src/Pure/Admin/build_log.scala
author wenzelm
Sat May 06 20:52:23 2017 +0200 (2017-05-06)
changeset 65748 1f4a80e80c88
parent 65741 cf42659364c9
child 65775 123f2c0995b8
permissions -rw-r--r--
tuned signature;
wenzelm@64160
     1
/*  Title:      Pure/Admin/build_log.scala
wenzelm@64045
     2
    Author:     Makarius
wenzelm@64045
     3
wenzelm@65608
     4
Management of build log files and database storage.
wenzelm@64045
     5
*/
wenzelm@64045
     6
wenzelm@64045
     7
package isabelle
wenzelm@64045
     8
wenzelm@64045
     9
wenzelm@64100
    10
import java.io.{File => JFile}
wenzelm@64100
    11
import java.time.ZoneId
wenzelm@64110
    12
import java.time.format.{DateTimeFormatter, DateTimeParseException}
wenzelm@64096
    13
import java.util.Locale
wenzelm@64061
    14
wenzelm@65600
    15
import scala.collection.immutable.SortedMap
wenzelm@64054
    16
import scala.collection.mutable
wenzelm@64054
    17
import scala.util.matching.Regex
wenzelm@64054
    18
wenzelm@64054
    19
wenzelm@64045
    20
object Build_Log
wenzelm@64045
    21
{
wenzelm@64298
    22
  /** content **/
wenzelm@64101
    23
wenzelm@64298
    24
  /* properties */
wenzelm@64150
    25
wenzelm@64298
    26
  object Prop
wenzelm@64298
    27
  {
wenzelm@65624
    28
    val build_tags = SQL.Column.string("build_tags")  // lines
wenzelm@65624
    29
    val build_args = SQL.Column.string("build_args")  // lines
wenzelm@65591
    30
    val build_group_id = SQL.Column.string("build_group_id")
wenzelm@65591
    31
    val build_id = SQL.Column.string("build_id")
wenzelm@65591
    32
    val build_engine = SQL.Column.string("build_engine")
wenzelm@65591
    33
    val build_host = SQL.Column.string("build_host")
wenzelm@65591
    34
    val build_start = SQL.Column.date("build_start")
wenzelm@65591
    35
    val build_end = SQL.Column.date("build_end")
wenzelm@65591
    36
    val isabelle_version = SQL.Column.string("isabelle_version")
wenzelm@65591
    37
    val afp_version = SQL.Column.string("afp_version")
wenzelm@65591
    38
wenzelm@65611
    39
    val all_props: List[SQL.Column] =
wenzelm@65591
    40
      List(build_tags, build_args, build_group_id, build_id, build_engine,
wenzelm@65591
    41
        build_host, build_start, build_end, isabelle_version, afp_version)
wenzelm@64298
    42
  }
wenzelm@64150
    43
wenzelm@64150
    44
wenzelm@64298
    45
  /* settings */
wenzelm@64080
    46
wenzelm@64081
    47
  object Settings
wenzelm@64081
    48
  {
wenzelm@65611
    49
    val ISABELLE_BUILD_OPTIONS = SQL.Column.string("ISABELLE_BUILD_OPTIONS")
wenzelm@65611
    50
    val ML_PLATFORM = SQL.Column.string("ML_PLATFORM")
wenzelm@65611
    51
    val ML_HOME = SQL.Column.string("ML_HOME")
wenzelm@65611
    52
    val ML_SYSTEM = SQL.Column.string("ML_SYSTEM")
wenzelm@65611
    53
    val ML_OPTIONS = SQL.Column.string("ML_OPTIONS")
wenzelm@65611
    54
wenzelm@65611
    55
    val ml_settings = List(ML_PLATFORM, ML_HOME, ML_SYSTEM, ML_OPTIONS)
wenzelm@65611
    56
    val all_settings = ISABELLE_BUILD_OPTIONS :: ml_settings
wenzelm@64081
    57
wenzelm@64081
    58
    type Entry = (String, String)
wenzelm@64081
    59
    type T = List[Entry]
wenzelm@64080
    60
wenzelm@64081
    61
    object Entry
wenzelm@64081
    62
    {
wenzelm@64081
    63
      def unapply(s: String): Option[Entry] =
wenzelm@64081
    64
        s.indexOf('=') match {
wenzelm@64081
    65
          case -1 => None
wenzelm@64081
    66
          case i =>
wenzelm@64081
    67
            val a = s.substring(0, i)
wenzelm@64081
    68
            val b = Library.perhaps_unquote(s.substring(i + 1))
wenzelm@64081
    69
            Some((a, b))
wenzelm@64081
    70
        }
wenzelm@64081
    71
      def apply(a: String, b: String): String = a + "=" + quote(b)
wenzelm@64081
    72
      def getenv(a: String): String = apply(a, Isabelle_System.getenv(a))
wenzelm@64081
    73
    }
wenzelm@64080
    74
wenzelm@64081
    75
    def show(): String =
wenzelm@64081
    76
      cat_lines(
wenzelm@65611
    77
        List(Entry.getenv(ISABELLE_BUILD_OPTIONS.name), "") :::
wenzelm@65611
    78
        ml_settings.map(c => Entry.getenv(c.name)))
wenzelm@64080
    79
  }
wenzelm@64080
    80
wenzelm@64080
    81
wenzelm@64298
    82
  /* file names */
wenzelm@64298
    83
wenzelm@64298
    84
  def log_date(date: Date): String =
wenzelm@64298
    85
    String.format(Locale.ROOT, "%s.%05d",
wenzelm@64298
    86
      DateTimeFormatter.ofPattern("yyyy-MM-dd").format(date.rep),
wenzelm@64298
    87
      new java.lang.Long((date.time - date.midnight.time).ms / 1000))
wenzelm@64298
    88
wenzelm@64298
    89
  def log_subdir(date: Date): Path =
wenzelm@64298
    90
    Path.explode("log") + Path.explode(date.rep.getYear.toString)
wenzelm@64298
    91
wenzelm@64298
    92
  def log_filename(engine: String, date: Date, more: List[String] = Nil): Path =
wenzelm@64298
    93
    Path.explode((engine :: log_date(date) :: more).mkString("", "_", ".log"))
wenzelm@64298
    94
wenzelm@64298
    95
wenzelm@64100
    96
wenzelm@64062
    97
  /** log file **/
wenzelm@64045
    98
wenzelm@64155
    99
  def print_date(date: Date): String = Log_File.Date_Format(date)
wenzelm@64155
   100
wenzelm@64062
   101
  object Log_File
wenzelm@64062
   102
  {
wenzelm@65607
   103
    /* log file */
wenzelm@65607
   104
wenzelm@65609
   105
    def plain_name(name: String): String =
wenzelm@65609
   106
    {
wenzelm@65609
   107
      List(".log", ".log.gz", ".log.xz", ".gz", ".xz").find(name.endsWith(_)) match {
wenzelm@65609
   108
        case Some(s) => Library.try_unsuffix(s, name).get
wenzelm@65609
   109
        case None => name
wenzelm@65609
   110
      }
wenzelm@65609
   111
    }
wenzelm@65609
   112
wenzelm@64062
   113
    def apply(name: String, lines: List[String]): Log_File =
wenzelm@65609
   114
      new Log_File(plain_name(name), lines)
wenzelm@64062
   115
wenzelm@64062
   116
    def apply(name: String, text: String): Log_File =
wenzelm@64063
   117
      Log_File(name, Library.trim_split_lines(text))
wenzelm@64090
   118
wenzelm@64090
   119
    def apply(file: JFile): Log_File =
wenzelm@64090
   120
    {
wenzelm@64090
   121
      val name = file.getName
wenzelm@65609
   122
      val text =
wenzelm@65609
   123
        if (name.endsWith(".gz")) File.read_gzip(file)
wenzelm@65609
   124
        else if (name.endsWith(".xz")) File.read_xz(file)
wenzelm@65609
   125
        else File.read(file)
wenzelm@65609
   126
      apply(name, text)
wenzelm@64090
   127
    }
wenzelm@64090
   128
wenzelm@64090
   129
    def apply(path: Path): Log_File = apply(path.file)
wenzelm@64101
   130
wenzelm@64110
   131
wenzelm@65607
   132
    /* log file collections */
wenzelm@65607
   133
wenzelm@65607
   134
    def is_log(file: JFile,
wenzelm@65607
   135
      prefixes: List[String] =
wenzelm@65664
   136
        List(Build_History.log_prefix, Identify.log_prefix, Isatest.log_prefix,
wenzelm@65664
   137
          AFP_Test.log_prefix, Jenkins.log_prefix),
wenzelm@65609
   138
      suffixes: List[String] = List(".log", ".log.gz", ".log.xz")): Boolean =
wenzelm@65607
   139
    {
wenzelm@65607
   140
      val name = file.getName
wenzelm@65639
   141
wenzelm@65607
   142
      prefixes.exists(name.startsWith(_)) &&
wenzelm@65639
   143
      suffixes.exists(name.endsWith(_)) &&
wenzelm@65639
   144
      name != "isatest.log" &&
wenzelm@65639
   145
      name != "afp-test.log" &&
wenzelm@65639
   146
      name != "main.log"
wenzelm@65607
   147
    }
wenzelm@65607
   148
wenzelm@65607
   149
    def find_files(dirs: Iterable[Path]): List[JFile] =
wenzelm@65607
   150
      dirs.iterator.flatMap(dir => File.find_files(dir.file, is_log(_))).toList
wenzelm@65607
   151
wenzelm@65607
   152
wenzelm@64110
   153
    /* date format */
wenzelm@64110
   154
wenzelm@64101
   155
    val Date_Format =
wenzelm@64101
   156
    {
wenzelm@64101
   157
      val fmts =
wenzelm@64101
   158
        Date.Formatter.variants(
wenzelm@64116
   159
          List("EEE MMM d HH:mm:ss O yyyy", "EEE MMM d HH:mm:ss VV yyyy"),
wenzelm@64104
   160
          List(Locale.ENGLISH, Locale.GERMAN)) :::
wenzelm@64110
   161
        List(
wenzelm@64110
   162
          DateTimeFormatter.RFC_1123_DATE_TIME,
wenzelm@64110
   163
          Date.Formatter.pattern("EEE MMM d HH:mm:ss yyyy").withZone(ZoneId.of("Europe/Berlin")))
wenzelm@64101
   164
wenzelm@64104
   165
      def tune_timezone(s: String): String =
wenzelm@64104
   166
        s match {
wenzelm@64101
   167
          case "CET" | "MET" => "GMT+1"
wenzelm@64101
   168
          case "CEST" | "MEST" => "GMT+2"
wenzelm@64104
   169
          case "EST" => "Europe/Berlin"
wenzelm@64104
   170
          case _ => s
wenzelm@64104
   171
        }
wenzelm@64104
   172
      def tune_weekday(s: String): String =
wenzelm@64104
   173
        s match {
wenzelm@64104
   174
          case "Die" => "Di"
wenzelm@64104
   175
          case "Mit" => "Mi"
wenzelm@64104
   176
          case "Don" => "Do"
wenzelm@64104
   177
          case "Fre" => "Fr"
wenzelm@64104
   178
          case "Sam" => "Sa"
wenzelm@64104
   179
          case "Son" => "So"
wenzelm@64104
   180
          case _ => s
wenzelm@64104
   181
        }
wenzelm@64104
   182
wenzelm@64104
   183
      def tune(s: String): String =
wenzelm@64104
   184
        Word.implode(
wenzelm@64104
   185
          Word.explode(s) match {
wenzelm@64104
   186
            case a :: "M\uFFFDr" :: bs => tune_weekday(a) :: "Mär" :: bs.map(tune_timezone(_))
wenzelm@64104
   187
            case a :: bs => tune_weekday(a) :: bs.map(tune_timezone(_))
wenzelm@64104
   188
            case Nil => Nil
wenzelm@64104
   189
          }
wenzelm@64104
   190
        )
wenzelm@64101
   191
wenzelm@64101
   192
      Date.Format.make(fmts, tune)
wenzelm@64101
   193
    }
wenzelm@64117
   194
wenzelm@64117
   195
wenzelm@64117
   196
    /* inlined content */
wenzelm@64117
   197
wenzelm@64119
   198
    def print_props(marker: String, props: Properties.T): String =
wenzelm@65624
   199
      marker + YXML.string_of_body(XML.Encode.properties(Properties.encode_lines(props)))
wenzelm@64102
   200
  }
wenzelm@64102
   201
wenzelm@64102
   202
  class Log_File private(val name: String, val lines: List[String])
wenzelm@64102
   203
  {
wenzelm@64102
   204
    log_file =>
wenzelm@64102
   205
wenzelm@64102
   206
    override def toString: String = name
wenzelm@64102
   207
wenzelm@64102
   208
    def text: String = cat_lines(lines)
wenzelm@64102
   209
wenzelm@64102
   210
    def err(msg: String): Nothing =
wenzelm@64102
   211
      error("Error in log file " + quote(name) + ": " + msg)
wenzelm@64102
   212
wenzelm@64102
   213
wenzelm@64102
   214
    /* date format */
wenzelm@64101
   215
wenzelm@64101
   216
    object Strict_Date
wenzelm@64101
   217
    {
wenzelm@64101
   218
      def unapply(s: String): Some[Date] =
wenzelm@64102
   219
        try { Some(Log_File.Date_Format.parse(s)) }
wenzelm@64101
   220
        catch { case exn: DateTimeParseException => log_file.err(exn.getMessage) }
wenzelm@64101
   221
    }
wenzelm@64101
   222
wenzelm@64101
   223
wenzelm@64062
   224
    /* inlined content */
wenzelm@64062
   225
wenzelm@64062
   226
    def find[A](f: String => Option[A]): Option[A] =
wenzelm@64062
   227
      lines.iterator.map(f).find(_.isDefined).map(_.get)
wenzelm@64062
   228
wenzelm@64196
   229
    def find_line(marker: String): Option[String] =
wenzelm@64196
   230
      find(Library.try_unprefix(marker, _))
wenzelm@64196
   231
wenzelm@65684
   232
    def find_match(regexes: List[Regex]): Option[String] =
wenzelm@65684
   233
      regexes match {
wenzelm@65684
   234
        case Nil => None
wenzelm@65684
   235
        case regex :: rest =>
wenzelm@65684
   236
          lines.iterator.map(regex.unapplySeq(_)).find(res => res.isDefined && res.get.length == 1).
wenzelm@65684
   237
            map(res => res.get.head) orElse find_match(rest)
wenzelm@65684
   238
      }
wenzelm@64062
   239
wenzelm@64062
   240
wenzelm@64062
   241
    /* settings */
wenzelm@64062
   242
wenzelm@64091
   243
    def get_setting(a: String): Option[Settings.Entry] =
wenzelm@64091
   244
      lines.find(_.startsWith(a + "=")) match {
wenzelm@64091
   245
        case Some(line) => Settings.Entry.unapply(line)
wenzelm@64091
   246
        case None => None
wenzelm@64091
   247
      }
wenzelm@64045
   248
wenzelm@65611
   249
    def get_all_settings: Settings.T =
wenzelm@65611
   250
      for { c <- Settings.all_settings; entry <- get_setting(c.name) }
wenzelm@65611
   251
      yield entry
wenzelm@64062
   252
wenzelm@64062
   253
wenzelm@64062
   254
    /* properties (YXML) */
wenzelm@64062
   255
wenzelm@64062
   256
    val xml_cache = new XML.Cache()
wenzelm@64062
   257
wenzelm@64062
   258
    def parse_props(text: String): Properties.T =
wenzelm@65624
   259
      xml_cache.props(Properties.decode_lines(XML.Decode.properties(YXML.parse_body(text))))
wenzelm@64062
   260
wenzelm@64119
   261
    def filter_props(marker: String): List[Properties.T] =
wenzelm@64196
   262
      for {
wenzelm@64196
   263
        line <- lines
wenzelm@64196
   264
        s <- Library.try_unprefix(marker, line)
wenzelm@64196
   265
        if YXML.detect(s)
wenzelm@64196
   266
      } yield parse_props(s)
wenzelm@64062
   267
wenzelm@64119
   268
    def find_props(marker: String): Option[Properties.T] =
wenzelm@64196
   269
      find_line(marker) match {
wenzelm@64196
   270
        case Some(text) if YXML.detect(text) => Some(parse_props(text))
wenzelm@64196
   271
        case _ => None
wenzelm@64196
   272
      }
wenzelm@64062
   273
wenzelm@64062
   274
wenzelm@64062
   275
    /* parse various formats */
wenzelm@64062
   276
wenzelm@64105
   277
    def parse_meta_info(): Meta_Info = Build_Log.parse_meta_info(log_file)
wenzelm@64105
   278
wenzelm@65646
   279
    def parse_build_info(ml_statistics: Boolean = false): Build_Info =
wenzelm@65646
   280
      Build_Log.parse_build_info(log_file, ml_statistics)
wenzelm@64105
   281
wenzelm@64082
   282
    def parse_session_info(
wenzelm@64082
   283
        command_timings: Boolean = false,
wenzelm@64082
   284
        ml_statistics: Boolean = false,
wenzelm@64082
   285
        task_statistics: Boolean = false): Session_Info =
wenzelm@65318
   286
      Build_Log.parse_session_info(log_file, command_timings, ml_statistics, task_statistics)
wenzelm@64045
   287
  }
wenzelm@64045
   288
wenzelm@64045
   289
wenzelm@64098
   290
wenzelm@65276
   291
  /** digested meta info: produced by Admin/build_history in log.xz file **/
wenzelm@64045
   292
wenzelm@64108
   293
  object Meta_Info
wenzelm@64099
   294
  {
wenzelm@64108
   295
    val empty: Meta_Info = Meta_Info(Nil, Nil)
wenzelm@64099
   296
  }
wenzelm@64098
   297
wenzelm@65611
   298
  sealed case class Meta_Info(props: Properties.T, settings: Settings.T)
wenzelm@64103
   299
  {
wenzelm@64103
   300
    def is_empty: Boolean = props.isEmpty && settings.isEmpty
wenzelm@65599
   301
wenzelm@65611
   302
    def get(c: SQL.Column): Option[String] =
wenzelm@65611
   303
      Properties.get(props, c.name) orElse
wenzelm@65611
   304
      Properties.get(settings, c.name)
wenzelm@65611
   305
wenzelm@65611
   306
    def get_date(c: SQL.Column): Option[Date] =
wenzelm@65611
   307
      get(c).map(Log_File.Date_Format.parse(_))
wenzelm@64103
   308
  }
wenzelm@64061
   309
wenzelm@65625
   310
  object Identify
wenzelm@65625
   311
  {
wenzelm@65625
   312
    val log_prefix = "isabelle_identify_"
wenzelm@65674
   313
wenzelm@65674
   314
    def engine(log_file: Log_File): String =
wenzelm@65674
   315
      if (log_file.name.startsWith(Jenkins.log_prefix)) "jenkins_identify"
wenzelm@65674
   316
      else "identify"
wenzelm@65674
   317
wenzelm@65674
   318
    def content(date: Date, isabelle_version: Option[String], afp_version: Option[String]): String =
wenzelm@65674
   319
      terminate_lines(
wenzelm@65674
   320
        List("isabelle_identify: " + Build_Log.print_date(date), "") :::
wenzelm@65674
   321
        isabelle_version.map("Isabelle version: " + _).toList :::
wenzelm@65674
   322
        afp_version.map("AFP version: " + _).toList)
wenzelm@65674
   323
wenzelm@65625
   324
    val Start = new Regex("""^isabelle_identify: (.+)$""")
wenzelm@65625
   325
    val No_End = new Regex("""$.""")
wenzelm@65684
   326
    val Isabelle_Version = List(new Regex("""^Isabelle version: (\S+)$"""))
wenzelm@65684
   327
    val AFP_Version = List(new Regex("""^AFP version: (\S+)$"""))
wenzelm@65625
   328
  }
wenzelm@65625
   329
wenzelm@64095
   330
  object Isatest
wenzelm@64095
   331
  {
wenzelm@65588
   332
    val log_prefix = "isatest-makeall-"
wenzelm@64108
   333
    val engine = "isatest"
wenzelm@64109
   334
    val Start = new Regex("""^------------------- starting test --- (.+) --- (.+)$""")
wenzelm@64109
   335
    val End = new Regex("""^------------------- test (?:successful|FAILED) --- (.+) --- .*$""")
wenzelm@65684
   336
    val Isabelle_Version = List(new Regex("""^Isabelle version: (\S+)$"""))
wenzelm@64095
   337
  }
wenzelm@64095
   338
wenzelm@64109
   339
  object AFP_Test
wenzelm@64061
   340
  {
wenzelm@65588
   341
    val log_prefix = "afp-test-devel-"
wenzelm@64108
   342
    val engine = "afp-test"
wenzelm@64109
   343
    val Start = new Regex("""^Start test(?: for .+)? at ([^,]+), (.*)$""")
wenzelm@64109
   344
    val Start_Old = new Regex("""^Start test(?: for .+)? at ([^,]+)$""")
wenzelm@64109
   345
    val End = new Regex("""^End test on (.+), .+, elapsed time:.*$""")
wenzelm@65684
   346
    val Isabelle_Version = List(new Regex("""^Isabelle version: .* -- hg id (\S+)$"""))
wenzelm@65684
   347
    val AFP_Version = List(new Regex("""^AFP version: .* -- hg id (\S+)$"""))
wenzelm@64104
   348
    val Bad_Init = new Regex("""^cp:.*: Disc quota exceeded$""")
wenzelm@64061
   349
  }
wenzelm@64061
   350
wenzelm@64110
   351
  object Jenkins
wenzelm@64110
   352
  {
wenzelm@65664
   353
    val log_prefix = "jenkins_"
wenzelm@64110
   354
    val engine = "jenkins"
wenzelm@65663
   355
    val Host = new Regex("""^Building remotely on (\S+) \((\S+)\).*$""")
wenzelm@65665
   356
    val Start = new Regex("""^(?:Started by an SCM change|Started from command line by admin|).*$""")
wenzelm@64110
   357
    val Start_Date = new Regex("""^Build started at (.+)$""")
wenzelm@64110
   358
    val No_End = new Regex("""$.""")
wenzelm@65674
   359
    val Isabelle_Version =
wenzelm@65684
   360
      List(new Regex("""^(?:Build for Isabelle id|Isabelle id) (\w+).*$"""),
wenzelm@65685
   361
        new Regex("""^ISABELLE_CI_REPO_ID="(\w+)".*$"""),
wenzelm@65685
   362
        new Regex("""^(\w{12}) tip.*$"""))
wenzelm@65674
   363
    val AFP_Version =
wenzelm@65684
   364
      List(new Regex("""^(?:Build for AFP id|AFP id) (\w+).*$"""),
wenzelm@65684
   365
        new Regex("""^ISABELLE_CI_AFP_ID="(\w+)".*$"""))
wenzelm@64110
   366
    val CONFIGURATION = "=== CONFIGURATION ==="
wenzelm@64110
   367
    val BUILD = "=== BUILD ==="
wenzelm@64110
   368
  }
wenzelm@64110
   369
wenzelm@64105
   370
  private def parse_meta_info(log_file: Log_File): Meta_Info =
wenzelm@64061
   371
  {
wenzelm@64108
   372
    def parse(engine: String, host: String, start: Date,
wenzelm@65684
   373
      End: Regex, Isabelle_Version: List[Regex], AFP_Version: List[Regex]): Meta_Info =
wenzelm@64091
   374
    {
wenzelm@64296
   375
      val build_id =
wenzelm@64296
   376
      {
wenzelm@65714
   377
        val prefix = proper_string(host) orElse proper_string(engine) getOrElse "build"
wenzelm@65714
   378
        prefix + ":" + start.time.ms
wenzelm@64296
   379
      }
wenzelm@65591
   380
      val build_engine = if (engine == "") Nil else List(Prop.build_engine.name -> engine)
wenzelm@65591
   381
      val build_host = if (host == "") Nil else List(Prop.build_host.name -> host)
wenzelm@64108
   382
wenzelm@65599
   383
      val start_date = List(Prop.build_start.name -> print_date(start))
wenzelm@64091
   384
      val end_date =
wenzelm@64091
   385
        log_file.lines.last match {
wenzelm@64109
   386
          case End(log_file.Strict_Date(end_date)) =>
wenzelm@65599
   387
            List(Prop.build_end.name -> print_date(end_date))
wenzelm@64091
   388
          case _ => Nil
wenzelm@64091
   389
        }
wenzelm@64091
   390
wenzelm@64091
   391
      val isabelle_version =
wenzelm@65591
   392
        log_file.find_match(Isabelle_Version).map(Prop.isabelle_version.name -> _)
wenzelm@64091
   393
      val afp_version =
wenzelm@65591
   394
        log_file.find_match(AFP_Version).map(Prop.afp_version.name -> _)
wenzelm@64062
   395
wenzelm@65591
   396
      Meta_Info((Prop.build_id.name -> build_id) :: build_engine ::: build_host :::
wenzelm@64108
   397
          start_date ::: end_date ::: isabelle_version.toList ::: afp_version.toList,
wenzelm@65611
   398
        log_file.get_all_settings)
wenzelm@64091
   399
    }
wenzelm@64091
   400
wenzelm@64091
   401
    log_file.lines match {
wenzelm@64119
   402
      case line :: _ if line.startsWith(Build_History.META_INFO_MARKER) =>
wenzelm@64119
   403
        Meta_Info(log_file.find_props(Build_History.META_INFO_MARKER).get,
wenzelm@65611
   404
          log_file.get_all_settings)
wenzelm@64117
   405
wenzelm@65625
   406
      case Identify.Start(log_file.Strict_Date(start)) :: _ =>
wenzelm@65674
   407
        parse(Identify.engine(log_file), "", start, Identify.No_End,
wenzelm@65625
   408
          Identify.Isabelle_Version, Identify.AFP_Version)
wenzelm@65625
   409
wenzelm@64109
   410
      case Isatest.Start(log_file.Strict_Date(start), host) :: _ =>
wenzelm@64109
   411
        parse(Isatest.engine, host, start, Isatest.End,
wenzelm@65684
   412
          Isatest.Isabelle_Version, Nil)
wenzelm@64099
   413
wenzelm@64109
   414
      case AFP_Test.Start(log_file.Strict_Date(start), host) :: _ =>
wenzelm@64109
   415
        parse(AFP_Test.engine, host, start, AFP_Test.End,
wenzelm@64109
   416
          AFP_Test.Isabelle_Version, AFP_Test.AFP_Version)
wenzelm@64099
   417
wenzelm@64109
   418
      case AFP_Test.Start_Old(log_file.Strict_Date(start)) :: _ =>
wenzelm@64109
   419
        parse(AFP_Test.engine, "", start, AFP_Test.End,
wenzelm@64109
   420
          AFP_Test.Isabelle_Version, AFP_Test.AFP_Version)
wenzelm@64099
   421
wenzelm@65665
   422
      case Jenkins.Start() :: _ =>
wenzelm@64110
   423
        log_file.lines.dropWhile(_ != Jenkins.BUILD) match {
wenzelm@64110
   424
          case Jenkins.BUILD :: _ :: Jenkins.Start_Date(log_file.Strict_Date(start)) :: _ =>
wenzelm@65663
   425
            val host =
wenzelm@65663
   426
              log_file.lines.takeWhile(_ != Jenkins.CONFIGURATION).collectFirst({
wenzelm@65663
   427
                case Jenkins.Host(a, b) => a + "." + b
wenzelm@65663
   428
              }).getOrElse("")
wenzelm@65663
   429
            parse(Jenkins.engine, host, start.to(ZoneId.of("Europe/Berlin")), Jenkins.No_End,
wenzelm@64110
   430
              Jenkins.Isabelle_Version, Jenkins.AFP_Version)
wenzelm@64110
   431
          case _ => Meta_Info.empty
wenzelm@64110
   432
        }
wenzelm@64110
   433
wenzelm@64341
   434
      case line :: _ if line.startsWith("\u0000") => Meta_Info.empty
wenzelm@64109
   435
      case List(Isatest.End(_)) => Meta_Info.empty
wenzelm@64109
   436
      case _ :: AFP_Test.Bad_Init() :: _ => Meta_Info.empty
wenzelm@64105
   437
      case Nil => Meta_Info.empty
wenzelm@64104
   438
wenzelm@64110
   439
      case _ => log_file.err("cannot detect log file format")
wenzelm@64061
   440
    }
wenzelm@64061
   441
  }
wenzelm@64061
   442
wenzelm@64085
   443
wenzelm@64098
   444
wenzelm@65276
   445
  /** build info: toplevel output of isabelle build or Admin/build_history **/
wenzelm@64119
   446
wenzelm@64119
   447
  val ML_STATISTICS_MARKER = "\fML_statistics = "
wenzelm@64119
   448
  val SESSION_NAME = "session_name"
wenzelm@64085
   449
wenzelm@64061
   450
  object Session_Status extends Enumeration
wenzelm@64061
   451
  {
wenzelm@65633
   452
    val existing, finished, failed, cancelled = Value
wenzelm@64061
   453
  }
wenzelm@64061
   454
wenzelm@65643
   455
  object Session_Entry
wenzelm@65643
   456
  {
wenzelm@65643
   457
    val empty: Session_Entry = Session_Entry()
wenzelm@65643
   458
  }
wenzelm@65643
   459
wenzelm@64085
   460
  sealed case class Session_Entry(
wenzelm@65643
   461
    chapter: String = "",
wenzelm@65643
   462
    groups: List[String] = Nil,
wenzelm@65643
   463
    threads: Option[Int] = None,
wenzelm@65643
   464
    timing: Timing = Timing.zero,
wenzelm@65643
   465
    ml_timing: Timing = Timing.zero,
wenzelm@65643
   466
    heap_size: Option[Long] = None,
wenzelm@65643
   467
    status: Option[Session_Status.Value] = None,
wenzelm@65643
   468
    ml_statistics: List[Properties.T] = Nil)
wenzelm@64085
   469
  {
wenzelm@65631
   470
    def proper_chapter: Option[String] = if (chapter == "") None else Some(chapter)
wenzelm@65631
   471
    def proper_groups: Option[String] = if (groups.isEmpty) None else Some(cat_lines(groups))
wenzelm@65643
   472
    def finished: Boolean = status == Some(Session_Status.finished)
wenzelm@64085
   473
  }
wenzelm@64054
   474
wenzelm@64085
   475
  sealed case class Build_Info(sessions: Map[String, Session_Entry])
wenzelm@64054
   476
  {
wenzelm@64085
   477
    def session(name: String): Session_Entry = sessions(name)
wenzelm@64085
   478
    def get_session(name: String): Option[Session_Entry] = sessions.get(name)
wenzelm@64054
   479
wenzelm@64089
   480
    def get_default[A](name: String, f: Session_Entry => A, x: A): A =
wenzelm@64085
   481
      get_session(name) match {
wenzelm@64089
   482
        case Some(entry) => f(entry)
wenzelm@64089
   483
        case None => x
wenzelm@64085
   484
      }
wenzelm@64085
   485
wenzelm@64119
   486
    def finished_sessions: List[String] = sessions.keySet.iterator.filter(finished(_)).toList
wenzelm@64089
   487
    def finished(name: String): Boolean = get_default(name, _.finished, false)
wenzelm@64089
   488
    def timing(name: String): Timing = get_default(name, _.timing, Timing.zero)
wenzelm@64089
   489
    def ml_timing(name: String): Timing = get_default(name, _.ml_timing, Timing.zero)
wenzelm@65052
   490
    def ml_statistics(name: String): ML_Statistics =
wenzelm@65052
   491
      get_default(name, entry => ML_Statistics(name, entry.ml_statistics), ML_Statistics.empty)
wenzelm@64054
   492
  }
wenzelm@64054
   493
wenzelm@65646
   494
  private def parse_build_info(log_file: Log_File, parse_ml_statistics: Boolean): Build_Info =
wenzelm@64085
   495
  {
wenzelm@64085
   496
    object Chapter_Name
wenzelm@64085
   497
    {
wenzelm@64085
   498
      def unapply(s: String): Some[(String, String)] =
wenzelm@64085
   499
        space_explode('/', s) match {
wenzelm@64085
   500
          case List(chapter, name) => Some((chapter, name))
wenzelm@64085
   501
          case _ => Some(("", s))
wenzelm@64085
   502
        }
wenzelm@64085
   503
    }
wenzelm@64054
   504
wenzelm@64085
   505
    val Session_No_Groups = new Regex("""^Session (\S+)$""")
wenzelm@64085
   506
    val Session_Groups = new Regex("""^Session (\S+) \((.*)\)$""")
wenzelm@64085
   507
    val Session_Finished1 =
wenzelm@64085
   508
      new Regex("""^Finished (\S+) \((\d+):(\d+):(\d+) elapsed time, (\d+):(\d+):(\d+) cpu time.*$""")
wenzelm@64085
   509
    val Session_Finished2 =
wenzelm@64085
   510
      new Regex("""^Finished (\S+) \((\d+):(\d+):(\d+) elapsed time.*$""")
wenzelm@64085
   511
    val Session_Timing =
wenzelm@65679
   512
      new Regex("""^Timing (\S+) \((\d+) threads, (\d+\.\d+)s elapsed time, (\d+\.\d+)s cpu time, (\d+\.\d+)s GC time.*$""")
wenzelm@64086
   513
    val Session_Started = new Regex("""^(?:Running|Building) (\S+) \.\.\.$""")
wenzelm@64085
   514
    val Session_Failed = new Regex("""^(\S+) FAILED""")
wenzelm@64085
   515
    val Session_Cancelled = new Regex("""^(\S+) CANCELLED""")
wenzelm@64120
   516
    val Heap = new Regex("""^Heap (\S+) \((\d+) bytes\)$""")
wenzelm@64085
   517
wenzelm@64085
   518
    var chapter = Map.empty[String, String]
wenzelm@64085
   519
    var groups = Map.empty[String, List[String]]
wenzelm@64085
   520
    var threads = Map.empty[String, Int]
wenzelm@64054
   521
    var timing = Map.empty[String, Timing]
wenzelm@64085
   522
    var ml_timing = Map.empty[String, Timing]
wenzelm@64086
   523
    var started = Set.empty[String]
wenzelm@64085
   524
    var failed = Set.empty[String]
wenzelm@64085
   525
    var cancelled = Set.empty[String]
wenzelm@65627
   526
    var heap_sizes = Map.empty[String, Long]
wenzelm@64119
   527
    var ml_statistics = Map.empty[String, List[Properties.T]]
wenzelm@64119
   528
wenzelm@64085
   529
    def all_sessions: Set[String] =
wenzelm@64120
   530
      chapter.keySet ++ groups.keySet ++ threads.keySet ++ timing.keySet ++ ml_timing.keySet ++
wenzelm@65627
   531
      failed ++ cancelled ++ started ++ heap_sizes.keySet ++ ml_statistics.keySet
wenzelm@64085
   532
wenzelm@64054
   533
wenzelm@64062
   534
    for (line <- log_file.lines) {
wenzelm@64054
   535
      line match {
wenzelm@64085
   536
        case Session_No_Groups(Chapter_Name(chapt, name)) =>
wenzelm@64085
   537
          chapter += (name -> chapt)
wenzelm@64085
   538
          groups += (name -> Nil)
wenzelm@64119
   539
wenzelm@64085
   540
        case Session_Groups(Chapter_Name(chapt, name), grps) =>
wenzelm@64085
   541
          chapter += (name -> chapt)
wenzelm@64085
   542
          groups += (name -> Word.explode(grps))
wenzelm@64119
   543
wenzelm@64086
   544
        case Session_Started(name) =>
wenzelm@64086
   545
          started += name
wenzelm@64119
   546
wenzelm@64054
   547
        case Session_Finished1(name,
wenzelm@64054
   548
            Value.Int(e1), Value.Int(e2), Value.Int(e3),
wenzelm@64054
   549
            Value.Int(c1), Value.Int(c2), Value.Int(c3)) =>
wenzelm@64054
   550
          val elapsed = Time.hms(e1, e2, e3)
wenzelm@64054
   551
          val cpu = Time.hms(c1, c2, c3)
wenzelm@64085
   552
          timing += (name -> Timing(elapsed, cpu, Time.zero))
wenzelm@64119
   553
wenzelm@64054
   554
        case Session_Finished2(name,
wenzelm@64054
   555
            Value.Int(e1), Value.Int(e2), Value.Int(e3)) =>
wenzelm@64054
   556
          val elapsed = Time.hms(e1, e2, e3)
wenzelm@64085
   557
          timing += (name -> Timing(elapsed, Time.zero, Time.zero))
wenzelm@64119
   558
wenzelm@64054
   559
        case Session_Timing(name,
wenzelm@64054
   560
            Value.Int(t), Value.Double(e), Value.Double(c), Value.Double(g)) =>
wenzelm@64054
   561
          val elapsed = Time.seconds(e)
wenzelm@64054
   562
          val cpu = Time.seconds(c)
wenzelm@64054
   563
          val gc = Time.seconds(g)
wenzelm@64085
   564
          ml_timing += (name -> Timing(elapsed, cpu, gc))
wenzelm@64054
   565
          threads += (name -> t)
wenzelm@64119
   566
wenzelm@64120
   567
        case Heap(name, Value.Long(size)) =>
wenzelm@64120
   568
          heap_sizes += (name -> size)
wenzelm@64120
   569
wenzelm@65646
   570
        case _
wenzelm@65646
   571
        if parse_ml_statistics && line.startsWith(ML_STATISTICS_MARKER) && YXML.detect(line) =>
wenzelm@64119
   572
          val (name, props) =
wenzelm@64119
   573
            Library.try_unprefix(ML_STATISTICS_MARKER, line).map(log_file.parse_props(_)) match {
wenzelm@64119
   574
              case Some((SESSION_NAME, session_name) :: props) => (session_name, props)
wenzelm@64119
   575
              case _ => log_file.err("malformed ML_statistics " + quote(line))
wenzelm@64119
   576
            }
wenzelm@64120
   577
          ml_statistics += (name -> (props :: ml_statistics.getOrElse(name, Nil)))
wenzelm@64119
   578
wenzelm@64054
   579
        case _ =>
wenzelm@64054
   580
      }
wenzelm@64054
   581
    }
wenzelm@64054
   582
wenzelm@64085
   583
    val sessions =
wenzelm@64085
   584
      Map(
wenzelm@64085
   585
        (for (name <- all_sessions.toList) yield {
wenzelm@64085
   586
          val status =
wenzelm@65633
   587
            if (failed(name)) Session_Status.failed
wenzelm@65633
   588
            else if (cancelled(name)) Session_Status.cancelled
wenzelm@64086
   589
            else if (timing.isDefinedAt(name) || ml_timing.isDefinedAt(name))
wenzelm@65633
   590
              Session_Status.finished
wenzelm@65633
   591
            else if (started(name)) Session_Status.failed
wenzelm@65633
   592
            else Session_Status.existing
wenzelm@64085
   593
          val entry =
wenzelm@64085
   594
            Session_Entry(
wenzelm@65643
   595
              chapter = chapter.getOrElse(name, ""),
wenzelm@65643
   596
              groups = groups.getOrElse(name, Nil),
wenzelm@65643
   597
              threads = threads.get(name),
wenzelm@65643
   598
              timing = timing.getOrElse(name, Timing.zero),
wenzelm@65643
   599
              ml_timing = ml_timing.getOrElse(name, Timing.zero),
wenzelm@65643
   600
              heap_size = heap_sizes.get(name),
wenzelm@65643
   601
              status = Some(status),
wenzelm@65643
   602
              ml_statistics = ml_statistics.getOrElse(name, Nil).reverse)
wenzelm@64085
   603
          (name -> entry)
wenzelm@64085
   604
        }):_*)
wenzelm@64085
   605
    Build_Info(sessions)
wenzelm@64054
   606
  }
wenzelm@64099
   607
wenzelm@64099
   608
wenzelm@64099
   609
wenzelm@65276
   610
  /** session info: produced by isabelle build as session log.gz file **/
wenzelm@64099
   611
wenzelm@64099
   612
  sealed case class Session_Info(
wenzelm@64099
   613
    session_timing: Properties.T,
wenzelm@64099
   614
    command_timings: List[Properties.T],
wenzelm@64099
   615
    ml_statistics: List[Properties.T],
wenzelm@64099
   616
    task_statistics: List[Properties.T])
wenzelm@64099
   617
wenzelm@64099
   618
  private def parse_session_info(
wenzelm@64099
   619
    log_file: Log_File,
wenzelm@64099
   620
    command_timings: Boolean,
wenzelm@64099
   621
    ml_statistics: Boolean,
wenzelm@64099
   622
    task_statistics: Boolean): Session_Info =
wenzelm@64099
   623
  {
wenzelm@65290
   624
    Session_Info(
wenzelm@65290
   625
      session_timing = log_file.find_props("\fTiming = ") getOrElse Nil,
wenzelm@65290
   626
      command_timings = if (command_timings) log_file.filter_props("\fcommand_timing = ") else Nil,
wenzelm@65290
   627
      ml_statistics = if (ml_statistics) log_file.filter_props(ML_STATISTICS_MARKER) else Nil,
wenzelm@65290
   628
      task_statistics = if (task_statistics) log_file.filter_props("\ftask_statistics = ") else Nil)
wenzelm@64099
   629
  }
wenzelm@65595
   630
wenzelm@65595
   631
wenzelm@65595
   632
wenzelm@65595
   633
  /** persistent store **/
wenzelm@65595
   634
wenzelm@65694
   635
  /* SQL data model */
wenzelm@65694
   636
wenzelm@65694
   637
  object Data
wenzelm@65694
   638
  {
wenzelm@65702
   639
    def build_log_table(name: String, columns: List[SQL.Column], body: String = ""): SQL.Table =
wenzelm@65700
   640
      SQL.Table("isabelle_build_log_" + name, columns, body)
wenzelm@65700
   641
wenzelm@65700
   642
wenzelm@65694
   643
    /* main content */
wenzelm@65694
   644
wenzelm@65694
   645
    val log_name = SQL.Column.string("log_name", primary_key = true)
wenzelm@65694
   646
    val session_name = SQL.Column.string("session_name", primary_key = true)
wenzelm@65694
   647
    val chapter = SQL.Column.string("chapter")
wenzelm@65694
   648
    val groups = SQL.Column.string("groups")
wenzelm@65694
   649
    val threads = SQL.Column.int("threads")
wenzelm@65694
   650
    val timing_elapsed = SQL.Column.long("timing_elapsed")
wenzelm@65694
   651
    val timing_cpu = SQL.Column.long("timing_cpu")
wenzelm@65694
   652
    val timing_gc = SQL.Column.long("timing_gc")
wenzelm@65694
   653
    val timing_factor = SQL.Column.double("timing_factor")
wenzelm@65694
   654
    val ml_timing_elapsed = SQL.Column.long("ml_timing_elapsed")
wenzelm@65694
   655
    val ml_timing_cpu = SQL.Column.long("ml_timing_cpu")
wenzelm@65694
   656
    val ml_timing_gc = SQL.Column.long("ml_timing_gc")
wenzelm@65694
   657
    val ml_timing_factor = SQL.Column.double("ml_timing_factor")
wenzelm@65694
   658
    val heap_size = SQL.Column.long("heap_size")
wenzelm@65694
   659
    val status = SQL.Column.string("status")
wenzelm@65694
   660
    val ml_statistics = SQL.Column.bytes("ml_statistics")
wenzelm@65694
   661
wenzelm@65694
   662
    val meta_info_table =
wenzelm@65702
   663
      build_log_table("meta_info", log_name :: Prop.all_props ::: Settings.all_settings)
wenzelm@65694
   664
wenzelm@65694
   665
    val sessions_table =
wenzelm@65702
   666
      build_log_table("sessions",
wenzelm@65694
   667
        List(log_name, session_name, chapter, groups, threads, timing_elapsed, timing_cpu,
wenzelm@65694
   668
          timing_gc, timing_factor, ml_timing_elapsed, ml_timing_cpu, ml_timing_gc, ml_timing_factor,
wenzelm@65694
   669
          heap_size, status))
wenzelm@65694
   670
wenzelm@65694
   671
    val ml_statistics_table =
wenzelm@65702
   672
      build_log_table("ml_statistics", List(log_name, session_name, ml_statistics))
wenzelm@65694
   673
wenzelm@65694
   674
wenzelm@65705
   675
    /* earliest pull date for repository version (PostgreSQL queries) */
wenzelm@65694
   676
wenzelm@65694
   677
    val pull_date = SQL.Column.date("pull_date")
wenzelm@65694
   678
wenzelm@65722
   679
    val pull_date_table: SQL.Table =
wenzelm@65709
   680
    {
wenzelm@65709
   681
      val version = Prop.isabelle_version
wenzelm@65722
   682
      build_log_table("pull_date", List(version.copy(primary_key = true), pull_date),
wenzelm@65701
   683
        "SELECT " + version + ", min(" + Prop.build_start + ") AS " + pull_date +
wenzelm@65701
   684
        " FROM " + meta_info_table +
wenzelm@65701
   685
        " WHERE " + version + " IS NOT NULL AND " + Prop.build_start + " IS NOT NULL" +
wenzelm@65701
   686
        " GROUP BY " + version)
wenzelm@65709
   687
    }
wenzelm@65694
   688
wenzelm@65736
   689
    def recent_time(days: Int): SQL.Source =
wenzelm@65736
   690
      "now() - INTERVAL '" + days.max(0) + " days'"
wenzelm@65736
   691
wenzelm@65709
   692
    def recent_table(days: Int): SQL.Table =
wenzelm@65702
   693
    {
wenzelm@65722
   694
      val table = pull_date_table
wenzelm@65709
   695
      SQL.Table("recent", table.columns,
wenzelm@65736
   696
        table.select(table.columns, "WHERE " + pull_date(table) + " > " + recent_time(days)))
wenzelm@65702
   697
    }
wenzelm@65694
   698
wenzelm@65705
   699
    def select_recent(table: SQL.Table, columns: List[SQL.Column], days: Int,
wenzelm@65736
   700
      distinct: Boolean = false, pull_date: Boolean = false): SQL.Source =
wenzelm@65702
   701
    {
wenzelm@65709
   702
      val recent = recent_table(days)
wenzelm@65705
   703
      val columns1 = if (pull_date) columns ::: List(Data.pull_date(recent)) else columns
wenzelm@65729
   704
      table.select(columns1, distinct = distinct) + " INNER JOIN " + recent.query_name +
wenzelm@65702
   705
      " ON " + Prop.isabelle_version(table) + " = " + Prop.isabelle_version(recent)
wenzelm@65702
   706
    }
wenzelm@65724
   707
wenzelm@65724
   708
wenzelm@65724
   709
    /* universal view on main data */
wenzelm@65724
   710
wenzelm@65724
   711
    val universal_table: SQL.Table =
wenzelm@65724
   712
    {
wenzelm@65724
   713
      val table1 = meta_info_table
wenzelm@65724
   714
      val table2 = pull_date_table
wenzelm@65724
   715
      val table3 = sessions_table
wenzelm@65724
   716
wenzelm@65724
   717
      val aux_columns = log_name :: pull_date :: meta_info_table.columns.tail
wenzelm@65724
   718
      val aux_table = SQL.Table("aux", aux_columns,
wenzelm@65724
   719
        SQL.select(aux_columns.take(2) ::: aux_columns.drop(2). map(_.apply(table1))) +
wenzelm@65724
   720
          table1 + " LEFT OUTER JOIN " + table2 + " ON " +
wenzelm@65724
   721
          Prop.isabelle_version(table1) + " = " + Prop.isabelle_version(table2))
wenzelm@65724
   722
wenzelm@65724
   723
      val columns = aux_columns ::: sessions_table.columns.tail
wenzelm@65724
   724
      SQL.Table("isabelle_build_log", columns,
wenzelm@65724
   725
        {
wenzelm@65729
   726
          SQL.select(log_name(aux_table) :: columns.tail) + aux_table.query_name +
wenzelm@65724
   727
          " INNER JOIN " + table3 + " ON " + log_name(aux_table) + " = " + log_name(table3)
wenzelm@65724
   728
        })
wenzelm@65724
   729
    }
wenzelm@65694
   730
  }
wenzelm@65694
   731
wenzelm@65694
   732
wenzelm@65694
   733
  /* database access */
wenzelm@65694
   734
wenzelm@65595
   735
  def store(options: Options): Store = new Store(options)
wenzelm@65595
   736
wenzelm@65595
   737
  class Store private[Build_Log](options: Options) extends Properties.Store
wenzelm@65595
   738
  {
wenzelm@65595
   739
    def open_database(
wenzelm@65595
   740
      user: String = options.string("build_log_database_user"),
wenzelm@65595
   741
      password: String = options.string("build_log_database_password"),
wenzelm@65595
   742
      database: String = options.string("build_log_database_name"),
wenzelm@65595
   743
      host: String = options.string("build_log_database_host"),
wenzelm@65595
   744
      port: Int = options.int("build_log_database_port"),
wenzelm@65595
   745
      ssh_host: String = options.string("build_log_ssh_host"),
wenzelm@65595
   746
      ssh_user: String = options.string("build_log_ssh_user"),
wenzelm@65595
   747
      ssh_port: Int = options.int("build_log_ssh_port")): PostgreSQL.Database =
wenzelm@65595
   748
    {
wenzelm@65595
   749
      PostgreSQL.open_database(
wenzelm@65595
   750
        user = user, password = password, database = database, host = host, port = port,
wenzelm@65595
   751
        ssh =
wenzelm@65595
   752
          if (ssh_host == "") None
wenzelm@65636
   753
          else Some(SSH.init_context(options).open_session(ssh_host, ssh_user, port)),
wenzelm@65636
   754
        ssh_close = true)
wenzelm@65595
   755
    }
wenzelm@65599
   756
wenzelm@65723
   757
    def update_database(db: PostgreSQL.Database, dirs: List[Path], ml_statistics: Boolean = false)
wenzelm@65694
   758
    {
wenzelm@65694
   759
      write_info(db, Log_File.find_files(dirs), ml_statistics = ml_statistics)
wenzelm@65694
   760
wenzelm@65723
   761
      db.create_view(Data.pull_date_table)
wenzelm@65724
   762
      db.create_view(Data.universal_table)
wenzelm@65694
   763
    }
wenzelm@65694
   764
wenzelm@65723
   765
    def snapshot_database(db: PostgreSQL.Database, sqlite_database: Path,
wenzelm@65705
   766
      days: Int = 100, ml_statistics: Boolean = false)
wenzelm@65694
   767
    {
wenzelm@65694
   768
      Isabelle_System.mkdirs(sqlite_database.dir)
wenzelm@65694
   769
      sqlite_database.file.delete
wenzelm@65694
   770
wenzelm@65694
   771
      using(SQLite.open_database(sqlite_database))(db2 =>
wenzelm@65694
   772
      {
wenzelm@65694
   773
        db.transaction {
wenzelm@65694
   774
          db2.transaction {
wenzelm@65705
   775
            // main content
wenzelm@65705
   776
            db2.create_table(Data.meta_info_table)
wenzelm@65705
   777
            db2.create_table(Data.sessions_table)
wenzelm@65705
   778
            db2.create_table(Data.ml_statistics_table)
wenzelm@65705
   779
wenzelm@65705
   780
            val recent_log_names =
wenzelm@65705
   781
              db.using_statement(
wenzelm@65705
   782
                Data.select_recent(
wenzelm@65740
   783
                  Data.meta_info_table, List(Data.log_name), days, distinct = true))(stmt =>
wenzelm@65740
   784
                    stmt.execute_query().iterator(_.string(Data.log_name)).toList)
wenzelm@65705
   785
wenzelm@65705
   786
            for (log_name <- recent_log_names) {
wenzelm@65705
   787
              read_meta_info(db, log_name).foreach(meta_info =>
wenzelm@65705
   788
                update_meta_info(db2, log_name, meta_info))
wenzelm@65705
   789
wenzelm@65705
   790
              update_sessions(db2, log_name, read_build_info(db, log_name))
wenzelm@65705
   791
wenzelm@65705
   792
              if (ml_statistics)
wenzelm@65705
   793
                update_ml_statistics(db2, log_name, read_build_info(db, log_name))
wenzelm@65705
   794
            }
wenzelm@65705
   795
wenzelm@65705
   796
            // pull_date
wenzelm@65694
   797
            {
wenzelm@65722
   798
              val table = Data.pull_date_table
wenzelm@65694
   799
              db2.create_table(table)
wenzelm@65698
   800
              db2.using_statement(table.insert())(stmt2 =>
wenzelm@65694
   801
              {
wenzelm@65709
   802
                db.using_statement(Data.recent_table(days).query)(stmt =>
wenzelm@65694
   803
                {
wenzelm@65740
   804
                  val res = stmt.execute_query()
wenzelm@65740
   805
                  while (res.next()) {
wenzelm@65748
   806
                    for ((c, i) <- table.columns.zipWithIndex) {
wenzelm@65748
   807
                      stmt2.string(i + 1) = res.get_string(c)
wenzelm@65748
   808
                    }
wenzelm@65740
   809
                    stmt2.execute()
wenzelm@65694
   810
                  }
wenzelm@65694
   811
                })
wenzelm@65694
   812
              })
wenzelm@65709
   813
            }
wenzelm@65705
   814
wenzelm@65705
   815
            // full view
wenzelm@65724
   816
            db2.create_view(Data.universal_table)
wenzelm@65694
   817
          }
wenzelm@65694
   818
        }
wenzelm@65708
   819
        db2.rebuild
wenzelm@65694
   820
      })
wenzelm@65694
   821
    }
wenzelm@65694
   822
wenzelm@65688
   823
    def domain(db: SQL.Database, table: SQL.Table, column: SQL.Column): Set[String] =
wenzelm@65698
   824
      db.using_statement(table.select(List(column), distinct = true))(stmt =>
wenzelm@65740
   825
        stmt.execute_query().iterator(_.string(column)).toSet)
wenzelm@65688
   826
wenzelm@65705
   827
    def update_meta_info(db: SQL.Database, log_name: String, meta_info: Meta_Info)
wenzelm@65632
   828
    {
wenzelm@65694
   829
      val table = Data.meta_info_table
wenzelm@65703
   830
      db.using_statement(db.insert_permissive(table))(stmt =>
wenzelm@65703
   831
      {
wenzelm@65748
   832
        stmt.string(1) = log_name
wenzelm@65703
   833
        for ((c, i) <- table.columns.tail.zipWithIndex) {
wenzelm@65703
   834
          if (c.T == SQL.Type.Date)
wenzelm@65748
   835
            stmt.date(i + 2) = meta_info.get_date(c)
wenzelm@65703
   836
          else
wenzelm@65748
   837
            stmt.string(i + 2) = meta_info.get(c)
wenzelm@65703
   838
        }
wenzelm@65703
   839
        stmt.execute()
wenzelm@65703
   840
      })
wenzelm@65600
   841
    }
wenzelm@65600
   842
wenzelm@65705
   843
    def update_sessions(db: SQL.Database, log_name: String, build_info: Build_Info)
wenzelm@65599
   844
    {
wenzelm@65694
   845
      val table = Data.sessions_table
wenzelm@65703
   846
      db.using_statement(db.insert_permissive(table))(stmt =>
wenzelm@65703
   847
      {
wenzelm@65703
   848
        val entries_iterator =
wenzelm@65703
   849
          if (build_info.sessions.isEmpty) Iterator("" -> Session_Entry.empty)
wenzelm@65703
   850
          else build_info.sessions.iterator
wenzelm@65703
   851
        for ((session_name, session) <- entries_iterator) {
wenzelm@65748
   852
          stmt.string(1) = log_name
wenzelm@65748
   853
          stmt.string(2) = session_name
wenzelm@65748
   854
          stmt.string(3) = session.proper_chapter
wenzelm@65748
   855
          stmt.string(4) = session.proper_groups
wenzelm@65748
   856
          stmt.int(5) = session.threads
wenzelm@65748
   857
          stmt.long(6) = session.timing.elapsed.proper_ms
wenzelm@65748
   858
          stmt.long(7) = session.timing.cpu.proper_ms
wenzelm@65748
   859
          stmt.long(8) = session.timing.gc.proper_ms
wenzelm@65748
   860
          stmt.double(9) = session.timing.factor
wenzelm@65748
   861
          stmt.long(10) = session.ml_timing.elapsed.proper_ms
wenzelm@65748
   862
          stmt.long(11) = session.ml_timing.cpu.proper_ms
wenzelm@65748
   863
          stmt.long(12) = session.ml_timing.gc.proper_ms
wenzelm@65748
   864
          stmt.double(13) = session.ml_timing.factor
wenzelm@65748
   865
          stmt.long(14) = session.heap_size
wenzelm@65748
   866
          stmt.string(15) = session.status.map(_.toString)
wenzelm@65703
   867
          stmt.execute()
wenzelm@65703
   868
        }
wenzelm@65703
   869
      })
wenzelm@65642
   870
    }
wenzelm@65642
   871
wenzelm@65705
   872
    def update_ml_statistics(db: SQL.Database, log_name: String, build_info: Build_Info)
wenzelm@65645
   873
    {
wenzelm@65694
   874
      val table = Data.ml_statistics_table
wenzelm@65703
   875
      db.using_statement(db.insert_permissive(table))(stmt =>
wenzelm@65703
   876
      {
wenzelm@65703
   877
        val ml_stats: List[(String, Option[Bytes])] =
wenzelm@65703
   878
          Par_List.map[(String, Session_Entry), (String, Option[Bytes])](
wenzelm@65703
   879
            { case (a, b) => (a, compress_properties(b.ml_statistics).proper) },
wenzelm@65703
   880
            build_info.sessions.iterator.filter(p => p._2.ml_statistics.nonEmpty).toList)
wenzelm@65703
   881
        val entries = if (ml_stats.nonEmpty) ml_stats else List("" -> None)
wenzelm@65703
   882
        for ((session_name, ml_statistics) <- entries) {
wenzelm@65748
   883
          stmt.string(1) = log_name
wenzelm@65748
   884
          stmt.string(2) = session_name
wenzelm@65748
   885
          stmt.bytes(3) = ml_statistics
wenzelm@65703
   886
          stmt.execute()
wenzelm@65703
   887
        }
wenzelm@65703
   888
      })
wenzelm@65645
   889
    }
wenzelm@65645
   890
wenzelm@65645
   891
    def write_info(db: SQL.Database, files: List[JFile], ml_statistics: Boolean = false)
wenzelm@65642
   892
    {
wenzelm@65705
   893
      abstract class Table_Status(table: SQL.Table)
wenzelm@65642
   894
      {
wenzelm@65688
   895
        db.create_table(table)
wenzelm@65694
   896
        private var known: Set[String] = domain(db, table, Data.log_name)
wenzelm@65688
   897
wenzelm@65642
   898
        def required(file: JFile): Boolean = !known(Log_File.plain_name(file.getName))
wenzelm@65705
   899
wenzelm@65705
   900
        def update_db(db: SQL.Database, log_file: Log_File): Unit
wenzelm@65642
   901
        def update(log_file: Log_File)
wenzelm@65642
   902
        {
wenzelm@65642
   903
          if (!known(log_file.name)) {
wenzelm@65642
   904
            update_db(db, log_file)
wenzelm@65642
   905
            known += log_file.name
wenzelm@65618
   906
          }
wenzelm@65614
   907
        }
wenzelm@65605
   908
      }
wenzelm@65642
   909
      val status =
wenzelm@65642
   910
        List(
wenzelm@65705
   911
          new Table_Status(Data.meta_info_table) {
wenzelm@65705
   912
            override def update_db(db: SQL.Database, log_file: Log_File): Unit =
wenzelm@65705
   913
              update_meta_info(db, log_file.name, log_file.parse_meta_info())
wenzelm@65705
   914
          },
wenzelm@65705
   915
          new Table_Status(Data.sessions_table) {
wenzelm@65705
   916
            override def update_db(db: SQL.Database, log_file: Log_File): Unit =
wenzelm@65705
   917
              update_sessions(db, log_file.name, log_file.parse_build_info())
wenzelm@65705
   918
          },
wenzelm@65705
   919
          new Table_Status(Data.ml_statistics_table) {
wenzelm@65705
   920
            override def update_db(db: SQL.Database, log_file: Log_File): Unit =
wenzelm@65705
   921
            if (ml_statistics) {
wenzelm@65705
   922
              update_ml_statistics(db, log_file.name,
wenzelm@65705
   923
                log_file.parse_build_info(ml_statistics = true))
wenzelm@65705
   924
            }
wenzelm@65705
   925
          })
wenzelm@65642
   926
wenzelm@65645
   927
      for (file_group <- files.filter(file => status.exists(_.required(file))).grouped(100)) {
wenzelm@65642
   928
        val log_files = Par_List.map[JFile, Log_File](Log_File.apply _, file_group)
wenzelm@65642
   929
        db.transaction { log_files.foreach(log_file => status.foreach(_.update(log_file))) }
wenzelm@65642
   930
      }
wenzelm@65605
   931
    }
wenzelm@65605
   932
wenzelm@65621
   933
    def read_meta_info(db: SQL.Database, log_name: String): Option[Meta_Info] =
wenzelm@65621
   934
    {
wenzelm@65694
   935
      val table = Data.meta_info_table
wenzelm@65642
   936
      val columns = table.columns.tail
wenzelm@65699
   937
      db.using_statement(table.select(columns, Data.log_name.where_equal(log_name)))(stmt =>
wenzelm@65621
   938
      {
wenzelm@65740
   939
        val res = stmt.execute_query()
wenzelm@65740
   940
        if (!res.next) None
wenzelm@65621
   941
        else {
wenzelm@65621
   942
          val results =
wenzelm@65642
   943
            columns.map(c => c.name ->
wenzelm@65621
   944
              (if (c.T == SQL.Type.Date)
wenzelm@65740
   945
                res.get_date(c).map(Log_File.Date_Format(_))
wenzelm@65621
   946
               else
wenzelm@65740
   947
                res.get_string(c)))
wenzelm@65621
   948
          val n = Prop.all_props.length
wenzelm@65621
   949
          val props = for ((x, Some(y)) <- results.take(n)) yield (x, y)
wenzelm@65621
   950
          val settings = for ((x, Some(y)) <- results.drop(n)) yield (x, y)
wenzelm@65621
   951
          Some(Meta_Info(props, settings))
wenzelm@65621
   952
        }
wenzelm@65621
   953
      })
wenzelm@65621
   954
    }
wenzelm@65621
   955
wenzelm@65621
   956
    def read_build_info(
wenzelm@65629
   957
      db: SQL.Database,
wenzelm@65629
   958
      log_name: String,
wenzelm@65629
   959
      session_names: List[String] = Nil,
wenzelm@65629
   960
      ml_statistics: Boolean = false): Build_Info =
wenzelm@65621
   961
    {
wenzelm@65694
   962
      val table1 = Data.sessions_table
wenzelm@65694
   963
      val table2 = Data.ml_statistics_table
wenzelm@65629
   964
wenzelm@65645
   965
      val where_log_name =
wenzelm@65699
   966
        Data.log_name(table1).where_equal(log_name) + " AND " +
wenzelm@65701
   967
        Data.session_name(table1) + " <> ''"
wenzelm@65621
   968
      val where =
wenzelm@65645
   969
        if (session_names.isEmpty) where_log_name
wenzelm@65621
   970
        else
wenzelm@65645
   971
          where_log_name + " AND " +
wenzelm@65701
   972
          session_names.map(a => Data.session_name(table1) + " = " + SQL.string(a)).
wenzelm@65621
   973
            mkString("(", " OR ", ")")
wenzelm@65629
   974
wenzelm@65645
   975
      val columns1 = table1.columns.tail.map(_.apply(table1))
wenzelm@65645
   976
      val (columns, from) =
wenzelm@65645
   977
        if (ml_statistics) {
wenzelm@65694
   978
          val columns = columns1 ::: List(Data.ml_statistics(table2))
wenzelm@65668
   979
          val join =
wenzelm@65738
   980
            table1 + " LEFT OUTER JOIN " + table2 + " ON " +
wenzelm@65738
   981
            Data.log_name(table1) + " = " + Data.log_name(table2) + " AND " +
wenzelm@65738
   982
            Data.session_name(table1) + " = " + Data.session_name(table2)
wenzelm@65668
   983
          (columns, SQL.enclose(join))
wenzelm@65645
   984
        }
wenzelm@65695
   985
        else (columns1, table1.ident)
wenzelm@65645
   986
wenzelm@65621
   987
      val sessions =
wenzelm@65698
   988
        db.using_statement(SQL.select(columns) + from + " " + where)(stmt =>
wenzelm@65621
   989
        {
wenzelm@65740
   990
          stmt.execute_query().iterator(res =>
wenzelm@65621
   991
          {
wenzelm@65740
   992
            val session_name = res.string(Data.session_name)
wenzelm@65626
   993
            val session_entry =
wenzelm@65626
   994
              Session_Entry(
wenzelm@65740
   995
                chapter = res.string(Data.chapter),
wenzelm@65740
   996
                groups = split_lines(res.string(Data.groups)),
wenzelm@65740
   997
                threads = res.get_int(Data.threads),
wenzelm@65741
   998
                timing = res.timing(Data.timing_elapsed, Data.timing_cpu, Data.timing_gc),
wenzelm@65626
   999
                ml_timing =
wenzelm@65741
  1000
                  res.timing(Data.ml_timing_elapsed, Data.ml_timing_cpu, Data.ml_timing_gc),
wenzelm@65740
  1001
                heap_size = res.get_long(Data.heap_size),
wenzelm@65740
  1002
                status = res.get_string(Data.status).map(Session_Status.withName(_)),
wenzelm@65629
  1003
                ml_statistics =
wenzelm@65740
  1004
                  if (ml_statistics) uncompress_properties(res.bytes(Data.ml_statistics))
wenzelm@65629
  1005
                  else Nil)
wenzelm@65626
  1006
            session_name -> session_entry
wenzelm@65621
  1007
          }).toMap
wenzelm@65621
  1008
        })
wenzelm@65621
  1009
      Build_Info(sessions)
wenzelm@65621
  1010
    }
wenzelm@65595
  1011
  }
wenzelm@64045
  1012
}