From 3bb40669d5efea707a90171ad2430a81fd4f4726 Mon Sep 17 00:00:00 2001
From: Fabio Alessandrelli <fabio.alessandrelli@gmail.com>
Date: Mon, 26 Apr 2021 09:33:28 +0200
Subject: [Net] Implement String::parse_url for parsing URLs.

Splits the URL into (scheme, host, port, path).
Supports both literal IPv4 and IPv6.
Strip credentials when present (e.g. http://user:pass@example.com/).

Use that function in both HTTPRequest and WebSocketClient.
---
 core/string/ustring.cpp | 65 +++++++++++++++++++++++++++++++++++++++++++++++++
 core/string/ustring.h   |  1 +
 2 files changed, 66 insertions(+)

(limited to 'core/string')

diff --git a/core/string/ustring.cpp b/core/string/ustring.cpp
index cf0040353d..108f90fd5d 100644
--- a/core/string/ustring.cpp
+++ b/core/string/ustring.cpp
@@ -240,6 +240,71 @@ String String::word_wrap(int p_chars_per_line) const {
 	return ret;
 }
 
+Error String::parse_url(String &r_scheme, String &r_host, int &r_port, String &r_path) const {
+	// Splits the URL into scheme, host, port, path. Strip credentials when present.
+	String base = *this;
+	r_scheme = "";
+	r_host = "";
+	r_port = 0;
+	r_path = "";
+	int pos = base.find("://");
+	// Scheme
+	if (pos != -1) {
+		r_scheme = base.substr(0, pos + 3).to_lower();
+		base = base.substr(pos + 3, base.length() - pos - 3);
+	}
+	pos = base.find("/");
+	// Path
+	if (pos != -1) {
+		r_path = base.substr(pos, base.length() - pos);
+		base = base.substr(0, pos);
+	}
+	// Host
+	pos = base.find("@");
+	if (pos != -1) {
+		// Strip credentials
+		base = base.substr(pos + 1, base.length() - pos - 1);
+	}
+	if (base.begins_with("[")) {
+		// Literal IPv6
+		pos = base.rfind("]");
+		if (pos == -1) {
+			return ERR_INVALID_PARAMETER;
+		}
+		r_host = base.substr(1, pos - 1);
+		base = base.substr(pos + 1, base.length() - pos - 1);
+	} else {
+		// Anything else
+		if (base.get_slice_count(":") > 1) {
+			return ERR_INVALID_PARAMETER;
+		}
+		pos = base.rfind(":");
+		if (pos == -1) {
+			r_host = base;
+			base = "";
+		} else {
+			r_host = base.substr(0, pos);
+			base = base.substr(pos, base.length() - pos);
+		}
+	}
+	if (r_host.is_empty()) {
+		return ERR_INVALID_PARAMETER;
+	}
+	r_host = r_host.to_lower();
+	// Port
+	if (base.begins_with(":")) {
+		base = base.substr(1, base.length() - 1);
+		if (!base.is_valid_integer()) {
+			return ERR_INVALID_PARAMETER;
+		}
+		r_port = base.to_int();
+		if (r_port < 1 || r_port > 65535) {
+			return ERR_INVALID_PARAMETER;
+		}
+	}
+	return OK;
+}
+
 void String::copy_from(const char *p_cstr) {
 	// copy Latin-1 encoded c-string directly
 	if (!p_cstr) {
diff --git a/core/string/ustring.h b/core/string/ustring.h
index 1e362d7683..a56845deff 100644
--- a/core/string/ustring.h
+++ b/core/string/ustring.h
@@ -416,6 +416,7 @@ public:
 	String c_unescape() const;
 	String json_escape() const;
 	String word_wrap(int p_chars_per_line) const;
+	Error parse_url(String &r_scheme, String &r_host, int &r_port, String &r_path) const;
 
 	String property_name_encode() const;
 
-- 
cgit v1.2.3