| author | wenzelm | 
| Wed, 16 Oct 2024 21:41:05 +0200 | |
| changeset 81177 | 137ea3d464be | 
| parent 80481 | 0e2b09fef3d2 | 
| child 82119 | b7929e1dc4fb | 
| permissions | -rw-r--r-- | 
| 48411 
5b3440850d36
more abstract file system operations in Scala, corresponding to ML version;
 wenzelm parents: diff
changeset | 1 | /* Title: Pure/General/file.scala | 
| 
5b3440850d36
more abstract file system operations in Scala, corresponding to ML version;
 wenzelm parents: diff
changeset | 2 | Author: Makarius | 
| 
5b3440850d36
more abstract file system operations in Scala, corresponding to ML version;
 wenzelm parents: diff
changeset | 3 | |
| 64698 | 4 | File-system operations. | 
| 48411 
5b3440850d36
more abstract file system operations in Scala, corresponding to ML version;
 wenzelm parents: diff
changeset | 5 | */ | 
| 
5b3440850d36
more abstract file system operations in Scala, corresponding to ML version;
 wenzelm parents: diff
changeset | 6 | |
| 
5b3440850d36
more abstract file system operations in Scala, corresponding to ML version;
 wenzelm parents: diff
changeset | 7 | package isabelle | 
| 
5b3440850d36
more abstract file system operations in Scala, corresponding to ML version;
 wenzelm parents: diff
changeset | 8 | |
| 
5b3440850d36
more abstract file system operations in Scala, corresponding to ML version;
 wenzelm parents: diff
changeset | 9 | |
| 79980 | 10 | import java.util.{Properties => JProperties}
 | 
| 48411 
5b3440850d36
more abstract file system operations in Scala, corresponding to ML version;
 wenzelm parents: diff
changeset | 11 | import java.io.{BufferedWriter, OutputStreamWriter, FileOutputStream, BufferedOutputStream,
 | 
| 51504 | 12 | OutputStream, InputStream, FileInputStream, BufferedInputStream, BufferedReader, | 
| 13 | InputStreamReader, File => JFile, IOException} | |
| 73317 | 14 | import java.nio.file.{StandardOpenOption, Path => JPath, Files, SimpleFileVisitor,
 | 
| 15 | FileVisitOption, FileVisitResult} | |
| 78169 
5ad1ae8626de
minor performance tuning: avoid external process;
 wenzelm parents: 
78161diff
changeset | 16 | import java.nio.file.attribute.{BasicFileAttributes, PosixFilePermission}
 | 
| 79659 
a4118f530263
clarified signature: avoid ill-defined type java.net.URL;
 wenzelm parents: 
79045diff
changeset | 17 | import java.net.URI | 
| 50684 
12b7e0b4a66e
support File.read_gzip as well, in accordance to File.write_gzip;
 wenzelm parents: 
50203diff
changeset | 18 | import java.util.zip.{GZIPInputStream, GZIPOutputStream}
 | 
| 69293 
72a9860f8602
clarified find_files: follow links by default, e.g. relevant for "~/cronjob/log";
 wenzelm parents: 
66693diff
changeset | 19 | import java.util.EnumSet | 
| 48411 
5b3440850d36
more abstract file system operations in Scala, corresponding to ML version;
 wenzelm parents: diff
changeset | 20 | |
| 76353 | 21 | import org.tukaani.xz | 
| 22 | import com.github.luben.zstd | |
| 76348 | 23 | |
| 48411 
5b3440850d36
more abstract file system operations in Scala, corresponding to ML version;
 wenzelm parents: diff
changeset | 24 | import scala.collection.mutable | 
| 
5b3440850d36
more abstract file system operations in Scala, corresponding to ML version;
 wenzelm parents: diff
changeset | 25 | |
| 
5b3440850d36
more abstract file system operations in Scala, corresponding to ML version;
 wenzelm parents: diff
changeset | 26 | |
| 75393 | 27 | object File {
 | 
| 60992 | 28 | /* standard path (Cygwin or Posix) */ | 
| 60988 | 29 | |
| 30 | def standard_path(path: Path): String = path.expand.implode | |
| 31 | ||
| 60992 | 32 | def standard_path(platform_path: String): String = | 
| 73911 | 33 | isabelle.setup.Environment.standard_path(platform_path) | 
| 60992 | 34 | |
| 35 | def standard_path(file: JFile): String = standard_path(file.getPath) | |
| 36 | ||
| 37 | def standard_url(name: String): String = | |
| 38 |     try {
 | |
| 79044 
8cc1ae43e12e
clarified signature: avoid deprecated URL constructors;
 wenzelm parents: 
78956diff
changeset | 39 | val url = new URI(name).toURL | 
| 
8cc1ae43e12e
clarified signature: avoid deprecated URL constructors;
 wenzelm parents: 
78956diff
changeset | 40 |       if (url.getProtocol == "file" && Url.is_wellformed_file(name)) {
 | 
| 64775 | 41 | standard_path(Url.parse_file(name)) | 
| 79044 
8cc1ae43e12e
clarified signature: avoid deprecated URL constructors;
 wenzelm parents: 
78956diff
changeset | 42 | } | 
| 60992 | 43 | else name | 
| 44 | } | |
| 79045 
24d04dd5bf01
more robust exception handling (amending 8cc1ae43e12e);
 wenzelm parents: 
79044diff
changeset | 45 |     catch { case exn: Throwable if Url.is_malformed(exn) => standard_path(name) }
 | 
| 60992 | 46 | |
| 47 | ||
| 48 | /* platform path (Windows or Posix) */ | |
| 49 | ||
| 50 | def platform_path(standard_path: String): String = | |
| 73911 | 51 | isabelle.setup.Environment.platform_path(standard_path) | 
| 60992 | 52 | |
| 53 | def platform_path(path: Path): String = platform_path(standard_path(path)) | |
| 60988 | 54 | def platform_file(path: Path): JFile = new JFile(platform_path(path)) | 
| 55 | ||
| 60992 | 56 | |
| 76884 | 57 | /* symbolic path representation, e.g. "~~/src/Pure/ROOT.ML" */ | 
| 58 | ||
| 59 |   def symbolic_path(path: Path): String = {
 | |
| 77218 | 60 |     val directories = space_explode(':', Isabelle_System.getenv("ISABELLE_DIRECTORIES")).reverse
 | 
| 76884 | 61 | val full_name = standard_path(path) | 
| 62 | directories.view.flatMap(a => | |
| 63 |       try {
 | |
| 64 | val b = standard_path(Path.explode(a)) | |
| 65 | if (full_name == b) Some(a) | |
| 66 |         else {
 | |
| 67 |           Library.try_unprefix(b + "/", full_name) match {
 | |
| 68 | case Some(name) => Some(a + "/" + name) | |
| 69 | case None => None | |
| 70 | } | |
| 71 | } | |
| 72 |       } catch { case ERROR(_) => None }).headOption.getOrElse(path.implode)
 | |
| 73 | } | |
| 74 | ||
| 75 | ||
| 66232 | 76 | /* platform files */ | 
| 77 | ||
| 78 | def absolute(file: JFile): JFile = file.toPath.toAbsolutePath.normalize.toFile | |
| 79 | def canonical(file: JFile): JFile = file.getCanonicalFile | |
| 80 | ||
| 81 | def path(file: JFile): Path = Path.explode(standard_path(file)) | |
| 76546 | 82 | def path(java_path: JPath): Path = path(java_path.toFile) | 
| 83 | ||
| 75701 | 84 | def uri(file: JFile): URI = file.toURI | 
| 85 | def uri(path: Path): URI = path.file.toURI | |
| 86 | ||
| 79659 
a4118f530263
clarified signature: avoid ill-defined type java.net.URL;
 wenzelm parents: 
79045diff
changeset | 87 | def url(file: JFile): Url = Url(uri(file)) | 
| 
a4118f530263
clarified signature: avoid ill-defined type java.net.URL;
 wenzelm parents: 
79045diff
changeset | 88 | def url(path: Path): Url = url(path.file) | 
| 75701 | 89 | |
| 66232 | 90 | |
| 75906 
2167b9e3157a
clarified signature: support for adhoc file types;
 wenzelm parents: 
75825diff
changeset | 91 | /* adhoc file types */ | 
| 
2167b9e3157a
clarified signature: support for adhoc file types;
 wenzelm parents: 
75825diff
changeset | 92 | |
| 
2167b9e3157a
clarified signature: support for adhoc file types;
 wenzelm parents: 
75825diff
changeset | 93 |   def is_ML(s: String): Boolean = s.endsWith(".ML")
 | 
| 
2167b9e3157a
clarified signature: support for adhoc file types;
 wenzelm parents: 
75825diff
changeset | 94 |   def is_bib(s: String): Boolean = s.endsWith(".bib")
 | 
| 
2167b9e3157a
clarified signature: support for adhoc file types;
 wenzelm parents: 
75825diff
changeset | 95 |   def is_dll(s: String): Boolean = s.endsWith(".dll")
 | 
| 
2167b9e3157a
clarified signature: support for adhoc file types;
 wenzelm parents: 
75825diff
changeset | 96 |   def is_exe(s: String): Boolean = s.endsWith(".exe")
 | 
| 
2167b9e3157a
clarified signature: support for adhoc file types;
 wenzelm parents: 
75825diff
changeset | 97 |   def is_gz(s: String): Boolean = s.endsWith(".gz")
 | 
| 
2167b9e3157a
clarified signature: support for adhoc file types;
 wenzelm parents: 
75825diff
changeset | 98 |   def is_html(s: String): Boolean = s.endsWith(".html")
 | 
| 
2167b9e3157a
clarified signature: support for adhoc file types;
 wenzelm parents: 
75825diff
changeset | 99 |   def is_jar(s: String): Boolean = s.endsWith(".jar")
 | 
| 
2167b9e3157a
clarified signature: support for adhoc file types;
 wenzelm parents: 
75825diff
changeset | 100 |   def is_java(s: String): Boolean = s.endsWith(".java")
 | 
| 
2167b9e3157a
clarified signature: support for adhoc file types;
 wenzelm parents: 
75825diff
changeset | 101 |   def is_node(s: String): Boolean = s.endsWith(".node")
 | 
| 
2167b9e3157a
clarified signature: support for adhoc file types;
 wenzelm parents: 
75825diff
changeset | 102 |   def is_pdf(s: String): Boolean = s.endsWith(".pdf")
 | 
| 
2167b9e3157a
clarified signature: support for adhoc file types;
 wenzelm parents: 
75825diff
changeset | 103 |   def is_png(s: String): Boolean = s.endsWith(".png")
 | 
| 76540 
83de6e9ae983
clarified signature: prefer Scala functions instead of shell scripts;
 wenzelm parents: 
76533diff
changeset | 104 |   def is_tar_bz2(s: String): Boolean = s.endsWith(".tar.bz2")
 | 
| 76533 | 105 |   def is_tar_gz(s: String): Boolean = s.endsWith(".tar.gz")
 | 
| 76540 
83de6e9ae983
clarified signature: prefer Scala functions instead of shell scripts;
 wenzelm parents: 
76533diff
changeset | 106 |   def is_tgz(s: String): Boolean = s.endsWith(".tgz")
 | 
| 75906 
2167b9e3157a
clarified signature: support for adhoc file types;
 wenzelm parents: 
75825diff
changeset | 107 |   def is_thy(s: String): Boolean = s.endsWith(".thy")
 | 
| 
2167b9e3157a
clarified signature: support for adhoc file types;
 wenzelm parents: 
75825diff
changeset | 108 |   def is_xz(s: String): Boolean = s.endsWith(".xz")
 | 
| 
2167b9e3157a
clarified signature: support for adhoc file types;
 wenzelm parents: 
75825diff
changeset | 109 |   def is_zip(s: String): Boolean = s.endsWith(".zip")
 | 
| 76348 | 110 |   def is_zst(s: String): Boolean = s.endsWith(".zst")
 | 
| 75906 
2167b9e3157a
clarified signature: support for adhoc file types;
 wenzelm parents: 
75825diff
changeset | 111 | |
| 
2167b9e3157a
clarified signature: support for adhoc file types;
 wenzelm parents: 
75825diff
changeset | 112 |   def is_backup(s: String): Boolean = s.endsWith("~") || s.endsWith(".orig")
 | 
| 
2167b9e3157a
clarified signature: support for adhoc file types;
 wenzelm parents: 
75825diff
changeset | 113 | |
| 
2167b9e3157a
clarified signature: support for adhoc file types;
 wenzelm parents: 
75825diff
changeset | 114 | |
| 66693 | 115 | /* relative paths */ | 
| 116 | ||
| 75393 | 117 |   def relative_path(base: Path, other: Path): Option[Path] = {
 | 
| 73945 | 118 | val base_path = base.java_path | 
| 119 | val other_path = other.java_path | |
| 66693 | 120 | if (other_path.startsWith(base_path)) | 
| 121 | Some(path(base_path.relativize(other_path).toFile)) | |
| 122 | else None | |
| 123 | } | |
| 124 | ||
| 125 | ||
| 62545 
8ebffdaf2ce2
Bash.process always uses a closed script instead of an open argument list, for extra robustness on Windows, where quoting is not well-defined;
 wenzelm parents: 
62544diff
changeset | 126 | /* bash path */ | 
| 60992 | 127 | |
| 64304 | 128 | def bash_path(path: Path): String = Bash.string(standard_path(path)) | 
| 129 | def bash_path(file: JFile): String = Bash.string(standard_path(file)) | |
| 60988 | 130 | |
| 72036 | 131 | def bash_platform_path(path: Path): String = Bash.string(platform_path(path)) | 
| 132 | ||
| 60988 | 133 | |
| 62829 
4141c2a8458b
clarified Isabelle tool wrapper: bash, Scala, no perl, no ML;
 wenzelm parents: 
62704diff
changeset | 134 | /* directory content */ | 
| 
4141c2a8458b
clarified Isabelle tool wrapper: bash, Scala, no perl, no ML;
 wenzelm parents: 
62704diff
changeset | 135 | |
| 75393 | 136 |   def read_dir(dir: Path): List[String] = {
 | 
| 69300 
8b6ab9989bcd
is_file/is_dir/read_dir: more uniform treatment of errors and boundary cases, notably for symlinks in ssh;
 wenzelm parents: 
69299diff
changeset | 137 |     if (!dir.is_dir) error("No such directory: " + dir.toString)
 | 
| 62829 
4141c2a8458b
clarified Isabelle tool wrapper: bash, Scala, no perl, no ML;
 wenzelm parents: 
62704diff
changeset | 138 | val files = dir.file.listFiles | 
| 
4141c2a8458b
clarified Isabelle tool wrapper: bash, Scala, no perl, no ML;
 wenzelm parents: 
62704diff
changeset | 139 | if (files == null) Nil | 
| 69427 
ff2f39a221d4
clarified operations: uniform sorting of results;
 wenzelm parents: 
69405diff
changeset | 140 | else files.toList.map(_.getName).sorted | 
| 62829 
4141c2a8458b
clarified Isabelle tool wrapper: bash, Scala, no perl, no ML;
 wenzelm parents: 
62704diff
changeset | 141 | } | 
| 48613 
232652ac346e
clarified directory content operations (similar to ML version);
 wenzelm parents: 
48550diff
changeset | 142 | |
| 76529 | 143 | def get_entry( | 
| 144 | dir: Path, | |
| 145 | pred: Path => Boolean = _ => true, | |
| 146 | title: String = "" | |
| 147 | ): Path = | |
| 148 |     read_dir(dir).filter(name => pred(dir + Path.basic(name))) match {
 | |
| 149 | case List(entry) => dir + Path.basic(entry) | |
| 150 | case bad => | |
| 151 |         error("Bad directory content in " + (if (title.nonEmpty) title else dir.toString) +
 | |
| 152 | "\nexpected a single entry, but found" + | |
| 153 | (if (bad.isEmpty) " nothing" | |
| 154 |            else bad.sorted.map(quote).mkString(":\n  ", "\n  ", "")))
 | |
| 72442 | 155 | } | 
| 156 | ||
| 76529 | 157 | def get_file(dir: Path, title: String = ""): Path = | 
| 158 | get_entry(dir, pred = _.is_file, title = title) | |
| 159 | ||
| 160 | def get_dir(dir: Path, title: String = ""): Path = | |
| 161 | get_entry(dir, pred = _.is_dir, title = title) | |
| 162 | ||
| 64932 | 163 | def find_files( | 
| 164 | start: JFile, | |
| 165 | pred: JFile => Boolean = _ => true, | |
| 69293 
72a9860f8602
clarified find_files: follow links by default, e.g. relevant for "~/cronjob/log";
 wenzelm parents: 
66693diff
changeset | 166 | include_dirs: Boolean = false, | 
| 75393 | 167 | follow_links: Boolean = false | 
| 168 |   ): List[JFile] = {
 | |
| 62443 
133f65ac17e5
just one File.find_files, based on Java 7 Files operations;
 wenzelm parents: 
62294diff
changeset | 169 | val result = new mutable.ListBuffer[JFile] | 
| 73340 | 170 | def check(file: JFile): Unit = if (pred(file)) result += file | 
| 62443 
133f65ac17e5
just one File.find_files, based on Java 7 Files operations;
 wenzelm parents: 
62294diff
changeset | 171 | |
| 64932 | 172 | if (start.isFile) check(start) | 
| 62443 
133f65ac17e5
just one File.find_files, based on Java 7 Files operations;
 wenzelm parents: 
62294diff
changeset | 173 |     else if (start.isDirectory) {
 | 
| 69293 
72a9860f8602
clarified find_files: follow links by default, e.g. relevant for "~/cronjob/log";
 wenzelm parents: 
66693diff
changeset | 174 | val options = | 
| 
72a9860f8602
clarified find_files: follow links by default, e.g. relevant for "~/cronjob/log";
 wenzelm parents: 
66693diff
changeset | 175 | if (follow_links) EnumSet.of(FileVisitOption.FOLLOW_LINKS) | 
| 
72a9860f8602
clarified find_files: follow links by default, e.g. relevant for "~/cronjob/log";
 wenzelm parents: 
66693diff
changeset | 176 | else EnumSet.noneOf(classOf[FileVisitOption]) | 
| 78243 | 177 | Files.walkFileTree(start.toPath, options, Int.MaxValue, | 
| 62443 
133f65ac17e5
just one File.find_files, based on Java 7 Files operations;
 wenzelm parents: 
62294diff
changeset | 178 |         new SimpleFileVisitor[JPath] {
 | 
| 75393 | 179 | override def preVisitDirectory( | 
| 180 | path: JPath, | |
| 181 | attrs: BasicFileAttributes | |
| 182 |           ): FileVisitResult = {
 | |
| 64932 | 183 | if (include_dirs) check(path.toFile) | 
| 184 | FileVisitResult.CONTINUE | |
| 185 | } | |
| 75393 | 186 | override def visitFile( | 
| 187 | path: JPath, | |
| 188 | attrs: BasicFileAttributes | |
| 189 |           ): FileVisitResult = {
 | |
| 69301 | 190 | val file = path.toFile | 
| 191 | if (include_dirs || !file.isDirectory) check(file) | |
| 62443 
133f65ac17e5
just one File.find_files, based on Java 7 Files operations;
 wenzelm parents: 
62294diff
changeset | 192 | FileVisitResult.CONTINUE | 
| 
133f65ac17e5
just one File.find_files, based on Java 7 Files operations;
 wenzelm parents: 
62294diff
changeset | 193 | } | 
| 
133f65ac17e5
just one File.find_files, based on Java 7 Files operations;
 wenzelm parents: 
62294diff
changeset | 194 | } | 
| 
133f65ac17e5
just one File.find_files, based on Java 7 Files operations;
 wenzelm parents: 
62294diff
changeset | 195 | ) | 
| 
133f65ac17e5
just one File.find_files, based on Java 7 Files operations;
 wenzelm parents: 
62294diff
changeset | 196 | } | 
| 
133f65ac17e5
just one File.find_files, based on Java 7 Files operations;
 wenzelm parents: 
62294diff
changeset | 197 | |
| 
133f65ac17e5
just one File.find_files, based on Java 7 Files operations;
 wenzelm parents: 
62294diff
changeset | 198 | result.toList | 
| 48613 
232652ac346e
clarified directory content operations (similar to ML version);
 wenzelm parents: 
48550diff
changeset | 199 | } | 
| 
232652ac346e
clarified directory content operations (similar to ML version);
 wenzelm parents: 
48550diff
changeset | 200 | |
| 
232652ac346e
clarified directory content operations (similar to ML version);
 wenzelm parents: 
48550diff
changeset | 201 | |
| 48411 
5b3440850d36
more abstract file system operations in Scala, corresponding to ML version;
 wenzelm parents: diff
changeset | 202 | /* read */ | 
| 
5b3440850d36
more abstract file system operations in Scala, corresponding to ML version;
 wenzelm parents: diff
changeset | 203 | |
| 65589 
f70c617e9c26
more robust treatment of non-UTF8 text files (cf. 3ed43cfc8b14), notably old log files in ISO-8859-15;
 wenzelm parents: 
64934diff
changeset | 204 | def read(file: JFile): String = Bytes.read(file).text | 
| 48913 
f686cb016c0c
more direct File.read_bytes -- avoid cumulative copying of StringBuilder;
 wenzelm parents: 
48613diff
changeset | 205 | def read(path: Path): String = read(path.file) | 
| 
f686cb016c0c
more direct File.read_bytes -- avoid cumulative copying of StringBuilder;
 wenzelm parents: 
48613diff
changeset | 206 | |
| 80481 
0e2b09fef3d2
more uniform Bytes.read_stream vs. File.read_stream;
 wenzelm parents: 
80441diff
changeset | 207 | def read_stream(stream: InputStream): String = Bytes.read_stream(stream).text | 
| 50684 
12b7e0b4a66e
support File.read_gzip as well, in accordance to File.write_gzip;
 wenzelm parents: 
50203diff
changeset | 208 | |
| 
12b7e0b4a66e
support File.read_gzip as well, in accordance to File.write_gzip;
 wenzelm parents: 
50203diff
changeset | 209 | def read_gzip(file: JFile): String = | 
| 
12b7e0b4a66e
support File.read_gzip as well, in accordance to File.write_gzip;
 wenzelm parents: 
50203diff
changeset | 210 | read_stream(new GZIPInputStream(new BufferedInputStream(new FileInputStream(file)))) | 
| 64000 | 211 | def read_gzip(path: Path): String = read_gzip(path.file) | 
| 51504 | 212 | |
| 64002 | 213 | def read_xz(file: JFile): String = | 
| 76353 | 214 | read_stream(new xz.XZInputStream(new BufferedInputStream(new FileInputStream(file)))) | 
| 64000 | 215 | def read_xz(path: Path): String = read_xz(path.file) | 
| 48411 
5b3440850d36
more abstract file system operations in Scala, corresponding to ML version;
 wenzelm parents: diff
changeset | 216 | |
| 76348 | 217 |   def read_zstd(file: JFile): String = {
 | 
| 76349 
b4daf7577ca0
clarified Zstd.init(): avoid accidential com.github.luben.zstd.util.Native.load() operation;
 wenzelm parents: 
76348diff
changeset | 218 | Zstd.init() | 
| 76353 | 219 | read_stream(new zstd.ZstdInputStream(new BufferedInputStream(new FileInputStream(file)))) | 
| 76348 | 220 | } | 
| 221 | def read_zstd(path: Path): String = read_zstd(path.file) | |
| 222 | ||
| 48411 
5b3440850d36
more abstract file system operations in Scala, corresponding to ML version;
 wenzelm parents: diff
changeset | 223 | |
| 50845 | 224 | /* read lines */ | 
| 225 | ||
| 75393 | 226 |   def read_line(reader: BufferedReader): Option[String] = {
 | 
| 69487 | 227 | val line = | 
| 228 |       try { reader.readLine}
 | |
| 229 |       catch { case _: IOException => null }
 | |
| 72698 | 230 | Option(line).map(Library.trim_line) | 
| 69487 | 231 | } | 
| 232 | ||
| 75393 | 233 |   def read_lines(reader: BufferedReader, progress: String => Unit): List[String] = {
 | 
| 50845 | 234 | val result = new mutable.ListBuffer[String] | 
| 69487 | 235 | var line: Option[String] = None | 
| 236 |     while ({ line = read_line(reader); line.isDefined }) {
 | |
| 237 | progress(line.get) | |
| 238 | result += line.get | |
| 50845 | 239 | } | 
| 73367 | 240 | reader.close() | 
| 50845 | 241 | result.toList | 
| 242 | } | |
| 243 | ||
| 244 | ||
| 79980 | 245 | /* read properties */ | 
| 246 | ||
| 247 |   def read_props(path: Path): JProperties = {
 | |
| 248 | val props = new JProperties | |
| 249 | props.load(Files.newBufferedReader(path.java_path)) | |
| 250 | props | |
| 251 | } | |
| 252 | ||
| 253 | ||
| 48411 
5b3440850d36
more abstract file system operations in Scala, corresponding to ML version;
 wenzelm parents: diff
changeset | 254 | /* write */ | 
| 
5b3440850d36
more abstract file system operations in Scala, corresponding to ML version;
 wenzelm parents: diff
changeset | 255 | |
| 71534 | 256 | def writer(file: JFile): BufferedWriter = | 
| 257 | new BufferedWriter(new OutputStreamWriter(new FileOutputStream(file), UTF8.charset)) | |
| 258 | ||
| 73340 | 259 | def write_file( | 
| 75393 | 260 | file: JFile, | 
| 261 | text: String, | |
| 262 | make_stream: OutputStream => OutputStream | |
| 263 |   ): Unit = {
 | |
| 51504 | 264 | val stream = make_stream(new FileOutputStream(file)) | 
| 69393 
ed0824ef337e
static type for Library.using: avoid Java 11 warnings on "Illegal reflective access";
 wenzelm parents: 
69301diff
changeset | 265 | using(new BufferedWriter(new OutputStreamWriter(stream, UTF8.charset)))(_.append(text)) | 
| 48411 
5b3440850d36
more abstract file system operations in Scala, corresponding to ML version;
 wenzelm parents: diff
changeset | 266 | } | 
| 
5b3440850d36
more abstract file system operations in Scala, corresponding to ML version;
 wenzelm parents: diff
changeset | 267 | |
| 73574 
12b3f78dde61
clarified signature: avoid overlap of String vs. Bytes (both are CharSequence);
 wenzelm parents: 
73367diff
changeset | 268 | def write(file: JFile, text: String): Unit = write_file(file, text, s => s) | 
| 
12b3f78dde61
clarified signature: avoid overlap of String vs. Bytes (both are CharSequence);
 wenzelm parents: 
73367diff
changeset | 269 | def write(path: Path, text: String): Unit = write(path.file, text) | 
| 48411 
5b3440850d36
more abstract file system operations in Scala, corresponding to ML version;
 wenzelm parents: diff
changeset | 270 | |
| 73574 
12b3f78dde61
clarified signature: avoid overlap of String vs. Bytes (both are CharSequence);
 wenzelm parents: 
73367diff
changeset | 271 | def write_gzip(file: JFile, text: String): Unit = | 
| 51504 | 272 | write_file(file, text, (s: OutputStream) => new GZIPOutputStream(new BufferedOutputStream(s))) | 
| 73574 
12b3f78dde61
clarified signature: avoid overlap of String vs. Bytes (both are CharSequence);
 wenzelm parents: 
73367diff
changeset | 273 | def write_gzip(path: Path, text: String): Unit = write_gzip(path.file, text) | 
| 48411 
5b3440850d36
more abstract file system operations in Scala, corresponding to ML version;
 wenzelm parents: diff
changeset | 274 | |
| 76351 
2cee31cd92f0
generic support for XZ and Zstd compression in Isabelle/Scala;
 wenzelm parents: 
76349diff
changeset | 275 | def write_xz(file: JFile, text: String, options: Compress.Options_XZ): Unit = | 
| 76353 | 276 | File.write_file(file, text, | 
| 277 | s => new xz.XZOutputStream(new BufferedOutputStream(s), options.make)) | |
| 76351 
2cee31cd92f0
generic support for XZ and Zstd compression in Isabelle/Scala;
 wenzelm parents: 
76349diff
changeset | 278 | def write_xz(file: JFile, text: String): Unit = write_xz(file, text, Compress.Options_XZ()) | 
| 
2cee31cd92f0
generic support for XZ and Zstd compression in Isabelle/Scala;
 wenzelm parents: 
76349diff
changeset | 279 | def write_xz(path: Path, text: String, options: Compress.Options_XZ): Unit = | 
| 64002 | 280 | write_xz(path.file, text, options) | 
| 76351 
2cee31cd92f0
generic support for XZ and Zstd compression in Isabelle/Scala;
 wenzelm parents: 
76349diff
changeset | 281 | def write_xz(path: Path, text: String): Unit = write_xz(path, text, Compress.Options_XZ()) | 
| 64000 | 282 | |
| 76351 
2cee31cd92f0
generic support for XZ and Zstd compression in Isabelle/Scala;
 wenzelm parents: 
76349diff
changeset | 283 |   def write_zstd(file: JFile, text: String, options: Compress.Options_Zstd): Unit = {
 | 
| 76349 
b4daf7577ca0
clarified Zstd.init(): avoid accidential com.github.luben.zstd.util.Native.load() operation;
 wenzelm parents: 
76348diff
changeset | 284 | Zstd.init() | 
| 76353 | 285 | File.write_file(file, text, | 
| 286 | s => new zstd.ZstdOutputStream(new BufferedOutputStream(s), options.level)) | |
| 76348 | 287 | } | 
| 76351 
2cee31cd92f0
generic support for XZ and Zstd compression in Isabelle/Scala;
 wenzelm parents: 
76349diff
changeset | 288 | def write_zstd(file: JFile, text: String): Unit = | 
| 
2cee31cd92f0
generic support for XZ and Zstd compression in Isabelle/Scala;
 wenzelm parents: 
76349diff
changeset | 289 | write_zstd(file, text, Compress.Options_Zstd()) | 
| 
2cee31cd92f0
generic support for XZ and Zstd compression in Isabelle/Scala;
 wenzelm parents: 
76349diff
changeset | 290 | def write_zstd(path: Path, text: String, options: Compress.Options_Zstd): Unit = | 
| 
2cee31cd92f0
generic support for XZ and Zstd compression in Isabelle/Scala;
 wenzelm parents: 
76349diff
changeset | 291 | write_zstd(path.file, text, options) | 
| 
2cee31cd92f0
generic support for XZ and Zstd compression in Isabelle/Scala;
 wenzelm parents: 
76349diff
changeset | 292 | def write_zstd(path: Path, text: String): Unit = | 
| 
2cee31cd92f0
generic support for XZ and Zstd compression in Isabelle/Scala;
 wenzelm parents: 
76349diff
changeset | 293 | write_zstd(path, text, Compress.Options_Zstd()) | 
| 76348 | 294 | |
| 75393 | 295 |   def write_backup(path: Path, text: String): Unit = {
 | 
| 73317 | 296 | if (path.is_file) Isabelle_System.move_file(path, path.backup) | 
| 62444 | 297 | write(path, text) | 
| 53336 | 298 | } | 
| 299 | ||
| 75393 | 300 |   def write_backup2(path: Path, text: String): Unit = {
 | 
| 73317 | 301 | if (path.is_file) Isabelle_System.move_file(path, path.backup2) | 
| 62444 | 302 | write(path, text) | 
| 58610 | 303 | } | 
| 304 | ||
| 48411 
5b3440850d36
more abstract file system operations in Scala, corresponding to ML version;
 wenzelm parents: diff
changeset | 305 | |
| 62703 | 306 | /* append */ | 
| 307 | ||
| 73574 
12b3f78dde61
clarified signature: avoid overlap of String vs. Bytes (both are CharSequence);
 wenzelm parents: 
73367diff
changeset | 308 | def append(file: JFile, text: String): Unit = | 
| 73627 | 309 | Files.write(file.toPath, UTF8.bytes(text), | 
| 62703 | 310 | StandardOpenOption.APPEND, StandardOpenOption.CREATE) | 
| 311 | ||
| 73574 
12b3f78dde61
clarified signature: avoid overlap of String vs. Bytes (both are CharSequence);
 wenzelm parents: 
73367diff
changeset | 312 | def append(path: Path, text: String): Unit = append(path.file, text) | 
| 62703 | 313 | |
| 314 | ||
| 75206 | 315 | /* change */ | 
| 316 | ||
| 75393 | 317 | def change( | 
| 318 | path: Path, | |
| 319 | init: Boolean = false, | |
| 320 | strict: Boolean = false | |
| 321 |   )(f: String => String): Unit = {
 | |
| 75208 | 322 | if (!path.is_file && init) write(path, "") | 
| 75206 | 323 | val x = read(path) | 
| 324 | val y = f(x) | |
| 325 | if (x != y) write(path, y) | |
| 75213 | 326 |     else if (strict) error("Unchanged file: " + path)
 | 
| 75206 | 327 | } | 
| 328 | ||
| 75213 | 329 | def change_lines(path: Path, init: Boolean = false, strict: Boolean = false)( | 
| 330 | f: List[String] => List[String]): Unit = | |
| 331 | change(path, init = init, strict = strict)(text => cat_lines(f(split_lines(text)))) | |
| 75206 | 332 | |
| 333 | ||
| 64213 | 334 | /* eq */ | 
| 48411 
5b3440850d36
more abstract file system operations in Scala, corresponding to ML version;
 wenzelm parents: diff
changeset | 335 | |
| 
5b3440850d36
more abstract file system operations in Scala, corresponding to ML version;
 wenzelm parents: diff
changeset | 336 | def eq(file1: JFile, file2: JFile): Boolean = | 
| 73318 | 337 |     try { Files.isSameFile(file1.toPath, file2.toPath) }
 | 
| 49673 
2a088cff1e7b
more robust File.eq, and thus File.copy of "~~/lib/logo/isabelle.gif";
 wenzelm parents: 
49610diff
changeset | 338 |     catch { case ERROR(_) => false }
 | 
| 48411 
5b3440850d36
more abstract file system operations in Scala, corresponding to ML version;
 wenzelm parents: diff
changeset | 339 | |
| 64213 | 340 | def eq(path1: Path, path2: Path): Boolean = eq(path1.file, path2.file) | 
| 341 | ||
| 342 | ||
| 64934 | 343 | /* eq_content */ | 
| 344 | ||
| 345 | def eq_content(file1: JFile, file2: JFile): Boolean = | |
| 346 | if (eq(file1, file2)) true | |
| 347 | else if (file1.length != file2.length) false | |
| 80378 
ab4badc7db7f
more efficient equals: avoid somewhat slow sha1_digest (see also 29b761e290c5, 306f273c91ec);
 wenzelm parents: 
80365diff
changeset | 348 | else Bytes.read(file1) == Bytes.read(file2) | 
| 64934 | 349 | |
| 350 | def eq_content(path1: Path, path2: Path): Boolean = eq_content(path1.file, path2.file) | |
| 351 | ||
| 352 | ||
| 69405 
22428643351f
more direct File.executable operation: avoid external process (on Unix);
 wenzelm parents: 
69402diff
changeset | 353 | /* permissions */ | 
| 
22428643351f
more direct File.executable operation: avoid external process (on Unix);
 wenzelm parents: 
69402diff
changeset | 354 | |
| 78169 
5ad1ae8626de
minor performance tuning: avoid external process;
 wenzelm parents: 
78161diff
changeset | 355 | private val restrict_perms: List[PosixFilePermission] = | 
| 
5ad1ae8626de
minor performance tuning: avoid external process;
 wenzelm parents: 
78161diff
changeset | 356 | List( | 
| 
5ad1ae8626de
minor performance tuning: avoid external process;
 wenzelm parents: 
78161diff
changeset | 357 | PosixFilePermission.GROUP_READ, | 
| 
5ad1ae8626de
minor performance tuning: avoid external process;
 wenzelm parents: 
78161diff
changeset | 358 | PosixFilePermission.GROUP_WRITE, | 
| 
5ad1ae8626de
minor performance tuning: avoid external process;
 wenzelm parents: 
78161diff
changeset | 359 | PosixFilePermission.GROUP_EXECUTE, | 
| 
5ad1ae8626de
minor performance tuning: avoid external process;
 wenzelm parents: 
78161diff
changeset | 360 | PosixFilePermission.OTHERS_READ, | 
| 
5ad1ae8626de
minor performance tuning: avoid external process;
 wenzelm parents: 
78161diff
changeset | 361 | PosixFilePermission.OTHERS_WRITE, | 
| 
5ad1ae8626de
minor performance tuning: avoid external process;
 wenzelm parents: 
78161diff
changeset | 362 | PosixFilePermission.OTHERS_EXECUTE) | 
| 
5ad1ae8626de
minor performance tuning: avoid external process;
 wenzelm parents: 
78161diff
changeset | 363 | |
| 
5ad1ae8626de
minor performance tuning: avoid external process;
 wenzelm parents: 
78161diff
changeset | 364 | def restrict(path: Path): Unit = | 
| 
5ad1ae8626de
minor performance tuning: avoid external process;
 wenzelm parents: 
78161diff
changeset | 365 |     if (Platform.is_windows) Isabelle_System.chmod("g-rwx,o-rwx", path)
 | 
| 
5ad1ae8626de
minor performance tuning: avoid external process;
 wenzelm parents: 
78161diff
changeset | 366 |     else {
 | 
| 
5ad1ae8626de
minor performance tuning: avoid external process;
 wenzelm parents: 
78161diff
changeset | 367 | val perms = Files.getPosixFilePermissions(path.java_path) | 
| 
5ad1ae8626de
minor performance tuning: avoid external process;
 wenzelm parents: 
78161diff
changeset | 368 | var perms_changed = false | 
| 
5ad1ae8626de
minor performance tuning: avoid external process;
 wenzelm parents: 
78161diff
changeset | 369 |       for (p <- restrict_perms if perms.contains(p)) {
 | 
| 
5ad1ae8626de
minor performance tuning: avoid external process;
 wenzelm parents: 
78161diff
changeset | 370 | perms.remove(p) | 
| 
5ad1ae8626de
minor performance tuning: avoid external process;
 wenzelm parents: 
78161diff
changeset | 371 | perms_changed = true | 
| 
5ad1ae8626de
minor performance tuning: avoid external process;
 wenzelm parents: 
78161diff
changeset | 372 | } | 
| 
5ad1ae8626de
minor performance tuning: avoid external process;
 wenzelm parents: 
78161diff
changeset | 373 | if (perms_changed) Files.setPosixFilePermissions(path.java_path, perms) | 
| 
5ad1ae8626de
minor performance tuning: avoid external process;
 wenzelm parents: 
78161diff
changeset | 374 | } | 
| 78161 | 375 | |
| 75393 | 376 |   def is_executable(path: Path): Boolean = {
 | 
| 69788 | 377 |     if (Platform.is_windows) Isabelle_System.bash("test -x " + bash_path(path)).check.ok
 | 
| 378 | else path.file.canExecute | |
| 379 | } | |
| 380 | ||
| 78298 
3b0f8f1010f2
clarified signature, with subtle change of semantics (amending 8b5a2e4b16d4);
 wenzelm parents: 
78243diff
changeset | 381 |   def set_executable(path: Path, reset: Boolean = false): Unit = {
 | 
| 
3b0f8f1010f2
clarified signature, with subtle change of semantics (amending 8b5a2e4b16d4);
 wenzelm parents: 
78243diff
changeset | 382 | if (Platform.is_windows) Isabelle_System.chmod(if (reset) "a-x" else "a+x", path) | 
| 
3b0f8f1010f2
clarified signature, with subtle change of semantics (amending 8b5a2e4b16d4);
 wenzelm parents: 
78243diff
changeset | 383 | else path.file.setExecutable(!reset, false) | 
| 69405 
22428643351f
more direct File.executable operation: avoid external process (on Unix);
 wenzelm parents: 
69402diff
changeset | 384 | } | 
| 74811 | 385 | |
| 386 | ||
| 387 | /* content */ | |
| 388 | ||
| 75825 | 389 | def content(path: Path, content: Bytes): Content = new Content(path, content) | 
| 390 | def content(path: Path, content: String): Content = new Content(path, Bytes(content)) | |
| 75824 | 391 | def content(path: Path, content: XML.Body): Content_XML = new Content_XML(path, content) | 
| 74811 | 392 | |
| 75825 | 393 |   final class Content private[File](val path: Path, val content: Bytes) {
 | 
| 75677 | 394 | override def toString: String = path.toString | 
| 74811 | 395 | |
| 77852 
df35b5b7b6a4
more direct hg_sync init via ssh (see also 721b3278c8e4);
 wenzelm parents: 
77218diff
changeset | 396 |     def write(dir: Path, ssh: SSH.System = SSH.Local): Unit = {
 | 
| 75508 | 397 | val full_path = dir + path | 
| 77852 
df35b5b7b6a4
more direct hg_sync init via ssh (see also 721b3278c8e4);
 wenzelm parents: 
77218diff
changeset | 398 | ssh.make_directory(ssh.expand_path(full_path).dir) | 
| 
df35b5b7b6a4
more direct hg_sync init via ssh (see also 721b3278c8e4);
 wenzelm parents: 
77218diff
changeset | 399 | ssh.write_bytes(full_path, content) | 
| 75508 | 400 | } | 
| 74811 | 401 | } | 
| 402 | ||
| 75676 | 403 |   final class Content_XML private[File](val path: Path, val content: XML.Body) {
 | 
| 75823 
6eb8d6cdb686
proper toString for Content_XML, which is not covered by trait Content;
 wenzelm parents: 
75701diff
changeset | 404 | override def toString: String = path.toString | 
| 
6eb8d6cdb686
proper toString for Content_XML, which is not covered by trait Content;
 wenzelm parents: 
75701diff
changeset | 405 | |
| 75825 | 406 | def output(out: XML.Body => String): Content = new Content(path, Bytes(out(content))) | 
| 74811 | 407 | } | 
| 77109 
e3a2b3536030
prefer typed bytes count, but retain toString of original Long for robustness of Java/Scala string composition;
 wenzelm parents: 
76884diff
changeset | 408 | |
| 
e3a2b3536030
prefer typed bytes count, but retain toString of original Long for robustness of Java/Scala string composition;
 wenzelm parents: 
76884diff
changeset | 409 | |
| 78956 | 410 | /* strict file size */ | 
| 77109 
e3a2b3536030
prefer typed bytes count, but retain toString of original Long for robustness of Java/Scala string composition;
 wenzelm parents: 
76884diff
changeset | 411 | |
| 78956 | 412 | def size(path: Path): Long = path.check_file.file.length | 
| 413 | def space(path: Path): Space = Space.bytes(size(path)) | |
| 48411 
5b3440850d36
more abstract file system operations in Scala, corresponding to ML version;
 wenzelm parents: diff
changeset | 414 | } |