git: fix parser

Accept ; as a comment starter.  Accept partially or fully quoted values.
Support the backslash escapes allowed by Git.

* src/git-parser.c (input): semicolon starts a comment, too.
(collect_substring): New function.
(collect_string): Rewrite via collect_substring.
(collect_value): New function.
(getvalue): Use collect_value.

1 files changed, 48 insertions, 5 deletions

diff --git a/src/git-parser.c b/src/git-parser.c
index 117a8fc..37eeab7 100644
--- a/src/git-parser.c
+++ b/src/git-parser.c
@@ -58,13 +58,13 @@ rawinput()
 }
 
 static int
 input()
 {
 	rawinput();
-	if (input_char == '#') {
+	if (input_char == '#' || input_char == ';') {
 		while (rawinput() && input_char != '\n')
 			;
 	}
 	return input_char;
 }
 
@@ -93,25 +93,41 @@ collect_tag()
 		grecs_txtacc_grow_char(acc, input_char);
 	grecs_txtacc_grow_char(acc, 0);
 	tok.tag = grecs_txtacc_finish(acc, 0);
 }
 
 static void
-collect_string()
+collect_substring()
 {
 	while (rawinput()) {
 		if (input_char == '\\') {
 			if (!input()) {
 				grecs_error(&grecs_current_locus, 0,
 					    "unexpected EOF in string");
 				break;
 			}
+			switch (input_char) {
+			case 'n':
+				input_char = '\n';
+				break;
+			case 't':
+				input_char = '\t';
+				break;
+			case 'b':
+				input_char = '\b';
+			}
 		} else if (input_char == '"')
 			break;
 		grecs_txtacc_grow_char(acc, input_char);
 	}
+}
+
+static void
+collect_string()
+{
+	collect_substring();
 	grecs_txtacc_grow_char(acc, 0);
 	tok.tag = grecs_txtacc_finish(acc, 0);
 }
 
 static void
 gettoken(void)
@@ -181,23 +197,50 @@ gettoken(void)
 	tok.chbuf[0] = input_char;
 	tok.chbuf[1] = 0;
 	tok.buf = tok.chbuf;
 	tok.type = input_char;
 }
 
+static void
+collect_value()
+{
+	do {
+		if (input_char == '"') {
+			collect_substring();
+			if (input_char == '"')
+				continue;
+			else
+				break;
+		}
+		if (input_char == '\\') {
+			if (!rawinput())
+				break;
+			switch (input_char) {
+			case 'n':
+				input_char = '\n';
+				break;
+			case 't':
+				input_char = '\t';
+				break;
+			case 'b':
+				input_char = '\b';
+			}
+		}
+		grecs_txtacc_grow_char(acc, input_char);
+	} while (input() && input_char != '\n');
+}
+
 static struct grecs_value *
 getvalue()
 {
 	int len;
 	struct grecs_value *val = grecs_malloc(sizeof(*val));
 	while (input() && ISSPACE(input_char) && input_char != '\n')
 		;
 	if (input_char != '\n')
-		do
-			grecs_txtacc_grow_char(acc, input_char);
-		while (input() && input_char != '\n');
+		collect_value();
 	grecs_txtacc_grow_char(acc, 0);
 	tok.type = TOK_VALUE;
 	tok.buf = grecs_txtacc_finish(acc, 1);
 	len = strlen(tok.buf);
 	while (len > 0 && ISSPACE(tok.buf[len-1]))
 		tok.buf[--len] = 0;