[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 {
    col: String,
    instance: String,
}
impl From<&str> for Entry {
    fn from(value: &str) -> Entry {
        Entry {
            col: String::from(value),
            instance: String::from(""),
        }
    }
}

#[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::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.col))
                .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::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::from("bar"), Entry::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::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	41	#[derive(Debug, PartialEq, Eq, Hash)]
e8657dff SW	42	struct Entry {
	43	col: String,
	44	instance: String,
	45	}
	46	impl From<&str> for Entry {
	47	fn from(value: &str) -> Entry {
	48	Entry {
	49	col: String::from(value),
	50	instance: String::from(""),
	51	}
	52	}
	53	}
14e9852b	54
75bb888a SW	55	#[derive(Debug, PartialEq, Eq)]
	56	struct RowInput {
	57	label: String,
14e9852b	58	entries: Vec<Entry>,
75bb888a SW	59	}
75bb888a SW	60
201b9ef3	61	struct Reader<Input: Iterator<Item = Result<String, std::io::Error>>> {
8110b492	62	input: std::iter::Enumerate<Input>,
201b9ef3 SW	63	row: Option<RowInput>,
	64	}
	65	impl<Input: Iterator<Item = Result<String, std::io::Error>>> Reader<Input> {
201b9ef3	66	fn new(input: Input) -> Self {
8110b492 SW	67	Self {
	68	input: input.enumerate(),
	69	row: None,
	70	}
201b9ef3 SW	71	}
	72	}
	73	impl<Input: Iterator<Item = Result<String, std::io::Error>>> Iterator for Reader<Input> {
	74	type Item = Result<RowInput, std::io::Error>;
	75	fn next(&mut self) -> Option<Self::Item> {
	76	loop {
1f6bd845 SW	77	match self
	78	.input
	79	.next()
8110b492	80	.map(\|(n, r)\| (n, r.map(\|line\| String::from(line.trim_end()))))
1f6bd845	81	{
201b9ef3	82	None => return Ok(std::mem::take(&mut self.row)).transpose(),
8110b492 SW	83	Some((_, Err(e))) => return Some(Err(e)),
8110b492 SW	84	Some((_, Ok(line))) if line.is_empty() && self.row.is_some() => {
201b9ef3 SW	85	return Ok(std::mem::take(&mut self.row)).transpose()
201b9ef3 SW	86	}
8110b492 SW	87	Some((_, Ok(line))) if line.is_empty() => {}
	88	Some((n, Ok(line))) if line.starts_with(' ') => match &mut self.row {
	89	None => {
	90	return Some(Err(std::io::Error::other(format!(
	91	"{}: Entry with no header",
	92	n + 1
	93	))))
	94	}
e8657dff	95	Some(ref mut row) => row.entries.push(Entry::from(line.trim())),
201b9ef3	96	},
8110b492	97	Some((_, Ok(line))) => {
201b9ef3 SW	98	let prev = std::mem::take(&mut self.row);
	99	self.row = Some(RowInput {
	100	label: line,
	101	entries: vec![],
	102	});
	103	if prev.is_some() {
	104	return Ok(prev).transpose();
	105	}
	106	}
	107	}
	108	}
	109	}
	110	}
	111
201b9ef3 SW	112	fn read_rows(input: impl std::io::Read) -> impl Iterator<Item = Result<RowInput, std::io::Error>> {
201b9ef3 SW	113	Reader::new(std::io::BufReader::new(input).lines())
75bb888a SW	114	}
75bb888a SW	115
58b5f36d SW	116	fn column_counts(rows: &[RowInput]) -> Vec<(usize, String)> {
	117	let mut counts: Vec<_> = rows
	118	.iter()
397ef957	119	.flat_map(\|r\| r.entries.iter().collect::<HashSet<_>>().into_iter())
58b5f36d	120	.fold(HashMap::new(), \|mut cs, e\| {
e8657dff	121	cs.entry(String::from(&e.col))
58b5f36d	122	.and_modify(\|n\| *n += 1)
f272e502	123	.or_insert(1);
58b5f36d	124	cs
f272e502	125	})
58b5f36d SW	126	.into_iter()
	127	.map(\|(col, n)\| (n, col))
	128	.collect();
	129	counts.sort();
	130	counts
f272e502 SW	131	}
f272e502 SW	132
4b99fb70 SW	133	/// # Errors
	134	///
	135	/// Will return `Err` if
	136	/// * there's an i/o error while reading `input`
	137	/// * the log has invalid syntax:
	138	/// * an indented line with no preceding non-indented line
	139	pub fn tablify(input: impl std::io::Read) -> Result<String, std::io::Error> {
	140	let rows = read_rows(input).collect::<Result<Vec<_>, _>>()?;
	141	let _columns = column_counts(&rows);
cc2378d5	142	Ok(String::from(HEADER) + "Hello, world!" + FOOTER)
ece97615	143	}
75bb888a SW	144
	145	#[cfg(test)]
	146	mod tests {
	147	use super::*;
	148
	149	#[test]
	150	fn test_read_rows() {
	151	assert_eq!(
201b9ef3	152	read_rows(&b"foo"[..]).flatten().collect::<Vec<_>>(),
75bb888a SW	153	vec![RowInput {
	154	label: String::from("foo"),
	155	entries: vec![]
	156	}]
	157	);
9dfa98b7	158	assert_eq!(
201b9ef3	159	read_rows(&b"bar"[..]).flatten().collect::<Vec<_>>(),
9dfa98b7 SW	160	vec![RowInput {
	161	label: String::from("bar"),
	162	entries: vec![]
	163	}]
	164	);
2aa9ef94	165	assert_eq!(
201b9ef3	166	read_rows(&b"foo\nbar\n"[..]).flatten().collect::<Vec<_>>(),
2aa9ef94 SW	167	vec![
	168	RowInput {
	169	label: String::from("foo"),
	170	entries: vec![]
	171	},
	172	RowInput {
	173	label: String::from("bar"),
	174	entries: vec![]
	175	}
	176	]
	177	);
201b9ef3 SW	178	assert_eq!(
	179	read_rows(&b"foo\n bar\n"[..]).flatten().collect::<Vec<_>>(),
	180	vec![RowInput {
	181	label: String::from("foo"),
e8657dff	182	entries: vec![Entry::from("bar")]
201b9ef3 SW	183	}]
	184	);
	185	assert_eq!(
	186	read_rows(&b"foo\n bar\n baz\n"[..])
	187	.flatten()
	188	.collect::<Vec<_>>(),
	189	vec![RowInput {
	190	label: String::from("foo"),
e8657dff	191	entries: vec![Entry::from("bar"), Entry::from("baz")]
201b9ef3 SW	192	}]
	193	);
	194	assert_eq!(
	195	read_rows(&b"foo\n\nbar\n"[..])
	196	.flatten()
	197	.collect::<Vec<_>>(),
	198	vec![
	199	RowInput {
	200	label: String::from("foo"),
	201	entries: vec![]
	202	},
	203	RowInput {
	204	label: String::from("bar"),
	205	entries: vec![]
	206	}
	207	]
	208	);
1f6bd845 SW	209	assert_eq!(
	210	read_rows(&b"foo\n \nbar\n"[..])
	211	.flatten()
	212	.collect::<Vec<_>>(),
	213	vec![
	214	RowInput {
	215	label: String::from("foo"),
	216	entries: vec![]
	217	},
	218	RowInput {
	219	label: String::from("bar"),
	220	entries: vec![]
	221	}
	222	]
	223	);
	224	assert_eq!(
	225	read_rows(&b"foo \n bar \n"[..])
	226	.flatten()
	227	.collect::<Vec<_>>(),
	228	vec![RowInput {
	229	label: String::from("foo"),
e8657dff	230	entries: vec![Entry::from("bar")]
1f6bd845 SW	231	}]
1f6bd845 SW	232	);
201b9ef3 SW	233
	234	let bad = read_rows(&b" foo"[..]).next().unwrap();
	235	assert!(bad.is_err());
8110b492	236	assert!(format!("{bad:?}").contains("1: Entry with no header"));
201b9ef3 SW	237
	238	let bad2 = read_rows(&b"foo\n\n bar"[..]).nth(1).unwrap();
	239	assert!(bad2.is_err());
8110b492	240	assert!(format!("{bad2:?}").contains("3: Entry with no header"));
75bb888a	241	}
f272e502 SW	242
	243	#[test]
	244	fn test_column_counts() {
	245	assert_eq!(
	246	column_counts(
	247	&read_rows(&b"foo\n bar\n baz\n"[..])
	248	.collect::<Result<Vec<_>, _>>()
	249	.unwrap()
	250	),
58b5f36d	251	vec![(1, String::from("bar")), (1, String::from("baz"))]
f272e502 SW	252	);
	253	assert_eq!(
	254	column_counts(
	255	&read_rows(&b"foo\n bar\n baz\nquux\n baz"[..])
	256	.collect::<Result<Vec<_>, _>>()
	257	.unwrap()
	258	),
58b5f36d	259	vec![(1, String::from("bar")), (2, String::from("baz"))]
f272e502	260	);
397ef957 SW	261	assert_eq!(
	262	column_counts(
	263	&read_rows(&b"foo\n bar\n bar\n baz\n bar\nquux\n baz"[..])
	264	.collect::<Result<Vec<_>, _>>()
	265	.unwrap()
	266	),
58b5f36d	267	vec![(1, String::from("bar")), (2, String::from("baz"))]
397ef957	268	);
f272e502	269	}
75bb888a	270	}