1 | #!/usr/bin/env python2
|
2 | """
|
3 | Render Oils value_t -> doc_t, so it can be pretty printed
|
4 | """
|
5 |
|
6 | from __future__ import print_function
|
7 |
|
8 | import math
|
9 |
|
10 | from _devbuild.gen.pretty_asdl import (doc, Measure, MeasuredDoc)
|
11 | from _devbuild.gen.value_asdl import Obj, value, value_e, value_t, value_str
|
12 | from data_lang import j8
|
13 | from data_lang import j8_lite
|
14 | from display.pretty import (_Break, _Concat, _Flat, _Group, _IfFlat, _Indent,
|
15 | _EmptyMeasure)
|
16 | from display import ansi
|
17 | from frontend import match
|
18 | from mycpp import mops
|
19 | from mycpp.mylib import log, tagswitch, iteritems
|
20 | from typing import cast, List, Dict
|
21 |
|
22 | import libc
|
23 |
|
24 | _ = log
|
25 |
|
26 |
|
27 | def ValType(val):
|
28 | # type: (value_t) -> str
|
29 | """Returns a user-facing string like Int, Eggex, BashArray, etc."""
|
30 | return value_str(val.tag(), dot=False)
|
31 |
|
32 |
|
33 | def FloatString(fl):
|
34 | # type: (float) -> str
|
35 |
|
36 | # Print in YSH syntax, similar to data_lang/j8.py
|
37 | if math.isinf(fl):
|
38 | s = 'INFINITY'
|
39 | if fl < 0:
|
40 | s = '-' + s
|
41 | elif math.isnan(fl):
|
42 | s = 'NAN'
|
43 | else:
|
44 | s = str(fl)
|
45 | return s
|
46 |
|
47 |
|
48 | #
|
49 | # Unicode Helpers
|
50 | #
|
51 |
|
52 |
|
53 | def TryUnicodeWidth(s):
|
54 | # type: (str) -> int
|
55 | try:
|
56 | width = libc.wcswidth(s)
|
57 | except UnicodeError:
|
58 | # e.g. en_US.UTF-8 locale missing, just return the number of bytes
|
59 | width = len(s)
|
60 |
|
61 | if width == -1: # non-printable wide char
|
62 | return len(s)
|
63 |
|
64 | return width
|
65 |
|
66 |
|
67 | def UText(string):
|
68 | # type: (str) -> MeasuredDoc
|
69 | """Print `string` (which must not contain a newline)."""
|
70 | return MeasuredDoc(doc.Text(string), Measure(TryUnicodeWidth(string), -1))
|
71 |
|
72 |
|
73 | class ValueEncoder:
|
74 | """Converts Oils values into `doc`s, which can then be pretty printed."""
|
75 |
|
76 | def __init__(self):
|
77 | # type: () -> None
|
78 |
|
79 | # Default values
|
80 | self.indent = 4
|
81 | self.use_styles = True
|
82 | # Tuned for 'data_lang/pretty-benchmark.sh float-demo'
|
83 | # TODO: might want options for float width
|
84 | self.max_tabular_width = 22
|
85 |
|
86 | self.ysh_style = True
|
87 |
|
88 | self.visiting = {} # type: Dict[int, bool]
|
89 |
|
90 | # These can be configurable later
|
91 | self.int_style = ansi.YELLOW
|
92 | self.float_style = ansi.BLUE
|
93 | self.null_style = ansi.RED
|
94 | self.bool_style = ansi.CYAN
|
95 | self.string_style = ansi.GREEN
|
96 | self.cycle_style = ansi.BOLD + ansi.BLUE
|
97 | self.type_style = ansi.MAGENTA
|
98 |
|
99 | def SetIndent(self, indent):
|
100 | # type: (int) -> None
|
101 | """Set the number of spaces per indent."""
|
102 | self.indent = indent
|
103 |
|
104 | def SetUseStyles(self, use_styles):
|
105 | # type: (bool) -> None
|
106 | """Print with ansi colors and styles, rather than plain text."""
|
107 | self.use_styles = use_styles
|
108 |
|
109 | def SetMaxTabularWidth(self, max_tabular_width):
|
110 | # type: (int) -> None
|
111 | """Set the maximum width that list elements can be, for them to be
|
112 | vertically aligned."""
|
113 | self.max_tabular_width = max_tabular_width
|
114 |
|
115 | def TypePrefix(self, type_str):
|
116 | # type: (str) -> List[MeasuredDoc]
|
117 | """Return docs for type string "(List)", which may break afterward."""
|
118 | type_name = self._Styled(self.type_style, UText(type_str))
|
119 |
|
120 | n = len(type_str)
|
121 | # Our maximum string is "Float"
|
122 | assert n <= 5, type_str
|
123 |
|
124 | # Start printing in column 8. Adjust to 6 because () takes 2 spaces.
|
125 | spaces = ' ' * (6 - n)
|
126 |
|
127 | mdocs = [UText("("), type_name, UText(")"), _Break(spaces)]
|
128 | return mdocs
|
129 |
|
130 | def Value(self, val):
|
131 | # type: (value_t) -> MeasuredDoc
|
132 | """Convert an Oils value into a `doc`, which can then be pretty printed."""
|
133 | self.visiting.clear()
|
134 | return self._Value(val)
|
135 |
|
136 | def _Styled(self, style, mdoc):
|
137 | # type: (str, MeasuredDoc) -> MeasuredDoc
|
138 | """Apply the ANSI style string to the given node, if use_styles is set."""
|
139 | if self.use_styles:
|
140 | return _Concat([
|
141 | MeasuredDoc(doc.Text(style), _EmptyMeasure()), mdoc,
|
142 | MeasuredDoc(doc.Text(ansi.RESET), _EmptyMeasure())
|
143 | ])
|
144 | else:
|
145 | return mdoc
|
146 |
|
147 | def _Surrounded(self, open, mdoc, close):
|
148 | # type: (str, MeasuredDoc, str) -> MeasuredDoc
|
149 | """Print one of two options (using '[', ']' for open, close):
|
150 |
|
151 | ```
|
152 | [mdoc]
|
153 | ------
|
154 | [
|
155 | mdoc
|
156 | ]
|
157 | ```
|
158 | """
|
159 | return _Group(
|
160 | _Concat([
|
161 | UText(open),
|
162 | _Indent(self.indent, _Concat([_Break(""), mdoc])),
|
163 | _Break(""),
|
164 | UText(close)
|
165 | ]))
|
166 |
|
167 | def _SurroundedAndPrefixed(self, open, prefix, sep, mdoc, close):
|
168 | # type: (str, MeasuredDoc, str, MeasuredDoc, str) -> MeasuredDoc
|
169 | """Print one of two options
|
170 | (using '[', 'prefix', ':', 'mdoc', ']' for open, prefix, sep, mdoc, close):
|
171 |
|
172 | ```
|
173 | [prefix:mdoc]
|
174 | ------
|
175 | [prefix
|
176 | mdoc
|
177 | ]
|
178 | ```
|
179 | """
|
180 | return _Group(
|
181 | _Concat([
|
182 | UText(open), prefix,
|
183 | _Indent(self.indent, _Concat([_Break(sep), mdoc])),
|
184 | _Break(""),
|
185 | UText(close)
|
186 | ]))
|
187 |
|
188 | def _Join(self, items, sep, space):
|
189 | # type: (List[MeasuredDoc], str, str) -> MeasuredDoc
|
190 | """Join `items`, using either 'sep+space' or 'sep+newline' between them.
|
191 |
|
192 | E.g., if sep and space are ',' and '_', print one of these two cases:
|
193 | ```
|
194 | first,_second,_third
|
195 | ------
|
196 | first,
|
197 | second,
|
198 | third
|
199 | ```
|
200 | """
|
201 | seq = [] # type: List[MeasuredDoc]
|
202 | for i, item in enumerate(items):
|
203 | if i != 0:
|
204 | seq.append(UText(sep))
|
205 | seq.append(_Break(space))
|
206 | seq.append(item)
|
207 | return _Concat(seq)
|
208 |
|
209 | def _Tabular(self, items, sep):
|
210 | # type: (List[MeasuredDoc], str) -> MeasuredDoc
|
211 | """Join `items` together, using one of three styles:
|
212 |
|
213 | (showing spaces as underscores for clarity)
|
214 | ```
|
215 | first,_second,_third,_fourth,_fifth,_sixth,_seventh,_eighth
|
216 | ------
|
217 | first,___second,__third,
|
218 | fourth,__fifth,___sixth,
|
219 | seventh,_eighth
|
220 | ------
|
221 | first,
|
222 | second,
|
223 | third,
|
224 | fourth,
|
225 | fifth,
|
226 | sixth,
|
227 | seventh,
|
228 | eighth
|
229 | ```
|
230 |
|
231 | The first "single line" style is used if the items fit on one line. The
|
232 | second "tabular' style is used if the flat width of all items is no
|
233 | greater than `self.max_tabular_width`. The third "multi line" style is
|
234 | used otherwise.
|
235 | """
|
236 |
|
237 | # Why not "just" use tabular alignment so long as two items fit on every
|
238 | # line? Because it isn't possible to check for that in the pretty
|
239 | # printing language. There are two sorts of conditionals we can do:
|
240 | #
|
241 | # A. Inside the pretty printing language, which supports exactly one
|
242 | # conditional: "does it fit on one line?".
|
243 | # B. Outside the pretty printing language we can run arbitrary Python
|
244 | # code, but we don't know how much space is available on the line
|
245 | # because it depends on the context in which we're printed, which may
|
246 | # vary.
|
247 | #
|
248 | # We're picking between the three styles, by using (A) to check if the
|
249 | # first style fits on one line, then using (B) with "are all the items
|
250 | # smaller than `self.max_tabular_width`?" to pick between style 2 and
|
251 | # style 3.
|
252 |
|
253 | if len(items) == 0:
|
254 | return UText("")
|
255 |
|
256 | max_flat_len = 0
|
257 | seq = [] # type: List[MeasuredDoc]
|
258 | for i, item in enumerate(items):
|
259 | if i != 0:
|
260 | seq.append(UText(sep))
|
261 | seq.append(_Break(" "))
|
262 | seq.append(item)
|
263 | max_flat_len = max(max_flat_len, item.measure.flat)
|
264 | non_tabular = _Concat(seq)
|
265 |
|
266 | sep_width = TryUnicodeWidth(sep)
|
267 | if max_flat_len + sep_width + 1 <= self.max_tabular_width:
|
268 | tabular_seq = [] # type: List[MeasuredDoc]
|
269 | for i, item in enumerate(items):
|
270 | tabular_seq.append(_Flat(item))
|
271 | if i != len(items) - 1:
|
272 | padding = max_flat_len - item.measure.flat + 1
|
273 | tabular_seq.append(UText(sep))
|
274 | tabular_seq.append(_Group(_Break(" " * padding)))
|
275 | tabular = _Concat(tabular_seq)
|
276 | return _Group(_IfFlat(non_tabular, tabular))
|
277 | else:
|
278 | return non_tabular
|
279 |
|
280 | def _DictKey(self, s):
|
281 | # type: (str) -> MeasuredDoc
|
282 | if match.IsValidVarName(s):
|
283 | encoded = s
|
284 | else:
|
285 | if self.ysh_style:
|
286 | encoded = j8_lite.YshEncodeString(s)
|
287 | else:
|
288 | # TODO: remove this dead branch after fixing tests
|
289 | encoded = j8_lite.EncodeString(s)
|
290 | return UText(encoded)
|
291 |
|
292 | def _StringLiteral(self, s):
|
293 | # type: (str) -> MeasuredDoc
|
294 | if self.ysh_style:
|
295 | # YSH r'' or b'' style
|
296 | encoded = j8_lite.YshEncodeString(s)
|
297 | else:
|
298 | # TODO: remove this dead branch after fixing tests
|
299 | encoded = j8_lite.EncodeString(s)
|
300 | return self._Styled(self.string_style, UText(encoded))
|
301 |
|
302 | def _BashStringLiteral(self, s):
|
303 | # type: (str) -> MeasuredDoc
|
304 |
|
305 | # '' or $'' style
|
306 | #
|
307 | # We mimic bash syntax by using $'\\' instead of b'\\'
|
308 | #
|
309 | # $ declare -a array=($'\\')
|
310 | # $ = array
|
311 | # (BashArray) (BashArray $'\\')
|
312 | #
|
313 | # $ declare -A assoc=([k]=$'\\')
|
314 | # $ = assoc
|
315 | # (BashAssoc) (BashAssoc ['k']=$'\\')
|
316 |
|
317 | encoded = j8_lite.ShellEncode(s)
|
318 | return self._Styled(self.string_style, UText(encoded))
|
319 |
|
320 | def _YshList(self, vlist):
|
321 | # type: (value.List) -> MeasuredDoc
|
322 | """Print a string literal."""
|
323 | if len(vlist.items) == 0:
|
324 | return UText("[]")
|
325 | mdocs = [self._Value(item) for item in vlist.items]
|
326 | return self._Surrounded("[", self._Tabular(mdocs, ","), "]")
|
327 |
|
328 | def _RawDict(self, d, open, close):
|
329 | # type: (Dict[str, value_t], str, str) -> MeasuredDoc
|
330 | mdocs = [] # type: List[MeasuredDoc]
|
331 | for k, v in iteritems(d):
|
332 | mdocs.append(
|
333 | _Concat([self._DictKey(k),
|
334 | UText(": "),
|
335 | self._Value(v)]))
|
336 | return self._Surrounded(open, self._Join(mdocs, ",", " "), close)
|
337 |
|
338 | def _YshDict(self, vdict):
|
339 | # type: (value.Dict) -> MeasuredDoc
|
340 | if len(vdict.d) == 0:
|
341 | return UText("{}")
|
342 | return self._RawDict(vdict.d, "{", "}")
|
343 |
|
344 | def _BashArray(self, varray):
|
345 | # type: (value.BashArray) -> MeasuredDoc
|
346 | type_name = self._Styled(self.type_style, UText("BashArray"))
|
347 | if len(varray.strs) == 0:
|
348 | return _Concat([UText("("), type_name, UText(")")])
|
349 | mdocs = [] # type: List[MeasuredDoc]
|
350 | for s in varray.strs:
|
351 | if s is None:
|
352 | mdocs.append(UText("null"))
|
353 | else:
|
354 | mdocs.append(self._BashStringLiteral(s))
|
355 | return self._SurroundedAndPrefixed("(", type_name, " ",
|
356 | self._Tabular(mdocs, ""), ")")
|
357 |
|
358 | def _BashAssoc(self, vassoc):
|
359 | # type: (value.BashAssoc) -> MeasuredDoc
|
360 | type_name = self._Styled(self.type_style, UText("BashAssoc"))
|
361 | if len(vassoc.d) == 0:
|
362 | return _Concat([UText("("), type_name, UText(")")])
|
363 | mdocs = [] # type: List[MeasuredDoc]
|
364 | for k2, v2 in iteritems(vassoc.d):
|
365 | mdocs.append(
|
366 | _Concat([
|
367 | UText("["),
|
368 | self._BashStringLiteral(k2),
|
369 | UText("]="),
|
370 | self._BashStringLiteral(v2)
|
371 | ]))
|
372 | return self._SurroundedAndPrefixed("(", type_name, " ",
|
373 | self._Join(mdocs, "", " "), ")")
|
374 |
|
375 | def _SparseArray(self, val):
|
376 | # type: (value.SparseArray) -> MeasuredDoc
|
377 | type_name = self._Styled(self.type_style, UText("SparseArray"))
|
378 | if len(val.d) == 0:
|
379 | return _Concat([UText("("), type_name, UText(")")])
|
380 | mdocs = [] # type: List[MeasuredDoc]
|
381 | for k2, v2 in iteritems(val.d):
|
382 | mdocs.append(
|
383 | _Concat([
|
384 | UText("["),
|
385 | self._Styled(self.int_style, UText(mops.ToStr(k2))),
|
386 | UText("]="),
|
387 | self._BashStringLiteral(v2)
|
388 | ]))
|
389 | return self._SurroundedAndPrefixed("(", type_name, " ",
|
390 | self._Join(mdocs, "", " "), ")")
|
391 |
|
392 | def _Obj(self, obj):
|
393 | # type: (Obj) -> MeasuredDoc
|
394 | chain = [] # type: List[MeasuredDoc]
|
395 | cur = obj
|
396 | while cur is not None:
|
397 | chain.append(self._RawDict(cur.d, "(", ")"))
|
398 | cur = cur.prototype
|
399 | if cur is not None:
|
400 | chain.append(UText(" --> "))
|
401 |
|
402 | return _Concat(chain)
|
403 |
|
404 | def _Value(self, val):
|
405 | # type: (value_t) -> MeasuredDoc
|
406 |
|
407 | with tagswitch(val) as case:
|
408 | if case(value_e.Null):
|
409 | return self._Styled(self.null_style, UText("null"))
|
410 |
|
411 | elif case(value_e.Bool):
|
412 | b = cast(value.Bool, val).b
|
413 | return self._Styled(self.bool_style,
|
414 | UText("true" if b else "false"))
|
415 |
|
416 | elif case(value_e.Int):
|
417 | i = cast(value.Int, val).i
|
418 | return self._Styled(self.int_style, UText(mops.ToStr(i)))
|
419 |
|
420 | elif case(value_e.Float):
|
421 | f = cast(value.Float, val).f
|
422 | return self._Styled(self.float_style, UText(FloatString(f)))
|
423 |
|
424 | elif case(value_e.Str):
|
425 | s = cast(value.Str, val).s
|
426 | return self._StringLiteral(s)
|
427 |
|
428 | elif case(value_e.Range):
|
429 | r = cast(value.Range, val)
|
430 | type_name = self._Styled(self.type_style, UText(ValType(r)))
|
431 | mdocs = [UText(str(r.lower)), UText("..<"), UText(str(r.upper))]
|
432 | return self._SurroundedAndPrefixed("(", type_name, " ",
|
433 | self._Join(mdocs, "", " "),
|
434 | ")")
|
435 |
|
436 | elif case(value_e.List):
|
437 | vlist = cast(value.List, val)
|
438 | heap_id = j8.HeapValueId(vlist)
|
439 | if self.visiting.get(heap_id, False):
|
440 | return _Concat([
|
441 | UText("["),
|
442 | self._Styled(self.cycle_style, UText("...")),
|
443 | UText("]")
|
444 | ])
|
445 | else:
|
446 | self.visiting[heap_id] = True
|
447 | result = self._YshList(vlist)
|
448 | self.visiting[heap_id] = False
|
449 | return result
|
450 |
|
451 | elif case(value_e.Dict):
|
452 | vdict = cast(value.Dict, val)
|
453 | heap_id = j8.HeapValueId(vdict)
|
454 | if self.visiting.get(heap_id, False):
|
455 | return _Concat([
|
456 | UText("{"),
|
457 | self._Styled(self.cycle_style, UText("...")),
|
458 | UText("}")
|
459 | ])
|
460 | else:
|
461 | self.visiting[heap_id] = True
|
462 | result = self._YshDict(vdict)
|
463 | self.visiting[heap_id] = False
|
464 | return result
|
465 |
|
466 | elif case(value_e.SparseArray):
|
467 | sparse = cast(value.SparseArray, val)
|
468 | return self._SparseArray(sparse)
|
469 |
|
470 | elif case(value_e.BashArray):
|
471 | varray = cast(value.BashArray, val)
|
472 | return self._BashArray(varray)
|
473 |
|
474 | elif case(value_e.BashAssoc):
|
475 | vassoc = cast(value.BashAssoc, val)
|
476 | return self._BashAssoc(vassoc)
|
477 |
|
478 | elif case(value_e.Obj):
|
479 | vaobj = cast(Obj, val)
|
480 | heap_id = j8.HeapValueId(vaobj)
|
481 | if self.visiting.get(heap_id, False):
|
482 | return _Concat([
|
483 | UText("("),
|
484 | self._Styled(self.cycle_style, UText("...")),
|
485 | UText(")")
|
486 | ])
|
487 | else:
|
488 | self.visiting[heap_id] = True
|
489 | result = self._Obj(vaobj)
|
490 | self.visiting[heap_id] = False
|
491 | return result
|
492 |
|
493 | else:
|
494 | type_name = self._Styled(self.type_style, UText(ValType(val)))
|
495 | id_str = j8.ValueIdString(val)
|
496 | return _Concat([UText("<"), type_name, UText(id_str + ">")])
|
497 |
|
498 |
|
499 | # vim: sw=4
|