test-c-tokenize.scm

(use gauche.test)
(use lang.c.parser)

(define (test-c-tokenize str expect)
  (test* (string-append "\"" str "\"") expect (c-tokenize (string->list str))))

(test-section "CARM 2.3 tokens")
(test-c-tokenize "forwhile;" '((ident forwhile) |;|))
(test-c-tokenize "forx;"     '((ident forx) |;|))
(test-c-tokenize "foox;"     '((ident foox) |;|))
(test-c-tokenize "xfor;"     '((ident xfor) |;|))
(test-c-tokenize "foo;"      '((ident foo) |;|))
(test-c-tokenize "b>x;"      '((ident b) > (ident  x) |;|))
(test-c-tokenize "b->x;"     '((ident b) -> (ident x) |;|))
(test-c-tokenize "b--x;"     '((ident b) -- (ident x) |;|))
(test-c-tokenize "b---x;"    '((ident b) --  - (ident x) |;|))

(test-section "integer constants")
(test-c-tokenize "0L;"      '((literal long-int     "0L") |;|))
(test-c-tokenize "0l;"      '((literal long-int     "0l") |;|))
(test-c-tokenize "0U;"      '((literal unsigned-int "0U") |;|))
(test-c-tokenize "0u;"      '((literal unsigned-int "0u") |;|))
(test-c-tokenize "1234;"    '((literal int          "1234") |;|))
(test-c-tokenize "012;"     '((literal int          "012" ) |;|))
(test-c-tokenize "0x12;"    '((literal int          "0x12") |;|))

(test-section "character constants")
(test-c-tokenize "'a';"     '((char (#\a))      |;|))
(test-c-tokenize "'A';"     '((char (#\A))      |;|))
(test-c-tokenize "' ';"     '((char (#\space))  |;|))
(test-c-tokenize "'?';"     '((char (#\?))      |;|))
(test-c-tokenize "'\\r';"   '((char (#\return)) |;|))
(test-c-tokenize "'\\0';"   '((char (#\null))   |;|))
(test-c-tokenize "'\"';"    '((char (#\"))      |;|))
(test-c-tokenize "'\\377';" '((char (#\xff))    |;|))
(test-c-tokenize "'%';"     '((char (#\%))      |;|))
(test-c-tokenize "'\\23';"  '((char (#\x13))    |;|))
(test-c-tokenize "'8';"     '((char (#\8))      |;|))
(test-c-tokenize "'\\\\';"  '((char (#\\))      |;|))
(test-c-tokenize "'ABCD';"  '((char (#\A #\B #\C #\D)) |;|))

(test-section "floating point constants")
(test-c-tokenize "0.;"       '((literal double "0."     )       |;|))
(test-c-tokenize "3e1;"      '((literal double "3e1"    )       |;|))
(test-c-tokenize "3.14159;"  '((literal double "3.14159")       |;|))
(test-c-tokenize ".0;"       '((literal double ".0"     )       |;|))
(test-c-tokenize "1.0E-3;"   '((literal double "1.0E-3" )       |;|))
(test-c-tokenize "1e-3;"     '((literal double "1e-3"   )       |;|))
(test-c-tokenize "1.0;"      '((literal double "1.0"    )       |;|))
(test-c-tokenize "0.00034;"  '((literal double "0.00034")       |;|))
(test-c-tokenize "2e+9;"     '((literal double "2e+9"   )       |;|))
(test-c-tokenize "1.0f;"     '((literal float  "1.0f"   )       |;|))
(test-c-tokenize "1.0e67L;"  '((literal long-double "1.0e67L")  |;|))
(test-c-tokenize "1.37E+6L;" '((literal long-double "1.37E+6L") |;|))
(test-c-tokenize "0E1L;"     '((literal long-double "0E1L"   )  |;|))
(test-c-tokenize "0x1.0p1;"  '((literal double "0x1.0p1")       |;|))
(test-c-tokenize "0x1.0;"    '((literal double "0x1.0")         |;|))

(test-section "string constants")
(test-c-tokenize "\"abra\""       '((string "abra")))
(test-c-tokenize "\"\";"          '((string "")     |;|))
(test-c-tokenize "\"\\\"\";"      '((string "\"")   |;|))
(test-c-tokenize "\"Copyright 2000 \\\nTexas Instruments. \""
           '((string "\"Copyright 2000 \\\nTexas Instruments. \"")))

(test-section "other")
(test-c-tokenize "X++Y;"     '((ident X) ++ (ident Y)  |;|))
(test-c-tokenize "-12ul;"    '(- (literal unsigned-long-int  "12ul") |;|))
(test-c-tokenize "x**2;"     '((ident x) * * (literal int  "2") |;|))
(test-c-tokenize "A*=B;"     '((ident A) *= (ident B)  |;|))