同样的代码和测试网页，在 Mac 下能用，在 Ubuntu 就出错。两台机器都是 Ruby2.0.0，locale 设置也一样。
用 Hpricot 解析网页。
网页是 gb2312 编码的，读取的时候转为 UTF-8 了。
y = Iconv.iconv("UTF-8//IGNORE", "gb2312",f.read)[0]
irb(main):069:0> y.encoding => #Encoding:UTF-8
irb(main):070:0> d1 = Hpricot(y)
ArgumentError: invalid byte sequence in US-ASCII
from /usr/local/lib/ruby/gems/2.0.0/gems/hpricot-0.8.6/lib/hpricot/builder.rb:10:in gsub'
from /usr/local/lib/ruby/gems/2.0.0/gems/hpricot-0.8.6/lib/hpricot/builder.rb:10:inuxs'
from /usr/local/lib/ruby/gems/2.0.0/gems/hpricot-0.8.6/lib/hpricot/inspect.rb:63:in block (2 levels) in pretty_print_stag'
from /usr/local/lib/ruby/gems/2.0.0/gems/hpricot-0.8.6/lib/hpricot/inspect.rb:60:ineach'
from /usr/local/lib/ruby/gems/2.0.0/gems/hpricot-0.8.6/lib/hpricot/inspect.rb:60:in block in pretty_print_stag'
from /usr/local/lib/ruby/2.0.0/prettyprint.rb:381:ingroup'
from /usr/local/lib/ruby/gems/2.0.0/gems/hpricot-0.8.6/lib/hpricot/inspect.rb:56:in pretty_print_stag'
from /usr/local/lib/ruby/gems/2.0.0/gems/hpricot-0.8.6/lib/hpricot/inspect.rb:41:inblock in pretty_print'
from /usr/local/lib/ruby/2.0.0/prettyprint.rb:381:in group'
from /usr/local/lib/ruby/gems/2.0.0/gems/hpricot-0.8.6/lib/hpricot/inspect.rb:40:inpretty_print'
from /usr/local/lib/ruby/2.0.0/pp.rb:155:in block in pp'
from /usr/local/lib/ruby/2.0.0/prettyprint.rb:381:ingroup'
from /usr/local/lib/ruby/2.0.0/pp.rb:155:in pp'
from /usr/local/lib/ruby/gems/2.0.0/gems/hpricot-0.8.6/lib/hpricot/inspect.rb:47:inblock (2 levels) in pretty_print'
from /usr/local/lib/ruby/gems/2.0.0/gems/hpricot-0.8.6/lib/hpricot/inspect.rb:47:in each'
from /usr/local/lib/ruby/gems/2.0.0/gems/hpricot-0.8.6/lib/hpricot/inspect.rb:47:inblock in pretty_print'
... 11 levels...
从出错信息看，貌似当 US-ASCII 处理了？
大家以后找工作看看他家主页是不是 UTF-8 编码的，不是的话再斟酌一下吧。
另外这个 Hpricot 项目已经被 closed 了，大家还是用 nokogiri 吧。
你确定网页是 gbk 编码的话那就
'''ruby site_body = site_body.force_encoding('gb2312')
然后
site_body_utf8 = site_body.encode('utf-8') '''
上面库出了什么问题不知道，但这个库处理后告诉 ruby 这个字符串是 ascii 码了，可 ruby 一处理发现编码的二进制字节不对然后就报告错误了
那个 iconv 是调命令行的 iconv 的吗？如果是的话，貌似 gb2312 编码得用 gbk 来读。
y = Iconv.iconv("UTF-8//IGNORE", "gbk",f.read)[0]
