Appearance
| 1 | namespace Syntax.Process.Printer is | |
| 2 | use Collections | |
| 3 | use Ghul | |
| 4 | ||
| 5 | use Trees | |
| 6 | use Source | |
| 7 | ||
| 8 | use Lexical.TRIVIA | |
| 9 | ||
| 10 | // The ghūl code formatter. | |
| 11 | // | |
| 12 | // Extends the plain GHUL printer (which the diagnostics path uses via | |
| 13 | // Node.to_string and must not be disturbed) and redirects its output | |
| 14 | // primitives into a DOC tree, so DOC_RENDERER can lay the result out | |
| 15 | // against a column budget. Comments and blank lines collected by the | |
| 16 | // tokenizer as TRIVIA are interleaved back in by source position. | |
| 17 | // | |
| 18 | // Selected visits are overridden to introduce soft breaks (GROUP/LINE) | |
| 19 | // where wrapping should happen; everything else inherits the printer's | |
| 20 | // structural layout unchanged. | |
| 21 | class FORMATTER: GHUL is | |
| 22 | _builder: DOC_BUILDER | |
| 23 | _width: int | |
| 24 | ||
| 25 | // Types a build settled at inferred sites, written in where the | |
| 26 | // source left them out. Null prints the source as written. | |
| 27 | annotations: InferredAnnotations? public | |
| 28 | ||
| 29 | _annotation(lookup: (InferredAnnotations) -> string?) -> string? => | |
| 30 | if let a = annotations then lookup(a) else null fi | |
| 31 | ||
| 32 | // Written annotations to leave out, so the source is printed with | |
| 33 | // less than its author wrote. Null prints every annotation. | |
| 34 | stripper: ANNOTATION_STRIPPER? public | |
| 35 | ||
| 36 | // Whether the variable being printed is a `let` local: the only | |
| 37 | // kind of Variables.VARIABLE whose type is an annotation rather | |
| 38 | // than a signature. | |
| 39 | _in_let: bool | |
| 40 | ||
| 41 | _strip(kind: string, location: Source.LOCATION, mode: string, from: Source.LOCATION, to: Source.LOCATION) -> bool => | |
| 42 | if let s = stripper then s.strip(kind, location, mode, from, to) else false fi | |
| 43 | ||
| 44 | _trivia: LIST[TRIVIA] | |
| 45 | _trivia_index: int | |
| 46 | ||
| 47 | // Source line of the furthest content emitted so far. Drives | |
| 48 | // trailing-comment detection: a comment starting on a line we have | |
| 49 | // already passed trails that line rather than leading the next node. | |
| 50 | _last_line: int | |
| 51 | ||
| 52 | // True between opening a block and emitting its first real content; | |
| 53 | // suppresses a blank line immediately inside a block. | |
| 54 | _at_block_start: bool | |
| 55 | ||
| 56 | // True when nothing has been written since the last line break. A | |
| 57 | // statement whose rendering ends in a block terminator - `od`, `esac` - | |
| 58 | // leaves the line open, and the statement after it must not be run onto | |
| 59 | // the same line. | |
| 60 | _at_line_start: bool | |
| 61 | ||
| 62 | // True while inside a union that declared a primary-constructor | |
| 63 | // header. Variants emitted under this flag can drop a sole | |
| 64 | // `(..)` field list — the rewriter implicitly splices the primary | |
| 65 | // params for a variant with no field list. Without a primary | |
| 66 | // header, a sole `..` is a hard error, and dropping the parens | |
| 67 | // would silently turn the error into valid empty-fields source. | |
| 68 | _union_has_primary_params: bool | |
| 69 | ||
| 70 | init(trivia: Iterable[TRIVIA]?, width: int) is | |
| 71 | super.init() | |
| 72 | ||
| 73 | _width = width | |
| 74 | _builder = DOC_BUILDER() | |
| 75 | ||
| 76 | _trivia = LIST[TRIVIA]() | |
| 77 | if trivia? then | |
| 78 | for t in trivia do | |
| 79 | _trivia.add(t) | |
| 80 | od | |
| 81 | fi | |
| 82 | si | |
| 83 | ||
| 84 | // Format a parsed file (or any node) and return the rendered text. | |
| 85 | format(node: Node) -> string is | |
| 86 | node.accept(self) | |
| 87 | flush_remaining() | |
| 88 | ||
| 89 | let text = DOC_RENDERER(_width).render(_builder.build()) | |
| 90 | ||
| 91 | // Strip trailing whitespace from every line (blank lines pick up | |
| 92 | // indentation from the renderer), then settle each terminator. | |
| 93 | let lines = Collections.LIST[string]() | |
| 94 | ||
| 95 | for line in text.split(['\n']) do | |
| 96 | lines.add(line.trim_end()) | |
| 97 | od | |
| 98 | ||
| 99 | _settle_terminators(lines) | |
| 100 | ||
| 101 | let result = System.Text.StringBuilder() | |
| 102 | let first mut = true | |
| 103 | for line in lines do | |
| 104 | if !first then | |
| 105 | result.append('\n') | |
| 106 | fi | |
| 107 | result.append(line) | |
| 108 | first = false | |
| 109 | od | |
| 110 | ||
| 111 | return "{result.to_string().trim_end()}\n" | |
| 112 | si | |
| 113 | ||
| 114 | // A statement terminator is written as this marker rather than as | |
| 115 | // `;`, because whether it is needed depends on what follows it, | |
| 116 | // which is not known when the statement is emitted. | |
| 117 | // _settle_terminators resolves every one of them once the text has | |
| 118 | // been laid out. | |
| 119 | // A control character, so it cannot occur in the source being | |
| 120 | // formatted. Held and searched for as a `char`: `index_of(string)` | |
| 121 | // compares by culture, under which a control character has no | |
| 122 | // collation weight and so matches at the start of any string at | |
| 123 | // all, while the `char` overload compares ordinally. | |
| 124 | TERMINATOR_MARKER: char static => cast char(1) | |
| 125 | ||
| 126 | write_terminator() is | |
| 127 | write(TERMINATOR_MARKER) | |
| 128 | write_line() | |
| 129 | si | |
| 130 | ||
| 131 | // A line break stands in for a statement terminator, so a marker | |
| 132 | // that nothing but a comment follows is dropped. The exception is | |
| 133 | // the one terminator that carries meaning: adjacent string literals | |
| 134 | // chain into a single literal across a line break, so a statement | |
| 135 | // ending on one and followed by a statement starting with one keeps | |
| 136 | // its `;`. A marker with code after it separates two statements on | |
| 137 | // one line and is kept. | |
| 138 | _settle_terminators(lines: Collections.LIST[string]) is | |
| 139 | let i mut = 0 | |
| 140 | ||
| 141 | while i < lines.count do | |
| 142 | let line = lines[i] | |
| 143 | let at = line.index_of(TERMINATOR_MARKER) | |
| 144 | ||
| 145 | if at >= 0 then | |
| 146 | let before = line[0..at] | |
| 147 | let after = line[at + 1..<0] | |
| 148 | ||
| 149 | let keep = | |
| 150 | !_is_only_a_comment(after) \/ | |
| 151 | (_ends_with_string(before) /\ | |
| 152 | _next_line_starts_with_string(lines, i)) | |
| 153 | ||
| 154 | lines[i] = if keep then "{before};{after}" else "{before}{after}" fi | |
| 155 | fi | |
| 156 | ||
| 157 | i = i + 1 | |
| 158 | od | |
| 159 | si | |
| 160 | ||
| 161 | // Whether the rest of a line, after a terminator marker, is nothing | |
| 162 | // but whitespace and a comment - in which case the line break is | |
| 163 | // what ends the statement. | |
| 164 | _is_only_a_comment(rest: string) -> bool is | |
| 165 | let trimmed = rest.trim() | |
| 166 | ||
| 167 | if trimmed.length == 0 \/ trimmed.starts_with("//") then | |
| 168 | return true | |
| 169 | fi | |
| 170 | ||
| 171 | return | |
| 172 | trimmed.starts_with("/*") /\ | |
| 173 | trimmed.ends_with("*/") /\ | |
| 174 | trimmed.index_of("*/") == trimmed.length - 2 | |
| 175 | si | |
| 176 | ||
| 177 | _ends_with_string(line: string) -> bool is | |
| 178 | let trimmed = line.trim_end() | |
| 179 | ||
| 180 | return trimmed.length > 0 /\ trimmed[trimmed.length - 1] == cast char(34) | |
| 181 | si | |
| 182 | ||
| 183 | // Comments are transparent to string chaining, so a comment line | |
| 184 | // between the two literals does not break the chain and is looked | |
| 185 | // through here too - a block comment as well as a line comment. | |
| 186 | // | |
| 187 | // A line this cannot classify - the opening line of a block comment | |
| 188 | // that runs on, say - is answered as though a literal followed. | |
| 189 | // The two errors are not equal: a terminator written where none was | |
| 190 | // needed costs a warning, while one dropped where it was needed | |
| 191 | // merges two literals into one and changes what the program says. | |
| 192 | _next_line_starts_with_string(lines: Collections.LIST[string], from: int) -> bool is | |
| 193 | let i mut = from + 1 | |
| 194 | ||
| 195 | while i < lines.count do | |
| 196 | let candidate = lines[i].trim() | |
| 197 | ||
| 198 | if candidate.length == 0 \/ _is_only_a_comment(candidate) then | |
| 199 | i = i + 1 | |
| 200 | elif candidate.starts_with("/*") then | |
| 201 | return true | |
| 202 | else | |
| 203 | return candidate[0] == cast char(34) | |
| 204 | fi | |
| 205 | od | |
| 206 | ||
| 207 | return false | |
| 208 | si | |
| 209 | ||
| 210 | // --- output primitives, redirected into the DOC builder --- | |
| 211 | ||
| 212 | write(value: string?) is | |
| 213 | if !value? then | |
| 214 | return | |
| 215 | fi | |
| 216 | ||
| 217 | if value.length > 0 then | |
| 218 | _at_block_start = false | |
| 219 | _at_line_start = false | |
| 220 | fi | |
| 221 | ||
| 222 | _builder.text(value) | |
| 223 | si | |
| 224 | ||
| 225 | write(c: char) is | |
| 226 | _at_block_start = false | |
| 227 | _at_line_start = false | |
| 228 | _builder.text("{c}") | |
| 229 | si | |
| 230 | ||
| 231 | // A name that collides with a keyword, or that would tokenise as a | |
| 232 | // number, only reached the tree via a backtick escape; the inherited | |
| 233 | // printer drops the backtick and the result no longer re-parses. | |
| 234 | // Restore it. Operator names are left to visit(Expressions.IDENTIFIER): | |
| 235 | // they must stay bare in definition and binary-operation positions, | |
| 236 | // which route through here too. | |
| 237 | write_name(name: string) is | |
| 238 | if Lexical.TOKENIZER.is_reserved_word(name) \/ Lexical.TOKENIZER.is_numeric_identifier(name) then | |
| 239 | write('`') | |
| 240 | fi | |
| 241 | write(name) | |
| 242 | si | |
| 243 | ||
| 244 | // An operator referred to as a plain value, rather than applied, | |
| 245 | // reached the tree via a backtick escape and must keep it to re-parse | |
| 246 | // as an identifier rather than an operator. This is the one identifier | |
| 247 | // position where an operator escapes; definition names and binary | |
| 248 | // operations render the operator bare through the inherited visits. | |
| 249 | visit(identifier: Expressions.IDENTIFIER) is | |
| 250 | location(identifier) | |
| 251 | let inner = identifier.identifier | |
| 252 | if !inner.is_qualified /\ Lexical.TOKENIZER.is_operator_name(inner.name) then | |
| 253 | write('`') | |
| 254 | write(inner.name) | |
| 255 | else | |
| 256 | inner.accept(self) | |
| 257 | fi | |
| 258 | si | |
| 259 | ||
| 260 | at_line_start: bool => _at_line_start | |
| 261 | ||
| 262 | write_line() is | |
| 263 | flush_trailing_comments() | |
| 264 | _builder.hard_line() | |
| 265 | _at_line_start = true | |
| 266 | si | |
| 267 | ||
| 268 | indent() is | |
| 269 | _builder.begin_nest(4) | |
| 270 | _at_block_start = true | |
| 271 | si | |
| 272 | ||
| 273 | outdent() is | |
| 274 | _builder.end_nest() | |
| 275 | si | |
| 276 | ||
| 277 | write_indent() is si | |
| 278 | ||
| 279 | // --- line tracking --- | |
| 280 | ||
| 281 | location(location: LOCATION) is | |
| 282 | note(location) | |
| 283 | si | |
| 284 | ||
| 285 | note(location: LOCATION?) is | |
| 286 | if location? /\ location.start_line > _last_line then | |
| 287 | _last_line = location.start_line | |
| 288 | fi | |
| 289 | si | |
| 290 | ||
| 291 | // --- trivia interleaving --- | |
| 292 | ||
| 293 | _has_trivia: bool => _trivia_index < _trivia.count | |
| 294 | ||
| 295 | // Whether any unemitted trivia still sits before `limit` (a | |
| 296 | // LOCATION.start value). A body with none can take a one-line | |
| 297 | // rendering, since there is no comment to keep inside it. | |
| 298 | _has_trivia_before(limit: int) -> bool => | |
| 299 | _has_trivia /\ _peek.location.start < limit | |
| 300 | ||
| 301 | _peek: TRIVIA => _trivia[_trivia_index] | |
| 302 | ||
| 303 | _consume() is | |
| 304 | _trivia_index = _trivia_index + 1 | |
| 305 | si | |
| 306 | ||
| 307 | // Emit any trivia positioned strictly before `limit` (a LOCATION.start | |
| 308 | // value) that is not a trailing comment of an already-emitted line. | |
| 309 | flush_leading(limit: int) is | |
| 310 | while _has_trivia /\ _peek.location.start < limit do | |
| 311 | let t = _peek | |
| 312 | ||
| 313 | if isa TRIVIA.BLANK_LINE(t) then | |
| 314 | // A blank line written inside an expression the layout | |
| 315 | // joins onto one line has nowhere to go, and emitting it | |
| 316 | // here puts it after the whole construct instead - one | |
| 317 | // more blank line on every reformat. Its source line has | |
| 318 | // already been passed, which is what distinguishes it | |
| 319 | // from a blank line genuinely separating two constructs. | |
| 320 | if !_at_block_start /\ _peek.location.start_line > _last_line then | |
| 321 | _builder.hard_line() | |
| 322 | fi | |
| 323 | _consume() | |
| 324 | else | |
| 325 | emit_comment(t) | |
| 326 | _builder.hard_line() | |
| 327 | _at_block_start = false | |
| 328 | _consume() | |
| 329 | fi | |
| 330 | od | |
| 331 | si | |
| 332 | ||
| 333 | // Emit comments that start on a line we have already emitted: they | |
| 334 | // trail that line. Called from write_line, before the newline. | |
| 335 | flush_trailing_comments() is | |
| 336 | while | |
| 337 | _has_trivia /\ | |
| 338 | _is_comment(_peek) /\ | |
| 339 | _peek.location.start_line <= _last_line | |
| 340 | do | |
| 341 | _builder.text(" ") | |
| 342 | emit_comment(_peek) | |
| 343 | _consume() | |
| 344 | od | |
| 345 | si | |
| 346 | ||
| 347 | // Emit everything left over (file-trailing comments). | |
| 348 | flush_remaining() is | |
| 349 | while _has_trivia do | |
| 350 | let t = _peek | |
| 351 | if _is_comment(t) then | |
| 352 | emit_comment(t) | |
| 353 | _builder.hard_line() | |
| 354 | fi | |
| 355 | _consume() | |
| 356 | od | |
| 357 | si | |
| 358 | ||
| 359 | _is_comment(t: TRIVIA) -> bool => | |
| 360 | isa TRIVIA.LINE_COMMENT(t) \/ isa TRIVIA.BLOCK_COMMENT(t) | |
| 361 | ||
| 362 | // The leading whitespace every non-blank continuation line of a | |
| 363 | // block comment has in common. | |
| 364 | _common_indent(lines: string[]) -> int is | |
| 365 | let common mut = -1 | |
| 366 | let first mut = true | |
| 367 | ||
| 368 | for line in lines do | |
| 369 | if !first /\ line.trim().length > 0 then | |
| 370 | let indent = line.length - line.trim_start().length | |
| 371 | ||
| 372 | if common < 0 \/ indent < common then | |
| 373 | common = indent | |
| 374 | fi | |
| 375 | fi | |
| 376 | ||
| 377 | first = false | |
| 378 | od | |
| 379 | ||
| 380 | return if common < 0 then 0 else common fi | |
| 381 | si | |
| 382 | ||
| 383 | _strip_indent(line: string, amount: int) -> string is | |
| 384 | let i mut = 0 | |
| 385 | ||
| 386 | while i < amount /\ i < line.length /\ char.is_white_space(line[i]) do | |
| 387 | i = i + 1 | |
| 388 | od | |
| 389 | ||
| 390 | return line[i..<0] | |
| 391 | si | |
| 392 | ||
| 393 | emit_comment(t: TRIVIA) is | |
| 394 | note(t.location) | |
| 395 | ||
| 396 | if let bc: TRIVIA.BLOCK_COMMENT = t then | |
| 397 | let lines = bc.text.split(['\n']) | |
| 398 | ||
| 399 | // A continuation line carries the indentation it was | |
| 400 | // written at, and the renderer indents it again, so | |
| 401 | // re-emitting it verbatim moves the comment four columns | |
| 402 | // right on every reformat. Strip what every continuation | |
| 403 | // line shares and let the renderer supply it instead; the | |
| 404 | // relative shape of the comment is preserved. | |
| 405 | let common = _common_indent(lines) | |
| 406 | ||
| 407 | let first mut = true | |
| 408 | for line in lines do | |
| 409 | if !first then | |
| 410 | _builder.hard_line() | |
| 411 | _builder.text(_strip_indent(line, common)) | |
| 412 | else | |
| 413 | _builder.text(line) | |
| 414 | fi | |
| 415 | first = false | |
| 416 | od | |
| 417 | elif let lc: TRIVIA.LINE_COMMENT = t then | |
| 418 | _builder.text(lc.text) | |
| 419 | elif let sb: TRIVIA.SHEBANG = t then | |
| 420 | _builder.text(sb.text) | |
| 421 | fi | |
| 422 | si | |
| 423 | si | |
| 424 | si |