normalize schema when parsing the IRI

RFC3986 in section 3.1 "Scheme" says that

> Although schemes are case-insensitive, the canonical form is
> lowercase and documents that specify schemes must do so with
> lowercase letters.  An implementation should accept uppercase
> letters as equivalent to lowercase in scheme names (e.g., allow
> "HTTP" as well as "http") for the sake of robustness but should only
> produce lowercase scheme names for consistency.

so we cope with that.  The other possibility would have been to use
strcasecmp instead of strcmp when checking on the protocol, but since
the "case" version, although popular, is not part of any standard
AFAIK I prefer downcasing while parsing and be done with it.

2 files changed, 15 insertions, 4 deletions

diff --git a/iri.c b/iri.c
index 1901dac..1b997dd 100644
--- a/iri.c
+++ b/iri.c
@@ -77,12 +77,19 @@ parse_scheme(struct parser *p)
 		return 0;
 	}
 
-	p->iri++;
-	while (isalnum(*p->iri)
+	do {
+		/* normalize the scheme (i.e. lowercase it)
+		 *
+		 * XXX: since we cannot have good things, tolower
+		 * depends on the LC_CTYPE locale.  The good things is
+		 * that we're sure p->iri points to something in the
+		 * ASCII range, so it shouldn't do weird stuff. */
+		*p->iri = tolower(*p->iri);
+		p->iri++;
+	} while (isalnum(*p->iri)
 	    || *p->iri == '+'
 	    || *p->iri == '-'
-	    || *p->iri == '.')
-		p->iri++;
+	    || *p->iri == '.');
 
 	if (*p->iri != ':') {
 		p->err = "illegal character in scheme";
diff --git a/iri_test.c b/iri_test.c
index 18a1020..e322813 100644
--- a/iri_test.c
+++ b/iri_test.c
@@ -97,6 +97,10 @@ main(void)
 	TEST("gemini:/omarpolo.com", FAIL, empty, "FAIL with invalid marker");
 	TEST("gemini//omarpolo.com", FAIL, empty, "FAIL with invalid marker");
 	TEST("h!!p://omarpolo.com", FAIL, empty, "FAIL with invalid schema");
+	TEST("GEMINI://omarpolo.com",
+	    PASS,
+	    IRI("gemini", "omarpolo.com", "", "", "", ""),
+	    "Schemas are case insensitive.");
 
 	/* authority */
 	TEST("gemini://omarpolo.com",