Skip to main content

max / pter

4.7 KB · 203 lines History Blame Raw
1 use pter::convert;
2
3 #[test]
4 fn empty_string() {
5 assert_eq!(convert(""), "");
6 }
7
8 #[test]
9 fn whitespace_only() {
10 assert_eq!(convert(" \n\t "), "");
11 }
12
13 #[test]
14 fn just_tags_no_content() {
15 assert_eq!(convert("<div><p><span></span></p></div>"), "");
16 }
17
18 #[test]
19 fn deeply_nested_divs() {
20 let mut html = String::new();
21 for _ in 0..100 {
22 html.push_str("<div>");
23 }
24 html.push_str("deep content");
25 for _ in 0..100 {
26 html.push_str("</div>");
27 }
28 let md = convert(&html);
29 assert!(md.contains("deep content"));
30 }
31
32 #[test]
33 fn deeply_nested_blockquotes() {
34 let mut html = String::new();
35 for _ in 0..20 {
36 html.push_str("<blockquote>");
37 }
38 html.push_str("very deep");
39 for _ in 0..20 {
40 html.push_str("</blockquote>");
41 }
42 let md = convert(&html);
43 assert!(md.contains("very deep"));
44 // Should have many > prefixes
45 assert!(md.contains("> > > > >"));
46 }
47
48 #[test]
49 fn deeply_nested_lists() {
50 let mut html = String::new();
51 for _ in 0..10 {
52 html.push_str("<ul><li>");
53 }
54 html.push_str("deep item");
55 for _ in 0..10 {
56 html.push_str("</li></ul>");
57 }
58 let md = convert(&html);
59 assert!(md.contains("deep item"));
60 }
61
62 #[test]
63 fn malformed_unclosed_tags() {
64 // html5ever auto-corrects these
65 let md = convert("<p>unclosed paragraph<p>another one");
66 assert!(md.contains("unclosed paragraph"));
67 assert!(md.contains("another one"));
68 }
69
70 #[test]
71 fn malformed_mismatched_tags() {
72 let md = convert("<b><i>crossed</b></i>");
73 assert!(md.contains("crossed"));
74 }
75
76 #[test]
77 fn only_script_content() {
78 assert_eq!(convert("<script>alert('xss')</script>"), "");
79 }
80
81 #[test]
82 fn only_style_content() {
83 assert_eq!(convert("<style>.x { color: red; }</style>"), "");
84 }
85
86 #[test]
87 fn only_tracking_pixels() {
88 let html = r#"
89 <img src="a.gif" width="1" height="1">
90 <img src="b.gif" width="1" height="1">
91 "#;
92 assert_eq!(convert(html), "");
93 }
94
95 #[test]
96 fn unicode_content() {
97 let md = convert("<p>日本語テスト 🎉 émojis café</p>");
98 assert!(md.contains("日本語テスト"));
99 assert!(md.contains("🎉"));
100 assert!(md.contains("café"));
101 }
102
103 #[test]
104 fn html_entities_numeric() {
105 let md = convert("<p>&#169; &#8212; &#x2019;</p>");
106 assert!(md.contains("©"));
107 assert!(md.contains("—"));
108 }
109
110 #[test]
111 fn large_input_doesnt_blow_up() {
112 let para = "<p>Hello world. This is a test paragraph with some content.</p>";
113 let html: String = para.repeat(1000);
114 let md = convert(&html);
115 assert!(md.contains("Hello world"));
116 // Should be proportional, not quadratic
117 assert!(md.len() < html.len());
118 }
119
120 #[test]
121 fn link_with_nested_formatting() {
122 let html = r#"<a href="https://example.com"><strong>bold link</strong></a>"#;
123 let md = convert(html);
124 assert!(md.contains("[**bold link**](https://example.com)"));
125 }
126
127 #[test]
128 fn image_with_no_alt() {
129 let md = convert(r#"<img src="photo.jpg">"#);
130 assert!(md.contains("![](photo.jpg)"));
131 }
132
133 #[test]
134 fn consecutive_inline_elements() {
135 let md = convert("<b>bold</b><i>italic</i><code>code</code>");
136 assert_eq!(md, "**bold***italic*`code`");
137 }
138
139 #[test]
140 fn table_with_empty_cells() {
141 let html = "<table><tr><th>A</th><th>B</th></tr>\
142 <tr><td></td><td>val</td></tr></table>";
143 let md = convert(html);
144 assert!(md.contains("| A | B |"));
145 assert!(md.contains("| | val |"));
146 }
147
148 #[test]
149 fn pre_with_html_inside() {
150 let html = "<pre>&lt;div&gt;not a tag&lt;/div&gt;</pre>";
151 let md = convert(html);
152 assert!(md.contains("```"));
153 assert!(md.contains("<div>not a tag</div>"));
154 }
155
156 #[test]
157 fn multiple_spaces_in_source() {
158 let md = convert("<p>word1 word2 word3</p>");
159 assert_eq!(md, "word1 word2 word3");
160 }
161
162 #[test]
163 fn newlines_in_source_collapsed() {
164 let md = convert("<p>line1\n\n\nline2</p>");
165 assert_eq!(md, "line1 line2");
166 }
167
168 #[test]
169 fn full_html_document() {
170 let html = r#"
171 <!DOCTYPE html>
172 <html lang="en">
173 <head>
174 <meta charset="UTF-8">
175 <title>Test Email</title>
176 <style>body { font-family: sans-serif; }</style>
177 </head>
178 <body>
179 <p>Hello!</p>
180 </body>
181 </html>
182 "#;
183 let md = convert(html);
184 assert_eq!(md, "Hello!");
185 }
186
187 #[test]
188 fn data_uri_image_not_tracking_pixel() {
189 // A data URI image that's not 1x1 should render
190 let html = r#"<img src="data:image/png;base64,iVBOR..." alt="inline" width="100">"#;
191 let md = convert(html);
192 assert!(md.contains("![inline]"));
193 }
194
195 #[test]
196 fn blockquote_with_paragraphs() {
197 let html = "<blockquote><p>First para</p><p>Second para</p></blockquote>";
198 let md = convert(html);
199 assert!(md.contains("> First para"));
200 assert!(md.contains("> "));
201 assert!(md.contains("> Second para"));
202 }
203