From: Luke Lau <luke.lau@intel.com>
Date: Tue, 23 Jul 2019 10:13:51 +0000 (+0100)
Subject: Add strings and print primitive
X-Git-Url: https://git.lukelau.me/?p=scheme.git;a=commitdiff_plain;h=703d22df97baba39426f890c6badd8d272e0bf45

Add strings and print primitive
---

diff --git a/codegen.scm b/codegen.scm
index 968a428..a1a0aa7 100644
--- a/codegen.scm
+++ b/codegen.scm
@@ -46,6 +46,25 @@
   (emit "not %rax")
   (emit "andq $1, %rax"))
 
+(define (codegen-print x si env)
+  (codegen-expr x si env) ; x should be a static-string, producing a label
+
+					; get the length of the null terminated string
+  (emit "mov %rax, %rdi")
+  (emit "xor %al, %al")   ; set %al to 0
+  (emit "mov $-1, %rcx") ; max search length = max int = -1
+  (emit "cld")           ; clear direction flag, search up in memory
+  (emit "repne scasb")   ; scan string, %rcx = -strlen - 1 - 1
+  
+  (emit "not %rcx")      ; -%rcx = strlen + 1
+  (emit "dec %rcx")
+  
+  (emit "mov %rcx, %rdx") ; number of bytes
+  (emit "mov %rax, %rsi") ; addr of string
+  (emit "mov $1, %rax") ; file handle 1 (stdout)
+  (emit "mov $1, %rdi") ; syscall 1 (write)
+  (emit "syscall"))
+
 (define (range s n)
   (if (= 0 n) '()
       (append (range s (- n 1))
@@ -175,6 +194,7 @@
 	 ('! (codegen-not (cadr e) si env))
 	 ('= (codegen-eq  (cadr e) (caddr e) si env))
 	 ('bool->int (codegen-expr (cadr e) si env))
+	 ('print (codegen-print (cadr e) si env))
 	 (else (codegen-call callee (cdr e) si env)))))
 
     ('let (codegen-let (let-bindings e)
@@ -186,10 +206,11 @@
 
     ('if (codegen-if (cadr e) (caddr e) (cadddr e) si env))
     
-    ('string-literal (emit "movq ~a, %rax" label))
     ('bool-literal (emit "movq $~a, %rax" (if e 1 0)))
     ('int-literal (emit "movq $~a, %rax" e))
     
+    ('static-string (emit "movq $~a, %rax" (cadr e))) ; move label
+
     (else (error #f "don't know how to codegen this"))))
 
 
@@ -238,7 +259,25 @@
   (let ((transformed (extract program)))
     (cons lambdas transformed)))
 
-;(define (extract-strings program))
+(define (extract-strings program)
+  (let ((cur-string 0)
+	(strings '())) ; assoc list of labels -> string
+    (define (fresh-string)
+      (set! cur-string (+ cur-string 1))
+      (format "string~a" (- cur-string 1)))
+    (define (extract e)
+      (case (ast-type e)
+	('string-literal
+	 (let ((label (fresh-string)))
+	   (set! strings (cons (cons label e) strings))
+	   `(static-string ,label)))
+	(else (ast-traverse extract e))))
+    (let ((transformed (extract program)))
+      (cons strings transformed))))
+
+(define (codegen-string-data s)
+  (emit "~a:" (car s))
+  (emit "\t.string \"~a\"" (cdr s)))
 
 (define (amd64-abi f)
 					; preserve registers
@@ -282,22 +321,30 @@
     (else (format "~a(%rsp)" (- n 6)))))
 
 (define (codegen program)
-  (let* ((extract-result (extract-lambdas program))
-	 (lambdas (car extract-result))
-	 (xform-prog (cdr extract-result)))
-    (emit ".text")
-    (emit ".p2align 4,,15")
+  (let* ((extract-res-0 (extract-strings program))
+	 (strings (car extract-res-0))
+	 (extract-res-1 (extract-lambdas (cdr extract-res-0)))
+	 (lambdas (car extract-res-1))
+	 (xform-prog (cdr extract-res-1)))
+
+    (emit "\t.globl _start")
+    (emit "\t.text")
+;    (emit ".p2align 4,,15") is this needed?
 
     (for-each codegen-lambda lambdas)
 
-    (emit ".globl _start")
+
     (emit "_start:")
     (codegen-expr xform-prog 0 '())
 
 					; exit syscall
     (emit "mov %rax, %rdi")
     (emit "mov $60, %rax")
-    (emit "syscall")))
+    (emit "syscall")
+
+    (emit "\t.data")
+
+    (for-each codegen-string-data strings)))
 
 (define (compile-to-binary program output)
   (when (not (eq? (typecheck program) 'int)) (error #f "not an int"))
diff --git a/sample.scm b/sample.scm
index 75e6043..03b240c 100644
--- a/sample.scm
+++ b/sample.scm
@@ -1 +1 @@
-(let ((a 3) (b 4)) (+ a b))
+(let () (print "hello world!") 0)
diff --git a/tests.scm b/tests.scm
index 563a6e7..7c73433 100644
--- a/tests.scm
+++ b/tests.scm
@@ -24,8 +24,7 @@
   (compile-to-binary prog "/tmp/test-prog")
   (system "/tmp/test-prog > /tmp/test-output.txt")
   (let ((str (read-file "/tmp/test-output.txt")))
-    (test (substring str 0 (- (string-length str) 1))
-	  output)))
+    (test str output)))
 
 (test (typecheck '(lambda (x) (+ ((lambda (y) (x y 3)) 5) 2)))
       '(abs (abs int (abs int int)) int))
@@ -37,4 +36,5 @@
 	      ((lambda (z) (+ 1 z)) (* y x)))
 	   1090)
 (test-prog '(if ((lambda (x) (= x 2)) 1) 0 (- 32 1)) 31)
+(test-prog-stdout '(if (= 3 2) 1 (let () (print "hello world!") 0)) "hello world!")
 
diff --git a/typecheck.scm b/typecheck.scm
index 46fddfe..7eb4fa9 100644
--- a/typecheck.scm
+++ b/typecheck.scm
@@ -10,6 +10,7 @@
   (case t
     ('int #t)
     ('bool #t)
+    ('void #t)
     (else #f)))
 
 (define (pretty-type t)
@@ -74,6 +75,7 @@
     ('! '(abs bool bool))
     ('= '(abs int (abs int bool)))
     ('bool->int '(abs bool int))
+    ('print '(abs string void))
     (else #f)))
 
 ; we typecheck the lambda calculus only (only single arg lambdas)
@@ -89,6 +91,7 @@
 	  (case (ast-type x)
 	   ('int-literal (list '() 'int))
 	   ('bool-literal (list '() 'bool))
+	   ('string-literal (list '() 'string))
 	   ('builtin (list '() (builtin-type x)))
 
 	   ('if