author | wenzelm |
Sat, 24 Jul 2021 15:38:41 +0200 | |
changeset 74056 | fb8d5c0133c9 |
parent 73359 | d8a0e996614b |
child 75393 | 87ebf5a50283 |
permissions | -rw-r--r-- |
36683 | 1 |
/* Title: Pure/General/pretty.scala |
2 |
Author: Makarius |
|
3 |
||
36687 | 4 |
Generic pretty printing module. |
36683 | 5 |
*/ |
6 |
||
7 |
package isabelle |
|
8 |
||
71601 | 9 |
import scala.annotation.tailrec |
10 |
||
36683 | 11 |
|
12 |
object Pretty |
|
13 |
{ |
|
61871 | 14 |
/* XML constructors */ |
15 |
||
61874 | 16 |
val space: XML.Body = List(XML.Text(Symbol.space)) |
17 |
def spaces(n: Int): XML.Body = |
|
18 |
if (n == 0) Nil |
|
19 |
else if (n == 1) space |
|
20 |
else List(XML.Text(Symbol.spaces(n))) |
|
48704
85a3de10567d
tuned signature -- make Pretty less dependent on Symbol;
wenzelm
parents:
45666
diff
changeset
|
21 |
|
61871 | 22 |
def block(consistent: Boolean, indent: Int, body: XML.Body): XML.Tree = |
23 |
XML.Elem(Markup.Block(consistent, indent), body) |
|
24 |
def block(indent: Int, body: XML.Body): XML.Tree = block(false, indent, body) |
|
25 |
def block(body: XML.Body): XML.Tree = block(2, body) |
|
48704
85a3de10567d
tuned signature -- make Pretty less dependent on Symbol;
wenzelm
parents:
45666
diff
changeset
|
26 |
|
61871 | 27 |
def brk(width: Int, indent: Int = 0): XML.Tree = |
28 |
XML.Elem(Markup.Break(width, indent), spaces(width)) |
|
29 |
||
69867 | 30 |
val fbrk: XML.Tree = XML.newline |
65130 | 31 |
def fbreaks(ts: List[XML.Tree]): XML.Body = Library.separate(fbrk, ts) |
61871 | 32 |
|
33 |
val Separator: XML.Body = List(XML.elem(Markup.SEPARATOR, space), fbrk) |
|
34 |
def separate(ts: List[XML.Tree]): XML.Body = Library.separate(Separator, ts.map(List(_))).flatten |
|
61864 | 35 |
|
48704
85a3de10567d
tuned signature -- make Pretty less dependent on Symbol;
wenzelm
parents:
45666
diff
changeset
|
36 |
|
51492
eaa1c4cc1106
more explicit Pretty.Metric, with clear distinction of unit (space width) vs. average char width (for visual adjustments) -- NB: Pretty formatting works via full space characters (despite a981a5c8a505 and 70f7483df9cb);
wenzelm
parents:
51470
diff
changeset
|
37 |
/* text metric -- standardized to width of space */ |
eaa1c4cc1106
more explicit Pretty.Metric, with clear distinction of unit (space width) vs. average char width (for visual adjustments) -- NB: Pretty formatting works via full space characters (despite a981a5c8a505 and 70f7483df9cb);
wenzelm
parents:
51470
diff
changeset
|
38 |
|
eaa1c4cc1106
more explicit Pretty.Metric, with clear distinction of unit (space width) vs. average char width (for visual adjustments) -- NB: Pretty formatting works via full space characters (despite a981a5c8a505 and 70f7483df9cb);
wenzelm
parents:
51470
diff
changeset
|
39 |
abstract class Metric |
eaa1c4cc1106
more explicit Pretty.Metric, with clear distinction of unit (space width) vs. average char width (for visual adjustments) -- NB: Pretty formatting works via full space characters (despite a981a5c8a505 and 70f7483df9cb);
wenzelm
parents:
51470
diff
changeset
|
40 |
{ |
eaa1c4cc1106
more explicit Pretty.Metric, with clear distinction of unit (space width) vs. average char width (for visual adjustments) -- NB: Pretty formatting works via full space characters (despite a981a5c8a505 and 70f7483df9cb);
wenzelm
parents:
51470
diff
changeset
|
41 |
val unit: Double |
eaa1c4cc1106
more explicit Pretty.Metric, with clear distinction of unit (space width) vs. average char width (for visual adjustments) -- NB: Pretty formatting works via full space characters (despite a981a5c8a505 and 70f7483df9cb);
wenzelm
parents:
51470
diff
changeset
|
42 |
def apply(s: String): Double |
eaa1c4cc1106
more explicit Pretty.Metric, with clear distinction of unit (space width) vs. average char width (for visual adjustments) -- NB: Pretty formatting works via full space characters (despite a981a5c8a505 and 70f7483df9cb);
wenzelm
parents:
51470
diff
changeset
|
43 |
} |
eaa1c4cc1106
more explicit Pretty.Metric, with clear distinction of unit (space width) vs. average char width (for visual adjustments) -- NB: Pretty formatting works via full space characters (despite a981a5c8a505 and 70f7483df9cb);
wenzelm
parents:
51470
diff
changeset
|
44 |
|
67896 | 45 |
object Default_Metric extends Metric |
51492
eaa1c4cc1106
more explicit Pretty.Metric, with clear distinction of unit (space width) vs. average char width (for visual adjustments) -- NB: Pretty formatting works via full space characters (despite a981a5c8a505 and 70f7483df9cb);
wenzelm
parents:
51470
diff
changeset
|
46 |
{ |
eaa1c4cc1106
more explicit Pretty.Metric, with clear distinction of unit (space width) vs. average char width (for visual adjustments) -- NB: Pretty formatting works via full space characters (despite a981a5c8a505 and 70f7483df9cb);
wenzelm
parents:
51470
diff
changeset
|
47 |
val unit = 1.0 |
eaa1c4cc1106
more explicit Pretty.Metric, with clear distinction of unit (space width) vs. average char width (for visual adjustments) -- NB: Pretty formatting works via full space characters (despite a981a5c8a505 and 70f7483df9cb);
wenzelm
parents:
51470
diff
changeset
|
48 |
def apply(s: String): Double = s.length.toDouble |
eaa1c4cc1106
more explicit Pretty.Metric, with clear distinction of unit (space width) vs. average char width (for visual adjustments) -- NB: Pretty formatting works via full space characters (despite a981a5c8a505 and 70f7483df9cb);
wenzelm
parents:
51470
diff
changeset
|
49 |
} |
eaa1c4cc1106
more explicit Pretty.Metric, with clear distinction of unit (space width) vs. average char width (for visual adjustments) -- NB: Pretty formatting works via full space characters (despite a981a5c8a505 and 70f7483df9cb);
wenzelm
parents:
51470
diff
changeset
|
50 |
|
eaa1c4cc1106
more explicit Pretty.Metric, with clear distinction of unit (space width) vs. average char width (for visual adjustments) -- NB: Pretty formatting works via full space characters (despite a981a5c8a505 and 70f7483df9cb);
wenzelm
parents:
51470
diff
changeset
|
51 |
|
36687 | 52 |
/* markup trees with physical blocks and breaks */ |
53 |
||
62820 | 54 |
private def force_nat(i: Int): Int = i max 0 |
55 |
||
61874 | 56 |
private sealed abstract class Tree { def length: Double } |
57 |
private case class Block( |
|
61871 | 58 |
markup: Option[(Markup, Option[XML.Body])], |
59 |
consistent: Boolean, indent: Int, body: List[Tree], length: Double) extends Tree |
|
61874 | 60 |
private case class Break(force: Boolean, width: Int, indent: Int) extends Tree |
61871 | 61 |
{ def length: Double = width.toDouble } |
61874 | 62 |
private case class Str(string: String, length: Double) extends Tree |
36683 | 63 |
|
61874 | 64 |
private val FBreak = Break(true, 1, 0) |
36683 | 65 |
|
61874 | 66 |
private def make_block( |
61871 | 67 |
markup: Option[(Markup, Option[XML.Body])], |
68 |
consistent: Boolean, |
|
69 |
indent: Int, |
|
70 |
body: List[Tree]): Tree = |
|
61883
c0f34fe6aa61
clarified length of block with pre-existant forced breaks;
wenzelm
parents:
61875
diff
changeset
|
71 |
{ |
62785
70b9c7d4ed7f
more robust pretty printing: permissive treatment of bad values;
wenzelm
parents:
61883
diff
changeset
|
72 |
val indent1 = force_nat(indent) |
70b9c7d4ed7f
more robust pretty printing: permissive treatment of bad values;
wenzelm
parents:
61883
diff
changeset
|
73 |
|
71601 | 74 |
@tailrec def body_length(prts: List[Tree], len: Double): Double = |
61883
c0f34fe6aa61
clarified length of block with pre-existant forced breaks;
wenzelm
parents:
61875
diff
changeset
|
75 |
{ |
c0f34fe6aa61
clarified length of block with pre-existant forced breaks;
wenzelm
parents:
61875
diff
changeset
|
76 |
val (line, rest) = |
c0f34fe6aa61
clarified length of block with pre-existant forced breaks;
wenzelm
parents:
61875
diff
changeset
|
77 |
Library.take_prefix[Tree]({ case Break(true, _, _) => false case _ => true }, prts) |
73359 | 78 |
val len1 = (line.foldLeft(0.0) { case (l, t) => l + t.length }) max len |
71781 | 79 |
(rest: @unchecked) match { |
61883
c0f34fe6aa61
clarified length of block with pre-existant forced breaks;
wenzelm
parents:
61875
diff
changeset
|
80 |
case Break(true, _, ind) :: rest1 => |
62785
70b9c7d4ed7f
more robust pretty printing: permissive treatment of bad values;
wenzelm
parents:
61883
diff
changeset
|
81 |
body_length(Break(false, indent1 + ind, 0) :: rest1, len1) |
61883
c0f34fe6aa61
clarified length of block with pre-existant forced breaks;
wenzelm
parents:
61875
diff
changeset
|
82 |
case Nil => len1 |
c0f34fe6aa61
clarified length of block with pre-existant forced breaks;
wenzelm
parents:
61875
diff
changeset
|
83 |
} |
c0f34fe6aa61
clarified length of block with pre-existant forced breaks;
wenzelm
parents:
61875
diff
changeset
|
84 |
} |
62785
70b9c7d4ed7f
more robust pretty printing: permissive treatment of bad values;
wenzelm
parents:
61883
diff
changeset
|
85 |
Block(markup, consistent, indent1, body, body_length(body, 0.0)) |
61883
c0f34fe6aa61
clarified length of block with pre-existant forced breaks;
wenzelm
parents:
61875
diff
changeset
|
86 |
} |
36683 | 87 |
|
61874 | 88 |
|
89 |
/* formatted output */ |
|
90 |
||
91 |
private sealed case class Text(tx: XML.Body = Nil, pos: Double = 0.0, nl: Int = 0) |
|
92 |
{ |
|
93 |
def newline: Text = copy(tx = fbrk :: tx, pos = 0.0, nl = nl + 1) |
|
94 |
def string(s: String, len: Double): Text = |
|
95 |
copy(tx = if (s == "") tx else XML.Text(s) :: tx, pos = pos + len) |
|
96 |
def blanks(wd: Int): Text = string(Symbol.spaces(wd), wd.toDouble) |
|
97 |
def content: XML.Body = tx.reverse |
|
98 |
} |
|
99 |
||
100 |
private def break_dist(trees: List[Tree], after: Double): Double = |
|
101 |
trees match { |
|
102 |
case (_: Break) :: _ => 0.0 |
|
103 |
case t :: ts => t.length + break_dist(ts, after) |
|
104 |
case Nil => after |
|
36689
379f5b1e7f91
replaced slightly odd fbreak markup by plain "\n", which also coincides with regular linebreaks produced outside the ML pretty engine;
wenzelm
parents:
36687
diff
changeset
|
105 |
} |
379f5b1e7f91
replaced slightly odd fbreak markup by plain "\n", which also coincides with regular linebreaks produced outside the ML pretty engine;
wenzelm
parents:
36687
diff
changeset
|
106 |
|
61874 | 107 |
private def force_break(tree: Tree): Tree = |
108 |
tree match { case Break(false, wd, ind) => Break(true, wd, ind) case _ => tree } |
|
71601 | 109 |
private def force_all(trees: List[Tree]): List[Tree] = trees.map(force_break) |
51570
3633828d80fc
basic support for Pretty.item, which is considered as logical markup and interpreted in Isabelle/Scala, but ignored elsewhere (TTY, latex etc.);
wenzelm
parents:
51569
diff
changeset
|
110 |
|
61874 | 111 |
private def force_next(trees: List[Tree]): List[Tree] = |
112 |
trees match { |
|
113 |
case Nil => Nil |
|
114 |
case (t: Break) :: ts => force_break(t) :: ts |
|
115 |
case t :: ts => t :: force_next(ts) |
|
116 |
} |
|
51570
3633828d80fc
basic support for Pretty.item, which is considered as logical markup and interpreted in Isabelle/Scala, but ignored elsewhere (TTY, latex etc.);
wenzelm
parents:
51569
diff
changeset
|
117 |
|
71601 | 118 |
val default_margin: Double = 76.0 |
119 |
val default_breakgain: Double = default_margin / 20 |
|
36820 | 120 |
|
67547
aefe7a7b330a
clarified breakgain: keeping it constant avoids margin fluctuation in Pretty_Tooltip vs. Pretty_Text_Area;
wenzelm
parents:
65130
diff
changeset
|
121 |
def formatted(input: XML.Body, |
67896 | 122 |
margin: Double = default_margin, |
123 |
breakgain: Double = default_breakgain, |
|
124 |
metric: Metric = Default_Metric): XML.Body = |
|
36687 | 125 |
{ |
51569
4e084727faae
maintain integer indentation during formatting -- it needs to be implemented by repeated spaces eventually;
wenzelm
parents:
51568
diff
changeset
|
126 |
val emergencypos = (margin / 2).round.toInt |
36687 | 127 |
|
61874 | 128 |
def make_tree(inp: XML.Body): List[Tree] = |
129 |
inp flatMap { |
|
130 |
case XML.Wrapped_Elem(markup, body1, body2) => |
|
131 |
List(make_block(Some(markup, Some(body1)), false, 0, make_tree(body2))) |
|
132 |
case XML.Elem(markup, body) => |
|
133 |
markup match { |
|
134 |
case Markup.Block(consistent, indent) => |
|
135 |
List(make_block(None, consistent, indent, make_tree(body))) |
|
136 |
case Markup.Break(width, indent) => |
|
62785
70b9c7d4ed7f
more robust pretty printing: permissive treatment of bad values;
wenzelm
parents:
61883
diff
changeset
|
137 |
List(Break(false, force_nat(width), force_nat(indent))) |
61874 | 138 |
case Markup(Markup.ITEM, _) => |
139 |
List(make_block(None, false, 2, |
|
140 |
make_tree(XML.elem(Markup.BULLET, space) :: space ::: body))) |
|
141 |
case _ => |
|
142 |
List(make_block(Some((markup, None)), false, 0, make_tree(body))) |
|
143 |
} |
|
144 |
case XML.Text(text) => |
|
145 |
Library.separate(FBreak, split_lines(text).map(s => Str(s, metric(s)))) |
|
36817
ed97e877ff2d
more precise pretty printing based on actual font metrics;
wenzelm
parents:
36763
diff
changeset
|
146 |
} |
ed97e877ff2d
more precise pretty printing based on actual font metrics;
wenzelm
parents:
36763
diff
changeset
|
147 |
|
61871 | 148 |
def format(trees: List[Tree], blockin: Int, after: Double, text: Text): Text = |
36687 | 149 |
trees match { |
150 |
case Nil => text |
|
151 |
||
61871 | 152 |
case Block(markup, consistent, indent, body, blen) :: ts => |
51569
4e084727faae
maintain integer indentation during formatting -- it needs to be implemented by repeated spaces eventually;
wenzelm
parents:
51568
diff
changeset
|
153 |
val pos1 = (text.pos + indent).ceil.toInt |
36687 | 154 |
val pos2 = pos1 % emergencypos |
61868 | 155 |
val blockin1 = if (pos1 < emergencypos) pos1 else pos2 |
61864 | 156 |
val d = break_dist(ts, after) |
157 |
val body1 = if (consistent && text.pos + blen > margin - d) force_all(body) else body |
|
61871 | 158 |
val btext = |
159 |
markup match { |
|
160 |
case None => format(body1, blockin1, d, text) |
|
161 |
case Some((m, markup_body)) => |
|
162 |
val btext0 = format(body1, blockin1, d, text.copy(tx = Nil)) |
|
163 |
val elem = |
|
164 |
markup_body match { |
|
165 |
case None => XML.Elem(m, btext0.content) |
|
166 |
case Some(b) => XML.Wrapped_Elem(m, b, btext0.content) |
|
167 |
} |
|
168 |
btext0.copy(tx = elem :: text.tx) |
|
169 |
} |
|
61864 | 170 |
val ts1 = if (text.nl < btext.nl) force_next(ts) else ts |
36687 | 171 |
format(ts1, blockin, after, btext) |
172 |
||
61871 | 173 |
case Break(force, wd, ind) :: ts => |
174 |
if (!force && |
|
175 |
text.pos + wd <= ((margin - break_dist(ts, after)) max (blockin + breakgain))) |
|
36687 | 176 |
format(ts, blockin, after, text.blanks(wd)) |
61862
e2a9e46ac0fb
support pretty break indent, like underlying ML systems;
wenzelm
parents:
55551
diff
changeset
|
177 |
else format(ts, blockin, after, text.newline.blanks(blockin + ind)) |
36687 | 178 |
|
61871 | 179 |
case Str(s, len) :: ts => format(ts, blockin, after, text.string(s, len)) |
36687 | 180 |
} |
61874 | 181 |
format(make_tree(input), 0, 0.0, Text()).content |
36687 | 182 |
} |
36734
d9b10c173330
Pretty.formatted operates directly on XML trees, treating XML.Elem like a pro-forma block of indentation 0, like the ML version;
wenzelm
parents:
36689
diff
changeset
|
183 |
|
67547
aefe7a7b330a
clarified breakgain: keeping it constant avoids margin fluctuation in Pretty_Tooltip vs. Pretty_Text_Area;
wenzelm
parents:
65130
diff
changeset
|
184 |
def string_of(input: XML.Body, |
67896 | 185 |
margin: Double = default_margin, |
186 |
breakgain: Double = default_breakgain, |
|
187 |
metric: Metric = Default_Metric): String = |
|
67547
aefe7a7b330a
clarified breakgain: keeping it constant avoids margin fluctuation in Pretty_Tooltip vs. Pretty_Text_Area;
wenzelm
parents:
65130
diff
changeset
|
188 |
XML.content(formatted(input, margin = margin, breakgain = breakgain, metric = metric)) |
36683 | 189 |
} |