[rbdr/page] / src / gemini_parser.rs

#[derive(PartialEq, Eq, Debug)]
pub enum GeminiLine {
    Text(String, bool),
    PreformattedToggle(bool, String),
    Heading(u8, String),
    Link(String, String),
    Quote(String),
    ListItem(String),
}

/// Parses gemtext source code into a vector of `GeminiLine` elements.
///
/// # Arguments
/// * `source` - A string slice that contains the gemtext
///
/// # Returns
/// A `Vec<GeminiLine>` containing the rendered HTML.
pub fn parse(source: &str) -> Vec<GeminiLine> {
    source
        .lines()
        .fold((Vec::new(), false), |(mut lines, is_preformatted), line| {
            let parsed = if is_preformatted {
                parse_preformatted_line(line)
            } else {
                parse_line(line)
            };

            let new_is_preformatted = match parsed {
                GeminiLine::PreformattedToggle(x, _) => x,
                _ => is_preformatted,
            };

            lines.push(parsed);
            (lines, new_is_preformatted)
        })
        .0
}

fn parse_preformatted_line(line: &str) -> GeminiLine {
    match line {
        s if s.starts_with("```") => GeminiLine::PreformattedToggle(false, String::new()),
        _ => GeminiLine::Text(line.to_string(), true),
    }
}

fn parse_line(line: &str) -> GeminiLine {
    match line {
        s if s.starts_with("###") => GeminiLine::Heading(3, s[3..].to_string()),
        s if s.starts_with("##") => GeminiLine::Heading(2, s[2..].to_string()),
        s if s.starts_with('#') => GeminiLine::Heading(1, s[1..].to_string()),
        s if s.starts_with("=>") => {
            let content = s[2..].trim();
            match content.split_once(char::is_whitespace) {
                Some((url, text)) => {
                    GeminiLine::Link(url.trim().to_string(), text.trim().to_string())
                }
                None => GeminiLine::Link(content.trim().to_string(), String::new()),
            }
        }
        s if s.starts_with("* ") => GeminiLine::ListItem(s[2..].to_string()),
        s if s.starts_with('>') => GeminiLine::Quote(s[1..].to_string()),
        s if s.starts_with("```") => GeminiLine::PreformattedToggle(true, s[3..].to_string()),
        _ => GeminiLine::Text(line.to_string(), false),
    }
}

#[cfg(test)]
mod tests {
    use super::*;

    #[test]
    fn test_headings() {
        assert_eq!(
            parse_line("### Heading"),
            GeminiLine::Heading(3, " Heading".to_string())
        );
        assert_eq!(
            parse_line("## Heading"),
            GeminiLine::Heading(2, " Heading".to_string())
        );
        assert_eq!(
            parse_line("# Heading"),
            GeminiLine::Heading(1, " Heading".to_string())
        );
        assert_eq!(parse_line("###"), GeminiLine::Heading(3, "".to_string()));
        assert_eq!(
            parse_line("#####"),
            GeminiLine::Heading(3, "##".to_string())
        );
        assert_eq!(parse_line("# "), GeminiLine::Heading(1, " ".to_string()));

        assert_eq!(
            parse_preformatted_line("### Heading"),
            GeminiLine::Text("### Heading".to_string(), true)
        );
        assert_eq!(
            parse_preformatted_line("## Heading"),
            GeminiLine::Text("## Heading".to_string(), true)
        );
        assert_eq!(
            parse_preformatted_line("# Heading"),
            GeminiLine::Text("# Heading".to_string(), true)
        );
    }

    #[test]
    fn test_links() {
        assert_eq!(
            parse_line("=> https://example.com Link text"),
            GeminiLine::Link("https://example.com".to_string(), "Link text".to_string())
        );
        assert_eq!(
            parse_line("=> /local/path"),
            GeminiLine::Link("/local/path".to_string(), "".to_string())
        );

        assert_eq!(
            parse_line("=>"),
            GeminiLine::Link("".to_string(), "".to_string())
        );
        assert_eq!(
            parse_line("=> "),
            GeminiLine::Link("".to_string(), "".to_string())
        );
        assert_eq!(
            parse_line("=>  multiple    spaces  in   text"),
            GeminiLine::Link("multiple".to_string(), "spaces  in   text".to_string())
        );

        assert_eq!(
            parse_preformatted_line("=> https://example.com Link text"),
            GeminiLine::Text("=> https://example.com Link text".to_string(), true)
        );
    }

    #[test]
    fn test_list_items() {
        assert_eq!(
            parse_line("* List item"),
            GeminiLine::ListItem("List item".to_string())
        );

        assert_eq!(parse_line("* "), GeminiLine::ListItem("".to_string()));
        assert_eq!(parse_line("*"), GeminiLine::Text("*".to_string(), false));
        assert_eq!(
            parse_line("*WithText"),
            GeminiLine::Text("*WithText".to_string(), false)
        );
        assert_eq!(
            parse_line("*  Multiple spaces"),
            GeminiLine::ListItem(" Multiple spaces".to_string())
        );
    }

    #[test]
    fn test_quotes() {
        assert_eq!(
            parse_line(">Quote text"),
            GeminiLine::Quote("Quote text".to_string())
        );

        assert_eq!(parse_line(">"), GeminiLine::Quote("".to_string()));
        assert_eq!(parse_line("> "), GeminiLine::Quote(" ".to_string()));
        assert_eq!(
            parse_line(">>Nested"),
            GeminiLine::Quote(">Nested".to_string())
        );
    }

    #[test]
    fn test_preformatted() {
        assert_eq!(
            parse_line("```alt-text"),
            GeminiLine::PreformattedToggle(true, "alt-text".to_string())
        );

        assert_eq!(
            parse_line("```"),
            GeminiLine::PreformattedToggle(true, "".to_string())
        );
        assert_eq!(
            parse_line("``` "),
            GeminiLine::PreformattedToggle(true, " ".to_string())
        );
        assert_eq!(
            parse_line("````"),
            GeminiLine::PreformattedToggle(true, "`".to_string())
        );

        assert_eq!(
            parse_preformatted_line("```alt-text"),
            GeminiLine::PreformattedToggle(false, "".to_string())
        );
        assert_eq!(
            parse_preformatted_line("```"),
            GeminiLine::PreformattedToggle(false, "".to_string())
        );
    }

    #[test]
    fn test_text() {
        // Normal case
        assert_eq!(
            parse_line("Regular text"),
            GeminiLine::Text("Regular text".to_string(), false)
        );

        // Edge cases
        assert_eq!(parse_line(""), GeminiLine::Text("".to_string(), false));
        assert_eq!(parse_line(" "), GeminiLine::Text(" ".to_string(), false));
        assert_eq!(parse_line("  "), GeminiLine::Text("  ".to_string(), false));
    }

    #[test]
    fn test_malformed_input() {
        assert_eq!(
            parse_line("= >Not a link"),
            GeminiLine::Text("= >Not a link".to_string(), false)
        );
        assert_eq!(
            parse_line("``Not preformatted"),
            GeminiLine::Text("``Not preformatted".to_string(), false)
        );
        assert_eq!(
            parse_line("** Not a list"),
            GeminiLine::Text("** Not a list".to_string(), false)
        );
    }

    #[test]
    fn test_full_document() {
        let input = "\
# Heading 1
## Heading 2
### Heading 3
Regular text
=> https://example.com Link text
* List item
>Quote
```alt
code
# Heading 1
## Heading 2
### Heading 3
=> https://example.com Link text
* List item
>Quote
```trailing alt";
        let result = parse(input);
        assert_eq!(
            result,
            vec![
                GeminiLine::Heading(1, " Heading 1".to_string()),
                GeminiLine::Heading(2, " Heading 2".to_string()),
                GeminiLine::Heading(3, " Heading 3".to_string()),
                GeminiLine::Text("Regular text".to_string(), false),
                GeminiLine::Link("https://example.com".to_string(), "Link text".to_string()),
                GeminiLine::ListItem("List item".to_string()),
                GeminiLine::Quote("Quote".to_string()),
                GeminiLine::PreformattedToggle(true, "alt".to_string()),
                GeminiLine::Text("code".to_string(), true),
                GeminiLine::Text("# Heading 1".to_string(), true),
                GeminiLine::Text("## Heading 2".to_string(), true),
                GeminiLine::Text("### Heading 3".to_string(), true),
                GeminiLine::Text("=> https://example.com Link text".to_string(), true),
                GeminiLine::Text("* List item".to_string(), true),
                GeminiLine::Text(">Quote".to_string(), true),
                GeminiLine::PreformattedToggle(false, "".to_string()),
            ]
        );
    }
}
Commit	Line	Data
260e8ec6 RBR	1	#[derive(PartialEq, Eq, Debug)]
	2	pub enum GeminiLine {
	3	Text(String, bool),
	4	PreformattedToggle(bool, String),
	5	Heading(u8, String),
	6	Link(String, String),
	7	Quote(String),
5732d284	8	ListItem(String),
8d4fac52 RBR	9	}
8d4fac52 RBR	10
8766e441	11	/// Parses gemtext source code into a vector of `GeminiLine` elements.
5732d284	12	///
260e8ec6 RBR	13	/// # Arguments
260e8ec6 RBR	14	/// * `source` - A string slice that contains the gemtext
5732d284	15	///
260e8ec6 RBR	16	/// # Returns
	17	/// A `Vec<GeminiLine>` containing the rendered HTML.
	18	pub fn parse(source: &str) -> Vec<GeminiLine> {
5732d284 RBR	19	source
	20	.lines()
	21	.fold((Vec::new(), false), \|(mut lines, is_preformatted), line\| {
	22	let parsed = if is_preformatted {
	23	parse_preformatted_line(line)
	24	} else {
	25	parse_line(line)
	26	};
	27
	28	let new_is_preformatted = match parsed {
	29	GeminiLine::PreformattedToggle(x, _) => x,
	30	_ => is_preformatted,
	31	};
	32
	33	lines.push(parsed);
	34	(lines, new_is_preformatted)
	35	})
260e8ec6	36	.0
6a8515fe RBR	37	}
6a8515fe RBR	38
260e8ec6 RBR	39	fn parse_preformatted_line(line: &str) -> GeminiLine {
	40	match line {
	41	s if s.starts_with("```") => GeminiLine::PreformattedToggle(false, String::new()),
	42	_ => GeminiLine::Text(line.to_string(), true),
6a8515fe	43	}
6a8515fe RBR	44	}
6a8515fe RBR	45
260e8ec6 RBR	46	fn parse_line(line: &str) -> GeminiLine {
	47	match line {
	48	s if s.starts_with("###") => GeminiLine::Heading(3, s[3..].to_string()),
	49	s if s.starts_with("##") => GeminiLine::Heading(2, s[2..].to_string()),
8766e441	50	s if s.starts_with('#') => GeminiLine::Heading(1, s[1..].to_string()),
260e8ec6 RBR	51	s if s.starts_with("=>") => {
	52	let content = s[2..].trim();
	53	match content.split_once(char::is_whitespace) {
5732d284 RBR	54	Some((url, text)) => {
	55	GeminiLine::Link(url.trim().to_string(), text.trim().to_string())
	56	}
260e8ec6	57	None => GeminiLine::Link(content.trim().to_string(), String::new()),
f0739225	58	}
5732d284	59	}
260e8ec6	60	s if s.starts_with("* ") => GeminiLine::ListItem(s[2..].to_string()),
8766e441	61	s if s.starts_with('>') => GeminiLine::Quote(s[1..].to_string()),
260e8ec6 RBR	62	s if s.starts_with("```") => GeminiLine::PreformattedToggle(true, s[3..].to_string()),
260e8ec6 RBR	63	_ => GeminiLine::Text(line.to_string(), false),
8d4fac52 RBR	64	}
	65	}
	66
260e8ec6 RBR	67	#[cfg(test)]
	68	mod tests {
	69	use super::*;
	70
	71	#[test]
	72	fn test_headings() {
5732d284 RBR	73	assert_eq!(
	74	parse_line("### Heading"),
	75	GeminiLine::Heading(3, " Heading".to_string())
	76	);
	77	assert_eq!(
	78	parse_line("## Heading"),
	79	GeminiLine::Heading(2, " Heading".to_string())
	80	);
	81	assert_eq!(
	82	parse_line("# Heading"),
	83	GeminiLine::Heading(1, " Heading".to_string())
	84	);
260e8ec6	85	assert_eq!(parse_line("###"), GeminiLine::Heading(3, "".to_string()));
5732d284 RBR	86	assert_eq!(
	87	parse_line("#####"),
	88	GeminiLine::Heading(3, "##".to_string())
	89	);
260e8ec6 RBR	90	assert_eq!(parse_line("# "), GeminiLine::Heading(1, " ".to_string()));
260e8ec6 RBR	91
5732d284 RBR	92	assert_eq!(
	93	parse_preformatted_line("### Heading"),
	94	GeminiLine::Text("### Heading".to_string(), true)
	95	);
	96	assert_eq!(
	97	parse_preformatted_line("## Heading"),
	98	GeminiLine::Text("## Heading".to_string(), true)
	99	);
	100	assert_eq!(
	101	parse_preformatted_line("# Heading"),
	102	GeminiLine::Text("# Heading".to_string(), true)
	103	);
260e8ec6 RBR	104	}
	105
	106	#[test]
	107	fn test_links() {
	108	assert_eq!(
	109	parse_line("=> https://example.com Link text"),
	110	GeminiLine::Link("https://example.com".to_string(), "Link text".to_string())
	111	);
	112	assert_eq!(
	113	parse_line("=> /local/path"),
	114	GeminiLine::Link("/local/path".to_string(), "".to_string())
	115	);
	116
	117	assert_eq!(
	118	parse_line("=>"),
	119	GeminiLine::Link("".to_string(), "".to_string())
	120	);
	121	assert_eq!(
	122	parse_line("=> "),
	123	GeminiLine::Link("".to_string(), "".to_string())
	124	);
	125	assert_eq!(
	126	parse_line("=> multiple spaces in text"),
	127	GeminiLine::Link("multiple".to_string(), "spaces in text".to_string())
	128	);
	129
	130	assert_eq!(
	131	parse_preformatted_line("=> https://example.com Link text"),
	132	GeminiLine::Text("=> https://example.com Link text".to_string(), true)
	133	);
	134	}
	135
	136	#[test]
	137	fn test_list_items() {
	138	assert_eq!(
	139	parse_line("* List item"),
	140	GeminiLine::ListItem("List item".to_string())
	141	);
	142
	143	assert_eq!(parse_line("* "), GeminiLine::ListItem("".to_string()));
	144	assert_eq!(parse_line(""), GeminiLine::Text("".to_string(), false));
5732d284 RBR	145	assert_eq!(
	146	parse_line("*WithText"),
	147	GeminiLine::Text("*WithText".to_string(), false)
	148	);
	149	assert_eq!(
	150	parse_line("* Multiple spaces"),
	151	GeminiLine::ListItem(" Multiple spaces".to_string())
	152	);
260e8ec6 RBR	153	}
	154
	155	#[test]
	156	fn test_quotes() {
	157	assert_eq!(
	158	parse_line(">Quote text"),
	159	GeminiLine::Quote("Quote text".to_string())
	160	);
	161
	162	assert_eq!(parse_line(">"), GeminiLine::Quote("".to_string()));
	163	assert_eq!(parse_line("> "), GeminiLine::Quote(" ".to_string()));
5732d284 RBR	164	assert_eq!(
	165	parse_line(">>Nested"),
	166	GeminiLine::Quote(">Nested".to_string())
	167	);
260e8ec6 RBR	168	}
	169
	170	#[test]
	171	fn test_preformatted() {
	172	assert_eq!(
	173	parse_line("```alt-text"),
	174	GeminiLine::PreformattedToggle(true, "alt-text".to_string())
	175	);
	176
5732d284 RBR	177	assert_eq!(
	178	parse_line("```"),
	179	GeminiLine::PreformattedToggle(true, "".to_string())
	180	);
	181	assert_eq!(
	182	parse_line("``` "),
	183	GeminiLine::PreformattedToggle(true, " ".to_string())
	184	);
	185	assert_eq!(
	186	parse_line("````"),
	187	GeminiLine::PreformattedToggle(true, "`".to_string())
	188	);
260e8ec6 RBR	189
	190	assert_eq!(
	191	parse_preformatted_line("```alt-text"),
	192	GeminiLine::PreformattedToggle(false, "".to_string())
	193	);
	194	assert_eq!(
	195	parse_preformatted_line("```"),
	196	GeminiLine::PreformattedToggle(false, "".to_string())
	197	);
260e8ec6 RBR	198	}
	199
	200	#[test]
	201	fn test_text() {
	202	// Normal case
	203	assert_eq!(
	204	parse_line("Regular text"),
	205	GeminiLine::Text("Regular text".to_string(), false)
	206	);
	207
	208	// Edge cases
	209	assert_eq!(parse_line(""), GeminiLine::Text("".to_string(), false));
	210	assert_eq!(parse_line(" "), GeminiLine::Text(" ".to_string(), false));
	211	assert_eq!(parse_line(" "), GeminiLine::Text(" ".to_string(), false));
	212	}
	213
	214	#[test]
	215	fn test_malformed_input() {
5732d284 RBR	216	assert_eq!(
	217	parse_line("= >Not a link"),
	218	GeminiLine::Text("= >Not a link".to_string(), false)
	219	);
	220	assert_eq!(
	221	parse_line("``Not preformatted"),
	222	GeminiLine::Text("``Not preformatted".to_string(), false)
	223	);
	224	assert_eq!(
	225	parse_line("** Not a list"),
	226	GeminiLine::Text("** Not a list".to_string(), false)
	227	);
260e8ec6 RBR	228	}
	229
	230	#[test]
	231	fn test_full_document() {
	232	let input = "\
	233	# Heading 1
	234	## Heading 2
	235	### Heading 3
	236	Regular text
	237	=> https://example.com Link text
	238	* List item
	239	>Quote
	240	```alt
	241	code
	242	# Heading 1
	243	## Heading 2
	244	### Heading 3
	245	=> https://example.com Link text
	246	* List item
	247	>Quote
	248	```trailing alt";
	249	let result = parse(input);
5732d284 RBR	250	assert_eq!(
	251	result,
	252	vec![
	253	GeminiLine::Heading(1, " Heading 1".to_string()),
	254	GeminiLine::Heading(2, " Heading 2".to_string()),
	255	GeminiLine::Heading(3, " Heading 3".to_string()),
	256	GeminiLine::Text("Regular text".to_string(), false),
	257	GeminiLine::Link("https://example.com".to_string(), "Link text".to_string()),
	258	GeminiLine::ListItem("List item".to_string()),
	259	GeminiLine::Quote("Quote".to_string()),
	260	GeminiLine::PreformattedToggle(true, "alt".to_string()),
	261	GeminiLine::Text("code".to_string(), true),
	262	GeminiLine::Text("# Heading 1".to_string(), true),
	263	GeminiLine::Text("## Heading 2".to_string(), true),
	264	GeminiLine::Text("### Heading 3".to_string(), true),
	265	GeminiLine::Text("=> https://example.com Link text".to_string(), true),
	266	GeminiLine::Text("* List item".to_string(), true),
	267	GeminiLine::Text(">Quote".to_string(), true),
	268	GeminiLine::PreformattedToggle(false, "".to_string()),
	269	]
	270	);
8d4fac52 RBR	271	}
8d4fac52 RBR	272	}