Support importing mbox containing multiple messages

2 files changed, 55 insertions, 19 deletions

diff --git a/lib/puke/core.rb b/lib/puke/core.rb
index 77652aa..4cbb0d5 100644
--- a/lib/puke/core.rb
+++ b/lib/puke/core.rb
@@ -57,20 +57,8 @@ module Puke
   end
 
   CreateMutex = Thread::Mutex.new
-  def create(io, is_raw = false)
-    t = Tempfile.open("unnamed", tmpdir)
-    if is_raw
-      # Is this really correct??
-      t << "From mboxrd Thu Jan  1 00:00:00 1970\n"
-      io.each_line do |line|
-        t << line.gsub(/\A(>*)From /, "\\1>From ").gsub(/\r\n\z/, "\n")
-      end
-      t << "\n"
-    else
-      IO.copy_stream(io, t)
-    end
-    t.rewind
-    mid, tid, subject, date = parse_file(t)
+  def create0(file)
+    mid, tid, subject, date = parse_file(file)
     fpath = File.join(datadir, mangle_mid(mid) + ".mbox")
 
     # A mutex is used here in case two threads try to store the same message
@@ -100,11 +88,61 @@ module Puke
         @gdbm["t#{tid}"] = l.map { |i| i[:mid] }.join("\0")
       end
       @gdbm["m#{mid}"] = [tid, subject, date.rfc2822].join("\0")
-      File.link(t, fpath)
+      File.link(file, fpath)
     }
     mid
-  ensure
-    t and t.close!
+  end
+
+  def create(io, is_raw = false)
+    if is_raw
+      Tempfile.create("unnamed", tmpdir) { |t|
+        t.binmode
+        # Is this really correct??
+        t << "From mboxrd Thu Jan  1 00:00:00 1970\n"
+        x = nil
+        io.each_line do |line|
+          x = line[-1] == "\n"
+          line.gsub!(/\A(>*)From /, "\\1>From ")
+          line.gsub!(/\r\n\z/, "\n")
+          t << line
+        end
+        if x
+          t << "\n\n"
+        else
+          t << "\n"
+        end
+        t.rewind
+        create0(t)
+      }
+    else
+      # io is already in mboxrd format and may contain multiple messages
+      mids = []
+      begin
+        t = nil
+        io.each_line do |line|
+          # Beginning of a new message
+          if line.start_with?("From ")
+            if t
+              t.rewind
+              mids << create0(t)
+              t.close!
+            end
+            t = Tempfile.open("unnamed", tmpdir)
+            t.binmode
+          end
+          raise "invalid mbox header" unless t
+          t << line
+        end
+        if t
+          t.rewind
+          mids << create0(t)
+          t.close!
+        end
+      ensure
+        t and not t.closed? and t.close!
+      end
+      mids
+    end
   end
 
   def parse_file(io)
diff --git a/puke-import b/puke-import
index ed362df..1a7c500 100755
--- a/puke-import
+++ b/puke-import
@@ -7,8 +7,6 @@ require "net/http"
 #
 # URI is the url for importing an message.
 # E.g. 'http://localhost:8080/new/secret-string'
-#
-# TODO: mboxrd containing multiple emails
 
 uri = URI.parse(ARGV.shift)
 http = Net::HTTP.new(uri.host, uri.port)