#!/usr/bin/ruby $LOAD_PATH.push '/usr/local/etc' class String def urlencode unpack('H*').first.upcase.gsub(/(..)/, "%\\1") end end class Time def rfc1123 utc.strftime('%a, %d %b %Y %H:%M:%S GMT') end def jst (utc + 32400).strftime('%Y-%m-%d %H:%M:%S') end def mysql utc.strftime('%Y-%m-%d %H:%M:%S') end def to_atom fmt = ('%%Y-%%m-%%dT%%H:%%M:%%S.%06uZ' % usec) utc.strftime(fmt) end MONTAB = { :jan => 1, :feb => 2, :mar => 3, :apr => 4, :may => 5, :jun => 6, :jul => 7, :aug => 8, :sep => 9, :oct => 10, :nov => 11, :dec => 12 } def Time.parse str begin case str when Mysql::Time then # considered UTC ymdhms = [:year, :month, :day, :hour, :minute, :second].map{|sym| str.send(sym) } STDERR.puts ymdhms.inspect if $DEBUG if ymdhms.first.zero? then nil else Time.gm(*ymdhms) end when /(\d+)\s+(\w+)\s+(\d+)\s+(\d+):(\d+):(\d+)\s+GMT/ then d, m, y, h, n, s = $1, $2, $3, $4, $5, $6 d, y, h, n, s = [d, y, h, n, s].map{|s| s.to_i} m = MONTAB[m.downcase.to_sym].to_i Time.gm(y, m, d, h, n, s) when /(\d+)-(\d+)-(\d+)[T ](\d+):(\d+):(\d+(?:\.\d+)?)Z/ then y, m, d, h, n = [$1, $2, $3, $4, $5].map{|s| s.to_i} s = $6.to_f Time.gm(y, m, d, h, n, s) when /(\d+)-(\d+)-(\d+)[T ](\d+):(\d+):(\d+(?:\.\d+)?)([-+]\d\d):?(\d\d)/ then y, m, d, h, n, zh, zn = [$1, $2, $3, $4, $5, $7, $8].map{|s| s.to_i} STDERR.puts [y, m, d, h, n, s, zh, zn].inspect if $DEBUG s = $6.to_f Time.gm(y, m, d, h, n, s) - (zh * 60 + zn) * 60 else raise "unknown datetime format '#{str}'" end rescue => e STDERR.puts "\t" + e.backtrace.join("\n\t") Time.now end end end module PSHBSpool class HtmlFeed def initialize url, css, title = 'title' require 'rexml/document' require 'digest/md5' @url, @css, @title = url, css, title @n = 0 @updated = Time.at(0) # to be completed by compile() @doc = @root = @ul = nil @cols = %w< updated title > end def add_col *name name.each{|n| @cols.push(n.to_s) } end def ins_link @cols.unshift('link') end COLN = { 'updated' => '発表時刻', #MAX(utime) 'title' => '表題', 'author' => '発表官署名', #pbof 'summary' => '見出し', 'Xstatus' => '運用種別', 'Xedof' => '編集官署名', 'Xitype' => '情報形態', 'Xevid' => '識別情報', 'Xexpire' => '有効期限', 'Xrtime' => '公式発表時刻', 'abr' => 'データ種類コード', } require 'pshbspool-cfg' PRMS = PSHBSpool::PRMS def compile return if @doc ths = @cols.map{|name| "#{COLN[name] or name}"}.join template = < #{@title}

#{@title}

Date: #{Time.now.jst}
Last-Modified: #{@updated.to_atom}

#{ths}

ENDXML @doc = REXML::Document.new(template) @root = @doc.root @ul = REXML::XPath.first(@root, '//table[@id="MainList"]/tbody') end def add_note url, text compile ul = REXML::XPath.first(@root, '//ul[@id="Notes"]') ul.add_element('li').add_element('a', 'href'=>url.to_s).add_text(text.to_s) end def to_s compile REXML::XPath.first(@root, '//span[@id="Updated"]').text = @updated.jst head = <@n.to_s) ent.add_attribute('data-ip', attrs['#postid'].to_s) if attrs['#postid'] @updated = [@updated, updated].compact.max attrs['updated'] = updated.jst for col in @cols if 'link' == col then td = ent.add_element('td', 'title'=>col) for mcol in attrs.keys.grep(/^=/).sort kw = mcol[1..-1] td.add_element('a', 'class'=>kw, 'href'=>attrs[mcol].to_s).add_text(kw) end next end collink = '@' + col if attrs[collink] then ent.add_element('td', 'title'=>col) \ .add_element('a', 'href'=>attrs[collink].to_s) \ .add_text(attrs[col].to_s) else ent.add_element('td', 'title'=>col).add_text(attrs[col].to_s) end end @ul.add_text("\n") if attrs['summary'] then ent = @ul.add_element('tr', 'data-ir'=>@n.to_s) ent.add_element('td', 'title'=>"summary", 'colspan'=>@cols.size.to_s, 'class'=>'summary').add_text(attrs['summary'].to_s) @ul.add_text("\n") end end def nrows compile @n + 0 end def banner text, alturl = nil compile ent = @ul.add_element('tr') td = ent.add_element('td', 'colspan'=>@cols.size) td.add_text(text) if alturl f = td.add_element('form', 'style'=>'display:inline;', 'action'=>alturl, 'method'=>'GET', 'id'=>"f#{@n}") f.add_element('input', 'type'=>'submit', 'value'=>"More") end true end end class App def initialize @method = ENV['REQUEST_METHOD'].to_s @qstr = ENV['QUERY_STRING'].to_s @path = ENV['PATH_INFO'].to_s @addr = ENV['REMOTE_ADDR'].to_s @ctype = ENV['CONTENT_TYPE'].to_s @clen = ENV['CONTENT_LENGTH'] @clen = @clen.to_i if @clen @reqbody = nil @myname = nil end def database $stderr.puts STORAGE.inspect if $DEBUG db = Mysql.connect(STORAGE[:srv], STORAGE[:usr], STORAGE[:pwd], STORAGE[:db]) begin yield db ensure db.close end end def prepare(db, sql) st = db.prepare(sql) begin yield st ensure st.close end end def myname return @myname if @myname host = ENV['SERVER_NAME'] || 'localhost' port = ENV['SERVER_PORT'] || '80' script = ENV['SCRIPT_NAME'] @myname = "http://#{host}:#{port}#{script}" if PRMS[:urlhook] PRMS[:urlhook].call(@myname) end @myname end JMXNS = { 'jmx' => 'http://xml.kishou.go.jp/jmaxml1/', 'ib' => 'http://xml.kishou.go.jp/jmaxml1/informationBasis1/' } XPATHS = { :title => '/jmx:Report/jmx:Control/jmx:Title', :status => '/jmx:Report/jmx:Control/jmx:Status', :edof => '/jmx:Report/jmx:Control/jmx:EditorialOffice', :evid => '/jmx:Report/ib:Head/ib:EventID', :utime => '/jmx:Report/jmx:Control/jmx:DateTime', :expire => '/jmx:Report/ib:Head/ib:ValidDateTime', :pbof => '/jmx:Report/jmx:Control/jmx:PublishingOffice', :rtime => '/jmx:Report/ib:Head/ib:ReportDateTime', :ikind => '/jmx:Report/ib:Head/ib:InfoKind', :hdline => '/jmx:Report/ib:Head/ib:Headline/ib:Text', :itype => '/jmx:Report/ib:Head/ib:InfoType', } def updatedb require 'rexml/document' n = 0 database {|db| rows = [] # phase 1 sql = <<-ENDSQL SELECT postid, uri, body FROM msgs WHERE postid > (SELECT IFNULL(MAX(postid), 0) FROM jmxmsgs) ORDER BY postid ASC LIMIT 40 ENDSQL prepare(db, sql) {|st| st.execute while row = st.fetch next unless /^urn:uuid/ === row[1] rows.push row end } # phase 2 db.autocommit(false) sql = <<-ENDSQL REPLACE INTO jmxmsgs(postid, uri, ins, title, status, edof, evid, utime, expire, pbof, rtime, ikind, hdline, itype) VALUES (?, ?, NOW(), ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?) ENDSQL prepare(db, sql) {|st| for row in rows postid, uri, body = row next if body.empty? h = {} begin d = REXML::Document.new(body) rescue REXML::ParseException => e next end for k, x in XPATHS v = REXML::XPath.first(d.root, x + '/text()', JMXNS) h[k] = v.to_s end utime = Time.parse(h[:utime]).mysql expire = if h[:expire].empty? then '0000-00-00T00:00:00' else Time.parse(h[:expire]).mysql end rtime = Time.parse(h[:rtime]).mysql #STDERR.puts h.inspect xrow = [postid, uri, h[:title], h[:status], h[:edof], h[:evid], utime, expire, h[:pbof], rtime, h[:ikind], h[:hdline], h[:itype] ] st.execute(*xrow) n = n.succ end } db.commit } < tnow STDERR.puts "CMP PASSTHRU" if $DEBUG end end def mlatest check_ims uri_this = File.join(myname, @path) title_feed = "新着電文" a = HtmlFeed.new(uri_this, './res/tab.css', title_feed) a.add_col(*%w( author Xstatus Xedof Xevid Xrtime )) a.ins_link database {|db| sql = <<-ENDSQL SELECT uri, title, status, evid, utime, pbof, rtime, edof, hdline, postid FROM jmxmsgs ORDER BY postid DESC LIMIT 50 ENDSQL prepare(db, sql) {|st| st.execute while row = st.fetch uri, title, status, evid, utime, pbof, rtime, edof, hdline, postid = row linkto = File.join(myname, "/msg/#{uri}") raw = File.join(File.dirname(myname), 'pshbspool', 'entry', uri) hdline = nil if hdline.empty? a.add_ent(Time.parse(utime), '=JMX'=>raw, 'title'=>title, '=HTML'=>linkto, 'author'=>pbof, 'summary'=>hdline, 'Xstatus'=>status, 'Xedof'=>edof, 'Xevid'=>evid, 'Xrtime'=> begin Time.parse(rtime).jst rescue '-' end, '#postid'=>postid.to_s) end } } uri = File.join(myname, '/titles.html') a.add_note(uri, '情報名称新着順一覧') a.to_s end def cutoff_time nposts utime = nil sql = <<-ENDSQL SELECT MAX(utime) FROM jmxmsgs WHERE postid <= (SELECT MAX(postid) - ? FROM jmxmsgs) ENDSQL database {|db| prepare(db, sql) {|st| st.execute(nposts) utime, dummy = st.fetch } } if utime then Time.parse(utime).jst else '-' end end def mlatest_bytitle(title, speedhack = false) check_ims uri_this = File.join(myname, @path) title_feed = "新着電文 (情報名称=#{title})" a = HtmlFeed.new(uri_this, '../res/tab.css', title_feed) a.add_col(*%w( author Xstatus Xedof Xevid Xrtime )) a.ins_link database {|db| sql = <<-ENDSQL SELECT uri, status, evid, utime, pbof, edof, rtime, hdline, jmxmsgs.title FROM jmxmsgs -- abrhack WHERE -- speedhack title = ? ORDER BY utime DESC LIMIT 21 ENDSQL if /^V/ === title then sql.sub!(/-- abrhack/, 'LEFT JOIN jmxxsl ON jmxmsgs.title = jmxxsl.title') sql.sub!(/title = \?/, 'abr = ?') end if speedhack then sql.sub!(/-- speedhack/, "postid > (SELECT MAX(postid) - 5000 FROM jmxmsgs) AND") end $stderr.puts sql if $DEBUG prepare(db, sql) {|st| st.execute(title) while row = st.fetch uri, status, evid, utime, pbof, edof, rtime, hdline, xtitle = row linkto = File.join(myname, "/msg/#{uri}") raw = File.join(File.dirname(myname), 'pshbspool', 'entry', uri) hdline = nil if hdline.empty? a.add_ent(Time.parse(utime), '=JMX'=>raw, 'title'=>xtitle, '=HTML'=>linkto, 'author'=>pbof, 'summary'=>hdline, 'Xstatus'=>status, 'Xedof'=>edof, 'Xevid'=>evid, 'Xrtime'=> begin Time.parse(rtime).jst rescue '-' end ) end } } uri = File.join(myname, '/titles.html') uri_t = File.join(myname, "/t1/#{title.urlencode}") a.add_note(uri, '情報名称新着順一覧') a.add_note(uri_t, "編集官署新着順一覧 (情報名称=#{title})") if a.nrows.zero? then if speedhack then url_a = File.join(myname, "/m6/#{title.urlencode}") ct = cutoff_time(5000) a.banner "検索範囲 #{ct} 以後にはみつかりません。", url_a else a.banner "みつかりません。" end end a.to_s end def mlatest_bytitle_edof(title, edof, speedhack = false) check_ims uri_this = File.join(myname, @path) title_feed = "新着電文 (情報名称=#{title} 編集官署=#{edof})" a = HtmlFeed.new(uri_this, '../../res/tab.css', title_feed) a.add_col(*%w( author Xstatus Xedof Xevid Xrtime )) a.ins_link database {|db| sql = <<-ENDSQL SELECT uri, title, status, evid, utime, pbof, rtime, hdline, postid FROM jmxmsgs WHERE -- speedhack title = ? AND edof = ? ORDER BY utime DESC LIMIT 21 ENDSQL if speedhack then sql.sub!(/-- speedhack/, "postid > (SELECT MAX(postid) - 20000 FROM jmxmsgs) AND") end prepare(db, sql) {|st| st.execute(title, edof) while row = st.fetch uri, title, status, evid, utime, pbof, rtime, hdline, postid = row linkto = File.join(myname, "/msg/#{uri}") raw = File.join(File.dirname(myname), 'pshbspool', 'entry', uri) hdline = nil if hdline.empty? a.add_ent(Time.parse(utime), '=JMX'=>raw, 'title'=>title, '=HTML'=>linkto, 'author'=>pbof, 'summary'=>hdline, 'Xstatus'=>status, 'Xedof'=>edof, 'Xevid'=>evid, 'Xrtime'=> begin Time.parse(rtime).jst rescue '-' end, '#postid'=>postid.to_s) end } } uri = File.join(myname, '/titles.html') uri_t = File.join(myname, "/t1/#{title.urlencode}") a.add_note(uri, '情報名称新着順一覧') a.add_note(uri_t, "編集官署新着順一覧 (情報名称=#{title})") if a.nrows.zero? then if speedhack then url_a = File.join(myname, "/m7/#{title.urlencode}/#{edof.urlencode}") ct = cutoff_time(20_000) a.banner "検索範囲 #{ct} 以後にはみつかりません。", url_a else a.banner "みつかりません。" end end a.to_s end def eolist_bytitle t check_ims uri_this = File.join(myname, @path) a = HtmlFeed.new(uri_this, '../res/tab.css', "編集官署新着順一覧 (情報名称=#{t})") a.add_col 'author' url_t = File.join(myname, "/titles.html") url_m = File.join(myname, "/m1/#{t.urlencode}") database {|db| sql = <<-ENDSQL SELECT edof, MAX(utime) AS updated FROM jmxmsgs WHERE title = ? GROUP BY edof ORDER BY updated DESC ENDSQL prepare(db, sql) {|st| st.execute(t) while row = st.fetch edof, updated = row uri = File.join(myname, "/m2/#{t.urlencode}/#{edof.urlencode}") a.add_ent(Time.parse(updated), 'title'=>t, '@title'=>url_m, 'author'=>edof, '@author'=>uri) end } } a.add_note(url_t, "情報名称新着順一覧") a.add_note(url_m, "新着電文 (情報名称=#{t})") a.to_s end def tlist check_ims uri_this = File.join(myname, @path) a = HtmlFeed.new(uri_this, './res/tab.css', "情報名称新着順一覧") a.add_col 'abr' upd = nil database {|db| sql = <<-ENDSQL SELECT jmxmsgs.title, MAX(utime) AS updated, abr FROM jmxmsgs LEFT JOIN jmxxsl ON jmxmsgs.title = jmxxsl.title GROUP BY jmxmsgs.title ORDER BY updated DESC ENDSQL prepare(db, sql) {|st| st.execute while row = st.fetch title, updated, abr = row uri = File.join(myname, "/t1/#{title.urlencode}") a.add_ent(Time.parse(updated), 'title'=>title, '@title'=>uri, 'abr'=>abr) end } } uri = File.join(myname, '/msgs.html') a.add_note(uri, '直近電文一覧') a.to_s end def showday yyyy, mm, dd check_ims day1 = "#{yyyy}-#{mm}-#{dd}" day2 = (Time.gm(yyyy.to_i, mm.to_i, dd.to_i) + 86400).strftime('%Y-%m-%d') body = [] insmax = "1999-12-31 12:34:56" $stderr.puts "#showday ins > #{day1} AND ins < #{day2}" if $stderr.tty? database {|db| sql = <<-ENDSQL SELECT uri, ins FROM msgs WHERE ins > ? and ins < ? ENDSQL prepare(db, sql) {|st| st.execute(day1, day2) while row = st.fetch uri, ins = row ins = ins.to_s insmax = ins if ins > insmax body.push [uri, "\t", ins].join end } } utime = Time.parse(insmax + 'Z') head = < e puts < e puts < e puts <