mail_address 1.1.2 → 1.2.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: ccb2c082070db454ef985f2a87325823c351097d
-  data.tar.gz: 4b6a338e58922edfba00921031ca331293f0b6fa
+  metadata.gz: 55a1aa4aa12065d46469e8c35b6041f37ee42cc1
+  data.tar.gz: b797ab4c2a23d3ce801beaf16172ed0b0ff4280b
 SHA512:
-  metadata.gz: fda758c1ac169360f0a18b9c8390ebf82022e471f70c19939b715b553f8b1ba8eb5a3ae68b2dd9405cef8b84bfff751e3a9c509a57e86e3f1e0e9fff57f69436
-  data.tar.gz: 47a765205ea2ae07b07cb8029245c14bdb96e9baa4c415395eb441ed5e5713b99c9a9b47078c9caf14fd88c7449b303fc066cce4547991aa119993e0c93efadb
+  metadata.gz: db93d3bf8aa49be0e99c1932ab0a6b2e226078e2768d97ec542aaa118c5dc65ec203fd9ed781ec17843390fcd78c2968501b6b61cd377953df9b6a079f685f4c
+  data.tar.gz: dfba207c8827f687e68401236ae672350d669f5734db8ff4a119bec122f1f0d7f820f5c6ae8d99927ac73b3d893a785009ba384404c59fbf81a3d1093572bcff

data/README.md

@@ -1,21 +1,21 @@
 # MailAddress [![Build Status](https://travis-ci.org/kizashi1122/mail_address.svg)](https://travis-ci.org/kizashi1122/mail_address) [![Coverage Status](https://coveralls.io/repos/kizashi1122/mail_address/badge.png)](https://coveralls.io/r/kizashi1122/mail_address)
 
-MailAddress is a simple email address parser.
-This library is implemented based on Perl Module Mail::Address and added some improvements.
+MailAddress is a port of Mail::Address from Perl.
 
-[mail](https://github.com/mikel/mail) is a great gem library. But some email addresses (mostly are violated RFC) are unparsable with mail gem which is strictly RFC compliant. In perl, [Mail::Address](http://search.cpan.org/~markov/MailTools-2.14/lib/Mail/Address.pod) is a very common library to parse email address. Mail::Address conviniently can parse even RFC-violated email addresses such as:
+[mail](https://github.com/mikel/mail) is a great gem library. But some email addresses are unparsable with it. In perl, [Mail::Address](http://search.cpan.org/~markov/MailTools-2.14/lib/Mail/Address.pod) is a very common library to parse an email address. Mail::Address conviniently can parse even NOT RFC-compliant email addresses such as:
 
 ```rb
 # mail gem cannot parse the following addresses
 Ello [Do Not Reply] <do-not-reply@ello.co> # [, ] are not permitted according to RFC5322
-大阪 太郎<osaka@example.com> # no whitespace just before <
+大阪 太郎<osaka@example.com> # no whitespace just before `<`
 ```
-
-So I straightforwardly converted Perl module Mail::Address to Ruby gem. Then I reviced it because original Mail::Address also has some bad points. For example:
+But Mail::Address has some bad points (below). These are fixed in MailAddress.
 
 - if no ending parenthesis in name part, cannot parse correctly.
 - Modifications of name part are too much.
 
+However, MailAddress module cannnot parse whitespace-separated addresses.
+Many people could paste email addresses from Excel or the other spreadsheets. To accomplish this, also ported from a parser part of [Google Closure Library](https://github.com/google/closure-library/blob/master/closure/goog/format/emailaddress.js).
 
 ## Installation
 
@@ -60,7 +60,7 @@ p addrs[1].host       # "example.jp"
 p addrs[1].user       # "osaka"
 p addrs[1].original   # "大阪 太郎 <osaka@example.jp>"
 ```
-`address.name` and `address.phrase` are almost same. 
+`address.name` and `address.phrase` are almost same.
 `address.phrase` keeps outermost double quotes or parentheses.
 
 ## Contributing

data/lib/mail_address.rb

@@ -3,5 +3,6 @@ require "mail_address/version"
 module MailAddress
   require "mail_address/address"
   require "mail_address/mail_address"
+  require "mail_address/simple_parser"
   require "mail_address/version"
 end

data/lib/mail_address/mail_address.rb

@@ -73,13 +73,12 @@ module MailAddress
 
   def self._tokenize(addresses)
     line = addresses.join(',') # $_
-    words, snippet, field = [], [], []
+    words = []
 
     line.sub!(/\A\s+/, '')
     line.gsub!(/[\r\n]+/,' ')
 
     while (line != '')
-      field = ''
       tmp = nil
       if (
           line.sub!(/\A("(?:[^"\\]+|\\.)*")(\s*)/, '')      || # "..."

data/lib/mail_address/simple_parser.rb

@@ -0,0 +1,133 @@
+module MailAddress
+
+  # --------------------------------------------------------------------------------------------------
+  # This module is ported from Google Closure JavaScript Library
+  #  -> https://github.com/google/closure-library/blob/master/closure/goog/format/emailaddress.js
+  # --------------------------------------------------------------------------------------------------
+
+  OPENERS_ = '"<(['
+  CLOSERS_ = '">)]'
+  # SPECIAL_CHARS = '()<>@:\\\".[]'
+  ADDRESS_SEPARATORS_ = ',;'
+  # CHARS_REQUIRE_QUOTES_ = SPECIAL_CHARS + ADDRESS_SEPARATORS_
+  ESCAPED_DOUBLE_QUOTES_ = /\\\"/
+  ESCAPED_BACKSLASHES_ = /\\\\/
+  LOCAL_PART_REGEXP_STR_ = '[+a-zA-Z0-9_.!#$%&\'*\\/=?^`{|}~-]+'
+  DOMAIN_PART_REGEXP_STR_ = '([a-zA-Z0-9-]+\\.)+[a-zA-Z0-9]{2,63}'
+  EMAIL_ADDRESS_ = Regexp.new('\\A' + LOCAL_PART_REGEXP_STR_ + '@' + DOMAIN_PART_REGEXP_STR_ + '\\z')
+
+  def self.parse_simple(str)
+    result = []
+    email = token = ''
+
+    # Remove non-UNIX-style newlines that would otherwise cause getToken_ to
+    # choke. Remove multiple consecutive whitespace characters for the same
+    # reason.
+    str = self.collapse_whitespace(str)
+    i = 0
+    while (i < str.length)
+      token = get_token(str, i)
+      if self.is_address_separator(token) || (token == ' ' && self.is_valid(self.parse_internal(email)))
+        if !self.is_empty_or_whitespace(email)
+          result.push(self.parse_internal(email))
+        end
+        email = ''
+        i += 1
+        next
+      end
+      email << token
+      i += token.length
+    end
+
+    # Add the final token.
+    if (!self.is_empty_or_whitespace(email))
+      result.push(self.parse_internal(email))
+    end
+    return result
+  end
+
+  def self.parse_internal(addr)
+    name = ''
+    address = ''
+    i = 0
+    while (i < addr.length)
+      token = get_token(addr, i)
+      if (token[0] == '<' && token.index('>'))
+        end_i = token.index('>')
+        address = token[1, end_i - 1]
+      elsif (address == '')
+        name << token
+      end
+      i += token.length
+    end
+
+    # Check if it's a simple email address of the form "jlim@google.com".
+    if (address == '' && name.index('@'))
+      address = name
+      name = ''
+    end
+
+    name = self.collapse_whitespace(name)
+    name = name[1 .. -2] if name.start_with?('\'') && name.end_with?('\'')
+    name = name[1 .. -2] if name.start_with?('"') && name.end_with?('"')
+
+    # Replace escaped quotes and slashes.
+    name = name.gsub(ESCAPED_DOUBLE_QUOTES_, '"')
+    name = name.gsub(ESCAPED_BACKSLASHES_, '\\')
+
+    #address = goog.string.collapseWhitespace(address);
+    address.strip!
+    MailAddress::Address.new(name, address, addr)
+  end
+
+  def self.get_token(str, pos)
+    ch = str[pos]
+    p = OPENERS_.index(ch)
+    return ch unless p
+
+    if (self.is_escaped_dbl_quote(str, pos))
+      # If an opener is an escaped quote we do not treat it as a real opener
+      # and keep accumulating the token.
+      return ch
+    end
+    closer_char = CLOSERS_[p]
+    end_pos = str.index(closer_char, pos + 1)
+
+    # If the closer is a quote we go forward skipping escaped quotes until we
+    # hit the real closing one.
+    while (end_pos >= 0 && self.is_escaped_dbl_quote(str, end_pos))
+      end_pos = str.index(closer_char, end_pos + 1)
+    end
+
+    token = (end_pos >= 0) ? str[pos .. end_pos] : ch
+    return token
+  end
+
+  def self.is_escaped_dbl_quote(str, pos)
+    return false if str[pos] != '"'
+    slash_count = 0
+
+    for idx in (pos - 1).downto(0)
+      break unless str[idx] == '\\'
+      slash_count += 1
+    end
+    (slash_count % 2) != 0
+  end
+
+  def self.collapse_whitespace(str)
+    str.gsub(/[\s\xc2\xa0]+/, ' ').strip
+  end
+
+  def self.is_empty_or_whitespace(str)
+    /\A[\s\xc2\xa0]*\z/ =~ str
+  end
+
+  def self.is_address_separator(ch)
+    ADDRESS_SEPARATORS_.include? ch
+  end
+
+  def self.is_valid(address)
+    EMAIL_ADDRESS_ =~ address.address
+  end
+
+end

data/lib/mail_address/version.rb

@@ -1,3 +1,3 @@
 module MailAddress
-  VERSION = "1.1.2"
+  VERSION = "1.2.0"
 end

data/spec/mail_address_spec.rb

@@ -445,12 +445,12 @@ describe MailAddress do
     line = 'john <john@example.com' # lack of right angle bracket
     expect {
       results = MailAddress.parse(line)
-    }.to raise_error(StandardError)
+    }.to raise_error(StandardError, "Unmatched '<>' in line")
 
     line = 'john <john@example.com> (last' # lack of right parenthesis
     expect {
       results = MailAddress.parse(line)
-    }.to raise_error(StandardError)
+    }.to raise_error(StandardError, "cannot find end paren")
   end
 
   it "unbelievable but existed address" do

data/spec/simple_parser_spec.rb

@@ -0,0 +1,86 @@
+# -*- coding: utf-8 -*-
+require 'spec_helper'
+require 'pp'
+
+#
+# These tests are almost ports of the following test code:
+#
+# https://github.com/google/closure-library/blob/master/closure/goog/format/emailaddress_test.js
+#
+
+describe MailAddress do
+
+  def assert_parsed_list(input, expected_list, opt_message = nil)
+    result = MailAddress.parse_simple input
+    expect(result.size).to eq(expected_list.size)
+    expected_list.each_with_index do |expected, index|
+      expect(result[index].address).to eq(expected)
+    end
+    result
+  end
+
+  it "simple parser - empty address" do
+    assert_parsed_list( '', [] )
+    assert_parsed_list( ',,', [] )
+  end
+
+  it "simple parser - single address" do
+    assert_parsed_list( '<foo@gmail.com>',   ['foo@gmail.com'] )
+    assert_parsed_list( '<foo@gmail.com>,',  ['foo@gmail.com'] )
+    assert_parsed_list( '<foo@gmail.com>, ', ['foo@gmail.com'] )
+    assert_parsed_list( ',<foo@gmail.com>',  ['foo@gmail.com'] )
+    assert_parsed_list( ' ,<foo@gmail.com>', ['foo@gmail.com'] )
+  end
+
+  it "simple parser - single address" do
+    assert_parsed_list( '<foo@gmail.com>, <bar@gmail.com>',   ['foo@gmail.com', 'bar@gmail.com'] )
+    assert_parsed_list( '<foo@gmail.com>, <bar@gmail.com>,',   ['foo@gmail.com', 'bar@gmail.com'] )
+    assert_parsed_list( '<foo@gmail.com>, <bar@gmail.com>, ',   ['foo@gmail.com', 'bar@gmail.com'] )
+    assert_parsed_list(
+      'John Doe <john@gmail.com>; Jane Doe <jane@gmail.com>, <jerry@gmail.com>',
+      ['john@gmail.com', 'jane@gmail.com', 'jerry@gmail.com']
+      )
+    assert_parsed_list(
+      'aaa@gmail.com, "bbb@gmail.com", <ccc@gmail.com>, (ddd@gmail.com), [eee@gmail.com]',
+      ['aaa@gmail.com', nil, 'ccc@gmail.com', nil, nil],
+      )
+  end
+
+  it "testparseListWithQuotedSpecialChars" do
+    res = assert_parsed_list(
+      'a\\"b\\"c <d@e.f>,"g\\"h\\"i\\\\" <j@k.l>',
+      ['d@e.f', 'j@k.l']
+      )
+    expect(res[0].phrase).to eq('a"b"c')
+    expect(res[1].phrase).to eq('g"h"i\\')
+  end
+
+  it "testparseListWithCommaInLocalPart" do
+    res = assert_parsed_list(
+      '"Doe, John" <doe.john@gmail.com>, <someone@gmail.com>',
+      ['doe.john@gmail.com', 'someone@gmail.com'])
+    expect(res[0].phrase).to eq('Doe, John')
+    expect(res[1].phrase).to eq('')
+  end
+
+  it "testparseListWithWhitespaceSeparatedEmails" do
+    res = assert_parsed_list(
+      'a@b.com <c@d.com> e@f.com "G H" <g@h.com> i@j.com',
+      ['a@b.com', 'c@d.com', 'e@f.com', 'g@h.com', 'i@j.com']);
+    expect(res[3].phrase).to eq('G H')
+  end
+
+  it "testparseListSystemNewlines" do
+    # These Windows newlines can be inserted in IE8, or copied-and-pasted from
+    # bad data on a Mac, as seen in bug 11081852.
+    assert_parsed_list("a@b.com\r\nc@d.com", ['a@b.com', 'c@d.com'],
+      'Failed to parse Windows newlines');
+    assert_parsed_list("a@b.com\nc@d.com", ['a@b.com', 'c@d.com'],
+      'Failed to parse *nix newlines');
+    assert_parsed_list("a@b.com\n\rc@d.com", ['a@b.com', 'c@d.com'],
+      'Failed to parse obsolete newlines');
+    assert_parsed_list("a@b.com\rc@d.com", ['a@b.com', 'c@d.com'],
+      'Failed to parse pre-OS X Mac newlines');
+  end
+
+end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: mail_address
 version: !ruby/object:Gem::Version
-  version: 1.1.2
+  version: 1.2.0
 platform: ruby
 authors:
 - Kizashi Nagata
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2015-04-27 00:00:00.000000000 Z
+date: 2015-05-26 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: bundler
@@ -90,9 +90,11 @@ files:
 - lib/mail_address.rb
 - lib/mail_address/address.rb
 - lib/mail_address/mail_address.rb
+- lib/mail_address/simple_parser.rb
 - lib/mail_address/version.rb
 - mail_address.gemspec
 - spec/mail_address_spec.rb
+- spec/simple_parser_spec.rb
 - spec/spec_helper.rb
 homepage: https://github.com/kizashi1122/mail_address
 licenses:
@@ -120,4 +122,5 @@ specification_version: 4
 summary: Simple Mail Address Parser
 test_files:
 - spec/mail_address_spec.rb
+- spec/simple_parser_spec.rb
 - spec/spec_helper.rb