.org/stdlib/libdoc/net/imap/rdoc/index.html
 #
 
+require 'date'
 require 'net/imap'
 require 'optparse'
 require 'ostruct'
@@ -17,11 +18,16 @@ class MyOptions
   def self.parse(args)
     opts = OpenStruct.new
     opts.mailboxes = Array.new
+    opts.delete = "none"
 
     # Fill in $opts values
     parser = OptionParser.new do |myparser|
       myparser.banner = "Usage: #{ File.basename($0) } [opts]"
       myparser.separator("Specific options:")
+      myparser.on("--delete POLICY", "-d",
+              "Set delete policy to 'none' or 'old'. Default is 'none'.") do |d|
+        opts.delete = d
+      end
       myparser.on("--username USERNAME", "-u",
               "Email account to fetch. (example: \
 RareCactus@gmail.com)") do |u|
@@ -84,27 +90,78 @@ def format_uid(uid)
   return sprintf("%006d", uid)
 end
 
+def format_date(date)
+  date.gsub!(' ', '_')
+end
+
+def get_sanitized_email_name(mailbox, arr)
+  msn = mailbox.dup
+  msn.gsub!(' ', '_')
+  msn.gsub!('/', '.')
+  return "#{msn}_#{format_date(arr["INTERNALDATE"])}_#{format_uid(arr["UID"])}"
+end
+
+def write_email_to_disk(mailbox, data)
+  arr = data[0].attr
+  filename = get_sanitized_email_name(mailbox, arr)
+  fp = File.open(filename, 'w')
+  fp.write(arr["RFC822.HEADER"])
+  fp.write(arr["RFC822.TEXT"])
+  fp.close
+end
+
 def snarf_mailbox(imap, mailbox)
-  imap.select(mailbox)
-  count = 0
-  total_count = 0
-  imap.search(["NOT", "DELETED"]).each do |message_id|
-    data = imap.fetch(message_id, [ "UID", "RFC822.HEADER", "RFC822.TEXT" ])
-    a = data[0].attr
-    filename = "#{mailbox}#{format_uid(a["UID"])}"
-    fp = File.open(filename, 'w')
-    fp.write(a["RFC822.HEADER"])
-    fp.write(a["RFC822.TEXT"])
-    fp.close
-    count = count + 1
-    total_count = total_count + 1
-    if (count > 10) then
-      count = 0
-      printf(".")
-      STDOUT.flush()
+  full_count = 0
+  first_time = true
+
+  searchterms = [ "NOT", "DELETED" ]
+  if $opts.delete == "old"
+    t = Date.today() - 365
+    time_str = t.strftime("%e-%b-%Y")
+    searchterms << "BEFORE" << time_str
+    prequel = "fetched and deleted: "
+  elsif $opts.delete == "none"
+    prequel = "fetched: "
+  else
+    raise "expected one of 'old', 'none' for delete argument."
+  end
+
+  while true
+    count = 0
+    msg_seqnos = Array.new
+
+    imap.select(mailbox)
+    imap.search(searchterms).each do |message_id|
+      if (first_time == true) then
+        # Print a dot immediately after making first contact with the server.
+        # It is reassuring to the user.
+        printf(".")
+        STDOUT.flush()
+        first_time = false
+      end
+      data = imap.fetch(message_id, 
+                [ "INTERNALDATE", "UID", "RFC822.HEADER", "RFC822.TEXT" ])
+      write_email_to_disk(mailbox, data)
+      count = count + 1
+      full_count = full_count + 1
+      msg_seqnos << data[0].seqno.to_i
+      #break if (count > 20)
+    end
+    if (count == 0) then
+      puts "#{prequel} #{full_count} messages from #{mailbox}"
+      return
+    end
+
+    # Print out a dot to signify progress
+    printf(".")
+    STDOUT.flush()
+
+    if ($opts.delete != "none"):
+      # Delete messages
+      imap.store(msg_seqnos, "+FLAGS", [:Deleted])
+      imap.expunge
     end
   end
-  puts "fetched #{total_count} messages from #{mailbox}"
 end
 
 # MAIN