TracePoint

Class

Document-class: TracePoint

A class that provides the functionality of Kernel#set_trace_func in a nice Object-Oriented API.

Example

We can use TracePoint to gather information specifically for exceptions:

trace = TracePoint.new(:raise) do |tp|
    p [tp.lineno, tp.event, tp.raised_exception]
end
#=> #<TracePoint:disabled>

trace.enable
#=> false

0 / 0
#=> [5, :raise, #<ZeroDivisionError: divided by 0>]

Events

If you don’t specify the type of events you want to listen for, TracePoint will include all available events.

Note do not depend on current event set, as this list is subject to change. Instead, it is recommended you specify the type of events you want to use.

To filter what is traced, you can pass any of the following as events:

:line

execute an expression or statement on a new line

:class

start a class or module definition

:end

finish a class or module definition

:call

call a Ruby method

:return

return from a Ruby method

:c_call

call a C-language routine

:c_return

return from a C-language routine

:raise

raise an exception

:b_call

event hook at block entry

:b_return

event hook at block ending

:a_call

event hook at all calls (call, b_call, and c_call)

:a_return

event hook at all returns (return, b_return, and c_return)

:thread_begin

event hook at thread beginning

:thread_end

event hook at thread ending

:fiber_switch

event hook at fiber switch

:script_compiled

new Ruby code compiled (with eval, load or require)

Class Methods

TracePoint.allow_reentry
::

In general, while a TracePoint callback is running, other registered callbacks are not called to avoid confusion by reentrance. This method allows the reentrance in a given block. This method should be used carefully, otherwise the callback can be easily called infinitely.

If this method is called when the reentrance is already allowed, it raises a RuntimeError.

TracePoint.new(*events) { |obj| block } → obj
::

Returns a new TracePoint object, not enabled by default.

Next, in order to activate the trace, you must use TracePoint#enable

trace = TracePoint.new(:call) do |tp|
    p [tp.lineno, tp.defined_class, tp.method_id, tp.event]
end
#=> #<TracePoint:disabled>

trace.enable
#=> false

puts "Hello, TracePoint!"
# ...
# [48, IRB::Notifier::AbstractNotifier, :printf, :call]
# ...

When you want to deactivate the trace, you must use TracePoint#disable

trace.disable

See Events at TracePoint for possible events and more information.

A block must be given, otherwise an ArgumentError is raised.

If the trace method isn’t included in the given events filter, a RuntimeError is raised.

TracePoint.trace(:line) do |tp|
    p tp.raised_exception
end
#=> RuntimeError: 'raised_exception' not supported by this event

If the trace method is called outside block, a RuntimeError is raised.

TracePoint.trace(:line) do |tp|
  $tp = tp
end
$tp.lineno #=> access from outside (RuntimeError)

Access from other threads is also forbidden.

TracePoint.stat → obj
::

Returns internal information of TracePoint.

The contents of the returned value are implementation specific. It may be changed in future.

This method is only for debugging TracePoint itself.

TracePoint.trace(*events) { |obj| block } → obj
::

A convenience method for TracePoint.new, that activates the trace automatically.

trace = TracePoint.trace(:call) { |tp| [tp.lineno, tp.event] }
#=> #<TracePoint:enabled>

trace.enabled? #=> true
Instance Methods

binding
#

Return the generated binding object from event.

Note that for c_call and c_return events, the binding returned is the binding of the nearest Ruby method calling the C method, since C methods themselves do not have bindings.

callee_id
#

Return the called name of the method being called

defined_class
#

Return class or module of the method being called.

class C; def foo; end; end
trace = TracePoint.new(:call) do |tp|
  p tp.defined_class #=> C
end.enable do
  C.new.foo
end

If method is defined by a module, then that module is returned.

module M; def foo; end; end
class C; include M; end;
trace = TracePoint.new(:call) do |tp|
  p tp.defined_class #=> M
end.enable do
  C.new.foo
end

Note: defined_class returns singleton class.

6th block parameter of Kernel#set_trace_func passes original class of attached by singleton class.

This is a difference between Kernel#set_trace_func and TracePoint.

class C; def self.foo; end; end
trace = TracePoint.new(:call) do |tp|
  p tp.defined_class #=> #<Class:C>
end.enable do
  C.foo
end

trace.disable → true or false

trace.disable { block } → obj
#

Deactivates the trace

Return true if trace was enabled. Return false if trace was disabled.

trace.enabled?      #=> true
trace.disable       #=> true (previous status)
trace.enabled?      #=> false
trace.disable       #=> false

If a block is given, the trace will only be disable within the scope of the block.

trace.enabled?
#=> true

trace.disable do
    trace.enabled?
    # only disabled for this block
end

trace.enabled?
#=> true

Note: You cannot access event hooks within the block.

trace.disable { p tp.lineno }
#=> RuntimeError: access from outside

trace.enable(target: nil, target_line: nil, target_thread: nil) → true or false

trace.enable(target: nil, target_line: nil, target_thread: nil) { block } → obj
#

Activates the trace.

Returns true if trace was enabled. Returns false if trace was disabled.

trace.enabled?  #=> false
trace.enable    #=> false (previous state)
                #   trace is enabled
trace.enabled?  #=> true
trace.enable    #=> true (previous state)
                #   trace is still enabled

If a block is given, the trace will only be enabled within the scope of the block.

trace.enabled?
#=> false

trace.enable do
  trace.enabled?
  # only enabled for this block
end

trace.enabled?
#=> false

target, target_line and target_thread parameters are used to limit tracing only to specified code objects. target should be a code object for which RubyVM::InstructionSequence.of will return an instruction sequence.

t = TracePoint.new(:line) { |tp| p tp }

def m1
  p 1
end

def m2
  p 2
end

t.enable(target: method(:m1))

m1
# prints #<TracePoint:line test.rb:4 in `m1'>
m2
# prints nothing

Note: You cannot access event hooks within the enable block.

trace.enable { p tp.lineno }
#=> RuntimeError: access from outside

trace.enabled? → true or false
#

The current status of the trace

eval_script
#

Compiled source code (String) on *eval methods on the :script_compiled event. If loaded from a file, it will return nil.

event
#

Type of event

See Events at TracePoint for more information.

trace.inspect → string
#

Return a string containing a human-readable TracePoint status.

instruction_sequence
#

Compiled instruction sequence represented by a RubyVM::InstructionSequence instance on the :script_compiled event.

Note that this method is MRI specific.

lineno
#

Line number of the event

method_id
#

Return the name at the definition of the method being called

parameters
#

Return the parameters definition of the method or block that the current hook belongs to. Format is the same as for Method#parameters

path
#

Path of the file being run

raised_exception
#

Value from exception raised on the :raise event

return_value
#

Return value from :return, c_return, and b_return event

self
#

Return the trace object during event

Same as the following, except it returns the correct object (the method receiver) for c_call and c_return events:

trace.binding.eval('self')