mailparser 0.4.22a

data/test/test_rfc2045.rb

@@ -0,0 +1,121 @@
+# Copyright (C) 2006-2010 TOMITA Masahiro
+# mailto:tommy@tmtm.org
+
+require "mailparser/rfc2045"
+
+unless ARGV.empty?
+  ARGV.each do |fname|
+    File.open(fname) do |f|
+      header = []
+      f.each do |line|
+        break if line.chomp.empty?
+        if line =~ /^\s/ then
+          header[-1] << line
+        else
+          header << line
+        end
+      end
+      header.each do |h|
+        begin
+          MailParser::RFC2045.parse(*h.split(/\s*:\s*/, 2))
+        rescue MailParser::ParseError => e
+          puts fname
+          p e
+          p h
+        end
+      end
+    end
+  end
+  exit
+end
+
+require "test/unit"
+
+class TC_RFC2045 < Test::Unit::TestCase
+  def test_content_type_noparams()
+    c = MailParser::RFC2045.parse("content-type", "text/plain")
+    assert_equal("text", c.type)
+    assert_equal("plain", c.subtype)
+    assert_equal({}, c.params)
+  end
+
+  def test_content_type_charset()
+    c = MailParser::RFC2045.parse("content-type", "text/plain; charset=euc-jp")
+    assert_equal("text", c.type)
+    assert_equal("plain", c.subtype)
+    assert_equal(1, c.params.size)
+    assert_equal("euc-jp", c.params["charset"])
+  end
+
+  def test_content_type_charset_upcase()
+    c = MailParser::RFC2045.parse("content-type", "TEXT/PLAIN; CHARSET=euc-jp")
+    assert_equal("text", c.type)
+    assert_equal("plain", c.subtype)
+    assert_equal(1, c.params.size)
+    assert_equal("euc-jp", c.params["charset"])
+  end
+
+  def test_content_type_charset_quote()
+    c = MailParser::RFC2045.parse("content-type", "text/plain; charset=\"euc-jp\"")
+    assert_equal("text", c.type)
+    assert_equal("plain", c.subtype)
+    assert_equal(1, c.params.size)
+    assert_equal("euc-jp", c.params["charset"])
+  end
+
+  def test_content_type_invalid()
+    assert_raises(MailParser::ParseError){MailParser::RFC2045.parse("content-type", "text/plain; name=\"hosts\"hoge")}
+  end
+
+  def test_content_description()
+    c = MailParser::RFC2045.parse("content-description", "(hoge fuga>")
+    assert_equal("(hoge fuga>", c)
+  end
+
+  def test_content_transfer_encoding()
+    c = MailParser::RFC2045.parse("content-transfer-encoding", "7bit")
+    assert_equal("7bit", c.mechanism)
+  end
+
+  def test_content_id()
+    c = MailParser::RFC2045.parse("content-id", "<a@b.c>")
+    assert_equal("a@b.c", c.msg_id)
+  end
+
+  def test_mime_version()
+    v = MailParser::RFC2045.parse("mime-version", "1.0")
+    assert_equal("1.0", v)
+    assert_equal([], v.comments)
+  end
+
+  def test_mime_version_with_comment()
+    v = MailParser::RFC2045.parse("mime-version", "1.0 (hoge)")
+    assert_equal("1.0", v)
+    assert_equal(["(hoge)"], v.comments)
+  end
+
+  def test_mime_version_with_comment2()
+    v = MailParser::RFC2045.parse("mime-version", "1.(hoge)0")
+    assert_equal("1.0", v)
+    assert_equal(["(hoge)"], v.comments)
+  end
+
+  def test_qp_decode()
+    assert_equal("IJKL", MailParser::RFC2045.qp_decode("=49=4A=4B=4C"))
+    assert_equal("IJKL", MailParser::RFC2045.qp_decode("=49=4a=4b=4c"))
+    assert_equal("abcdefg", MailParser::RFC2045.qp_decode("abcd=\r\nefg"))
+    assert_equal("abcdefg", MailParser::RFC2045.qp_decode("abcd=  \r\nefg"))
+    assert_equal("abcdefg", MailParser::RFC2045.qp_decode("abcd=\nefg"))
+    assert_equal("abcdefg", MailParser::RFC2045.qp_decode("abcd=  \nefg"))
+    assert_equal("abcd\r\nefg", MailParser::RFC2045.qp_decode("abcd=\r\n\r\nefg"))
+    assert_equal("abcd\r\nefg", MailParser::RFC2045.qp_decode("abcd=  \r\n\r\nefg"))
+  end
+
+  def test_b64_decode()
+    assert_equal("abcdefg", MailParser::RFC2045.b64_decode("YWJjZGVmZw=="))
+    assert_equal("01234567890123456789", MailParser::RFC2045.b64_decode("MDEyMzQ1Njc4OTAxMjM0NTY3ODk="))
+    assert_equal("01234567890123456789", MailParser::RFC2045.b64_decode("MDEyMzQ1Njc4\nOTAxMjM0NTY3ODk="))
+    assert_equal("01234567890123456789", MailParser::RFC2045.b64_decode("MDEyMzQ1Nj\nc4OTAxMjM0NTY3ODk="))
+  end
+end
+

data/test/test_rfc2047.rb

@@ -0,0 +1,118 @@
+# -*- coding: utf-8 -*-
+# Copyright (C) 2006-2010 TOMITA Masahiro
+# mailto:tommy@tmtm.org
+
+require "mailparser/rfc2047"
+require "test/unit"
+
+class TC_RFC2047 < Test::Unit::TestCase
+  def test_q_decode_ascii()
+    assert_equal("hoge", MailParser::RFC2047.q_decode("hoge"))
+  end
+
+  def test_q_decode_space()
+    assert_equal("a b", MailParser::RFC2047.q_decode("a_b"))
+  end
+
+  def test_q_decode_utf8()
+    assert_equal("とみた", MailParser::RFC2047.q_decode("=E3=81=A8=E3=81=BF=E3=81=9F"))
+  end
+
+  def test_q_decode_utf8_ascii()
+    assert_equal("とaみbた", MailParser::RFC2047.q_decode("=E3=81=A8a=E3=81=BFb=E3=81=9F"))
+  end
+
+  def test_q_decode_end_equal()
+    assert_equal("abc", MailParser::RFC2047.q_decode("abc="))
+  end
+
+  def test_b_decode_ascii()
+    assert_equal("hoge", MailParser::RFC2047.b_decode("aG9nZQ=="))
+  end
+
+  def test_b_decode_utf8()
+    assert_equal("とみた", MailParser::RFC2047.b_decode("44Go44G/44Gf"))
+  end
+
+  def test_b_decode_invalid_space()
+    assert_equal("とみた", MailParser::RFC2047.b_decode("44Go 44 G/4 4Gf"))
+  end
+
+  def test_split_decode_q_ascii()
+    s = MailParser::RFC2047.split_decode("=?us-ascii?q?hoge?=")
+    assert_equal(1, s.size)
+    assert_equal("hoge", s[0])
+    assert_equal("us-ascii", s[0].charset)
+    assert_equal("=?us-ascii?q?hoge?=", s[0].raw)
+  end
+
+  def test_split_decode_q_ascii_upcase()
+    s = MailParser::RFC2047.split_decode("=?US-ASCII?Q?hoge?=")
+    assert_equal(1, s.size)
+    assert_equal("hoge", s[0])
+    assert_equal("us-ascii", s[0].charset)
+    assert_equal("=?US-ASCII?Q?hoge?=", s[0].raw)
+  end
+
+  def test_decode_q_ascii()
+    s = MailParser::RFC2047.decode("=?us-ascii?q?hoge?=")
+    assert_equal("hoge", s)
+  end
+
+  def test_decode_q_b()
+    s = MailParser::RFC2047.decode("=?us-ascii?q?hoge?= =?us-ascii?b?aG9nZQ==?=")
+    assert_equal("hogehoge", s)
+  end
+
+  def test_decode_plain()
+    s = MailParser::RFC2047.decode("abcdefg")
+    assert_equal("abcdefg", s)
+  end
+
+  def test_decode_encode_plain()
+    s = MailParser::RFC2047.decode("012345 =?us-ascii?q?hoge?= abcdefg")
+    assert_equal("012345 hoge abcdefg", s)
+  end
+
+  def test_decode_encode_plain2()
+    s = MailParser::RFC2047.decode("=?us-ascii?q?hoge?= abcdefg =?us-ascii?q?fuga?=")
+    assert_equal("hoge abcdefg fuga", s)
+  end
+
+  def test_decode_unknown_charset()
+    s = MailParser::RFC2047.decode("=?hoge?q?hoge?=")
+    assert_equal("hoge", s)
+  end
+
+  def test_decode_unknown_charset2()
+    s = MailParser::RFC2047.decode("=?hoge?q?hoge?=", "UTF-8")
+    assert_equal("=?hoge?q?hoge?=", s)
+  end
+
+  def test_decode_unknown_charset3()
+    s = MailParser::RFC2047.decode("abcdefg =?hoge?q?hoge?= =?us-ascii?q?fuga?=")
+    assert_equal("abcdefg hogefuga", s)
+  end
+
+  def test_decode_unknown_charset4()
+    s = MailParser::RFC2047.decode("abcdefg =?hoge?q?hoge?= =?us-ascii?q?fuga?=", "UTF-8")
+    assert_equal("abcdefg =?hoge?q?hoge?= fuga", s)
+  end
+
+  def test_decode_sjis()
+    s = MailParser::RFC2047.decode("=?sjis?b?h0A=?=", "UTF-8")
+    assert_equal("\xe2\x91\xa0", s)
+  end
+
+  def test_decode_iso2022jp()
+    s = MailParser::RFC2047.decode("=?iso-2022-jp?b?GyRCLSEbKEI=?=", "UTF-8")
+    assert_equal("\xe2\x91\xa0", s)
+  end
+
+  def test_decode_charset_converter()
+    proc = Proc.new{|f,t,s| s.gsub(/o/, "X")}
+    s = MailParser::RFC2047.decode("=?us-ascii?q?hoge?=", :output_charset=>"utf-8", :charset_converter=>proc)
+    assert_equal("hXge", s)
+  end
+
+end

data/test/test_rfc2183.rb

@@ -0,0 +1,60 @@
+# Copyright (C) 2006-2010 TOMITA Masahiro
+# mailto:tommy@tmtm.org
+
+require "mailparser/rfc2183"
+
+unless ARGV.empty?
+  ARGV.each do |fname|
+    File.open(fname) do |f|
+      header = []
+      f.each do |line|
+        break if line.chomp.empty?
+        if line =~ /^\s/ then
+          header[-1] << line
+        else
+          header << line
+        end
+      end
+      header.each do |h|
+        begin
+          MailParser::RFC2183.parse(*h.split(/\s*:\s*/, 2))
+        rescue MailParser::ParseError => e
+          puts fname
+          p e
+          p h
+        end
+      end
+    end
+  end
+  exit
+end
+
+require "test/unit"
+
+class TC_RFC2183 < Test::Unit::TestCase
+  def test_content_disposition_noparams()
+    c = MailParser::RFC2183.parse("content-disposition", "inline")
+    assert_equal("inline", c.type)
+    assert_equal({}, c.params)
+  end
+
+  def test_content_disposition_charset()
+    c = MailParser::RFC2183.parse("content-disposition", "inline; filename=hoge")
+    assert_equal("inline", c.type)
+    assert_equal(1, c.params.size)
+    assert_equal("hoge", c.params["filename"])
+  end
+
+  def test_content_disposition_charset_upcase()
+    c = MailParser::RFC2183.parse("content-disposition", "INLINE; FILENAME=HOGE")
+    assert_equal("inline", c.type)
+    assert_equal("HOGE", c.params["filename"])
+  end
+
+  def test_content_disposition_charset_quote()
+    c = MailParser::RFC2183.parse("content-disposition", "inline; filename=\"hoge\"")
+    assert_equal("inline", c.type)
+    assert_equal(1, c.params.size)
+    assert_equal("hoge", c.params["filename"])
+  end
+end

data/test/test_rfc2231.rb

@@ -0,0 +1,167 @@
+# Copyright (C) 2006-2010 TOMITA Masahiro
+# mailto:tommy@tmtm.org
+
+require "mailparser/rfc2231"
+require "test/unit"
+
+class TC_RFC2231 < Test::Unit::TestCase
+  def test_parse_param()
+    params = {
+      "hoge" => "fuga",
+      "foo"  => "bar",
+    }
+    h = MailParser::RFC2231.parse_param(params)
+    assert_equal("fuga", h["hoge"])
+    assert_equal(nil, h["hoge"].charset)
+    assert_equal(nil, h["hoge"].language)
+    assert_equal("bar", h["foo"])
+    assert_equal(nil, h["foo"].charset)
+    assert_equal(nil, h["foo"].language)
+  end
+
+  def test_parse_param2()
+    params = {
+      "hoge*0" => "fuga",
+      "hoge*1" => "bar",
+      "foo"  => "bar",
+    }
+    h = MailParser::RFC2231.parse_param(params)
+    assert_equal("fugabar", h["hoge"])
+    assert_equal(nil, h["hoge"].charset)
+    assert_equal(nil, h["hoge"].language)
+    assert_equal("bar", h["foo"])
+    assert_equal(nil, h["foo"].charset)
+    assert_equal(nil, h["foo"].language)
+  end
+
+  def test_parse_param3()
+    params = {
+      "hoge*0*" => "''fuga",
+      "hoge*1"  => "bar",
+      "foo"  => "bar",
+    }
+    h = MailParser::RFC2231.parse_param(params)
+    assert_equal("fugabar", h["hoge"])
+    assert_equal("", h["hoge"].charset)
+    assert_equal("", h["hoge"].language)
+    assert_equal("bar", h["foo"])
+    assert_equal(nil, h["foo"].charset)
+    assert_equal(nil, h["foo"].language)
+  end
+
+  def test_parse_param4()
+    params = {
+      "hoge*0*" => "euc-jp'ja'fuga",
+      "hoge*1"  => "bar",
+      "foo"  => "bar",
+    }
+    h = MailParser::RFC2231.parse_param(params)
+    assert_equal("fugabar", h["hoge"])
+    assert_equal("euc-jp", h["hoge"].charset)
+    assert_equal("ja", h["hoge"].language)
+    assert_equal("bar", h["foo"])
+    assert_equal(nil, h["foo"].charset)
+    assert_equal(nil, h["foo"].language)
+  end
+
+  def test_parse_param5()
+    params = {
+      "hoge*0*" => "''%30%31%32%33",
+      "hoge*1"  => "%34%35%36",
+      "foo"  => "bar",
+    }
+    h = MailParser::RFC2231.parse_param(params)
+    assert_equal("0123%34%35%36", h["hoge"])
+    assert_equal("", h["hoge"].charset)
+    assert_equal("", h["hoge"].language)
+    assert_equal("bar", h["foo"])
+    assert_equal(nil, h["foo"].charset)
+    assert_equal(nil, h["foo"].language)
+  end
+
+  def test_parse_param6()
+    params = {
+      "hoge*" => "''fuga",
+      "foo"  => "bar",
+    }
+    h = MailParser::RFC2231.parse_param(params)
+    assert_equal("fuga", h["hoge"])
+    assert_equal("", h["hoge"].charset)
+    assert_equal("", h["hoge"].language)
+    assert_equal("bar", h["foo"])
+    assert_equal(nil, h["foo"].charset)
+    assert_equal(nil, h["foo"].language)
+  end
+
+  def test_parse_param7()
+    params = {
+      "hoge*" => "fuga",
+    }
+    assert_raises(MailParser::ParseError){MailParser::RFC2231.parse_param(params)}
+  end
+
+  def test_parse_param7_1()
+    params = {
+      "hoge*" => "fuga",
+    }
+    h = MailParser::RFC2231.parse_param(params, false)
+    assert_equal("fuga", h["hoge"])
+  end
+
+  def test_parse_param8()
+    params = {
+      "hoge*0*" => "fuga",
+    }
+    assert_raises(MailParser::ParseError){MailParser::RFC2231.parse_param(params)}
+  end
+
+  def test_parse_param8_1()
+    params = {
+      "hoge*0*" => "fuga",
+    }
+    h = MailParser::RFC2231.parse_param(params, false)
+    assert_equal("fuga", h["hoge"])
+  end
+
+  def test_parse_param_charset_converter
+    params = {
+      "hoge*" => "xcharset'xlang'%41%42%43",
+    }
+    h = MailParser::RFC2231.parse_param(params, :output_charset=>"xx", :charset_converter=>Proc.new{|f,t,s| s.downcase})
+    assert_equal "abc", h["hoge"]
+  end
+
+  def test_rfc_example()
+    params = {
+      "URL*0" => "ftp://",
+      "URL*1" => "cs.utk.edu.pub/moore/bulk-mailer/buik-mailer.tar"
+    }
+    h = MailParser::RFC2231.parse_param(params)
+    assert_equal("ftp://cs.utk.edu.pub/moore/bulk-mailer/buik-mailer.tar", h["URL"])
+    assert_equal(nil, h["URL"].charset)
+    assert_equal(nil, h["URL"].language)
+  end
+
+  def test_rfc_example2()
+    params = {
+      "title*" => "us-ascii'en-us'This%20is%20%2A%2A%2Afun%2A%2A%2A"
+    }
+    h = MailParser::RFC2231.parse_param(params)
+    assert_equal("This is ***fun***", h["title"])
+    assert_equal("us-ascii", h["title"].charset)
+    assert_equal("en-us", h["title"].language)
+  end
+
+  def test_rfc_example3()
+    params = {
+      "title*0*" => "us-ascii'en'This%20is%20even%20more%20",
+      "title*1*" => "%2A%2A%2Afun%2A%2A%2A%20",
+      "title*2"  => "isn't it!"
+    }
+    h = MailParser::RFC2231.parse_param(params)
+    assert_equal("This is even more ***fun*** isn't it!", h["title"])
+    assert_equal("us-ascii", h["title"].charset)
+    assert_equal("en", h["title"].language)
+  end
+
+end