| author | paulson | 
| Mon, 08 Nov 2021 09:31:26 +0000 | |
| changeset 74730 | 25f5f1fa31bb | 
| parent 74056 | fb8d5c0133c9 | 
| child 75107 | 7c0217c8b8a5 | 
| permissions | -rw-r--r-- | 
| 43600 | 1 | /* Title: Pure/General/path.scala | 
| 2 | Author: Makarius | |
| 3 | ||
| 43601 | 4 | Algebra of file-system paths: basic POSIX notation, extended by named | 
| 5 | roots (e.g. //foo) and variables (e.g. $BAR). | |
| 43600 | 6 | */ | 
| 7 | ||
| 8 | package isabelle | |
| 9 | ||
| 10 | ||
| 73897 | 11 | import java.util.{Map => JMap}
 | 
| 48409 | 12 | import java.io.{File => JFile}
 | 
| 73945 | 13 | import java.nio.file.{Path => JPath}
 | 
| 48373 | 14 | |
| 43697 
77ce24aa1770
explicit Document.Node.Header, with master_dir and thy_name;
 wenzelm parents: 
43670diff
changeset | 15 | import scala.util.matching.Regex | 
| 
77ce24aa1770
explicit Document.Node.Header, with master_dir and thy_name;
 wenzelm parents: 
43670diff
changeset | 16 | |
| 
77ce24aa1770
explicit Document.Node.Header, with master_dir and thy_name;
 wenzelm parents: 
43670diff
changeset | 17 | |
| 43600 | 18 | object Path | 
| 19 | {
 | |
| 20 | /* path elements */ | |
| 21 | ||
| 45244 | 22 | sealed abstract class Elem | 
| 60215 | 23 | private case class Root(name: String) extends Elem | 
| 24 | private case class Basic(name: String) extends Elem | |
| 25 | private case class Variable(name: String) extends Elem | |
| 43600 | 26 | private case object Parent extends Elem | 
| 27 | ||
| 28 | private def err_elem(msg: String, s: String): Nothing = | |
| 69547 | 29 | error(msg + " path element " + quote(s)) | 
| 30 | ||
| 69548 
415dc92050a6
more strict check: avoid confusion of Path.basic with Path.current / Path.parent;
 wenzelm parents: 
69547diff
changeset | 31 |   private val illegal_elem = Set("", "~", "~~", ".", "..")
 | 
| 69550 
57ff523d9008
reject further illegal chars according to https://docs.microsoft.com/en-us/windows/desktop/fileio/naming-a-file
 wenzelm parents: 
69549diff
changeset | 32 | private val illegal_char = "/\\$:\"'<>|?*" | 
| 43600 | 33 | |
| 34 | private def check_elem(s: String): String = | |
| 69547 | 35 |     if (illegal_elem.contains(s)) err_elem("Illegal", s)
 | 
| 52106 | 36 |     else {
 | 
| 69550 
57ff523d9008
reject further illegal chars according to https://docs.microsoft.com/en-us/windows/desktop/fileio/naming-a-file
 wenzelm parents: 
69549diff
changeset | 37 |       for (c <- s) {
 | 
| 
57ff523d9008
reject further illegal chars according to https://docs.microsoft.com/en-us/windows/desktop/fileio/naming-a-file
 wenzelm parents: 
69549diff
changeset | 38 | if (c.toInt < 32) | 
| 
57ff523d9008
reject further illegal chars according to https://docs.microsoft.com/en-us/windows/desktop/fileio/naming-a-file
 wenzelm parents: 
69549diff
changeset | 39 |           err_elem("Illegal control character " + c.toInt + " in", s)
 | 
| 
57ff523d9008
reject further illegal chars according to https://docs.microsoft.com/en-us/windows/desktop/fileio/naming-a-file
 wenzelm parents: 
69549diff
changeset | 40 | if (illegal_char.contains(c)) | 
| 
57ff523d9008
reject further illegal chars according to https://docs.microsoft.com/en-us/windows/desktop/fileio/naming-a-file
 wenzelm parents: 
69549diff
changeset | 41 |           err_elem("Illegal character " + quote(c.toString) + " in", s)
 | 
| 69547 | 42 | } | 
| 52106 | 43 | s | 
| 44 | } | |
| 43600 | 45 | |
| 46 | private def root_elem(s: String): Elem = Root(check_elem(s)) | |
| 47 | private def basic_elem(s: String): Elem = Basic(check_elem(s)) | |
| 48 | private def variable_elem(s: String): Elem = Variable(check_elem(s)) | |
| 49 | ||
| 50 | private def apply_elem(y: Elem, xs: List[Elem]): List[Elem] = | |
| 51 |     (y, xs) match {
 | |
| 52 | case (Root(_), _) => List(y) | |
| 53 | case (Parent, Root(_) :: _) => xs | |
| 54 | case (Parent, Basic(_) :: rest) => rest | |
| 55 | case _ => y :: xs | |
| 56 | } | |
| 57 | ||
| 58 | private def norm_elems(elems: List[Elem]): List[Elem] = | |
| 73361 | 59 | elems.foldRight(List.empty[Elem])(apply_elem) | 
| 43600 | 60 | |
| 56844 | 61 | private def implode_elem(elem: Elem, short: Boolean): String = | 
| 43600 | 62 |     elem match {
 | 
| 63 |       case Root("") => ""
 | |
| 64 | case Root(s) => "//" + s | |
| 65 | case Basic(s) => s | |
| 56844 | 66 |       case Variable("USER_HOME") if short => "~"
 | 
| 67 |       case Variable("ISABELLE_HOME") if short => "~~"
 | |
| 43600 | 68 | case Variable(s) => "$" + s | 
| 69 | case Parent => ".." | |
| 70 | } | |
| 71 | ||
| 72962 | 72 | private def squash_elem(elem: Elem): String = | 
| 73 |     elem match {
 | |
| 74 |       case Root("") => "ROOT"
 | |
| 75 | case Root(s) => "SERVER_" + s | |
| 76 | case Basic(s) => s | |
| 77 | case Variable(s) => s | |
| 78 | case Parent => "PARENT" | |
| 79 | } | |
| 80 | ||
| 43600 | 81 | |
| 82 | /* path constructors */ | |
| 83 | ||
| 45244 | 84 | val current: Path = new Path(Nil) | 
| 85 |   val root: Path = new Path(List(Root("")))
 | |
| 86 | def named_root(s: String): Path = new Path(List(root_elem(s))) | |
| 71601 | 87 | def make(elems: List[String]): Path = new Path(elems.reverse.map(basic_elem)) | 
| 45244 | 88 | def basic(s: String): Path = new Path(List(basic_elem(s))) | 
| 89 | def variable(s: String): Path = new Path(List(variable_elem(s))) | |
| 90 | val parent: Path = new Path(List(Parent)) | |
| 43600 | 91 | |
| 73522 | 92 |   val USER_HOME: Path = variable("USER_HOME")
 | 
| 93 |   val ISABELLE_HOME: Path = variable("ISABELLE_HOME")
 | |
| 94 | ||
| 43600 | 95 | |
| 96 | /* explode */ | |
| 97 | ||
| 98 | def explode(str: String): Path = | |
| 99 |   {
 | |
| 52106 | 100 | def explode_elem(s: String): Elem = | 
| 101 |       try {
 | |
| 102 | if (s == "..") Parent | |
| 103 |         else if (s == "~") Variable("USER_HOME")
 | |
| 104 |         else if (s == "~~") Variable("ISABELLE_HOME")
 | |
| 105 |         else if (s.startsWith("$")) variable_elem(s.substring(1))
 | |
| 106 | else basic_elem(s) | |
| 107 | } | |
| 108 |       catch { case ERROR(msg) => cat_error(msg, "The error(s) above occurred in " + quote(str)) }
 | |
| 109 | ||
| 43670 
7f933761764b
prefer space_explode/split_lines as in Isabelle/ML;
 wenzelm parents: 
43669diff
changeset | 110 |     val ss = space_explode('/', str)
 | 
| 43600 | 111 | val r = ss.takeWhile(_.isEmpty).length | 
| 112 | val es = ss.dropWhile(_.isEmpty) | |
| 113 | val (roots, raw_elems) = | |
| 114 | if (r == 0) (Nil, es) | |
| 115 |       else if (r == 1) (List(Root("")), es)
 | |
| 116 |       else if (es.isEmpty) (List(Root("")), Nil)
 | |
| 117 | else (List(root_elem(es.head)), es.tail) | |
| 52106 | 118 | val elems = raw_elems.filterNot(s => s.isEmpty || s == ".").map(explode_elem) | 
| 119 | ||
| 63866 | 120 | new Path(norm_elems(elems reverse_::: roots)) | 
| 43600 | 121 | } | 
| 43669 | 122 | |
| 55879 
ac979f750c1a
clarified path checks: avoid crash of rendering due to spurious errors;
 wenzelm parents: 
55555diff
changeset | 123 | def is_wellformed(str: String): Boolean = | 
| 48484 | 124 |     try { explode(str); true } catch { case ERROR(_) => false }
 | 
| 125 | ||
| 55879 
ac979f750c1a
clarified path checks: avoid crash of rendering due to spurious errors;
 wenzelm parents: 
55555diff
changeset | 126 | def is_valid(str: String): Boolean = | 
| 
ac979f750c1a
clarified path checks: avoid crash of rendering due to spurious errors;
 wenzelm parents: 
55555diff
changeset | 127 |     try { explode(str).expand; true } catch { case ERROR(_) => false }
 | 
| 
ac979f750c1a
clarified path checks: avoid crash of rendering due to spurious errors;
 wenzelm parents: 
55555diff
changeset | 128 | |
| 43669 | 129 | def split(str: String): List[Path] = | 
| 43670 
7f933761764b
prefer space_explode/split_lines as in Isabelle/ML;
 wenzelm parents: 
43669diff
changeset | 130 |     space_explode(':', str).filterNot(_.isEmpty).map(explode)
 | 
| 48457 | 131 | |
| 132 | ||
| 133 | /* encode */ | |
| 134 | ||
| 135 | val encode: XML.Encode.T[Path] = (path => XML.Encode.string(path.implode)) | |
| 69551 
adb52af5ba55
exclude file name components that are special on Windows;
 wenzelm parents: 
69550diff
changeset | 136 | |
| 
adb52af5ba55
exclude file name components that are special on Windows;
 wenzelm parents: 
69550diff
changeset | 137 | |
| 
adb52af5ba55
exclude file name components that are special on Windows;
 wenzelm parents: 
69550diff
changeset | 138 | /* reserved names */ | 
| 
adb52af5ba55
exclude file name components that are special on Windows;
 wenzelm parents: 
69550diff
changeset | 139 | |
| 
adb52af5ba55
exclude file name components that are special on Windows;
 wenzelm parents: 
69550diff
changeset | 140 | private val reserved_windows: Set[String] = | 
| 
adb52af5ba55
exclude file name components that are special on Windows;
 wenzelm parents: 
69550diff
changeset | 141 |     Set("CON", "PRN", "AUX", "NUL",
 | 
| 
adb52af5ba55
exclude file name components that are special on Windows;
 wenzelm parents: 
69550diff
changeset | 142 | "COM1", "COM2", "COM3", "COM4", "COM5", "COM6", "COM7", "COM8", "COM9", | 
| 
adb52af5ba55
exclude file name components that are special on Windows;
 wenzelm parents: 
69550diff
changeset | 143 | "LPT1", "LPT2", "LPT3", "LPT4", "LPT5", "LPT6", "LPT7", "LPT8", "LPT9") | 
| 
adb52af5ba55
exclude file name components that are special on Windows;
 wenzelm parents: 
69550diff
changeset | 144 | |
| 
adb52af5ba55
exclude file name components that are special on Windows;
 wenzelm parents: 
69550diff
changeset | 145 | def is_reserved(name: String): Boolean = | 
| 
adb52af5ba55
exclude file name components that are special on Windows;
 wenzelm parents: 
69550diff
changeset | 146 | Long_Name.explode(name).exists(a => reserved_windows.contains(Word.uppercase(a))) | 
| 69904 | 147 | |
| 148 | ||
| 149 | /* case-insensitive names */ | |
| 150 | ||
| 73340 | 151 | def check_case_insensitive(paths: List[Path]): Unit = | 
| 69904 | 152 |   {
 | 
| 153 | val table = | |
| 73359 | 154 |       paths.foldLeft(Multi_Map.empty[String, String]) { case (tab, path) =>
 | 
| 69904 | 155 | val name = path.expand.implode | 
| 156 | tab.insert(Word.lowercase(name), name) | |
| 73359 | 157 | } | 
| 69904 | 158 | val collisions = | 
| 159 |       (for { (_, coll) <- table.iterator_list if coll.length > 1 } yield coll).toList.flatten
 | |
| 160 |     if (collisions.nonEmpty) {
 | |
| 161 |       error(("Collision of file names due case-insensitivity:" :: collisions).mkString("\n  "))
 | |
| 162 | } | |
| 163 | } | |
| 43600 | 164 | } | 
| 165 | ||
| 43669 | 166 | |
| 72746 | 167 | final class Path private(protected val elems: List[Path.Elem]) // reversed elements | 
| 43600 | 168 | {
 | 
| 72746 | 169 | override def hashCode: Int = elems.hashCode | 
| 170 | override def equals(that: Any): Boolean = | |
| 171 |     that match {
 | |
| 172 | case other: Path => elems == other.elems | |
| 173 | case _ => false | |
| 174 | } | |
| 175 | ||
| 43600 | 176 | def is_current: Boolean = elems.isEmpty | 
| 59319 | 177 | def is_absolute: Boolean = elems.nonEmpty && elems.last.isInstanceOf[Path.Root] | 
| 65559 | 178 |   def is_root: Boolean = elems match { case List(Path.Root(_)) => true case _ => false }
 | 
| 43600 | 179 |   def is_basic: Boolean = elems match { case List(Path.Basic(_)) => true case _ => false }
 | 
| 72572 | 180 | def starts_basic: Boolean = elems.nonEmpty && elems.last.isInstanceOf[Path.Basic] | 
| 43600 | 181 | |
| 73360 | 182 | def +(other: Path): Path = new Path(other.elems.foldRight(elems)(Path.apply_elem)) | 
| 43600 | 183 | |
| 184 | ||
| 43604 | 185 | /* implode */ | 
| 43600 | 186 | |
| 56844 | 187 | private def gen_implode(short: Boolean): String = | 
| 43600 | 188 |     elems match {
 | 
| 189 | case Nil => "." | |
| 190 |       case List(Path.Root("")) => "/"
 | |
| 56844 | 191 |       case _ => elems.map(Path.implode_elem(_, short)).reverse.mkString("/")
 | 
| 43600 | 192 | } | 
| 56844 | 193 | def implode: String = gen_implode(false) | 
| 194 | def implode_short: String = gen_implode(true) | |
| 43600 | 195 | |
| 43652 | 196 | override def toString: String = quote(implode) | 
| 43600 | 197 | |
| 198 | ||
| 199 | /* base element */ | |
| 200 | ||
| 201 | private def split_path: (Path, String) = | |
| 202 |     elems match {
 | |
| 45244 | 203 | case Path.Basic(s) :: xs => (new Path(xs), s) | 
| 43604 | 204 |       case _ => error("Cannot split path into dir/base: " + toString)
 | 
| 43600 | 205 | } | 
| 206 | ||
| 207 | def dir: Path = split_path._1 | |
| 45244 | 208 | def base: Path = new Path(List(Path.Basic(split_path._2))) | 
| 43600 | 209 | |
| 74056 | 210 | def ends_with(a: String): Boolean = | 
| 211 |     elems match {
 | |
| 212 | case Path.Basic(b) :: _ => b.endsWith(a) | |
| 213 | case _ => false | |
| 214 | } | |
| 215 |   def is_java: Boolean = ends_with(".java")
 | |
| 216 |   def is_scala: Boolean = ends_with(".scala")
 | |
| 217 | ||
| 43600 | 218 | def ext(e: String): Path = | 
| 219 | if (e == "") this | |
| 220 |     else {
 | |
| 221 | val (prfx, s) = split_path | |
| 43604 | 222 | prfx + Path.basic(s + "." + e) | 
| 43600 | 223 | } | 
| 43604 | 224 | |
| 72575 | 225 |   def xz: Path = ext("xz")
 | 
| 73660 | 226 |   def xml: Path = ext("xml")
 | 
| 72962 | 227 |   def html: Path = ext("html")
 | 
| 72574 
d892f6d66402
build documents in Isabelle/Scala, based on generated tex files as session exports;
 wenzelm parents: 
72572diff
changeset | 228 |   def tex: Path = ext("tex")
 | 
| 
d892f6d66402
build documents in Isabelle/Scala, based on generated tex files as session exports;
 wenzelm parents: 
72572diff
changeset | 229 |   def pdf: Path = ext("pdf")
 | 
| 72776 | 230 |   def thy: Path = ext("thy")
 | 
| 73628 | 231 |   def tar: Path = ext("tar")
 | 
| 232 |   def gz: Path = ext("gz")
 | |
| 73691 
2f9877db82a1
reimplemented Mirabelle as Isabelle/ML presentation hook + Isabelle/Scala tool, but sledgehammer is still inactive;
 wenzelm parents: 
73660diff
changeset | 233 |   def log: Path = ext("log")
 | 
| 72574 
d892f6d66402
build documents in Isabelle/Scala, based on generated tex files as session exports;
 wenzelm parents: 
72572diff
changeset | 234 | |
| 53336 | 235 | def backup: Path = | 
| 236 |   {
 | |
| 237 | val (prfx, s) = split_path | |
| 238 | prfx + Path.basic(s + "~") | |
| 239 | } | |
| 240 | ||
| 58610 | 241 | def backup2: Path = | 
| 242 |   {
 | |
| 243 | val (prfx, s) = split_path | |
| 244 | prfx + Path.basic(s + "~~") | |
| 245 | } | |
| 246 | ||
| 72464 | 247 | def platform_exe: Path = | 
| 248 |     if (Platform.is_windows) ext("exe") else this
 | |
| 249 | ||
| 43697 
77ce24aa1770
explicit Document.Node.Header, with master_dir and thy_name;
 wenzelm parents: 
43670diff
changeset | 250 |   private val Ext = new Regex("(.*)\\.([^.]*)")
 | 
| 
77ce24aa1770
explicit Document.Node.Header, with master_dir and thy_name;
 wenzelm parents: 
43670diff
changeset | 251 | |
| 
77ce24aa1770
explicit Document.Node.Header, with master_dir and thy_name;
 wenzelm parents: 
43670diff
changeset | 252 | def split_ext: (Path, String) = | 
| 
77ce24aa1770
explicit Document.Node.Header, with master_dir and thy_name;
 wenzelm parents: 
43670diff
changeset | 253 |   {
 | 
| 
77ce24aa1770
explicit Document.Node.Header, with master_dir and thy_name;
 wenzelm parents: 
43670diff
changeset | 254 | val (prefix, base) = split_path | 
| 
77ce24aa1770
explicit Document.Node.Header, with master_dir and thy_name;
 wenzelm parents: 
43670diff
changeset | 255 |     base match {
 | 
| 
77ce24aa1770
explicit Document.Node.Header, with master_dir and thy_name;
 wenzelm parents: 
43670diff
changeset | 256 | case Ext(b, e) => (prefix + Path.basic(b), e) | 
| 56556 | 257 | case _ => (prefix + Path.basic(base), "") | 
| 43697 
77ce24aa1770
explicit Document.Node.Header, with master_dir and thy_name;
 wenzelm parents: 
43670diff
changeset | 258 | } | 
| 
77ce24aa1770
explicit Document.Node.Header, with master_dir and thy_name;
 wenzelm parents: 
43670diff
changeset | 259 | } | 
| 
77ce24aa1770
explicit Document.Node.Header, with master_dir and thy_name;
 wenzelm parents: 
43670diff
changeset | 260 | |
| 69367 | 261 | def drop_ext: Path = split_ext._1 | 
| 262 | def get_ext: String = split_ext._2 | |
| 263 | ||
| 72962 | 264 | def squash: Path = new Path(elems.map(elem => Path.Basic(Path.squash_elem(elem)))) | 
| 265 | ||
| 43604 | 266 | |
| 267 | /* expand */ | |
| 268 | ||
| 73897 | 269 | def expand_env(env: JMap[String, String]): Path = | 
| 43604 | 270 |   {
 | 
| 271 | def eval(elem: Path.Elem): List[Path.Elem] = | |
| 272 |       elem match {
 | |
| 43664 | 273 | case Path.Variable(s) => | 
| 64228 
b46969a851a9
expand relatively to given environment, notably remote HOME;
 wenzelm parents: 
63866diff
changeset | 274 | val path = Path.explode(Isabelle_System.getenv_strict(s, env)) | 
| 48658 | 275 | if (path.elems.exists(_.isInstanceOf[Path.Variable])) | 
| 73715 
bf51c23f3f99
clarified signature -- avoid odd warning about scala/bug#6675;
 wenzelm parents: 
73712diff
changeset | 276 |             error("Illegal path variable nesting: " + Properties.Eq(s, path.toString))
 | 
| 48658 | 277 | else path.elems | 
| 43604 | 278 | case x => List(x) | 
| 279 | } | |
| 280 | ||
| 71383 | 281 | new Path(Path.norm_elems(elems.flatMap(eval))) | 
| 43604 | 282 | } | 
| 48373 | 283 | |
| 64228 
b46969a851a9
expand relatively to given environment, notably remote HOME;
 wenzelm parents: 
63866diff
changeset | 284 | def expand: Path = expand_env(Isabelle_System.settings()) | 
| 
b46969a851a9
expand relatively to given environment, notably remote HOME;
 wenzelm parents: 
63866diff
changeset | 285 | |
| 69366 | 286 | def file_name: String = expand.base.implode | 
| 287 | ||
| 48373 | 288 | |
| 72784 | 289 | /* implode wrt. given directories */ | 
| 48548 | 290 | |
| 72784 | 291 | def implode_symbolic: String = | 
| 292 |   {
 | |
| 293 | val directories = | |
| 294 |       Library.space_explode(':', Isabelle_System.getenv("ISABELLE_DIRECTORIES")).reverse
 | |
| 295 | val full_name = expand.implode | |
| 296 | directories.view.flatMap(a => | |
| 297 |       try {
 | |
| 298 | val b = Path.explode(a).expand.implode | |
| 299 | if (full_name == b) Some(a) | |
| 300 |         else {
 | |
| 301 |           Library.try_unprefix(b + "/", full_name) match {
 | |
| 302 | case Some(name) => Some(a + "/" + name) | |
| 303 | case None => None | |
| 304 | } | |
| 305 | } | |
| 306 |       } catch { case ERROR(_) => None }).headOption.getOrElse(implode)
 | |
| 307 | } | |
| 308 | ||
| 309 | def position: Position.T = Position.File(implode_symbolic) | |
| 48548 | 310 | |
| 311 | ||
| 66232 | 312 | /* platform files */ | 
| 48373 | 313 | |
| 60988 | 314 | def file: JFile = File.platform_file(this) | 
| 48548 | 315 | def is_file: Boolean = file.isFile | 
| 316 | def is_dir: Boolean = file.isDirectory | |
| 65833 | 317 | |
| 73945 | 318 | def java_path: JPath = file.toPath | 
| 319 | ||
| 66232 | 320 | def absolute_file: JFile = File.absolute(file) | 
| 321 | def canonical_file: JFile = File.canonical(file) | |
| 67181 | 322 | |
| 323 | def absolute: Path = File.path(absolute_file) | |
| 324 | def canonical: Path = File.path(canonical_file) | |
| 43600 | 325 | } |