strip leading whitespace from triple-quoted strings (closes #70)

For example,

    s = """
        a
         b
        """

is now equivalent to "a\n b".

Author: nolta

base/exports.jl

@@ -1100,6 +1100,8 @@ export
 
 # Macros
     @str,
+    @mstr,
+    @imstr,
     @I_str,
     @E_str,
     @B_str,

base/string.jl

@@ -652,9 +652,53 @@ function interp_parse_bytes(s::String)
     interp_parse(s, unescape_string, writer)
 end
 
+## multiline strings ##
+
+function multiline_lstrip(s::String)
+    if length(s) == 0 || !isspace(s[1])
+        return s
+    end
+    lines = split(s, '\n')
+
+    # trim leading,trailing whitespace
+    a,b = 1,length(lines)
+    if b == 1 return s end
+    if lstrip(lines[a]) == "" a += 1 end
+    if lstrip(lines[b]) == "" b -= 1 end
+    if a > b return s end
+
+    # find prefix
+    first_line = lines[a]
+    n = 0
+    for c in first_line
+        if isspace(c)
+            n += 1
+        else
+            break
+        end
+    end
+    prefix = (n == 0) ? "" : first_line[1:n]
+
+    # output string
+    prefix_len = length(prefix)
+    buf = memio(length(s) - (b-a+1)*prefix_len, false)
+    for i = a:b
+        line = lines[i]
+        if begins_with(line, prefix)
+            print(buf, line[prefix_len+1:end])
+        else
+            print(buf, line)
+        end
+        if i != b print(buf, '\n') end
+    end
+    takebuf_string(buf)
+end
+
 ## core string macros ##
 
 macro   str(s); interp_parse(s); end
+macro  mstr(s); multiline_lstrip(s); end
+macro imstr(s); interp_parse(multiline_lstrip(s)); end
 macro I_str(s); interp_parse(s, x->unescape_chars(x,"\"")); end
 macro E_str(s); check_utf8(unescape_string(s)); end
 macro B_str(s); interp_parse_bytes(s); end

src/julia-parser.scm

@@ -800,16 +800,19 @@
 		   (if (and (symbol? ex) (not (operator? ex))
 			    (not (ts:space? s)))
 		       ;; custom prefixed string literals, x"s" => @x_str "s"
-		       (let ((str (begin (take-token s)
-					 (parse-string-literal s)))
-			     (macname (symbol (string #\@ ex '_str))))
-			 (let ((nxt (peek-token s)))
-			   (if (and (symbol? nxt) (not (operator? nxt))
-				    (not (ts:space? s)))
-			       ;; string literal suffix, "s"x
-			       (loop `(macrocall ,macname ,(car str)
-						 ,(string (take-token s))))
-			       (loop `(macrocall ,macname ,(car str))))))
+                       (let* ((str (begin (take-token s)
+                                          (parse-string-literal s)))
+                              (nxt (peek-token s))
+                              (macname (symbol (string #\@ ex '_str)))
+                              (macstr (if (triplequote-string-literal? str)
+                                          `(macrocall @mstr ,(car str))
+                                          (car str))))
+                         (if (and (symbol? nxt) (not (operator? nxt))
+                                  (not (ts:space? s)))
+                             ;; string literal suffix, "s"x
+                             (loop `(macrocall ,macname ,macstr
+                                               ,(string (take-token s))))
+                             (loop `(macrocall ,macname ,macstr))))
 		       ex))
 		  (else ex))))))))
 
@@ -1312,7 +1315,7 @@
       c))
 
 (define (take-char p)
-    (begin (read-char p) p))
+  (begin (read-char p) p))
 
 ; reads a raw string literal with no processing.
 ; quote can be escaped with \, but the \ is left in place.
@@ -1322,7 +1325,7 @@
     (if (eqv? (peek-char p) #\")
         (if (eqv? (peek-char (take-char p)) #\")
             (parse-string-literal-3 (take-char p))
-            (cons "" #f))
+            (cons "" (cons #f #f)))
         (parse-string-literal-1 p))))
 
 (define (parse-string-literal-1 p)
@@ -1340,7 +1343,7 @@
 			   (set! interpolate #t))
 		       (write-char (not-eof-3 c) b)))
 		 (loop (read-char p)))))
-    (cons (io.tostring! b) interpolate)))
+    (cons (io.tostring! b) (cons interpolate #f))))
 
 (define (parse-string-literal-3 p)
   (let ((b (open-output-string))
@@ -1366,7 +1369,10 @@
                            (set! interpolate #t))
                        (write-char (not-eof-3 c) b)))
                  (loop (read-char p)))))
-    (cons (io.tostring! b) interpolate)))
+    (cons (io.tostring! b) (cons interpolate #t))))
+
+(define (interpolate-string-literal? s) (cadr s))
+(define (triplequote-string-literal? s) (cddr s))
 
 (define (not-eof-1 c)
   (if (eof-object? c)
@@ -1542,12 +1548,16 @@
 	  ((eqv? t #\")
 	   (take-token s)
 	   (let ((ps (parse-string-literal s)))
-	     (if (cdr ps)
-		 `(macrocall @str ,(car ps))
-		 (let ((str (unescape-string (car ps))))
-		   (if (not (string.isutf8 str))
-		       (error "invalid UTF-8 sequence"))
-		   str))))
+             (if (interpolate-string-literal? ps)
+                 (if (triplequote-string-literal? ps)
+                     `(macrocall @imstr ,(car ps))
+                     `(macrocall @str ,(car ps)))
+                 (let ((str (unescape-string (car ps))))
+                   (if (not (string.isutf8 str))
+                       (error "invalid UTF-8 sequence"))
+                   (if (triplequote-string-literal? ps)
+                       `(macrocall @mstr ,str)
+                       str)))))
 
 	  ;; macro call
 	  ((eqv? t #\@)

src/julia-syntax.scm

@@ -2121,7 +2121,8 @@ So far only the second case can actually occur.
 
 (define (julia-expand-strs e)
   (cond ((not (pair? e))     e)
-	((and (eq? (car e) 'macrocall) (eq? (cadr e) '@str))
+	((and (eq? (car e) 'macrocall) (or (eq? (cadr e) '@str)
+                                           (eq? (cadr e) '@mstr)))
 	 ;; expand macro
 	 (let ((form
 		(apply invoke-julia-macro (cadr e) (cddr e))))

test/strings.jl

@@ -508,3 +508,22 @@ str = "s\u2200"
 @test """ab""c""" == "ab\"\"c"
 @test """ab"\"c""" == "ab\"\"c"
 @test """abc\"""" == "abc\""
+n = 3
+@test """$n""" == "$n"
+@test E"""$n""" == E"$n"
+@test """
+     a
+     b
+
+     c
+    """ == "a\nb\n\nc"
+@test """x
+     a
+    """ == "x\n     a\n    "
+@test """
+     $n
+   """ == "3"
+@test E"""
+     $n
+   """ == E"$n"
+