Class StarTableFactory

java.lang.Object
uk.ac.starlink.table.StarTableFactory

public class StarTableFactory extends Object
Manufactures StarTable objects from generic inputs. This factory delegates the actual table creation to external TableBuilder objects, each of which knows how to read a particular table format from an input data stream. Various makeStarTable methods are offered, which construct StarTables from different types of object, such as URL and DataSource. Each of these comes in two types: automatic format detection and named format. Additionally, a list of TableScheme objects is maintained, each of which can produce a table from an opaque specification string of the form :<scheme-name>:<spec>.

In the case of a named format, a specifier must be given for the format in which the table to be read is held. This may be one of the following:

  • The format name - this is a short mnemonic string like "fits" which is returned by the TableBuilder's getFormatName method - it is matched case insensitively. This must be one of the builders known to the factory.
  • The classname of a suitable TableBuilder (the class must implement TableBuilder and have no-arg constructor). Such a class must be on the classpath, but need not have been specified previously to the factory.
  • The empty string or null or AUTO_HANDLER - in this case automatic format detection is used.

In the case of automatic format detection (no format specified), the factory hands the table location to each of the handlers in the default handler list in turn, and if any of them can make a table out of it, it is returned.

In either case, failure to make a table will usually result in a TableFormatException, though if an error in actual I/O is encountered an IOException may be thrown instead.

By default, if the corresponding classes are present, the following TableBuilders are installed in the default handler list (used by default in automatic format detection):

  • uk.ac.starlink.votable.FitsPlusTableBuilder (format name="fits-plus")
  • uk.ac.starlink.votable.ColFitsPlusTableBuilder (format name="colfits-plus")
  • uk.ac.starlink.fits.ColFitsTableBuilder (format name="colfits-basic")
  • uk.ac.starlink.fits.FitsTableBuilder (format name="fits")
  • uk.ac.starlink.votable.VOTableBuilder (format name="votable")
  • uk.ac.starlink.cdf.CdfTableBuilder (format name="cdf")
  • uk.ac.starlink.ecsv.EcsvTableBuilder (format name="ecsv")
  • uk.ac.starlink.pds4.Pds4TableBuilder (format name="pds4")
  • MrtTableBuilder (format name="mrt")
  • uk.ac.starlink.parquet.ParquetTableBuilder (format name="parquet")
  • uk.ac.starlink.feather.FeatherTableBuilder (format name="feather")

The following additional handlers are installed in the known handler list (not used by default but available by specifying the format name):

Additionally, any classes named in the startable.readers system property (as a colon-separated list) which implement the TableBuilder interface and have a no-arg constructor will be instantiated and added to the known handler list.

Some makeStarTable methods take a location String rather than an input stream or DataSource; these may either give a URL or filename, or a scheme-based location of the form :<scheme-name>:<scheme-specification>, for instance "jdbc://localhost/db1#SELECT id FROM gsc". There is a theoretical risk of a namespace clash between input-yielding URLs, or even filenames, and scheme-based locations, but if scheme names avoid obvious values like "http" and "C" this is not likely to cause problems in practice.

The following TableSchemes are installed by default:

Additionally, any classes named in the startable.schemes system property (as a colon-separated list) which implement the TableScheme interface and have a no-arg constructor will be instantiated and added to the known scheme list.

The factory has a flag requireRandom which determines whether the makeStarTable methods are guaranteed to return tables which provide random access (StarTable.isRandom()==true). NOTE the meaning (and name) of this flag has changed as of STIL version 2.1. Previously it was only a hint that random tables were preferred. Now setting it true guarantees that all tables returned by the factory are random.

Author:
Mark Taylor (Starlink)
  • Field Details

    • KNOWN_BUILDERS_PROPERTY

      public static final String KNOWN_BUILDERS_PROPERTY
      System property which can contain a list of TableBuilder classnames for addition to the known (non-automatically detected) handler list.
      See Also:
    • SCHEMES_PROPERTY

      public static final String SCHEMES_PROPERTY
      System property which can contain a list of TableScheme classnames for addition to the default list.
      See Also:
    • AUTO_HANDLER

      public static final String AUTO_HANDLER
      Special handler identifier which signifies automatic format detection.
      See Also:
  • Constructor Details

    • StarTableFactory

      public StarTableFactory()
      Constructs a StarTableFactory with a default list of builders which is not guaranteed to construct random-access tables.
    • StarTableFactory

      public StarTableFactory(boolean requireRandom)
      Constructs a StarTableFactory with a default list of builders specifying whether it will return random-access tables.
      Parameters:
      requireRandom - whether random-access tables will be constructed
    • StarTableFactory

      public StarTableFactory(StarTableFactory fact)
      Constructs a StarTableFactory which is a copy of an existing one.
      Parameters:
      fact - instance to copy
  • Method Details

    • getDefaultBuilders

      public List<TableBuilder> getDefaultBuilders()
      Gets the list of builders which are used for automatic format detection. Builders earlier in the list are given a chance to make the table before ones later in the list. This list can be modified to change the behaviour of the factory.
      Returns:
      a mutable list of TableBuilder objects used to construct StarTables
    • setDefaultBuilders

      public void setDefaultBuilders(TableBuilder[] builders)
      Sets the list of builders which actually do the table construction. Builders earlier in the list are given a chance to make the table before ones later in the list.
      Parameters:
      builders - an array of TableBuilder objects used to construct StarTables
    • getKnownBuilders

      public List<TableBuilder> getKnownBuilders()
      Gets the list of builders which are available for selection by format name. This is initially set to the list of default builders plus a few others. This list can be modified to change the behaviour of the factory.
      Returns:
      a mutable list of TableBuilder objects which may be specified for table building
    • setKnownBuilders

      public void setKnownBuilders(TableBuilder[] builders)
      Sets the list of builders which are available for selection by format name. This is initially set to the list of default builders plus a few others.
      Parameters:
      builders - an array of TableBuilder objects used to construct StarTables
    • getKnownFormats

      public List<String> getKnownFormats()
      Returns the list of format names, one for each of the handlers returned by getKnownBuilders().
      Returns:
      list of format name strings
    • getSchemes

      public Map<String,TableScheme> getSchemes()
      Returns a schemeName->scheme map indicating the TableSchemes in use by this factory. This map is mutable, so entries may be added or removed, but NOTE that the map keys are assumed to be equivalent to the getSchemeName return value for their value, so modify it with care. Consider using the addScheme(uk.ac.starlink.table.TableScheme) method for adding entries.
      Returns:
      map of scheme names to schemes
    • addScheme

      public void addScheme(TableScheme scheme)
      Safely adds a table scheme for use by this factory. It is equivalent to getSchemes().put(scheme.getSchemeName(),scheme).
      Parameters:
      scheme - new scheme for use
    • setRequireRandom

      public void setRequireRandom(boolean requireRandom)
      Sets whether random-access tables will be constructed by this factory. If this flag is set true then any table returned by the various makeStarTable methods is guaranteed to provide random access (its StarTable.isRandom() method will return true). If the flag is false, then returned tables may or may not be random-access.
      Parameters:
      requireRandom - whether this factory will create random-access tables
    • requireRandom

      public boolean requireRandom()
      Returns the requireRandom flag. If this flag is set true then any table returned by the various makeStarTable methods is guaranteed to provide random access (its StarTable.isRandom() method will return true). If the flag is false, then returned tables may or may not be random-access.
      Returns:
      whether this factory will create random-access tables
    • setStoragePolicy

      public void setStoragePolicy(StoragePolicy policy)
      Sets the storage policy. This may be used to determine what kind of scratch storage is used when constructing tables.
      Parameters:
      policy - the new storage policy object
    • getStoragePolicy

      public StoragePolicy getStoragePolicy()
      Returns the current storage policy. This may be used to determine what kind of scratch storage is used when constructing tables. If it has not been set explicitly, the default policy is used (StoragePolicy.getDefaultPolicy()).
      Returns:
      storage policy object
    • setPreparation

      public void setPreparation(TablePreparation tablePrep)
      Sets a table preparation object that is invoked on each table created by this factory. Any previous value is overwritten. Null is allowed.
      Parameters:
      tablePrep - new table preparation, or null
    • getPreparation

      public TablePreparation getPreparation()
      Returns the current table preparation object, if any. By default, null is returned.
      Returns:
      table preparation, or null
    • setInputRestriction

      public void setInputRestriction(Predicate<uk.ac.starlink.util.DataSource> restriction)
      Sets an object that can control access to input data. If a non-null value is set, then any attempt to read a table from a resource such as a file or URL will first test it using the supplied predicate. If its test method returns false, the table read attempt will fail with an IOException.
      Parameters:
      restriction - policy for restricting DataSource access, or null for no restrictions
    • getInputRestriction

      public Predicate<uk.ac.starlink.util.DataSource> getInputRestriction()
      Returns the object controlling access to input data. By default this returns null, meaning no access controls.
      Returns:
      policy for restricting DataSource access, or null
      See Also:
    • randomTable

      public StarTable randomTable(StarTable table) throws IOException
      Returns a table based on a given table and guaranteed to have random access. If the original table table has random access then it is returned, otherwise a new random access table is built using its data.

      This convenience method is equivalent to getStoragePolicy().randomTable(table).

      Parameters:
      table - original table
      Returns:
      a table with the same data as table and with isRandom()==true
      Throws:
      IOException
    • makeStarTable

      public StarTable makeStarTable(uk.ac.starlink.util.DataSource datsrc) throws TableFormatException, IOException
      Constructs a StarTable from a DataSource object using automatic format detection.
      Parameters:
      datsrc - the data source containing the table data
      Returns:
      a new StarTable view of the resource datsrc
      Throws:
      TableFormatException - if none of the default handlers could turn datsrc into a table
      IOException - if an I/O error is encountered
    • makeStarTables

      public TableSequence makeStarTables(uk.ac.starlink.util.DataSource datsrc) throws TableFormatException, IOException
      Constructs a sequence of StarTables from a DataSource using automatic format detection. Only certain formats (those whose handlers implement MultiTableBuilder) will be capable of returning a sequence having more than one element.
      Parameters:
      datsrc - the data source containing the table data
      Returns:
      a sequence of tables loaded from datsrc
      Throws:
      TableFormatException - if none of the default handlers could turn datsrc into a table
      IOException - if an I/O error is encountered
    • makeStarTable

      public StarTable makeStarTable(String location) throws TableFormatException, IOException
      Constructs a StarTable from a location string without format specification. The location string can represent a filename or URL, or a scheme-based specification of the form <scheme>:<scheme-spec> corresponding to one of the installed schemes.
      Parameters:
      location - the name of the table resource
      Returns:
      a new StarTable view of the resource at location
      Throws:
      TableFormatException - if no handler capable of turning location into a table is available
      IOException - if one of the handlers encounters an error constructing a table
    • makeStarTable

      @Deprecated public StarTable makeStarTable(URL url) throws IOException
      Deprecated.
      Use makeStarTable(new URLDataSource(url))
      Constructs a StarTable from a URL using automatic format detection.
      Parameters:
      url - the URL where the table lives
      Returns:
      a new StarTable view of the resource at url
      Throws:
      TableFormatException - if no handler capable of turning datsrc into a table is available
      IOException - if one of the handlers encounters an error constructing a table
    • makeStarTable

      public StarTable makeStarTable(uk.ac.starlink.util.DataSource datsrc, String handler) throws TableFormatException, IOException
      Constructs a StarTable from a DataSource using a named table input handler. The input handler may be named either using its format name (as returned from the TableBuilder.getFormatName() method) or by giving the full class name of the handler. In the latter case this factory does not need to have been informed about the handler previously. If null or the empty string or the special value AUTO_HANDLER is supplied for handler, it will fall back on automatic format detection.
      Parameters:
      datsrc - the data source containing the table data
      handler - specifier for the handler which can handle tables of the right format
      Returns:
      a new StarTable view of the resource datsrc
      Throws:
      TableFormatException - if datsrc does not contain a table in the format named by handler
      IOException - if an I/O error is encountered
    • makeStarTables

      public TableSequence makeStarTables(uk.ac.starlink.util.DataSource datsrc, String handler) throws TableFormatException, IOException
      Constructs a sequence of StarTables from a DataSource using a named table input handler. The input handler may be named either using its format name (as returned from the TableBuilder.getFormatName() method) or by giving the full class name of the handler. In the latter case this factory does not need to have been informed about the handler previously. If null or the empty string or the special value AUTO_HANDLER is supplied for handler, it will fall back on automatic format detection.

      If the handler does not implement the MultiTableBuilder interface, then the returned sequence will contain a single table.

      Parameters:
      datsrc - the data source containing the table data
      handler - specifier for the handler which can handle tables of the right format
      Returns:
      a sequence of StarTables loaded from datsrc
      Throws:
      TableFormatException - if datsrc does not contain a table in the format named by handler
      IOException - if an I/O error is encountered
    • makeStarTables

      public TableSequence makeStarTables(String location, String handler) throws TableFormatException, IOException
      Constructs a sequence of StarTables from a location string using a named table input handler. The input handler may be named either using its format name (as returned from the TableBuilder.getFormatName() method) or by giving the full class name of the handler. In the latter case this factory does not need to have been informed about the handler previously. If null or the empty string or the special value AUTO_HANDLER is supplied for handler, it will fall back on automatic format detection.

      Alternatively, the location string can be a scheme-based specification, in which case the handler is ignored.

      Parameters:
      location - the name of the table resource
      handler - specifier for the handler which can handle tables of the right format
      Returns:
      a new StarTable view of the resource at location
      Throws:
      TableFormatException - if location does not point to a table in the format named by handler
      IOException - if an I/O error is encountered
    • makeStarTable

      public StarTable makeStarTable(String location, String handler) throws TableFormatException, IOException
      Constructs a StarTable from a location string using a named table input handler. The input handler may be named either using its format name (as returned from the TableBuilder.getFormatName() method) or by giving the full class name of the handler. In the latter case this factory does not need to have been informed about the handler previously. If null or the empty string or the special value AUTO_HANDLER is supplied for handler, it will fall back on automatic format detection.

      A location of "-" means standard input - in this case the handler must be specified.

      Alternatively, the location string can be a scheme-based specification, in which case the handler is ignored.

      Parameters:
      location - the name of the table resource
      handler - specifier for the handler which can handle tables of the right format
      Returns:
      a new StarTable view of the resource at location
      Throws:
      TableFormatException - if location does not point to a table in the format named by handler
      IOException - if an I/O error is encountered
    • makeStarTable

      @Deprecated public StarTable makeStarTable(URL url, String handler) throws TableFormatException, IOException
      Deprecated.
      Use makeStarTable(new URLDataSource(url),handler)
      Constructs a StarTable from a URL using a named table input handler. The input handler may be named either using its format name (as returned from the TableBuilder.getFormatName() method) or by giving the full class name of the handler. In the latter case this factory does not need to have been informed about the handler previously. If null or the empty string or the special value AUTO_HANDLER is supplied for handler, it will fall back on automatic format detection.
      Parameters:
      url - the URL where the table lives
      handler - specifier for the handler which can handle tables of the right format
      Returns:
      a new StarTable view of the resource at url
      Throws:
      TableFormatException - if the resource at url cannot be turned into a table by handler
      IOException - if an I/O error is encountered
    • makeStarTable

      public StarTable makeStarTable(InputStream in, TableBuilder builder) throws TableFormatException, IOException
      Attempts to read and return a StarTable from an input stream. This is not always possible, since certain table handlers may required more than one pass through the input data. The handler must be specified (automatic format detection cannot be used on a stream). The input stream will be decompressed and buffered if necessary.
      Parameters:
      in - input stream
      builder - handler which understands the data in in
      Returns:
      a table read from the stream if it could be done
      Throws:
      TableFormatException - if builder needs more than one pass of the data, or the stream is in some way malformed
      IOException - for other I/O errors
      See Also:
    • makeStarTable

      public StarTable makeStarTable(Transferable trans) throws IOException
      Constructs a StarTable from a Transferable object using automatic format detection. In conjunction with a suitable TransferHandler this makes it easy to accept drop of an object representing a table which has been dragged from another application.

      The implementation of this method currently tries the following on a given transferable to turn it into a table:

      • If it finds a URL object, passes that to the URL factory method
      • If it finds a transferable that will supply an InputStream, turns it into a DataSource and passes that to the DataSource constructor
      Parameters:
      trans - the Transferable object to construct a table from
      Returns:
      a new StarTable constructed from the Transferable
      Throws:
      TableFormatException - if no table can be constructed
      IOException
      See Also:
    • canImport

      public boolean canImport(DataFlavor[] flavors)
      Indicates whether a particular set of DataFlavor ojects offered by a Transferable is suitable for attempting to turn the Transferable into a StarTable.

      Each of the builder objects is queried about whether it can import the given flavour, and if one says it can, a true value is returned. A true value is also returned if one of the flavours has a representation class of URL.

      Parameters:
      flavors - the data flavours offered
    • getJDBCHandler

      public JDBCHandler getJDBCHandler()
      Returns the JDBC handler object used by this factory.
      Returns:
      the JDBC handler
    • setJDBCHandler

      public void setJDBCHandler(JDBCHandler handler)
      Sets the JDBC handler object used by this factory.
      Parameters:
      handler - the JDBC handler
    • getTableBuilder

      public TableBuilder getTableBuilder(String name) throws TableFormatException
      Returns a table handler with a given name. This name may be either its format name (as returned from the TableBuilder.getFormatName() method) or by giving the full class name of the handler. In the latter case this factory does not need to have been informed about the handler previously.
      Parameters:
      name - specification of the handler required
      Returns:
      TableBuilder specified by name
      Throws:
      TableFormatException - if name doesn't name any available handler
    • parseSchemeLocation

      public static String[] parseSchemeLocation(String location)
      Parses a scheme-format table specification as a scheme name and a scheme-specific part. Normally schemes are of the form ":<scheme-name>:<scheme-specific-part>", but as a special case the initial colon may be omitted for JDBC (backward compatibility).
      Parameters:
      location - table specification
      Returns:
      if the location is syntactically a scheme, a 2-element array giving [scheme-name,scheme-specific-part]; otherwise null