[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: Option<String>,
}
impl From<&str> for Entry {
    fn from(value: &str) -> Entry {
        match value.split_once(':') {
            None => Entry {
                col: String::from(value),
                instance: None,
            },
            Some((col, instance)) => Entry {
                col: String::from(col.trim()),
                instance: Some(String::from(instance.trim())),
            },
        }
    }
}

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

/// # 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_order(&rows);
    Ok(String::from(HEADER) + "Hello, world!" + FOOTER)
}

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

    #[test]
    fn test_parse_entry() {
        assert_eq!(
            Entry::from("foo"),
            Entry {
                col: String::from("foo"),
                instance: None
            }
        );
        assert_eq!(
            Entry::from("foo:bar"),
            Entry {
                col: String::from("foo"),
                instance: Some(String::from("bar"))
            }
        );
        assert_eq!(
            Entry::from("foo: bar"),
            Entry {
                col: String::from("foo"),
                instance: Some(String::from("bar"))
            }
        );
    }

    #[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"))]
        );
        assert_eq!(
            column_counts(
                &read_rows(&b"foo\n bar: 1\n bar: 2\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 {
e8657dff SW	43	col: String,
b8907770	44	instance: Option<String>,
e8657dff SW	45	}
	46	impl From<&str> for Entry {
	47	fn from(value: &str) -> Entry {
b8907770 SW	48	match value.split_once(':') {
	49	None => Entry {
	50	col: String::from(value),
	51	instance: None,
	52	},
	53	Some((col, instance)) => Entry {
0d999bc3 SW	54	col: String::from(col.trim()),
0d999bc3 SW	55	instance: Some(String::from(instance.trim())),
b8907770	56	},
e8657dff SW	57	}
	58	}
	59	}
14e9852b	60
75bb888a SW	61	#[derive(Debug, PartialEq, Eq)]
	62	struct RowInput {
	63	label: String,
14e9852b	64	entries: Vec<Entry>,
75bb888a SW	65	}
75bb888a SW	66
201b9ef3	67	struct Reader<Input: Iterator<Item = Result<String, std::io::Error>>> {
8110b492	68	input: std::iter::Enumerate<Input>,
201b9ef3 SW	69	row: Option<RowInput>,
	70	}
	71	impl<Input: Iterator<Item = Result<String, std::io::Error>>> Reader<Input> {
201b9ef3	72	fn new(input: Input) -> Self {
8110b492 SW	73	Self {
	74	input: input.enumerate(),
	75	row: None,
	76	}
201b9ef3 SW	77	}
	78	}
	79	impl<Input: Iterator<Item = Result<String, std::io::Error>>> Iterator for Reader<Input> {
	80	type Item = Result<RowInput, std::io::Error>;
	81	fn next(&mut self) -> Option<Self::Item> {
	82	loop {
1f6bd845 SW	83	match self
	84	.input
	85	.next()
8110b492	86	.map(\|(n, r)\| (n, r.map(\|line\| String::from(line.trim_end()))))
1f6bd845	87	{
201b9ef3	88	None => return Ok(std::mem::take(&mut self.row)).transpose(),
8110b492 SW	89	Some((_, Err(e))) => return Some(Err(e)),
8110b492 SW	90	Some((_, Ok(line))) if line.is_empty() && self.row.is_some() => {
201b9ef3 SW	91	return Ok(std::mem::take(&mut self.row)).transpose()
201b9ef3 SW	92	}
8110b492 SW	93	Some((_, Ok(line))) if line.is_empty() => {}
	94	Some((n, Ok(line))) if line.starts_with(' ') => match &mut self.row {
	95	None => {
	96	return Some(Err(std::io::Error::other(format!(
	97	"{}: Entry with no header",
	98	n + 1
	99	))))
	100	}
e8657dff	101	Some(ref mut row) => row.entries.push(Entry::from(line.trim())),
201b9ef3	102	},
8110b492	103	Some((_, Ok(line))) => {
201b9ef3 SW	104	let prev = std::mem::take(&mut self.row);
	105	self.row = Some(RowInput {
	106	label: line,
	107	entries: vec![],
	108	});
	109	if prev.is_some() {
	110	return Ok(prev).transpose();
	111	}
	112	}
	113	}
	114	}
	115	}
	116	}
	117
201b9ef3 SW	118	fn read_rows(input: impl std::io::Read) -> impl Iterator<Item = Result<RowInput, std::io::Error>> {
201b9ef3 SW	119	Reader::new(std::io::BufReader::new(input).lines())
75bb888a SW	120	}
75bb888a SW	121
58b5f36d SW	122	fn column_counts(rows: &[RowInput]) -> Vec<(usize, String)> {
	123	let mut counts: Vec<_> = rows
	124	.iter()
b8907770 SW	125	.flat_map(\|r\| {
	126	r.entries
	127	.iter()
	128	.map(\|e\| &e.col)
	129	.collect::<HashSet<_>>()
	130	.into_iter()
	131	})
	132	.fold(HashMap::new(), \|mut cs, col\| {
	133	cs.entry(String::from(col))
58b5f36d	134	.and_modify(\|n\| *n += 1)
f272e502	135	.or_insert(1);
58b5f36d	136	cs
f272e502	137	})
58b5f36d SW	138	.into_iter()
	139	.map(\|(col, n)\| (n, col))
	140	.collect();
	141	counts.sort();
	142	counts
f272e502	143	}
d22b2e05 SW	144	fn column_order(rows: &[RowInput]) -> Vec<String> {
	145	column_counts(rows)
	146	.into_iter()
	147	.map(\|(_, col)\| col)
	148	.collect()
	149	}
f272e502	150
4b99fb70 SW	151	/// # Errors
	152	///
	153	/// Will return `Err` if
	154	/// * there's an i/o error while reading `input`
	155	/// * the log has invalid syntax:
	156	/// * an indented line with no preceding non-indented line
	157	pub fn tablify(input: impl std::io::Read) -> Result<String, std::io::Error> {
	158	let rows = read_rows(input).collect::<Result<Vec<_>, _>>()?;
d22b2e05	159	let _columns = column_order(&rows);
cc2378d5	160	Ok(String::from(HEADER) + "Hello, world!" + FOOTER)
ece97615	161	}
75bb888a SW	162
	163	#[cfg(test)]
	164	mod tests {
	165	use super::*;
	166
b8907770 SW	167	#[test]
	168	fn test_parse_entry() {
	169	assert_eq!(
	170	Entry::from("foo"),
	171	Entry {
	172	col: String::from("foo"),
	173	instance: None
	174	}
	175	);
	176	assert_eq!(
	177	Entry::from("foo:bar"),
	178	Entry {
	179	col: String::from("foo"),
	180	instance: Some(String::from("bar"))
	181	}
	182	);
0d999bc3 SW	183	assert_eq!(
	184	Entry::from("foo: bar"),
	185	Entry {
	186	col: String::from("foo"),
	187	instance: Some(String::from("bar"))
	188	}
	189	);
b8907770 SW	190	}
b8907770 SW	191
75bb888a SW	192	#[test]
	193	fn test_read_rows() {
	194	assert_eq!(
201b9ef3	195	read_rows(&b"foo"[..]).flatten().collect::<Vec<_>>(),
75bb888a SW	196	vec![RowInput {
	197	label: String::from("foo"),
	198	entries: vec![]
	199	}]
	200	);
9dfa98b7	201	assert_eq!(
201b9ef3	202	read_rows(&b"bar"[..]).flatten().collect::<Vec<_>>(),
9dfa98b7 SW	203	vec![RowInput {
	204	label: String::from("bar"),
	205	entries: vec![]
	206	}]
	207	);
2aa9ef94	208	assert_eq!(
201b9ef3	209	read_rows(&b"foo\nbar\n"[..]).flatten().collect::<Vec<_>>(),
2aa9ef94 SW	210	vec![
	211	RowInput {
	212	label: String::from("foo"),
	213	entries: vec![]
	214	},
	215	RowInput {
	216	label: String::from("bar"),
	217	entries: vec![]
	218	}
	219	]
	220	);
201b9ef3 SW	221	assert_eq!(
	222	read_rows(&b"foo\n bar\n"[..]).flatten().collect::<Vec<_>>(),
	223	vec![RowInput {
	224	label: String::from("foo"),
e8657dff	225	entries: vec![Entry::from("bar")]
201b9ef3 SW	226	}]
	227	);
	228	assert_eq!(
	229	read_rows(&b"foo\n bar\n baz\n"[..])
	230	.flatten()
	231	.collect::<Vec<_>>(),
	232	vec![RowInput {
	233	label: String::from("foo"),
e8657dff	234	entries: vec![Entry::from("bar"), Entry::from("baz")]
201b9ef3 SW	235	}]
	236	);
	237	assert_eq!(
	238	read_rows(&b"foo\n\nbar\n"[..])
	239	.flatten()
	240	.collect::<Vec<_>>(),
	241	vec![
	242	RowInput {
	243	label: String::from("foo"),
	244	entries: vec![]
	245	},
	246	RowInput {
	247	label: String::from("bar"),
	248	entries: vec![]
	249	}
	250	]
	251	);
1f6bd845 SW	252	assert_eq!(
	253	read_rows(&b"foo\n \nbar\n"[..])
	254	.flatten()
	255	.collect::<Vec<_>>(),
	256	vec![
	257	RowInput {
	258	label: String::from("foo"),
	259	entries: vec![]
	260	},
	261	RowInput {
	262	label: String::from("bar"),
	263	entries: vec![]
	264	}
	265	]
	266	);
	267	assert_eq!(
	268	read_rows(&b"foo \n bar \n"[..])
	269	.flatten()
	270	.collect::<Vec<_>>(),
	271	vec![RowInput {
	272	label: String::from("foo"),
e8657dff	273	entries: vec![Entry::from("bar")]
1f6bd845 SW	274	}]
1f6bd845 SW	275	);
201b9ef3 SW	276
	277	let bad = read_rows(&b" foo"[..]).next().unwrap();
	278	assert!(bad.is_err());
8110b492	279	assert!(format!("{bad:?}").contains("1: Entry with no header"));
201b9ef3 SW	280
	281	let bad2 = read_rows(&b"foo\n\n bar"[..]).nth(1).unwrap();
	282	assert!(bad2.is_err());
8110b492	283	assert!(format!("{bad2:?}").contains("3: Entry with no header"));
75bb888a	284	}
f272e502 SW	285
	286	#[test]
	287	fn test_column_counts() {
	288	assert_eq!(
	289	column_counts(
	290	&read_rows(&b"foo\n bar\n baz\n"[..])
	291	.collect::<Result<Vec<_>, _>>()
	292	.unwrap()
	293	),
58b5f36d	294	vec![(1, String::from("bar")), (1, String::from("baz"))]
f272e502 SW	295	);
	296	assert_eq!(
	297	column_counts(
	298	&read_rows(&b"foo\n bar\n baz\nquux\n baz"[..])
	299	.collect::<Result<Vec<_>, _>>()
	300	.unwrap()
	301	),
58b5f36d	302	vec![(1, String::from("bar")), (2, String::from("baz"))]
f272e502	303	);
397ef957 SW	304	assert_eq!(
	305	column_counts(
	306	&read_rows(&b"foo\n bar\n bar\n baz\n bar\nquux\n baz"[..])
	307	.collect::<Result<Vec<_>, _>>()
	308	.unwrap()
	309	),
58b5f36d	310	vec![(1, String::from("bar")), (2, String::from("baz"))]
397ef957	311	);
b8907770 SW	312	assert_eq!(
	313	column_counts(
	314	&read_rows(&b"foo\n bar: 1\n bar: 2\n baz\n bar\nquux\n baz"[..])
	315	.collect::<Result<Vec<_>, _>>()
	316	.unwrap()
	317	),
	318	vec![(1, String::from("bar")), (2, String::from("baz"))]
	319	);
f272e502	320	}
75bb888a	321	}