parse.y: add heredoc <<~ syntax (Feature #9098)

bjmllr · bjmllr · commit 61a35adbf9b2 · 2015-04-20T07:21:33.000-04:00
diff --git a/parse.y b/parse.y
@@ -5177,6 +5177,7 @@ static int parser_tokadd_string(struct parser_params*,int,int,int,long*,rb_encod
 static void parser_tokaddmbc(struct parser_params *parser, int c, rb_encoding *enc);
 static int parser_parse_string(struct parser_params*,NODE*);
 static int parser_here_document(struct parser_params*,NODE*);
+static VALUE parser_heredoc_dedent(VALUE);
 
 
 # define nextc()                      parser_nextc(parser)
@@ -5699,6 +5700,7 @@ rb_parser_compile_file_path(volatile VALUE vparser, VALUE fname, VALUE file, int
 #define STR_FUNC_QWORDS 0x08
 #define STR_FUNC_SYMBOL 0x10
 #define STR_FUNC_INDENT 0x20
+#define STR_FUNC_DEDENT 0x40
 
 enum string_type {
     str_squote = (0),
@@ -5725,6 +5727,10 @@ parser_str_new(const char *p, long n, rb_encoding *enc, int func, rb_encoding *e
 	}
     }
 
+    if (func & STR_FUNC_DEDENT) {
+        return parser_heredoc_dedent(str);
+    }
+
     return str;
 }
 
@@ -6478,7 +6484,11 @@ parser_heredoc_identifier(struct parser_params *parser)
     if (c == '-') {
 	c = nextc();
 	func = STR_FUNC_INDENT;
+    } else if (c == '~') {
+        c = nextc();
+        func = STR_FUNC_INDENT | STR_FUNC_DEDENT;
     }
+    
     switch (c) {
       case '\'':
 	func |= str_squote; goto quoted;
@@ -6502,7 +6512,9 @@ parser_heredoc_identifier(struct parser_params *parser)
       default:
 	if (!parser_is_identchar()) {
 	    pushback(c);
-	    if (func & STR_FUNC_INDENT) {
+	    if (func & STR_FUNC_DEDENT) {
+		pushback('~');
+	    } else if (func & STR_FUNC_INDENT) {
 		pushback('-');
 	    }
 	    return 0;
@@ -6550,6 +6562,44 @@ parser_heredoc_restore(struct parser_params *parser, NODE *here)
     ripper_flush(parser);
 }
 
+static VALUE
+parser_heredoc_dedent(VALUE input)
+{
+    char *str = RSTRING_PTR(input), *p, *out_p;
+    int len = RSTRING_LEN(input), indent = len, line_indent = 0, lines = 0;
+    char *end = &str[len];
+    VALUE output;
+
+    p = str;
+    while (p < end) {
+        lines++;
+        line_indent = 0;
+        while (p < end && (*p == ' ' || *p == '\t')) {
+            line_indent++;
+            p++;
+        }
+        if (p < end && line_indent < indent) indent = line_indent;
+        if (indent == 0) break;
+
+        while (p < end && *p != '\r' && *p != '\n') p++;
+        if (p < end && *p == '\r') p++;
+        if (p < end && *p == '\n') p++;
+    }
+
+    output = rb_str_new(0, len - (lines * indent));
+    out_p = RSTRING_PTR(output);
+
+    p = str;
+    while (p < end) {
+        p = p + indent;
+        while (p < end && *p != '\r' && *p != '\n') *out_p++ = *p++;
+        if (p < end && *p == '\r') *out_p++ = *p++;
+        if (p < end && *p == '\n') *out_p++ = *p++;
+    }
+
+    return output;
+}
+
 static int
 parser_whole_match_p(struct parser_params *parser,
     const char *eos, long len, int indent)
diff --git a/test/ruby/test_syntax.rb b/test/ruby/test_syntax.rb
@@ -425,6 +425,20 @@ def test_lineno_command_call_quote
     assert_equal(expected, actual, "#{Bug7559}: ")
   end
 
+  def test_dedented_heredoc_without_indentation
+    assert_equal(" y\nz\n", <<~eos)
+ y
+z
+    eos
+  end
+
+  def test_dedented_heredoc_with_indentation
+    assert_equal(" a\nb\n", <<~eos)
+     a
+    b
+    eos
+  end
+
   def test_lineno_after_heredoc
     bug7559 = '[ruby-dev:46737]'
     expected, _, actual = __LINE__, <<eom, __LINE__
