git.scottworley.com Git - tablify/blob - src/lib.rs

   1 use std::borrow::ToOwned;
   2 use std::collections::HashMap;
   3 use std::fmt::Write;
   4 use std::io::BufRead;
   5 use std::iter::Iterator;
   6
   7 pub struct Config {
   8     pub column_threshold: usize,
   9 }
  10
  11 const HEADER: &str = r#"<!DOCTYPE html>
  12 <html>
  13 <head>
  14   <meta charset="utf-8">
  15   <meta name="viewport" content="width=device-width, initial-scale=1">
  16   <style>
  17     td { text-align: center; }
  18     /* h/t https://wabain.github.io/2019/10/13/css-rotated-table-header.html */
  19     th, td { white-space: nowrap; }
  20     th { text-align: left; font-weight: normal; }
  21     table { border-collapse: collapse }
  22     tr.key > th { height: 10em; vertical-align: bottom; line-height: 1 }
  23     tr.key > th > div { width: 1em; }
  24     tr.key > th > div > div { width: 5em; transform-origin: bottom left; transform: translateX(1em) rotate(-65deg) }
  25     td { border: thin solid gray; }
  26     td.leftover { text-align: left; border: none; padding-left: .4em; }
  27     td.yes { border: thin solid gray; background-color: #ddd; }
  28     /* h/t https://stackoverflow.com/questions/5687035/css-bolding-some-text-without-changing-its-containers-size/46452396#46452396 */
  29     .highlight { text-shadow: -0.06ex 0 black, 0.06ex 0 black; }
  30   </style>
  31   <script>
  32     function highlight(id)       { const e = document.getElementById(id); if (e) { e.classList.add(   "highlight"); } }
  33     function clear_highlight(id) { const e = document.getElementById(id); if (e) { e.classList.remove("highlight"); } }
  34     function h2(a, b)  {       highlight(a);       highlight(b); }
  35     function ch2(a, b) { clear_highlight(a); clear_highlight(b); }
  36   </script>
  37 </head>
  38 <body>
  39   <table>
  40     <tbody>
  41 "#;
  42 const FOOTER: &str = "    </tbody>
  43   </table>
  44 </body>
  45 </html>";
  46
  47 #[derive(PartialEq, Eq, Debug)]
  48 pub struct HTML(String);
  49 impl HTML {
  50     fn escape(value: &str) -> HTML {
  51         let mut escaped: String = String::new();
  52         for c in value.chars() {
  53             match c {
  54                 '>' => escaped.push_str("&gt;"),
  55                 '<' => escaped.push_str("&lt;"),
  56                 '\'' => escaped.push_str("&#39;"),
  57                 '"' => escaped.push_str("&quot;"),
  58                 '&' => escaped.push_str("&amp;"),
  59                 ok_c => escaped.push(ok_c),
  60             }
  61         }
  62         HTML(escaped)
  63     }
  64 }
  65 impl From<&str> for HTML {
  66     fn from(value: &str) -> HTML {
  67         HTML(String::from(value))
  68     }
  69 }
  70 impl FromIterator<HTML> for HTML {
  71     fn from_iter<T>(iter: T) -> HTML
  72     where
  73         T: IntoIterator<Item = HTML>,
  74     {
  75         HTML(iter.into_iter().map(|html| html.0).collect::<String>())
  76     }
  77 }
  78 impl std::fmt::Display for HTML {
  79     fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
  80         write!(f, "{}", self.0)
  81     }
  82 }
  83
  84 #[derive(Debug, PartialEq, Eq)]
  85 enum InputLine<'a> {
  86     Blank,
  87     RowHeader(&'a str),
  88     Entry(&'a str, Option<&'a str>),
  89 }
  90 impl<'a> From<&'a str> for InputLine<'a> {
  91     fn from(value: &'a str) -> InputLine<'a> {
  92         let trimmed = value.trim_end();
  93         if trimmed.is_empty() {
  94             InputLine::Blank
  95         } else if !trimmed.starts_with(' ') {
  96             InputLine::RowHeader(value.trim())
  97         } else {
  98             match value.split_once(':') {
  99                 None => InputLine::Entry(value.trim(), None),
 100                 Some((col, instance)) => InputLine::Entry(col.trim(), Some(instance.trim())),
 101             }
 102         }
 103     }
 104 }
 105
 106 #[derive(Debug, PartialEq, Eq)]
 107 struct Row {
 108     label: String,
 109     entries: HashMap<String, Vec<Option<String>>>,
 110 }
 111
 112 struct Reader<Input: Iterator<Item = Result<String, std::io::Error>>> {
 113     input: std::iter::Enumerate<Input>,
 114     row: Option<Row>,
 115 }
 116 impl<Input: Iterator<Item = Result<String, std::io::Error>>> Reader<Input> {
 117     fn new(input: Input) -> Self {
 118         Self {
 119             input: input.enumerate(),
 120             row: None,
 121         }
 122     }
 123 }
 124 impl<Input: Iterator<Item = Result<String, std::io::Error>>> Iterator for Reader<Input> {
 125     type Item = Result<Row, std::io::Error>;
 126     fn next(&mut self) -> Option<Self::Item> {
 127         loop {
 128             match self.input.next() {
 129                 None => return Ok(std::mem::take(&mut self.row)).transpose(),
 130                 Some((_, Err(e))) => return Some(Err(e)),
 131                 Some((n, Ok(line))) => match InputLine::from(line.as_ref()) {
 132                     InputLine::Blank if self.row.is_some() => {
 133                         return Ok(std::mem::take(&mut self.row)).transpose()
 134                     }
 135                     InputLine::Blank => {}
 136                     InputLine::Entry(col, instance) => match &mut self.row {
 137                         None => {
 138                             return Some(Err(std::io::Error::other(format!(
 139                                 "{}: Entry with no header",
 140                                 n + 1
 141                             ))))
 142                         }
 143                         Some(ref mut row) => {
 144                             row.entries
 145                                 .entry(col.to_owned())
 146                                 .and_modify(|is| is.push(instance.map(ToOwned::to_owned)))
 147                                 .or_insert_with(|| vec![instance.map(ToOwned::to_owned)]);
 148                         }
 149                     },
 150                     InputLine::RowHeader(row) => {
 151                         let prev = std::mem::take(&mut self.row);
 152                         self.row = Some(Row {
 153                             label: row.to_owned(),
 154                             entries: HashMap::new(),
 155                         });
 156                         if prev.is_some() {
 157                             return Ok(prev).transpose();
 158                         }
 159                     }
 160                 },
 161             }
 162         }
 163     }
 164 }
 165
 166 fn read_rows(input: impl std::io::Read) -> impl Iterator<Item = Result<Row, std::io::Error>> {
 167     Reader::new(std::io::BufReader::new(input).lines())
 168 }
 169
 170 fn column_counts(rows: &[Row]) -> Vec<(usize, String)> {
 171     let mut counts: Vec<_> = rows
 172         .iter()
 173         .flat_map(|r| r.entries.keys())
 174         .fold(HashMap::new(), |mut cs, col| {
 175             cs.entry(col.to_owned())
 176                 .and_modify(|n| *n += 1)
 177                 .or_insert(1);
 178             cs
 179         })
 180         .into_iter()
 181         .map(|(col, n)| (n, col))
 182         .collect();
 183     counts.sort_unstable_by(|(an, acol), (bn, bcol)| bn.cmp(an).then(acol.cmp(bcol)));
 184     counts
 185 }
 186 fn column_order(config: &Config, rows: &[Row]) -> Vec<String> {
 187     column_counts(rows)
 188         .into_iter()
 189         .filter_map(|(n, col)| (n >= config.column_threshold).then_some(col))
 190         .collect()
 191 }
 192
 193 fn render_one_instance(instance: &Option<String>) -> HTML {
 194     match instance {
 195         None => HTML::from("✓"),
 196         Some(instance) => HTML::escape(instance.as_ref()),
 197     }
 198 }
 199
 200 fn render_instances(instances: &[Option<String>]) -> HTML {
 201     let all_empty = instances.iter().all(Option::is_none);
 202     if all_empty && instances.len() == 1 {
 203         HTML::from("")
 204     } else if all_empty {
 205         HTML(format!("{}", instances.len()))
 206     } else {
 207         HTML(
 208             instances
 209                 .iter()
 210                 .map(render_one_instance)
 211                 .map(|html| html.0) // Waiting for slice_concat_trait to stabilize
 212                 .collect::<Vec<_>>()
 213                 .join(" "),
 214         )
 215     }
 216 }
 217
 218 fn render_cell(col: &str, row: &mut Row) -> HTML {
 219     let row_label = HTML::escape(row.label.as_ref());
 220     let col_label = HTML::escape(col);
 221     let instances: Option<&Vec<Option<String>>> = row.entries.get(col);
 222     let class = HTML::from(if instances.is_none() { "" } else { "yes" });
 223     let contents = match instances {
 224         None => HTML::from(""),
 225         Some(is) => render_instances(is),
 226     };
 227     row.entries.remove(col);
 228     HTML(format!(
 229         r#"<td class="{class}" onmouseover="h2('{row_label}','{col_label}')" onmouseout="ch2('{row_label}','{col_label}')">{contents}</td>"#
 230     ))
 231 }
 232
 233 fn render_leftover(notcol: &str, instances: &[Option<String>]) -> HTML {
 234     let label = HTML::escape(notcol);
 235     let rest = render_instances(instances);
 236     if rest == HTML::from("") {
 237         HTML(format!("{label}"))
 238     } else {
 239         HTML(format!("{label}: {rest}"))
 240     }
 241 }
 242
 243 fn render_all_leftovers(row: &Row) -> HTML {
 244     let mut order: Vec<_> = row.entries.keys().collect();
 245     order.sort_unstable();
 246     HTML(
 247         order
 248             .into_iter()
 249             .map(|notcol| render_leftover(notcol, row.entries.get(notcol).expect("Key vanished?!")))
 250             .map(|html| html.0) // Waiting for slice_concat_trait to stabilize
 251             .collect::<Vec<_>>()
 252             .join(", "),
 253     )
 254 }
 255
 256 fn render_row(columns: &[String], row: &mut Row) -> HTML {
 257     let row_label = HTML::escape(row.label.as_ref());
 258     let cells = columns
 259         .iter()
 260         .map(|col| render_cell(col, row))
 261         .collect::<HTML>();
 262     let leftovers = render_all_leftovers(row);
 263     HTML(format!(
 264         "<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"
 265     ))
 266 }
 267
 268 fn render_column_headers(columns: &[String]) -> HTML {
 269     HTML(
 270         String::from(r#"<tr class="key"><th></th>"#)
 271             + &columns.iter().fold(String::new(), |mut acc, col| {
 272                 let col_header = HTML::escape(col.as_ref());
 273                 write!(
 274                     &mut acc,
 275                     r#"<th id="{col_header}"><div><div>{col_header}</div></div></th>"#
 276                 )
 277                 .unwrap();
 278                 acc
 279             })
 280             + "</tr>\n",
 281     )
 282 }
 283
 284 /// # Errors
 285 ///
 286 /// Will return `Err` if
 287 ///   * there's an i/o error while reading `input`
 288 ///   * the log has invalid syntax:
 289 ///     * an indented line with no preceding non-indented line
 290 pub fn tablify(config: &Config, input: impl std::io::Read) -> Result<HTML, std::io::Error> {
 291     let rows = read_rows(input).collect::<Result<Vec<_>, _>>()?;
 292     let columns = column_order(config, &rows);
 293     Ok(HTML(format!(
 294         "{HEADER}{}{}{FOOTER}",
 295         render_column_headers(&columns),
 296         rows.into_iter()
 297             .map(|mut r| render_row(&columns, &mut r))
 298             .collect::<HTML>()
 299     )))
 300 }
 301
 302 #[cfg(test)]
 303 mod tests {
 304     use super::*;
 305
 306     #[test]
 307     fn test_parse_line() {
 308         assert_eq!(InputLine::from(""), InputLine::Blank);
 309         assert_eq!(InputLine::from(" "), InputLine::Blank);
 310         assert_eq!(InputLine::from("foo"), InputLine::RowHeader("foo"));
 311         assert_eq!(InputLine::from("foo "), InputLine::RowHeader("foo"));
 312         assert_eq!(InputLine::from(" foo"), InputLine::Entry("foo", None));
 313         assert_eq!(
 314             InputLine::from(" foo:bar"),
 315             InputLine::Entry("foo", Some("bar"))
 316         );
 317         assert_eq!(
 318             InputLine::from(" foo: bar"),
 319             InputLine::Entry("foo", Some("bar"))
 320         );
 321         assert_eq!(
 322             InputLine::from(" foo: bar "),
 323             InputLine::Entry("foo", Some("bar"))
 324         );
 325         assert_eq!(
 326             InputLine::from(" foo:  bar "),
 327             InputLine::Entry("foo", Some("bar"))
 328         );
 329         assert_eq!(
 330             InputLine::from(" foo :  bar "),
 331             InputLine::Entry("foo", Some("bar"))
 332         );
 333     }
 334
 335     #[test]
 336     fn test_read_rows() {
 337         assert_eq!(
 338             read_rows(&b"foo"[..]).flatten().collect::<Vec<_>>(),
 339             vec![Row {
 340                 label: "foo".to_owned(),
 341                 entries: HashMap::new(),
 342             }]
 343         );
 344         assert_eq!(
 345             read_rows(&b"bar"[..]).flatten().collect::<Vec<_>>(),
 346             vec![Row {
 347                 label: "bar".to_owned(),
 348                 entries: HashMap::new(),
 349             }]
 350         );
 351         assert_eq!(
 352             read_rows(&b"foo\nbar\n"[..]).flatten().collect::<Vec<_>>(),
 353             vec![
 354                 Row {
 355                     label: "foo".to_owned(),
 356                     entries: HashMap::new(),
 357                 },
 358                 Row {
 359                     label: "bar".to_owned(),
 360                     entries: HashMap::new(),
 361                 }
 362             ]
 363         );
 364         assert_eq!(
 365             read_rows(&b"foo\n bar\n"[..]).flatten().collect::<Vec<_>>(),
 366             vec![Row {
 367                 label: "foo".to_owned(),
 368                 entries: HashMap::from([("bar".to_owned(), vec![None])]),
 369             }]
 370         );
 371         assert_eq!(
 372             read_rows(&b"foo\n bar\n baz\n"[..])
 373                 .flatten()
 374                 .collect::<Vec<_>>(),
 375             vec![Row {
 376                 label: "foo".to_owned(),
 377                 entries: HashMap::from([
 378                     ("bar".to_owned(), vec![None]),
 379                     ("baz".to_owned(), vec![None])
 380                 ]),
 381             }]
 382         );
 383         assert_eq!(
 384             read_rows(&b"foo\n\nbar\n"[..])
 385                 .flatten()
 386                 .collect::<Vec<_>>(),
 387             vec![
 388                 Row {
 389                     label: "foo".to_owned(),
 390                     entries: HashMap::new(),
 391                 },
 392                 Row {
 393                     label: "bar".to_owned(),
 394                     entries: HashMap::new(),
 395                 }
 396             ]
 397         );
 398         assert_eq!(
 399             read_rows(&b"foo\n \nbar\n"[..])
 400                 .flatten()
 401                 .collect::<Vec<_>>(),
 402             vec![
 403                 Row {
 404                     label: "foo".to_owned(),
 405                     entries: HashMap::new(),
 406                 },
 407                 Row {
 408                     label: "bar".to_owned(),
 409                     entries: HashMap::new(),
 410                 }
 411             ]
 412         );
 413         assert_eq!(
 414             read_rows(&b"foo  \n bar  \n"[..])
 415                 .flatten()
 416                 .collect::<Vec<_>>(),
 417             vec![Row {
 418                 label: "foo".to_owned(),
 419                 entries: HashMap::from([("bar".to_owned(), vec![None])]),
 420             }]
 421         );
 422
 423         let bad = read_rows(&b" foo"[..]).next().unwrap();
 424         assert!(bad.is_err());
 425         assert!(format!("{bad:?}").contains("1: Entry with no header"));
 426
 427         let bad2 = read_rows(&b"foo\n\n bar"[..]).nth(1).unwrap();
 428         assert!(bad2.is_err());
 429         assert!(format!("{bad2:?}").contains("3: Entry with no header"));
 430     }
 431
 432     #[test]
 433     fn test_column_counts() {
 434         assert_eq!(
 435             column_counts(
 436                 &read_rows(&b"foo\n bar\n baz\n"[..])
 437                     .collect::<Result<Vec<_>, _>>()
 438                     .unwrap()
 439             ),
 440             vec![(1, String::from("bar")), (1, String::from("baz"))]
 441         );
 442         assert_eq!(
 443             column_counts(
 444                 &read_rows(&b"foo\n bar\n baz\nquux\n baz"[..])
 445                     .collect::<Result<Vec<_>, _>>()
 446                     .unwrap()
 447             ),
 448             vec![(2, String::from("baz")), (1, String::from("bar"))]
 449         );
 450         assert_eq!(
 451             column_counts(
 452                 &read_rows(&b"foo\n bar\n bar\n baz\n bar\nquux\n baz"[..])
 453                     .collect::<Result<Vec<_>, _>>()
 454                     .unwrap()
 455             ),
 456             vec![(2, String::from("baz")), (1, String::from("bar"))]
 457         );
 458         assert_eq!(
 459             column_counts(
 460                 &read_rows(&b"foo\n bar: 1\n bar: 2\n baz\n bar\nquux\n baz"[..])
 461                     .collect::<Result<Vec<_>, _>>()
 462                     .unwrap()
 463             ),
 464             vec![(2, String::from("baz")), (1, String::from("bar"))]
 465         );
 466     }
 467
 468     #[test]
 469     fn test_render_cell() {
 470         assert_eq!(
 471             render_cell(
 472                 "foo",
 473                 &mut Row {
 474                     label: "nope".to_owned(),
 475                     entries: HashMap::new(),
 476                 }
 477             ),
 478             HTML::from(
 479                 r#"<td class="" onmouseover="h2('nope','foo')" onmouseout="ch2('nope','foo')"></td>"#
 480             )
 481         );
 482         assert_eq!(
 483             render_cell(
 484                 "foo",
 485                 &mut Row {
 486                     label: "nope".to_owned(),
 487                     entries: HashMap::from([("bar".to_owned(), vec![None])]),
 488                 }
 489             ),
 490             HTML::from(
 491                 r#"<td class="" onmouseover="h2('nope','foo')" onmouseout="ch2('nope','foo')"></td>"#
 492             )
 493         );
 494         assert_eq!(
 495             render_cell(
 496                 "foo",
 497                 &mut Row {
 498                     label: "nope".to_owned(),
 499                     entries: HashMap::from([("foo".to_owned(), vec![None])]),
 500                 }
 501             ),
 502             HTML::from(
 503                 r#"<td class="yes" onmouseover="h2('nope','foo')" onmouseout="ch2('nope','foo')"></td>"#
 504             )
 505         );
 506         assert_eq!(
 507             render_cell(
 508                 "foo",
 509                 &mut Row {
 510                     label: "nope".to_owned(),
 511                     entries: HashMap::from([("foo".to_owned(), vec![None, None])]),
 512                 }
 513             ),
 514             HTML::from(
 515                 r#"<td class="yes" onmouseover="h2('nope','foo')" onmouseout="ch2('nope','foo')">2</td>"#
 516             )
 517         );
 518         assert_eq!(
 519             render_cell(
 520                 "foo",
 521                 &mut Row {
 522                     label: "nope".to_owned(),
 523                     entries: HashMap::from([(
 524                         "foo".to_owned(),
 525                         vec![Some("5".to_owned()), Some("10".to_owned())]
 526                     )]),
 527                 }
 528             ),
 529             HTML::from(
 530                 r#"<td class="yes" onmouseover="h2('nope','foo')" onmouseout="ch2('nope','foo')">5 10</td>"#
 531             )
 532         );
 533         assert_eq!(
 534             render_cell(
 535                 "foo",
 536                 &mut Row {
 537                     label: "nope".to_owned(),
 538                     entries: HashMap::from([("foo".to_owned(), vec![Some("5".to_owned()), None])]),
 539                 }
 540             ),
 541             HTML::from(
 542                 r#"<td class="yes" onmouseover="h2('nope','foo')" onmouseout="ch2('nope','foo')">5 ✓</td>"#
 543             )
 544         );
 545         assert_eq!(
 546             render_cell(
 547                 "heart",
 548                 &mut Row {
 549                     label: "nope".to_owned(),
 550                     entries: HashMap::from([("heart".to_owned(), vec![Some("<3".to_owned())])]),
 551                 }
 552             ),
 553             HTML::from(
 554                 r#"<td class="yes" onmouseover="h2('nope','heart')" onmouseout="ch2('nope','heart')">&lt;3</td>"#
 555             )
 556         );
 557         assert_eq!(
 558             render_cell(
 559                 "foo",
 560                 &mut Row {
 561                     label: "bob's".to_owned(),
 562                     entries: HashMap::from([("foo".to_owned(), vec![None])]),
 563                 }
 564             ),
 565             HTML::from(
 566                 r#"<td class="yes" onmouseover="h2('bob&#39;s','foo')" onmouseout="ch2('bob&#39;s','foo')"></td>"#
 567             )
 568         );
 569         let mut r = Row {
 570             label: "nope".to_owned(),
 571             entries: HashMap::from([
 572                 ("foo".to_owned(), vec![None]),
 573                 ("baz".to_owned(), vec![None]),
 574             ]),
 575         };
 576         assert_eq!(r.entries.len(), 2);
 577         render_cell("foo", &mut r);
 578         assert_eq!(r.entries.len(), 1);
 579         render_cell("bar", &mut r);
 580         assert_eq!(r.entries.len(), 1);
 581         render_cell("baz", &mut r);
 582         assert_eq!(r.entries.len(), 0);
 583     }
 584
 585     #[test]
 586     fn test_render_leftovers() {
 587         assert_eq!(
 588             render_all_leftovers(&Row {
 589                 label: "nope".to_owned(),
 590                 entries: HashMap::from([("foo".to_owned(), vec![None])]),
 591             }),
 592             HTML::from("foo")
 593         );
 594         assert_eq!(
 595             render_all_leftovers(&Row {
 596                 label: "nope".to_owned(),
 597                 entries: HashMap::from([
 598                     ("foo".to_owned(), vec![None]),
 599                     ("bar".to_owned(), vec![None])
 600                 ]),
 601             }),
 602             HTML::from("bar, foo")
 603         );
 604         assert_eq!(
 605             render_all_leftovers(&Row {
 606                 label: "nope".to_owned(),
 607                 entries: HashMap::from([
 608                     ("foo".to_owned(), vec![None]),
 609                     ("bar".to_owned(), vec![None, None])
 610                 ]),
 611             }),
 612             HTML::from("bar: 2, foo")
 613         );
 614     }
 615
 616     #[test]
 617     fn test_render_row() {
 618         assert_eq!(
 619             render_row(
 620                 &["foo".to_owned()],
 621                 &mut Row {
 622                     label: "nope".to_owned(),
 623                     entries: HashMap::from([("bar".to_owned(), vec![None])]),
 624                 }
 625             ),
 626             HTML::from(
 627                 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>
 628 "#
 629             )
 630         );
 631     }
 632 }