# Copyright (c) 2010 Flowerfire, Inc.  All Rights Reserved.
ice_cast = {

  plugin_version = "1.5"

  info.1.manufacturer = "Xiph Foundation"
  info.1.device = "Icecast"
  info.1.version = "2"

  #            - 1.0 -     - Initial creation.
  # 2008-02-20 - 1.1 - KBB - Made autodetect_regular_expression more restrictive, since this
  # format is so consistent. Fixed bug where user agent field was not being set because of an
  # extra field being gathered. (My original assumption was that there was a format variant that
  # lacked the server_domain field, but we have many samples of icecast logs and they are all the
  # same, so I think it was a bug. If this breaks backward compatibility to an earlier default,
  # or a common variant, then that must be addressed. I don't know if icecast logging uses format
  # strings like Apache.)
  # 2008-02-20 - 1.2 - KBB - Based on information from icecast.org forum,
  # http://icecast.imux.net/viewtopic.php?t=655&highlight=log+format, am now gathering final
  # numeric field as the duration in seconds. I have used this to replace the previous duration
  # which as calculated from the size and an assumption about the speed.
  # 2008-02-20 - 1.3 - GMF - Moved "accesses=1" to its own filter, so it can be automatically disabled
  #                          when accesses does not exist.
  # 2011-09-28 - 1.3.1 - KBB - 18446744073709551615 shows up as the duration in logs. This is the largest
  # unsigned 64 bit integer, and it is used in SQL to represent NULL, so it shows up as something
  # enormous in v7 and zero in v8.5. We now set the duration to zero explicitly in the parsing filter.
  # 2011-11-17 - 1.3.2 - KBB - 18446744073709551614 also shows up as the duration in logs, so generalizing
  # the previous fix to set anything that is less than zero seconds or over five years to zero.
  # 2013-05-31 - 1.3.3 - MSG - Increased the suppress bottom value of page field to 9.
  # 2014-09-16 - 1.4 - gas - added media reports
  # 2014-10-10 - 1.5 - gas - removed location field, no longer needed - media_reports snapon does geo
  # 2015-01-14 - 1.6 - ILD - Added User Agent snapon and removed old UA calculation method

  # The name of the log format
  log.format.format_label = "IceCast Log Format"
  log.miscellaneous.log_data_type = "http_access"  
  log.miscellaneous.log_format_type = "media_server"

  # The log is in this format if any of the first ten lines match this regular expression
  #211.11.11.111 - - [09/Jan/2008:16:55:34 -0500] "GET /sample.mp3 HTTP/1.1" 200 11360 "-" "NSPlayer/11.0.5721.5145 WMFSDK/11.0" 1
  #log.format.autodetect_regular_expression = "^[^ ]* [^ ]* .* \\[../.../....:..:..:...*\\] \"[A-Z]* [^ ]* HTTP[^\"]*\" [0-9]* [-0-9]* [^ ]* \".*\" [0-9][0-9]*$"
  log.format.autodetect_regular_expression = '^[^ ]+ [^ ]+ .* \\[../.../....:..:..:...*\\] "[A-Z]+ [^ ]+ HTTP[^"]*" [0-9]+ [-0-9]+ ("[^"]+"|[^ ]+) "[^"]+" [0-9]+$'

  # Treat fields surrounded by square brackets (e.g. the date/time field) as a single quoted field.
  log.format.treat_brackets_as_quotes = "true"
  log.format.common_log_format = "true"

  # The format of dates and times in this log
  log.format.date_format = "dd/mmm/yyyy:hh:mm:ss"
  log.format.time_format = "dd/mmm/yyyy:hh:mm:ss"

  # Log fields
  log.fields = {

    hostname = {
      type = "host"
      index = 1
    } # hostname

    authenticated_user = {
      index = 3
    } # authenticated_user

    date_time = {
      index = 4
      subindex = 1
    } # date_time

    operation = {
      index = 5
      subindex = 1
    } # operation

    page = {
      type = "page"
      index = 5
      subindex = 2
      hierarchy_dividers = "/?"
      left_to_right = true
      leading_divider = "true"
    } # page

    protocol = {
      index = 5
      subindex = 3
    } # protocol

    server_response = {
      type = "response"
      index = 6
    } # server_response

    size = {
      index = 7
    } # size

#    server_domain = {
#      index = 8
#    } # server_domain

    referrer = {
      type = "URL"
#      index = 9
      index = 8
      hierarchy_dividers = "/?"
      left_to_right = true
      leading_divider = "false"
    } # referrer

    agent = {
#     type = "agent"
#      index = 10
      index = 9
    } # agent

    duration = {
      index = 10
    }

    more_than_15_minutes = ""

    #total_duration_96kbps = ""
    fifteen_minute_sessions = ""

  } # log.fields

  # Database fields
  database.fields = {
    date_time = ""
    day_of_week = ""
    hour_of_day = ""
    page = {
      label = "$lang_stats.field_labels.page"
      log_field = "page"
      type = "string"
      suppress_top = 0
      suppress_bottom = 9
      display_format_type = "page"
    } # page
    file_type = ""
    worm = ""
    screen_dimensions = ""
    screen_depth = ""
    hostname = ""
    domain_description = ""
    #location = ""

    referrer = {
      suppress_top = 1
      suppress_bottom = 3
    } # referrer

    referrer_description = ""
    search_engine = ""
    search_phrase = ""
#    web_browser = ""
#    operating_system = ""
#    spider = ""
    authenticated_user = ""
    server_response = ""

    more_than_15_minutes = ""

  } # database.fields

  log.parsing_filters.parse = `

# Get search engine and search phrase information from the referrer field (before it gets simplified).
if (get_search_engine_info(referrer)) then (
  search_engine = volatile.search_engine;
  search_phrase = volatile.search_phrase;
);

# Get web browser, operating system, web browser, and spider information from the user-agent field.
#get_user_agent_info(agent);
#web_browser = volatile.web_browser;
#operating_system = volatile.operating_system;
#spider = volatile.spider;

#166.166.166.166 - - [06/Jun/2011:11:00:02 -0400] "GET /lizard-radio HTTP/1.1" 200 71112 "-" "LIZRadio/3.3.3 LIZNetwork/5.5.5 Lizard/11.11.11" 18446744073709551614
#if (duration eq '18446744073709551615') then (
if ((duration > 5*365*24*60*60) or (duration < 0)) then (
  duration = 0;
);

# Compute the duration in seconds fo the stream, based on the assumption that
# it was delivered at a continuous 96kbps.
#total_duration_96kbps = (size * 8.0) / (96.0*1024.0);

# Determine if this is more than 15 minutes, based on 96kbps
#more_than_15_minutes = total_duration_96kbps > (15*60);
more_than_15_minutes = duration > (15*60);
if (more_than_15_minutes) then
  fifteen_minute_sessions = 1;
`

  # Log Filters
  log.filters = {

    simplify_referrer = {
      label = "$lang_admin.log_filters.simplify_referrer_label"
      comment = "$lang_admin.log_filters.simplify_referrer_comment"
      value = "if (referrer eq '-') then referrer = '(no referrer)' else if (matches_regular_expression(referrer, '^([^:]+://[^/]+/)')) then referrer = $1 . '(omitted)'"
    } # simplify_referrer

    internal_referrer = {
      label = "$lang_admin.log_filters.internal_referrer_label"
      comment = "$lang_admin.log_filters.internal_referrer_comment"
      value = "if (contains(referrer, 'mydomain.com/')) then referrer = '(internal referrer)';"
      disabled = true
    } # internal_referrer

    not_authenticated = {
      label = "$lang_admin.log_filters.not_authenticated_label"
      comment = "$lang_admin.log_filters.not_authenticated_comment"
      value = "if (authenticated_user eq '-') then authenticated_user = '(not authenticated)';"
    } # not_authenticated

    set_page_for_worm = {
      label = "$lang_admin.log_filters.set_page_for_worm_label"
      comment = "$lang_admin.log_filters.set_page_for_worm_comment"
      value = "if (starts_with(worm, '(')) then '' else page = '(worm)';"
    } # set_page_for_worm

    remove_query = {
      label = "$lang_admin.log_filters.remove_query_label"
      comment = "$lang_admin.log_filters.remove_query_comment"
      value = "if (contains(page, '?')) then page = substr(page, 0, index(page, '?') + 1) . '(parameters)';"
    } # remove_query

    detect_page_views = {
      label = '$lang_admin.log_filters.detect_page_views_label'
      comment = '$lang_admin.log_filters.detect_page_views_comment'
      value = "if ((file_type eq 'JPEG') or (file_type eq 'JPG') or (file_type eq 'GIF') or (file_type eq 'ICO') or (file_type eq 'PNG') or (file_type eq 'CSS') or (file_type eq 'SWF') or (file_type eq 'JS')) then page_views = 0; else page_views = 1;"
    } # detect_page_views

    strip_non_page_views = {
      label = '$lang_admin.log_filters.strip_non_page_views_label'
      comment = '$lang_admin.log_filters.strip_non_page_views_comment'
      value = "if (page_views == 0) then page = substr(page, 0, last_index(page, '/') + 1) . '(nonpage)';"
    } # strip_non_page_views

    mark_entry = {
      label = '$lang_admin.log_filters.mark_entry_label'
      comment = '$lang_admin.log_filters.mark_entry_comment'
      value = 'accesses = 1;'
    } # mark_entry

  } # log.filters

  log.field_options = {

    sessions_page_field = "page"
    sessions_visitor_id_field = "hostname"
    sessions_event_field = "page_views"

  } # log.field_options

  database.numerical_fields = {

    accesses = {
      requires_log_field = false
      entries_field = true
    } # accesses

    page_views = {
      default = true
      requires_log_field = false
    } # page_views

    visitors = {
      log_field = "hostname"
      type = "unique"
    } # visitors

    size = {
      type = "int"
      integer_bits = 64
      display_format_type = "bandwidth"
    } # size

#    total_duration_96kbps = {
#      type = "int"
#      integer_bits = 64
#      display_format_type = "duration_compact"
#    }

#    average_duration_96kbps = {
#      type = "int"
#      integer_bits = 64
#      display_format_type = "duration_compact"
#      log_field = "total_duration_96kbps"
#      aggregation_method = "average"
#      average_denominator_field = "accesses"
#    }

    duration = {
      type = "int"
      default = true
      integer_bits = 64
      display_format_type = "duration_compact"
    } # duration

    average_duration = {
      label = "$lang_stats.field_labels.average $lang_stats.field_labels.duration"
      type = "int"
      integer_bits = 64
      display_format_type = "duration_compact"
      log_field = "duration"
      aggregation_method = "average"
      average_denominator_field = "accesses"
    } # average_duration

    fifteen_minute_sessions = ""

  } # database.numerical_fields

  create_profile_wizard_options = {

    # How the reports should be grouped in the report menu
    report_groups = {
      date_time_group = ""
    } # report_groups
		snapons = {
      # Attach a media_reports snapon
      media_reports = {
        snapon = "media_reports"
        name = "media_reports"
        label = "$lang_admin.snapons.media_reports.label"
        parameters = {
          user_field.parameter_value = "authenticated_user"
          client_ip_field.parameter_value = "hostname"
          duration_field.parameter_value = "duration"
          have_stream_name_field.parameter_value = true
          stream_name_field.parameter_value = "page"

          have_stream_id_field.parameter_value = false
#          stream_id_field.parameter_value = "c_playerid"

          # Do the default concurrent connections analysis; it's based on a single duration field, which we have.
          attach_concurrent_events.parameter_value = true
  
        } # parameters

      } # media_reports
    	user_agent_analysis = {
        snapon = "user_agent_analysis"
        name = "user_agent_analysis"
        label = "$lang_admin.snapons.user_agent_analysis.label"
        parameters = {
          user_agent_field.parameter_value = "agent"
        } # parameters
      } # user_agent_analysis
     } # snapons
  } # create_profile_wizard_options

} # ice_cast