From 07221d401de8d633cc58fec8c313617b0716ef1d Mon Sep 17 00:00:00 2001
From: Rose Hogenson <rhogenson@posteo.net>
Date: Thu, 3 Mar 2022 13:51:41 -0800
Subject: Write the compiler frontend.

---
 csc/compiler.csc | 89 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 89 insertions(+)
 create mode 100644 csc/compiler.csc

(limited to 'csc/compiler.csc')

diff --git a/csc/compiler.csc b/csc/compiler.csc
new file mode 100644
index 0000000..5486344
--- /dev/null
+++ b/csc/compiler.csc
@@ -0,0 +1,89 @@
+(define-library (csc compiler)
+  (import (scheme base)
+          (only (csc hash-map)
+            hash-bytevector
+            insert
+            key-not-found-error?
+            lookup
+            make-map
+            map-for-each
+            merge)
+          (only (csc ir1)
+            toplevel-define?)
+          (only (csc loop)
+            loop
+            return)
+          (only (csc macros)
+            builtins-environment
+            expand)
+          (only (csc match) match))
+  (begin
+
+
+    (define (hash-symbol s)
+      (hash-bytevector (string->utf8 (symbol->string s))))
+
+
+    (define (cmp-symbol s1 s2)
+      (string<? (symbol->string s1) (symbol->string s2)))
+
+
+    (define-record-type <syntax-error>
+      (make-syntax-error msg irritants)
+      syntax-error?)
+
+
+    (define (raise-syntax-error msg . irritants)
+      (raise (make-syntax-error msg irritants)))
+
+
+    (define (parse-import-set expr)
+      (match expr
+        (((! 'only) import-set . idents)
+          (loop with bindings = (parse-import-set import-set)
+                and new-bindings = (make-map hash-symbol cmp-symbol)
+                for ident in idents
+                do (set! new-bindings
+                     (insert new-bindings
+                       ident
+                       (guard (e ((key-not-found-error? e) (raise-syntax-error "unknown symbol in `only' form" ident)))
+                         (lookup bindings ident))))
+                finally (return new-bindings)))
+        ((! '(csc builtins))
+          builtins-environment)
+        (_ (raise-syntax-error "unknown or unsupported import set form" expr))))
+          
+
+
+    (define (parse-import expr)
+      (match expr
+        (((! 'import) . import-sets)
+          (loop with bindings = (make-map hash-symbol cmp-symbol)
+                for import-set in import-sets
+                do (set! bindings
+                     (merge bindings (parse-import-set import-set)))
+                finally (return bindings)))
+        (_ (raise-syntax-error "expected import form" expr))))
+
+
+    (define-record-type <program>
+      (make-program imports body)
+      program?
+      (imports program-imports)
+      (body program-body))
+
+
+    ; A Scheme program consists of one or more import declarations
+    ; followed by a sequence of expressions and definitions.
+    ;   -- R7RS
+    (define (program->ir1 program)
+      (loop with bindings = (make-map hash-symbol cmp-symbol)
+            for body on program
+            for expr = (car body)
+            do (match expr
+                 (((! 'import) . import-sets)
+                   (set! bindings
+                     (merge bindings (parse-import expr))))
+                 (_ (loop for expr in body
+                          for expanded-expr = (expand expr bindings)
+                          if (toplevel-define? expanded-expr)
-- 
cgit v1.3.1