编辑代码

data = "192139-92-7（100g）
五甲基环戊二烯基双(三苯基膦)氯化钌(II) CAS：92361-49-4 100g
异常数据测试
四氯钯酸铵 13820-40-1（100g）"
cas_pattern = /[0-9]+[-][0-9]+[-][0-9]+/
spec_pattern = /[a-zA-Z]/
data.split("\n").map{|line| line.strip}.each do |line|
    cas_matches = line.match cas_pattern
    if cas_matches === nil || cas_matches.length == 0
        puts "'#{line}' cas解析失败!!!"
        next
    end
    cas_begin_index = cas_matches.begin(0)
    cas_end_index = cas_matches.end(0)
    product_name = line[0, cas_begin_index].to_s.gsub("CAS"," ").gsub("："," ").gsub(":"," ").strip
    cas = line[cas_begin_index,cas_end_index - cas_begin_index].to_s.strip
    specification = line[cas_end_index, line.length-1].to_s.gsub("（","").gsub("）","").gsub("(","").gsub(")","").strip
    spec_matches = specification.match spec_pattern
    if spec_matches === nil || spec_matches.length == 0
        puts "'#{line}' 规格解析失败!!!"
        next
    end
    unit_begin_index = spec_matches.begin(0)
    weight = specification[0, unit_begin_index]
    unit = specification[unit_begin_index, specification.length - 1]
    puts "#{product_name}\t#{cas}\t#{weight}\t#{unit}"
end