git.scottworley.com Git - tablify/blame_incremental - 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),
                instance: Some(String::from(instance)),
            },
        }
    }
}

#[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
}

/// # 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_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"))
            }
        );
    }

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