| author | wenzelm | 
| Sun, 21 Aug 2022 15:00:14 +0200 | |
| changeset 75952 | 864b10457a7d | 
| parent 75393 | 87ebf5a50283 | 
| child 78613 | 60561d28569b | 
| permissions | -rw-r--r-- | 
| 56599 | 1 | /* Title: Pure/General/word.scala | 
| 2 | Author: Makarius | |
| 3 | ||
| 56747 | 4 | Support for words within Unicode text. | 
| 56599 | 5 | */ | 
| 6 | ||
| 7 | package isabelle | |
| 8 | ||
| 62812 
ce22e5c3d4ce
more robust display of bidirectional Unicode text: enforce left-to-right;
 wenzelm parents: 
59319diff
changeset | 9 | import java.text.Bidi | 
| 56599 | 10 | import java.util.Locale | 
| 11 | ||
| 12 | ||
| 75393 | 13 | object Word {
 | 
| 62812 
ce22e5c3d4ce
more robust display of bidirectional Unicode text: enforce left-to-right;
 wenzelm parents: 
59319diff
changeset | 14 | /* directionality */ | 
| 
ce22e5c3d4ce
more robust display of bidirectional Unicode text: enforce left-to-right;
 wenzelm parents: 
59319diff
changeset | 15 | |
| 
ce22e5c3d4ce
more robust display of bidirectional Unicode text: enforce left-to-right;
 wenzelm parents: 
59319diff
changeset | 16 | def bidi_detect(str: String): Boolean = | 
| 
ce22e5c3d4ce
more robust display of bidirectional Unicode text: enforce left-to-right;
 wenzelm parents: 
59319diff
changeset | 17 | str.exists(c => c >= 0x590) && Bidi.requiresBidi(str.toArray, 0, str.length) | 
| 
ce22e5c3d4ce
more robust display of bidirectional Unicode text: enforce left-to-right;
 wenzelm parents: 
59319diff
changeset | 18 | |
| 
ce22e5c3d4ce
more robust display of bidirectional Unicode text: enforce left-to-right;
 wenzelm parents: 
59319diff
changeset | 19 | def bidi_override(str: String): String = | 
| 
ce22e5c3d4ce
more robust display of bidirectional Unicode text: enforce left-to-right;
 wenzelm parents: 
59319diff
changeset | 20 | if (bidi_detect(str)) "\u200E\u202D" + str + "\u202C" else str | 
| 
ce22e5c3d4ce
more robust display of bidirectional Unicode text: enforce left-to-right;
 wenzelm parents: 
59319diff
changeset | 21 | |
| 
ce22e5c3d4ce
more robust display of bidirectional Unicode text: enforce left-to-right;
 wenzelm parents: 
59319diff
changeset | 22 | |
| 56600 | 23 | /* case */ | 
| 24 | ||
| 56599 | 25 | def lowercase(str: String): String = str.toLowerCase(Locale.ROOT) | 
| 26 | def uppercase(str: String): String = str.toUpperCase(Locale.ROOT) | |
| 27 | ||
| 28 | def capitalize(str: String): String = | |
| 29 | if (str.length == 0) str | |
| 56601 | 30 |     else {
 | 
| 31 | val n = Character.charCount(str.codePointAt(0)) | |
| 56602 | 32 | uppercase(str.substring(0, n)) + lowercase(str.substring(n)) | 
| 56601 | 33 | } | 
| 34 | ||
| 56609 
5ac67041ccf8
capitalize more carefully, e.g. relevant for option "ML_exception_trace";
 wenzelm parents: 
56602diff
changeset | 35 | def perhaps_capitalize(str: String): String = | 
| 64610 | 36 | if (Codepoint.iterator(str).forall(c => Character.isLowerCase(c) || Character.isDigit(c))) | 
| 57087 
16536c15d749
capitalize even more carefully (see 5ac67041ccf8), e.g. relevant for option "z3_non_commercial" and prospective "MaSh";
 wenzelm parents: 
56792diff
changeset | 37 | capitalize(str) | 
| 
16536c15d749
capitalize even more carefully (see 5ac67041ccf8), e.g. relevant for option "z3_non_commercial" and prospective "MaSh";
 wenzelm parents: 
56792diff
changeset | 38 | else str | 
| 56609 
5ac67041ccf8
capitalize more carefully, e.g. relevant for option "ML_exception_trace";
 wenzelm parents: 
56602diff
changeset | 39 | |
| 56601 | 40 | sealed abstract class Case | 
| 41 | case object Lowercase extends Case | |
| 42 | case object Uppercase extends Case | |
| 43 | case object Capitalized extends Case | |
| 44 | ||
| 75393 | 45 |   object Case {
 | 
| 56601 | 46 | def apply(c: Case, str: String): String = | 
| 47 |       c match {
 | |
| 48 | case Lowercase => lowercase(str) | |
| 49 | case Uppercase => uppercase(str) | |
| 50 | case Capitalized => capitalize(str) | |
| 51 | } | |
| 52 | def unapply(str: String): Option[Case] = | |
| 59319 | 53 |       if (str.nonEmpty) {
 | 
| 71601 | 54 | if (Codepoint.iterator(str).forall(Character.isLowerCase)) Some(Lowercase) | 
| 55 | else if (Codepoint.iterator(str).forall(Character.isUpperCase)) Some(Uppercase) | |
| 56601 | 56 |         else {
 | 
| 64610 | 57 | val it = Codepoint.iterator(str) | 
| 73344 | 58 | if (Character.isUpperCase(it.next()) && it.forall(Character.isLowerCase)) | 
| 56601 | 59 | Some(Capitalized) | 
| 60 | else None | |
| 61 | } | |
| 62 | } | |
| 63 | else None | |
| 64 | } | |
| 56599 | 65 | |
| 56600 | 66 | |
| 67 | /* sequence of words */ | |
| 68 | ||
| 69 |   def implode(words: Iterable[String]): String = words.iterator.mkString(" ")
 | |
| 70 | ||
| 71 | def explode(sep: Char => Boolean, text: String): List[String] = | |
| 72 | Library.separated_chunks(sep, text).map(_.toString).filter(_ != "").toList | |
| 73 | ||
| 74 | def explode(sep: Char, text: String): List[String] = | |
| 75 | explode(_ == sep, text) | |
| 76 | ||
| 77 | def explode(text: String): List[String] = | |
| 71867 | 78 | explode(Character.isWhitespace _, text) | 
| 63450 | 79 | |
| 80 | ||
| 81 | /* brackets */ | |
| 82 | ||
| 71866 | 83 |   val open_brackets = "([{«‹⟨⌈⌊⦇⟦⦃⟪"
 | 
| 84 | val close_brackets = ")]}»›⟩⌉⌋⦈⟧⦄⟫" | |
| 56599 | 85 | } |