koruri/source/trend_word.rb

†

source code †

   require 'rubygems'
   require 'dbi'
   require 'mecab_util.rb'
   
   # ---------------------------------------
   # --- define
   # ---------------------------------------
   DBNAME = 'YOURDB'
   DBLOGIN = 'LOGIN'
   DBPASS = 'PASS'
   VERB = false
   
   # ---------------------------------------
   # --- subroutine
   # ---------------------------------------
   
   def add_trend_noun(db, noun)
     id1 = 0
     sql = 'SELECT id FROM trend_word WHERE word=?'
     sth = db.execute(sql, noun)
     sth.each do |row|
       id1 = row[0].to_i
     end
     sth.finish
     if id1 == 0 then
       sql = 'INSERT INTO trend_word VALUES (NULL,?,?)'
       db.do(sql, noun, Time.now.to_i)
     else
       sql = 'UPDATE trend_word SET lastupdate=? WHERE id=?'
       db.do(sql, Time.now.to_i, id1)
     end
   end
   
   def process_t_cloud(db, tweet)
     line = []
     tweet.each do |chunk|
       line << chunk
     end
     spot = line[0].split
     trendword = spot[0]
     if trendword =~ /^[^@#]/ then
       puts trendword if VERB
       add_trend_noun(db, trendword)
     end
   end
   
   def process_buzztter(db, tweet)
     chunk = tweet.split
     chunk.each do |line|
       line.gsub!(/,$/, '')
       next if line =~ /^http/
       next if line =~ /^HOT:/
       next if line =~ /^[@#\[]/
       token = MecabParse.new(line)
       nounlist = token.get_noun
       nounlist.each do |yomi|
         noun = yomi.split
         trendword = noun[0]
         next if trendword =~ /^[@#\[]/
         next if trendword =~ /:/
         puts trendword if VERB
         add_trend_noun(db, trendword)
       end 
     end
   end
   
   # ---------------------------------------
   # --- main routine
   # ---------------------------------------
   
   # --- open DB
   db = DBI.connect(DBNAME, DBLOGIN, DBPASS)
   db.do('SET NAMES utf8')
   
   # --- tweet query
   sql = "SELECT id,text FROM tweet_log WHERE process=1 AND screen_name='t_cloud'"
   item_num = 0
   begin
     puts "open t_cloud log" if VERB
     sth = db.execute(sql)
     puts "start" if VERB
     sth.each do |row|
       process_t_cloud(db, row[1])
       donesql = 'UPDATE tweet_log SET process=2 WHERE id=?'
       db.do(donesql, row[0].to_i)
     end
     sth.finish
   end
   
   sql = "SELECT id,text FROM tweet_log WHERE process=1 AND screen_name='buzztter'"
   item_num = 0
   begin
     puts "open buzztter log" if VERB
     sth = db.execute(sql)
     puts "start" if VERB
     sth.each do |row|
       process_buzztter(db, row[1])
       donesql = 'UPDATE tweet_log SET process=2 WHERE id=?'
       db.do(donesql, row[0].to_i)
     end
     sth.finish
   end