[tablify] / src / lib.rs

use std::collections::{HashMap, HashSet};
use std::io::BufRead;
use std::iter::Iterator;

const HEADER: &str = "<!DOCTYPE html>
<html>
<head>
  <meta charset=\"utf-8\">
  <meta name=\"viewport\" content=\"width=device-width, initial-scale=1\">
  <style>
    /* h/t https://wabain.github.io/2019/10/13/css-rotated-table-header.html */
    th, td { white-space: nowrap; }
    th { text-align: left; font-weight: normal; }
    table { border-collapse: collapse }
    tr.key > th { height: 8em; vertical-align: bottom; line-height: 1 }
    tr.key > th > div { width: 1em; }
    tr.key > th > div > div { width: 5em; transform-origin: bottom left; transform: translateX(1em) rotate(-65deg) }
    td { border: thin solid gray; }
    td.numeric { text-align: right; }
    td.yes { border: thin solid gray; background-color: gray; }
    td.spacer { border: none; }
    /* h/t https://stackoverflow.com/questions/5687035/css-bolding-some-text-without-changing-its-containers-size/46452396#46452396 */
    .highlight { text-shadow: -0.06ex 0 black, 0.06ex 0 black; }
    img { height: 1.2em; }
  </style>
  <script>
    function highlight(id)       { const e = document.getElementById(id); if (e) { e.classList.add(   \"highlight\"); } }
    function clear_highlight(id) { const e = document.getElementById(id); if (e) { e.classList.remove(\"highlight\"); } }
    function h2(a, b)  {       highlight(a);       highlight(b); }
    function ch2(a, b) { clear_highlight(a); clear_highlight(b); }
  </script>
</head>
<body>
  <table>
    <tbody>";
const FOOTER: &str = "    </tbody>
  </table>
</body>
</html>";

#[derive(Debug, PartialEq, Eq)]
struct RowInput {
    label: String,
    entries: Vec<String>,
}

struct Reader<Input: Iterator<Item = Result<String, std::io::Error>>> {
    input: std::iter::Enumerate<Input>,
    row: Option<RowInput>,
}
impl<Input: Iterator<Item = Result<String, std::io::Error>>> Reader<Input> {
    fn new(input: Input) -> Self {
        Self {
            input: input.enumerate(),
            row: None,
        }
    }
}
impl<Input: Iterator<Item = Result<String, std::io::Error>>> Iterator for Reader<Input> {
    type Item = Result<RowInput, std::io::Error>;
    fn next(&mut self) -> Option<Self::Item> {
        loop {
            match self
                .input
                .next()
                .map(|(n, r)| (n, r.map(|line| String::from(line.trim_end()))))
            {
                None => return Ok(std::mem::take(&mut self.row)).transpose(),
                Some((_, Err(e))) => return Some(Err(e)),
                Some((_, Ok(line))) if line.is_empty() && self.row.is_some() => {
                    return Ok(std::mem::take(&mut self.row)).transpose()
                }
                Some((_, Ok(line))) if line.is_empty() => {}
                Some((n, Ok(line))) if line.starts_with(' ') => match &mut self.row {
                    None => {
                        return Some(Err(std::io::Error::other(format!(
                            "{}: Entry with no header",
                            n + 1
                        ))))
                    }
                    Some(ref mut row) => row.entries.push(String::from(line.trim())),
                },
                Some((_, Ok(line))) => {
                    let prev = std::mem::take(&mut self.row);
                    self.row = Some(RowInput {
                        label: line,
                        entries: vec![],
                    });
                    if prev.is_some() {
                        return Ok(prev).transpose();
                    }
                }
            }
        }
    }
}

fn read_rows(input: impl std::io::Read) -> impl Iterator<Item = Result<RowInput, std::io::Error>> {
    Reader::new(std::io::BufReader::new(input).lines())
}

fn column_counts(rows: &[RowInput]) -> Vec<(usize, String)> {
    let mut counts: Vec<_> = rows
        .iter()
        .flat_map(|r| r.entries.iter().collect::<HashSet<_>>().into_iter())
        .fold(HashMap::new(), |mut cs, e| {
            cs.entry(String::from(e))
                .and_modify(|n| *n += 1)
                .or_insert(1);
            cs
        })
        .into_iter()
        .map(|(col, n)| (n, col))
        .collect();
    counts.sort();
    counts
}

/// # Errors
///
/// Will return `Err` if
///   * there's an i/o error while reading `input`
///   * the log has invalid syntax:
///     * an indented line with no preceding non-indented line
pub fn tablify(input: impl std::io::Read) -> Result<String, std::io::Error> {
    let rows = read_rows(input).collect::<Result<Vec<_>, _>>()?;
    let _columns = column_counts(&rows);
    Ok(String::from(HEADER) + "Hello, world!" + FOOTER)
}

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

    #[test]
    fn test_read_rows() {
        assert_eq!(
            read_rows(&b"foo"[..]).flatten().collect::<Vec<_>>(),
            vec![RowInput {
                label: String::from("foo"),
                entries: vec![]
            }]
        );
        assert_eq!(
            read_rows(&b"bar"[..]).flatten().collect::<Vec<_>>(),
            vec![RowInput {
                label: String::from("bar"),
                entries: vec![]
            }]
        );
        assert_eq!(
            read_rows(&b"foo\nbar\n"[..]).flatten().collect::<Vec<_>>(),
            vec![
                RowInput {
                    label: String::from("foo"),
                    entries: vec![]
                },
                RowInput {
                    label: String::from("bar"),
                    entries: vec![]
                }
            ]
        );
        assert_eq!(
            read_rows(&b"foo\n bar\n"[..]).flatten().collect::<Vec<_>>(),
            vec![RowInput {
                label: String::from("foo"),
                entries: vec![String::from("bar")]
            }]
        );
        assert_eq!(
            read_rows(&b"foo\n bar\n baz\n"[..])
                .flatten()
                .collect::<Vec<_>>(),
            vec![RowInput {
                label: String::from("foo"),
                entries: vec![String::from("bar"), String::from("baz")]
            }]
        );
        assert_eq!(
            read_rows(&b"foo\n\nbar\n"[..])
                .flatten()
                .collect::<Vec<_>>(),
            vec![
                RowInput {
                    label: String::from("foo"),
                    entries: vec![]
                },
                RowInput {
                    label: String::from("bar"),
                    entries: vec![]
                }
            ]
        );
        assert_eq!(
            read_rows(&b"foo\n \nbar\n"[..])
                .flatten()
                .collect::<Vec<_>>(),
            vec![
                RowInput {
                    label: String::from("foo"),
                    entries: vec![]
                },
                RowInput {
                    label: String::from("bar"),
                    entries: vec![]
                }
            ]
        );
        assert_eq!(
            read_rows(&b"foo  \n bar  \n"[..])
                .flatten()
                .collect::<Vec<_>>(),
            vec![RowInput {
                label: String::from("foo"),
                entries: vec![String::from("bar")]
            }]
        );

        let bad = read_rows(&b" foo"[..]).next().unwrap();
        assert!(bad.is_err());
        assert!(format!("{bad:?}").contains("1: Entry with no header"));

        let bad2 = read_rows(&b"foo\n\n bar"[..]).nth(1).unwrap();
        assert!(bad2.is_err());
        assert!(format!("{bad2:?}").contains("3: Entry with no header"));
    }

    #[test]
    fn test_column_counts() {
        assert_eq!(
            column_counts(
                &read_rows(&b"foo\n bar\n baz\n"[..])
                    .collect::<Result<Vec<_>, _>>()
                    .unwrap()
            ),
            vec![(1, String::from("bar")), (1, String::from("baz"))]
        );
        assert_eq!(
            column_counts(
                &read_rows(&b"foo\n bar\n baz\nquux\n baz"[..])
                    .collect::<Result<Vec<_>, _>>()
                    .unwrap()
            ),
            vec![(1, String::from("bar")), (2, String::from("baz"))]
        );
        assert_eq!(
            column_counts(
                &read_rows(&b"foo\n bar\n bar\n baz\n bar\nquux\n baz"[..])
                    .collect::<Result<Vec<_>, _>>()
                    .unwrap()
            ),
            vec![(1, String::from("bar")), (2, String::from("baz"))]
        );
    }
}
Commit	Line	Data
397ef957	1	use std::collections::{HashMap, HashSet};
9dfa98b7	2	use std::io::BufRead;
75bb888a SW	3	use std::iter::Iterator;
75bb888a SW	4
cc2378d5 SW	5	const HEADER: &str = "<!DOCTYPE html>
	6	<html>
	7	<head>
	8	<meta charset=\"utf-8\">
	9	<meta name=\"viewport\" content=\"width=device-width, initial-scale=1\">
	10	<style>
	11	/* h/t https://wabain.github.io/2019/10/13/css-rotated-table-header.html */
	12	th, td { white-space: nowrap; }
	13	th { text-align: left; font-weight: normal; }
	14	table { border-collapse: collapse }
	15	tr.key > th { height: 8em; vertical-align: bottom; line-height: 1 }
	16	tr.key > th > div { width: 1em; }
	17	tr.key > th > div > div { width: 5em; transform-origin: bottom left; transform: translateX(1em) rotate(-65deg) }
	18	td { border: thin solid gray; }
	19	td.numeric { text-align: right; }
	20	td.yes { border: thin solid gray; background-color: gray; }
	21	td.spacer { border: none; }
	22	/* h/t https://stackoverflow.com/questions/5687035/css-bolding-some-text-without-changing-its-containers-size/46452396#46452396 */
	23	.highlight { text-shadow: -0.06ex 0 black, 0.06ex 0 black; }
	24	img { height: 1.2em; }
	25	</style>
	26	<script>
	27	function highlight(id) { const e = document.getElementById(id); if (e) { e.classList.add( \"highlight\"); } }
	28	function clear_highlight(id) { const e = document.getElementById(id); if (e) { e.classList.remove(\"highlight\"); } }
	29	function h2(a, b) { highlight(a); highlight(b); }
	30	function ch2(a, b) { clear_highlight(a); clear_highlight(b); }
	31	</script>
	32	</head>
	33	<body>
	34	<table>
	35	<tbody>";
	36	const FOOTER: &str = " </tbody>
	37	</table>
	38	</body>
	39	</html>";
	40
75bb888a SW	41	#[derive(Debug, PartialEq, Eq)]
	42	struct RowInput {
	43	label: String,
	44	entries: Vec<String>,
	45	}
	46
201b9ef3	47	struct Reader<Input: Iterator<Item = Result<String, std::io::Error>>> {
8110b492	48	input: std::iter::Enumerate<Input>,
201b9ef3 SW	49	row: Option<RowInput>,
	50	}
	51	impl<Input: Iterator<Item = Result<String, std::io::Error>>> Reader<Input> {
201b9ef3	52	fn new(input: Input) -> Self {
8110b492 SW	53	Self {
	54	input: input.enumerate(),
	55	row: None,
	56	}
201b9ef3 SW	57	}
	58	}
	59	impl<Input: Iterator<Item = Result<String, std::io::Error>>> Iterator for Reader<Input> {
	60	type Item = Result<RowInput, std::io::Error>;
	61	fn next(&mut self) -> Option<Self::Item> {
	62	loop {
1f6bd845 SW	63	match self
	64	.input
	65	.next()
8110b492	66	.map(\|(n, r)\| (n, r.map(\|line\| String::from(line.trim_end()))))
1f6bd845	67	{
201b9ef3	68	None => return Ok(std::mem::take(&mut self.row)).transpose(),
8110b492 SW	69	Some((_, Err(e))) => return Some(Err(e)),
8110b492 SW	70	Some((_, Ok(line))) if line.is_empty() && self.row.is_some() => {
201b9ef3 SW	71	return Ok(std::mem::take(&mut self.row)).transpose()
201b9ef3 SW	72	}
8110b492 SW	73	Some((_, Ok(line))) if line.is_empty() => {}
	74	Some((n, Ok(line))) if line.starts_with(' ') => match &mut self.row {
	75	None => {
	76	return Some(Err(std::io::Error::other(format!(
	77	"{}: Entry with no header",
	78	n + 1
	79	))))
	80	}
201b9ef3 SW	81	Some(ref mut row) => row.entries.push(String::from(line.trim())),
201b9ef3 SW	82	},
8110b492	83	Some((_, Ok(line))) => {
201b9ef3 SW	84	let prev = std::mem::take(&mut self.row);
	85	self.row = Some(RowInput {
	86	label: line,
	87	entries: vec![],
	88	});
	89	if prev.is_some() {
	90	return Ok(prev).transpose();
	91	}
	92	}
	93	}
	94	}
	95	}
	96	}
	97
201b9ef3 SW	98	fn read_rows(input: impl std::io::Read) -> impl Iterator<Item = Result<RowInput, std::io::Error>> {
201b9ef3 SW	99	Reader::new(std::io::BufReader::new(input).lines())
75bb888a SW	100	}
75bb888a SW	101
58b5f36d SW	102	fn column_counts(rows: &[RowInput]) -> Vec<(usize, String)> {
	103	let mut counts: Vec<_> = rows
	104	.iter()
397ef957	105	.flat_map(\|r\| r.entries.iter().collect::<HashSet<_>>().into_iter())
58b5f36d SW	106	.fold(HashMap::new(), \|mut cs, e\| {
	107	cs.entry(String::from(e))
	108	.and_modify(\|n\| *n += 1)
f272e502	109	.or_insert(1);
58b5f36d	110	cs
f272e502	111	})
58b5f36d SW	112	.into_iter()
	113	.map(\|(col, n)\| (n, col))
	114	.collect();
	115	counts.sort();
	116	counts
f272e502 SW	117	}
f272e502 SW	118
4b99fb70 SW	119	/// # Errors
	120	///
	121	/// Will return `Err` if
	122	/// * there's an i/o error while reading `input`
	123	/// * the log has invalid syntax:
	124	/// * an indented line with no preceding non-indented line
	125	pub fn tablify(input: impl std::io::Read) -> Result<String, std::io::Error> {
	126	let rows = read_rows(input).collect::<Result<Vec<_>, _>>()?;
	127	let _columns = column_counts(&rows);
cc2378d5	128	Ok(String::from(HEADER) + "Hello, world!" + FOOTER)
ece97615	129	}
75bb888a SW	130
	131	#[cfg(test)]
	132	mod tests {
	133	use super::*;
	134
	135	#[test]
	136	fn test_read_rows() {
	137	assert_eq!(
201b9ef3	138	read_rows(&b"foo"[..]).flatten().collect::<Vec<_>>(),
75bb888a SW	139	vec![RowInput {
	140	label: String::from("foo"),
	141	entries: vec![]
	142	}]
	143	);
9dfa98b7	144	assert_eq!(
201b9ef3	145	read_rows(&b"bar"[..]).flatten().collect::<Vec<_>>(),
9dfa98b7 SW	146	vec![RowInput {
	147	label: String::from("bar"),
	148	entries: vec![]
	149	}]
	150	);
2aa9ef94	151	assert_eq!(
201b9ef3	152	read_rows(&b"foo\nbar\n"[..]).flatten().collect::<Vec<_>>(),
2aa9ef94 SW	153	vec![
	154	RowInput {
	155	label: String::from("foo"),
	156	entries: vec![]
	157	},
	158	RowInput {
	159	label: String::from("bar"),
	160	entries: vec![]
	161	}
	162	]
	163	);
201b9ef3 SW	164	assert_eq!(
	165	read_rows(&b"foo\n bar\n"[..]).flatten().collect::<Vec<_>>(),
	166	vec![RowInput {
	167	label: String::from("foo"),
	168	entries: vec![String::from("bar")]
	169	}]
	170	);
	171	assert_eq!(
	172	read_rows(&b"foo\n bar\n baz\n"[..])
	173	.flatten()
	174	.collect::<Vec<_>>(),
	175	vec![RowInput {
	176	label: String::from("foo"),
	177	entries: vec![String::from("bar"), String::from("baz")]
	178	}]
	179	);
	180	assert_eq!(
	181	read_rows(&b"foo\n\nbar\n"[..])
	182	.flatten()
	183	.collect::<Vec<_>>(),
	184	vec![
	185	RowInput {
	186	label: String::from("foo"),
	187	entries: vec![]
	188	},
	189	RowInput {
	190	label: String::from("bar"),
	191	entries: vec![]
	192	}
	193	]
	194	);
1f6bd845 SW	195	assert_eq!(
	196	read_rows(&b"foo\n \nbar\n"[..])
	197	.flatten()
	198	.collect::<Vec<_>>(),
	199	vec![
	200	RowInput {
	201	label: String::from("foo"),
	202	entries: vec![]
	203	},
	204	RowInput {
	205	label: String::from("bar"),
	206	entries: vec![]
	207	}
	208	]
	209	);
	210	assert_eq!(
	211	read_rows(&b"foo \n bar \n"[..])
	212	.flatten()
	213	.collect::<Vec<_>>(),
	214	vec![RowInput {
	215	label: String::from("foo"),
	216	entries: vec![String::from("bar")]
	217	}]
	218	);
201b9ef3 SW	219
	220	let bad = read_rows(&b" foo"[..]).next().unwrap();
	221	assert!(bad.is_err());
8110b492	222	assert!(format!("{bad:?}").contains("1: Entry with no header"));
201b9ef3 SW	223
	224	let bad2 = read_rows(&b"foo\n\n bar"[..]).nth(1).unwrap();
	225	assert!(bad2.is_err());
8110b492	226	assert!(format!("{bad2:?}").contains("3: Entry with no header"));
75bb888a	227	}
f272e502 SW	228
	229	#[test]
	230	fn test_column_counts() {
	231	assert_eq!(
	232	column_counts(
	233	&read_rows(&b"foo\n bar\n baz\n"[..])
	234	.collect::<Result<Vec<_>, _>>()
	235	.unwrap()
	236	),
58b5f36d	237	vec![(1, String::from("bar")), (1, String::from("baz"))]
f272e502 SW	238	);
	239	assert_eq!(
	240	column_counts(
	241	&read_rows(&b"foo\n bar\n baz\nquux\n baz"[..])
	242	.collect::<Result<Vec<_>, _>>()
	243	.unwrap()
	244	),
58b5f36d	245	vec![(1, String::from("bar")), (2, String::from("baz"))]
f272e502	246	);
397ef957 SW	247	assert_eq!(
	248	column_counts(
	249	&read_rows(&b"foo\n bar\n bar\n baz\n bar\nquux\n baz"[..])
	250	.collect::<Result<Vec<_>, _>>()
	251	.unwrap()
	252	),
58b5f36d	253	vec![(1, String::from("bar")), (2, String::from("baz"))]
397ef957	254	);
f272e502	255	}
75bb888a	256	}