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

  plugin_version = "1.2"

  info.1.manufacturer = "Microsoft"
  info.1.device = "IIS (with syslog)"
  info.1.version = "6"

  # 2008-06-06 - 1.0 - GMF - Initial implementation (copied from iisweb.cfg, and modified for syslog stuff).
  # 2010-05-10 - 1.1 - Benson - Fixed if date format have a space character like "May  9 2010".
  # 2010-10-01 - 1.2 - MSG - Edited info lines


  # The name of the log format
  log.format.format_label = "IIS Extended (W3C) Web Server Log Format (logged through a syslog server)"
  log.miscellaneous.log_data_type = "syslog_required"
  log.miscellaneous.log_format_type = "web_server"

  # Autodetect based on the Software line, but require a character before it, so it won't match normal IIS data.
  log.format.autodetect_regular_expression = ".#Software: Microsoft Internet Information (Server|Services) [4-9]"

  # Look through a lot of lines to try to find a #Fields line
  log.format.autodetect_lines = "1000"

  # IIS uses does not use quotes to quote fields, so treat them as actual field values if we see them
  log.format.ignore_quotes = "true"

  log.format.parse_only_with_filters = "true"

  # Log fields
  log.fields.broken_link = ""

  # Database fields
  database.fields.broken_link = ""

  # Parse W3C header when we find it (during profile creation).
  log.filter_preprocessor = `

if (!node_exists('v.initialized')) then (

v.initialized = true;
include 'util.parse_w3c';

); # if not initialized

if (matches_regular_expression(current_log_line(), '.#Fields: (.*)$')) then (

  string fields = $1;
#echo("fields: " . fields);
  parse_w3c_fields(fields);

  # Trim trailing spaces
  if (matches_regular_expression(fields, '^(.*[^ ]+) *$')) then
    fields = $1;

  # Build a regular expression to detect these lines
  node field;
  string regexp = "^";
  string field_setting_code = "";
  int field_number = 1;
  v.fields_node = '';
  split(fields, ' ', 'v.fields_node');
#  echo('fields_node: ' . node_as_string('v.fields_node'));
#  foreach field ('profiles.' . volatile.new_profile_name . '.log.fields') (
  foreach field 'v.fields_node' (
    string field_name = replace_all(node_value(field), '-', '_');
#    echo("log field: " . field_name);
    if (regexp ne "^") then
      regexp .= " ";
    regexp .= "([^ ]+)";
    field_setting_code .= "  set_collected_field('', '" . field_name . "', $" . field_number . ");\n";
    field_number++;
#    echo(field);
  );
#  echo("regexp: " . regexp);

  # Build the parsing filter to extract this data
  string filter_code = "
if (matches_regular_expression(v.syslog_message, '" . regexp . "')) then (
" . field_setting_code . "
  accept_collected_entry('', false);
);
";

  # Install the filter in the new profile, if we're creating one, or the current profile, if we're building a database
#debug  echo("filter_code: " . filter_code);
  if (node_exists('volatile.new_profile_name')) then
    ('profiles.' . volatile.new_profile_name . '.log.parsing_filters.parse_data_line') = filter_code;
  else
    ('profiles.' . internal.profile_name . '.log.parsing_filters.parse_data_line') = filter_code;

  'reject';

); # if #Fields

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


  log.parsing_filters.remove_date_time = `
if (matches_regular_expression(v.syslog_message, '^[A-Z][a-z][a-z] [0-9| ][0-9] [0-9][0-9]:[0-9][0-9]:[0-9][0-9] [^ ]+ [^[]+[[][^]]+[]] (.*)$')) then 
  v.syslog_message = $1;
`

  log.parsing_filters.parse_data_line = `
error("No #Fields line detected in syslog data--cannot parse log data without a #Fields line");
`

  # Get search engine and search phrase information from the referrer field (before it gets simplified).
  log.parsing_filters.compute_se_sp = {
    value = `
if (get_search_engine_info(referrer)) then (
  search_engine = volatile.search_engine;
  search_phrase = volatile.search_phrase;
);
`
    requires_fields = {
      referrer = true
    }
  }

  # Get web browser, operating system, web browser, and spider information from the user-agent field.
  log.parsing_filters.derive_from_user_agent = {
    value = `get_user_agent_info(replace_all(cs_user_agent, '+', ' '));
web_browser = volatile.web_browser;
operating_system = volatile.operating_system;
spider = volatile.spider;
`
    requires_fields = {
      cs_user_agent = true
    }
  }


  # Log Filters
  log.filters = {

    detect_broken_links = {
      label = "$lang_admin.log_filters.detect_broken_links_label"
      comment = "$lang_admin.log_filters.detect_broken_links_comment"
      value = "if (sc_status eq '404') then broken_link = cs_uri_stem;"
      requires_fields = {
        sc_status = true
        cs_uri_stem = true
      }
    } # detect_broken_links

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

    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)'"
      requires_fields = {
        referrer = true
      }
    } # 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
      requires_fields = {
        referrer = true
      }
    } # internal_referrer

    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 cs_uri_stem = '(worm)';"
      requires_fields = {
#        cs_uri_stem = true
        worm = true
      }
    } # set_page_for_worm

    # This filter tacks the page parameters ("URL query") onto the end of the page field
    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 = true
      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 = true
      requires_fields = {
        cs_uri_query = true
        cs_uri_stem = true
      }
    } # add_cs_uri_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;"
      requires_fields = {
        file_type = true
      }
    } # 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 cs_uri_stem = substr(cs_uri_stem, 0, last_index(cs_uri_stem, '/') + 1) . '(nonpage)';"
      requires_fields = {
        cs_uri_stem = true
      }
    } # strip_non_page_views

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

  } # log.filters

  log.field_options = {

    sessions_page_field = "cs_uri_stem"
    sessions_visitor_id_field = "c_ip"
    sessions_event_field = "page_views"

  } # log.field_options


  database.numerical_fields = {

    hits = {
      label = "$lang_stats.field_labels.hits"
      default = false
      requires_log_field = false
      type = "int"
      display_format_type = "integer"
      entries_field = true
    } # hits

    page_views = {
      label = "$lang_stats.field_labels.page_views"
      default = true
      requires_log_field = false
      type = "int"
      display_format_type = "integer"
    } # page_views

    visitors = {
      label = "$lang_stats.field_labels.visitors"
      default = false
      requires_log_field = true
      log_field = "c_ip"
      type = "unique"
      display_format_type = "integer"
    } # visitors

    sc_bytes = {
      label = "$lang_stats.field_labels.sc_bytes"
      default = false
      log_field = "sc_bytes"
      requires_log_field = true
      type = "int"
      integer_bits = 64
      display_format_type = "bandwidth"
    }

    cs_bytes = {
      label = "$lang_stats.field_labels.cs_bytes"
      default = false
      log_field = "cs_bytes"
      requires_log_field = true
      type = "int"
      integer_bits = 64
      display_format_type = "bandwidth"
    }

    time_taken = {
      label = $lang_stats.field_labels.time_taken
      default = false
      requires_log_field = true
      type = "int"
      integer_bits = 64
      display_format_type = duration_milliseconds
    } # 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 = {
    date_time_tracking = true
    host_tracking = true

    # How the reports should be grouped in the report menu
    report_groups = {
      date_time_group = ""
      content_group = {
        file_type = true
        cs_uri_stem = true
        cs_uri_query = true
        broken_link = true
      }
      visitor_demographics_group = {
        c_ip = true
        domain_description = true
        location = true
        organization = true
        isp = true
        domain = true
        cs_username = true
      }
      visitor_systems_group = {
        screen_dimensions = true
        screen_depth = true
        web_browser = true
        operating_system = true
      }
      referrer_group = {
        referrer = true
        search_engine = true
        search_phrase = true
        referrer_description = true
        search_phrase_by_search_engine = true
      }
      server_group = {
        s_sitename = true
        s_computername = true
        s_ip = true
        s_port = true
        cs_host = true
      }
      other_group = {
        cs_version = true
        worm = true
        spider = true
        cs_method = true
        sc_status = true
        sc_substatus = true
        sc_win32_status = true
        cs_cookie = true
      }
    } # report_groups

    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

##
##  Uncommenting this section will provide enhanced granularity visitor tracking by over-riding
##  default visitor tracking and uniqueness using c_ip with c_ip + cs_user_agent in
##  all reports including sessions analysis reporting
##
##  log.fields.visitor_id = ""
##  database.fields.visitor_id = ""
##  database.numerical_fields.visitors.log_field = "visitor_id"
##
##  log.filters.visitor_id = {
##      label = 'Client IP + user Agent'
##      comment = 'Provides better tracking of unique visitors when multiple users are coming from the same client IP (ISP/NAT/FIREWALL/ROUTER)'
##      value = 'visitor_id = c_ip . "+" . cs_user_agent;'
##  } # log.filters.visitor_id
##
##  log.field_options.sessions_visitor_id_field = "visitor_id"

} # iisweb_with_syslog