Tokenize nesting? HEAD main

1 files changed, 55 insertions, 3 deletions

@@ -10,6 +10,9 @@
 BEGIN {
         nested = 0
         buffer = ""
+        OFS = "\t"
+        STDERR = "/dev/stderr"
+        PRGN = "squawk"
 }
 
 {
@@ -17,15 +20,64 @@ BEGIN {
 }
 
 END {
-        ast = read(buffer)
+        read(buffer, ast)
         eval(ast)
 }
 
 
-function eval(buf)
+function die(message, errcode)
 {
+        eprint(PRGN " ERROR" (message ? ": " message : ""))
+        exit (errcode ? errcode : 1)
 }
 
-function read(buf)
+function eprint(message)
 {
+        # Print MESSAGE to STDERR.
+        print(message) > STDERR
+}
+
+function eval(ast)
+{
+        # Evaluate multi-dimensional array AST.
+        for (w in ast) {
+                print ast[w]
+        }
+}
+
+function read(buf, ast)
+{
+        # Read string BUF into multi-dimensional array AST.
+        split(buf, b, "")
+        w = 1
+        word = ""
+        # Tokenize
+        for (c in b) {
+                # print c, b[c]
+                if (b[c] == "\\") {
+                        word = word b[c++]
+                } else if (b[c] == "(") {
+                        if (word) {
+                                ast[w++] = word
+                                word = ""
+                        }
+                        ast[w++] = "(" nested++
+                } else if (b[c] == ")") {
+                        if (word) {
+                                ast[w++] = word
+                                word = ""
+                        }
+                        ast[w++] = ")" --nested
+                } else if (b[c] ~ /[ \t\r\n\f\v]/) {
+                        if (word) {
+                                ast[w++] = word
+                                word = ""
+                        }
+                } else {
+                        word = word b[c]
+                }
+                if (nested < 0) {
+                        die("Unmatched paren at " c)
+                }
+        }
 }