| author | wenzelm | 
| Sat, 18 Mar 2017 20:35:58 +0100 | |
| changeset 65312 | 34d56ca5b548 | 
| parent 64610 | 1b89608974e9 | 
| child 71601 | 97ccf48c2f0c | 
| 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 | ||
| 13 | object Word | |
| 14 | {
 | |
| 62812 
ce22e5c3d4ce
more robust display of bidirectional Unicode text: enforce left-to-right;
 wenzelm parents: 
59319diff
changeset | 15 | /* directionality */ | 
| 
ce22e5c3d4ce
more robust display of bidirectional Unicode text: enforce left-to-right;
 wenzelm parents: 
59319diff
changeset | 16 | |
| 
ce22e5c3d4ce
more robust display of bidirectional Unicode text: enforce left-to-right;
 wenzelm parents: 
59319diff
changeset | 17 | def bidi_detect(str: String): Boolean = | 
| 
ce22e5c3d4ce
more robust display of bidirectional Unicode text: enforce left-to-right;
 wenzelm parents: 
59319diff
changeset | 18 | 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 | 19 | |
| 
ce22e5c3d4ce
more robust display of bidirectional Unicode text: enforce left-to-right;
 wenzelm parents: 
59319diff
changeset | 20 | def bidi_override(str: String): String = | 
| 
ce22e5c3d4ce
more robust display of bidirectional Unicode text: enforce left-to-right;
 wenzelm parents: 
59319diff
changeset | 21 | 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 | 22 | |
| 
ce22e5c3d4ce
more robust display of bidirectional Unicode text: enforce left-to-right;
 wenzelm parents: 
59319diff
changeset | 23 | |
| 56600 | 24 | /* case */ | 
| 25 | ||
| 56599 | 26 | def lowercase(str: String): String = str.toLowerCase(Locale.ROOT) | 
| 27 | def uppercase(str: String): String = str.toUpperCase(Locale.ROOT) | |
| 28 | ||
| 29 | def capitalize(str: String): String = | |
| 30 | if (str.length == 0) str | |
| 56601 | 31 |     else {
 | 
| 32 | val n = Character.charCount(str.codePointAt(0)) | |
| 56602 | 33 | uppercase(str.substring(0, n)) + lowercase(str.substring(n)) | 
| 56601 | 34 | } | 
| 35 | ||
| 56609 
5ac67041ccf8
capitalize more carefully, e.g. relevant for option "ML_exception_trace";
 wenzelm parents: 
56602diff
changeset | 36 | def perhaps_capitalize(str: String): String = | 
| 64610 | 37 | 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 | 38 | capitalize(str) | 
| 
16536c15d749
capitalize even more carefully (see 5ac67041ccf8), e.g. relevant for option "z3_non_commercial" and prospective "MaSh";
 wenzelm parents: 
56792diff
changeset | 39 | else str | 
| 56609 
5ac67041ccf8
capitalize more carefully, e.g. relevant for option "ML_exception_trace";
 wenzelm parents: 
56602diff
changeset | 40 | |
| 56601 | 41 | sealed abstract class Case | 
| 42 | case object Lowercase extends Case | |
| 43 | case object Uppercase extends Case | |
| 44 | case object Capitalized extends Case | |
| 45 | ||
| 46 | object Case | |
| 47 |   {
 | |
| 48 | def apply(c: Case, str: String): String = | |
| 49 |       c match {
 | |
| 50 | case Lowercase => lowercase(str) | |
| 51 | case Uppercase => uppercase(str) | |
| 52 | case Capitalized => capitalize(str) | |
| 53 | } | |
| 54 | def unapply(str: String): Option[Case] = | |
| 59319 | 55 |       if (str.nonEmpty) {
 | 
| 64610 | 56 | if (Codepoint.iterator(str).forall(Character.isLowerCase(_))) Some(Lowercase) | 
| 57 | else if (Codepoint.iterator(str).forall(Character.isUpperCase(_))) Some(Uppercase) | |
| 56601 | 58 |         else {
 | 
| 64610 | 59 | val it = Codepoint.iterator(str) | 
| 56601 | 60 | if (Character.isUpperCase(it.next) && it.forall(Character.isLowerCase(_))) | 
| 61 | Some(Capitalized) | |
| 62 | else None | |
| 63 | } | |
| 64 | } | |
| 65 | else None | |
| 66 | } | |
| 56599 | 67 | |
| 56600 | 68 | |
| 69 | /* sequence of words */ | |
| 70 | ||
| 71 |   def implode(words: Iterable[String]): String = words.iterator.mkString(" ")
 | |
| 72 | ||
| 73 | def explode(sep: Char => Boolean, text: String): List[String] = | |
| 74 | Library.separated_chunks(sep, text).map(_.toString).filter(_ != "").toList | |
| 75 | ||
| 76 | def explode(sep: Char, text: String): List[String] = | |
| 77 | explode(_ == sep, text) | |
| 78 | ||
| 79 | def explode(text: String): List[String] = | |
| 56747 | 80 | explode(Character.isWhitespace(_), text) | 
| 63450 | 81 | |
| 82 | ||
| 83 | /* brackets */ | |
| 84 | ||
| 85 |   val open_brackets = "([{«‹⟨⌈⌊⦇⟦⦃"
 | |
| 86 | val close_brackets = ")]}»›⟩⌉⌋⦈⟧⦄" | |
| 56599 | 87 | } |