neinomaten/mediawikibot.rb

#------------------------------------------------------------------------------
# TODO:
# - Fix HTML parsing, no regexen! See wpEditToken+wpEdittime
# - EditToken-stuff in its own method?

module MediaWikiBot

require 'cgi'
require 'erb'
require 'http-access2'

class WikiBot

  include ERB::Util # for url_encode()

  def initialize(wiki)
    @wiki = wiki

    @client = HTTPAccess2::Client.new()
    @client.set_cookie_store("cookie.dat")
  end

  #------------------------------------------------------------------------------
  # FIXME: This should really be in HTTPAccess2::Client

  def post_form (url, post_vars) 
    body = ""
    post_vars.keys.each do |var|
      body += "&" unless body == ""
      body += url_encode(var) + "=" + url_encode(post_vars[var])
    end

    result = @client.post(url, body,
      [[ "Content-Type", "application/x-www-form-urlencoded" ]])

    return result.content
  end

  #------------------------------------------------------------------------------
  # HTTP authentication

  def set_basic_auth(user_id, passwd)
    @client.set_basic_auth(@wiki, user_id, passwd)
  end

  #------------------------------------------------------------------------------
  # MediaWiki stuff: Could be useful for other wiki bots

  def url_no_redirect (title) 
    return @wiki + "index.php?title=" + url_encode(title) + "&redirect=no"
  end
  
  def url_page (title)
    return @wiki + "index.php?title=" + url_encode(title)
  end

  def url_raw (title)
    return @wiki + "index.php?title=" + url_encode(title) + "&action=raw"
  end

  def url_what_links_here (title)
    return @wiki \
        + "index.php?title=Special:Whatlinkshere&target=" + url_encode(title)
  end 

  def url_delete (title)
    return @wiki + "index.php?title=" + url_encode(title) + "&action=delete"
  end

  def url_submitlogin
    return @wiki + "index.php?title=Special:Userlogin&action=submitlogin"
  end

  def url_protect(title)
    return @wiki + "index.php?title=" + url_encode(title) + "&action=protect"
  end

  def login(wiki_name, wiki_password)

    post_form(url_submitlogin(), 
      { "wpName" => wiki_name,
        "wpPassword" => wiki_password,
        "wpLoginattempt" => "" })
  end

  def is_redirect? (title)
    return get_raw(title) =~ /^#REDIRECT[: ]/
  end

  def is_not_linked? (title) 
    return get_what_links_here(title).size() == 0
  end

  def get_raw (title)
    return @client.get_content(url_raw(title))
  end

  def get_allpages
    # FIXME: There seems to be no MediaWiki API? Let's scrape it up from 
    # HTML for now.

    allpages = []

    allpages_page = @client.get_content(@wiki + "Special:Allpages")
    allpages_page.scan(/title="(.*?)"/) do |m|
      title = CGI.unescapeHTML(m[0])
      allpages.push(title)
    end

    return allpages
  end

  def get_redirect (title)
    return get_raw(title).scan(/^#REDIRECT[: ]\[\[(.*)\]\]/)[0][0] # first match
  end

  def get_what_links_here (title)
    # FIXME: There seems to be no MediaWiki API? Let's scrape it up from 
    # HTML for now.
    
    what_links_here = []
    
    what_links_here_page = @client.get_content(url_what_links_here(title))
    what_links_here_page.scan(/<li><a href=".*?" title="(.*?)"/) do |m|
      title = CGI.unescapeHTML(m[0])
      what_links_here.push(title)
    end

    return what_links_here
  end 

  def delete (title, reason)
    $stderr.print("Deleting '", title, "'\n")
    token_page = post_form(url_delete(title),
      { "wpReason" => reason,
        "wpConfirm" => "1",})
    token = token_page.scan(/name='wpEditToken' value="(.*?)"/)[0][0]
    post_form(url_delete(title),
      { "wpReason" => reason,
        "wpConfirm" => "1",
        "wpEditToken" => token })
  end

  def replace (replace_where, replace_what, replace_with, reason)
    # " " could be "_"
    # FIXME: Shouldn't be done here
    replace_what = replace_what.gsub(/ /, "[ _]")

    $stderr.print("Replacing /", replace_what, "/ with '",
      replace_with, "' in '", replace_where, "'.\n")

    replace_what_re = Regexp.new(replace_what)
    before = get_raw(replace_where)
    after = before.gsub(replace_what_re, replace_with)

    edit(replace_where, after, reason)
  end

  def url_edit(title)
    return @wiki + "index.php?title=" + url_encode(title)+ "&action=edit" 
  end

  def url_edit_submit(title)
    return @wiki + "index.php?title=" + url_encode(title)+ "&action=submit" 
  end

  def edit (title, body, summary)
    $stderr.print("Submitting '", title, "'.\n")

    token_page = @client.get_content(url_edit(title))
    while ! token_page.match(/value="(.*?)" name="wpEdittime" /) 
      # FIXME: This workaround loop really fucking sucks.
      token_page = @client.get_content(url_edit(title))
    end
    time = token_page.scan(/value="(.*?)" name="wpEdittime" /)[0][0]

    # No token for MediaWiki 1.3.x
    if token_page =~ /wpEditToken/ 
      token = token_page.scan(/value="(.*?)" name="wpEditToken" /)[0][0]
    end

    post_form(url_edit_submit(title),
      { "wpTextbox1" => body,
        "wpSummary" => summary,
        "wpEditToken" => token, 
        "wpEdittime" => time, 
        "wpSave" => "save" })
  end

  def get_what_uses_template(title)
    return get_what_links_here(title)
  end

  def parse_template(title, template)
    template_re = Regexp.new("\\\{\\\{" + template + "(.*?)\\\}\\\}")

    fields = {}

    fields_string = get_raw(title).scan(template_re)[0][0]
    fields_string.split(/\|/).each do |field|
      (key, value) = field.split(/=/)
      fields[key] = value
    end

    return fields
  end
  
  def get_category_articles(category)
    # FIXME: There seems to be no MediaWiki API? Let's scrape it up from 
    # HTML for now.

    articles = []

    category_page = @client.get_content(@wiki + url_encode(category))
    category_page.scan(/<li><a href=".*?" title="(.*?)"/) do |m|
      title = CGI.unescapeHTML(m[0])
      articles.push(title)
    end

    return articles
  end

  def get_categories(title)

    if is_redirect?(title) 
      return []
    end
    
    categories = []
    categories_html = @client.get_content(url_page(title)).scan(/<div id="catlinks">(.*?)<\/div>/)[0][0]
    categories_html.scan(/title="(.*?)"/) do |m|
      category = CGI.unescapeHTML(m[0])
      categories.push(category)
    end

    return categories[1..-1]
  end
    
  def is_in_category?(title, category)
    
    if is_redirect?(title) 
      return false
    end

    get_categories(title).member?(category)
  end

  def is_protected?(title)
    return @client.get_content(url_page(title)) =~ /action=unprotect/
  end

  def protect(title, reason)
    
    token_page = @client.get_content(url_protect(title))
    # No token for MediaWiki 1.3.x
    if token_page =~ /wpEditToken/ 
      token = token_page.scan(/name='wpEditToken' value="(.*?)"/)[0][0]
    end

    post_form(url_protect(title),
      { "wpReasonProtect" => reason,
        "wpConfirmProtectB" => "confirm",
        "wpEditToken" => token })
  end

end # class WikiBot

end # module MediaWikiBot