# Copyright (c) 2014 Flowerfire, Inc.  All Rights Reserved.
akamai_web = {

  plugin_version = "1.1"

  info.1.manufacturer = "Akamai"
  info.1.device = "Web Server Log (W3C)"
  info.1.version = ""

  # 2014-04-07 - 1.0 - GMF - Created by cloning iisweb.cfg (v2.0.5)
  # 2014-04-25 - 1.1 - GMF - Added average_time_taken

  # The name of the log format
  log.format.format_label = "Akamai Web Server (W3C) Log Format"
  log.miscellaneous.log_data_type = "generic_w3c"  
  log.miscellaneous.log_format_type = "web_server"

  # The log is in this format if any of the first ten lines match this regular expression.
  # 2014-04-07 - GMF - Autodetecting from #Fields now (No #Software line).  Might be better to autodetect from the data; in particular, this has quoted fields while normal W3C Web doesn't quote fields
##Fields: date time cs-ip cs-method cs-uri sc-status sc-bytes time-taken cs(Referer) cs(User-Agent) cs(Cookie)"
#2009-08-24	11:08:10	66.249.71.182	GET	/www.defense.gov/	200	12226	0	"-"	"Mozilla/5.0 (compatible; Googlebot/2.1; +http://www.google.com/bot.html)"	"-"
  log.format.autodetect_regular_expression = "^#Fields: date time cs-ip "

  # The format of dates and times in this log
  log.format.ignore_format_lines = "true"

  # 2014-04-07 - GMF - Akamai W3C logs quote fields (unlike normal W3C) [ThreadID:1299722]
  log.format.ignore_quotes = "false"

  # 2013-02-15 - GMF - Transplanted this from generic_w3c
  # This handles #Fields lines, and creates log and database fields from them
  log.filter_preprocessor = `

if (matches_regular_expression(current_log_line(), '^#Fields: +([^ ].*)$')) then (
  string fields = $1;
  string fieldname;
  v.logfieldindex = 1;
  string numerical_fields = "profiles." . internal.profile_name . ".database.numerical_fields";

  # Send this line to all parsing servers
  distribute_format_line(current_log_line());

  # Reset all log fields to 0/0 so none carry over from the previous format line definition
  node log_fields = "profiles"{internal.profile_name}{"log"}{"fields"};
  node log_field;
  foreach log_field log_fields (
    @log_field{"index"} = 0;
    @log_field{"subindex"} = 0;
  );

  # This subroutine creates a database field
  subroutine(create_database_field(string fieldname, int top, int bottom), (
    #echo("create_database_field: " . fieldname); # debug
    debug_message("create_database_field(" . fieldname . ")\n");
    string databasefieldpath = "profiles." . internal.profile_name . ".database.fields." . fieldname;
    (databasefieldpath . "") = "";
    node databasefield = databasefieldpath;
    if (top ne 0) then
      set_subnode_value(databasefield, "suppress_top", top);
    if (bottom ne 0) then
      set_subnode_value(databasefield, "suppress_bottom", bottom);
    databasefield;
  ));

  subroutine(create_log_field(string fieldname, string type, bool withindex), (
    debug_message("create_log_field(" . fieldname . "; type=" . type . ")\n");
#    echo("create_log_field(" . fieldname . "; type=" . type . ")");
    string logfieldpath = "profiles." . internal.profile_name . ".log.fields." . fieldname;
    (logfieldpath . "") = "";
    node logfield = logfieldpath;
    if (withindex) then (
      set_subnode_value(logfield, "index", v.logfieldindex);
      v.logfieldindex++;
    );
    set_subnode_value(logfield, "subindex", 0);
    if (type ne '') then
      set_subnode_value(logfield, "type", type);
    logfield;
  ));

  # Get the log format separator. Assume at least two fields.
  # 2010-02-01 - GMF - Changed the default from " " to "" (any whitespace).  With " " as the default, data which
  #                    has a spaces in the header, but tabs in the data, doesn't parse, because this assumes space
  #                    separators.  Truly generic W3C header parsing appears to be impossible, which is why we can do
  #                    per-plugin parsing now; but that's just moved the problem here, where we'll still have to
  #                    fight it forever.  Hopefully, "" gives better coverage to the universe of W3C variants than " ".
#  log.format.field_separator = " ";
  log.format.field_separator = "";
  if (matches_regular_expression(fields, '^[^ 	]+	[^ 	]+')) then (
    log.format.field_separator = "	";
  );

  # Extract the fields on at a time
  while (matches_regular_expression(fields, '^([^ 	]+)[ 	](.*)$') or matches_regular_expression(fields, '^([^ 	]+)$')) (

    string unconverted_fieldname = $1;
    fields = $2;

    # Clean up the field name
    fieldname = '';
    for (int i = 0; i < length(unconverted_fieldname); i++) (
      string c = lowercase(substr(unconverted_fieldname, i, 1));
      if (!matches_regular_expression(c, '^[a-z0-9]$')) then
        c = '_';
      fieldname .= c;
    );
    while (matches_regular_expression(fieldname, '^(.*)_$'))
      fieldname = $1;

    # Get the log field type
    string log_field_type = '';
#    if (fieldname eq 'cs_referrer') then (
#      fieldname = 'cs_referer';
#    );
#    if (fieldname eq 'cs_uri') then (
#      fieldname = 'cs_uri_stem';
#    );
    # 2013-12-23 - GMF - Don't make cs_referrer type "referrer" anymore; we now do all referrer derivation in snapons, and this will create duplicate fields.

    # Create the log field
    # 2013-05-15 - GMF - We need to create these sc_bytes and cs_bytes log fields here, because otherwise web_server_package won't create the database fields.  Why were they excluded?  Maybe when this was copied from another plug-in which computes them from another field?
#    if ((fieldname ne "sc_bytes") and (fieldname ne "cs_bytes")) then
    create_log_field(fieldname, log_field_type, true);

    # If we're creating a profile, create the database fields too.
    if (node_exists("volatile.creating_profile")) then (

      if (false) then ( )

      # Don't track these fields in the database
      else if ((fieldname eq "sc_bytes") or (fieldname eq "cs_bytes") or (fieldname eq 'cs_uri_query') or (fieldname eq 'cs_cookie')) then
      ( );

      # Handle date by creating date_time and derived database fields
      else if (fieldname eq "date") then (
#        create_log_field('localtime', '', false); # placeholder - 7/Nov/2006 - KBB
        create_database_field('date_time', 0,  0);
        create_database_field('day_of_week', 0,  0);
        create_database_field('hour_of_day', 0,  0);
      ); # if date

      else if (fieldname eq "time") then (
        create_database_field('date_time', 0,  0);
        create_database_field('day_of_week', 0,  0);
        create_database_field('hour_of_day', 0,  0);
      ); # if time

      # Don't include agent in database
      else if (fieldname eq "cs_user_agent") then (
#        create_database_field('operating_system', 0,  0);
      );

      # Create database field cs_referer and derived fields for referrer
      else if (fieldname eq "cs_referer") then (
        create_database_field('cs_referer', 1,  9);
      );

      else if (fieldname eq "cs_uri") then (
        create_database_field('cs_uri', 0,  9);
      );

      # Don't add a database field for numerical fields
      else if (subnode_exists(numerical_fields, fieldname)) then (
        debug_message("Not adding numerical field: " . fieldname . "\n");
      );

      # Create a normal database field
      else
        create_database_field(fieldname, 0,  0);

    ); # if creating profile

  ); # while another field

  # Don't parse the #Fields line as a data line
  'reject';

); # if #Fields

# Don't parse any other # lines as data lines
else if (starts_with(current_log_line(), '#')) then (
  'reject';
);
`

#  log.fields = {
#    hits = ""
#  } # log.fields

  # Log Filters
  log.filters = {

#    empty_uri_query = {
#      label = "$lang_admin.log_filters.empty_uri_query_label"
#      comment = "$lang_admin.log_filters.empty_uri_query_comment"
#      value = "if (cs_uri_query eq '-') then cs_uri_query = '(empty)';"
#      disabled = false
#      requires_fields = {
#        cs_uri_query = true
#      }
#    } # empty_uri_query

#    add_cs_uri_query = {
#      label = "$lang_admin.log_filters.add_cs_uri_query_label"
#      comment = "$lang_admin.log_filters.add_cs_uri_query_comment"
#      value = "cs_uri_stem = cs_uri_stem . '?' . cs_uri_query"
#      disabled = false
#      requires_fields = {
#        cs_uri_query = true
#        cs_uri_stem = true
#      }
#    } # add_cs_uri_query

  } # log.filters

  database.numerical_fields = {

#    hits = ""

    time_taken = {
      type = "int"
      integer_bits = 64
      display_format_type = duration_milliseconds
    } # time_taken

    average_time_taken = {
      type = "int"
      integer_bits = 64
      display_format_type = duration_milliseconds
      aggregation_method = "average"
      average_denominator_field = "hits"
      log_field = "time_taken"
      requires_log_field = false
    } # average_time_taken

#    time_taken_avg = {
#      label = "$lang_stats.field_labels.average $lang_stats.field_labels.time_taken"
#      default = false
#      log_field = "time_taken"
#      requires_log_field = true
#      type = "int"
#      integer_bits = 64
#      aggregation_method = "average"
#      average_denominator_field = "hits"
#      display_format_type = duration_milliseconds
#    } # time_taken

  } # database.numerical_fields


  create_profile_wizard_options = {

    # How the reports should be grouped in the report menu
    report_groups = {
      date_time_group = ""
      content_group = {
        cs_uri = true
      }
      visitor_demographics_group = {
        cs_ip = true
      }
      referrer_group = {
        cs_referer = true
      }
      other_group = {
        cs_method = true
        sc_status = true
        cs_cookie = true
      }
    } # report_groups

    snapons = {

      # Attach a web_server_package snapon
      web_server_package = {
	snapon = "web_server_package"
	name = "web_server_package"
	label = "$lang_admin.snapons.web_server_package.label"
        parameters = {
          user_agent_field.parameter_value = "cs_user_agent"
          page_field.parameter_value = "cs_uri"
          client_ip_field.parameter_value = "cs_ip"
          server_response_field.parameter_value = "sc_status"
          referrer_field.parameter_value = "cs_referer"
          authenticated_user_field.parameter_value = "cs_username"
          hits_field = {
            parameter_value = "$lang_stats.field_labels.hits"
            final_node_name = "hits"
          } # hits
          visitors_field = {
            parameter_value = "$lang_stats.field_labels.unique_client_ips"
            final_node_name = "unique_client_ips"
          } # visitors_field
          sc_bytes_field = {
            parameter_value = "$lang_stats.field_labels.sc_bytes"
            final_node_name = "sc_bytes"
          } # sc_bytes_field
          cs_bytes_field = {
            parameter_value = "$lang_stats.field_labels.cs_bytes"
            final_node_name = "cs_bytes"
          } # cs_bytes_field
        } # parameters
      } # web_server_package

    } # snapons

#    final_step = `
#include "templates.admin.profiles.setup_reports_util";
#
#string profile = "profiles." . volatile.new_profile_name;
#
## Remove cs_uri_query and cs_cookie from the database fields
#delete_database_field(profile, 'cs_uri_query');
#delete_database_field(profile, 'cs_cookie');
#
## Create the standard reports
#add_standard_reports(profile);
#
## Delete referrer filters if there's no referrer field
##if (!node_exists(profile . ".database.fields.referrer")) then (
##  (profile . ".log.filters.no_referrer.disabled") = true;
##  (profile . ".log.filters.internal_referrer.disabled") = true;
##);
#
#    ` # end final_step

  } # create_profile_wizard_options

} # akamai_web