assert(♥)

Pyparsing differs from yeanpypa and regexen in that it assumes whitespace as an implicit delimiter, and leaves it out of the parsed results. Here is my pyparsing version of your TAP parser (you can find the full script at the pyparsing Wiki, http://pyparsing.wikispaces.com/space/showimage/TAP.py):

# newlines are significant whitespace in this parser, so set 
# default skippable whitespace to just spaces and tabs
ParserElement.setDefaultWhitespaceChars(" \t")
NL = LineEnd().suppress()

integer = Word(nums)
plan = '1..' + integer("ubound")

OK,NOT_OK = map(Literal,['ok','not ok'])
testStatus = (OK | NOT_OK)

description = Regex("[^#\n]+")
description.setParseAction(lambda t:t[0].lstrip('- '))

TODO,SKIP = map(CaselessLiteral,'TODO SKIP'.split())
directive = Group(Suppress('#') + (TODO + restOfLine | 
    FollowedBy(SKIP) + 
	restOfLine.copy().setParseAction(lambda t:['SKIP',t[0]]) ))

testLine = Group(testStatus("passed") +
    Optional(integer)("testNumber") + 
    Optional(description)("description") + 
    Optional(directive)("directive")
    )
bailLine = Group(Literal("Bail out!")("BAIL") + 
		    empty + Optional(restOfLine)("reason"))

tapOutput = Optional(Group(plan)("plan") + NL) & \
	    Group(OneOrMore((testLine|bailLine) + NL))("tests")

The quoted strings embedded within the grammar indicate field names that can be used to directly access strings within the parsed results. Here is the code that tests and processes the TAP output:

def tallyResults(results):
    failedTests = []
    skippedTests = []
    todoTests = []
    bonusTests = []
    if results.plan:
	expected = range(1, int(results.plan.ubound)+1)
    else:
	expected = range(1,len(results.tests)+1)
    for i,res in enumerate(results.tests):
	# test for bail out
	if res.BAIL:
	    print "Test suite aborted: " + res.reason
	    failedTests += expected[i:]
	    break
	
	#~ print res.dump()
	testnum = i+1
	if res.testNumber != "":
	    if testnum != int(res.testNumber):
		print "ERROR! test %(testNumber)s out of sequence" % res
	    testnum = int(res.testNumber)
	passed = (res.passed=="ok")
	skipped = todo = False
	if res.directive:
	    skipped = (res.directive[0][0]=='SKIP')
	    todo = (res.directive[0][0]=='TODO')
	if not passed: failedTests.append(testnum)
	if skipped: skippedTests.append(testnum)
	if todo: todoTests.append(testnum)
	if todo and passed: bonusTests.append(testnum)
    
    if failedTests:
	print "Failed tests:", failedTests
    if skippedTests:
	print "SKIPPED:", skippedTests
    if todoTests:
	print "TODO:", todoTests
    if bonusTests:
	print "BONUS:", bonusTests
    if (set(failedTests)-set(todoTests) == set()):
	print "PASSED"
    else:
	print "FAILED"

if __name__ == "__main__":
    test1 = """\
	1..4
	ok 1 - Input file opened
	not ok 2 - First line of the input valid
	ok 3 - Read the rest of the file
	not ok 4 - Summarized correctly # TODO Not written yet
	"""
    test2 = """\
	ok 1
	not ok 2 some description # TODO with a directive
	ok 3 a description only, no directive
	ok 4 # TODO directive only
	ok a description only, no directive
	ok # SKIP only a directive, no description
	ok
	"""
	
    for test in (test1,test2):
	print test
	testResults = tapOutput.parseString(test)
	tallyResults(testResults)
	print

These tests print out:

    
	1..4
	ok 1 - Input file opened
	not ok 2 - First line of the input valid
	ok 3 - Read the rest of the file
	not ok 4 - Summarized correctly # TODO Not written yet
	
Failed tests: [2, 4]
TODO: [4]
FAILED

	ok 1
	not ok 2 some description # TODO with a directive
	ok 3 a description only, no directive
	ok 4 # TODO directive only
	ok a description only, no directive
	ok # SKIP only a directive, no description
	ok
	
Failed tests: [2]
SKIPPED: [6]
TODO: [2, 4]
BONUS: [4]
PASSED

You are absolutely right, pyparsing is mostly an uber-verbose regex. But that verbosity really helps when the time comes to go back and make changes or enhancements to a parser that you wrote months ago.