| author | wenzelm | 
| Tue, 10 Dec 2024 21:06:04 +0100 | |
| changeset 81572 | 693a95492008 | 
| parent 81395 | d9f791f75b8b | 
| child 81685 | 13bd6223691d | 
| 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 | |
| 75393 | 12 | object Pretty {
 | 
| 61871 | 13 | /* XML constructors */ | 
| 14 | ||
| 61874 | 15 | val space: XML.Body = List(XML.Text(Symbol.space)) | 
| 16 | def spaces(n: Int): XML.Body = | |
| 17 | if (n == 0) Nil | |
| 18 | else if (n == 1) space | |
| 19 | else List(XML.Text(Symbol.spaces(n))) | |
| 48704 
85a3de10567d
tuned signature -- make Pretty less dependent on Symbol;
 wenzelm parents: 
45666diff
changeset | 20 | |
| 80940 | 21 | val bullet: XML.Body = XML.elem(Markup.BULLET, space) :: space | 
| 22 | ||
| 81121 
7cacedbddba7
support for pretty blocks that are "open" and thus have no impact on formatting, only on markup;
 wenzelm parents: 
81120diff
changeset | 23 | def block(body: XML.Body, | 
| 
7cacedbddba7
support for pretty blocks that are "open" and thus have no impact on formatting, only on markup;
 wenzelm parents: 
81120diff
changeset | 24 | consistent: Boolean = false, | 
| 
7cacedbddba7
support for pretty blocks that are "open" and thus have no impact on formatting, only on markup;
 wenzelm parents: 
81120diff
changeset | 25 | indent: Int = 2 | 
| 81395 | 26 | ): XML.Elem = XML.Elem(Markup.Block(consistent = consistent, indent = indent), body) | 
| 48704 
85a3de10567d
tuned signature -- make Pretty less dependent on Symbol;
 wenzelm parents: 
45666diff
changeset | 27 | |
| 81395 | 28 | def string(s: String): XML.Elem = block(XML.string(s), indent = 0) | 
| 29 | ||
| 30 | def brk(width: Int, indent: Int = 0): XML.Elem = | |
| 81120 | 31 | XML.Elem(Markup.Break(width = width, indent = indent), spaces(width)) | 
| 61871 | 32 | |
| 69867 | 33 | val fbrk: XML.Tree = XML.newline | 
| 65130 | 34 | def fbreaks(ts: List[XML.Tree]): XML.Body = Library.separate(fbrk, ts) | 
| 61871 | 35 | |
| 36 | val Separator: XML.Body = List(XML.elem(Markup.SEPARATOR, space), fbrk) | |
| 75958 | 37 | def separate(ts: List[XML.Tree], sep: XML.Body = Separator): XML.Body = | 
| 38 | Library.separate(sep, ts.map(List(_))).flatten | |
| 39 | ||
| 40 |   val comma: XML.Body = List(XML.Text(","), brk(1))
 | |
| 41 | def commas(ts: List[XML.Tree]): XML.Body = separate(ts, sep = comma) | |
| 42 | ||
| 43 | def `enum`(ts: List[XML.Tree], | |
| 44 |     bg: String = "(",
 | |
| 45 | en: String = ")", | |
| 46 | sep: XML.Body = comma, | |
| 47 | indent: Int = 2 | |
| 81395 | 48 | ): XML.Elem = Pretty.block(XML.enclose(bg, en, separate(ts, sep = sep)), indent = indent) | 
| 61864 | 49 | |
| 48704 
85a3de10567d
tuned signature -- make Pretty less dependent on Symbol;
 wenzelm parents: 
45666diff
changeset | 50 | |
| 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: 
51470diff
changeset | 51 | /* 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: 
51470diff
changeset | 52 | |
| 75393 | 53 |   abstract class Metric {
 | 
| 81340 | 54 | def unit: Double | 
| 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: 
51470diff
changeset | 55 | 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: 
51470diff
changeset | 56 | } | 
| 
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: 
51470diff
changeset | 57 | |
| 
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: 
51470diff
changeset | 58 | |
| 36687 | 59 | /* markup trees with physical blocks and breaks */ | 
| 60 | ||
| 62820 | 61 | private def force_nat(i: Int): Int = i max 0 | 
| 62 | ||
| 61874 | 63 |   private sealed abstract class Tree { def length: Double }
 | 
| 64 | private case class Block( | |
| 80800 | 65 | markup: Markup, | 
| 66 | markup_body: Option[XML.Body], | |
| 81121 
7cacedbddba7
support for pretty blocks that are "open" and thus have no impact on formatting, only on markup;
 wenzelm parents: 
81120diff
changeset | 67 | open_block: Boolean, | 
| 80798 | 68 | consistent: Boolean, | 
| 69 | indent: Int, | |
| 70 | body: List[Tree], | |
| 71 | length: Double | |
| 72 | ) extends Tree | |
| 75393 | 73 |   private case class Break(force: Boolean, width: Int, indent: Int) extends Tree {
 | 
| 74 | def length: Double = width.toDouble | |
| 75 | } | |
| 61874 | 76 | private case class Str(string: String, length: Double) extends Tree | 
| 36683 | 77 | |
| 61874 | 78 | private val FBreak = Break(true, 1, 0) | 
| 36683 | 79 | |
| 80940 | 80 | private def make_block(body: List[Tree], | 
| 81 | markup: Markup = Markup.Empty, | |
| 82 | markup_body: Option[XML.Body] = None, | |
| 81121 
7cacedbddba7
support for pretty blocks that are "open" and thus have no impact on formatting, only on markup;
 wenzelm parents: 
81120diff
changeset | 83 | open_block: Boolean = false, | 
| 80940 | 84 | consistent: Boolean = false, | 
| 85 | indent: Int = 0 | |
| 75393 | 86 |   ): Tree = {
 | 
| 62785 
70b9c7d4ed7f
more robust pretty printing: permissive treatment of bad values;
 wenzelm parents: 
61883diff
changeset | 87 | val indent1 = force_nat(indent) | 
| 
70b9c7d4ed7f
more robust pretty printing: permissive treatment of bad values;
 wenzelm parents: 
61883diff
changeset | 88 | |
| 75393 | 89 |     @tailrec def body_length(prts: List[Tree], len: Double): Double = {
 | 
| 61883 
c0f34fe6aa61
clarified length of block with pre-existant forced breaks;
 wenzelm parents: 
61875diff
changeset | 90 | val (line, rest) = | 
| 
c0f34fe6aa61
clarified length of block with pre-existant forced breaks;
 wenzelm parents: 
61875diff
changeset | 91 |         Library.take_prefix[Tree]({ case Break(true, _, _) => false case _ => true }, prts)
 | 
| 73359 | 92 |       val len1 = (line.foldLeft(0.0) { case (l, t) => l + t.length }) max len
 | 
| 71781 | 93 |       (rest: @unchecked) match {
 | 
| 61883 
c0f34fe6aa61
clarified length of block with pre-existant forced breaks;
 wenzelm parents: 
61875diff
changeset | 94 | case Break(true, _, ind) :: rest1 => | 
| 62785 
70b9c7d4ed7f
more robust pretty printing: permissive treatment of bad values;
 wenzelm parents: 
61883diff
changeset | 95 | body_length(Break(false, indent1 + ind, 0) :: rest1, len1) | 
| 61883 
c0f34fe6aa61
clarified length of block with pre-existant forced breaks;
 wenzelm parents: 
61875diff
changeset | 96 | case Nil => len1 | 
| 
c0f34fe6aa61
clarified length of block with pre-existant forced breaks;
 wenzelm parents: 
61875diff
changeset | 97 | } | 
| 
c0f34fe6aa61
clarified length of block with pre-existant forced breaks;
 wenzelm parents: 
61875diff
changeset | 98 | } | 
| 81121 
7cacedbddba7
support for pretty blocks that are "open" and thus have no impact on formatting, only on markup;
 wenzelm parents: 
81120diff
changeset | 99 | Block(markup, markup_body, open_block, consistent, indent1, body, body_length(body, 0.0)) | 
| 61883 
c0f34fe6aa61
clarified length of block with pre-existant forced breaks;
 wenzelm parents: 
61875diff
changeset | 100 | } | 
| 36683 | 101 | |
| 61874 | 102 | |
| 80845 | 103 | /* no formatting */ | 
| 76086 
338adf8d423c
support Pretty.unformatted, similar to ML version;
 wenzelm parents: 
75958diff
changeset | 104 | |
| 80871 | 105 | def output_content(pure: Boolean, output: XML.Body): String = | 
| 106 | XML.content(if (pure) Protocol_Message.clean_output(output) else output) | |
| 107 | ||
| 80807 | 108 | def unbreakable(input: XML.Body): XML.Body = | 
| 76086 
338adf8d423c
support Pretty.unformatted, similar to ML version;
 wenzelm parents: 
75958diff
changeset | 109 |     input flatMap {
 | 
| 
338adf8d423c
support Pretty.unformatted, similar to ML version;
 wenzelm parents: 
75958diff
changeset | 110 | case XML.Wrapped_Elem(markup, body1, body2) => | 
| 80807 | 111 | List(XML.Wrapped_Elem(markup, body1, unbreakable(body2))) | 
| 112 | case XML.Elem(Markup.Break(width, _), _) => spaces(width) | |
| 113 | case XML.Elem(markup, body) => List(XML.Elem(markup, unbreakable(body))) | |
| 76086 
338adf8d423c
support Pretty.unformatted, similar to ML version;
 wenzelm parents: 
75958diff
changeset | 114 | case XML.Text(text) => XML.string(split_lines(text).mkString(Symbol.space)) | 
| 
338adf8d423c
support Pretty.unformatted, similar to ML version;
 wenzelm parents: 
75958diff
changeset | 115 | } | 
| 80807 | 116 | |
| 80871 | 117 | def unformatted_string_of(input: XML.Body, pure: Boolean = false): String = | 
| 118 | output_content(pure, unbreakable(input)) | |
| 76086 
338adf8d423c
support Pretty.unformatted, similar to ML version;
 wenzelm parents: 
75958diff
changeset | 119 | |
| 
338adf8d423c
support Pretty.unformatted, similar to ML version;
 wenzelm parents: 
75958diff
changeset | 120 | |
| 80845 | 121 | /* formatting */ | 
| 61874 | 122 | |
| 75393 | 123 |   private sealed case class Text(tx: XML.Body = Nil, pos: Double = 0.0, nl: Int = 0) {
 | 
| 61874 | 124 | def newline: Text = copy(tx = fbrk :: tx, pos = 0.0, nl = nl + 1) | 
| 125 | def string(s: String, len: Double): Text = | |
| 126 | copy(tx = if (s == "") tx else XML.Text(s) :: tx, pos = pos + len) | |
| 127 | def blanks(wd: Int): Text = string(Symbol.spaces(wd), wd.toDouble) | |
| 128 | def content: XML.Body = tx.reverse | |
| 129 | } | |
| 130 | ||
| 131 | private def break_dist(trees: List[Tree], after: Double): Double = | |
| 132 |     trees match {
 | |
| 133 | case (_: Break) :: _ => 0.0 | |
| 134 | case t :: ts => t.length + break_dist(ts, after) | |
| 135 | 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: 
36687diff
changeset | 136 | } | 
| 
379f5b1e7f91
replaced slightly odd fbreak markup by plain "\n", which also coincides with regular linebreaks produced outside the ML pretty engine;
 wenzelm parents: 
36687diff
changeset | 137 | |
| 61874 | 138 | private def force_break(tree: Tree): Tree = | 
| 139 |     tree match { case Break(false, wd, ind) => Break(true, wd, ind) case _ => tree }
 | |
| 71601 | 140 | 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: 
51569diff
changeset | 141 | |
| 61874 | 142 | private def force_next(trees: List[Tree]): List[Tree] = | 
| 143 |     trees match {
 | |
| 144 | case Nil => Nil | |
| 145 | case (t: Break) :: ts => force_break(t) :: ts | |
| 146 | case t :: ts => t :: force_next(ts) | |
| 147 | } | |
| 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: 
51569diff
changeset | 148 | |
| 71601 | 149 | val default_margin: Double = 76.0 | 
| 150 | val default_breakgain: Double = default_margin / 20 | |
| 36820 | 151 | |
| 67547 
aefe7a7b330a
clarified breakgain: keeping it constant avoids margin fluctuation in Pretty_Tooltip vs. Pretty_Text_Area;
 wenzelm parents: 
65130diff
changeset | 152 | def formatted(input: XML.Body, | 
| 67896 | 153 | margin: Double = default_margin, | 
| 154 | breakgain: Double = default_breakgain, | |
| 81346 | 155 | metric: Metric = Codepoint.Metric | 
| 75393 | 156 |   ): XML.Body = {
 | 
| 51569 
4e084727faae
maintain integer indentation during formatting -- it needs to be implemented by repeated spaces eventually;
 wenzelm parents: 
51568diff
changeset | 157 | val emergencypos = (margin / 2).round.toInt | 
| 36687 | 158 | |
| 61874 | 159 | def make_tree(inp: XML.Body): List[Tree] = | 
| 160 |       inp flatMap {
 | |
| 161 | case XML.Wrapped_Elem(markup, body1, body2) => | |
| 80940 | 162 | List(make_block(make_tree(body2), markup = markup, markup_body = Some(body1))) | 
| 61874 | 163 | case XML.Elem(markup, body) => | 
| 164 |           markup match {
 | |
| 81294 
108284c8cbfd
removed obsolete markup for "open_block" (see also d5ad89fda714): Isabelle/Scala directly supports XML.Elem pretty-printing;
 wenzelm parents: 
81121diff
changeset | 165 | case Markup.Block(consistent, indent) => | 
| 81121 
7cacedbddba7
support for pretty blocks that are "open" and thus have no impact on formatting, only on markup;
 wenzelm parents: 
81120diff
changeset | 166 | List( | 
| 
7cacedbddba7
support for pretty blocks that are "open" and thus have no impact on formatting, only on markup;
 wenzelm parents: 
81120diff
changeset | 167 | make_block(make_tree(body), | 
| 81294 
108284c8cbfd
removed obsolete markup for "open_block" (see also d5ad89fda714): Isabelle/Scala directly supports XML.Elem pretty-printing;
 wenzelm parents: 
81121diff
changeset | 168 | consistent = consistent, indent = indent, open_block = false)) | 
| 61874 | 169 | case Markup.Break(width, indent) => | 
| 62785 
70b9c7d4ed7f
more robust pretty printing: permissive treatment of bad values;
 wenzelm parents: 
61883diff
changeset | 170 | List(Break(false, force_nat(width), force_nat(indent))) | 
| 61874 | 171 | case Markup(Markup.ITEM, _) => | 
| 80941 
fd7a70babec1
more markup <expression kind="item"> in Isabelle/Scala, with pro-forma Markup_Kind.setup in Isabelle/ML;
 wenzelm parents: 
80940diff
changeset | 172 | List(make_block(make_tree(bullet ::: body), | 
| 
fd7a70babec1
more markup <expression kind="item"> in Isabelle/Scala, with pro-forma Markup_Kind.setup in Isabelle/ML;
 wenzelm parents: 
80940diff
changeset | 173 | markup = Markup.Expression.item, indent = 2)) | 
| 61874 | 174 | case _ => | 
| 81121 
7cacedbddba7
support for pretty blocks that are "open" and thus have no impact on formatting, only on markup;
 wenzelm parents: 
81120diff
changeset | 175 | List(make_block(make_tree(body), markup = markup, open_block = true)) | 
| 61874 | 176 | } | 
| 177 | case XML.Text(text) => | |
| 178 | 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: 
36763diff
changeset | 179 | } | 
| 
ed97e877ff2d
more precise pretty printing based on actual font metrics;
 wenzelm parents: 
36763diff
changeset | 180 | |
| 61871 | 181 | def format(trees: List[Tree], blockin: Int, after: Double, text: Text): Text = | 
| 36687 | 182 |       trees match {
 | 
| 183 | case Nil => text | |
| 184 | ||
| 81121 
7cacedbddba7
support for pretty blocks that are "open" and thus have no impact on formatting, only on markup;
 wenzelm parents: 
81120diff
changeset | 185 | case (block: Block) :: ts if block.open_block => | 
| 
7cacedbddba7
support for pretty blocks that are "open" and thus have no impact on formatting, only on markup;
 wenzelm parents: 
81120diff
changeset | 186 | val btext = format(block.body, blockin, break_dist(ts, after), text.copy(tx = Nil)) | 
| 
7cacedbddba7
support for pretty blocks that are "open" and thus have no impact on formatting, only on markup;
 wenzelm parents: 
81120diff
changeset | 187 | val ts1 = if (text.nl < btext.nl) force_next(ts) else ts | 
| 
7cacedbddba7
support for pretty blocks that are "open" and thus have no impact on formatting, only on markup;
 wenzelm parents: 
81120diff
changeset | 188 | val btext1 = btext.copy(tx = XML.Elem(block.markup, btext.content) :: text.tx) | 
| 
7cacedbddba7
support for pretty blocks that are "open" and thus have no impact on formatting, only on markup;
 wenzelm parents: 
81120diff
changeset | 189 | format(ts1, blockin, after, btext1) | 
| 
7cacedbddba7
support for pretty blocks that are "open" and thus have no impact on formatting, only on markup;
 wenzelm parents: 
81120diff
changeset | 190 | |
| 
7cacedbddba7
support for pretty blocks that are "open" and thus have no impact on formatting, only on markup;
 wenzelm parents: 
81120diff
changeset | 191 | case Block(markup, markup_body, _, consistent, indent, body, blen) :: ts => | 
| 51569 
4e084727faae
maintain integer indentation during formatting -- it needs to be implemented by repeated spaces eventually;
 wenzelm parents: 
51568diff
changeset | 192 | val pos1 = (text.pos + indent).ceil.toInt | 
| 36687 | 193 | val pos2 = pos1 % emergencypos | 
| 61868 | 194 | val blockin1 = if (pos1 < emergencypos) pos1 else pos2 | 
| 61864 | 195 | val d = break_dist(ts, after) | 
| 196 | val body1 = if (consistent && text.pos + blen > margin - d) force_all(body) else body | |
| 61871 | 197 | val btext = | 
| 80800 | 198 | if (markup.is_empty && markup_body.isEmpty) format(body1, blockin1, d, text) | 
| 199 |             else {
 | |
| 200 | val btext0 = format(body1, blockin1, d, text.copy(tx = Nil)) | |
| 201 | val elem = | |
| 202 |                 markup_body match {
 | |
| 203 | case None => XML.Elem(markup, btext0.content) | |
| 204 | case Some(body1) => XML.Wrapped_Elem(markup, body1, btext0.content) | |
| 205 | } | |
| 206 | btext0.copy(tx = elem :: text.tx) | |
| 61871 | 207 | } | 
| 61864 | 208 | val ts1 = if (text.nl < btext.nl) force_next(ts) else ts | 
| 36687 | 209 | format(ts1, blockin, after, btext) | 
| 210 | ||
| 61871 | 211 | case Break(force, wd, ind) :: ts => | 
| 212 | if (!force && | |
| 213 | text.pos + wd <= ((margin - break_dist(ts, after)) max (blockin + breakgain))) | |
| 36687 | 214 | format(ts, blockin, after, text.blanks(wd)) | 
| 61862 
e2a9e46ac0fb
support pretty break indent, like underlying ML systems;
 wenzelm parents: 
55551diff
changeset | 215 | else format(ts, blockin, after, text.newline.blanks(blockin + ind)) | 
| 36687 | 216 | |
| 61871 | 217 | case Str(s, len) :: ts => format(ts, blockin, after, text.string(s, len)) | 
| 36687 | 218 | } | 
| 61874 | 219 | format(make_tree(input), 0, 0.0, Text()).content | 
| 36687 | 220 | } | 
| 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: 
36689diff
changeset | 221 | |
| 67547 
aefe7a7b330a
clarified breakgain: keeping it constant avoids margin fluctuation in Pretty_Tooltip vs. Pretty_Text_Area;
 wenzelm parents: 
65130diff
changeset | 222 | def string_of(input: XML.Body, | 
| 80871 | 223 | margin: Double = default_margin, | 
| 224 | breakgain: Double = default_breakgain, | |
| 81346 | 225 | metric: Metric = Codepoint.Metric, | 
| 80871 | 226 | pure: Boolean = false | 
| 227 |   ): String = {
 | |
| 228 | output_content(pure, formatted(input, margin = margin, breakgain = breakgain, metric = metric)) | |
| 229 | } | |
| 36683 | 230 | } |