#!/usr/bin/env node

var fs = require('fs')
var path = require('path')
var URL = require('url')
var http = require('http')
var https = require('https')
var crypto = require('crypto')
var readline = require('readline')
var os = require('os')

var ssbClient = require('ssb-client')
var pull = require('pull-stream')

var pkg = require('./package')

var userAgentBase = pkg.name + '/' + pkg.version
var userAgentContact
var userAgentBot = false

function estimateMessageSize(content) {
  var draftMsg = {
    key: '%0000000000000000000000000000000000000000000=.sha256',
    value: {
      previous: '%0000000000000000000000000000000000000000000=.sha256',
      author: '@0000000000000000000000000000000000000000000=.ed25519',
      sequence: 100000,
      timestamp: 1000000000000.0001,
      hash: 'sha256',
      content: content,
      signature: '00000000000000000000000000000000000000000000000000000000000000000000000000000000000000==.sig.ed25519'
    }
  }
  return JSON.stringify(draftMsg, null, 2).length
}

function getJson(url, cb) {
  var opts = URL.parse(url)
  opts.headers = {
    'User-Agent': userAgentBase
      + (userAgentContact ? ' (' + userAgentContact + ')' : '')
      + (userAgentBot ? ' bot' : '')
  }
  var h = opts.protocol === 'https:' ? https : http
  h.get(opts, function (res) {
    if (res.statusCode !== 200) {
      console.error(res.headers, url)
      return cb(new Error('HTTP ' + res.statusCode + ' ' + res.statusMessage))
    }
    var bufs = []
    res.on('data', function (buf) {
      bufs.push(buf)
    })
    res.on('end', function () {
      res.removeListener('error', cb)
      var buf = Buffer.concat(bufs)
      bufs = null
      var data
      try {
        data = JSON.parse(buf.toString('utf8'))
      } catch(e) {
        return cb(e)
      }
      cb(null, data)
    })
    res.on('error', cb)
  })
}

function publishDrafts(sbot, drafts, cb) {
  var draftIdIndex = {}
  drafts.forEach(function (draft, i) {
    draftIdIndex[draft.draftId] = i
  })
  var ids = []

  function replaceDraftIds(obj) {
    if (typeof obj === 'string') {
      var i = draftIdIndex[obj]
      if (typeof i === 'number') {
        var id = ids[i]
        if (!id) throw new ReferenceError('draft referernces unknown message')
        return id
      }
    } else if (Array.isArray(obj)) {
      return obj.map(replaceDraftIds)
    } else if (obj !== null && typeof obj === 'object') {
      var o = {}
      for (var k in obj) o[k] = replaceDraftIds(obj[k])
      return o
    }
    return obj
  }

  pull(
    pull.values(drafts),
    pull.asyncMap(function (draft, cb) {
      var content = replaceDraftIds(draft.content)
      sbot.publish(content, function (err, msg) {
        if (err) return cb(err)
        ids.push(msg.key)
        cb(null, msg)
      })
    }),
    pull.collect(cb)
  )
}

var args = process.argv.slice(2)
var yes = false
var dry = false
var help = false
var urls = []
args.forEach(function (arg) {
  if (arg[0] === '-') switch (arg) {
    case '-n': return dry = true
    case '-y': return yes = true
    case '-h': return help = true
    default: throw 'Unknown argument: ' + arg
  } else urls.push(arg)
})

if (help) {
  process.stdout.write(fs.readFileSync(path.join(__dirname, 'usage.txt')))
  process.exit(0)
}

ssbClient(function (err, sbot, config) {
  if (err) throw err
  var conf = config.wikimedia || {}
  userAgentContact = conf.contact
  userAgentBot = conf.bot

  if (urls.length === 0) {
    var pagesFile = path.join(config.path, 'wikimedia-pages.txt')
    var pagesData = fs.readFileSync(pagesFile, 'utf8')
    urls = pagesData.split('\n').filter(RegExp.prototype.test.bind(/[^#]/))
    if (!urls.length) {
      console.log('No pages to sync.')
      return sbot.close()
    }
  }

  var pagesInfo = urls.map(function (page) {
    var m = /^(https?:\/\/.*?)(\/wiki)?\/(.*)$/.exec(page)
    if (!m) throw 'Unable to parse page URL ' + page
    return {
      site: m[1] + '/',
      api: m[1] + (m[2] ? '/w' : '/wiki') + '/api.php',
      title: m[3]
    }
  })
  var pagesInfoByApi = {}
  pagesInfo.forEach(function (pageInfo) {
    var infos = pagesInfoByApi[pageInfo.api] || (pagesInfoByApi[pageInfo.api] = [])
    infos.push(pageInfo)
  })
  console.log('Normalizing titles...')
  var waiting = 0
  for (var api in pagesInfoByApi) (function (api) {
    var pagesInfoForApi = pagesInfoByApi[api]
    var pagesInfoForApiByTitle = {}
    var titles = pagesInfoForApi.map(function (info) {
      pagesInfoForApiByTitle[info.title] = info
      return info.title
    })
    var url = api + '?format=json&action=query' +
      '&titles=' + encodeURIComponent('\x1f' + titles.join('\x1f')) +
      '&' // trailing & needed for some reason
    waiting++
    getJson(url, function (err, data) {
      if (err) throw err
      if (data.warnings) console.trace('Warnings:', data.warnings)
      if (data.query.normalized) data.query.normalized.forEach(function (norm) {
        var info = pagesInfoForApiByTitle[norm.from]
        if (!info) {
          console.error(JSON.stringify({titles: titles, response: data}, 0, 2))
          throw new Error('Unexpected title in server response')
        }
        // console.log('Normalized title', norm.from, norm.to)
        info.title = norm.to
      })
      if (!--waiting) next()
    })
  }(api))

  function next() {
    console.log('Getting revisions...')
    var userHashes = {}
    pull(
      pull.values(pagesInfo),
      pull.asyncMap(function (pageInfo, cb) {
        // Calculate blob id for page URL + title, for linking
        pull(
          pull.once(pageInfo.site + '\t' + pageInfo.title),
          sbot.blobs.add(function (err, hash) {
            pageInfo.hash = hash
            cb(null, pageInfo)
          })
        )
      }),
      pull.asyncMap(function (pageInfo, cb) {
        // Get previous messages for this page.
        // Simple solution: find the revision with latest timestamp.
        var maxRevTs = ''
        var maxRevMsgId
        pull(
          sbot.links({
            dest: pageInfo.hash,
            rel: 'pageId',
            values: true,
            meta: false
          }),
          pull.filter(function (msg) {
            var c = msg && msg.value && msg.value.content
            return c
              && c.type === 'wikimedia/revisions'
              && c.site === pageInfo.site
              && c.title === pageInfo.title
          }),
          pull.drain(function (msg) {
            var c = msg && msg.value && msg.value.content
            var revs = Array.isArray(c.revisions) && c.revisions
            if (revs) revs.forEach(function (rev) {
              if (rev && rev.timestamp > maxRevTs) {
                maxRevTs = rev.timestamp
                maxRevMsgId == msg.key
              }
            })
          }, function (err) {
            if (err) return cb(err)
            pageInfo.latestMsgId = maxRevMsgId
            pageInfo.latestRevTs = maxRevTs
            cb(null, pageInfo)
          })
        )
      }),
      pull.map(function (pageInfo) {
        // Get new revisions.
        var rvcontinue, rvdone
        var rvstart = pageInfo.latestRevTs
        var prevId = pageInfo.latestMsgId
        var aborted
        var revisions = pull(
          function (abort, cb) {
            if (aborted = abort) return cb(abort)
            if (rvdone) return cb(true)

            console.log('Getting revisions for', pageInfo.title + '...',
              rvstart || '', rvcontinue || '')
            var url = api + '?format=json&action=query&prop=revisions&rvslots=*'
              + '&titles=' + encodeURIComponent(pageInfo.title)
              + '&rvprop=ids|timestamp|comment|user|sha1|size|slotsha1|slotsize|content|roles|flags|tags'
              + '&rvdir=newer'
              + (rvcontinue ? '&rvcontinue=' + rvcontinue : '')
              + (rvstart ? '&rvstart=' + rvstart : '')
              + '&rvlimit=50'
            getJson(url, function (err, data) {
              if (aborted) return err && console.trace(err)
              if (err) return cb(err)

              var warnings = data.warnings
              if (warnings) {
                if (warnings.main) {
                  if (warnings.main['*'] === 'Unrecognized parameter: rvslots.') {
                    delete warnings.main['*']
                    if (Object.keys(warnings.main).length === 0) {
                      delete warnings.main
                    }
                  }
                }
                if (warnings.revisions) {
                  if (warnings.revisions['*'] === 'Unrecognized values for parameter "rvprop": slotsha1, slotsize, roles.') {
                    delete warnings.revisions['*']
                    if (Object.keys(warnings.revisions).length === 0) {
                      delete warnings.revisions
                    }
                  }
                }
                if (Object.keys(warnings).length > 0) {
                  console.trace('Warnings:', warnings)
                }
              }

              rvcontinue = data.continue && data.continue.rvcontinue
              if (!rvcontinue) rvdone = true
              var page
              if (data.query) for (var pageid in data.query.pages) {
                page = data.query.pages[pageid]
                if (page.title === pageInfo.title) break
                else page = null
              }
              if (!page) {
                console.trace(data.query.pages, pageInfo)
                return cb(new Error('Unable to find page'))
              }
              var revs = page.revisions || []
              console.log('Got ' + revs.length + ' revisions')
              cb(null, revs)
            })
          },
          pull.flatten(),

          pull.through(function (rev) {
            if (!rev.slots) {
               // old API does not use slots.
               // Transform result to be forward-compatible.
              rev.slots = {
                main: {
                  size: rev.size,
                  sha1: rev.sha1,
                  contentmodel: rev.contentmodel,
                  contentformat: rev.contentformat,
                  '*': rev['*']
                }
              }
              delete rev.contentmodel
              delete rev.contentformat
              delete rev['*']
            }
            // duplicate values supplied in new API in slotsize and slotsha1
            delete rev.sha1
            delete rev.size
          }),

          pull.asyncMap(function (rev, cb) {
            // Calculate blob id for user page URL + title, for linking
            var hash = userHashes[rev.user]
            if (hash) {
              rev.userId = hash
              return cb(null, rev)
            }
            pull(
              pull.once(pageInfo.site + '\tUser:' + rev.user),
              sbot.blobs.add(function (err, hash) {
                rev.userId = userHashes[rev.user] = hash
                cb(null, rev)
              })
            )
          }),

          pull.asyncMap(function (rev, cb) {
            var waiting = 0
            for (var slot in rev.slots) (function (slot) {
              waiting++
              var slotInfo = rev.slots[slot]
              var content = slotInfo['*']
              if (!content) {
                console.trace(slotInfo)
                return cb(new Error('Missing content'))
              }
              var sha1 = crypto.createHash('sha1').update(content).digest('hex')
              if (sha1 !== slotInfo.sha1) {
                console.trace(slotInfo, sha1)
                return cb(new Error('Mismatched content sha1'))
              }
              pull(
                pull.once(content),
                sbot.blobs.add(function (err, hash) {
                  if (err) return cb(err)
                  slotInfo.link = hash
                  delete slotInfo['*']
                  if (!--waiting) cb(null, rev)
                })
              )
            }(slot))
          })
        )

        var queuedRevisions = []
        var ended
        function cbDraft(content, cb) {
          if (!content.revisions.length) {
            console.log('No revisions for', pageInfo.title)
            return cb(true)
          }
          console.log('Prepared a message',
            'with', content.revisions.length, 'revisions',
            'for', pageInfo.title)
          prevId = '%' + crypto.createHash('sha256').update(JSON.stringify(content)).digest('base64') + '.draft6'
          cb(null, {
            draftId: prevId,
            content: content
          })
        }

        return function (abort, cb) {
          if (abort) return revisions(abort, cb)
          if (ended) return cb(true)
          var content = {
            type: 'wikimedia/revisions',
            site: pageInfo.site,
            title: pageInfo.title,
            pageId: pageInfo.hash,
            parents: prevId ? [prevId] : undefined,
            revisions: queuedRevisions.splice(0)
          }
          revisions(null, function next(end, revision) {
            if (ended = end) return cbDraft(content, cb)
            content.revisions.push(revision)
            if (estimateMessageSize(content) > 8192) {
              queuedRevisions.push(content.revisions.pop())
              // console.log('filled msg for ', pageInfo.title, ' with ', content.revisions.length, 'revisions')
              return cbDraft(content, cb)
            }
            revisions(null, next)
          })
        }
      }),
      pull.flatten(),

      pull.collect(function (err, drafts) {
        if (err) throw err
        if (dry) {
          console.log(JSON.stringify(drafts, 0, 2))
          return sbot.close()
        }
        if (!drafts.length) {
          console.log('No messages to publish.')
          return sbot.close()
        }
        if (yes) return confirmed(true)
        var rl = readline.createInterface({
          input: process.stdin,
          output: process.stdout
        })
        rl.question('Publish ' + drafts.length + ' messages? [Y/n] ', function (answer) {
          rl.close()
          confirmed(!/^n/i.test(answer))
        })
        function confirmed(yes) {
          if (!yes) return sbot.close()
          publishDrafts(sbot, drafts, function (err, msgs) {
            if (err) throw err
            console.log('Published:\n' + msgs.map(function (msg) {
              return msg.key
            }.join('\n')))
            sbot.close()
          })
        }
      })
    )
  }
})