"{ Package: 'stx:goodies/petitparser/compiler/tests' }"

"{ NameSpace: Smalltalk }"

TestCase subclass:#PEGFsaScannerIntegrationTest

	instanceVariableNames:'fsa fsaGenerator parser scanner result compiled'

	classVariableNames:''

	poolDictionaries:''

	category:'PetitCompiler-Tests-Scanner'

!

!PEGFsaScannerIntegrationTest methodsFor:'as yet unclassified'!

compile

    | ppcTree |

    compiled ifTrue: [ ^ self ].

    ppcTree := parser asCompilerTree.

    fsa := ppcTree asFsa.

    fsa name: #nextToken.

    fsa finalStates do: [ :s | s isFailure ifFalse: [s retval: #token ]].

    scanner := ((PPCScannerCodeGenerator new)

        generate: fsa).

    compiled := true

!

failScan: stream

    self compile.

    scanner initialize.

    scanner stream: stream asPetitStream. 

    result := scanner nextToken.

    self assert: result isEmpty

!

scan: stream token: token

    self scan: stream token: token position: stream size.

!

scan: stream token: token position: position

    self compile.

    scanner initialize.

    scanner stream: stream asPetitStream. 

    result := scanner nextToken.

    self assert: result isCollection description: 'no collection returned as a result!!'.

    self assert: (result isEmpty not) description: 'no token found'.

    self assert: (result at: token) = position.

!

setUp

    compiled := false.

    fsaGenerator := PEGFsaGenerator new.

!

testA

    parser := 'a' asParser.

    self compile.

    self assert: fsa isDeterministic.

    self assert: fsa isWithoutEpsilons.	

    self failScan: ''.

    self failScan: 'b'.

    self scan: 'a' token: #token position: 1.

    self scan: 'aaa' token: #token position: 1.

!

testAAA_Aplusnot

    parser := 'aaa' asParser not, $a asParser plus.

    self compile.

    self assert: fsa isDeterministic.

    self assert: fsa isWithoutEpsilons.	

    self scan: 'a' token: #token.

    self scan: 'aa' token: #token.

    self failScan: ''.

    self failScan: 'aaa'.

    self failScan: 'aaaa'.

    self failScan: 'aaaaa'.

!

testAAplus_A

    parser := 'aa' asParser plus, $a asParser.

    self scan: 'aaa' token: #token.

    self scan: 'aaaaa' token: #token.

    self failScan: 'a'.

    self failScan: 'aa'.

    self failScan: 'aaaa'.

!

testAAplus_B

    parser := 'aa' asParser plus, $b asParser.

    self scan: 'aab' token: #token.

    self scan: 'aaaab' token: #token.

    self failScan: 'ab'.

    self failScan: 'aaab'.

    self failScan: 'aac'.

!

testAAstar_A

    parser := 'aa' asParser star, $a asParser.

    self scan: 'a' token: #token.

    self scan: 'aaa' token: #token.

    self scan: 'aaaaa' token: #token.

    self scan: 'aaaaaaa' token: #token.

    self failScan: 'aa'.

    self failScan: 'aaaa'.

!

testAAstar_B

    parser := 'aa' asParser star, $b asParser.

    self scan: 'b' token: #token.

    self scan: 'aab' token: #token.

    self scan: 'aaaab' token: #token.

    self scan: 'aaaaaab' token: #token.

    self failScan: 'ab'.

    self failScan: 'aaa'.

!

testAB

    parser := 'ab' asParser.

    self compile.

    self assert: fsa isDeterministic.

    self assert: fsa isWithoutEpsilons.	

    self failScan: ''.

    self failScan: 'b'.

    self scan: 'ab' token: #token position: 2.

    self scan: 'aba' token: #token position: 2.

!

testA_BCorCD_D

    parser := $a asParser, ('bc' asParser / 'cd' asParser), $d asParser.

    self scan: 'abcd' token: #token.

    self scan: 'acdd' token: #token.

    self failScan: 'abdd'.

    self failScan: 'ad'.

    self failScan: 'aacd'.

!

testA_BCorCDplus_D

    parser := $a asParser, ('bc' asParser / 'cd' asParser) plus, $d asParser.

    self scan: 'abcd' token: #token.

    self scan: 'acdd' token: #token.

    self scan: 'abcbccdd' token: #token.

    self scan: 'acdcdbcbcd' token: #token.

    self failScan: 'abdd'.

    self failScan: 'ad'.

    self failScan: 'abcccd'.

!

testA_BCorCDstar_D

    parser := $a asParser, ('bc' asParser / 'cd' asParser) star, $d asParser.

    self scan: 'ad' token: #token.

    self scan: 'abcd' token: #token.

    self scan: 'acdd' token: #token.

    self scan: 'abcbccdd' token: #token.

    self scan: 'acdcdbcbcd' token: #token.

    self failScan: 'abdd'.

    self failScan: 'abcccd'.

!

testA_Bnot

    parser := 'a' asParser, $b asParser not.

    self compile.

    self assert: fsa isDeterministic.

    self assert: fsa isWithoutEpsilons.	

    self failScan: 'ab'.

    self failScan: 'bb'.

    self scan: 'a' token: #token position: 1.

    self scan: 'ac' token: #token position: 1.

!

testA_Boptional

    parser := $a asParser, $b asParser optional.

    self compile.

    self assert: fsa isDeterministic.

    self assert: fsa isWithoutEpsilons.	

    self failScan: ''.

    self failScan: 'b'.

    self scan: 'ab' token: #token position: 2.

    self scan: 'ac' token: #token position: 1.

    self scan: 'a' token: #token position: 1.

!

testA_BorC_D

    parser := $a asParser, ($b asParser / $c asParser), $d asParser.

    self scan: 'abd' token: #token.

    self scan: 'acd' token: #token.

    self failScan: 'a'.

    self failScan: 'abc'.

    self failScan: 'add'.

!

testA_BorCplus_D

    parser := $a asParser, ($b asParser / $c asParser) plus, $d asParser.

    self scan: 'abd' token: #token.

    self scan: 'acd' token: #token.

    self scan: 'abcbcd' token: #token.

    self scan: 'acbcbcd' token: #token.

    self failScan: 'a'.

    self failScan: 'ad'.

    self failScan: 'abc'.

    self failScan: 'aad'.

!

testA_BorCstar_D

    parser := $a asParser, ($b asParser / $c asParser) star, $d asParser.

    self scan: 'ad' token: #token.

    self scan: 'abd' token: #token.

    self scan: 'acd' token: #token.

    self scan: 'abcbcd' token: #token.

    self scan: 'acbcbcd' token: #token.

    self failScan: 'a'.

    self failScan: 'abc'.

    self failScan: 'aad'.

!

testAorAA

    parser := 'a' asParser / 'aa' asParser.

    self compile.

    self assert: fsa isDeterministic.

    self assert: fsa isWithoutEpsilons.	

    self failScan: ''.

    self failScan: 'b'.

    self scan: 'aa' token: #token position: 1.

    self scan: 'a' token: #token position: 1.

!

testAorAX_X

    parser := ('a' asParser / 'ax' asParser), $x asParser.

    self compile.

    self assert: fsa isDeterministic.

    self assert: fsa isWithoutEpsilons.	

    self scan: 'ax' token: #token position: 2.

    self scan: 'axx' token: #token position: 2.	

    self failScan: 'a'.

    self failScan: 'x'.

    self failScan: ''.		

!

testAorB

    parser := $a asParser / $b asParser.

    self compile.

    self assert: fsa isDeterministic.

    self assert: fsa isWithoutEpsilons.	

    self failScan: ''.

    self failScan: 'c'.

    self scan: 'aa' token: #token position: 1.

    self scan: 'bb' token: #token position: 1.

!

testAplus_B

    parser := $a asParser plus, $b asParser.

    self scan: 'ab' token: #token.

    self scan: 'aab' token: #token.

    self scan: 'aaab' token: #token.

    self failScan: 'b'.

    self failScan: 'ac'.

!

testAstar_A

    parser := $a asParser star, $a asParser.

    self failScan: 'a'.

    self failScan: 'aa'.

    self failScan: 'ac'.

!

testAstar_B

    parser := $a asParser star, $b asParser.

    self scan: 'b' token: #token.

    self scan: 'ab' token: #token.

    self scan: 'aab' token: #token.

    self failScan: ''.

    self failScan: 'ac'.

!

testAstar_Bnot

    parser := 'a' asParser star, $b asParser not.

    self compile.

    self assert: fsa isDeterministic.

    self assert: fsa isWithoutEpsilons.	

    self failScan: 'aaab'.

    self failScan: 'b'.

    self scan: '' token: #token position: 0.

    self scan: 'a' token: #token position: 1.

    self scan: 'aac' token: #token position: 2.

    self scan: 'aaaac' token: #token position: 4.

!

testFoo

    parser := 'foo' asParser.

    self scan: 'foo' token: #token.

    self scan: 'foobar' token: #token position: 3.

    self failScan: 'bar'.

    self failScan: 'fo'.

!

testNumber

    parser := #digit asParser plus.

    self compile.

    self assert: fsa isDeterministic.

    self assert: fsa isWithoutEpsilons.	

    self failScan: ''.

    self failScan: 'b'.

    self scan: '12' token: #token position: 2.

    self scan: '2312' token: #token position: 4.

!

testSmalltalkIdentifier

    parser := #letter asParser, #word asParser star, $: asParser not.

    self compile.

    self assert: fsa isDeterministic.

    self assert: fsa isWithoutEpsilons.	

    self scan: 'a' token: #token.

    self scan: 'hithere' token: #token.

    self scan: 'hi123' token: #token.

    self failScan: ''.

    self failScan: 'aaa:'.

    self failScan: '123'.	

! !