Package ai.pipestream.parsed.data.tika.v1

Interface TikaResponseOrBuilder

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

  • Method Details

    • getDocId

      String getDocId()
       Document identifier
      string doc_id = 1 [json_name = "docId"];
      Returns:
      The docId.

    • getDocIdBytes

      com.google.protobuf.ByteString getDocIdBytes()
       Document identifier
      string doc_id = 1 [json_name = "docId"];
      Returns:
      The bytes for docId.

    • getParseId

      String getParseId()
      string parse_id = 2 [json_name = "parseId"];
      Returns:
      The parseId.

    • getParseIdBytes

      com.google.protobuf.ByteString getParseIdBytes()
      string parse_id = 2 [json_name = "parseId"];
      Returns:
      The bytes for parseId.

    • hasContent

      boolean hasContent()
       Core content (always present if extraction succeeded)
      .ai.pipestream.parsed.data.tika.v1.TikaContent content = 3 [json_name = "content"];
      Returns:
      Whether the content field is set.

    • getContent

      TikaContent getContent()
       Core content (always present if extraction succeeded)
      .ai.pipestream.parsed.data.tika.v1.TikaContent content = 3 [json_name = "content"];
      Returns:
      The content.

    • getContentOrBuilder

      TikaContentOrBuilder getContentOrBuilder()
       Core content (always present if extraction succeeded)
      .ai.pipestream.parsed.data.tika.v1.TikaContent content = 3 [json_name = "content"];

    • hasDublinCore

      boolean hasDublinCore()
       Dublin Core metadata - standardized fields present in most documents
      .ai.pipestream.parsed.data.dublin.v1.DublinCoreMetadata dublin_core = 4 [json_name = "dublinCore"];
      Returns:
      Whether the dublinCore field is set.

    • getDublinCore

      DublinCoreMetadata getDublinCore()
       Dublin Core metadata - standardized fields present in most documents
      .ai.pipestream.parsed.data.dublin.v1.DublinCoreMetadata dublin_core = 4 [json_name = "dublinCore"];
      Returns:
      The dublinCore.

    • getDublinCoreOrBuilder

      DublinCoreMetadataOrBuilder getDublinCoreOrBuilder()
       Dublin Core metadata - standardized fields present in most documents
      .ai.pipestream.parsed.data.dublin.v1.DublinCoreMetadata dublin_core = 4 [json_name = "dublinCore"];

    • hasOffice

      boolean hasOffice()
      .ai.pipestream.parsed.data.office.v1.OfficeMetadata office = 5 [json_name = "office"];
      Returns:
      Whether the office field is set.

    • getOffice

      OfficeMetadata getOffice()
      .ai.pipestream.parsed.data.office.v1.OfficeMetadata office = 5 [json_name = "office"];
      Returns:
      The office.

    • getOfficeOrBuilder

      OfficeMetadataOrBuilder getOfficeOrBuilder()
      .ai.pipestream.parsed.data.office.v1.OfficeMetadata office = 5 [json_name = "office"];

    • hasPdf

      boolean hasPdf()
      .ai.pipestream.parsed.data.pdf.v1.PdfMetadata pdf = 7 [json_name = "pdf"];
      Returns:
      Whether the pdf field is set.

    • getPdf

      PdfMetadata getPdf()
      .ai.pipestream.parsed.data.pdf.v1.PdfMetadata pdf = 7 [json_name = "pdf"];
      Returns:
      The pdf.

    • getPdfOrBuilder

      PdfMetadataOrBuilder getPdfOrBuilder()
      .ai.pipestream.parsed.data.pdf.v1.PdfMetadata pdf = 7 [json_name = "pdf"];

    • hasImage

      boolean hasImage()
      .ai.pipestream.parsed.data.image.v1.ImageMetadata image = 8 [json_name = "image"];
      Returns:
      Whether the image field is set.

    • getImage

      ImageMetadata getImage()
      .ai.pipestream.parsed.data.image.v1.ImageMetadata image = 8 [json_name = "image"];
      Returns:
      The image.

    • getImageOrBuilder

      ImageMetadataOrBuilder getImageOrBuilder()
      .ai.pipestream.parsed.data.image.v1.ImageMetadata image = 8 [json_name = "image"];

    • hasEmail

      boolean hasEmail()
      .ai.pipestream.parsed.data.email.v1.EmailMetadata email = 9 [json_name = "email"];
      Returns:
      Whether the email field is set.

    • getEmail

      EmailMetadata getEmail()
      .ai.pipestream.parsed.data.email.v1.EmailMetadata email = 9 [json_name = "email"];
      Returns:
      The email.

    • getEmailOrBuilder

      EmailMetadataOrBuilder getEmailOrBuilder()
      .ai.pipestream.parsed.data.email.v1.EmailMetadata email = 9 [json_name = "email"];

    • hasMedia

      boolean hasMedia()
      .ai.pipestream.parsed.data.media.v1.MediaMetadata media = 10 [json_name = "media"];
      Returns:
      Whether the media field is set.

    • getMedia

      MediaMetadata getMedia()
      .ai.pipestream.parsed.data.media.v1.MediaMetadata media = 10 [json_name = "media"];
      Returns:
      The media.

    • getMediaOrBuilder

      MediaMetadataOrBuilder getMediaOrBuilder()
      .ai.pipestream.parsed.data.media.v1.MediaMetadata media = 10 [json_name = "media"];

    • hasHtml

      boolean hasHtml()
      .ai.pipestream.parsed.data.html.v1.HtmlMetadata html = 11 [json_name = "html"];
      Returns:
      Whether the html field is set.

    • getHtml

      HtmlMetadata getHtml()
      .ai.pipestream.parsed.data.html.v1.HtmlMetadata html = 11 [json_name = "html"];
      Returns:
      The html.

    • getHtmlOrBuilder

      HtmlMetadataOrBuilder getHtmlOrBuilder()
      .ai.pipestream.parsed.data.html.v1.HtmlMetadata html = 11 [json_name = "html"];

    • hasRtf

      boolean hasRtf()
      .ai.pipestream.parsed.data.rtf.v1.RtfMetadata rtf = 17 [json_name = "rtf"];
      Returns:
      Whether the rtf field is set.

    • getRtf

      RtfMetadata getRtf()
      .ai.pipestream.parsed.data.rtf.v1.RtfMetadata rtf = 17 [json_name = "rtf"];
      Returns:
      The rtf.

    • getRtfOrBuilder

      RtfMetadataOrBuilder getRtfOrBuilder()
      .ai.pipestream.parsed.data.rtf.v1.RtfMetadata rtf = 17 [json_name = "rtf"];

    • hasDatabase

      boolean hasDatabase()
      .ai.pipestream.parsed.data.database.v1.DatabaseMetadata database = 18 [json_name = "database"];
      Returns:
      Whether the database field is set.

    • getDatabase

      DatabaseMetadata getDatabase()
      .ai.pipestream.parsed.data.database.v1.DatabaseMetadata database = 18 [json_name = "database"];
      Returns:
      The database.

    • getDatabaseOrBuilder

      DatabaseMetadataOrBuilder getDatabaseOrBuilder()
      .ai.pipestream.parsed.data.database.v1.DatabaseMetadata database = 18 [json_name = "database"];

    • hasFont

      boolean hasFont()
      .ai.pipestream.parsed.data.tika.font.v1.FontMetadata font = 19 [json_name = "font"];
      Returns:
      Whether the font field is set.

    • getFont

      FontMetadata getFont()
      .ai.pipestream.parsed.data.tika.font.v1.FontMetadata font = 19 [json_name = "font"];
      Returns:
      The font.

    • getFontOrBuilder

      FontMetadataOrBuilder getFontOrBuilder()
      .ai.pipestream.parsed.data.tika.font.v1.FontMetadata font = 19 [json_name = "font"];

    • hasEpub

      boolean hasEpub()
      .ai.pipestream.parsed.data.epub.v1.EpubMetadata epub = 20 [json_name = "epub"];
      Returns:
      Whether the epub field is set.

    • getEpub

      EpubMetadata getEpub()
      .ai.pipestream.parsed.data.epub.v1.EpubMetadata epub = 20 [json_name = "epub"];
      Returns:
      The epub.

    • getEpubOrBuilder

      EpubMetadataOrBuilder getEpubOrBuilder()
      .ai.pipestream.parsed.data.epub.v1.EpubMetadata epub = 20 [json_name = "epub"];

    • hasWarc

      boolean hasWarc()
      .ai.pipestream.parsed.data.warc.v1.WarcMetadata warc = 21 [json_name = "warc"];
      Returns:
      Whether the warc field is set.

    • getWarc

      WarcMetadata getWarc()
      .ai.pipestream.parsed.data.warc.v1.WarcMetadata warc = 21 [json_name = "warc"];
      Returns:
      The warc.

    • getWarcOrBuilder

      WarcMetadataOrBuilder getWarcOrBuilder()
      .ai.pipestream.parsed.data.warc.v1.WarcMetadata warc = 21 [json_name = "warc"];

    • hasClimateForecast

      boolean hasClimateForecast()
      .ai.pipestream.parsed.data.climate.v1.ClimateForcastMetadata climate_forecast = 22 [json_name = "climateForecast"];
      Returns:
      Whether the climateForecast field is set.

    • getClimateForecast

      ClimateForcastMetadata getClimateForecast()
      .ai.pipestream.parsed.data.climate.v1.ClimateForcastMetadata climate_forecast = 22 [json_name = "climateForecast"];
      Returns:
      The climateForecast.

    • getClimateForecastOrBuilder

      ClimateForcastMetadataOrBuilder getClimateForecastOrBuilder()
      .ai.pipestream.parsed.data.climate.v1.ClimateForcastMetadata climate_forecast = 22 [json_name = "climateForecast"];

    • hasCreativeCommons

      boolean hasCreativeCommons()
      .ai.pipestream.parsed.data.creative_commons.v1.CreativeCommonsMetadata creative_commons = 23 [json_name = "creativeCommons"];
      Returns:
      Whether the creativeCommons field is set.

    • getCreativeCommons

      CreativeCommonsMetadata getCreativeCommons()
      .ai.pipestream.parsed.data.creative_commons.v1.CreativeCommonsMetadata creative_commons = 23 [json_name = "creativeCommons"];
      Returns:
      The creativeCommons.

    • getCreativeCommonsOrBuilder

      CreativeCommonsMetadataOrBuilder getCreativeCommonsOrBuilder()
      .ai.pipestream.parsed.data.creative_commons.v1.CreativeCommonsMetadata creative_commons = 23 [json_name = "creativeCommons"];

    • hasGeneric

      boolean hasGeneric()
      .ai.pipestream.parsed.data.generic.v1.GenericMetadata generic = 24 [json_name = "generic"];
      Returns:
      Whether the generic field is set.

    • getGeneric

      GenericMetadata getGeneric()
      .ai.pipestream.parsed.data.generic.v1.GenericMetadata generic = 24 [json_name = "generic"];
      Returns:
      The generic.

    • getGenericOrBuilder

      GenericMetadataOrBuilder getGenericOrBuilder()
      .ai.pipestream.parsed.data.generic.v1.GenericMetadata generic = 24 [json_name = "generic"];

    • getMetadataList

      List<TikaMetadataEntry> getMetadataList()
       Additional metadata as typed entries (for less common fields)
      repeated .ai.pipestream.parsed.data.tika.v1.TikaMetadataEntry metadata = 12 [json_name = "metadata"];

    • getMetadata

      TikaMetadataEntry getMetadata(int index)
       Additional metadata as typed entries (for less common fields)
      repeated .ai.pipestream.parsed.data.tika.v1.TikaMetadataEntry metadata = 12 [json_name = "metadata"];

    • getMetadataCount

      int getMetadataCount()
       Additional metadata as typed entries (for less common fields)
      repeated .ai.pipestream.parsed.data.tika.v1.TikaMetadataEntry metadata = 12 [json_name = "metadata"];

    • getMetadataOrBuilderList

      List<? extends TikaMetadataEntryOrBuilder> getMetadataOrBuilderList()
       Additional metadata as typed entries (for less common fields)
      repeated .ai.pipestream.parsed.data.tika.v1.TikaMetadataEntry metadata = 12 [json_name = "metadata"];

    • getMetadataOrBuilder

      TikaMetadataEntryOrBuilder getMetadataOrBuilder(int index)
       Additional metadata as typed entries (for less common fields)
      repeated .ai.pipestream.parsed.data.tika.v1.TikaMetadataEntry metadata = 12 [json_name = "metadata"];

    • hasStatus

      boolean hasStatus()
       Parse status and statistics
      .ai.pipestream.parsed.data.tika.v1.TikaParseStatus status = 13 [json_name = "status"];
      Returns:
      Whether the status field is set.

    • getStatus

      TikaParseStatus getStatus()
       Parse status and statistics
      .ai.pipestream.parsed.data.tika.v1.TikaParseStatus status = 13 [json_name = "status"];
      Returns:
      The status.

    • getStatusOrBuilder

      TikaParseStatusOrBuilder getStatusOrBuilder()
       Parse status and statistics
      .ai.pipestream.parsed.data.tika.v1.TikaParseStatus status = 13 [json_name = "status"];

    • getRawPropertiesCount

      int getRawPropertiesCount()
       Raw key-value pairs for anything not captured in typed metadata
      map<string, string> raw_properties = 14 [json_name = "rawProperties"];

    • containsRawProperties

      boolean containsRawProperties(String key)
       Raw key-value pairs for anything not captured in typed metadata
      map<string, string> raw_properties = 14 [json_name = "rawProperties"];

    • getRawProperties

      @Deprecated Map<String,String> getRawProperties()
      Deprecated.
      Use getRawPropertiesMap() instead.

    • getRawPropertiesMap

      Map<String,String> getRawPropertiesMap()
       Raw key-value pairs for anything not captured in typed metadata
      map<string, string> raw_properties = 14 [json_name = "rawProperties"];

    • getRawPropertiesOrDefault

      String getRawPropertiesOrDefault(String key, String defaultValue)
       Raw key-value pairs for anything not captured in typed metadata
      map<string, string> raw_properties = 14 [json_name = "rawProperties"];

    • getRawPropertiesOrThrow

      String getRawPropertiesOrThrow(String key)
       Raw key-value pairs for anything not captured in typed metadata
      map<string, string> raw_properties = 14 [json_name = "rawProperties"];

    • getEmbeddedDocsList

      List<TikaEmbeddedDoc> getEmbeddedDocsList()
       Embedded documents if found
      repeated .ai.pipestream.parsed.data.tika.v1.TikaEmbeddedDoc embedded_docs = 15 [json_name = "embeddedDocs"];

    • getEmbeddedDocs

      TikaEmbeddedDoc getEmbeddedDocs(int index)
       Embedded documents if found
      repeated .ai.pipestream.parsed.data.tika.v1.TikaEmbeddedDoc embedded_docs = 15 [json_name = "embeddedDocs"];

    • getEmbeddedDocsCount

      int getEmbeddedDocsCount()
       Embedded documents if found
      repeated .ai.pipestream.parsed.data.tika.v1.TikaEmbeddedDoc embedded_docs = 15 [json_name = "embeddedDocs"];

    • getEmbeddedDocsOrBuilderList

      List<? extends TikaEmbeddedDocOrBuilder> getEmbeddedDocsOrBuilderList()
       Embedded documents if found
      repeated .ai.pipestream.parsed.data.tika.v1.TikaEmbeddedDoc embedded_docs = 15 [json_name = "embeddedDocs"];

    • getEmbeddedDocsOrBuilder

      TikaEmbeddedDocOrBuilder getEmbeddedDocsOrBuilder(int index)
       Embedded documents if found
      repeated .ai.pipestream.parsed.data.tika.v1.TikaEmbeddedDoc embedded_docs = 15 [json_name = "embeddedDocs"];

    • hasParsedAt

      boolean hasParsedAt()
      .google.protobuf.Timestamp parsed_at = 16 [json_name = "parsedAt"];
      Returns:
      Whether the parsedAt field is set.

    • getParsedAt

      com.google.protobuf.Timestamp getParsedAt()
      .google.protobuf.Timestamp parsed_at = 16 [json_name = "parsedAt"];
      Returns:
      The parsedAt.

    • getParsedAtOrBuilder

      com.google.protobuf.TimestampOrBuilder getParsedAtOrBuilder()
      .google.protobuf.Timestamp parsed_at = 16 [json_name = "parsedAt"];

    • getDocumentMetadataCase

      TikaResponse.DocumentMetadataCase getDocumentMetadataCase()