Don't be O(n^2) unnecessarily

[tablify] / src / lib.rs

diff --git a/src/lib.rs b/src/lib.rs
index 370eb5ddd553a80f2cb4ce75c24703bf285e7b91..b406f957f50b457cdc53bbc70746b26176bc1d5e 100644 (file)
--- a/src/lib.rs
+++ b/src/lib.rs
@@ -1,4 +1,5 @@
-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;
 use std::fmt::Write;
 use std::io::BufRead;
 use std::iter::Iterator;


@@ -75,35 +76,37 @@ impl std::fmt::Display for HTML {
     }
 }
 
     }
 }
 

-#[derive(Debug, PartialEq, Eq, Hash)]
-struct Entry {
-    col: String,
-    instance: Option<String>,
+#[derive(Debug, PartialEq, Eq)]
+enum InputLine<'a> {
+    Blank,
+    RowHeader(&'a str),
+    Entry(&'a str, Option<&'a str>),

 }
 }

-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())),
-            },
+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)]
         }
     }
 }
 
 #[derive(Debug, PartialEq, Eq)]

-struct RowInput {
+struct Row {

     label: String,
     label: String,

-    entries: Vec<Entry>,
+    entries: HashMap<String, Vec<Option<String>>>,

 }
 
 struct Reader<Input: Iterator<Item = Result<String, std::io::Error>>> {
     input: std::iter::Enumerate<Input>,
 }
 
 struct Reader<Input: Iterator<Item = Result<String, std::io::Error>>> {
     input: std::iter::Enumerate<Input>,

-    row: Option<RowInput>,
+    row: Option<Row>,

 }
 impl<Input: Iterator<Item = Result<String, std::io::Error>>> Reader<Input> {
     fn new(input: Input) -> Self {
 }
 impl<Input: Iterator<Item = Result<String, std::io::Error>>> Reader<Input> {
     fn new(input: Input) -> Self {


@@ -114,60 +117,57 @@ impl<Input: Iterator<Item = Result<String, std::io::Error>>> Reader<Input> {
     }
 }
 impl<Input: Iterator<Item = Result<String, std::io::Error>>> Iterator for Reader<Input> {
     }
 }
 impl<Input: Iterator<Item = Result<String, std::io::Error>>> Iterator for Reader<Input> {

-    type Item = Result<RowInput, std::io::Error>;
+    type Item = Result<Row, std::io::Error>;

     fn next(&mut self) -> Option<Self::Item> {
         loop {
     fn next(&mut self) -> Option<Self::Item> {
         loop {

-            match self
-                .input
-                .next()
-                .map(|(n, r)| (n, r.map(|line| String::from(line.trim_end()))))
-            {
+            match self.input.next() {

                 None => return Ok(std::mem::take(&mut self.row)).transpose(),
                 Some((_, Err(e))) => return Some(Err(e)),
                 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((n, Ok(line))) => match InputLine::from(line.as_ref()) {
+                    InputLine::Blank if self.row.is_some() => {
+                        return Ok(std::mem::take(&mut self.row)).transpose()

                     }
                     }

-                    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();
+                    InputLine::Blank => {}
+                    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).transpose();
+                        }

                     }
                     }

-                }
+                },

             }
         }
     }
 }
 
             }
         }
     }
 }
 

-fn read_rows(input: impl std::io::Read) -> impl Iterator<Item = Result<RowInput, std::io::Error>> {
+fn read_rows(input: impl std::io::Read) -> impl Iterator<Item = Result<Row, std::io::Error>> {

     Reader::new(std::io::BufReader::new(input).lines())
 }
 
     Reader::new(std::io::BufReader::new(input).lines())
 }
 

-fn column_counts(rows: &[RowInput]) -> Vec<(usize, String)> {
+fn column_counts(rows: &[Row]) -> Vec<(usize, String)> {

     let mut counts: Vec<_> = rows
         .iter()
     let mut counts: Vec<_> = rows
         .iter()

-        .flat_map(|r| {
-            r.entries
-                .iter()
-                .map(|e| &e.col)
-                .collect::<HashSet<_>>()
-                .into_iter()
-        })
+        .flat_map(|r| r.entries.keys())

         .fold(HashMap::new(), |mut cs, col| {
         .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
                 .and_modify(|n| *n += 1)
                 .or_insert(1);
             cs


@@ -178,35 +178,39 @@ fn column_counts(rows: &[RowInput]) -> Vec<(usize, String)> {
     counts.sort_unstable_by(|(an, acol), (bn, bcol)| bn.cmp(an).then(acol.cmp(bcol)));
     counts
 }
     counts.sort_unstable_by(|(an, acol), (bn, bcol)| bn.cmp(an).then(acol.cmp(bcol)));
     counts
 }

-fn column_order(rows: &[RowInput]) -> Vec<String> {
+fn column_order(rows: &[Row]) -> Vec<String> {

     column_counts(rows)
         .into_iter()
         .map(|(_, col)| col)
         .collect()
 }
 
     column_counts(rows)
         .into_iter()
         .map(|(_, col)| col)
         .collect()
 }
 

-fn render_instance(entry: &Entry) -> HTML {
-    match &entry.instance {
+fn render_instance(instance: &Option<String>) -> HTML {
+    match instance {

         None => HTML::from("✓"),
         Some(instance) => HTML::escape(instance.as_ref()),
     }
 }
 
         None => HTML::from("✓"),
         Some(instance) => HTML::escape(instance.as_ref()),
     }
 }
 

-fn render_cell(col: &str, row: &RowInput) -> HTML {
+fn render_cell(col: &str, row: &Row) -> HTML {

     let row_label = HTML::escape(row.label.as_ref());
     let col_label = HTML::escape(col);
     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) {
+    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) {

         HTML::from("")
     } else if all_empty {
         HTML::from("")
     } else if all_empty {

-        HTML(format!("{}", entries.len()))
+        HTML(format!("{}", instances.unwrap().len()))

     } else {
         HTML(
     } else {
         HTML(

-            entries
+            instances
+                .unwrap()

                 .iter()
                 .iter()

-                .map(|i| render_instance(i))
+                .map(render_instance)

                 .map(|html| html.0) // Waiting for slice_concat_trait to stabilize
                 .collect::<Vec<_>>()
                 .join(" "),
                 .map(|html| html.0) // Waiting for slice_concat_trait to stabilize
                 .collect::<Vec<_>>()
                 .join(" "),


@@ -215,8 +219,7 @@ fn render_cell(col: &str, row: &RowInput) -> HTML {
     HTML(format!("<td class=\"{class}\" onmouseover=\"h2('{row_label}','{col_label}')\" onmouseout=\"ch2('{row_label}','{col_label}')\">{contents}</td>"))
 }
 
     HTML(format!("<td class=\"{class}\" onmouseover=\"h2('{row_label}','{col_label}')\" onmouseout=\"ch2('{row_label}','{col_label}')\">{contents}</td>"))
 }
 

-fn render_row(columns: &[String], row: &RowInput) -> HTML {
-    // This is O(n^2) & doesn't need to be
+fn render_row(columns: &[String], row: &Row) -> HTML {

     let row_label = HTML::escape(row.label.as_ref());
     HTML(format!(
         "<tr><th id=\"{row_label}\">{row_label}</th>{}</tr>\n",
     let row_label = HTML::escape(row.label.as_ref());
     HTML(format!(
         "<tr><th id=\"{row_label}\">{row_label}</th>{}</tr>\n",


@@ -266,27 +269,31 @@ mod tests {
     use super::*;
 
     #[test]
     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!(
         assert_eq!(

-            Entry::from("foo"),
-            Entry {
-                col: String::from("foo"),
-                instance: None
-            }
+            InputLine::from(" foo:bar"),
+            InputLine::Entry("foo", Some("bar"))

         );
         assert_eq!(
         );
         assert_eq!(

-            Entry::from("foo:bar"),
-            Entry {
-                col: String::from("foo"),
-                instance: Some(String::from("bar"))
-            }
+            InputLine::from(" foo: bar"),
+            InputLine::Entry("foo", Some("bar"))

         );
         assert_eq!(
         );
         assert_eq!(

-            Entry::from("foo: bar"),
-            Entry {
-                col: String::from("foo"),
-                instance: Some(String::from("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"))

         );
     }
 
         );
     }
 


@@ -294,45 +301,48 @@ mod tests {
     fn test_read_rows() {
         assert_eq!(
             read_rows(&b"foo"[..]).flatten().collect::<Vec<_>>(),
     fn test_read_rows() {
         assert_eq!(
             read_rows(&b"foo"[..]).flatten().collect::<Vec<_>>(),

-            vec![RowInput {
-                label: String::from("foo"),
-                entries: vec![]
+            vec![Row {
+                label: "foo".to_owned(),
+                entries: HashMap::new(),

             }]
         );
         assert_eq!(
             read_rows(&b"bar"[..]).flatten().collect::<Vec<_>>(),
             }]
         );
         assert_eq!(
             read_rows(&b"bar"[..]).flatten().collect::<Vec<_>>(),

-            vec![RowInput {
-                label: String::from("bar"),
-                entries: vec![]
+            vec![Row {
+                label: "bar".to_owned(),
+                entries: HashMap::new(),

             }]
         );
         assert_eq!(
             read_rows(&b"foo\nbar\n"[..]).flatten().collect::<Vec<_>>(),
             vec![
             }]
         );
         assert_eq!(
             read_rows(&b"foo\nbar\n"[..]).flatten().collect::<Vec<_>>(),
             vec![

-                RowInput {
-                    label: String::from("foo"),
-                    entries: vec![]
+                Row {
+                    label: "foo".to_owned(),
+                    entries: HashMap::new(),

                 },
                 },

-                RowInput {
-                    label: String::from("bar"),
-                    entries: vec![]
+                Row {
+                    label: "bar".to_owned(),
+                    entries: HashMap::new(),

                 }
             ]
         );
         assert_eq!(
             read_rows(&b"foo\n bar\n"[..]).flatten().collect::<Vec<_>>(),
                 }
             ]
         );
         assert_eq!(
             read_rows(&b"foo\n bar\n"[..]).flatten().collect::<Vec<_>>(),

-            vec![RowInput {
-                label: String::from("foo"),
-                entries: vec![Entry::from("bar")]
+            vec![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<_>>(),
             }]
         );
         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")]
+            vec![Row {
+                label: "foo".to_owned(),
+                entries: HashMap::from([
+                    ("bar".to_owned(), vec![None]),
+                    ("baz".to_owned(), vec![None])
+                ]),

             }]
         );
         assert_eq!(
             }]
         );
         assert_eq!(


@@ -340,13 +350,13 @@ mod tests {
                 .flatten()
                 .collect::<Vec<_>>(),
             vec![
                 .flatten()
                 .collect::<Vec<_>>(),
             vec![

-                RowInput {
-                    label: String::from("foo"),
-                    entries: vec![]
+                Row {
+                    label: "foo".to_owned(),
+                    entries: HashMap::new(),

                 },
                 },

-                RowInput {
-                    label: String::from("bar"),
-                    entries: vec![]
+                Row {
+                    label: "bar".to_owned(),
+                    entries: HashMap::new(),

                 }
             ]
         );
                 }
             ]
         );


@@ -355,13 +365,13 @@ mod tests {
                 .flatten()
                 .collect::<Vec<_>>(),
             vec![
                 .flatten()
                 .collect::<Vec<_>>(),
             vec![

-                RowInput {
-                    label: String::from("foo"),
-                    entries: vec![]
+                Row {
+                    label: "foo".to_owned(),
+                    entries: HashMap::new(),

                 },
                 },

-                RowInput {
-                    label: String::from("bar"),
-                    entries: vec![]
+                Row {
+                    label: "bar".to_owned(),
+                    entries: HashMap::new(),

                 }
             ]
         );
                 }
             ]
         );


@@ -369,9 +379,9 @@ mod tests {
             read_rows(&b"foo  \n bar  \n"[..])
                 .flatten()
                 .collect::<Vec<_>>(),
             read_rows(&b"foo  \n bar  \n"[..])
                 .flatten()
                 .collect::<Vec<_>>(),

-            vec![RowInput {
-                label: String::from("foo"),
-                entries: vec![Entry::from("bar")]
+            vec![Row {
+                label: "foo".to_owned(),
+                entries: HashMap::from([("bar".to_owned(), vec![None])]),

             }]
         );
 
             }]
         );
 


@@ -425,9 +435,9 @@ mod tests {
         assert_eq!(
             render_cell(
                 "foo",
         assert_eq!(
             render_cell(
                 "foo",

-                &RowInput {
-                    label: String::from("nope"),
-                    entries: vec![]
+                &Row {
+                    label: "nope".to_owned(),
+                    entries: HashMap::new(),

                 }
             ),
             HTML::from("<td class=\"\" onmouseover=\"h2('nope','foo')\" onmouseout=\"ch2('nope','foo')\"></td>")
                 }
             ),
             HTML::from("<td class=\"\" onmouseover=\"h2('nope','foo')\" onmouseout=\"ch2('nope','foo')\"></td>")


@@ -435,9 +445,9 @@ mod tests {
         assert_eq!(
             render_cell(
                 "foo",
         assert_eq!(
             render_cell(
                 "foo",

-                &RowInput {
-                    label: String::from("nope"),
-                    entries: vec![Entry::from("bar")]
+                &Row {
+                    label: "nope".to_owned(),
+                    entries: HashMap::from([("bar".to_owned(), vec![None])]),

                 }
             ),
             HTML::from("<td class=\"\" onmouseover=\"h2('nope','foo')\" onmouseout=\"ch2('nope','foo')\"></td>")
                 }
             ),
             HTML::from("<td class=\"\" onmouseover=\"h2('nope','foo')\" onmouseout=\"ch2('nope','foo')\"></td>")


@@ -445,9 +455,9 @@ mod tests {
         assert_eq!(
             render_cell(
                 "foo",
         assert_eq!(
             render_cell(
                 "foo",

-                &RowInput {
-                    label: String::from("nope"),
-                    entries: vec![Entry::from("foo")]
+                &Row {
+                    label: "nope".to_owned(),
+                    entries: HashMap::from([("foo".to_owned(), vec![None])]),

                 }
             ),
             HTML::from("<td class=\"yes\" onmouseover=\"h2('nope','foo')\" onmouseout=\"ch2('nope','foo')\"></td>")
                 }
             ),
             HTML::from("<td class=\"yes\" onmouseover=\"h2('nope','foo')\" onmouseout=\"ch2('nope','foo')\"></td>")


@@ -455,9 +465,9 @@ mod tests {
         assert_eq!(
             render_cell(
                 "foo",
         assert_eq!(
             render_cell(
                 "foo",

-                &RowInput {
-                    label: String::from("nope"),
-                    entries: vec![Entry::from("foo"), Entry::from("foo")]
+                &Row {
+                    label: "nope".to_owned(),
+                    entries: HashMap::from([("foo".to_owned(), vec![None, None])]),

                 }
             ),
             HTML::from("<td class=\"yes\" onmouseover=\"h2('nope','foo')\" onmouseout=\"ch2('nope','foo')\">2</td>")
                 }
             ),
             HTML::from("<td class=\"yes\" onmouseover=\"h2('nope','foo')\" onmouseout=\"ch2('nope','foo')\">2</td>")


@@ -465,9 +475,9 @@ mod tests {
         assert_eq!(
             render_cell(
                 "foo",
         assert_eq!(
             render_cell(
                 "foo",

-                &RowInput {
-                    label: String::from("nope"),
-                    entries: vec![Entry::from("foo: 5"), Entry::from("foo: 10")]
+                &Row {
+                    label: "nope".to_owned(),
+                    entries: HashMap::from([("foo".to_owned(), vec![Some("5".to_owned()), Some("10".to_owned())])]),

                 }
             ),
             HTML::from("<td class=\"yes\" onmouseover=\"h2('nope','foo')\" onmouseout=\"ch2('nope','foo')\">5 10</td>")
                 }
             ),
             HTML::from("<td class=\"yes\" onmouseover=\"h2('nope','foo')\" onmouseout=\"ch2('nope','foo')\">5 10</td>")


@@ -475,9 +485,9 @@ mod tests {
         assert_eq!(
             render_cell(
                 "foo",
         assert_eq!(
             render_cell(
                 "foo",

-                &RowInput {
-                    label: String::from("nope"),
-                    entries: vec![Entry::from("foo: 5"), Entry::from("foo")]
+                &Row {
+                    label: "nope".to_owned(),
+                    entries: HashMap::from([("foo".to_owned(), vec![Some("5".to_owned()), None])]),

                 }
             ),
             HTML::from("<td class=\"yes\" onmouseover=\"h2('nope','foo')\" onmouseout=\"ch2('nope','foo')\">5 ✓</td>")
                 }
             ),
             HTML::from("<td class=\"yes\" onmouseover=\"h2('nope','foo')\" onmouseout=\"ch2('nope','foo')\">5 ✓</td>")


@@ -485,9 +495,9 @@ mod tests {
         assert_eq!(
             render_cell(
                 "heart",
         assert_eq!(
             render_cell(
                 "heart",

-                &RowInput {
-                    label: String::from("nope"),
-                    entries: vec![Entry::from("heart: <3")]
+                &Row {
+                    label: "nope".to_owned(),
+                    entries: HashMap::from([("heart".to_owned(), vec![Some("<3".to_owned())])]),

                 }
             ),
             HTML::from("<td class=\"yes\" onmouseover=\"h2('nope','heart')\" onmouseout=\"ch2('nope','heart')\">&lt;3</td>")
                 }
             ),
             HTML::from("<td class=\"yes\" onmouseover=\"h2('nope','heart')\" onmouseout=\"ch2('nope','heart')\">&lt;3</td>")


@@ -495,9 +505,9 @@ mod tests {
         assert_eq!(
             render_cell(
                 "foo",
         assert_eq!(
             render_cell(
                 "foo",

-                &RowInput {
-                    label: String::from("bob's"),
-                    entries: vec![Entry::from("foo")]
+                &Row {
+                    label: "bob's".to_owned(),
+                    entries: HashMap::from([("foo".to_owned(), vec![None])]),

                 }
             ),
             HTML::from("<td class=\"yes\" onmouseover=\"h2('bob&#39;s','foo')\" onmouseout=\"ch2('bob&#39;s','foo')\"></td>")
                 }
             ),
             HTML::from("<td class=\"yes\" onmouseover=\"h2('bob&#39;s','foo')\" onmouseout=\"ch2('bob&#39;s','foo')\"></td>")