search sombrerowebdoc {

  document sombrerowebdoc {
    version : 0

  struct sombreroattribute {
    field attributetype type int {}
    field intvalue      type int {}
    field huffmanindex  type int {}
    field lexicon       type string {}
    field name          type string {}
    field strvalue      type string {}
  }

  annotation sombreroTOKEN { }

  annotation sombreroPHRASE {
    field prefixused           type int {}
    field suffixused           type int {}
    field attributes           type array<sombreroattribute> {}
    field annotationreferences type array<annotationreference<sombreroTOKEN>> {}
  }

  struct sombrerocontentfilter {
    field id    type string {}
    field value type string {}
  }

  struct sombrerorule {
    field name           type string {}
    field location       type string {}
    field contentfilters type array<sombrerocontentfilter> {}
  }

  struct sombrerometrics {
    field executiontime type double {}
  }

  struct sombreroextraction {
    field adjuncts type string {}
    field metrics  type sombrerometrics {}
    field status   type string {}
  }


    field url type string {}

    field raw_html type string {}

    field decoded_html type string {}

    field rules type array<sombrerorule> {}

    field extraction type sombreroextraction {}

    field content_language type string {}

    field dupdetectshinglesketch type string {}

    field mlprisma type string {}

    field raw_bytes type raw {}

    field charset_encoding type string {}

    field meta_language type string {}

    field language type string {}

    field region type string {}

  }
}