Rake 脚本
desc "Update departments and classes listing"
task :update_classes => :environment do
require 'rubygems'
require 'mechanize'
require 'nokogiri'
def getSubject(agent, subject)
#subj_form.field_with(:name=>"SUBJ").options.first.unselect
page_form = agent.get('https://was.nd.edu/reg/srch/ClassSearchServlet').form()
select = page_form.field_with(:name => "SUBJ")
select.select_none()
pp subject
#pp select.
select.options_with(:value => "#{subject}").each do |field|
field.select
end
response = agent.submit(page_form)
#pp page.body
results = response.body
pp "Response submitted succesfully!"
doc = Nokogiri::HTML(results, 'UTF-8')
pp "Passed the doc"
parseHTML(doc, subject)
end
def parseHTML(doc, subject)
#Find nodes by XPATH
pp "Enter the parseHTML function"
rows = doc.xpath('//table[@id="resulttable"]/tbody/tr')
pp "Trying to find department by key . . . "
#@department = Department.where(:key => subject)
pp "Success!"
#cells = rows.css('td')
details = rows.collect do |row|
detail = {}
[
[:subject, subject],
[:course, 'td[1]/a[1]'],
[:section, 'td[1]/a[1]'],
[:title, 'td[2]'],
[:credits, 'td[3]'],
[:status, 'td[4]'],
[:max_spots, 'td[5]'],
[:open_spots, 'td[6]'],
[:xlst, 'td[7]'],
[:crn, 'td[8]'],
[:syl, 'td[9]'],
[:instructor, 'td[10]/a'],
[:when, 'td[11]'],
[:begin, 'td[12]'],
[:end, 'td[13]'],
[:location, 'td[14]'],
].each do |name, xpath|
# Want to loop through each cell/detail, then perform actions accordingly
if name == :course
detail[name] = row.xpath(xpath).text.strip.split('-').first.strip
#pp detail[name]
elsif name == :section
detail[name] = row.xpath(xpath).text.strip.split('-').last.strip
#pp detail[name]
elsif name == :subject
detail[name] = subject
else
detail[name] = row.xpath(xpath).text.strip
#pp detail[name]
end
end
detail
end
#pp details
# Let's build our Courses array!
prev = ''
#c = Course.new(nil, nil, nil)
details.each do |d|
#temp = Course.new(d[:title], d[:course], d[:credits])
#pp prev
#unless @courses.empty?
#Check if the course number already exists in the array
if d[:course] == prev
#loc = @courses.find_index{|course| course.name == d[:title]}
#@course = Course.where(:coursenum => d[:course]).first
# s = Section.new(d[:section], d[:open_spots], d[:max_spots], d[:crn], d[:instructor], d[:when], d[:location])
#c.add(s)
#@section = Section.new(:num => d[:section], :opspots => d[:open_spots], :maxspots => d[:max_spots], :crn => d[:crn], :instructor => d[:instructor], :time => d[:when], :location => d[:location])
@section = @course.sections.new(:num => d[:section], :opspots => d[:open_spots], :maxspots => d[:max_spots], :crn => d[:crn], :instructor => d[:instructor], :time => d[:when], :location => d[:location])
@section.save
@course.save
@dept.save
#@department.save
else
#c = Course.new( d[:title], d[:course], d[:credits])
#s = Section.new(d[:section], d[:open_spots], d[:max_spots], d[:crn], d[:instructor], d[:when], d[:location])
##c.add(s)
#add(c)
#@course = Course.new( :name => d[:title], :coursenum => d[:course], :credit => d[:credits])
#@dept.save
@course = @dept.courses.new(:name => d[:title], :coursenum => d[:course], :credit => d[:credits])
#@course.sections.build(:course_id => @course.id)
#@section = Section.new(:num => d[:section], :opspots => d[:open_spots], :maxspots => d[:max_spots], :crn => d[:crn], :instructor => d[:instructor], :time => d[:when], :location => d[:location])
#@section.save
@section = @course.sections.new(:num => d[:section], :opspots => d[:open_spots], :maxspots => d[:max_spots], :crn => d[:crn], :instructor => d[:instructor], :time => d[:when], :location => d[:location])
@section.save
#@course.save
@course.save
@dept.save
end
unless d[:course] == nil
prev = d[:course]
end
end
#exportHash(subject, @courses)
#@courses.each do |co|
# pp co
#end
end
#def exportHash(subject, courselist)
# File.open("courses/#{subject}.yml", File::WRONLY|File::CREAT) do |file|
# file.write courselist.to_yaml
#end
#end
#end
a = Mechanize.new
page = a.get('https://was.nd.edu/reg/srch/ClassSearchServlet')
subj_form = page.form()
options = subj_form.field_with(:name=>"SUBJ").options
options.each { |op|
@dept = Department.new(:deptname => op.text.strip, :key => op.value.strip)
@dept.save
getSubject(a, op)
#pp op.text.strip
#getSubject(a, op)
}
end
所有模型、部门、课程和部分。
class Course < ActiveRecord::Base
belongs_to :department
attr_accessor :department_id
attr_accessible :coursenum, :credit, :name, :course_id, :department
validates :coursenum, :presence => true, :uniqueness => true
validates :credit, :presence => true
validates :name, :presence => true
has_many :sections, :autosave => true
accepts_nested_attributes_for :sections
end
class Department < ActiveRecord::Base
attr_accessible :deptname, :key
validates :key, :presence => true, :uniqueness => true
validates :deptname, :presence => true, :uniqueness => true
has_many :courses, :autosave => true
accepts_nested_attributes_for :courses
end
class Section < ActiveRecord::Base
attr_accessible :crn, :instructor, :location, :maxspots, :num, :opspots, :time
validates :crn, :presence => true, :uniqueness => true
validates :instructor, :presence => true
validates :location, :presence => true
validates :maxspots, :presence => true
validates :num, :presence => true
validates :opspots, :presence => true
validates :time, :presence => true
belongs_to :course
attr_accessor :course_id
#validates :num, :presence => true, :uniqueness => true
end
没有抛出错误，但我不知道为什么它不存储数据。正如代码所示，我正在使用 Mechanize/Nokogiri 来解析一个 HTML 页面，该页面是一个表单自动化的类和部分，但它似乎没有将新课程存储在正确的部门中，如果有的话。. . 我已经坚持了好几个小时，这让我发疯。