diff --git a/src/lib.rs b/src/lib.rs
index 28985c1..aee105b 100644
--- a/src/lib.rs
+++ b/src/lib.rs
@@ -206,116 +206,140 @@ pub fn html2text(html: &str) -> String {
#[cfg(test)]
mod tests {
use super::*;
- const cases: &[(&str, &str)] = &[
- ("blah", "blah"),
+
+ macro_rules! test {
+ ($name:ident, $from:literal, $to:literal $(,)?) => {
+ #[test]
+ fn $name() {
+ assert_eq!(&html2text($from), $to);
+ }
+ };
+ ($($name:ident: $from:literal to $to:literal,)* $(,)?) => {
+ $(test!{$name, $from, $to})*
+ };
+ }
+
+ test! {
+ plaintext: "blah" to "blah",
+ tag: "
" to "",
+ tag_contents: "simple text
" to "simple text",
// links
- ("", ""),
- ("simple text
", "simple text"),
- ("click here", "click test"),
- ("click here", "click test"),
- (
- "click here",
- "click ents/'x'",
- ),
- ("click here", "click "),
- (
- "click here or here",
- "click test",
- ),
- (
- "click news",
- "click http://bit.ly/2n4wXRs",
- ),
- ("yet, not yet", "/wiki/yet#English, /wiki/not_yet#English"),
-
+ link:
+ "click here"
+ to "click test",
+ links_ignore_attributes:
+ "click here"
+ to "click test",
+ link_entities_in_url:
+ "click here"
+ to "click ents/'x'",
+ link_javascript:
+ "click here"
+ to "click ",
+ link_ignore_content_tags:
+ "click here or here"
+ to "click test",
+ link_absolute_url:
+ "click news"
+ to "click http://bit.ly/2n4wXRs",
+ link_ignore_attributes_2:
+ "yet, not yet"
+ to "/wiki/yet#English, /wiki/not_yet#English",
// inlines
- ("strong text", "strong text"),
- ("some div
", "some div"),
+ ignore_inline:
+ "strong text"
+ to "strong text",
+ ignore_inline_attributes:
+ "some div
"
+ to "some div",
// lines breaks and spaces
- ("should ignore more spaces", "should ignore more spaces"),
- ("should \nignore \r\nnew lines", "should ignore new lines"),
- ("a\nb\nc", "a b c"),
- ("two
line
breaks", "two\r\nline\r\nbreaks"),
- ("two
paragraphs
", "two\r\n\r\nparagraphs"),
+ collapse_spaces:
+ "should ignore more spaces" to "should ignore more spaces",
+ collapse_linebreaks:
+ "a\nb\nc" to "a b c",
+ collapse_mixed:
+ "should \nignore \r\nnew lines" to "should ignore new lines",
+ br_tag:
+ "two
line
breaks" to "two\r\nline\r\nbreaks",
+ paragraph:
+ "two
paragraphs
" to "two\r\n\r\nparagraphs",
// Headers
- ("First
main text", "First\r\n\r\nmain text"),
- (
- "FirstSecond
next section",
- "First\r\n\r\nSecond\r\n\r\nnext section",
- ),
- ("Second
next section", "Second\r\n\r\nnext section"),
- (
- "SecondThird
next section",
- "Second\r\n\r\nThird\r\n\r\nnext section",
- ),
- ("Third
next section", "Third\r\n\r\nnext section"),
- (
- "ThirdFourth
next section",
- "Third\r\n\r\nFourth\r\n\r\nnext section",
- ),
- ("Fourth
next section", "Fourth\r\n\r\nnext section"),
- (
- "FourthFifth
next section",
- "Fourth\r\n\r\nFifth\r\n\r\nnext section",
- ),
- ("Fifth
next section", "Fifth\r\n\r\nnext section"),
- (
- "FifthSixth
next section",
- "Fifth\r\n\r\nSixth\r\n\r\nnext section",
- ),
- ("Sixth
next section", "Sixth\r\n\r\nnext section"),
- ("Not Headernext section", "Not Headernext section"),
+ h1:
+ "First
main text" to "First\r\n\r\nmain text",
+ h2_inline:
+ "FirstSecond
next section"
+ to "First\r\n\r\nSecond\r\n\r\nnext section",
+ h2:
+ "Second
next section" to "Second\r\n\r\nnext section",
+ h3_inline:
+ "SecondThird
next section"
+ to "Second\r\n\r\nThird\r\n\r\nnext section",
+ h3:
+ "Third
next section" to "Third\r\n\r\nnext section",
+ h4_inline:
+ "ThirdFourth
next section"
+ to "Third\r\n\r\nFourth\r\n\r\nnext section",
+ h4:
+ "Fourth
next section" to "Fourth\r\n\r\nnext section",
+ h5_inline:
+ "FourthFifth
next section"
+ to "Fourth\r\n\r\nFifth\r\n\r\nnext section",
+ h5:
+ "Fifth
next section" to "Fifth\r\n\r\nnext section",
+ h6_inline:
+ "FifthSixth
next section"
+ to "Fifth\r\n\r\nSixth\r\n\r\nnext section",
+ h6:
+ "Sixth
next section" to "Sixth\r\n\r\nnext section",
+ no_h7:
+ "Not Headernext section" to "Not Headernext section",
// html entitites
- ("two spaces", "two spaces"),
- ("© 2017 K3A", "© 2017 K3A"),
- ("<printtag>", ""),
- (
- "would you pay in ¢, £, ¥ or €?",
- "would you pay in ¢, £, ¥ or €?",
- ),
- (
- "Tom & Jerry is not an entity",
- "Tom & Jerry is not an entity",
- ),
- ("this &neither; as you see", "this &neither; as you see"),
- (
- "list of items",
- "list of items\r\nOne\r\nTwo\r\nThree\r\n",
- ),
- ("fish & chips", "fish & chips"),
- (
- ""I'm sorry, Dave. I'm afraid I can't do that." – HAL, 2001: A Space Odyssey",
- "\"I'm sorry, Dave. I'm afraid I can't do that.\" – HAL, 2001: A Space Odyssey",
- ),
- ("Google ®", "Google ®"),
- (
- "⁌ decimal and hex entities supported ⁍",
- "⁌ decimal and hex entities supported ⁍",
- ),
+ entity_nbsp:
+ "two spaces" to "two spaces",
+ entity_copy:
+ "© 2017 K3A" to "© 2017 K3A",
+ entity_tag:
+ "<printtag>" to "",
+ entity_currencies:
+ "would you pay in ¢, £, ¥ or €?"
+ to "would you pay in ¢, £, ¥ or €?",
+ ampersand_not_entity:
+ "Tom & Jerry is not an entity" to "Tom & Jerry is not an entity",
+ entity_unknown:
+ "this &neither; as you see" to "this &neither; as you see",
+ entity_amp:
+ "fish & chips" to "fish & chips",
+ unordered_list:
+ "list of items"
+ to "list of items\r\nOne\r\nTwo\r\nThree\r\n",
+ entity_quot:
+ ""I'm sorry, Dave. I'm afraid I can't do that." – HAL, 2001: A Space Odyssey"
+ to "\"I'm sorry, Dave. I'm afraid I can't do that.\" – HAL, 2001: A Space Odyssey",
+ entity_reg:
+ "Google ®" to "Google ®",
// Large entity
- ("&abcdefghij;", "&abcdefghij;"),
+ entity_large_unknown:
+ "&abcdefghij;" to "&abcdefghij;",
// Numeric HTML entities
- (
- "'single quotes' and 츝",
- "'single quotes' and 츝",
- ),
+ entity_numeric:
+ "⁌ decimal and hex entities supported ⁍"
+ to "⁌ decimal and hex entities supported ⁍",
+ entity_numeric_2:
+ "'single quotes' and 츝"
+ to "'single quotes' and 츝",
// full thml structure
- ("", ""),
- ("Goodx", "x"),
- (
- "we are not interested in scripts",
- "we are not interested in scripts",
- ),
+ empty: "" to "",
+ full_html:
+ "Goodx" to "x",
+ ignore_script:
+ "we are not interested in scripts"
+ to "we are not interested in scripts",
// custom html tags
- ("hello", "hello"),
- ("hello", "hello"),
- ("hello", "hello"),
- ];
-
- #[test]
- fn test_all() {
- for case in cases {
- assert_eq!(&html2text(case.0), case.1);
- }
+ ignore_unknown_tag:
+ "hello" to "hello",
+ ignore_unknown_tag_whitespace:
+ "hello" to "hello",
+ ignore_unknown_tag_attributes:
+ "hello" to "hello",
}
}