Package org.apache.tika.server.core.resource

Class PipesParsingHelper

java.lang.Object
org.apache.tika.server.core.resource.PipesParsingHelper
public class PipesParsingHelper extends Object
Helper class for pipes-based parsing in tika-server endpoints. Handles temp file management, FetchEmitTuple creation, and result processing.

The helper manages a dedicated temp directory for input files. A file-system-fetcher is configured with basePath pointing to this directory, ensuring child processes can only access files within the designated temp directory (no absolute paths).

  • Field Details

    • DEFAULT_FETCHER_ID

      public static final String DEFAULT_FETCHER_ID
      The fetcher ID used for reading temp files. This fetcher is configured with basePath = inputTempDirectory.
      See Also:

    • UNPACK_EMITTER_ID

      public static final String UNPACK_EMITTER_ID
      Name of the file-system emitter used for UNPACK mode. This emitter must be configured in tika-config.json with a basePath pointing to a writable temp directory.
      See Also:

  • Constructor Details

    • PipesParsingHelper

      public PipesParsingHelper(PipesParser pipesParser, PipesConfig pipesConfig, Path inputTempDirectory, Path unpackEmitterBasePath)
      Creates a PipesParsingHelper.
      Parameters:
      pipesParser - the PipesParser instance
      pipesConfig - the PipesConfig instance
      inputTempDirectory - the temp directory for input files. The file-system-fetcher is configured with basePath = this directory.
      unpackEmitterBasePath - the basePath where the unpack-emitter writes files. This is where the server will find the zip files created by UNPACK mode. May be null if UNPACK mode won't be used.

  • Method Details

    • shutdown

      public void shutdown()
      Closes the shared PipesParser (destroying its forked workers) and deletes the input and unpack temp directories. Invoked from the server's ordered shutdown sequence, after the HTTP endpoint has stopped -- so workers are torn down only once no new request can arrive.

    • parse

      public List<Metadata> parse(TikaInputStream tis, Metadata metadata, ParseContext parseContext, ParseMode parseMode) throws IOException
      Parses content using pipes-based parsing with process isolation.

      This method spools the input to the dedicated temp directory and uses a relative filename in the FetchKey. The file-system-fetcher is configured with basePath pointing to this directory, so the child process can only access files there.

      The caller is responsible for closing the TikaInputStream.

      Parameters:
      tis - the TikaInputStream containing the content to parse
      metadata - metadata to pass to the parser (may include filename, content-type, etc.)
      parseContext - parse context with handler configuration
      parseMode - the parse mode (RMETA or CONCATENATE)
      Returns:
      list of metadata objects from parsing
      Throws:
      IOException - if temp file operations fail
      TikaServerParseException - if parsing fails

    • responseBuilder

      public static jakarta.ws.rs.core.Response.ResponseBuilder responseBuilder(PipesResult.RESULT_STATUS status, long maxWaitForClientMillis)
      Response builder for a pipes status, carrying Retry-After on the two families where the server knows the condition is transient. Without it, a client loop's only options are to give up or to hammer a pool that is already saturated.

    • getPipesParser

      public PipesParser getPipesParser()
      Gets the PipesParser instance.

    • getPipesConfig

      public PipesConfig getPipesConfig()
      Gets the PipesConfig instance.

    • rejectReservedComponentIds

      public static void rejectReservedComponentIds(FetchEmitTuple t)
      Backstop for ids the tuple deserializer cannot reach. fetcher and emitter are already refused there for any tuple parsed from a request; the UnpackConfig emitter is buried in a parse-context component, so it is checked only here.
      Throws:
      jakarta.ws.rs.BadRequestException - if a caller-supplied tuple names a server-internal component.

    • parseUnpack

      public PipesParsingHelper.UnpackResult parseUnpack(TikaInputStream tis, Metadata metadata, ParseContext parseContext, boolean saveAll) throws IOException
      Parses content using UNPACK mode and returns a path to the zip file containing extracted embedded documents.

      This method: 1. Spools input to the dedicated temp directory 2. Configures UnpackConfig with zipEmbeddedFiles=true 3. The pipes child process extracts embedded files and creates a zip 4. The zip is emitted to the configured file-system emitter 5. Returns the path to the zip file for streaming

      The caller is responsible for deleting the zip file after streaming.

      Parameters:
      tis - the TikaInputStream containing the content to parse
      metadata - metadata to pass to the parser
      parseContext - parse context (may contain UnpackConfig, UnpackSelector, EmbeddedLimits)
      saveAll - if true, includes container text and metadata in the zip
      Returns:
      UnpackResult containing path to zip file and metadata list
      Throws:
      IOException - if parsing or file operations fail