Package ai.pipestream.parsed.data.warc.v1

Class WarcMetadata

java.lang.Object
com.google.protobuf.AbstractMessageLite
com.google.protobuf.AbstractMessage
com.google.protobuf.GeneratedMessage
ai.pipestream.parsed.data.warc.v1.WarcMetadata
All Implemented Interfaces:
WarcMetadataOrBuilder, com.google.protobuf.Message, com.google.protobuf.MessageLite, com.google.protobuf.MessageLiteOrBuilder, com.google.protobuf.MessageOrBuilder, Serializable
@Generated public final class WarcMetadata extends com.google.protobuf.GeneratedMessage implements WarcMetadataOrBuilder
 Comprehensive WARC (Web ARChive) metadata as extracted by Tika from WARC and ARC files.
 Based on Tika's WARC interface, WARCParser implementation, and WARC/ARC specifications.
 This is a 1:1 mapping to all fields that Tika can extract from web archive files.
 WARC files are the standard format for web archiving, containing:
 - Web pages and their resources
 - HTTP request/response pairs
 - Metadata about the crawl process
 - Relationships between archived resources
 Critical for digital preservation and web content analysis.
Protobuf type ai.pipestream.parsed.data.warc.v1.WarcMetadata
See Also:

  • Field Details

    • WARC_WARNINGS_FIELD_NUMBER

      public static final int WARC_WARNINGS_FIELD_NUMBER
      See Also:

    • WARC_RECORD_CONTENT_TYPE_FIELD_NUMBER

      public static final int WARC_RECORD_CONTENT_TYPE_FIELD_NUMBER
      See Also:

    • WARC_PAYLOAD_CONTENT_TYPE_FIELD_NUMBER

      public static final int WARC_PAYLOAD_CONTENT_TYPE_FIELD_NUMBER
      See Also:

    • WARC_RECORD_ID_FIELD_NUMBER

      public static final int WARC_RECORD_ID_FIELD_NUMBER
      See Also:

    • WARC_TYPE_FIELD_NUMBER

      public static final int WARC_TYPE_FIELD_NUMBER
      See Also:

    • WARC_TARGET_URI_FIELD_NUMBER

      public static final int WARC_TARGET_URI_FIELD_NUMBER
      See Also:

    • WARC_DATE_FIELD_NUMBER

      public static final int WARC_DATE_FIELD_NUMBER
      See Also:

    • WARC_CONTENT_LENGTH_FIELD_NUMBER

      public static final int WARC_CONTENT_LENGTH_FIELD_NUMBER
      See Also:

    • WARC_FILENAME_FIELD_NUMBER

      public static final int WARC_FILENAME_FIELD_NUMBER
      See Also:

    • WARC_REFERS_TO_FIELD_NUMBER

      public static final int WARC_REFERS_TO_FIELD_NUMBER
      See Also:

    • WARC_CONCURRENT_TO_FIELD_NUMBER

      public static final int WARC_CONCURRENT_TO_FIELD_NUMBER
      See Also:

    • WARC_WARCINFO_ID_FIELD_NUMBER

      public static final int WARC_WARCINFO_ID_FIELD_NUMBER
      See Also:

    • WARC_IP_ADDRESS_FIELD_NUMBER

      public static final int WARC_IP_ADDRESS_FIELD_NUMBER
      See Also:

    • WARC_BLOCK_DIGEST_FIELD_NUMBER

      public static final int WARC_BLOCK_DIGEST_FIELD_NUMBER
      See Also:

    • WARC_PAYLOAD_DIGEST_FIELD_NUMBER

      public static final int WARC_PAYLOAD_DIGEST_FIELD_NUMBER
      See Also:

    • WARC_TRUNCATED_FIELD_NUMBER

      public static final int WARC_TRUNCATED_FIELD_NUMBER
      See Also:

    • WARC_IDENTIFIED_PAYLOAD_TYPE_FIELD_NUMBER

      public static final int WARC_IDENTIFIED_PAYLOAD_TYPE_FIELD_NUMBER
      See Also:

    • HTTP_STATUS_CODE_FIELD_NUMBER

      public static final int HTTP_STATUS_CODE_FIELD_NUMBER
      See Also:

    • HTTP_STATUS_REASON_FIELD_NUMBER

      public static final int HTTP_STATUS_REASON_FIELD_NUMBER
      See Also:

    • HTTP_HEADERS_FIELD_NUMBER

      public static final int HTTP_HEADERS_FIELD_NUMBER
      See Also:

    • WARC_CREATED_BY_FIELD_NUMBER

      public static final int WARC_CREATED_BY_FIELD_NUMBER
      See Also:

    • WARC_FORMAT_VERSION_FIELD_NUMBER

      public static final int WARC_FORMAT_VERSION_FIELD_NUMBER
      See Also:

    • WARC_COLLECTION_FIELD_NUMBER

      public static final int WARC_COLLECTION_FIELD_NUMBER
      See Also:

    • WARC_CRAWL_ID_FIELD_NUMBER

      public static final int WARC_CRAWL_ID_FIELD_NUMBER
      See Also:

    • WARC_ROBOT_POLICY_FIELD_NUMBER

      public static final int WARC_ROBOT_POLICY_FIELD_NUMBER
      See Also:

    • FETCH_TIMESTAMP_FIELD_NUMBER

      public static final int FETCH_TIMESTAMP_FIELD_NUMBER
      See Also:

    • CONTENT_LANGUAGE_FIELD_NUMBER

      public static final int CONTENT_LANGUAGE_FIELD_NUMBER
      See Also:

    • CONTENT_ENCODING_FIELD_NUMBER

      public static final int CONTENT_ENCODING_FIELD_NUMBER
      See Also:

    • ESTIMATED_WORD_COUNT_FIELD_NUMBER

      public static final int ESTIMATED_WORD_COUNT_FIELD_NUMBER
      See Also:

    • ESTIMATED_CHARACTER_COUNT_FIELD_NUMBER

      public static final int ESTIMATED_CHARACTER_COUNT_FIELD_NUMBER
      See Also:

    • IS_TRUNCATED_FIELD_NUMBER

      public static final int IS_TRUNCATED_FIELD_NUMBER
      See Also:

    • COMPRESSION_METHOD_FIELD_NUMBER

      public static final int COMPRESSION_METHOD_FIELD_NUMBER
      See Also:

    • HTTP_METHOD_FIELD_NUMBER

      public static final int HTTP_METHOD_FIELD_NUMBER
      See Also:

    • HTTP_VERSION_FIELD_NUMBER

      public static final int HTTP_VERSION_FIELD_NUMBER
      See Also:

    • REDIRECT_CHAIN_FIELD_NUMBER

      public static final int REDIRECT_CHAIN_FIELD_NUMBER
      See Also:

    • COOKIES_FIELD_NUMBER

      public static final int COOKIES_FIELD_NUMBER
      See Also:

    • REFERRER_URL_FIELD_NUMBER

      public static final int REFERRER_URL_FIELD_NUMBER
      See Also:

    • USER_AGENT_FIELD_NUMBER

      public static final int USER_AGENT_FIELD_NUMBER
      See Also:

    • RECORD_POSITION_FIELD_NUMBER

      public static final int RECORD_POSITION_FIELD_NUMBER
      See Also:

    • RECORD_SIZE_FIELD_NUMBER

      public static final int RECORD_SIZE_FIELD_NUMBER
      See Also:

    • TOTAL_RECORDS_IN_FILE_FIELD_NUMBER

      public static final int TOTAL_RECORDS_IN_FILE_FIELD_NUMBER
      See Also:

    • RECORD_INDEX_FIELD_NUMBER

      public static final int RECORD_INDEX_FIELD_NUMBER
      See Also:

    • HAS_METADATA_RECORDS_FIELD_NUMBER

      public static final int HAS_METADATA_RECORDS_FIELD_NUMBER
      See Also:

    • HAS_REQUEST_RECORDS_FIELD_NUMBER

      public static final int HAS_REQUEST_RECORDS_FIELD_NUMBER
      See Also:

    • VALIDATION_ERRORS_FIELD_NUMBER

      public static final int VALIDATION_ERRORS_FIELD_NUMBER
      See Also:

    • VALIDATION_WARNINGS_FIELD_NUMBER

      public static final int VALIDATION_WARNINGS_FIELD_NUMBER
      See Also:

    • COMPLETENESS_SCORE_FIELD_NUMBER

      public static final int COMPLETENESS_SCORE_FIELD_NUMBER
      See Also:

    • CHECKSUM_VERIFIED_FIELD_NUMBER

      public static final int CHECKSUM_VERIFIED_FIELD_NUMBER
      See Also:

    • ARCHIVE_INSTITUTION_FIELD_NUMBER

      public static final int ARCHIVE_INSTITUTION_FIELD_NUMBER
      See Also:

    • ARCHIVE_COLLECTION_NAME_FIELD_NUMBER

      public static final int ARCHIVE_COLLECTION_NAME_FIELD_NUMBER
      See Also:

    • ACCESS_RIGHTS_FIELD_NUMBER

      public static final int ACCESS_RIGHTS_FIELD_NUMBER
      See Also:

    • PRESERVATION_ACTIONS_FIELD_NUMBER

      public static final int PRESERVATION_ACTIONS_FIELD_NUMBER
      See Also:

    • MIGRATION_HISTORY_FIELD_NUMBER

      public static final int MIGRATION_HISTORY_FIELD_NUMBER
      See Also:

    • DETECTED_MIME_TYPE_FIELD_NUMBER

      public static final int DETECTED_MIME_TYPE_FIELD_NUMBER
      See Also:

    • FORMAT_VERSION_FIELD_NUMBER

      public static final int FORMAT_VERSION_FIELD_NUMBER
      See Also:

    • TECHNICAL_PROPERTIES_FIELD_NUMBER

      public static final int TECHNICAL_PROPERTIES_FIELD_NUMBER
      See Also:

    • EMBEDDED_RESOURCES_FIELD_NUMBER

      public static final int EMBEDDED_RESOURCES_FIELD_NUMBER
      See Also:

    • ADDITIONAL_METADATA_FIELD_NUMBER

      public static final int ADDITIONAL_METADATA_FIELD_NUMBER
      See Also:

    • BASE_FIELDS_FIELD_NUMBER

      public static final int BASE_FIELDS_FIELD_NUMBER
      See Also:

  • Method Details

    • getDescriptor

      public static final com.google.protobuf.Descriptors.Descriptor getDescriptor()

    • internalGetFieldAccessorTable

      protected com.google.protobuf.GeneratedMessage.FieldAccessorTable internalGetFieldAccessorTable()
      Specified by:
      internalGetFieldAccessorTable in class com.google.protobuf.GeneratedMessage

    • getWarcWarningsList

      public com.google.protobuf.ProtocolStringList getWarcWarningsList()
       WARC warnings encountered during parsing
 Multiple warnings can be present
 WARC.WARC_WARNING
      repeated string warc_warnings = 1 [json_name = "warcWarnings"];
      Specified by:
      getWarcWarningsList in interface WarcMetadataOrBuilder
      Returns:
      A list containing the warcWarnings.

    • getWarcWarningsCount

      public int getWarcWarningsCount()
       WARC warnings encountered during parsing
 Multiple warnings can be present
 WARC.WARC_WARNING
      repeated string warc_warnings = 1 [json_name = "warcWarnings"];
      Specified by:
      getWarcWarningsCount in interface WarcMetadataOrBuilder
      Returns:
      The count of warcWarnings.

    • getWarcWarnings

      public String getWarcWarnings(int index)
       WARC warnings encountered during parsing
 Multiple warnings can be present
 WARC.WARC_WARNING
      repeated string warc_warnings = 1 [json_name = "warcWarnings"];
      Specified by:
      getWarcWarnings in interface WarcMetadataOrBuilder
      Parameters:
      index - The index of the element to return.
      Returns:
      The warcWarnings at the given index.

    • getWarcWarningsBytes

      public com.google.protobuf.ByteString getWarcWarningsBytes(int index)
       WARC warnings encountered during parsing
 Multiple warnings can be present
 WARC.WARC_WARNING
      repeated string warc_warnings = 1 [json_name = "warcWarnings"];
      Specified by:
      getWarcWarningsBytes in interface WarcMetadataOrBuilder
      Parameters:
      index - The index of the value to return.
      Returns:
      The bytes of the warcWarnings at the given index.

    • hasWarcRecordContentType

      public boolean hasWarcRecordContentType()
       Content type of the WARC record itself
 e.g., "application/http; msgtype=response"
 WARC.WARC_RECORD_CONTENT_TYPE
      optional string warc_record_content_type = 2 [json_name = "warcRecordContentType"];
      Specified by:
      hasWarcRecordContentType in interface WarcMetadataOrBuilder
      Returns:
      Whether the warcRecordContentType field is set.

    • getWarcRecordContentType

      public String getWarcRecordContentType()
       Content type of the WARC record itself
 e.g., "application/http; msgtype=response"
 WARC.WARC_RECORD_CONTENT_TYPE
      optional string warc_record_content_type = 2 [json_name = "warcRecordContentType"];
      Specified by:
      getWarcRecordContentType in interface WarcMetadataOrBuilder
      Returns:
      The warcRecordContentType.

    • getWarcRecordContentTypeBytes

      public com.google.protobuf.ByteString getWarcRecordContentTypeBytes()
       Content type of the WARC record itself
 e.g., "application/http; msgtype=response"
 WARC.WARC_RECORD_CONTENT_TYPE
      optional string warc_record_content_type = 2 [json_name = "warcRecordContentType"];
      Specified by:
      getWarcRecordContentTypeBytes in interface WarcMetadataOrBuilder
      Returns:
      The bytes for warcRecordContentType.

    • hasWarcPayloadContentType

      public boolean hasWarcPayloadContentType()
       Content type of the payload within the WARC record
 e.g., "text/html", "image/jpeg"
 WARC.WARC_PAYLOAD_CONTENT_TYPE
      optional string warc_payload_content_type = 3 [json_name = "warcPayloadContentType"];
      Specified by:
      hasWarcPayloadContentType in interface WarcMetadataOrBuilder
      Returns:
      Whether the warcPayloadContentType field is set.

    • getWarcPayloadContentType

      public String getWarcPayloadContentType()
       Content type of the payload within the WARC record
 e.g., "text/html", "image/jpeg"
 WARC.WARC_PAYLOAD_CONTENT_TYPE
      optional string warc_payload_content_type = 3 [json_name = "warcPayloadContentType"];
      Specified by:
      getWarcPayloadContentType in interface WarcMetadataOrBuilder
      Returns:
      The warcPayloadContentType.

    • getWarcPayloadContentTypeBytes

      public com.google.protobuf.ByteString getWarcPayloadContentTypeBytes()
       Content type of the payload within the WARC record
 e.g., "text/html", "image/jpeg"
 WARC.WARC_PAYLOAD_CONTENT_TYPE
      optional string warc_payload_content_type = 3 [json_name = "warcPayloadContentType"];
      Specified by:
      getWarcPayloadContentTypeBytes in interface WarcMetadataOrBuilder
      Returns:
      The bytes for warcPayloadContentType.

    • hasWarcRecordId

      public boolean hasWarcRecordId()
       Unique identifier for this WARC record
 Format: <urn:uuid:...> or similar URI
 WARC.WARC_RECORD_ID
      optional string warc_record_id = 4 [json_name = "warcRecordId"];
      Specified by:
      hasWarcRecordId in interface WarcMetadataOrBuilder
      Returns:
      Whether the warcRecordId field is set.

    • getWarcRecordId

      public String getWarcRecordId()
       Unique identifier for this WARC record
 Format: <urn:uuid:...> or similar URI
 WARC.WARC_RECORD_ID
      optional string warc_record_id = 4 [json_name = "warcRecordId"];
      Specified by:
      getWarcRecordId in interface WarcMetadataOrBuilder
      Returns:
      The warcRecordId.

    • getWarcRecordIdBytes

      public com.google.protobuf.ByteString getWarcRecordIdBytes()
       Unique identifier for this WARC record
 Format: <urn:uuid:...> or similar URI
 WARC.WARC_RECORD_ID
      optional string warc_record_id = 4 [json_name = "warcRecordId"];
      Specified by:
      getWarcRecordIdBytes in interface WarcMetadataOrBuilder
      Returns:
      The bytes for warcRecordId.

    • hasWarcType

      public boolean hasWarcType()
       Type of WARC record (response, request, metadata, resource, etc.)
 WARC-Type header
      optional string warc_type = 5 [json_name = "warcType"];
      Specified by:
      hasWarcType in interface WarcMetadataOrBuilder
      Returns:
      Whether the warcType field is set.

    • getWarcType

      public String getWarcType()
       Type of WARC record (response, request, metadata, resource, etc.)
 WARC-Type header
      optional string warc_type = 5 [json_name = "warcType"];
      Specified by:
      getWarcType in interface WarcMetadataOrBuilder
      Returns:
      The warcType.

    • getWarcTypeBytes

      public com.google.protobuf.ByteString getWarcTypeBytes()
       Type of WARC record (response, request, metadata, resource, etc.)
 WARC-Type header
      optional string warc_type = 5 [json_name = "warcType"];
      Specified by:
      getWarcTypeBytes in interface WarcMetadataOrBuilder
      Returns:
      The bytes for warcType.

    • hasWarcTargetUri

      public boolean hasWarcTargetUri()
       Target URI that was archived
 The original URL of the web resource
 WARC-Target-URI
      optional string warc_target_uri = 6 [json_name = "warcTargetUri"];
      Specified by:
      hasWarcTargetUri in interface WarcMetadataOrBuilder
      Returns:
      Whether the warcTargetUri field is set.

    • getWarcTargetUri

      public String getWarcTargetUri()
       Target URI that was archived
 The original URL of the web resource
 WARC-Target-URI
      optional string warc_target_uri = 6 [json_name = "warcTargetUri"];
      Specified by:
      getWarcTargetUri in interface WarcMetadataOrBuilder
      Returns:
      The warcTargetUri.

    • getWarcTargetUriBytes

      public com.google.protobuf.ByteString getWarcTargetUriBytes()
       Target URI that was archived
 The original URL of the web resource
 WARC-Target-URI
      optional string warc_target_uri = 6 [json_name = "warcTargetUri"];
      Specified by:
      getWarcTargetUriBytes in interface WarcMetadataOrBuilder
      Returns:
      The bytes for warcTargetUri.

    • hasWarcDate

      public boolean hasWarcDate()
       Date and time when the resource was captured
 WARC-Date
      optional .google.protobuf.Timestamp warc_date = 7 [json_name = "warcDate"];
      Specified by:
      hasWarcDate in interface WarcMetadataOrBuilder
      Returns:
      Whether the warcDate field is set.

    • getWarcDate

      public com.google.protobuf.Timestamp getWarcDate()
       Date and time when the resource was captured
 WARC-Date
      optional .google.protobuf.Timestamp warc_date = 7 [json_name = "warcDate"];
      Specified by:
      getWarcDate in interface WarcMetadataOrBuilder
      Returns:
      The warcDate.

    • getWarcDateOrBuilder

      public com.google.protobuf.TimestampOrBuilder getWarcDateOrBuilder()
       Date and time when the resource was captured
 WARC-Date
      optional .google.protobuf.Timestamp warc_date = 7 [json_name = "warcDate"];
      Specified by:
      getWarcDateOrBuilder in interface WarcMetadataOrBuilder

    • hasWarcContentLength

      public boolean hasWarcContentLength()
       Length of the record payload in bytes
 Content-Length
      optional int64 warc_content_length = 8 [json_name = "warcContentLength"];
      Specified by:
      hasWarcContentLength in interface WarcMetadataOrBuilder
      Returns:
      Whether the warcContentLength field is set.

    • getWarcContentLength

      public long getWarcContentLength()
       Length of the record payload in bytes
 Content-Length
      optional int64 warc_content_length = 8 [json_name = "warcContentLength"];
      Specified by:
      getWarcContentLength in interface WarcMetadataOrBuilder
      Returns:
      The warcContentLength.

    • hasWarcFilename

      public boolean hasWarcFilename()
       WARC filename where this record is stored
 WARC-Filename
      optional string warc_filename = 9 [json_name = "warcFilename"];
      Specified by:
      hasWarcFilename in interface WarcMetadataOrBuilder
      Returns:
      Whether the warcFilename field is set.

    • getWarcFilename

      public String getWarcFilename()
       WARC filename where this record is stored
 WARC-Filename
      optional string warc_filename = 9 [json_name = "warcFilename"];
      Specified by:
      getWarcFilename in interface WarcMetadataOrBuilder
      Returns:
      The warcFilename.

    • getWarcFilenameBytes

      public com.google.protobuf.ByteString getWarcFilenameBytes()
       WARC filename where this record is stored
 WARC-Filename
      optional string warc_filename = 9 [json_name = "warcFilename"];
      Specified by:
      getWarcFilenameBytes in interface WarcMetadataOrBuilder
      Returns:
      The bytes for warcFilename.

    • hasWarcRefersTo

      public boolean hasWarcRefersTo()
       WARC record ID that this record refers to
 Used for linking related records
 WARC-Refers-To
      optional string warc_refers_to = 10 [json_name = "warcRefersTo"];
      Specified by:
      hasWarcRefersTo in interface WarcMetadataOrBuilder
      Returns:
      Whether the warcRefersTo field is set.

    • getWarcRefersTo

      public String getWarcRefersTo()
       WARC record ID that this record refers to
 Used for linking related records
 WARC-Refers-To
      optional string warc_refers_to = 10 [json_name = "warcRefersTo"];
      Specified by:
      getWarcRefersTo in interface WarcMetadataOrBuilder
      Returns:
      The warcRefersTo.

    • getWarcRefersToBytes

      public com.google.protobuf.ByteString getWarcRefersToBytes()
       WARC record ID that this record refers to
 Used for linking related records
 WARC-Refers-To
      optional string warc_refers_to = 10 [json_name = "warcRefersTo"];
      Specified by:
      getWarcRefersToBytes in interface WarcMetadataOrBuilder
      Returns:
      The bytes for warcRefersTo.

    • hasWarcConcurrentTo

      public boolean hasWarcConcurrentTo()
       WARC record ID that this record is concurrent with
 WARC-Concurrent-To
      optional string warc_concurrent_to = 11 [json_name = "warcConcurrentTo"];
      Specified by:
      hasWarcConcurrentTo in interface WarcMetadataOrBuilder
      Returns:
      Whether the warcConcurrentTo field is set.

    • getWarcConcurrentTo

      public String getWarcConcurrentTo()
       WARC record ID that this record is concurrent with
 WARC-Concurrent-To
      optional string warc_concurrent_to = 11 [json_name = "warcConcurrentTo"];
      Specified by:
      getWarcConcurrentTo in interface WarcMetadataOrBuilder
      Returns:
      The warcConcurrentTo.

    • getWarcConcurrentToBytes

      public com.google.protobuf.ByteString getWarcConcurrentToBytes()
       WARC record ID that this record is concurrent with
 WARC-Concurrent-To
      optional string warc_concurrent_to = 11 [json_name = "warcConcurrentTo"];
      Specified by:
      getWarcConcurrentToBytes in interface WarcMetadataOrBuilder
      Returns:
      The bytes for warcConcurrentTo.

    • hasWarcWarcinfoId

      public boolean hasWarcWarcinfoId()
       WARC record ID of the warcinfo record for this crawl
 WARC-Warcinfo-ID
      optional string warc_warcinfo_id = 12 [json_name = "warcWarcinfoId"];
      Specified by:
      hasWarcWarcinfoId in interface WarcMetadataOrBuilder
      Returns:
      Whether the warcWarcinfoId field is set.

    • getWarcWarcinfoId

      public String getWarcWarcinfoId()
       WARC record ID of the warcinfo record for this crawl
 WARC-Warcinfo-ID
      optional string warc_warcinfo_id = 12 [json_name = "warcWarcinfoId"];
      Specified by:
      getWarcWarcinfoId in interface WarcMetadataOrBuilder
      Returns:
      The warcWarcinfoId.

    • getWarcWarcinfoIdBytes

      public com.google.protobuf.ByteString getWarcWarcinfoIdBytes()
       WARC record ID of the warcinfo record for this crawl
 WARC-Warcinfo-ID
      optional string warc_warcinfo_id = 12 [json_name = "warcWarcinfoId"];
      Specified by:
      getWarcWarcinfoIdBytes in interface WarcMetadataOrBuilder
      Returns:
      The bytes for warcWarcinfoId.

    • hasWarcIpAddress

      public boolean hasWarcIpAddress()
       IP address of the server that provided this resource
 WARC-IP-Address
      optional string warc_ip_address = 13 [json_name = "warcIpAddress"];
      Specified by:
      hasWarcIpAddress in interface WarcMetadataOrBuilder
      Returns:
      Whether the warcIpAddress field is set.

    • getWarcIpAddress

      public String getWarcIpAddress()
       IP address of the server that provided this resource
 WARC-IP-Address
      optional string warc_ip_address = 13 [json_name = "warcIpAddress"];
      Specified by:
      getWarcIpAddress in interface WarcMetadataOrBuilder
      Returns:
      The warcIpAddress.

    • getWarcIpAddressBytes

      public com.google.protobuf.ByteString getWarcIpAddressBytes()
       IP address of the server that provided this resource
 WARC-IP-Address
      optional string warc_ip_address = 13 [json_name = "warcIpAddress"];
      Specified by:
      getWarcIpAddressBytes in interface WarcMetadataOrBuilder
      Returns:
      The bytes for warcIpAddress.

    • hasWarcBlockDigest

      public boolean hasWarcBlockDigest()
       Block digest (hash) of the payload
 WARC-Block-Digest
      optional string warc_block_digest = 14 [json_name = "warcBlockDigest"];
      Specified by:
      hasWarcBlockDigest in interface WarcMetadataOrBuilder
      Returns:
      Whether the warcBlockDigest field is set.

    • getWarcBlockDigest

      public String getWarcBlockDigest()
       Block digest (hash) of the payload
 WARC-Block-Digest
      optional string warc_block_digest = 14 [json_name = "warcBlockDigest"];
      Specified by:
      getWarcBlockDigest in interface WarcMetadataOrBuilder
      Returns:
      The warcBlockDigest.

    • getWarcBlockDigestBytes

      public com.google.protobuf.ByteString getWarcBlockDigestBytes()
       Block digest (hash) of the payload
 WARC-Block-Digest
      optional string warc_block_digest = 14 [json_name = "warcBlockDigest"];
      Specified by:
      getWarcBlockDigestBytes in interface WarcMetadataOrBuilder
      Returns:
      The bytes for warcBlockDigest.

    • hasWarcPayloadDigest

      public boolean hasWarcPayloadDigest()
       Payload digest (hash) of the actual content
 WARC-Payload-Digest
      optional string warc_payload_digest = 15 [json_name = "warcPayloadDigest"];
      Specified by:
      hasWarcPayloadDigest in interface WarcMetadataOrBuilder
      Returns:
      Whether the warcPayloadDigest field is set.

    • getWarcPayloadDigest

      public String getWarcPayloadDigest()
       Payload digest (hash) of the actual content
 WARC-Payload-Digest
      optional string warc_payload_digest = 15 [json_name = "warcPayloadDigest"];
      Specified by:
      getWarcPayloadDigest in interface WarcMetadataOrBuilder
      Returns:
      The warcPayloadDigest.

    • getWarcPayloadDigestBytes

      public com.google.protobuf.ByteString getWarcPayloadDigestBytes()
       Payload digest (hash) of the actual content
 WARC-Payload-Digest
      optional string warc_payload_digest = 15 [json_name = "warcPayloadDigest"];
      Specified by:
      getWarcPayloadDigestBytes in interface WarcMetadataOrBuilder
      Returns:
      The bytes for warcPayloadDigest.

    • hasWarcTruncated

      public boolean hasWarcTruncated()
       Truncation reason if record was truncated
 WARC-Truncated
      optional string warc_truncated = 16 [json_name = "warcTruncated"];
      Specified by:
      hasWarcTruncated in interface WarcMetadataOrBuilder
      Returns:
      Whether the warcTruncated field is set.

    • getWarcTruncated

      public String getWarcTruncated()
       Truncation reason if record was truncated
 WARC-Truncated
      optional string warc_truncated = 16 [json_name = "warcTruncated"];
      Specified by:
      getWarcTruncated in interface WarcMetadataOrBuilder
      Returns:
      The warcTruncated.

    • getWarcTruncatedBytes

      public com.google.protobuf.ByteString getWarcTruncatedBytes()
       Truncation reason if record was truncated
 WARC-Truncated
      optional string warc_truncated = 16 [json_name = "warcTruncated"];
      Specified by:
      getWarcTruncatedBytes in interface WarcMetadataOrBuilder
      Returns:
      The bytes for warcTruncated.

    • hasWarcIdentifiedPayloadType

      public boolean hasWarcIdentifiedPayloadType()
       Identified payload type (may differ from Content-Type)
 WARC-Identified-Payload-Type
      optional string warc_identified_payload_type = 17 [json_name = "warcIdentifiedPayloadType"];
      Specified by:
      hasWarcIdentifiedPayloadType in interface WarcMetadataOrBuilder
      Returns:
      Whether the warcIdentifiedPayloadType field is set.

    • getWarcIdentifiedPayloadType

      public String getWarcIdentifiedPayloadType()
       Identified payload type (may differ from Content-Type)
 WARC-Identified-Payload-Type
      optional string warc_identified_payload_type = 17 [json_name = "warcIdentifiedPayloadType"];
      Specified by:
      getWarcIdentifiedPayloadType in interface WarcMetadataOrBuilder
      Returns:
      The warcIdentifiedPayloadType.

    • getWarcIdentifiedPayloadTypeBytes

      public com.google.protobuf.ByteString getWarcIdentifiedPayloadTypeBytes()
       Identified payload type (may differ from Content-Type)
 WARC-Identified-Payload-Type
      optional string warc_identified_payload_type = 17 [json_name = "warcIdentifiedPayloadType"];
      Specified by:
      getWarcIdentifiedPayloadTypeBytes in interface WarcMetadataOrBuilder
      Returns:
      The bytes for warcIdentifiedPayloadType.

    • hasHttpStatusCode

      public boolean hasHttpStatusCode()
       HTTP status code (200, 404, 500, etc.)
 HTTP status
      optional int32 http_status_code = 18 [json_name = "httpStatusCode"];
      Specified by:
      hasHttpStatusCode in interface WarcMetadataOrBuilder
      Returns:
      Whether the httpStatusCode field is set.

    • getHttpStatusCode

      public int getHttpStatusCode()
       HTTP status code (200, 404, 500, etc.)
 HTTP status
      optional int32 http_status_code = 18 [json_name = "httpStatusCode"];
      Specified by:
      getHttpStatusCode in interface WarcMetadataOrBuilder
      Returns:
      The httpStatusCode.

    • hasHttpStatusReason

      public boolean hasHttpStatusReason()
       HTTP status reason phrase ("OK", "Not Found", etc.)
 HTTP reason phrase
      optional string http_status_reason = 19 [json_name = "httpStatusReason"];
      Specified by:
      hasHttpStatusReason in interface WarcMetadataOrBuilder
      Returns:
      Whether the httpStatusReason field is set.

    • getHttpStatusReason

      public String getHttpStatusReason()
       HTTP status reason phrase ("OK", "Not Found", etc.)
 HTTP reason phrase
      optional string http_status_reason = 19 [json_name = "httpStatusReason"];
      Specified by:
      getHttpStatusReason in interface WarcMetadataOrBuilder
      Returns:
      The httpStatusReason.

    • getHttpStatusReasonBytes

      public com.google.protobuf.ByteString getHttpStatusReasonBytes()
       HTTP status reason phrase ("OK", "Not Found", etc.)
 HTTP reason phrase
      optional string http_status_reason = 19 [json_name = "httpStatusReason"];
      Specified by:
      getHttpStatusReasonBytes in interface WarcMetadataOrBuilder
      Returns:
      The bytes for httpStatusReason.

    • getHttpHeadersList

      public List<WarcHttpHeader> getHttpHeadersList()
       HTTP response headers
      repeated .ai.pipestream.parsed.data.warc.v1.WarcHttpHeader http_headers = 20 [json_name = "httpHeaders"];
      Specified by:
      getHttpHeadersList in interface WarcMetadataOrBuilder

    • getHttpHeadersOrBuilderList

      public List<? extends WarcHttpHeaderOrBuilder> getHttpHeadersOrBuilderList()
       HTTP response headers
      repeated .ai.pipestream.parsed.data.warc.v1.WarcHttpHeader http_headers = 20 [json_name = "httpHeaders"];
      Specified by:
      getHttpHeadersOrBuilderList in interface WarcMetadataOrBuilder

    • getHttpHeadersCount

      public int getHttpHeadersCount()
       HTTP response headers
      repeated .ai.pipestream.parsed.data.warc.v1.WarcHttpHeader http_headers = 20 [json_name = "httpHeaders"];
      Specified by:
      getHttpHeadersCount in interface WarcMetadataOrBuilder

    • getHttpHeaders

      public WarcHttpHeader getHttpHeaders(int index)
       HTTP response headers
      repeated .ai.pipestream.parsed.data.warc.v1.WarcHttpHeader http_headers = 20 [json_name = "httpHeaders"];
      Specified by:
      getHttpHeaders in interface WarcMetadataOrBuilder

    • getHttpHeadersOrBuilder

      public WarcHttpHeaderOrBuilder getHttpHeadersOrBuilder(int index)
       HTTP response headers
      repeated .ai.pipestream.parsed.data.warc.v1.WarcHttpHeader http_headers = 20 [json_name = "httpHeaders"];
      Specified by:
      getHttpHeadersOrBuilder in interface WarcMetadataOrBuilder

    • hasWarcCreatedBy

      public boolean hasWarcCreatedBy()
       Software/crawler that created this WARC file
 Software info
      optional string warc_created_by = 21 [json_name = "warcCreatedBy"];
      Specified by:
      hasWarcCreatedBy in interface WarcMetadataOrBuilder
      Returns:
      Whether the warcCreatedBy field is set.

    • getWarcCreatedBy

      public String getWarcCreatedBy()
       Software/crawler that created this WARC file
 Software info
      optional string warc_created_by = 21 [json_name = "warcCreatedBy"];
      Specified by:
      getWarcCreatedBy in interface WarcMetadataOrBuilder
      Returns:
      The warcCreatedBy.

    • getWarcCreatedByBytes

      public com.google.protobuf.ByteString getWarcCreatedByBytes()
       Software/crawler that created this WARC file
 Software info
      optional string warc_created_by = 21 [json_name = "warcCreatedBy"];
      Specified by:
      getWarcCreatedByBytes in interface WarcMetadataOrBuilder
      Returns:
      The bytes for warcCreatedBy.

    • hasWarcFormatVersion

      public boolean hasWarcFormatVersion()
       Version of the WARC format specification used
 WARC format version
      optional string warc_format_version = 22 [json_name = "warcFormatVersion"];
      Specified by:
      hasWarcFormatVersion in interface WarcMetadataOrBuilder
      Returns:
      Whether the warcFormatVersion field is set.

    • getWarcFormatVersion

      public String getWarcFormatVersion()
       Version of the WARC format specification used
 WARC format version
      optional string warc_format_version = 22 [json_name = "warcFormatVersion"];
      Specified by:
      getWarcFormatVersion in interface WarcMetadataOrBuilder
      Returns:
      The warcFormatVersion.

    • getWarcFormatVersionBytes

      public com.google.protobuf.ByteString getWarcFormatVersionBytes()
       Version of the WARC format specification used
 WARC format version
      optional string warc_format_version = 22 [json_name = "warcFormatVersion"];
      Specified by:
      getWarcFormatVersionBytes in interface WarcMetadataOrBuilder
      Returns:
      The bytes for warcFormatVersion.

    • hasWarcCollection

      public boolean hasWarcCollection()
       Crawl or collection identifier
 Collection name
      optional string warc_collection = 23 [json_name = "warcCollection"];
      Specified by:
      hasWarcCollection in interface WarcMetadataOrBuilder
      Returns:
      Whether the warcCollection field is set.

    • getWarcCollection

      public String getWarcCollection()
       Crawl or collection identifier
 Collection name
      optional string warc_collection = 23 [json_name = "warcCollection"];
      Specified by:
      getWarcCollection in interface WarcMetadataOrBuilder
      Returns:
      The warcCollection.

    • getWarcCollectionBytes

      public com.google.protobuf.ByteString getWarcCollectionBytes()
       Crawl or collection identifier
 Collection name
      optional string warc_collection = 23 [json_name = "warcCollection"];
      Specified by:
      getWarcCollectionBytes in interface WarcMetadataOrBuilder
      Returns:
      The bytes for warcCollection.

    • hasWarcCrawlId

      public boolean hasWarcCrawlId()
       Crawl job or session identifier
 Crawl identifier
      optional string warc_crawl_id = 24 [json_name = "warcCrawlId"];
      Specified by:
      hasWarcCrawlId in interface WarcMetadataOrBuilder
      Returns:
      Whether the warcCrawlId field is set.

    • getWarcCrawlId

      public String getWarcCrawlId()
       Crawl job or session identifier
 Crawl identifier
      optional string warc_crawl_id = 24 [json_name = "warcCrawlId"];
      Specified by:
      getWarcCrawlId in interface WarcMetadataOrBuilder
      Returns:
      The warcCrawlId.

    • getWarcCrawlIdBytes

      public com.google.protobuf.ByteString getWarcCrawlIdBytes()
       Crawl job or session identifier
 Crawl identifier
      optional string warc_crawl_id = 24 [json_name = "warcCrawlId"];
      Specified by:
      getWarcCrawlIdBytes in interface WarcMetadataOrBuilder
      Returns:
      The bytes for warcCrawlId.

    • hasWarcRobotPolicy

      public boolean hasWarcRobotPolicy()
       Robot policy information (robots.txt compliance)
 Robot policy
      optional string warc_robot_policy = 25 [json_name = "warcRobotPolicy"];
      Specified by:
      hasWarcRobotPolicy in interface WarcMetadataOrBuilder
      Returns:
      Whether the warcRobotPolicy field is set.

    • getWarcRobotPolicy

      public String getWarcRobotPolicy()
       Robot policy information (robots.txt compliance)
 Robot policy
      optional string warc_robot_policy = 25 [json_name = "warcRobotPolicy"];
      Specified by:
      getWarcRobotPolicy in interface WarcMetadataOrBuilder
      Returns:
      The warcRobotPolicy.

    • getWarcRobotPolicyBytes

      public com.google.protobuf.ByteString getWarcRobotPolicyBytes()
       Robot policy information (robots.txt compliance)
 Robot policy
      optional string warc_robot_policy = 25 [json_name = "warcRobotPolicy"];
      Specified by:
      getWarcRobotPolicyBytes in interface WarcMetadataOrBuilder
      Returns:
      The bytes for warcRobotPolicy.

    • hasFetchTimestamp

      public boolean hasFetchTimestamp()
       Fetch timestamp (may differ from WARC-Date)
      optional .google.protobuf.Timestamp fetch_timestamp = 26 [json_name = "fetchTimestamp"];
      Specified by:
      hasFetchTimestamp in interface WarcMetadataOrBuilder
      Returns:
      Whether the fetchTimestamp field is set.

    • getFetchTimestamp

      public com.google.protobuf.Timestamp getFetchTimestamp()
       Fetch timestamp (may differ from WARC-Date)
      optional .google.protobuf.Timestamp fetch_timestamp = 26 [json_name = "fetchTimestamp"];
      Specified by:
      getFetchTimestamp in interface WarcMetadataOrBuilder
      Returns:
      The fetchTimestamp.

    • getFetchTimestampOrBuilder

      public com.google.protobuf.TimestampOrBuilder getFetchTimestampOrBuilder()
       Fetch timestamp (may differ from WARC-Date)
      optional .google.protobuf.Timestamp fetch_timestamp = 26 [json_name = "fetchTimestamp"];
      Specified by:
      getFetchTimestampOrBuilder in interface WarcMetadataOrBuilder

    • hasContentLanguage

      public boolean hasContentLanguage()
       Detected language of the content
      optional string content_language = 27 [json_name = "contentLanguage"];
      Specified by:
      hasContentLanguage in interface WarcMetadataOrBuilder
      Returns:
      Whether the contentLanguage field is set.

    • getContentLanguage

      public String getContentLanguage()
       Detected language of the content
      optional string content_language = 27 [json_name = "contentLanguage"];
      Specified by:
      getContentLanguage in interface WarcMetadataOrBuilder
      Returns:
      The contentLanguage.

    • getContentLanguageBytes

      public com.google.protobuf.ByteString getContentLanguageBytes()
       Detected language of the content
      optional string content_language = 27 [json_name = "contentLanguage"];
      Specified by:
      getContentLanguageBytes in interface WarcMetadataOrBuilder
      Returns:
      The bytes for contentLanguage.

    • hasContentEncoding

      public boolean hasContentEncoding()
       Character encoding of the content
      optional string content_encoding = 28 [json_name = "contentEncoding"];
      Specified by:
      hasContentEncoding in interface WarcMetadataOrBuilder
      Returns:
      Whether the contentEncoding field is set.

    • getContentEncoding

      public String getContentEncoding()
       Character encoding of the content
      optional string content_encoding = 28 [json_name = "contentEncoding"];
      Specified by:
      getContentEncoding in interface WarcMetadataOrBuilder
      Returns:
      The contentEncoding.

    • getContentEncodingBytes

      public com.google.protobuf.ByteString getContentEncodingBytes()
       Character encoding of the content
      optional string content_encoding = 28 [json_name = "contentEncoding"];
      Specified by:
      getContentEncodingBytes in interface WarcMetadataOrBuilder
      Returns:
      The bytes for contentEncoding.

    • hasEstimatedWordCount

      public boolean hasEstimatedWordCount()
       Estimated word count of textual content
      optional int64 estimated_word_count = 29 [json_name = "estimatedWordCount"];
      Specified by:
      hasEstimatedWordCount in interface WarcMetadataOrBuilder
      Returns:
      Whether the estimatedWordCount field is set.

    • getEstimatedWordCount

      public long getEstimatedWordCount()
       Estimated word count of textual content
      optional int64 estimated_word_count = 29 [json_name = "estimatedWordCount"];
      Specified by:
      getEstimatedWordCount in interface WarcMetadataOrBuilder
      Returns:
      The estimatedWordCount.

    • hasEstimatedCharacterCount

      public boolean hasEstimatedCharacterCount()
       Estimated character count of textual content
      optional int64 estimated_character_count = 30 [json_name = "estimatedCharacterCount"];
      Specified by:
      hasEstimatedCharacterCount in interface WarcMetadataOrBuilder
      Returns:
      Whether the estimatedCharacterCount field is set.

    • getEstimatedCharacterCount

      public long getEstimatedCharacterCount()
       Estimated character count of textual content
      optional int64 estimated_character_count = 30 [json_name = "estimatedCharacterCount"];
      Specified by:
      getEstimatedCharacterCount in interface WarcMetadataOrBuilder
      Returns:
      The estimatedCharacterCount.

    • hasIsTruncated

      public boolean hasIsTruncated()
       Whether the content appears to be truncated
      optional bool is_truncated = 31 [json_name = "isTruncated"];
      Specified by:
      hasIsTruncated in interface WarcMetadataOrBuilder
      Returns:
      Whether the isTruncated field is set.

    • getIsTruncated

      public boolean getIsTruncated()
       Whether the content appears to be truncated
      optional bool is_truncated = 31 [json_name = "isTruncated"];
      Specified by:
      getIsTruncated in interface WarcMetadataOrBuilder
      Returns:
      The isTruncated.

    • hasCompressionMethod

      public boolean hasCompressionMethod()
       Compression method used (if any)
      optional string compression_method = 32 [json_name = "compressionMethod"];
      Specified by:
      hasCompressionMethod in interface WarcMetadataOrBuilder
      Returns:
      Whether the compressionMethod field is set.

    • getCompressionMethod

      public String getCompressionMethod()
       Compression method used (if any)
      optional string compression_method = 32 [json_name = "compressionMethod"];
      Specified by:
      getCompressionMethod in interface WarcMetadataOrBuilder
      Returns:
      The compressionMethod.

    • getCompressionMethodBytes

      public com.google.protobuf.ByteString getCompressionMethodBytes()
       Compression method used (if any)
      optional string compression_method = 32 [json_name = "compressionMethod"];
      Specified by:
      getCompressionMethodBytes in interface WarcMetadataOrBuilder
      Returns:
      The bytes for compressionMethod.

    • hasHttpMethod

      public boolean hasHttpMethod()
       HTTP method used for the request (GET, POST, etc.)
      optional string http_method = 33 [json_name = "httpMethod"];
      Specified by:
      hasHttpMethod in interface WarcMetadataOrBuilder
      Returns:
      Whether the httpMethod field is set.

    • getHttpMethod

      public String getHttpMethod()
       HTTP method used for the request (GET, POST, etc.)
      optional string http_method = 33 [json_name = "httpMethod"];
      Specified by:
      getHttpMethod in interface WarcMetadataOrBuilder
      Returns:
      The httpMethod.

    • getHttpMethodBytes

      public com.google.protobuf.ByteString getHttpMethodBytes()
       HTTP method used for the request (GET, POST, etc.)
      optional string http_method = 33 [json_name = "httpMethod"];
      Specified by:
      getHttpMethodBytes in interface WarcMetadataOrBuilder
      Returns:
      The bytes for httpMethod.

    • hasHttpVersion

      public boolean hasHttpVersion()
       HTTP version (1.0, 1.1, 2.0)
      optional string http_version = 34 [json_name = "httpVersion"];
      Specified by:
      hasHttpVersion in interface WarcMetadataOrBuilder
      Returns:
      Whether the httpVersion field is set.

    • getHttpVersion

      public String getHttpVersion()
       HTTP version (1.0, 1.1, 2.0)
      optional string http_version = 34 [json_name = "httpVersion"];
      Specified by:
      getHttpVersion in interface WarcMetadataOrBuilder
      Returns:
      The httpVersion.

    • getHttpVersionBytes

      public com.google.protobuf.ByteString getHttpVersionBytes()
       HTTP version (1.0, 1.1, 2.0)
      optional string http_version = 34 [json_name = "httpVersion"];
      Specified by:
      getHttpVersionBytes in interface WarcMetadataOrBuilder
      Returns:
      The bytes for httpVersion.

    • getRedirectChainList

      public List<WarcRedirect> getRedirectChainList()
       Redirect chain information
      repeated .ai.pipestream.parsed.data.warc.v1.WarcRedirect redirect_chain = 35 [json_name = "redirectChain"];
      Specified by:
      getRedirectChainList in interface WarcMetadataOrBuilder

    • getRedirectChainOrBuilderList

      public List<? extends WarcRedirectOrBuilder> getRedirectChainOrBuilderList()
       Redirect chain information
      repeated .ai.pipestream.parsed.data.warc.v1.WarcRedirect redirect_chain = 35 [json_name = "redirectChain"];
      Specified by:
      getRedirectChainOrBuilderList in interface WarcMetadataOrBuilder

    • getRedirectChainCount

      public int getRedirectChainCount()
       Redirect chain information
      repeated .ai.pipestream.parsed.data.warc.v1.WarcRedirect redirect_chain = 35 [json_name = "redirectChain"];
      Specified by:
      getRedirectChainCount in interface WarcMetadataOrBuilder

    • getRedirectChain

      public WarcRedirect getRedirectChain(int index)
       Redirect chain information
      repeated .ai.pipestream.parsed.data.warc.v1.WarcRedirect redirect_chain = 35 [json_name = "redirectChain"];
      Specified by:
      getRedirectChain in interface WarcMetadataOrBuilder

    • getRedirectChainOrBuilder

      public WarcRedirectOrBuilder getRedirectChainOrBuilder(int index)
       Redirect chain information
      repeated .ai.pipestream.parsed.data.warc.v1.WarcRedirect redirect_chain = 35 [json_name = "redirectChain"];
      Specified by:
      getRedirectChainOrBuilder in interface WarcMetadataOrBuilder

    • getCookiesList

      public List<WarcCookie> getCookiesList()
       Cookies associated with this request/response
      repeated .ai.pipestream.parsed.data.warc.v1.WarcCookie cookies = 36 [json_name = "cookies"];
      Specified by:
      getCookiesList in interface WarcMetadataOrBuilder

    • getCookiesOrBuilderList

      public List<? extends WarcCookieOrBuilder> getCookiesOrBuilderList()
       Cookies associated with this request/response
      repeated .ai.pipestream.parsed.data.warc.v1.WarcCookie cookies = 36 [json_name = "cookies"];
      Specified by:
      getCookiesOrBuilderList in interface WarcMetadataOrBuilder

    • getCookiesCount

      public int getCookiesCount()
       Cookies associated with this request/response
      repeated .ai.pipestream.parsed.data.warc.v1.WarcCookie cookies = 36 [json_name = "cookies"];
      Specified by:
      getCookiesCount in interface WarcMetadataOrBuilder

    • getCookies

      public WarcCookie getCookies(int index)
       Cookies associated with this request/response
      repeated .ai.pipestream.parsed.data.warc.v1.WarcCookie cookies = 36 [json_name = "cookies"];
      Specified by:
      getCookies in interface WarcMetadataOrBuilder

    • getCookiesOrBuilder

      public WarcCookieOrBuilder getCookiesOrBuilder(int index)
       Cookies associated with this request/response
      repeated .ai.pipestream.parsed.data.warc.v1.WarcCookie cookies = 36 [json_name = "cookies"];
      Specified by:
      getCookiesOrBuilder in interface WarcMetadataOrBuilder

    • hasReferrerUrl

      public boolean hasReferrerUrl()
       Referrer URL (if available)
      optional string referrer_url = 37 [json_name = "referrerUrl"];
      Specified by:
      hasReferrerUrl in interface WarcMetadataOrBuilder
      Returns:
      Whether the referrerUrl field is set.

    • getReferrerUrl

      public String getReferrerUrl()
       Referrer URL (if available)
      optional string referrer_url = 37 [json_name = "referrerUrl"];
      Specified by:
      getReferrerUrl in interface WarcMetadataOrBuilder
      Returns:
      The referrerUrl.

    • getReferrerUrlBytes

      public com.google.protobuf.ByteString getReferrerUrlBytes()
       Referrer URL (if available)
      optional string referrer_url = 37 [json_name = "referrerUrl"];
      Specified by:
      getReferrerUrlBytes in interface WarcMetadataOrBuilder
      Returns:
      The bytes for referrerUrl.

    • hasUserAgent

      public boolean hasUserAgent()
       User agent string used for the request
      optional string user_agent = 38 [json_name = "userAgent"];
      Specified by:
      hasUserAgent in interface WarcMetadataOrBuilder
      Returns:
      Whether the userAgent field is set.

    • getUserAgent

      public String getUserAgent()
       User agent string used for the request
      optional string user_agent = 38 [json_name = "userAgent"];
      Specified by:
      getUserAgent in interface WarcMetadataOrBuilder
      Returns:
      The userAgent.

    • getUserAgentBytes

      public com.google.protobuf.ByteString getUserAgentBytes()
       User agent string used for the request
      optional string user_agent = 38 [json_name = "userAgent"];
      Specified by:
      getUserAgentBytes in interface WarcMetadataOrBuilder
      Returns:
      The bytes for userAgent.

    • hasRecordPosition

      public boolean hasRecordPosition()
       Position of this record within the WARC file
      optional int64 record_position = 39 [json_name = "recordPosition"];
      Specified by:
      hasRecordPosition in interface WarcMetadataOrBuilder
      Returns:
      Whether the recordPosition field is set.

    • getRecordPosition

      public long getRecordPosition()
       Position of this record within the WARC file
      optional int64 record_position = 39 [json_name = "recordPosition"];
      Specified by:
      getRecordPosition in interface WarcMetadataOrBuilder
      Returns:
      The recordPosition.

    • hasRecordSize

      public boolean hasRecordSize()
       Size of the complete WARC record (headers + payload)
      optional int64 record_size = 40 [json_name = "recordSize"];
      Specified by:
      hasRecordSize in interface WarcMetadataOrBuilder
      Returns:
      Whether the recordSize field is set.

    • getRecordSize

      public long getRecordSize()
       Size of the complete WARC record (headers + payload)
      optional int64 record_size = 40 [json_name = "recordSize"];
      Specified by:
      getRecordSize in interface WarcMetadataOrBuilder
      Returns:
      The recordSize.

    • hasTotalRecordsInFile

      public boolean hasTotalRecordsInFile()
       Number of records in the WARC file
      optional int32 total_records_in_file = 41 [json_name = "totalRecordsInFile"];
      Specified by:
      hasTotalRecordsInFile in interface WarcMetadataOrBuilder
      Returns:
      Whether the totalRecordsInFile field is set.

    • getTotalRecordsInFile

      public int getTotalRecordsInFile()
       Number of records in the WARC file
      optional int32 total_records_in_file = 41 [json_name = "totalRecordsInFile"];
      Specified by:
      getTotalRecordsInFile in interface WarcMetadataOrBuilder
      Returns:
      The totalRecordsInFile.

    • hasRecordIndex

      public boolean hasRecordIndex()
       Index of this record within the file (0-based)
      optional int32 record_index = 42 [json_name = "recordIndex"];
      Specified by:
      hasRecordIndex in interface WarcMetadataOrBuilder
      Returns:
      Whether the recordIndex field is set.

    • getRecordIndex

      public int getRecordIndex()
       Index of this record within the file (0-based)
      optional int32 record_index = 42 [json_name = "recordIndex"];
      Specified by:
      getRecordIndex in interface WarcMetadataOrBuilder
      Returns:
      The recordIndex.

    • hasHasMetadataRecords

      public boolean hasHasMetadataRecords()
       Whether this record has associated metadata records
      optional bool has_metadata_records = 43 [json_name = "hasMetadataRecords"];
      Specified by:
      hasHasMetadataRecords in interface WarcMetadataOrBuilder
      Returns:
      Whether the hasMetadataRecords field is set.

    • getHasMetadataRecords

      public boolean getHasMetadataRecords()
       Whether this record has associated metadata records
      optional bool has_metadata_records = 43 [json_name = "hasMetadataRecords"];
      Specified by:
      getHasMetadataRecords in interface WarcMetadataOrBuilder
      Returns:
      The hasMetadataRecords.

    • hasHasRequestRecords

      public boolean hasHasRequestRecords()
       Whether this record has associated request records
      optional bool has_request_records = 44 [json_name = "hasRequestRecords"];
      Specified by:
      hasHasRequestRecords in interface WarcMetadataOrBuilder
      Returns:
      Whether the hasRequestRecords field is set.

    • getHasRequestRecords

      public boolean getHasRequestRecords()
       Whether this record has associated request records
      optional bool has_request_records = 44 [json_name = "hasRequestRecords"];
      Specified by:
      getHasRequestRecords in interface WarcMetadataOrBuilder
      Returns:
      The hasRequestRecords.

    • getValidationErrorsList

      public com.google.protobuf.ProtocolStringList getValidationErrorsList()
       Validation errors found in the WARC record
      repeated string validation_errors = 45 [json_name = "validationErrors"];
      Specified by:
      getValidationErrorsList in interface WarcMetadataOrBuilder
      Returns:
      A list containing the validationErrors.

    • getValidationErrorsCount

      public int getValidationErrorsCount()
       Validation errors found in the WARC record
      repeated string validation_errors = 45 [json_name = "validationErrors"];
      Specified by:
      getValidationErrorsCount in interface WarcMetadataOrBuilder
      Returns:
      The count of validationErrors.

    • getValidationErrors

      public String getValidationErrors(int index)
       Validation errors found in the WARC record
      repeated string validation_errors = 45 [json_name = "validationErrors"];
      Specified by:
      getValidationErrors in interface WarcMetadataOrBuilder
      Parameters:
      index - The index of the element to return.
      Returns:
      The validationErrors at the given index.

    • getValidationErrorsBytes

      public com.google.protobuf.ByteString getValidationErrorsBytes(int index)
       Validation errors found in the WARC record
      repeated string validation_errors = 45 [json_name = "validationErrors"];
      Specified by:
      getValidationErrorsBytes in interface WarcMetadataOrBuilder
      Parameters:
      index - The index of the value to return.
      Returns:
      The bytes of the validationErrors at the given index.

    • getValidationWarningsList

      public com.google.protobuf.ProtocolStringList getValidationWarningsList()
       Validation warnings found in the WARC record
      repeated string validation_warnings = 46 [json_name = "validationWarnings"];
      Specified by:
      getValidationWarningsList in interface WarcMetadataOrBuilder
      Returns:
      A list containing the validationWarnings.

    • getValidationWarningsCount

      public int getValidationWarningsCount()
       Validation warnings found in the WARC record
      repeated string validation_warnings = 46 [json_name = "validationWarnings"];
      Specified by:
      getValidationWarningsCount in interface WarcMetadataOrBuilder
      Returns:
      The count of validationWarnings.

    • getValidationWarnings

      public String getValidationWarnings(int index)
       Validation warnings found in the WARC record
      repeated string validation_warnings = 46 [json_name = "validationWarnings"];
      Specified by:
      getValidationWarnings in interface WarcMetadataOrBuilder
      Parameters:
      index - The index of the element to return.
      Returns:
      The validationWarnings at the given index.

    • getValidationWarningsBytes

      public com.google.protobuf.ByteString getValidationWarningsBytes(int index)
       Validation warnings found in the WARC record
      repeated string validation_warnings = 46 [json_name = "validationWarnings"];
      Specified by:
      getValidationWarningsBytes in interface WarcMetadataOrBuilder
      Parameters:
      index - The index of the value to return.
      Returns:
      The bytes of the validationWarnings at the given index.

    • hasCompletenessScore

      public boolean hasCompletenessScore()
       Completeness score (0.0 to 1.0) indicating how complete the capture is
      optional double completeness_score = 47 [json_name = "completenessScore"];
      Specified by:
      hasCompletenessScore in interface WarcMetadataOrBuilder
      Returns:
      Whether the completenessScore field is set.

    • getCompletenessScore

      public double getCompletenessScore()
       Completeness score (0.0 to 1.0) indicating how complete the capture is
      optional double completeness_score = 47 [json_name = "completenessScore"];
      Specified by:
      getCompletenessScore in interface WarcMetadataOrBuilder
      Returns:
      The completenessScore.

    • hasChecksumVerified

      public boolean hasChecksumVerified()
       Whether the content matches expected checksums
      optional bool checksum_verified = 48 [json_name = "checksumVerified"];
      Specified by:
      hasChecksumVerified in interface WarcMetadataOrBuilder
      Returns:
      Whether the checksumVerified field is set.

    • getChecksumVerified

      public boolean getChecksumVerified()
       Whether the content matches expected checksums
      optional bool checksum_verified = 48 [json_name = "checksumVerified"];
      Specified by:
      getChecksumVerified in interface WarcMetadataOrBuilder
      Returns:
      The checksumVerified.

    • hasArchiveInstitution

      public boolean hasArchiveInstitution()
       Institution or organization that created the archive
      optional string archive_institution = 49 [json_name = "archiveInstitution"];
      Specified by:
      hasArchiveInstitution in interface WarcMetadataOrBuilder
      Returns:
      Whether the archiveInstitution field is set.

    • getArchiveInstitution

      public String getArchiveInstitution()
       Institution or organization that created the archive
      optional string archive_institution = 49 [json_name = "archiveInstitution"];
      Specified by:
      getArchiveInstitution in interface WarcMetadataOrBuilder
      Returns:
      The archiveInstitution.

    • getArchiveInstitutionBytes

      public com.google.protobuf.ByteString getArchiveInstitutionBytes()
       Institution or organization that created the archive
      optional string archive_institution = 49 [json_name = "archiveInstitution"];
      Specified by:
      getArchiveInstitutionBytes in interface WarcMetadataOrBuilder
      Returns:
      The bytes for archiveInstitution.

    • hasArchiveCollectionName

      public boolean hasArchiveCollectionName()
       Archive collection or series name
      optional string archive_collection_name = 50 [json_name = "archiveCollectionName"];
      Specified by:
      hasArchiveCollectionName in interface WarcMetadataOrBuilder
      Returns:
      Whether the archiveCollectionName field is set.

    • getArchiveCollectionName

      public String getArchiveCollectionName()
       Archive collection or series name
      optional string archive_collection_name = 50 [json_name = "archiveCollectionName"];
      Specified by:
      getArchiveCollectionName in interface WarcMetadataOrBuilder
      Returns:
      The archiveCollectionName.

    • getArchiveCollectionNameBytes

      public com.google.protobuf.ByteString getArchiveCollectionNameBytes()
       Archive collection or series name
      optional string archive_collection_name = 50 [json_name = "archiveCollectionName"];
      Specified by:
      getArchiveCollectionNameBytes in interface WarcMetadataOrBuilder
      Returns:
      The bytes for archiveCollectionName.

    • hasAccessRights

      public boolean hasAccessRights()
       Access restrictions or rights information
      optional string access_rights = 51 [json_name = "accessRights"];
      Specified by:
      hasAccessRights in interface WarcMetadataOrBuilder
      Returns:
      Whether the accessRights field is set.

    • getAccessRights

      public String getAccessRights()
       Access restrictions or rights information
      optional string access_rights = 51 [json_name = "accessRights"];
      Specified by:
      getAccessRights in interface WarcMetadataOrBuilder
      Returns:
      The accessRights.

    • getAccessRightsBytes

      public com.google.protobuf.ByteString getAccessRightsBytes()
       Access restrictions or rights information
      optional string access_rights = 51 [json_name = "accessRights"];
      Specified by:
      getAccessRightsBytes in interface WarcMetadataOrBuilder
      Returns:
      The bytes for accessRights.

    • getPreservationActionsList

      public com.google.protobuf.ProtocolStringList getPreservationActionsList()
       Preservation actions taken on this record
      repeated string preservation_actions = 52 [json_name = "preservationActions"];
      Specified by:
      getPreservationActionsList in interface WarcMetadataOrBuilder
      Returns:
      A list containing the preservationActions.

    • getPreservationActionsCount

      public int getPreservationActionsCount()
       Preservation actions taken on this record
      repeated string preservation_actions = 52 [json_name = "preservationActions"];
      Specified by:
      getPreservationActionsCount in interface WarcMetadataOrBuilder
      Returns:
      The count of preservationActions.

    • getPreservationActions

      public String getPreservationActions(int index)
       Preservation actions taken on this record
      repeated string preservation_actions = 52 [json_name = "preservationActions"];
      Specified by:
      getPreservationActions in interface WarcMetadataOrBuilder
      Parameters:
      index - The index of the element to return.
      Returns:
      The preservationActions at the given index.

    • getPreservationActionsBytes

      public com.google.protobuf.ByteString getPreservationActionsBytes(int index)
       Preservation actions taken on this record
      repeated string preservation_actions = 52 [json_name = "preservationActions"];
      Specified by:
      getPreservationActionsBytes in interface WarcMetadataOrBuilder
      Parameters:
      index - The index of the value to return.
      Returns:
      The bytes of the preservationActions at the given index.

    • getMigrationHistoryList

      public List<WarcMigration> getMigrationHistoryList()
       Migration history if the record has been migrated
      repeated .ai.pipestream.parsed.data.warc.v1.WarcMigration migration_history = 53 [json_name = "migrationHistory"];
      Specified by:
      getMigrationHistoryList in interface WarcMetadataOrBuilder

    • getMigrationHistoryOrBuilderList

      public List<? extends WarcMigrationOrBuilder> getMigrationHistoryOrBuilderList()
       Migration history if the record has been migrated
      repeated .ai.pipestream.parsed.data.warc.v1.WarcMigration migration_history = 53 [json_name = "migrationHistory"];
      Specified by:
      getMigrationHistoryOrBuilderList in interface WarcMetadataOrBuilder

    • getMigrationHistoryCount

      public int getMigrationHistoryCount()
       Migration history if the record has been migrated
      repeated .ai.pipestream.parsed.data.warc.v1.WarcMigration migration_history = 53 [json_name = "migrationHistory"];
      Specified by:
      getMigrationHistoryCount in interface WarcMetadataOrBuilder

    • getMigrationHistory

      public WarcMigration getMigrationHistory(int index)
       Migration history if the record has been migrated
      repeated .ai.pipestream.parsed.data.warc.v1.WarcMigration migration_history = 53 [json_name = "migrationHistory"];
      Specified by:
      getMigrationHistory in interface WarcMetadataOrBuilder

    • getMigrationHistoryOrBuilder

      public WarcMigrationOrBuilder getMigrationHistoryOrBuilder(int index)
       Migration history if the record has been migrated
      repeated .ai.pipestream.parsed.data.warc.v1.WarcMigration migration_history = 53 [json_name = "migrationHistory"];
      Specified by:
      getMigrationHistoryOrBuilder in interface WarcMetadataOrBuilder

    • hasDetectedMimeType

      public boolean hasDetectedMimeType()
       MIME type as detected by content analysis
      optional string detected_mime_type = 54 [json_name = "detectedMimeType"];
      Specified by:
      hasDetectedMimeType in interface WarcMetadataOrBuilder
      Returns:
      Whether the detectedMimeType field is set.

    • getDetectedMimeType

      public String getDetectedMimeType()
       MIME type as detected by content analysis
      optional string detected_mime_type = 54 [json_name = "detectedMimeType"];
      Specified by:
      getDetectedMimeType in interface WarcMetadataOrBuilder
      Returns:
      The detectedMimeType.

    • getDetectedMimeTypeBytes

      public com.google.protobuf.ByteString getDetectedMimeTypeBytes()
       MIME type as detected by content analysis
      optional string detected_mime_type = 54 [json_name = "detectedMimeType"];
      Specified by:
      getDetectedMimeTypeBytes in interface WarcMetadataOrBuilder
      Returns:
      The bytes for detectedMimeType.

    • hasFormatVersion

      public boolean hasFormatVersion()
       File format version (if applicable)
      optional string format_version = 55 [json_name = "formatVersion"];
      Specified by:
      hasFormatVersion in interface WarcMetadataOrBuilder
      Returns:
      Whether the formatVersion field is set.

    • getFormatVersion

      public String getFormatVersion()
       File format version (if applicable)
      optional string format_version = 55 [json_name = "formatVersion"];
      Specified by:
      getFormatVersion in interface WarcMetadataOrBuilder
      Returns:
      The formatVersion.

    • getFormatVersionBytes

      public com.google.protobuf.ByteString getFormatVersionBytes()
       File format version (if applicable)
      optional string format_version = 55 [json_name = "formatVersion"];
      Specified by:
      getFormatVersionBytes in interface WarcMetadataOrBuilder
      Returns:
      The bytes for formatVersion.

    • getTechnicalPropertiesList

      public List<WarcTechnicalProperty> getTechnicalPropertiesList()
       Technical properties specific to the content type
      repeated .ai.pipestream.parsed.data.warc.v1.WarcTechnicalProperty technical_properties = 56 [json_name = "technicalProperties"];
      Specified by:
      getTechnicalPropertiesList in interface WarcMetadataOrBuilder

    • getTechnicalPropertiesOrBuilderList

      public List<? extends WarcTechnicalPropertyOrBuilder> getTechnicalPropertiesOrBuilderList()
       Technical properties specific to the content type
      repeated .ai.pipestream.parsed.data.warc.v1.WarcTechnicalProperty technical_properties = 56 [json_name = "technicalProperties"];
      Specified by:
      getTechnicalPropertiesOrBuilderList in interface WarcMetadataOrBuilder

    • getTechnicalPropertiesCount

      public int getTechnicalPropertiesCount()
       Technical properties specific to the content type
      repeated .ai.pipestream.parsed.data.warc.v1.WarcTechnicalProperty technical_properties = 56 [json_name = "technicalProperties"];
      Specified by:
      getTechnicalPropertiesCount in interface WarcMetadataOrBuilder

    • getTechnicalProperties

      public WarcTechnicalProperty getTechnicalProperties(int index)
       Technical properties specific to the content type
      repeated .ai.pipestream.parsed.data.warc.v1.WarcTechnicalProperty technical_properties = 56 [json_name = "technicalProperties"];
      Specified by:
      getTechnicalProperties in interface WarcMetadataOrBuilder

    • getTechnicalPropertiesOrBuilder

      public WarcTechnicalPropertyOrBuilder getTechnicalPropertiesOrBuilder(int index)
       Technical properties specific to the content type
      repeated .ai.pipestream.parsed.data.warc.v1.WarcTechnicalProperty technical_properties = 56 [json_name = "technicalProperties"];
      Specified by:
      getTechnicalPropertiesOrBuilder in interface WarcMetadataOrBuilder

    • getEmbeddedResourcesList

      public List<WarcEmbeddedResource> getEmbeddedResourcesList()
       Embedded resources found within this record
      repeated .ai.pipestream.parsed.data.warc.v1.WarcEmbeddedResource embedded_resources = 57 [json_name = "embeddedResources"];
      Specified by:
      getEmbeddedResourcesList in interface WarcMetadataOrBuilder

    • getEmbeddedResourcesOrBuilderList

      public List<? extends WarcEmbeddedResourceOrBuilder> getEmbeddedResourcesOrBuilderList()
       Embedded resources found within this record
      repeated .ai.pipestream.parsed.data.warc.v1.WarcEmbeddedResource embedded_resources = 57 [json_name = "embeddedResources"];
      Specified by:
      getEmbeddedResourcesOrBuilderList in interface WarcMetadataOrBuilder

    • getEmbeddedResourcesCount

      public int getEmbeddedResourcesCount()
       Embedded resources found within this record
      repeated .ai.pipestream.parsed.data.warc.v1.WarcEmbeddedResource embedded_resources = 57 [json_name = "embeddedResources"];
      Specified by:
      getEmbeddedResourcesCount in interface WarcMetadataOrBuilder

    • getEmbeddedResources

      public WarcEmbeddedResource getEmbeddedResources(int index)
       Embedded resources found within this record
      repeated .ai.pipestream.parsed.data.warc.v1.WarcEmbeddedResource embedded_resources = 57 [json_name = "embeddedResources"];
      Specified by:
      getEmbeddedResources in interface WarcMetadataOrBuilder

    • getEmbeddedResourcesOrBuilder

      public WarcEmbeddedResourceOrBuilder getEmbeddedResourcesOrBuilder(int index)
       Embedded resources found within this record
      repeated .ai.pipestream.parsed.data.warc.v1.WarcEmbeddedResource embedded_resources = 57 [json_name = "embeddedResources"];
      Specified by:
      getEmbeddedResourcesOrBuilder in interface WarcMetadataOrBuilder

    • hasAdditionalMetadata

      public boolean hasAdditionalMetadata()
       === Catch-all for unmapped WARC-specific metadata ===
      .google.protobuf.Struct additional_metadata = 200 [json_name = "additionalMetadata"];
      Specified by:
      hasAdditionalMetadata in interface WarcMetadataOrBuilder
      Returns:
      Whether the additionalMetadata field is set.

    • getAdditionalMetadata

      public com.google.protobuf.Struct getAdditionalMetadata()
       === Catch-all for unmapped WARC-specific metadata ===
      .google.protobuf.Struct additional_metadata = 200 [json_name = "additionalMetadata"];
      Specified by:
      getAdditionalMetadata in interface WarcMetadataOrBuilder
      Returns:
      The additionalMetadata.

    • getAdditionalMetadataOrBuilder

      public com.google.protobuf.StructOrBuilder getAdditionalMetadataOrBuilder()
       === Catch-all for unmapped WARC-specific metadata ===
      .google.protobuf.Struct additional_metadata = 200 [json_name = "additionalMetadata"];
      Specified by:
      getAdditionalMetadataOrBuilder in interface WarcMetadataOrBuilder

    • hasBaseFields

      public boolean hasBaseFields()
       Base Tika fields common to all document types
      .ai.pipestream.parsed.data.tika.base.v1.TikaBaseFields base_fields = 201 [json_name = "baseFields"];
      Specified by:
      hasBaseFields in interface WarcMetadataOrBuilder
      Returns:
      Whether the baseFields field is set.

    • getBaseFields

      public TikaBaseFields getBaseFields()
       Base Tika fields common to all document types
      .ai.pipestream.parsed.data.tika.base.v1.TikaBaseFields base_fields = 201 [json_name = "baseFields"];
      Specified by:
      getBaseFields in interface WarcMetadataOrBuilder
      Returns:
      The baseFields.

    • getBaseFieldsOrBuilder

      public TikaBaseFieldsOrBuilder getBaseFieldsOrBuilder()
       Base Tika fields common to all document types
      .ai.pipestream.parsed.data.tika.base.v1.TikaBaseFields base_fields = 201 [json_name = "baseFields"];
      Specified by:
      getBaseFieldsOrBuilder in interface WarcMetadataOrBuilder

    • isInitialized

      public final boolean isInitialized()
      Specified by:
      isInitialized in interface com.google.protobuf.MessageLiteOrBuilder
      Overrides:
      isInitialized in class com.google.protobuf.GeneratedMessage

    • writeTo

      public void writeTo(com.google.protobuf.CodedOutputStream output) throws IOException
      Specified by:
      writeTo in interface com.google.protobuf.MessageLite
      Overrides:
      writeTo in class com.google.protobuf.GeneratedMessage
      Throws:
      IOException

    • getSerializedSize

      public int getSerializedSize()
      Specified by:
      getSerializedSize in interface com.google.protobuf.MessageLite
      Overrides:
      getSerializedSize in class com.google.protobuf.GeneratedMessage

    • equals

      public boolean equals(Object obj)
      Specified by:
      equals in interface com.google.protobuf.Message
      Overrides:
      equals in class com.google.protobuf.AbstractMessage

    • hashCode

      public int hashCode()
      Specified by:
      hashCode in interface com.google.protobuf.Message
      Overrides:
      hashCode in class com.google.protobuf.AbstractMessage

    • parseFrom

      public static WarcMetadata parseFrom(ByteBuffer data) throws com.google.protobuf.InvalidProtocolBufferException
      Throws:
      com.google.protobuf.InvalidProtocolBufferException

    • parseFrom

      public static WarcMetadata parseFrom(ByteBuffer data, com.google.protobuf.ExtensionRegistryLite extensionRegistry) throws com.google.protobuf.InvalidProtocolBufferException
      Throws:
      com.google.protobuf.InvalidProtocolBufferException

    • parseFrom

      public static WarcMetadata parseFrom(com.google.protobuf.ByteString data) throws com.google.protobuf.InvalidProtocolBufferException
      Throws:
      com.google.protobuf.InvalidProtocolBufferException

    • parseFrom

      public static WarcMetadata parseFrom(com.google.protobuf.ByteString data, com.google.protobuf.ExtensionRegistryLite extensionRegistry) throws com.google.protobuf.InvalidProtocolBufferException
      Throws:
      com.google.protobuf.InvalidProtocolBufferException

    • parseFrom

      public static WarcMetadata parseFrom(byte[] data) throws com.google.protobuf.InvalidProtocolBufferException
      Throws:
      com.google.protobuf.InvalidProtocolBufferException

    • parseFrom

      public static WarcMetadata parseFrom(byte[] data, com.google.protobuf.ExtensionRegistryLite extensionRegistry) throws com.google.protobuf.InvalidProtocolBufferException
      Throws:
      com.google.protobuf.InvalidProtocolBufferException

    • parseFrom

      public static WarcMetadata parseFrom(InputStream input) throws IOException
      Throws:
      IOException

    • parseFrom

      public static WarcMetadata parseFrom(InputStream input, com.google.protobuf.ExtensionRegistryLite extensionRegistry) throws IOException
      Throws:
      IOException

    • parseDelimitedFrom

      public static WarcMetadata parseDelimitedFrom(InputStream input) throws IOException
      Throws:
      IOException

    • parseDelimitedFrom

      public static WarcMetadata parseDelimitedFrom(InputStream input, com.google.protobuf.ExtensionRegistryLite extensionRegistry) throws IOException
      Throws:
      IOException

    • parseFrom

      public static WarcMetadata parseFrom(com.google.protobuf.CodedInputStream input) throws IOException
      Throws:
      IOException

    • parseFrom

      public static WarcMetadata parseFrom(com.google.protobuf.CodedInputStream input, com.google.protobuf.ExtensionRegistryLite extensionRegistry) throws IOException
      Throws:
      IOException

    • newBuilderForType

      public WarcMetadata.Builder newBuilderForType()
      Specified by:
      newBuilderForType in interface com.google.protobuf.Message
      Specified by:
      newBuilderForType in interface com.google.protobuf.MessageLite

    • newBuilder

      public static WarcMetadata.Builder newBuilder()

    • newBuilder

      public static WarcMetadata.Builder newBuilder(WarcMetadata prototype)

    • toBuilder

      public WarcMetadata.Builder toBuilder()
      Specified by:
      toBuilder in interface com.google.protobuf.Message
      Specified by:
      toBuilder in interface com.google.protobuf.MessageLite

    • newBuilderForType

      protected WarcMetadata.Builder newBuilderForType(com.google.protobuf.AbstractMessage.BuilderParent parent)
      Overrides:
      newBuilderForType in class com.google.protobuf.AbstractMessage

    • getDefaultInstance

      public static WarcMetadata getDefaultInstance()

    • parser

      public static com.google.protobuf.Parser<WarcMetadata> parser()

    • getParserForType

      public com.google.protobuf.Parser<WarcMetadata> getParserForType()
      Specified by:
      getParserForType in interface com.google.protobuf.Message
      Specified by:
      getParserForType in interface com.google.protobuf.MessageLite
      Overrides:
      getParserForType in class com.google.protobuf.GeneratedMessage

    • getDefaultInstanceForType

      public WarcMetadata getDefaultInstanceForType()
      Specified by:
      getDefaultInstanceForType in interface com.google.protobuf.MessageLiteOrBuilder
      Specified by:
      getDefaultInstanceForType in interface com.google.protobuf.MessageOrBuilder