[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, Hash)]
struct Entry(String);

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

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(Entry(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.0))
                .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![Entry(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![Entry(String::from("bar")), Entry(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![Entry(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
14e9852b SW	41	#[derive(Debug, PartialEq, Eq, Hash)]
	42	struct Entry(String);
	43
75bb888a SW	44	#[derive(Debug, PartialEq, Eq)]
	45	struct RowInput {
	46	label: String,
14e9852b	47	entries: Vec<Entry>,
75bb888a SW	48	}
75bb888a SW	49
201b9ef3	50	struct Reader<Input: Iterator<Item = Result<String, std::io::Error>>> {
8110b492	51	input: std::iter::Enumerate<Input>,
201b9ef3 SW	52	row: Option<RowInput>,
	53	}
	54	impl<Input: Iterator<Item = Result<String, std::io::Error>>> Reader<Input> {
201b9ef3	55	fn new(input: Input) -> Self {
8110b492 SW	56	Self {
	57	input: input.enumerate(),
	58	row: None,
	59	}
201b9ef3 SW	60	}
	61	}
	62	impl<Input: Iterator<Item = Result<String, std::io::Error>>> Iterator for Reader<Input> {
	63	type Item = Result<RowInput, std::io::Error>;
	64	fn next(&mut self) -> Option<Self::Item> {
	65	loop {
1f6bd845 SW	66	match self
	67	.input
	68	.next()
8110b492	69	.map(\|(n, r)\| (n, r.map(\|line\| String::from(line.trim_end()))))
1f6bd845	70	{
201b9ef3	71	None => return Ok(std::mem::take(&mut self.row)).transpose(),
8110b492 SW	72	Some((_, Err(e))) => return Some(Err(e)),
8110b492 SW	73	Some((_, Ok(line))) if line.is_empty() && self.row.is_some() => {
201b9ef3 SW	74	return Ok(std::mem::take(&mut self.row)).transpose()
201b9ef3 SW	75	}
8110b492 SW	76	Some((_, Ok(line))) if line.is_empty() => {}
	77	Some((n, Ok(line))) if line.starts_with(' ') => match &mut self.row {
	78	None => {
	79	return Some(Err(std::io::Error::other(format!(
	80	"{}: Entry with no header",
	81	n + 1
	82	))))
	83	}
14e9852b	84	Some(ref mut row) => row.entries.push(Entry(String::from(line.trim()))),
201b9ef3	85	},
8110b492	86	Some((_, Ok(line))) => {
201b9ef3 SW	87	let prev = std::mem::take(&mut self.row);
	88	self.row = Some(RowInput {
	89	label: line,
	90	entries: vec![],
	91	});
	92	if prev.is_some() {
	93	return Ok(prev).transpose();
	94	}
	95	}
	96	}
	97	}
	98	}
	99	}
	100
201b9ef3 SW	101	fn read_rows(input: impl std::io::Read) -> impl Iterator<Item = Result<RowInput, std::io::Error>> {
201b9ef3 SW	102	Reader::new(std::io::BufReader::new(input).lines())
75bb888a SW	103	}
75bb888a SW	104
58b5f36d SW	105	fn column_counts(rows: &[RowInput]) -> Vec<(usize, String)> {
	106	let mut counts: Vec<_> = rows
	107	.iter()
397ef957	108	.flat_map(\|r\| r.entries.iter().collect::<HashSet<_>>().into_iter())
58b5f36d	109	.fold(HashMap::new(), \|mut cs, e\| {
14e9852b	110	cs.entry(String::from(&e.0))
58b5f36d	111	.and_modify(\|n\| *n += 1)
f272e502	112	.or_insert(1);
58b5f36d	113	cs
f272e502	114	})
58b5f36d SW	115	.into_iter()
	116	.map(\|(col, n)\| (n, col))
	117	.collect();
	118	counts.sort();
	119	counts
f272e502 SW	120	}
f272e502 SW	121
4b99fb70 SW	122	/// # Errors
	123	///
	124	/// Will return `Err` if
	125	/// * there's an i/o error while reading `input`
	126	/// * the log has invalid syntax:
	127	/// * an indented line with no preceding non-indented line
	128	pub fn tablify(input: impl std::io::Read) -> Result<String, std::io::Error> {
	129	let rows = read_rows(input).collect::<Result<Vec<_>, _>>()?;
	130	let _columns = column_counts(&rows);
cc2378d5	131	Ok(String::from(HEADER) + "Hello, world!" + FOOTER)
ece97615	132	}
75bb888a SW	133
	134	#[cfg(test)]
	135	mod tests {
	136	use super::*;
	137
	138	#[test]
	139	fn test_read_rows() {
	140	assert_eq!(
201b9ef3	141	read_rows(&b"foo"[..]).flatten().collect::<Vec<_>>(),
75bb888a SW	142	vec![RowInput {
	143	label: String::from("foo"),
	144	entries: vec![]
	145	}]
	146	);
9dfa98b7	147	assert_eq!(
201b9ef3	148	read_rows(&b"bar"[..]).flatten().collect::<Vec<_>>(),
9dfa98b7 SW	149	vec![RowInput {
	150	label: String::from("bar"),
	151	entries: vec![]
	152	}]
	153	);
2aa9ef94	154	assert_eq!(
201b9ef3	155	read_rows(&b"foo\nbar\n"[..]).flatten().collect::<Vec<_>>(),
2aa9ef94 SW	156	vec![
	157	RowInput {
	158	label: String::from("foo"),
	159	entries: vec![]
	160	},
	161	RowInput {
	162	label: String::from("bar"),
	163	entries: vec![]
	164	}
	165	]
	166	);
201b9ef3 SW	167	assert_eq!(
	168	read_rows(&b"foo\n bar\n"[..]).flatten().collect::<Vec<_>>(),
	169	vec![RowInput {
	170	label: String::from("foo"),
14e9852b	171	entries: vec![Entry(String::from("bar"))]
201b9ef3 SW	172	}]
	173	);
	174	assert_eq!(
	175	read_rows(&b"foo\n bar\n baz\n"[..])
	176	.flatten()
	177	.collect::<Vec<_>>(),
	178	vec![RowInput {
	179	label: String::from("foo"),
14e9852b	180	entries: vec![Entry(String::from("bar")), Entry(String::from("baz"))]
201b9ef3 SW	181	}]
	182	);
	183	assert_eq!(
	184	read_rows(&b"foo\n\nbar\n"[..])
	185	.flatten()
	186	.collect::<Vec<_>>(),
	187	vec![
	188	RowInput {
	189	label: String::from("foo"),
	190	entries: vec![]
	191	},
	192	RowInput {
	193	label: String::from("bar"),
	194	entries: vec![]
	195	}
	196	]
	197	);
1f6bd845 SW	198	assert_eq!(
	199	read_rows(&b"foo\n \nbar\n"[..])
	200	.flatten()
	201	.collect::<Vec<_>>(),
	202	vec![
	203	RowInput {
	204	label: String::from("foo"),
	205	entries: vec![]
	206	},
	207	RowInput {
	208	label: String::from("bar"),
	209	entries: vec![]
	210	}
	211	]
	212	);
	213	assert_eq!(
	214	read_rows(&b"foo \n bar \n"[..])
	215	.flatten()
	216	.collect::<Vec<_>>(),
	217	vec![RowInput {
	218	label: String::from("foo"),
14e9852b	219	entries: vec![Entry(String::from("bar"))]
1f6bd845 SW	220	}]
1f6bd845 SW	221	);
201b9ef3 SW	222
	223	let bad = read_rows(&b" foo"[..]).next().unwrap();
	224	assert!(bad.is_err());
8110b492	225	assert!(format!("{bad:?}").contains("1: Entry with no header"));
201b9ef3 SW	226
	227	let bad2 = read_rows(&b"foo\n\n bar"[..]).nth(1).unwrap();
	228	assert!(bad2.is_err());
8110b492	229	assert!(format!("{bad2:?}").contains("3: Entry with no header"));
75bb888a	230	}
f272e502 SW	231
	232	#[test]
	233	fn test_column_counts() {
	234	assert_eq!(
	235	column_counts(
	236	&read_rows(&b"foo\n bar\n baz\n"[..])
	237	.collect::<Result<Vec<_>, _>>()
	238	.unwrap()
	239	),
58b5f36d	240	vec![(1, String::from("bar")), (1, String::from("baz"))]
f272e502 SW	241	);
	242	assert_eq!(
	243	column_counts(
	244	&read_rows(&b"foo\n bar\n baz\nquux\n baz"[..])
	245	.collect::<Result<Vec<_>, _>>()
	246	.unwrap()
	247	),
58b5f36d	248	vec![(1, String::from("bar")), (2, String::from("baz"))]
f272e502	249	);
397ef957 SW	250	assert_eq!(
	251	column_counts(
	252	&read_rows(&b"foo\n bar\n bar\n baz\n bar\nquux\n baz"[..])
	253	.collect::<Result<Vec<_>, _>>()
	254	.unwrap()
	255	),
58b5f36d	256	vec![(1, String::from("bar")), (2, String::from("baz"))]
397ef957	257	);
f272e502	258	}
75bb888a	259	}