Package ai.pipestream.parsed.data.generic.v1

Interface GenericMetadataOrBuilder

All Superinterfaces:
com.google.protobuf.MessageLiteOrBuilder, com.google.protobuf.MessageOrBuilder
All Known Implementing Classes:
GenericMetadata, GenericMetadata.Builder
@Generated public interface GenericMetadataOrBuilder extends com.google.protobuf.MessageOrBuilder

  • Method Details

    • hasDocumentFormat

      boolean hasDocumentFormat()
       Detected document format or type
 e.g., "WordPerfect Document", "Quattro Pro Spreadsheet", "PST Mail Archive"
      optional string document_format = 1 [json_name = "documentFormat"];
      Returns:
      Whether the documentFormat field is set.

    • getDocumentFormat

      String getDocumentFormat()
       Detected document format or type
 e.g., "WordPerfect Document", "Quattro Pro Spreadsheet", "PST Mail Archive"
      optional string document_format = 1 [json_name = "documentFormat"];
      Returns:
      The documentFormat.

    • getDocumentFormatBytes

      com.google.protobuf.ByteString getDocumentFormatBytes()
       Detected document format or type
 e.g., "WordPerfect Document", "Quattro Pro Spreadsheet", "PST Mail Archive"
      optional string document_format = 1 [json_name = "documentFormat"];
      Returns:
      The bytes for documentFormat.

    • hasFormatVersion

      boolean hasFormatVersion()
       Format version (if detectable)
 e.g., "5.1", "12.0", "2003"
      optional string format_version = 2 [json_name = "formatVersion"];
      Returns:
      Whether the formatVersion field is set.

    • getFormatVersion

      String getFormatVersion()
       Format version (if detectable)
 e.g., "5.1", "12.0", "2003"
      optional string format_version = 2 [json_name = "formatVersion"];
      Returns:
      The formatVersion.

    • getFormatVersionBytes

      com.google.protobuf.ByteString getFormatVersionBytes()
       Format version (if detectable)
 e.g., "5.1", "12.0", "2003"
      optional string format_version = 2 [json_name = "formatVersion"];
      Returns:
      The bytes for formatVersion.

    • hasDetectedMimeType

      boolean hasDetectedMimeType()
       MIME type as detected by Tika
      optional string detected_mime_type = 3 [json_name = "detectedMimeType"];
      Returns:
      Whether the detectedMimeType field is set.

    • getDetectedMimeType

      String getDetectedMimeType()
       MIME type as detected by Tika
      optional string detected_mime_type = 3 [json_name = "detectedMimeType"];
      Returns:
      The detectedMimeType.

    • getDetectedMimeTypeBytes

      com.google.protobuf.ByteString getDetectedMimeTypeBytes()
       MIME type as detected by Tika
      optional string detected_mime_type = 3 [json_name = "detectedMimeType"];
      Returns:
      The bytes for detectedMimeType.

    • hasFileExtension

      boolean hasFileExtension()
       File extension (if available)
      optional string file_extension = 4 [json_name = "fileExtension"];
      Returns:
      Whether the fileExtension field is set.

    • getFileExtension

      String getFileExtension()
       File extension (if available)
      optional string file_extension = 4 [json_name = "fileExtension"];
      Returns:
      The fileExtension.

    • getFileExtensionBytes

      com.google.protobuf.ByteString getFileExtensionBytes()
       File extension (if available)
      optional string file_extension = 4 [json_name = "fileExtension"];
      Returns:
      The bytes for fileExtension.

    • hasCreatingApplication

      boolean hasCreatingApplication()
       Application that created the document (if detectable)
      optional string creating_application = 5 [json_name = "creatingApplication"];
      Returns:
      Whether the creatingApplication field is set.

    • getCreatingApplication

      String getCreatingApplication()
       Application that created the document (if detectable)
      optional string creating_application = 5 [json_name = "creatingApplication"];
      Returns:
      The creatingApplication.

    • getCreatingApplicationBytes

      com.google.protobuf.ByteString getCreatingApplicationBytes()
       Application that created the document (if detectable)
      optional string creating_application = 5 [json_name = "creatingApplication"];
      Returns:
      The bytes for creatingApplication.

    • hasApplicationVersion

      boolean hasApplicationVersion()
       Application version
      optional string application_version = 6 [json_name = "applicationVersion"];
      Returns:
      Whether the applicationVersion field is set.

    • getApplicationVersion

      String getApplicationVersion()
       Application version
      optional string application_version = 6 [json_name = "applicationVersion"];
      Returns:
      The applicationVersion.

    • getApplicationVersionBytes

      com.google.protobuf.ByteString getApplicationVersionBytes()
       Application version
      optional string application_version = 6 [json_name = "applicationVersion"];
      Returns:
      The bytes for applicationVersion.

    • hasContentLength

      boolean hasContentLength()
       Estimated content length in characters
      optional int64 content_length = 7 [json_name = "contentLength"];
      Returns:
      Whether the contentLength field is set.

    • getContentLength

      long getContentLength()
       Estimated content length in characters
      optional int64 content_length = 7 [json_name = "contentLength"];
      Returns:
      The contentLength.

    • hasContentLanguage

      boolean hasContentLanguage()
       Detected content language
      optional string content_language = 8 [json_name = "contentLanguage"];
      Returns:
      Whether the contentLanguage field is set.

    • getContentLanguage

      String getContentLanguage()
       Detected content language
      optional string content_language = 8 [json_name = "contentLanguage"];
      Returns:
      The contentLanguage.

    • getContentLanguageBytes

      com.google.protobuf.ByteString getContentLanguageBytes()
       Detected content language
      optional string content_language = 8 [json_name = "contentLanguage"];
      Returns:
      The bytes for contentLanguage.

    • hasContentEncoding

      boolean hasContentEncoding()
       Content encoding (if applicable)
      optional string content_encoding = 9 [json_name = "contentEncoding"];
      Returns:
      Whether the contentEncoding field is set.

    • getContentEncoding

      String getContentEncoding()
       Content encoding (if applicable)
      optional string content_encoding = 9 [json_name = "contentEncoding"];
      Returns:
      The contentEncoding.

    • getContentEncodingBytes

      com.google.protobuf.ByteString getContentEncodingBytes()
       Content encoding (if applicable)
      optional string content_encoding = 9 [json_name = "contentEncoding"];
      Returns:
      The bytes for contentEncoding.

    • hasIsEncrypted

      boolean hasIsEncrypted()
       Whether the document appears to be encrypted or password-protected
      optional bool is_encrypted = 10 [json_name = "isEncrypted"];
      Returns:
      Whether the isEncrypted field is set.

    • getIsEncrypted

      boolean getIsEncrypted()
       Whether the document appears to be encrypted or password-protected
      optional bool is_encrypted = 10 [json_name = "isEncrypted"];
      Returns:
      The isEncrypted.

    • hasHasEmbeddedContent

      boolean hasHasEmbeddedContent()
       Whether the document contains embedded objects or attachments
      optional bool has_embedded_content = 11 [json_name = "hasEmbeddedContent"];
      Returns:
      Whether the hasEmbeddedContent field is set.

    • getHasEmbeddedContent

      boolean getHasEmbeddedContent()
       Whether the document contains embedded objects or attachments
      optional bool has_embedded_content = 11 [json_name = "hasEmbeddedContent"];
      Returns:
      The hasEmbeddedContent.

    • hasFileSize

      boolean hasFileSize()
       File size in bytes (if available)
      optional int64 file_size = 12 [json_name = "fileSize"];
      Returns:
      Whether the fileSize field is set.

    • getFileSize

      long getFileSize()
       File size in bytes (if available)
      optional int64 file_size = 12 [json_name = "fileSize"];
      Returns:
      The fileSize.

    • hasCompressionMethod

      boolean hasCompressionMethod()
       Compression method used (if any)
      optional string compression_method = 13 [json_name = "compressionMethod"];
      Returns:
      Whether the compressionMethod field is set.

    • getCompressionMethod

      String getCompressionMethod()
       Compression method used (if any)
      optional string compression_method = 13 [json_name = "compressionMethod"];
      Returns:
      The compressionMethod.

    • getCompressionMethodBytes

      com.google.protobuf.ByteString getCompressionMethodBytes()
       Compression method used (if any)
      optional string compression_method = 13 [json_name = "compressionMethod"];
      Returns:
      The bytes for compressionMethod.

    • hasContentChecksum

      boolean hasContentChecksum()
       Checksum or hash of the content (if computed)
      optional string content_checksum = 14 [json_name = "contentChecksum"];
      Returns:
      Whether the contentChecksum field is set.

    • getContentChecksum

      String getContentChecksum()
       Checksum or hash of the content (if computed)
      optional string content_checksum = 14 [json_name = "contentChecksum"];
      Returns:
      The contentChecksum.

    • getContentChecksumBytes

      com.google.protobuf.ByteString getContentChecksumBytes()
       Checksum or hash of the content (if computed)
      optional string content_checksum = 14 [json_name = "contentChecksum"];
      Returns:
      The bytes for contentChecksum.

    • hasChecksumAlgorithm

      boolean hasChecksumAlgorithm()
       Checksum algorithm used
      optional string checksum_algorithm = 15 [json_name = "checksumAlgorithm"];
      Returns:
      Whether the checksumAlgorithm field is set.

    • getChecksumAlgorithm

      String getChecksumAlgorithm()
       Checksum algorithm used
      optional string checksum_algorithm = 15 [json_name = "checksumAlgorithm"];
      Returns:
      The checksumAlgorithm.

    • getChecksumAlgorithmBytes

      com.google.protobuf.ByteString getChecksumAlgorithmBytes()
       Checksum algorithm used
      optional string checksum_algorithm = 15 [json_name = "checksumAlgorithm"];
      Returns:
      The bytes for checksumAlgorithm.

    • hasTikaParserClass

      boolean hasTikaParserClass()
       Specific Tika parser that handled this document
      optional string tika_parser_class = 16 [json_name = "tikaParserClass"];
      Returns:
      Whether the tikaParserClass field is set.

    • getTikaParserClass

      String getTikaParserClass()
       Specific Tika parser that handled this document
      optional string tika_parser_class = 16 [json_name = "tikaParserClass"];
      Returns:
      The tikaParserClass.

    • getTikaParserClassBytes

      com.google.protobuf.ByteString getTikaParserClassBytes()
       Specific Tika parser that handled this document
      optional string tika_parser_class = 16 [json_name = "tikaParserClass"];
      Returns:
      The bytes for tikaParserClass.

    • getParserOptionsList

      List<String> getParserOptionsList()
       Parser configuration or options used
      repeated string parser_options = 17 [json_name = "parserOptions"];
      Returns:
      A list containing the parserOptions.

    • getParserOptionsCount

      int getParserOptionsCount()
       Parser configuration or options used
      repeated string parser_options = 17 [json_name = "parserOptions"];
      Returns:
      The count of parserOptions.

    • getParserOptions

      String getParserOptions(int index)
       Parser configuration or options used
      repeated string parser_options = 17 [json_name = "parserOptions"];
      Parameters:
      index - The index of the element to return.
      Returns:
      The parserOptions at the given index.

    • getParserOptionsBytes

      com.google.protobuf.ByteString getParserOptionsBytes(int index)
       Parser configuration or options used
      repeated string parser_options = 17 [json_name = "parserOptions"];
      Parameters:
      index - The index of the value to return.
      Returns:
      The bytes of the parserOptions at the given index.

    • hasParseStatus

      boolean hasParseStatus()
       Whether parsing was successful or partial
      optional string parse_status = 18 [json_name = "parseStatus"];
      Returns:
      Whether the parseStatus field is set.

    • getParseStatus

      String getParseStatus()
       Whether parsing was successful or partial
      optional string parse_status = 18 [json_name = "parseStatus"];
      Returns:
      The parseStatus.

    • getParseStatusBytes

      com.google.protobuf.ByteString getParseStatusBytes()
       Whether parsing was successful or partial
      optional string parse_status = 18 [json_name = "parseStatus"];
      Returns:
      The bytes for parseStatus.

    • getParserMessagesList

      List<String> getParserMessagesList()
       Any parser-specific warnings or messages
      repeated string parser_messages = 19 [json_name = "parserMessages"];
      Returns:
      A list containing the parserMessages.

    • getParserMessagesCount

      int getParserMessagesCount()
       Any parser-specific warnings or messages
      repeated string parser_messages = 19 [json_name = "parserMessages"];
      Returns:
      The count of parserMessages.

    • getParserMessages

      String getParserMessages(int index)
       Any parser-specific warnings or messages
      repeated string parser_messages = 19 [json_name = "parserMessages"];
      Parameters:
      index - The index of the element to return.
      Returns:
      The parserMessages at the given index.

    • getParserMessagesBytes

      com.google.protobuf.ByteString getParserMessagesBytes(int index)
       Any parser-specific warnings or messages
      repeated string parser_messages = 19 [json_name = "parserMessages"];
      Parameters:
      index - The index of the value to return.
      Returns:
      The bytes of the parserMessages at the given index.

    • hasAllMetadata

      boolean hasAllMetadata()
       All metadata extracted by Tika as key-value pairs
 This is the primary extensibility mechanism - any metadata not covered
 by the strongly-typed fields above will be captured here
      .google.protobuf.Struct all_metadata = 20 [json_name = "allMetadata"];
      Returns:
      Whether the allMetadata field is set.

    • getAllMetadata

      com.google.protobuf.Struct getAllMetadata()
       All metadata extracted by Tika as key-value pairs
 This is the primary extensibility mechanism - any metadata not covered
 by the strongly-typed fields above will be captured here
      .google.protobuf.Struct all_metadata = 20 [json_name = "allMetadata"];
      Returns:
      The allMetadata.

    • getAllMetadataOrBuilder

      com.google.protobuf.StructOrBuilder getAllMetadataOrBuilder()
       All metadata extracted by Tika as key-value pairs
 This is the primary extensibility mechanism - any metadata not covered
 by the strongly-typed fields above will be captured here
      .google.protobuf.Struct all_metadata = 20 [json_name = "allMetadata"];

    • hasDocumentProperties

      boolean hasDocumentProperties()
       Document properties (for formats that have property-based metadata)
      .google.protobuf.Struct document_properties = 21 [json_name = "documentProperties"];
      Returns:
      Whether the documentProperties field is set.

    • getDocumentProperties

      com.google.protobuf.Struct getDocumentProperties()
       Document properties (for formats that have property-based metadata)
      .google.protobuf.Struct document_properties = 21 [json_name = "documentProperties"];
      Returns:
      The documentProperties.

    • getDocumentPropertiesOrBuilder

      com.google.protobuf.StructOrBuilder getDocumentPropertiesOrBuilder()
       Document properties (for formats that have property-based metadata)
      .google.protobuf.Struct document_properties = 21 [json_name = "documentProperties"];

    • hasCustomMetadata

      boolean hasCustomMetadata()
       Custom metadata fields (user-defined or application-specific)
      .google.protobuf.Struct custom_metadata = 22 [json_name = "customMetadata"];
      Returns:
      Whether the customMetadata field is set.

    • getCustomMetadata

      com.google.protobuf.Struct getCustomMetadata()
       Custom metadata fields (user-defined or application-specific)
      .google.protobuf.Struct custom_metadata = 22 [json_name = "customMetadata"];
      Returns:
      The customMetadata.

    • getCustomMetadataOrBuilder

      com.google.protobuf.StructOrBuilder getCustomMetadataOrBuilder()
       Custom metadata fields (user-defined or application-specific)
      .google.protobuf.Struct custom_metadata = 22 [json_name = "customMetadata"];

    • hasTechnicalProperties

      boolean hasTechnicalProperties()
       Technical properties specific to the file format
      .google.protobuf.Struct technical_properties = 23 [json_name = "technicalProperties"];
      Returns:
      Whether the technicalProperties field is set.

    • getTechnicalProperties

      com.google.protobuf.Struct getTechnicalProperties()
       Technical properties specific to the file format
      .google.protobuf.Struct technical_properties = 23 [json_name = "technicalProperties"];
      Returns:
      The technicalProperties.

    • getTechnicalPropertiesOrBuilder

      com.google.protobuf.StructOrBuilder getTechnicalPropertiesOrBuilder()
       Technical properties specific to the file format
      .google.protobuf.Struct technical_properties = 23 [json_name = "technicalProperties"];

    • hasBaseFields

      boolean hasBaseFields()
       Base Tika fields common to all document types
      .ai.pipestream.parsed.data.tika.base.v1.TikaBaseFields base_fields = 100 [json_name = "baseFields"];
      Returns:
      Whether the baseFields field is set.

    • getBaseFields

      TikaBaseFields getBaseFields()
       Base Tika fields common to all document types
      .ai.pipestream.parsed.data.tika.base.v1.TikaBaseFields base_fields = 100 [json_name = "baseFields"];
      Returns:
      The baseFields.

    • getBaseFieldsOrBuilder

      TikaBaseFieldsOrBuilder getBaseFieldsOrBuilder()
       Base Tika fields common to all document types
      .ai.pipestream.parsed.data.tika.base.v1.TikaBaseFields base_fields = 100 [json_name = "baseFields"];