X-Git-Url: http://git.scottworley.com/tablify/blobdiff_plain/8bf0d5b1daaf4841b1c593630d3c862e72985dda..12e913005080a9f60750d807c220a58704429361:/src/lib.rs?ds=sidebyside
diff --git a/src/lib.rs b/src/lib.rs
index 7292f02..dbcfc7b 100644
--- a/src/lib.rs
+++ b/src/lib.rs
@@ -1,30 +1,37 @@
-use std::collections::{HashMap, HashSet};
+use std::borrow::ToOwned;
+use std::collections::HashMap;
use std::fmt::Write;
use std::io::BufRead;
use std::iter::Iterator;
-const HEADER: &str = "
+pub struct Config {
+ pub column_threshold: usize,
+}
+
+const HEADER: &str = r#"
-
-
+
+
@@ -32,7 +39,7 @@ const HEADER: &str = "
-";
+"#;
const FOOTER: &str = "
@@ -75,37 +82,45 @@ impl std::fmt::Display for HTML {
}
}
-#[derive(Debug, PartialEq, Eq, Hash)]
-struct Entry<'a> {
- col: &'a str,
- instance: Option<&'a str>,
+#[derive(Debug, PartialEq, Eq)]
+enum InputLine<'a> {
+ Blank,
+ RowHeader(&'a str),
+ Entry(&'a str, Option<&'a str>),
}
-impl<'a> From<&'a str> for Entry<'a> {
- fn from(value: &'a str) -> Entry<'a> {
- match value.split_once(':') {
- None => Entry {
- col: value,
- instance: None,
- },
- Some((col, instance)) => Entry {
- col: col.trim(),
- instance: Some(instance.trim()),
- },
+impl<'a> From<&'a str> for InputLine<'a> {
+ fn from(value: &'a str) -> InputLine<'a> {
+ let trimmed = value.trim_end();
+ if trimmed.is_empty() {
+ InputLine::Blank
+ } else if !trimmed.starts_with(' ') {
+ InputLine::RowHeader(value.trim())
+ } else {
+ match value.split_once(':') {
+ None => InputLine::Entry(value.trim(), None),
+ Some((col, instance)) => InputLine::Entry(col.trim(), Some(instance.trim())),
+ }
}
}
}
#[derive(Debug, PartialEq, Eq)]
-struct RowInput<'a> {
- label: &'a str,
- entries: Vec>,
+struct Row {
+ label: String,
+ entries: HashMap>>,
}
-struct Reader<'a, Input: Iterator- >> {
+#[derive(Debug, PartialEq, Eq)]
+enum Rowlike {
+ Row(Row),
+ Spacer,
+}
+
+struct Reader>> {
input: std::iter::Enumerate,
- row: Option>,
+ row: Option
,
}
-impl<'a, Input: Iterator- >> Reader<'a, Input> {
+impl>> Reader {
fn new(input: Input) -> Self {
Self {
input: input.enumerate(),
@@ -113,61 +128,62 @@ impl<'a, Input: Iterator
- >> Reader<'a, Inpu
}
}
}
-impl<'a, Input: Iterator
- >> Iterator for Reader<'a, Input> {
- type Item = Result, std::io::Error>;
+impl>> Iterator for Reader {
+ type Item = Result;
fn next(&mut self) -> Option {
loop {
match self.input.next() {
- None => return Ok(std::mem::take(&mut self.row)).transpose(),
+ None => return Ok(std::mem::take(&mut self.row).map(Rowlike::Row)).transpose(),
Some((_, Err(e))) => return Some(Err(e)),
- Some((_, Ok(line))) if line.trim_end().is_empty() && self.row.is_some() => {
- return Ok(std::mem::take(&mut self.row)).transpose()
- }
- Some((_, Ok(line))) if line.trim_end().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((n, Ok(line))) => match InputLine::from(line.as_ref()) {
+ InputLine::Blank if self.row.is_some() => {
+ return Ok(std::mem::take(&mut self.row).map(Rowlike::Row)).transpose()
}
- // TODO: Don't leak
- Some(ref mut row) => row.entries.push(Entry::from(line.leak().trim())),
- },
- Some((_, Ok(line))) => {
- let prev = std::mem::take(&mut self.row);
- self.row = Some(RowInput {
- // TODO: Don't leak
- label: line.leak().trim_end(),
- entries: vec![],
- });
- if prev.is_some() {
- return Ok(prev).transpose();
+ InputLine::Blank => return Some(Ok(Rowlike::Spacer)),
+ InputLine::Entry(col, instance) => 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
+ .entry(col.to_owned())
+ .and_modify(|is| is.push(instance.map(ToOwned::to_owned)))
+ .or_insert_with(|| vec![instance.map(ToOwned::to_owned)]);
+ }
+ },
+ InputLine::RowHeader(row) => {
+ let prev = std::mem::take(&mut self.row);
+ self.row = Some(Row {
+ label: row.to_owned(),
+ entries: HashMap::new(),
+ });
+ if prev.is_some() {
+ return Ok(prev.map(Rowlike::Row)).transpose();
+ }
}
- }
+ },
}
}
}
}
-fn read_rows(
- input: impl std::io::Read,
-) -> impl Iterator
- , std::io::Error>> {
- Reader::new(std::io::BufReader::new(input).lines())
+fn read_rows(input: impl std::io::Read) -> Result, std::io::Error> {
+ Reader::new(std::io::BufReader::new(input).lines()).collect::, _>>()
}
-fn column_counts(rows: &[RowInput]) -> Vec<(usize, String)> {
+fn column_counts(rows: &[Rowlike]) -> Vec<(usize, String)> {
+ let empty = HashMap::new();
let mut counts: Vec<_> = rows
.iter()
- .flat_map(|r| {
- r.entries
- .iter()
- .map(|e| &e.col)
- .collect::>()
- .into_iter()
+ .flat_map(|rl| match rl {
+ Rowlike::Row(r) => r.entries.keys(),
+ Rowlike::Spacer => empty.keys(),
})
.fold(HashMap::new(), |mut cs, col| {
- cs.entry(String::from(*col))
+ cs.entry(col.to_owned())
.and_modify(|n| *n += 1)
.or_insert(1);
cs
@@ -178,63 +194,101 @@ fn column_counts(rows: &[RowInput]) -> Vec<(usize, String)> {
counts.sort_unstable_by(|(an, acol), (bn, bcol)| bn.cmp(an).then(acol.cmp(bcol)));
counts
}
-fn column_order(rows: &[RowInput]) -> Vec {
+fn column_order(config: &Config, rows: &[Rowlike]) -> Vec {
column_counts(rows)
.into_iter()
- .map(|(_, col)| col)
+ .filter_map(|(n, col)| (n >= config.column_threshold).then_some(col))
.collect()
}
-fn render_instance(entry: &Entry) -> HTML {
- match &entry.instance {
+fn render_one_instance(instance: &Option) -> HTML {
+ match instance {
None => HTML::from("â"),
Some(instance) => HTML::escape(instance.as_ref()),
}
}
-fn render_cell(col: &str, row: &RowInput) -> HTML {
- let row_label = HTML::escape(row.label.as_ref());
- let col_label = HTML::escape(col);
- let entries: Vec<&Entry> = row.entries.iter().filter(|e| e.col == col).collect();
- let class = HTML::from(if entries.is_empty() { "" } else { "yes" });
- let all_empty = entries.iter().all(|e| e.instance.is_none());
- let contents = if entries.is_empty() || (all_empty && entries.len() == 1) {
+fn render_instances(instances: &[Option]) -> HTML {
+ let all_empty = instances.iter().all(Option::is_none);
+ if all_empty && instances.len() == 1 {
HTML::from("")
} else if all_empty {
- HTML(format!("{}", entries.len()))
+ HTML(format!("{}", instances.len()))
} else {
HTML(
- entries
+ instances
.iter()
- .map(|i| render_instance(i))
+ .map(render_one_instance)
.map(|html| html.0) // Waiting for slice_concat_trait to stabilize
.collect::>()
.join(" "),
)
- };
- HTML(format!("
| {contents} | "))
+ }
}
-fn render_row(columns: &[String], row: &RowInput) -> HTML {
- // This is O(n^2) & doesn't need to be
+fn render_cell(col: &str, row: &mut Row) -> HTML {
let row_label = HTML::escape(row.label.as_ref());
+ let col_label = HTML::escape(col);
+ let instances: Option<&Vec>> = row.entries.get(col);
+ let class = HTML::from(if instances.is_none() { "" } else { "yes" });
+ let contents = match instances {
+ None => HTML::from(""),
+ Some(is) => render_instances(is),
+ };
+ row.entries.remove(col);
HTML(format!(
- "| {row_label} | {}
|---|
\n",
- &columns
- .iter()
- .map(|col| render_cell(col, row))
- .collect::()
+ r#"{contents} | "#
))
}
+fn render_leftover(notcol: &str, instances: &[Option]) -> HTML {
+ let label = HTML::escape(notcol);
+ let rest = render_instances(instances);
+ if rest == HTML::from("") {
+ HTML(format!("{label}"))
+ } else {
+ HTML(format!("{label}: {rest}"))
+ }
+}
+
+fn render_all_leftovers(row: &Row) -> HTML {
+ let mut order: Vec<_> = row.entries.keys().collect();
+ order.sort_unstable();
+ HTML(
+ order
+ .into_iter()
+ .map(|notcol| render_leftover(notcol, row.entries.get(notcol).expect("Key vanished?!")))
+ .map(|html| html.0) // Waiting for slice_concat_trait to stabilize
+ .collect::>()
+ .join(", "),
+ )
+}
+
+fn render_row(columns: &[String], rowlike: &mut Rowlike) -> HTML {
+ match rowlike {
+ Rowlike::Spacer => HTML::from(" |
|---|
\n"),
+ Rowlike::Row(row) => {
+ let row_label = HTML::escape(row.label.as_ref());
+ let cells = columns
+ .iter()
+ .map(|col| render_cell(col, row))
+ .collect::();
+ let leftovers = render_all_leftovers(row);
+ HTML(format!(
+ "| {row_label} | {cells}{leftovers} |
|---|
\n"
+ ))
+ }
+ }
+}
+
fn render_column_headers(columns: &[String]) -> HTML {
HTML(
- String::from(" | ")
+ String::from(r#"
|---|
| "#)
+ &columns.iter().fold(String::new(), |mut acc, col| {
let col_header = HTML::escape(col.as_ref());
write!(
&mut acc,
- ""
+ r#""#
)
.unwrap();
acc
@@ -249,14 +303,14 @@ fn render_column_headers(columns: &[String]) -> HTML {
/// * 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 {
- let rows = read_rows(input).collect::, _>>()?;
- let columns = column_order(&rows);
+pub fn tablify(config: &Config, input: impl std::io::Read) -> Result {
+ let rows = read_rows(input)?;
+ let columns = column_order(config, &rows);
Ok(HTML(format!(
"{HEADER}{}{}{FOOTER}",
render_column_headers(&columns),
rows.into_iter()
- .map(|r| render_row(&columns, &r))
+ .map(|mut r| render_row(&columns, &mut r))
.collect::()
)))
}
@@ -266,120 +320,133 @@ mod tests {
use super::*;
#[test]
- fn test_parse_entry() {
+ fn test_parse_line() {
+ assert_eq!(InputLine::from(""), InputLine::Blank);
+ assert_eq!(InputLine::from(" "), InputLine::Blank);
+ assert_eq!(InputLine::from("foo"), InputLine::RowHeader("foo"));
+ assert_eq!(InputLine::from("foo "), InputLine::RowHeader("foo"));
+ assert_eq!(InputLine::from(" foo"), InputLine::Entry("foo", None));
assert_eq!(
- Entry::from("foo"),
- Entry {
- col: "foo",
- instance: None
- }
+ InputLine::from(" foo:bar"),
+ InputLine::Entry("foo", Some("bar"))
);
assert_eq!(
- Entry::from("foo:bar"),
- Entry {
- col: "foo",
- instance: Some("bar")
- }
+ InputLine::from(" foo: bar"),
+ InputLine::Entry("foo", Some("bar"))
);
assert_eq!(
- Entry::from("foo: bar"),
- Entry {
- col: "foo",
- instance: Some("bar")
- }
+ InputLine::from(" foo: bar "),
+ InputLine::Entry("foo", Some("bar"))
+ );
+ assert_eq!(
+ InputLine::from(" foo: bar "),
+ InputLine::Entry("foo", Some("bar"))
+ );
+ assert_eq!(
+ InputLine::from(" foo : bar "),
+ InputLine::Entry("foo", Some("bar"))
);
}
#[test]
fn test_read_rows() {
assert_eq!(
- read_rows(&b"foo"[..]).flatten().collect::>(),
- vec![RowInput {
- label: "foo",
- entries: vec![]
- }]
+ read_rows(&b"foo"[..]).unwrap(),
+ vec![Rowlike::Row(Row {
+ label: "foo".to_owned(),
+ entries: HashMap::new(),
+ })]
);
assert_eq!(
- read_rows(&b"bar"[..]).flatten().collect::>(),
- vec![RowInput {
- label: "bar",
- entries: vec![]
- }]
+ read_rows(&b"bar"[..]).unwrap(),
+ vec![Rowlike::Row(Row {
+ label: "bar".to_owned(),
+ entries: HashMap::new(),
+ })]
);
assert_eq!(
- read_rows(&b"foo\nbar\n"[..]).flatten().collect::>(),
+ read_rows(&b"foo\nbar\n"[..]).unwrap(),
vec![
- RowInput {
- label: "foo",
- entries: vec![]
- },
- RowInput {
- label: "bar",
- entries: vec![]
- }
+ Rowlike::Row(Row {
+ label: "foo".to_owned(),
+ entries: HashMap::new(),
+ }),
+ Rowlike::Row(Row {
+ label: "bar".to_owned(),
+ entries: HashMap::new(),
+ })
]
);
assert_eq!(
- read_rows(&b"foo\n bar\n"[..]).flatten().collect::>(),
- vec![RowInput {
- label: "foo",
- entries: vec![Entry::from("bar")]
- }]
+ read_rows(&b"foo\n bar\n"[..]).unwrap(),
+ vec![Rowlike::Row(Row {
+ label: "foo".to_owned(),
+ entries: HashMap::from([("bar".to_owned(), vec![None])]),
+ })]
);
assert_eq!(
- read_rows(&b"foo\n bar\n baz\n"[..])
- .flatten()
- .collect::>(),
- vec![RowInput {
- label: "foo",
- entries: vec![Entry::from("bar"), Entry::from("baz")]
- }]
+ read_rows(&b"foo\n bar\n baz\n"[..]).unwrap(),
+ vec![Rowlike::Row(Row {
+ label: "foo".to_owned(),
+ entries: HashMap::from([
+ ("bar".to_owned(), vec![None]),
+ ("baz".to_owned(), vec![None])
+ ]),
+ })]
);
assert_eq!(
- read_rows(&b"foo\n\nbar\n"[..])
- .flatten()
- .collect::>(),
+ read_rows(&b"foo\n\nbar\n"[..]).unwrap(),
vec![
- RowInput {
- label: "foo",
- entries: vec![]
- },
- RowInput {
- label: "bar",
- entries: vec![]
- }
+ Rowlike::Row(Row {
+ label: "foo".to_owned(),
+ entries: HashMap::new(),
+ }),
+ Rowlike::Row(Row {
+ label: "bar".to_owned(),
+ entries: HashMap::new(),
+ })
]
);
assert_eq!(
- read_rows(&b"foo\n \nbar\n"[..])
- .flatten()
- .collect::>(),
+ read_rows(&b"foo\n\n\nbar\n"[..]).unwrap(),
vec![
- RowInput {
- label: "foo",
- entries: vec![]
- },
- RowInput {
- label: "bar",
- entries: vec![]
- }
+ Rowlike::Row(Row {
+ label: "foo".to_owned(),
+ entries: HashMap::new(),
+ }),
+ Rowlike::Spacer,
+ Rowlike::Row(Row {
+ label: "bar".to_owned(),
+ entries: HashMap::new(),
+ })
+ ]
+ );
+ assert_eq!(
+ read_rows(&b"foo\n \nbar\n"[..]).unwrap(),
+ vec![
+ Rowlike::Row(Row {
+ label: "foo".to_owned(),
+ entries: HashMap::new(),
+ }),
+ Rowlike::Row(Row {
+ label: "bar".to_owned(),
+ entries: HashMap::new(),
+ })
]
);
assert_eq!(
- read_rows(&b"foo \n bar \n"[..])
- .flatten()
- .collect::>(),
- vec![RowInput {
- label: "foo",
- entries: vec![Entry::from("bar")]
- }]
+ read_rows(&b"foo \n bar \n"[..]).unwrap(),
+ vec![Rowlike::Row(Row {
+ label: "foo".to_owned(),
+ entries: HashMap::from([("bar".to_owned(), vec![None])]),
+ })]
);
- let bad = read_rows(&b" foo"[..]).next().unwrap();
+ let bad = read_rows(&b" foo"[..]);
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();
+ let bad2 = read_rows(&b"foo\n\n bar"[..]);
assert!(bad2.is_err());
assert!(format!("{bad2:?}").contains("3: Entry with no header"));
}
@@ -387,34 +454,20 @@ mod tests {
#[test]
fn test_column_counts() {
assert_eq!(
- column_counts(
- &read_rows(&b"foo\n bar\n baz\n"[..])
- .collect::, _>>()
- .unwrap()
- ),
+ column_counts(&read_rows(&b"foo\n bar\n baz\n"[..]).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::, _>>()
- .unwrap()
- ),
+ column_counts(&read_rows(&b"foo\n bar\n baz\nquux\n baz"[..]).unwrap()),
vec![(2, String::from("baz")), (1, String::from("bar"))]
);
assert_eq!(
- column_counts(
- &read_rows(&b"foo\n bar\n bar\n baz\n bar\nquux\n baz"[..])
- .collect::, _>>()
- .unwrap()
- ),
+ column_counts(&read_rows(&b"foo\n bar\n bar\n baz\n bar\nquux\n baz"[..]).unwrap()),
vec![(2, String::from("baz")), (1, String::from("bar"))]
);
assert_eq!(
column_counts(
- &read_rows(&b"foo\n bar: 1\n bar: 2\n baz\n bar\nquux\n baz"[..])
- .collect::, _>>()
- .unwrap()
+ &read_rows(&b"foo\n bar: 1\n bar: 2\n baz\n bar\nquux\n baz"[..]).unwrap()
),
vec![(2, String::from("baz")), (1, String::from("bar"))]
);
@@ -425,82 +478,163 @@ mod tests {
assert_eq!(
render_cell(
"foo",
- &RowInput {
- label: "nope",
- entries: vec![]
+ &mut Row {
+ label: "nope".to_owned(),
+ entries: HashMap::new(),
}
),
- HTML::from(" | ")
+ HTML::from(
+ r#" | "#
+ )
);
assert_eq!(
render_cell(
"foo",
- &RowInput {
- label: "nope",
- entries: vec![Entry::from("bar")]
+ &mut Row {
+ label: "nope".to_owned(),
+ entries: HashMap::from([("bar".to_owned(), vec![None])]),
}
),
- HTML::from(" | ")
+ HTML::from(
+ r#" | "#
+ )
);
assert_eq!(
render_cell(
"foo",
- &RowInput {
- label: "nope",
- entries: vec![Entry::from("foo")]
+ &mut Row {
+ label: "nope".to_owned(),
+ entries: HashMap::from([("foo".to_owned(), vec![None])]),
}
),
- HTML::from(" | ")
+ HTML::from(
+ r#" | "#
+ )
);
assert_eq!(
render_cell(
"foo",
- &RowInput {
- label: "nope",
- entries: vec![Entry::from("foo"), Entry::from("foo")]
+ &mut Row {
+ label: "nope".to_owned(),
+ entries: HashMap::from([("foo".to_owned(), vec![None, None])]),
}
),
- HTML::from("2 | ")
+ HTML::from(
+ r#"2 | "#
+ )
);
assert_eq!(
render_cell(
"foo",
- &RowInput {
- label: "nope",
- entries: vec![Entry::from("foo: 5"), Entry::from("foo: 10")]
+ &mut Row {
+ label: "nope".to_owned(),
+ entries: HashMap::from([(
+ "foo".to_owned(),
+ vec![Some("5".to_owned()), Some("10".to_owned())]
+ )]),
}
),
- HTML::from("5 10 | ")
+ HTML::from(
+ r#"5 10 | "#
+ )
);
assert_eq!(
render_cell(
"foo",
- &RowInput {
- label: "nope",
- entries: vec![Entry::from("foo: 5"), Entry::from("foo")]
+ &mut Row {
+ label: "nope".to_owned(),
+ entries: HashMap::from([("foo".to_owned(), vec![Some("5".to_owned()), None])]),
}
),
- HTML::from("5 â | ")
+ HTML::from(
+ r#"5 â | "#
+ )
);
assert_eq!(
render_cell(
"heart",
- &RowInput {
- label: "nope",
- entries: vec![Entry::from("heart: <3")]
+ &mut Row {
+ label: "nope".to_owned(),
+ entries: HashMap::from([("heart".to_owned(), vec![Some("<3".to_owned())])]),
}
),
- HTML::from("<3 | ")
+ HTML::from(
+ r#"<3 | "#
+ )
);
assert_eq!(
render_cell(
"foo",
- &RowInput {
- label: "bob's",
- entries: vec![Entry::from("foo")]
+ &mut Row {
+ label: "bob's".to_owned(),
+ entries: HashMap::from([("foo".to_owned(), vec![None])]),
}
),
- HTML::from(" | ")
+ HTML::from(
+ r#" | "#
+ )
+ );
+ let mut r = Row {
+ label: "nope".to_owned(),
+ entries: HashMap::from([
+ ("foo".to_owned(), vec![None]),
+ ("baz".to_owned(), vec![None]),
+ ]),
+ };
+ assert_eq!(r.entries.len(), 2);
+ render_cell("foo", &mut r);
+ assert_eq!(r.entries.len(), 1);
+ render_cell("bar", &mut r);
+ assert_eq!(r.entries.len(), 1);
+ render_cell("baz", &mut r);
+ assert_eq!(r.entries.len(), 0);
+ }
+
+ #[test]
+ fn test_render_leftovers() {
+ assert_eq!(
+ render_all_leftovers(&Row {
+ label: "nope".to_owned(),
+ entries: HashMap::from([("foo".to_owned(), vec![None])]),
+ }),
+ HTML::from("foo")
+ );
+ assert_eq!(
+ render_all_leftovers(&Row {
+ label: "nope".to_owned(),
+ entries: HashMap::from([
+ ("foo".to_owned(), vec![None]),
+ ("bar".to_owned(), vec![None])
+ ]),
+ }),
+ HTML::from("bar, foo")
+ );
+ assert_eq!(
+ render_all_leftovers(&Row {
+ label: "nope".to_owned(),
+ entries: HashMap::from([
+ ("foo".to_owned(), vec![None]),
+ ("bar".to_owned(), vec![None, None])
+ ]),
+ }),
+ HTML::from("bar: 2, foo")
+ );
+ }
+
+ #[test]
+ fn test_render_row() {
+ assert_eq!(
+ render_row(
+ &["foo".to_owned()],
+ &mut Rowlike::Row(Row {
+ label: "nope".to_owned(),
+ entries: HashMap::from([("bar".to_owned(), vec![None])]),
+ })
+ ),
+ HTML::from(
+ r#"| nope | | bar |
|---|
+"#
+ )
);
}
}
|---|