From 90f4eb831fd6f1e8a293851f1b8f2ed5238f5e58 Mon Sep 17 00:00:00 2001
From: Linnnus <linnnus@users.noreply.github.com>
Date: Fri, 16 Feb 2024 22:56:49 +0100
Subject: fix(creole): Add special case for "." at end of raw URL

---
 src/creole-test.c |  6 ++++++
 src/creole.c      | 33 ++++++++++++++++++++++++++++-----
 2 files changed, 34 insertions(+), 5 deletions(-)

diff --git a/src/creole-test.c b/src/creole-test.c
index 76d62af..ee29e54 100644
--- a/src/creole-test.c
+++ b/src/creole-test.c
@@ -107,6 +107,12 @@ struct {
 		.output  =  "<p><a href=\"mailto:quandale@dingle.com\">"
 		            "mailto:quandale@dingle.com</a></p>"
 	},
+	{ // This test captures a non-standard (?) special case in the parser.
+		.name    =  "Raw URL followed by full stop",
+		.input   =  "My favorite website is https://wiki.c2.com/.",
+		.output  =  "<p>My favorite website is <a href=\"https://wiki.c2.com/\">"
+		            "https://wiki.c2.com/</a>.</p>"
+	},
 	{
 		.name    =  "Unnamed URL",
 		.input   =  "[[http //example.com/examplepage]]",
diff --git a/src/creole.c b/src/creole.c
index 107b50b..462dc50 100644
--- a/src/creole.c
+++ b/src/creole.c
@@ -199,12 +199,28 @@ int do_raw_url(const char *begin, const char *end, bool new_block, FILE *out)
 	}
 	p += 1;
 
-	// Eat the remainder of the URI.
-	// This is not technically correct, but it's a good enough heuristic.
-	const char *q = p;
-	while (q < end && !isspace(*q)) {
-		q += 1;
+        // Eat the remainder of the URI, purely going by what "legal" URI
+        // characters it contains.
+	// See: <https://stackoverflow.com/a/7109208>
+        const char *q = p;
+	while (q < end) {
+		switch (*q) {
+			case '0' ... '9':
+			case 'a' ... 'z':
+			case 'A' ... 'Z':
+			case '-': case '.': case '_': case '~':
+			case ':': case '/': case '?': case '#':
+			case '[': case ']': case '@': case '!':
+			case '$': case '&': case '\'': case '(':
+			case ')': case '*': case '+': case ',':
+			case ';': case '%': case '=':
+				q += 1;
+				break;
+			default:
+				goto end_url;
+		}
 	}
+end_url:
 
         // If there is nothing following the colon, don't accept it as a raw
         // url. Otherwise we'd incorrectly find a link with the "said" protocol
@@ -213,6 +229,13 @@ int do_raw_url(const char *begin, const char *end, bool new_block, FILE *out)
 		return 0;
 	}
 
+        // Special case: If we end on a ".", assume it's a full stop at the end
+        // of a sentence. Here's an example:
+	// My favorite webside is https://cohost.org/.
+        if (q[-1] == '.') {
+		q -= 1;
+	}
+
 	fputs("<a href=\"", out);
 	hprint(out, begin, q);
 	fputs("\">", out);
-- 
cgit v1.2.3