2.5.6 Blobs

Disclaimer:

The blob API for C++ is not completely general, but is designed to make a specific use case easier to write. For other use cases, the underlying C API can still be used. The use case is:

• The blob contains the foreign object (e.g., contains a pointer to a database connection), plus optionally some other data.
• The blob is created by a predicate that makes the foreign object and stores it (or a pointer to it) within the blob - for example, making a connection to a database or compiling a regular expression into an internal form.
• Optionally, there is a predicate that deletes the foreign object, such as a file or database connection close().
• The blob will not be subclassed.
• The blob is defined as a subclass of PlBlob, which provides a number of fields and methods, of which a few can be overridden in the blob (notably: write_fields(), compare_fields(), save(), load(), and the destructor).
• The blob must have a default constructor that sets all the fields to appropriate initial values.^{10This is used by the load() method, which by default throws an error.}
• The blob's constructor throws an exception and cleans up any resources if it cannot create the blob.^{11This is not a strong requirement, but the code is simpler if this style is used.}
• The foreign object can be deleted when the blob is deleted. That is, the foreign object is created using the new operator and passes ownership to the blob.
• The blob's allocation is controlled by Prolog and its destructor is envoked when the blob is garbage collected. Optionally, the predicate that deletes the foreign object deletes the foreign object and the Prolog garbage collector only frees the blob.

A Prolog blob consists of five parts:

• A PL_blob_t structure that defines the callbacks.
• A structure that contains the blob data.
• A "create" or "open" predicate that unifies one of its arguments with a newly created blob that contains the foreign object.
• (Optionally) a "close" predicate that does the opposite of the "create" or "open" predicate.
• Predicates that manipulate the foreign object (e.g., for a file-like object, these could be read, write, seek, etc.).

For the PL_blob_t structure, the C++ API provides a set of template functions that allow easily setting up the callbacks, and defining the corresponding methods in the blob "contents" class. The C interface allows more flexibility by allowing some of the callbacks to default; however, the C++ API for blobs provides suitable callbacks for all of them, so usually the programmer will specify all the template callbacks using the PL_BLOB_DEFINITION(blob_class,blob_name) macro.

For the data, which is subclassed from PlBlob, the programmer defines the various fields, a constructor that initializes them, and a destructor. Optionally, methods can be defined for one of more of blob compare_fields(), write_fields(), save(), load(). More details on these are given later.

There is a mismatch between how Prolog does memory management (and garbage collection) and how C++ does it. In particular, Prolog assumes that cleanup will be done in the release() function associated with the blob whereas C++ typically does cleanup in a destructor. The blob interface gets around this mismatch by providing a default release() function that assumes that the blob was created using PL_BLOB_NOCOPY and manages memory using a std::unique_ptr.

The C blob interface has a flag that determines how memory is managed: PL_BLOB_NOCOPY. The PL_BLOB_DEFINITION() macro sets this, so Prolog does not do a call to free() when the blob is garbage collected; instead, it lets the blob's release() free the memory, which is done by calling the C++ destructor.

The C++ API for blobs only supports blobs with PL_BLOB_NOCOPY.^{12The API can probably also support blobs with PL_BLOB_UNIQUE, but there seems to be little point in setting this flag for non-text blobs.}

2.5.6.1 How to define a blob using C++

TL;DR: Use PL_BLOB_DEFINITION() to define the blob with the flag PL_BLOB_NOCOPY and the default PlBlob wrappers; define your struct as a subclass of PlBlob with no copy constructor, move constructor, or assignment operator; create blob using std::unique_ptr<PlBlob>(new ...), call PlTerm::unify_blob(). Optionally, define one or more of: compare_fields(), write_fields(), save(), load() methods (these are described after the sample code).

2.5.6.2 The life of a PlBlob

In this section, the blob is of type MyBlob, a subclass of PlBlob.

A blob is typically created by calling a predicate that does the following:

• Creates the blob using auto ref = std::unique_ptr<PlBlob>(new MyBlob>(...)) (std::make_unique() can't be used because it returns type std::unique_ptr<MyBlob> but PlTerm::unify_blob() requires a std::unique_ptr<PlBlob> and C++'s type inferencing can't figure out that this is a covariant type).

• Calls PlTerm::unify_blob(ref), using PlCheckFail() to throw an exception if it fails. This, in turn, calls:
  • PlBlobV<MyBlob>acquire(), which calls
  • MyBlob::acquire(), which sets MyBlob::symbol_ (MyBlob::symbol_ is usually accessed using the method MyBlob::symbol_term()). If this all succeeds, PlTerm::unify_blob(ref) calls ref.release() to pass ownership to the Prolog blob. If you wish to use std::make_unique<MyBlob>(), you could instead do:

          auto ref = std::make_unique<MyBlob>(...);
             ... // code that accesses fields in *ref
          std::unique_ptr<PlBlob> refb(ref.release()); // transfer ownership of ptr
             // from here on, can't access fields in *ref
          return A2.unify_blob(refb);
          

At this point, the blob is owned by Prolog and will be freed by its atom garbage collector, which will call the blob's destructor.

Whenever a predicate is called with the blob as an argument (e.g., as A1), the blob can be accessed by PlBlobv<MyBlob>::cast_check(A1.as_atom()).

Within a method, the Prolog blob can be accessed as a term (e.g., for constructing an error term) using the method MyBlob::symbol_term(). This field is initialized by the call to PlTerm::unify_blob(); if MyBlob::symbol_term() is called before a successful call to PlTerm::unify_blob(), MyBlob::symbol_term() returns a PlTerm_var.

When the atom garbage collector runs, it frees the blob by first calling the release() callback, which does delete, which calls the destructor MyBlob:: MyBlob(). Note that C++ destructors are not supposed to raise exception; they also should not cause a Prolog error, which could cause deadlock unless the real work is done in another thread.

Often it is desired to release the resources before the garbage collector runs. To do this, the programmer can provide a "close" predicate which is the inverse of the "open" predicate that created the blob. This typically has the same logic as the destructor, except that it can raise a Prolog error.

2.5.6.3 C++ exceptions and blobs

When a blob is used in the context of a PREDICATE() macro, it can raise a C++ exception (PlFail or PlException) and the PREDICATE() code will convert it to the appropriate Prolog failure or error; memory allocation exceptions are also handled.

Blobs have callbacks, which can run outside the context of a PREDICATE(). Their exception handling is as follows:

• acquire(), which is called from PlBlobV<MyBlob>::acquire(), can throw a C++ exception.
• compare_fields(), which is called from PlBlobV<MyBlob>::compare(), should not throw an exception. A Prolog error won't work as it uses "raw pointers" and thus a GC or stack shift triggered by creating the exception will upset the system.
• write_fields(), which is called from PlBlobV<MyBlob>::write(), can throw an exception, just like code inside a PREDICATE(). In particular, you can wrap calls to Sfprintf() in PlCheckFail(), although the calling context will check for errors on the stream, so checking the Sfprintf() result isn't necessary.
• save() can throw a C++ exception, including PlFail().
• load() can throw a C++ exception, which is converted to a return value of PlAtom::null, which is interpreted by Prolog as failure.

2.5.6.4 Sample PlBlob code

Here is minimal sample code for creating a blob that owns a connection to a database. It has a single field (connection) and defines compare_fields() and write_fields().

struct MyConnection
{ std::string name;

  explicit MyConnection();
  explicit MyConnection(const std::string& _name);
  ~MyConnection() { }
  bool open();
  bool close() noexcept;
  void portray(PlStream& strm) const;
};

struct MyBlob;

static PL_blob_t my_blob = PL_BLOB_DEFINITION(MyBlob, "my_blob");

struct MyBlob : public PlBlob
{ std::unique_ptr<MyConnection> connection;

  explicit MyBlob()
    : PlBlob(&my_blob) { }

  explicit MyBlob(const std::string& connection_name)
    : PlBlob(&my_blob),
      connection(std::make_unique<MyConnection>(connection_name))
  { if ( !connection->open() )
      throw MyBlobError("my_blob_open_error");
  }

  PL_BLOB_SIZE

  ~MyBlob() noexcept
  { if ( !close() )
      Sdprintf("***ERROR: Close MyBlob failed: %s\n", name().c_str()); // Can't use PL_warning()
  }

  inline std::string
  name() const
  { return connection ? connection->name : "";
  }

  bool close() noexcept
  { if ( !connection )
      return true;
    bool rc = connection->close();
    connection.reset(); // Can be omitted, leaving deletion to ~MyBlob()
    return rc;
  }

  PlException MyBlobError(const char* error) const
  { return PlGeneralError(PlCompound(error, PlTermv(symbol_term())));
  }

  int compare_fields(const PlBlob* _b_data) const override
  { auto b_data = static_cast<const MyBlob*>(_b_data); // See note about cast
    return name().compare(b_data->name());
  }

  bool write_fields(IOSTREAM *s, int flags) const override
  { PlStream strm(s);
    strm.printf(",");
    return write_fields_only(strm);
  }

  bool write_fields_only(PlStream& strm) const
  { if ( connection )
      connection->portray(strm);
    else
      strm.printf("closed");
    return true;
  }

  bool portray(PlStream& strm) const
  { strm.printf("MyBlob(");
    write_fields_only(strm);
    strm.printf(")");
    return true;
  }
};

// %! create_my_blob(+Name: atom, -MyBlob) is semidet.
PREDICATE(create_my_blob, 2)
{ // Allocating the blob uses std::unique_ptr<MyBlob> so that it'll be
  // deleted if an error happens - the auto-deletion is disabled by
  // ref.release() before returning success.

  auto ref = std::unique_ptr<PlBlob>(new MyBlob(A1.as_atom().as_string()));
  return A2.unify_blob(&ref);
}

// %! close_my_blob(+MyBlob) is det.
// % Close the connection, silently succeeding if is already
// % closed; throw an exception if something goes wrong.
PREDICATE(close_my_blob, 1)
{ auto ref = PlBlobV<MyBlob>::cast_ex(A1, my_blob);
  if ( !ref->close() )
    throw ref->MyBlobError("my_blob_close_error");
  return true;
}

// %! portray_my_blob(+Stream, +MyBlob) is det.
// % Hook predicate for
// %   user:portray(MyBlob) :-
// %     blob(MyBlob, my_blob), !,
// %     portray_my_blob(current_output, MyBlob).
PREDICATE(portray_my_blob, 2)
{ auto ref = PlBlobV<MyBlob>::cast_ex(A2, my_blob);
  PlStream strm(A1, 0);
  return ref->portray(strm);
}

2.5.6.5 Discussion of the sample PlBlob code

• PL_BLOB_DEFINITION(MyBlob, "my_blob") creates a PL_blob_t structure with the wrapper functions and flags set to PL_BLOB_NOCOPY. It should be declared outside the PlBlob class and should not be marked const - otherwise, a runtime error can occur.^{13The cause of the runtime error is not clear, but possibly has to do with the order of initializing globals, which is unspecified for C++.}

• The MyBlob struct is a subclass of PlBlob. See below for a discussion of the default behaviors.
  • MyBlob contains a pointer to a MyConnection object and keeps a copy of the connection's name. The MyConnection object is handled by a std::unique_ptr smart pointer, so that it is automatically freed when the MyBlob object is freed.

  • A default constructor is defined - this is needed for the load() and save() methods; it invokes the PlBlob constructor.

  • The MyBlob class must not provide a copy or move constructor, nor an assignment operator (PlBlob has these as delete, so if you try to use one of these, you will get a compile-time error).

  • PlBlob’s constructor sets blob_t_ to a pointer to the my_blob definition. This is used for run-time consistency checking by the various callback functions and for constructing error terms (see PlBlob::symbol_term()).

  • PlBlob’s acquire() is called by PlBlobV<MyBlob>::awcuire() and fills in the symbol field. MyBlob must not override this - it is not a virtual method.

  • PlBlob::symbol_term() Creates a term from the blob, for use in error terms. It is always safe to use this; if the symbol hasn't been set (because acquire() hasn't been called), symbol_term() returns a "var" term.

  • The MyBlob(connection_name) constructor creates a MyConnection object. If this fails, an exception is thrown. The constructor then calls MyConnection::open() and throws an exception if that fails. (The code would be similar if instead the constructor for MyConnection also did an open and threw an exception on failure.)

  • The PL_BLOB_SIZE is boilerplate that defines a blob_size_() method that is used when the blob is created.

  • The destructor MyBlob() is called when the blob is released by the garbage collector and in turn calls the MyBlob::close(), throwing away the result. If there is an error, a message is printed because there is no other way report the error. For this reason, it is preferred that the program explicitly calls the close_my_blob/1 predicate, which can raise an error. One way of doing this is by using the at_halt/1 hook.

  • The MyBlob::close() method is called by either the destructor or by the close_my_blob/1 predicate. Because it can be called by the garbage collector, which does not provide the usual environment and which may also be in a different thread, the only Prolog function that can be called is PlAtom::unregister_ref(); and the MyBlob::close() method must not throw an exception.^{14It isn't enough to just catch exceptions; for example, if the code does throw PlUnknownError("..."), that will try to create a Prolog term, which will crash because the environment for creating terms is not available.} Because there is no mechanism for reporting an error, the destructor prints a message on failure (calling PL_warning() would cause a crash).

    PlBlob::close() calls MyConnection::close() and then frees the object. Error handling is left to the caller because of the possibility that this is called in the context of garbage collection. It is not necessary to free the MyConnection object here - if it is not freed, the std::unique_ptr<MyConnection>’s destructor would free it.

  • PlBlob::MyBlobError() is a convenience method for creating errror terms.

  • PlBlob::compare_fields() makes the blob comparison function more deterministic by comparing the name fields; if the names are the same, the comparison will be done by comparing the addresses of the blobs (which is the default behavior for blobs defined using the C API). PlBlob::compare_fields() is called by PlBlobV<PlBlob>::compare(), which provides the default comparison if PlBlob::compare_fields() returns 0 ("equal").

    The _b_data argument is of type const PlBlob* - this is cast to const MyBlob* using a static_cast. This is safe because Prolog guarantees that PlBlobV<PlBlob>::compare() will only be called if both blobs are of the same type.

  • PlBlob::write_fields() outputs the name and the status of the connection, in addition to the default of outputting the blob type and its address. This is for illustrative purposes only; an alternative is to have a my_blob_properties/2 predicate to provide the information.

    The flags argument is the same as given to PlBlobV<PlBlob>::write(), which is a bitwise or of zero or more of the PL_WRT_* flags that were passed in to the caling PL_write_term() (defined in SWI-Prolog.h). The flags do not have the PL_WRT_NEWLINE bit set, so it is safe to call PlTerm::write() and there is no need for writing a trailing newline.

    If anything in PlBlob::write_fields() throws a C++ exception, it will be caught by the calling PlBlobV<PlBlob>::write() and handled appropriately.

  • PlBlob::save() and PlBlob::load() are not defined, so the defaults are used - they throw an error on an attempt to save the blob (e.g., by using
    indexqsave_program/[1,2]qsave_program/[1,2]).^{15If these weren't specified, the defaults would save the internal form of the blob, which is probably not what you want.}

• create_my_blob/2 predicate:
  • std::unique_ptr<PlBlob>() creates a MyBlob that is deleted when it goes out of scope. If an exception occurs between the creation of the blob or if the call to unify_blob() fails, the pointer will be automatically freed (and the MyBlob destructor will be called).

    If PlTerm::unify_blob() is called with a pointer to a std::unique_ptr, it takes ownership of the object by calling std::unique_ptr<PlBlob>::release(). This sets ref to nullptr, so any attempt to use ref after a successful call to PlTerm::unify_blob() will be an error.

    If you wish to create a MyBlob object instead of a PlBlob object, a slightly different form is used:

          auto ref = std::make_unique<MyBlob>(...);
            ...
          std::unique_ptr<PlBlob> refb(ref.release());
          PlCheckFail(A2.unify_blob(&refb));
          return true;
          

• close_my_blob/1 predicate:
  • The argument is turned into a MyBlob pointer using the PlBlobV<MyBlob>::cast_ex() function, which will throw a type_error if the argument isn't a blob of the expected type.

  • The MyBlob::close() method is called - if it fails, a Prolog error is thrown.

2.5.6.6 Identifying blobs by atoms

Passing a blob around can be inconvenient; there is an easy way to identify a blob by an atom. An example of this is with streams, which are identified by atoms such as user_input.

A utility class AtomMap is provided for this situation. See section 2.20.4.