fmt/pl: add pl_strip_html()

include/re_fmt.h

@@ -58,6 +58,7 @@ const char *pl_strstr(const struct pl *pl, const char *str);
 int      pl_trim(struct pl *pl);
 int      pl_ltrim(struct pl *pl);
 int      pl_rtrim(struct pl *pl);
+void     pl_strip_html(struct pl *pl);
 
 /** Advance pl position/length by +/- N bytes */
 static inline void pl_advance(struct pl *pl, ssize_t n)

src/fmt/pl.c

@@ -902,3 +902,51 @@ int pl_trim(struct pl *pl)
 
 	return err;
 }
+
+
+/**
+ * Strip HTML tags from a pointer-length string in-place
+ *
+ * @param pl Pointer-length string
+ */
+void pl_strip_html(struct pl *pl)
+{
+	if (!pl)
+		return;
+
+	const char *r = pl->p;
+	bool in_tag   = false;
+
+	/* lookup first possible html tag */
+	r = memchr(r, '<', pl->l);
+	if (!r)
+		return;
+
+	char *w = (char *)r;
+
+	/* Reference: https://html.spec.whatwg.org/multipage/parsing.html */
+	for (size_t len = pl->l - (size_t)(r - pl->p); len--; r++) {
+		if (in_tag) {
+			if (*r == '>')
+				in_tag = false;
+			continue;
+		}
+
+		/* 13.2.5.1 Data state */
+		if (*r == '<' && len >= 1) {
+			/* 13.2.5.6 Tag open state */
+			unsigned char n = *(r + 1);
+			if (isalpha(n) || n == '/' || n == '!' || n == '?') {
+				in_tag = true;
+				continue;
+			}
+		}
+
+		unsigned char c = (unsigned char)*r;
+		if (c >= 0x20 || c == '\n' || c == '\r' || c == '\t') {
+			*w++ = *r;
+		}
+	}
+
+	pl->l = (size_t)(w - pl->p);
+}

test/fmt.c

@@ -143,6 +143,14 @@ int test_fmt_pl(void)
 	if (NULL != pl_strstr(&pl1, str0))
 		goto out;
 
+	/* pl_strip_html */
+	struct pl pl_html = PL_INIT;
+	char str_html[]	  = "abc <script>alert(1)</script> <= test <><a";
+	pl_set_str(&pl_html, str_html);
+	pl_strip_html(&pl_html);
+	TEST_EQUALS(23, pl_html.l);
+	TEST_EQUALS(0, pl_strcmp(&pl_html, "abc alert(1) <= test <>"));
+
 	return 0;
  out:
 	return EINVAL;