Class: Webhookdb::Replicator::IcalendarCalendarV1

Inherits:
Base
  • Object
show all
Includes:
Appydays::Loggable
Defined in:
lib/webhookdb/replicator/icalendar_calendar_v1.rb

Defined Under Namespace

Classes: EventProcessor, Upserter

Constant Summary collapse

RECURRENCE_PROJECTION = 
5.years

Constants inherited from Base

Base::MAX_INDEX_NAME_LENGTH

Constants included from DBAdapter::ColumnTypes

DBAdapter::ColumnTypes::BIGINT, DBAdapter::ColumnTypes::BIGINT_ARRAY, DBAdapter::ColumnTypes::BOOLEAN, DBAdapter::ColumnTypes::COLUMN_TYPES, DBAdapter::ColumnTypes::DATE, DBAdapter::ColumnTypes::DECIMAL, DBAdapter::ColumnTypes::DOUBLE, DBAdapter::ColumnTypes::FLOAT, DBAdapter::ColumnTypes::INTEGER, DBAdapter::ColumnTypes::INTEGER_ARRAY, DBAdapter::ColumnTypes::OBJECT, DBAdapter::ColumnTypes::TEXT, DBAdapter::ColumnTypes::TEXT_ARRAY, DBAdapter::ColumnTypes::TIMESTAMP, DBAdapter::ColumnTypes::UUID

Instance Attribute Summary

Attributes inherited from Base

#service_integration

Class Method Summary collapse

Instance Method Summary collapse

Methods inherited from Base

#_any_subscriptions_to_notify?, #_backfill_state_change_fields, #_backfillers, #_clear_backfill_information, #_clear_webook_information, #_coalesce_excluded_on_update, #_enqueue_backfill_jobs, #_extra_index_specs, #_fetch_enrichment, #_find_dependency_candidate, #_notify_dependents, #_parallel_backfill, #_prepare_for_insert, #_publish_rowupsert, #_store_enrichment_body?, #_to_json, #_upsert_conflict_target, #_upsert_webhook, #_upsert_webhook_single_resource, #_verify_backfill_err_msg, #_webhook_state_change_fields, #admin_dataset, #avoid_writes?, #backfill, #backfill_not_supported_message, #calculate_and_backfill_state_machine, #calculate_backfill_state_machine, #calculate_dependency_state_machine_step, #calculate_preferred_create_state_machine, chunked_row_update_bounds, #clear_backfill_information, #clear_webhook_information, #create_table, #create_table_modification, #create_table_partitions, #data_column, #dbadapter_table, #denormalized_columns, #descriptor, #dispatch_request_to, #enqueue_sync_targets, #enrichment_column, #ensure_all_columns, #ensure_all_columns_modification, #existing_partitions, #find_dependent, #find_dependent!, #indices, #initialize, #on_backfill_error, #on_dependency_webhook_upsert, #partition?, #partitioning, #preferred_create_state_machine_method, #preprocess_headers_for_logging, #primary_key_column, #process_state_change, #process_webhooks_synchronously?, #qualified_table_sequel_identifier, #readonly_dataset, #remote_key_column, #requires_sequence?, #resource_name_plural, #resource_name_singular, #schema_and_table_symbols, #storable_columns, #synchronous_processing_response_body, #timestamp_column, #upsert_webhook_body, #verify_backfill_credentials, #webhook_endpoint, #webhook_response, #with_advisory_lock

Constructor Details

This class inherits a constructor from Webhookdb::Replicator::Base

Class Method Details

.descriptor ⇒ Webhookdb::Replicator::Descriptor

Returns:



18
19
20
21
22
23
24
25
26
27
28
 
# File 'lib/webhookdb/replicator/icalendar_calendar_v1.rb', line 18

def self.descriptor
  return Webhookdb::Replicator::Descriptor.new(
    name: "icalendar_calendar_v1",
    ctor: ->(sint) { Webhookdb::Replicator::IcalendarCalendarV1.new(sint) },
    feature_roles: [],
    resource_name_singular: "iCalendar Calendar",
    supports_webhooks: true,
    description: "Fetch and convert an icalendar format file into a schematized and queryable database table.",
    api_docs_url: "https://icalendar.org/",
  )
end

Instance Method Details

#_clean_ics_url(url) ⇒ Object

We get all sorts of strange urls, fix up what we can.



272
273
274
275
276
277
 
# File 'lib/webhookdb/replicator/icalendar_calendar_v1.rb', line 272

def _clean_ics_url(url)
  u = URI(url)
  # https://xyz.com:80 is invalid, set it to 443 which yields https://xyz.com
  u.port = 443 if u.scheme == "https" && u.port == 80
  return u.to_s
end

#_denormalized_columns ⇒ Object 



70
71
72
73
74
75
76
77
78
79
80
81
82
 
# File 'lib/webhookdb/replicator/icalendar_calendar_v1.rb', line 70

def _denormalized_columns
  col = Webhookdb::Replicator::Column
  return [
    col.new(:row_created_at, TIMESTAMP, index: true, optional: true, defaulter: :now),
    col.new(:row_updated_at, TIMESTAMP, index: true, optional: true, defaulter: :now),
    col.new(:last_synced_at, TIMESTAMP, index: true, optional: true),
    col.new(:ics_url, TEXT, converter: col.converter_gsub("^webcal", "https")),
    col.new(:event_count, INTEGER, optional: true),
    col.new(:feed_bytes, INTEGER, optional: true),
    col.new(:last_sync_duration_ms, INTEGER, optional: true),
    col.new(:last_fetch_context, OBJECT, optional: true),
  ]
end

#_handle_down_error(e, request_url:, calendar_external_id:) ⇒ Object 



279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
 
# File 'lib/webhookdb/replicator/icalendar_calendar_v1.rb', line 279

def _handle_down_error(e, request_url:, calendar_external_id:)
  case e
    when Down::TooManyRedirects
      response_status = 301
      response_body = "<too many redirects>"
    when Down::NotModified
      # Do not alert on 304, but do log
      self.logger.info("icalendar_fetch_not_modified", response_status: 304, request_url:, calendar_external_id:)
      return
    when Down::SSLError
      # Most SSL errors are transient and can be retried, but some are due to a long-term misconfiguration.
      # Handle these with an alert, like if we had a 404, which indicates a longer-term issue.
      is_fatal =
        # There doesn't appear to be a way to allow unsafe legacy content negotiation on a per-request basis,
        # it is compiled into OpenSSL (may be wrong about this).
        e.to_s.include?("unsafe legacy renegotiation disabled") ||
        # Certificate failures are not transient
        e.to_s.include?("certificate verify failed")
      if is_fatal
        response_status = 0
        response_body = e.to_s
      else
        self._handle_retryable_down_error!(e, request_url:, calendar_external_id:)
      end
    when Down::TimeoutError, Down::ConnectionError, Down::InvalidUrl,
      Errno::ECONNRESET,
      URI::InvalidURIError,
      HTTPX::NativeResolveError, HTTPX::InsecureRedirectError,
      HTTPX::Connection::HTTP2::Error,
      EOFError
      response_status = 0
      response_body = e.to_s
    when Down::ClientError
      raise e if e.response.nil?
      response_status = e.response.status.to_i
      self._handle_retryable_down_error!(e, request_url:, calendar_external_id:) if
        self._retryable_client_error?(e, request_url:)
      # These are all the errors we've seen, we can't do anything about.
      # In theory we should do this for ALL 4xx errors,
      # but we'd rather error on the WebhookDB side until we're sure
      # we want to ignore things.
      expected_errors = [
        400, 401, 402, 403, # Common access problems we can't do anything about
        404, 405, # Fundamental issues with the URL given
        409, 410, # More access problems
        417, # If someone uses an Outlook HTML calendar, fetch gives us a 417
        422, # Sometimes used instead of 404
        429, # Usually 429s are retried (as above), but in some cases they're not.
        500, 503, 504, # Intermittent server issues, usually
        599, # Represents a timeout in icalproxy
      ]
      # For most client errors, we can't do anything about it. For example,
      # and 'unshared' URL could result in a 401, 403, 404, or even a 405.
      # For now, other client errors, we can raise on,
      # in case it's something we can fix/work around.
      # For example, it's possible something like a 415 is a WebhookDB issue.
      if response_status == 421 && (origin_err = e.response.headers["Ical-Proxy-Origin-Error"])
        response_status = origin_err.to_i
      end
      raise e unless expected_errors.include?(response_status)
      response_body = self._safe_read_body(e)
    when Down::ServerError
      response_status = e.response.status.to_i
      response_body = self._safe_read_body(e)
    else
      response_body = nil
      response_status = nil
  end
  raise e if response_status.nil?
  loggable_body = response_body && response_body[..256]
  self.logger.warn("icalendar_fetch_error",
                   response_body: loggable_body, response_status:, request_url:, calendar_external_id:,)
  message = Webhookdb::Messages::ErrorIcalendarFetch.new(
    self.service_integration,
    calendar_external_id,
    response_status:,
    response_body:,
    request_url:,
    request_method: "GET",
  )
  self.service_integration.organization.alerting.dispatch_alert(message, separate_connection: false)
end

#_handle_retryable_down_error!(e, request_url:, calendar_external_id:) ⇒ Object

Raises:

  • (Amigo::Retry::OrDie) 


381
382
383
384
385
386
387
388
389
390
391
392
393
394
 
# File 'lib/webhookdb/replicator/icalendar_calendar_v1.rb', line 381

def _handle_retryable_down_error!(e, request_url:, calendar_external_id:)
  # Retry on these, which are hopefully transient.
  # For now, if they aren't transient, die so we see the job.
  # We will probably need to do an alert if the retries on exhausted instead.
  retry_in = rand(4..60).minutes
  self.logger.debug(
    "icalendar_fetch_error_retry",
    response_status: e.respond_to?(:response) ? e.response&.status : 0,
    request_url:,
    calendar_external_id:,
    retry_at: Time.now + retry_in,
  )
  raise Amigo::Retry::OrDie.new(10, retry_in)
end

#_make_ics_request(request_url, last_fetch_context) ⇒ Object 



253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
 
# File 'lib/webhookdb/replicator/icalendar_calendar_v1.rb', line 253

def _make_ics_request(request_url, last_fetch_context)
  # Some servers require a VERY explicit accept header,
  # so tell them we prefer icalendar here.
  # Using Httpx, Accept-Encoding is gzip,deflate
  # which seems fine (server should use identity as worst case).
  headers = {
    "Accept" => "text/calendar,*/*",
  }
  headers["If-None-Match"] = last_fetch_context["etag"] if last_fetch_context & ["etag"]
  if (proxy_url = Webhookdb::Icalendar.proxy_url).present?
    request_url = "#{proxy_url.delete_suffix('/')}/?url=#{URI.encode_www_form_component(request_url)}"
    headers["Authorization"] = "Apikey #{Webhookdb::Icalendar.proxy_api_key}" if
      Webhookdb::Icalendar.proxy_api_key.present?
  end
  resp = Webhookdb::Http.chunked_download(request_url, rewindable: false, headers:)
  return resp
end

#_remote_key_column ⇒ Object 



66
67
68
 
# File 'lib/webhookdb/replicator/icalendar_calendar_v1.rb', line 66

def _remote_key_column
  return Webhookdb::Replicator::Column.new(:external_id, TEXT)
end

#_resource_and_event(request) ⇒ Object 



88
89
90
 
# File 'lib/webhookdb/replicator/icalendar_calendar_v1.rb', line 88

def _resource_and_event(request)
  return request.body, nil
end

#_resource_to_data(resource) ⇒ Object 



103
104
105
106
107
108
 
# File 'lib/webhookdb/replicator/icalendar_calendar_v1.rb', line 103

def _resource_to_data(resource, *)
  data = resource.dup
  # Remove the client-provided webhook fields.
  data.clear
  return data
end

#_retryable_client_error?(e, request_url:) ⇒ Boolean

Returns:

  • (Boolean) 


370
371
372
373
374
375
376
377
378
379
 
# File 'lib/webhookdb/replicator/icalendar_calendar_v1.rb', line 370

def _retryable_client_error?(e, request_url:)
  code = e.response.status.to_i
  # This is a bad domain that returns 429 for most requests.
  # Tell the org admins it won't sync.
  return false if code == 429 && request_url.start_with?("https://ical.schedulestar.com")
  # Other 429s can be retried.
  return true if code == 429
  # Otherwise, handle the client error normally, by telling the org admins, or raising.
  return false
end

#_safe_read_body(e) ⇒ Object

We can hit an error while reading the error body, since it was opened as a stream. Ignore those errors.



364
365
366
367
368
 
# File 'lib/webhookdb/replicator/icalendar_calendar_v1.rb', line 364

def _safe_read_body(e)
  return e.response.body.to_s
rescue OpenSSL::SSL::SSLError, HTTPX::Error
  return "<error reading body>"
end

#_sync_row(row, dep, now:) ⇒ Object 



216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
 
# File 'lib/webhookdb/replicator/icalendar_calendar_v1.rb', line 216

def _sync_row(row, dep, now:)
  calendar_external_id = row.fetch(:external_id)
  begin
    request_url = self._clean_ics_url(row.fetch(:ics_url))
    io = self._make_ics_request(request_url, row.fetch(:last_fetch_context))
  rescue Down::Error,
         URI::InvalidURIError,
         HTTPX::NativeResolveError,
         HTTPX::InsecureRedirectError,
         HTTPX::Connection::HTTP2::Error,
         EOFError => e
    self._handle_down_error(e, request_url:, calendar_external_id:)
    return
  end

  upserter = Upserter.new(dep.replicator, calendar_external_id, now:)
  processor = EventProcessor.new(io:, upserter:, headers: io.data[:headers])
  processor.process
  # Delete all the extra replicator rows, and cancel all the rows that weren't upserted.
  dep.replicator.admin_dataset do |ds|
    ds = ds.where(calendar_external_id:)
    if (delete_condition = processor.delete_condition)
      ds.where(delete_condition).delete
    end
    # Update both the status, and set the data json to match.
    # Only update rows not already CANCELLED.
    ds = ds.exclude(Sequel[compound_identity: processor.upserted_identities])
    ds = ds.where(Sequel[status: nil] | ~Sequel[status: "CANCELLED"])
    ds.update(
      status: "CANCELLED",
      data: Sequel.lit('data || \'{"STATUS":{"v":"CANCELLED"}}\'::jsonb'),
      row_updated_at: now,
    )
  end
  return processor
end

#_timestamp_column_name ⇒ Object 



84
85
86
 
# File 'lib/webhookdb/replicator/icalendar_calendar_v1.rb', line 84

def _timestamp_column_name
  return :row_updated_at
end

#_update_where_expr ⇒ Object 



92
93
94
 
# File 'lib/webhookdb/replicator/icalendar_calendar_v1.rb', line 92

def _update_where_expr
  return self.qualified_table_sequel_identifier[:row_updated_at] < Sequel[:excluded][:row_updated_at]
end

#_upsert_update_expr(inserting, **_kwargs) ⇒ Object 



96
97
98
99
100
101
 
# File 'lib/webhookdb/replicator/icalendar_calendar_v1.rb', line 96

def _upsert_update_expr(inserting, **_kwargs)
  update = super
  # Only set created_at if it's not set so the initial insert isn't modified.
  self._coalesce_excluded_on_update(update, [:row_created_at])
  return update
end

#_webhook_response(request) ⇒ Object 



32
33
34
 
# File 'lib/webhookdb/replicator/icalendar_calendar_v1.rb', line 32

def _webhook_response(request)
  return Webhookdb::WebhookResponse.for_standard_secret(request, self.service_integration.webhook_secret)
end

#calculate_webhook_state_machine ⇒ Object 



36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
 
# File 'lib/webhookdb/replicator/icalendar_calendar_v1.rb', line 36

def calculate_webhook_state_machine
  step = Webhookdb::Replicator::StateMachineStep.new
  if self.service_integration.webhook_secret.blank?
    self.service_integration.save_changes
    step.output = %(You are about to add support for replicating iCalendar (.ics) URLs into WebhookDB.

We have detailed instructions on this process
at https://docs.webhookdb.com/guides/icalendar/.

The first step is to generate a secret you will use for signing
API requests you send to WebhookDB. You can use '#{Webhookdb::Id.rand_enc(16)}'
or generate your own value.
Copy and paste or enter a new value, and press enter.)
    return step.secret_prompt("secret").webhook_secret(self.service_integration)
  end
  step.output = %(
All set! Here is the endpoint to send requests to
from your backend. Refer to https://docs.webhookdb.com/guides/icalendar/
for details on the format of the request:

#{self.webhook_endpoint}

The secret to use for signing is:

#{self.service_integration.webhook_secret}

#{self._query_help_output})
  return step.completed
end

#delete_data_for_external_id(external_id) ⇒ Object 



136
137
138
139
140
141
142
143
144
145
146
147
 
# File 'lib/webhookdb/replicator/icalendar_calendar_v1.rb', line 136

def delete_data_for_external_id(external_id)
  relevant_integrations = self.service_integration.recursive_dependents.
    filter { |d| Webhookdb::Icalendar::EVENT_REPLICATORS.include?(d.service_name) }
  self.admin_dataset do |ds|
    ds.db.transaction do
      ds.where(external_id:).delete
      relevant_integrations.each do |sint|
        ds.db[sint.replicator.qualified_table_sequel_identifier].where(calendar_external_id: external_id).delete
      end
    end
  end
end

#documentation_url ⇒ Object 



15
 
# File 'lib/webhookdb/replicator/icalendar_calendar_v1.rb', line 15

def documentation_url = Webhookdb::Icalendar::DOCUMENTATION_URL

#feed_changed?(row) ⇒ Boolean

Return true if the data in the feed has changed from what was last synced, or false if it has not so the sync can be skipped. This operation is meant to be resource-light (most of the work is the HTTP request), so should be done in a threadpool.

  • If we have no previous fetch context, we sync.

  • If the fetch errors, sync, because we want the normal error handler to figure it out (alert admins, etc).

  • If the last fetch’s content type and length is different from the current, we sync.

  • Download the bytes. If the hash of the bytes is different from what was last processed, sync. Since this involves reading the streaming body, we must return a copy of the body (a StringIO).

Returns:

  • (Boolean) 


688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
 
# File 'lib/webhookdb/replicator/icalendar_calendar_v1.rb', line 688

def feed_changed?(row)
  last_fetch = row.fetch(:last_fetch_context)
  return true if last_fetch.nil? || last_fetch.empty?

  begin
    url = self._clean_ics_url(row.fetch(:ics_url))
    resp = self._make_ics_request(url, last_fetch)
  rescue Down::NotModified
    return false
  rescue StandardError
    return true
  end
  headers = resp.data[:headers] || {}
  content_type_match = headers["Content-Type"] == last_fetch["content_type"] &&
    headers["Content-Length"] == last_fetch["content_length"]
  return true unless content_type_match
  last_hash = last_fetch["hash"]
  return true if last_hash.nil?

  hash = Digest::MD5.new
  while (line = resp.gets)
    hash.update(line)
  end
  return hash.hexdigest != last_hash
end

#rows_needing_sync(dataset, now: Time.now) ⇒ Object 



131
132
133
134
 
# File 'lib/webhookdb/replicator/icalendar_calendar_v1.rb', line 131

def rows_needing_sync(dataset, now: Time.now)
  cutoff = now - Webhookdb::Icalendar.sync_period_hours.hours
  return dataset.where(Sequel[last_synced_at: nil] | Sequel.expr { last_synced_at < cutoff })
end

#sync_row(row, force: false, now: Time.now) ⇒ Object 



168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
 
# File 'lib/webhookdb/replicator/icalendar_calendar_v1.rb', line 168

def sync_row(row, force: false, now: Time.now)
  Appydays::Loggable.with_log_tags(icalendar_url: row.fetch(:ics_url)) do
    last_synced_at = row.fetch(:last_synced_at)
    should_sync = force ||
      last_synced_at.nil? ||
      # If a proxy is configured, we always want to try to sync,
      # since this could have come from a webhook, but also the proxy feed refresh TTL
      # is likely much lower than ICALENDAR_SYNC_PERIOD_HOURS so it's good to check on it.
      # The check is very fast (should 304) so is safe to do relatively often.
      Webhookdb::Icalendar.proxy_url.present? ||
      last_synced_at < (now - Webhookdb::Icalendar.sync_period_hours.hours)
    unless should_sync
      self.logger.info("skip_sync_recently_synced", last_synced_at:)
      return
    end
    self.with_advisory_lock(row.fetch(:pk)) do
      start = Time.now
      if (dep = self.find_dependent(Webhookdb::Icalendar::EVENT_REPLICATORS))
        if dep.replicator.avoid_writes?
          # Check if this table is being vacuumed/etc. We use this instead of a semaphore job,
          # since it's a better fit for icalendar, which is pre-scheduled, rather than reactive.
          # That is, when we receive webhooks, a semaphore job gives us a more predictable rate;
          # but icalendar rate is negotiated in advance (when enqueing jobs),
          # and we can be more 'helpful' to something like a vacuum by not running any jobs at all.
          self.logger.info("skip_sync_table_locked")
          raise Amigo::Retry::Retry, 60.seconds + (rand * 10.seconds)
        end
        processor = self._sync_row(row, dep, now:)
      end
      self.admin_dataset do |ds|
        ds.where(pk: row.fetch(:pk)).
          update(
            last_synced_at: now,
            event_count: processor&.upserted_identities&.count,
            feed_bytes: processor&.read_bytes,
            last_sync_duration_ms: (Time.now - start).in_milliseconds,
            last_fetch_context: {
              "hash" => processor&.feed_hash,
              "content_type" => processor&.headers&.fetch("Content-Type", nil),
              "content_length" => processor&.headers&.fetch("Content-Length", nil),
              "etag" => processor&.headers&.fetch("Etag", nil),
            }.to_json,
          )
      end
    end
  end
end

#upsert_has_deps? ⇒ Boolean

Returns:

  • (Boolean) 


30
 
# File 'lib/webhookdb/replicator/icalendar_calendar_v1.rb', line 30

def upsert_has_deps? = true

#upsert_webhook(request) ⇒ Object 



110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
 
# File 'lib/webhookdb/replicator/icalendar_calendar_v1.rb', line 110

def upsert_webhook(request)
  request_type = request.body.fetch("type")
  external_id = request.body.fetch("external_id")
  case request_type
    when "SYNC"
      super
      Webhookdb::Jobs::IcalendarSync.perform_async(self.service_integration.id, external_id)
      return
    when "DELETE"
      self.delete_data_for_external_id(external_id)
      return
    when "__WHDB_UNIT_TEST"
      unless Webhookdb::RACK_ENV == "test"
        raise "someone tried to use the special unit test google event type outside of unit tests"
      end
      return super
    else
      raise ArgumentError, "Unknown request type: #{request_type}"
  end
end