]> git.scottworley.com Git - tablify/blobdiff - src/lib.rs
Read column threshold from `!col_threshold <N>` in input
[tablify] / src / lib.rs
index 47e1d3a9eb1fa10fac60a16cf66153b55d5d725f..b52806a744df217691d50d43d471f01cdbd15e0d 100644 (file)
@@ -4,7 +4,20 @@ use std::fmt::Write;
 use std::io::BufRead;
 use std::iter::Iterator;
 
-pub struct Config {}
+#[derive(PartialEq, Eq, Debug)]
+struct Config {
+    column_threshold: usize,
+}
+impl Config {
+    fn apply_command(&mut self, cmd: &str) -> Result<(), std::io::Error> {
+        if let Some(threshold) = cmd.strip_prefix("col_threshold ") {
+            self.column_threshold = threshold
+                .parse()
+                .map_err(|e| std::io::Error::new(std::io::ErrorKind::InvalidInput, e))?;
+        }
+        Ok(())
+    }
+}
 
 const HEADER: &str = r#"<!DOCTYPE html>
 <html>
@@ -16,11 +29,13 @@ const HEADER: &str = r#"<!DOCTYPE html>
     /* 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; }
+    th.spacer_row { height: .3em; }
     table { border-collapse: collapse }
     tr.key > th { height: 10em; 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.leftover { text-align: left; border: none; padding-left: .4em; }
     td.yes { border: thin solid gray; background-color: #ddd; }
     /* 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; }
@@ -83,12 +98,15 @@ enum InputLine<'a> {
     Blank,
     RowHeader(&'a str),
     Entry(&'a str, Option<&'a str>),
+    Command(&'a str),
 }
 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 let Some(cmd) = trimmed.strip_prefix('!') {
+            InputLine::Command(cmd)
         } else if !trimmed.starts_with(' ') {
             InputLine::RowHeader(value.trim())
         } else {
@@ -106,30 +124,45 @@ struct Row {
     entries: HashMap<String, Vec<Option<String>>>,
 }
 
-struct Reader<Input: Iterator<Item = Result<String, std::io::Error>>> {
+#[derive(Debug, PartialEq, Eq)]
+enum Rowlike {
+    Row(Row),
+    Spacer,
+}
+
+struct Reader<'cfg, Input: Iterator<Item = Result<String, std::io::Error>>> {
     input: std::iter::Enumerate<Input>,
     row: Option<Row>,
+    config: &'cfg mut Config,
 }
-impl<Input: Iterator<Item = Result<String, std::io::Error>>> Reader<Input> {
-    fn new(input: Input) -> Self {
+impl<'cfg, Input: Iterator<Item = Result<String, std::io::Error>>> Reader<'cfg, Input> {
+    fn new(config: &'cfg mut Config, input: Input) -> Self {
         Self {
             input: input.enumerate(),
             row: None,
+            config,
         }
     }
 }
-impl<Input: Iterator<Item = Result<String, std::io::Error>>> Iterator for Reader<Input> {
-    type Item = Result<Row, std::io::Error>;
+impl<'cfg, Input: Iterator<Item = Result<String, std::io::Error>>> Iterator
+    for Reader<'cfg, Input>
+{
+    type Item = Result<Rowlike, std::io::Error>;
     fn next(&mut self) -> Option<Self::Item> {
         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((n, Ok(line))) => match InputLine::from(line.as_ref()) {
+                    InputLine::Command(cmd) => {
+                        if let Err(e) = self.config.apply_command(cmd) {
+                            return Some(Err(e));
+                        }
+                    }
                     InputLine::Blank if self.row.is_some() => {
-                        return Ok(std::mem::take(&mut self.row)).transpose()
+                        return Ok(std::mem::take(&mut self.row).map(Rowlike::Row)).transpose()
                     }
-                    InputLine::Blank => {}
+                    InputLine::Blank => return Some(Ok(Rowlike::Spacer)),
                     InputLine::Entry(col, instance) => match &mut self.row {
                         None => {
                             return Some(Err(std::io::Error::other(format!(
@@ -151,7 +184,7 @@ impl<Input: Iterator<Item = Result<String, std::io::Error>>> Iterator for Reader
                             entries: HashMap::new(),
                         });
                         if prev.is_some() {
-                            return Ok(prev).transpose();
+                            return Ok(prev.map(Rowlike::Row)).transpose();
                         }
                     }
                 },
@@ -160,14 +193,24 @@ impl<Input: Iterator<Item = Result<String, std::io::Error>>> Iterator for Reader
     }
 }
 
-fn read_rows(input: impl std::io::Read) -> impl Iterator<Item = Result<Row, std::io::Error>> {
-    Reader::new(std::io::BufReader::new(input).lines())
+fn read_input(input: impl std::io::Read) -> Result<(Vec<Rowlike>, Config), std::io::Error> {
+    let mut config = Config {
+        column_threshold: 2,
+    };
+    let reader = Reader::new(&mut config, std::io::BufReader::new(input).lines());
+    reader
+        .collect::<Result<Vec<_>, _>>()
+        .map(|rows| (rows, config))
 }
 
-fn column_counts(rows: &[Row]) -> 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.keys())
+        .flat_map(|rl| match rl {
+            Rowlike::Row(r) => r.entries.keys(),
+            Rowlike::Spacer => empty.keys(),
+        })
         .fold(HashMap::new(), |mut cs, col| {
             cs.entry(col.to_owned())
                 .and_modify(|n| *n += 1)
@@ -180,43 +223,46 @@ fn column_counts(rows: &[Row]) -> Vec<(usize, String)> {
     counts.sort_unstable_by(|(an, acol), (bn, bcol)| bn.cmp(an).then(acol.cmp(bcol)));
     counts
 }
-fn column_order(rows: &[Row]) -> Vec<String> {
+fn column_order(config: &Config, rows: &[Rowlike]) -> Vec<String> {
     column_counts(rows)
         .into_iter()
-        .map(|(_, col)| col)
+        .filter_map(|(n, col)| (n >= config.column_threshold).then_some(col))
         .collect()
 }
 
-fn render_instance(instance: &Option<String>) -> HTML {
+fn render_one_instance(instance: &Option<String>) -> HTML {
     match instance {
         None => HTML::from("✓"),
         Some(instance) => HTML::escape(instance.as_ref()),
     }
 }
 
-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<Option<String>>> = row.entries.get(col);
-    let class = HTML::from(if instances.is_none() { "" } else { "yes" });
-    let all_empty = instances
-        .iter()
-        .flat_map(|is| is.iter())
-        .all(Option::is_none);
-    let contents = if instances.is_none() || (all_empty && instances.unwrap().len() == 1) {
+fn render_instances(instances: &[Option<String>]) -> HTML {
+    let all_empty = instances.iter().all(Option::is_none);
+    if all_empty && instances.len() == 1 {
         HTML::from("")
     } else if all_empty {
-        HTML(format!("{}", instances.unwrap().len()))
+        HTML(format!("{}", instances.len()))
     } else {
         HTML(
             instances
-                .unwrap()
                 .iter()
-                .map(render_instance)
+                .map(render_one_instance)
                 .map(|html| html.0) // Waiting for slice_concat_trait to stabilize
                 .collect::<Vec<_>>()
                 .join(" "),
         )
+    }
+}
+
+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<Option<String>>> = 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!(
@@ -224,15 +270,44 @@ fn render_cell(col: &str, row: &mut Row) -> HTML {
     ))
 }
 
-fn render_row(columns: &[String], row: &mut Row) -> HTML {
-    let row_label = HTML::escape(row.label.as_ref());
-    HTML(format!(
-        "<tr><th id=\"{row_label}\">{row_label}</th>{}</tr>\n",
-        &columns
-            .iter()
-            .map(|col| render_cell(col, row))
-            .collect::<HTML>()
-    ))
+fn render_leftover(notcol: &str, instances: &[Option<String>]) -> 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::<Vec<_>>()
+            .join(", "),
+    )
+}
+
+fn render_row(columns: &[String], rowlike: &mut Rowlike) -> HTML {
+    match rowlike {
+        Rowlike::Spacer => HTML::from("<tr><th class=\"spacer_row\"></th></tr>\n"),
+        Rowlike::Row(row) => {
+            let row_label = HTML::escape(row.label.as_ref());
+            let cells = columns
+                .iter()
+                .map(|col| render_cell(col, row))
+                .collect::<HTML>();
+            let leftovers = render_all_leftovers(row);
+            HTML(format!(
+                "<tr><th id=\"{row_label}\">{row_label}</th>{cells}<td class=\"leftover\" onmouseover=\"highlight('{row_label}')\" onmouseout=\"clear_highlight('{row_label}')\">{leftovers}</td></tr>\n"
+            ))
+        }
+    }
 }
 
 fn render_column_headers(columns: &[String]) -> HTML {
@@ -257,9 +332,9 @@ 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(config: &Config, input: impl std::io::Read) -> Result<HTML, std::io::Error> {
-    let rows = read_rows(input).collect::<Result<Vec<_>, _>>()?;
-    let columns = column_order(&rows);
+pub fn tablify(input: impl std::io::Read) -> Result<HTML, std::io::Error> {
+    let (rows, config) = read_input(input)?;
+    let columns = column_order(&config, &rows);
     Ok(HTML(format!(
         "{HEADER}{}{}{FOOTER}",
         render_column_headers(&columns),
@@ -302,134 +377,146 @@ mod tests {
         );
     }
 
+    fn read_rows(input: impl std::io::Read) -> Result<Vec<Rowlike>, std::io::Error> {
+        read_input(input).map(|(rows, _)| rows)
+    }
+    fn read_config(input: impl std::io::Read) -> Result<Config, std::io::Error> {
+        read_input(input).map(|(_, config)| config)
+    }
     #[test]
     fn test_read_rows() {
         assert_eq!(
-            read_rows(&b"foo"[..]).flatten().collect::<Vec<_>>(),
-            vec![Row {
+            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<_>>(),
-            vec![Row {
+            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::<Vec<_>>(),
+            read_rows(&b"foo\nbar\n"[..]).unwrap(),
             vec![
-                Row {
+                Rowlike::Row(Row {
                     label: "foo".to_owned(),
                     entries: HashMap::new(),
-                },
-                Row {
+                }),
+                Rowlike::Row(Row {
                     label: "bar".to_owned(),
                     entries: HashMap::new(),
-                }
+                })
             ]
         );
         assert_eq!(
-            read_rows(&b"foo\n bar\n"[..]).flatten().collect::<Vec<_>>(),
-            vec![Row {
+            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<_>>(),
-            vec![Row {
+            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::<Vec<_>>(),
+            read_rows(&b"foo\n\nbar\n"[..]).unwrap(),
             vec![
-                Row {
+                Rowlike::Row(Row {
                     label: "foo".to_owned(),
                     entries: HashMap::new(),
-                },
-                Row {
+                }),
+                Rowlike::Row(Row {
                     label: "bar".to_owned(),
                     entries: HashMap::new(),
-                }
+                })
             ]
         );
         assert_eq!(
-            read_rows(&b"foo\n \nbar\n"[..])
-                .flatten()
-                .collect::<Vec<_>>(),
+            read_rows(&b"foo\n\n\nbar\n"[..]).unwrap(),
             vec![
-                Row {
+                Rowlike::Row(Row {
                     label: "foo".to_owned(),
                     entries: HashMap::new(),
-                },
-                Row {
+                }),
+                Rowlike::Spacer,
+                Rowlike::Row(Row {
                     label: "bar".to_owned(),
                     entries: HashMap::new(),
-                }
+                })
             ]
         );
         assert_eq!(
-            read_rows(&b"foo  \n bar  \n"[..])
-                .flatten()
-                .collect::<Vec<_>>(),
-            vec![Row {
+            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"[..]).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"));
     }
 
+    #[test]
+    fn test_read_config() {
+        assert_eq!(
+            read_config(&b"!col_threshold 10"[..]).unwrap(),
+            Config {
+                column_threshold: 10
+            }
+        );
+
+        let bad_num = read_config(&b"!col_threshold foo"[..]);
+        assert!(bad_num.is_err());
+        assert!(format!("{bad_num:?}").contains("Parse"));
+    }
+
     #[test]
     fn test_column_counts() {
         assert_eq!(
-            column_counts(
-                &read_rows(&b"foo\n bar\n baz\n"[..])
-                    .collect::<Result<Vec<_>, _>>()
-                    .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::<Result<Vec<_>, _>>()
-                    .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::<Result<Vec<_>, _>>()
-                    .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::<Result<Vec<_>, _>>()
-                    .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"))]
         );
@@ -552,18 +639,49 @@ mod tests {
         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 Row {
+                &mut Rowlike::Row(Row {
                     label: "nope".to_owned(),
                     entries: HashMap::from([("bar".to_owned(), vec![None])]),
-                }
+                })
             ),
             HTML::from(
-                r#"<tr><th id="nope">nope</th><td class="" onmouseover="h2('nope','foo')" onmouseout="ch2('nope','foo')"></td></tr>
+                r#"<tr><th id="nope">nope</th><td class="" onmouseover="h2('nope','foo')" onmouseout="ch2('nope','foo')"></td><td class="leftover" onmouseover="highlight('nope')" onmouseout="clear_highlight('nope')">bar</td></tr>
 "#
             )
         );