schema

Documentation

Index

• Variables
• type Cdx
  • func (*Cdx) Descriptor() ([]byte, []int)
  • func (x *Cdx) GetCle() int64
  • func (x *Cdx) GetDig() string
  • func (x *Cdx) GetHsc() int32
  • func (x *Cdx) GetMct() string
  • func (x *Cdx) GetPle() int64
  • func (x *Cdx) GetRct() string
  • func (x *Cdx) GetRef() string
  • func (x *Cdx) GetRid() string
  • func (x *Cdx) GetRle() int64
  • func (x *Cdx) GetRod() *timestamppb.Timestamp
  • func (x *Cdx) GetRoi() string
  • func (x *Cdx) GetRou() string
  • func (x *Cdx) GetSha() string
  • func (x *Cdx) GetSrt() string
  • func (x *Cdx) GetSsu() string
  • func (x *Cdx) GetSts() *timestamppb.Timestamp
  • func (x *Cdx) GetUri() string
  • func (*Cdx) ProtoMessage()
  • func (x *Cdx) ProtoReflect() protoreflect.Message
  • func (x *Cdx) Reset()
  • func (x *Cdx) String() string
• type Fileinfo
  • func (*Fileinfo) Descriptor() ([]byte, []int)
  • func (x *Fileinfo) GetLastModified() *timestamppb.Timestamp
  • func (x *Fileinfo) GetName() string
  • func (x *Fileinfo) GetPath() string
  • func (x *Fileinfo) GetSize() int64
  • func (*Fileinfo) ProtoMessage()
  • func (x *Fileinfo) ProtoReflect() protoreflect.Message
  • func (x *Fileinfo) Reset()
  • func (x *Fileinfo) String() string

Variables

var File_cdx_proto protoreflect.FileDescriptor

var File_fileinfo_proto protoreflect.FileDescriptor

Types

type Cdx

type Cdx struct {

	// uri (required) - The value should be the non-transformed URI used for the searchable URI (first sortable field).
	Uri string `protobuf:"bytes,1,opt,name=uri,proto3" json:"uri,omitempty"`
	// sha (recommended) - A Base32 encoded SHA-1 digest of the payload that this record refers to. Omit if the URI has
	// no intrinsic payload. For revisit records, this is the digest of the original payload. The algorithm prefix
	// (e.g. sha-1) is not included in this field. See dig for alternative hashing algorithms.
	Sha string `protobuf:"bytes,2,opt,name=sha,proto3" json:"sha,omitempty"`
	// dig - A Base32 encoded output of a hashing algorithm applied to the URI’s payload. This should include a prefix
	// indicating the algorithm.
	Dig string `protobuf:"bytes,3,opt,name=dig,proto3" json:"dig,omitempty"`
	// hsc - HTTP Status Code. Applicable for response records for HTTP(S) URIs.
	Hsc int32 `protobuf:"varint,4,opt,name=hsc,proto3" json:"hsc,omitempty"`
	// mct - Media Content Type (MIME type). For HTTP(S) response records this is typically the “Content-Type” from
	// the HTTP header. This field, however, does not specify the origin of the information. It may be used to include
	// content type that was derived from content analysis or other sources.
	Mct string `protobuf:"bytes,5,opt,name=mct,proto3" json:"mct,omitempty"`
	// ref (required) - A URI that resolves to the resource that this record refers to. This can be any well defined
	// URI scheme. For the most common web archive use case of warc filename plus offset, see Appendix C. For other
	// use cases, existing schemes can be used or new ones devised.
	Ref string `protobuf:"bytes,6,opt,name=ref,proto3" json:"ref,omitempty"`
	// rid (recommended) - Record ID. Typically WARC-Record-ID or equivalent if not using WARCs. In a mixed environment,
	// you should ensure that record ID is unique.
	Rid string `protobuf:"bytes,7,opt,name=rid,proto3" json:"rid,omitempty"`
	// cle - Content Length. The length of the content (uncompressed), ignoring WARC headers, but including any
	// HTTP headers or similar.
	Cle int64 `protobuf:"varint,8,opt,name=cle,proto3" json:"cle,omitempty"`
	// ple - Payload Length. The length of the payload (uncompressed). The exact meaning will vary by content type,
	// but the common case is the length of the document, excluding any HTTP headers in a HTTP response record.
	Ple int64 `protobuf:"varint,9,opt,name=ple,proto3" json:"ple,omitempty"`
	// rle - Record Length. The length of the record that this line refers to. This is the entire record
	// (including e.g. WARC headers) as written on disk (compressed if stored compressed).
	Rle int64 `protobuf:"varint,10,opt,name=rle,proto3" json:"rle,omitempty"`
	// rct - Record Concurrant To. The record ID of another record that the current record is considered to be
	// ‘concurrant’ to. See further WARC chapter 5.7 (WARC-Concurrent-To).
	Rct string `protobuf:"bytes,11,opt,name=rct,proto3" json:"rct,omitempty"`
	// rou (recommended) - Revisit Original URI. Only valid for records of type revisit. Contains the URI of
	// the record that this record is considered a revisit of.
	Rou string `protobuf:"bytes,12,opt,name=rou,proto3" json:"rou,omitempty"`
	// rod (recommended) - Revisit Original Date. Only valid for records of type revisit. Contains the
	// timestamp (equivalent to sortable field #2) of the record that this record is considered a revisit of.
	Rod *timestamppb.Timestamp `protobuf:"bytes,13,opt,name=rod,proto3" json:"rod,omitempty"`
	// roi - Revisit Original record ID. Only valid for records of type revisit. Contains the record ID of
	// the record that this record is considered a revisit of.
	Roi string `protobuf:"bytes,14,opt,name=roi,proto3" json:"roi,omitempty"`
	// Searchable URI - ssu (sortable searchable URI)
	Ssu string `protobuf:"bytes,15,opt,name=ssu,proto3" json:"ssu,omitempty"`
	// Timestamp - sts (sortable timestamp)
	Sts *timestamppb.Timestamp `protobuf:"bytes,16,opt,name=sts,proto3" json:"sts,omitempty"`
	// Record Type - srt (sortable record type)
	Srt string `protobuf:"bytes,17,opt,name=srt,proto3" json:"srt,omitempty"`
	// contains filtered or unexported fields
}

func (*Cdx) Descriptor

func (*Cdx) Descriptor() ([]byte, []int)

Deprecated: Use Cdx.ProtoReflect.Descriptor instead.

func (*Cdx) GetCle

func (x *Cdx) GetCle() int64

func (*Cdx) GetDig

func (x *Cdx) GetDig() string

func (*Cdx) GetHsc

func (x *Cdx) GetHsc() int32

func (*Cdx) GetMct

func (x *Cdx) GetMct() string

func (*Cdx) GetPle

func (x *Cdx) GetPle() int64

func (*Cdx) GetRct

func (x *Cdx) GetRct() string

func (*Cdx) GetRef

func (x *Cdx) GetRef() string

func (*Cdx) GetRid

func (x *Cdx) GetRid() string

func (*Cdx) GetRle

func (x *Cdx) GetRle() int64

func (*Cdx) GetRod

func (x *Cdx) GetRod() *timestamppb.Timestamp

func (*Cdx) GetRoi

func (x *Cdx) GetRoi() string

func (*Cdx) GetRou

func (x *Cdx) GetRou() string

func (*Cdx) GetSha

func (x *Cdx) GetSha() string

func (*Cdx) GetSrt

func (x *Cdx) GetSrt() string

func (*Cdx) GetSsu

func (x *Cdx) GetSsu() string

func (*Cdx) GetSts

func (x *Cdx) GetSts() *timestamppb.Timestamp

func (*Cdx) GetUri

func (x *Cdx) GetUri() string

func (*Cdx) ProtoMessage

func (*Cdx) ProtoMessage()

func (*Cdx) ProtoReflect

func (x *Cdx) ProtoReflect() protoreflect.Message

func (*Cdx) Reset

func (x *Cdx) Reset()

func (*Cdx) String

func (x *Cdx) String() string

type Fileinfo

type Fileinfo struct {

	// Filename
	Name string `protobuf:"bytes,1,opt,name=name,proto3" json:"name,omitempty"`
	// Full path
	Path string `protobuf:"bytes,2,opt,name=path,proto3" json:"path,omitempty"`
	// Last modified
	LastModified *timestamppb.Timestamp `protobuf:"bytes,3,opt,name=last_modified,json=lastModified,proto3" json:"last_modified,omitempty"`
	// File size
	Size int64 `protobuf:"varint,4,opt,name=size,proto3" json:"size,omitempty"`
	// contains filtered or unexported fields
}

func (*Fileinfo) Descriptor

func (*Fileinfo) Descriptor() ([]byte, []int)

Deprecated: Use Fileinfo.ProtoReflect.Descriptor instead.

func (*Fileinfo) GetLastModified

func (x *Fileinfo) GetLastModified() *timestamppb.Timestamp

func (*Fileinfo) GetName

func (x *Fileinfo) GetName() string

func (*Fileinfo) GetPath

func (x *Fileinfo) GetPath() string

func (*Fileinfo) GetSize

func (x *Fileinfo) GetSize() int64

func (*Fileinfo) ProtoMessage

func (*Fileinfo) ProtoMessage()

func (*Fileinfo) ProtoReflect

func (x *Fileinfo) ProtoReflect() protoreflect.Message

func (*Fileinfo) Reset

func (x *Fileinfo) Reset()

func (*Fileinfo) String

func (x *Fileinfo) String() string