1// Copyright 2011 The Go Authors. All rights reserved.
2// Use of this source code is governed by a BSD-style
3// license that can be found in the LICENSE file.
4
5package template
6
7import (
8	"bytes"
9	"fmt"
10	"strings"
11)
12
13// urlFilter returns its input unless it contains an unsafe scheme in which
14// case it defangs the entire URL.
15//
16// Schemes that cause unintended side effects that are irreversible without user
17// interaction are considered unsafe. For example, clicking on a "javascript:"
18// link can immediately trigger JavaScript code execution.
19//
20// This filter conservatively assumes that all schemes other than the following
21// are unsafe:
22//    * http:   Navigates to a new website, and may open a new window or tab.
23//              These side effects can be reversed by navigating back to the
24//              previous website, or closing the window or tab. No irreversible
25//              changes will take place without further user interaction with
26//              the new website.
27//    * https:  Same as http.
28//    * mailto: Opens an email program and starts a new draft. This side effect
29//              is not irreversible until the user explicitly clicks send; it
30//              can be undone by closing the email program.
31//
32// To allow URLs containing other schemes to bypass this filter, developers must
33// explicitly indicate that such a URL is expected and safe by encapsulating it
34// in a template.URL value.
35func urlFilter(args ...any) string {
36	s, t := stringify(args...)
37	if t == contentTypeURL {
38		return s
39	}
40	if !isSafeURL(s) {
41		return "#" + filterFailsafe
42	}
43	return s
44}
45
46// isSafeURL is true if s is a relative URL or if URL has a protocol in
47// (http, https, mailto).
48func isSafeURL(s string) bool {
49	if protocol, _, ok := strings.Cut(s, ":"); ok && !strings.Contains(protocol, "/") {
50		if !strings.EqualFold(protocol, "http") && !strings.EqualFold(protocol, "https") && !strings.EqualFold(protocol, "mailto") {
51			return false
52		}
53	}
54	return true
55}
56
57// urlEscaper produces an output that can be embedded in a URL query.
58// The output can be embedded in an HTML attribute without further escaping.
59func urlEscaper(args ...any) string {
60	return urlProcessor(false, args...)
61}
62
63// urlNormalizer normalizes URL content so it can be embedded in a quote-delimited
64// string or parenthesis delimited url(...).
65// The normalizer does not encode all HTML specials. Specifically, it does not
66// encode '&' so correct embedding in an HTML attribute requires escaping of
67// '&' to '&'.
68func urlNormalizer(args ...any) string {
69	return urlProcessor(true, args...)
70}
71
72// urlProcessor normalizes (when norm is true) or escapes its input to produce
73// a valid hierarchical or opaque URL part.
74func urlProcessor(norm bool, args ...any) string {
75	s, t := stringify(args...)
76	if t == contentTypeURL {
77		norm = true
78	}
79	var b bytes.Buffer
80	if processURLOnto(s, norm, &b) {
81		return b.String()
82	}
83	return s
84}
85
86// processURLOnto appends a normalized URL corresponding to its input to b
87// and reports whether the appended content differs from s.
88func processURLOnto(s string, norm bool, b *bytes.Buffer) bool {
89	b.Grow(len(s) + 16)
90	written := 0
91	// The byte loop below assumes that all URLs use UTF-8 as the
92	// content-encoding. This is similar to the URI to IRI encoding scheme
93	// defined in section 3.1 of  RFC 3987, and behaves the same as the
94	// EcmaScript builtin encodeURIComponent.
95	// It should not cause any misencoding of URLs in pages with
96	// Content-type: text/html;charset=UTF-8.
97	for i, n := 0, len(s); i < n; i++ {
98		c := s[i]
99		switch c {
100		// Single quote and parens are sub-delims in RFC 3986, but we
101		// escape them so the output can be embedded in single
102		// quoted attributes and unquoted CSS url(...) constructs.
103		// Single quotes are reserved in URLs, but are only used in
104		// the obsolete "mark" rule in an appendix in RFC 3986
105		// so can be safely encoded.
106		case '!', '#', '$', '&', '*', '+', ',', '/', ':', ';', '=', '?', '@', '[', ']':
107			if norm {
108				continue
109			}
110		// Unreserved according to RFC 3986 sec 2.3
111		// "For consistency, percent-encoded octets in the ranges of
112		// ALPHA (%41-%5A and %61-%7A), DIGIT (%30-%39), hyphen (%2D),
113		// period (%2E), underscore (%5F), or tilde (%7E) should not be
114		// created by URI producers
115		case '-', '.', '_', '~':
116			continue
117		case '%':
118			// When normalizing do not re-encode valid escapes.
119			if norm && i+2 < len(s) && isHex(s[i+1]) && isHex(s[i+2]) {
120				continue
121			}
122		default:
123			// Unreserved according to RFC 3986 sec 2.3
124			if 'a' <= c && c <= 'z' {
125				continue
126			}
127			if 'A' <= c && c <= 'Z' {
128				continue
129			}
130			if '0' <= c && c <= '9' {
131				continue
132			}
133		}
134		b.WriteString(s[written:i])
135		fmt.Fprintf(b, "%%%02x", c)
136		written = i + 1
137	}
138	b.WriteString(s[written:])
139	return written != 0
140}
141
142// Filters and normalizes srcset values which are comma separated
143// URLs followed by metadata.
144func srcsetFilterAndEscaper(args ...any) string {
145	s, t := stringify(args...)
146	switch t {
147	case contentTypeSrcset:
148		return s
149	case contentTypeURL:
150		// Normalizing gets rid of all HTML whitespace
151		// which separate the image URL from its metadata.
152		var b bytes.Buffer
153		if processURLOnto(s, true, &b) {
154			s = b.String()
155		}
156		// Additionally, commas separate one source from another.
157		return strings.ReplaceAll(s, ",", "%2c")
158	}
159
160	var b bytes.Buffer
161	written := 0
162	for i := 0; i < len(s); i++ {
163		if s[i] == ',' {
164			filterSrcsetElement(s, written, i, &b)
165			b.WriteString(",")
166			written = i + 1
167		}
168	}
169	filterSrcsetElement(s, written, len(s), &b)
170	return b.String()
171}
172
173// Derived from https://play.golang.org/p/Dhmj7FORT5
174const htmlSpaceAndASCIIAlnumBytes = "\x00\x36\x00\x00\x01\x00\xff\x03\xfe\xff\xff\x07\xfe\xff\xff\x07"
175
176// isHTMLSpace is true iff c is a whitespace character per
177// https://infra.spec.whatwg.org/#ascii-whitespace
178func isHTMLSpace(c byte) bool {
179	return (c <= 0x20) && 0 != (htmlSpaceAndASCIIAlnumBytes[c>>3]&(1<<uint(c&0x7)))
180}
181
182func isHTMLSpaceOrASCIIAlnum(c byte) bool {
183	return (c < 0x80) && 0 != (htmlSpaceAndASCIIAlnumBytes[c>>3]&(1<<uint(c&0x7)))
184}
185
186func filterSrcsetElement(s string, left int, right int, b *bytes.Buffer) {
187	start := left
188	for start < right && isHTMLSpace(s[start]) {
189		start++
190	}
191	end := right
192	for i := start; i < right; i++ {
193		if isHTMLSpace(s[i]) {
194			end = i
195			break
196		}
197	}
198	if url := s[start:end]; isSafeURL(url) {
199		// If image metadata is only spaces or alnums then
200		// we don't need to URL normalize it.
201		metadataOk := true
202		for i := end; i < right; i++ {
203			if !isHTMLSpaceOrASCIIAlnum(s[i]) {
204				metadataOk = false
205				break
206			}
207		}
208		if metadataOk {
209			b.WriteString(s[left:start])
210			processURLOnto(url, true, b)
211			b.WriteString(s[end:right])
212			return
213		}
214	}
215	b.WriteString("#")
216	b.WriteString(filterFailsafe)
217}
218