src/Pure/Admin/build_log.scala
author wenzelm
Sat Oct 14 20:58:52 2017 +0200 (21 months ago)
changeset 66863 6acd1a2bd146
parent 66858 2ca6f0275de7
child 66873 9953ae603a23
permissions -rw-r--r--
clarified afp_pull_date: both repository versions are relevant;
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@66046
   259
      try {
wenzelm@66046
   260
        xml_cache.props(Properties.decode_lines(XML.Decode.properties(YXML.parse_body(text))))
wenzelm@66046
   261
      }
wenzelm@66046
   262
      catch { case _: XML.Error => log_file.err("malformed properties") }
wenzelm@64062
   263
wenzelm@65934
   264
    def filter_lines(marker: String): List[String] =
wenzelm@65934
   265
      for (line <- lines; s <- Library.try_unprefix(marker, line)) yield s
wenzelm@65934
   266
wenzelm@64119
   267
    def filter_props(marker: String): List[Properties.T] =
wenzelm@65934
   268
      for (s <- filter_lines(marker) if YXML.detect(s)) yield parse_props(s)
wenzelm@64062
   269
wenzelm@64119
   270
    def find_props(marker: String): Option[Properties.T] =
wenzelm@64196
   271
      find_line(marker) match {
wenzelm@64196
   272
        case Some(text) if YXML.detect(text) => Some(parse_props(text))
wenzelm@64196
   273
        case _ => None
wenzelm@64196
   274
      }
wenzelm@64062
   275
wenzelm@64062
   276
wenzelm@64062
   277
    /* parse various formats */
wenzelm@64062
   278
wenzelm@64105
   279
    def parse_meta_info(): Meta_Info = Build_Log.parse_meta_info(log_file)
wenzelm@64105
   280
wenzelm@65646
   281
    def parse_build_info(ml_statistics: Boolean = false): Build_Info =
wenzelm@65646
   282
      Build_Log.parse_build_info(log_file, ml_statistics)
wenzelm@64105
   283
wenzelm@64082
   284
    def parse_session_info(
wenzelm@64082
   285
        command_timings: Boolean = false,
wenzelm@64082
   286
        ml_statistics: Boolean = false,
wenzelm@64082
   287
        task_statistics: Boolean = false): Session_Info =
wenzelm@65318
   288
      Build_Log.parse_session_info(log_file, command_timings, ml_statistics, task_statistics)
wenzelm@64045
   289
  }
wenzelm@64045
   290
wenzelm@64045
   291
wenzelm@64098
   292
wenzelm@65276
   293
  /** digested meta info: produced by Admin/build_history in log.xz file **/
wenzelm@64045
   294
wenzelm@64108
   295
  object Meta_Info
wenzelm@64099
   296
  {
wenzelm@64108
   297
    val empty: Meta_Info = Meta_Info(Nil, Nil)
wenzelm@64099
   298
  }
wenzelm@64098
   299
wenzelm@65611
   300
  sealed case class Meta_Info(props: Properties.T, settings: Settings.T)
wenzelm@64103
   301
  {
wenzelm@64103
   302
    def is_empty: Boolean = props.isEmpty && settings.isEmpty
wenzelm@65599
   303
wenzelm@65611
   304
    def get(c: SQL.Column): Option[String] =
wenzelm@65611
   305
      Properties.get(props, c.name) orElse
wenzelm@65611
   306
      Properties.get(settings, c.name)
wenzelm@65611
   307
wenzelm@65611
   308
    def get_date(c: SQL.Column): Option[Date] =
wenzelm@65611
   309
      get(c).map(Log_File.Date_Format.parse(_))
wenzelm@64103
   310
  }
wenzelm@64061
   311
wenzelm@65625
   312
  object Identify
wenzelm@65625
   313
  {
wenzelm@65625
   314
    val log_prefix = "isabelle_identify_"
wenzelm@65674
   315
wenzelm@65674
   316
    def engine(log_file: Log_File): String =
wenzelm@65674
   317
      if (log_file.name.startsWith(Jenkins.log_prefix)) "jenkins_identify"
wenzelm@65674
   318
      else "identify"
wenzelm@65674
   319
wenzelm@65674
   320
    def content(date: Date, isabelle_version: Option[String], afp_version: Option[String]): String =
wenzelm@65674
   321
      terminate_lines(
wenzelm@65674
   322
        List("isabelle_identify: " + Build_Log.print_date(date), "") :::
wenzelm@65674
   323
        isabelle_version.map("Isabelle version: " + _).toList :::
wenzelm@65674
   324
        afp_version.map("AFP version: " + _).toList)
wenzelm@65674
   325
wenzelm@65625
   326
    val Start = new Regex("""^isabelle_identify: (.+)$""")
wenzelm@65625
   327
    val No_End = new Regex("""$.""")
wenzelm@65684
   328
    val Isabelle_Version = List(new Regex("""^Isabelle version: (\S+)$"""))
wenzelm@65684
   329
    val AFP_Version = List(new Regex("""^AFP version: (\S+)$"""))
wenzelm@65625
   330
  }
wenzelm@65625
   331
wenzelm@64095
   332
  object Isatest
wenzelm@64095
   333
  {
wenzelm@65588
   334
    val log_prefix = "isatest-makeall-"
wenzelm@64108
   335
    val engine = "isatest"
wenzelm@64109
   336
    val Start = new Regex("""^------------------- starting test --- (.+) --- (.+)$""")
wenzelm@64109
   337
    val End = new Regex("""^------------------- test (?:successful|FAILED) --- (.+) --- .*$""")
wenzelm@65684
   338
    val Isabelle_Version = List(new Regex("""^Isabelle version: (\S+)$"""))
wenzelm@64095
   339
  }
wenzelm@64095
   340
wenzelm@64109
   341
  object AFP_Test
wenzelm@64061
   342
  {
wenzelm@65588
   343
    val log_prefix = "afp-test-devel-"
wenzelm@64108
   344
    val engine = "afp-test"
wenzelm@64109
   345
    val Start = new Regex("""^Start test(?: for .+)? at ([^,]+), (.*)$""")
wenzelm@64109
   346
    val Start_Old = new Regex("""^Start test(?: for .+)? at ([^,]+)$""")
wenzelm@64109
   347
    val End = new Regex("""^End test on (.+), .+, elapsed time:.*$""")
wenzelm@65684
   348
    val Isabelle_Version = List(new Regex("""^Isabelle version: .* -- hg id (\S+)$"""))
wenzelm@65684
   349
    val AFP_Version = List(new Regex("""^AFP version: .* -- hg id (\S+)$"""))
wenzelm@64104
   350
    val Bad_Init = new Regex("""^cp:.*: Disc quota exceeded$""")
wenzelm@64061
   351
  }
wenzelm@64061
   352
wenzelm@64110
   353
  object Jenkins
wenzelm@64110
   354
  {
wenzelm@65664
   355
    val log_prefix = "jenkins_"
wenzelm@64110
   356
    val engine = "jenkins"
wenzelm@65663
   357
    val Host = new Regex("""^Building remotely on (\S+) \((\S+)\).*$""")
wenzelm@65665
   358
    val Start = new Regex("""^(?:Started by an SCM change|Started from command line by admin|).*$""")
wenzelm@64110
   359
    val Start_Date = new Regex("""^Build started at (.+)$""")
wenzelm@64110
   360
    val No_End = new Regex("""$.""")
wenzelm@65674
   361
    val Isabelle_Version =
wenzelm@65684
   362
      List(new Regex("""^(?:Build for Isabelle id|Isabelle id) (\w+).*$"""),
wenzelm@65685
   363
        new Regex("""^ISABELLE_CI_REPO_ID="(\w+)".*$"""),
wenzelm@65685
   364
        new Regex("""^(\w{12}) tip.*$"""))
wenzelm@65674
   365
    val AFP_Version =
wenzelm@65684
   366
      List(new Regex("""^(?:Build for AFP id|AFP id) (\w+).*$"""),
wenzelm@65684
   367
        new Regex("""^ISABELLE_CI_AFP_ID="(\w+)".*$"""))
wenzelm@64110
   368
    val CONFIGURATION = "=== CONFIGURATION ==="
wenzelm@64110
   369
    val BUILD = "=== BUILD ==="
wenzelm@64110
   370
  }
wenzelm@64110
   371
wenzelm@64105
   372
  private def parse_meta_info(log_file: Log_File): Meta_Info =
wenzelm@64061
   373
  {
wenzelm@64108
   374
    def parse(engine: String, host: String, start: Date,
wenzelm@65684
   375
      End: Regex, Isabelle_Version: List[Regex], AFP_Version: List[Regex]): Meta_Info =
wenzelm@64091
   376
    {
wenzelm@64296
   377
      val build_id =
wenzelm@64296
   378
      {
wenzelm@65714
   379
        val prefix = proper_string(host) orElse proper_string(engine) getOrElse "build"
wenzelm@65714
   380
        prefix + ":" + start.time.ms
wenzelm@64296
   381
      }
wenzelm@65591
   382
      val build_engine = if (engine == "") Nil else List(Prop.build_engine.name -> engine)
wenzelm@65591
   383
      val build_host = if (host == "") Nil else List(Prop.build_host.name -> host)
wenzelm@64108
   384
wenzelm@65599
   385
      val start_date = List(Prop.build_start.name -> print_date(start))
wenzelm@64091
   386
      val end_date =
wenzelm@64091
   387
        log_file.lines.last match {
wenzelm@64109
   388
          case End(log_file.Strict_Date(end_date)) =>
wenzelm@65599
   389
            List(Prop.build_end.name -> print_date(end_date))
wenzelm@64091
   390
          case _ => Nil
wenzelm@64091
   391
        }
wenzelm@64091
   392
wenzelm@64091
   393
      val isabelle_version =
wenzelm@65591
   394
        log_file.find_match(Isabelle_Version).map(Prop.isabelle_version.name -> _)
wenzelm@64091
   395
      val afp_version =
wenzelm@65591
   396
        log_file.find_match(AFP_Version).map(Prop.afp_version.name -> _)
wenzelm@64062
   397
wenzelm@65591
   398
      Meta_Info((Prop.build_id.name -> build_id) :: build_engine ::: build_host :::
wenzelm@64108
   399
          start_date ::: end_date ::: isabelle_version.toList ::: afp_version.toList,
wenzelm@65611
   400
        log_file.get_all_settings)
wenzelm@64091
   401
    }
wenzelm@64091
   402
wenzelm@64091
   403
    log_file.lines match {
wenzelm@64119
   404
      case line :: _ if line.startsWith(Build_History.META_INFO_MARKER) =>
wenzelm@64119
   405
        Meta_Info(log_file.find_props(Build_History.META_INFO_MARKER).get,
wenzelm@65611
   406
          log_file.get_all_settings)
wenzelm@64117
   407
wenzelm@65625
   408
      case Identify.Start(log_file.Strict_Date(start)) :: _ =>
wenzelm@65674
   409
        parse(Identify.engine(log_file), "", start, Identify.No_End,
wenzelm@65625
   410
          Identify.Isabelle_Version, Identify.AFP_Version)
wenzelm@65625
   411
wenzelm@64109
   412
      case Isatest.Start(log_file.Strict_Date(start), host) :: _ =>
wenzelm@64109
   413
        parse(Isatest.engine, host, start, Isatest.End,
wenzelm@65684
   414
          Isatest.Isabelle_Version, Nil)
wenzelm@64099
   415
wenzelm@64109
   416
      case AFP_Test.Start(log_file.Strict_Date(start), host) :: _ =>
wenzelm@64109
   417
        parse(AFP_Test.engine, host, start, AFP_Test.End,
wenzelm@64109
   418
          AFP_Test.Isabelle_Version, AFP_Test.AFP_Version)
wenzelm@64099
   419
wenzelm@64109
   420
      case AFP_Test.Start_Old(log_file.Strict_Date(start)) :: _ =>
wenzelm@64109
   421
        parse(AFP_Test.engine, "", start, AFP_Test.End,
wenzelm@64109
   422
          AFP_Test.Isabelle_Version, AFP_Test.AFP_Version)
wenzelm@64099
   423
wenzelm@65665
   424
      case Jenkins.Start() :: _ =>
wenzelm@64110
   425
        log_file.lines.dropWhile(_ != Jenkins.BUILD) match {
wenzelm@64110
   426
          case Jenkins.BUILD :: _ :: Jenkins.Start_Date(log_file.Strict_Date(start)) :: _ =>
wenzelm@65663
   427
            val host =
wenzelm@65663
   428
              log_file.lines.takeWhile(_ != Jenkins.CONFIGURATION).collectFirst({
wenzelm@65663
   429
                case Jenkins.Host(a, b) => a + "." + b
wenzelm@65663
   430
              }).getOrElse("")
wenzelm@65663
   431
            parse(Jenkins.engine, host, start.to(ZoneId.of("Europe/Berlin")), Jenkins.No_End,
wenzelm@64110
   432
              Jenkins.Isabelle_Version, Jenkins.AFP_Version)
wenzelm@64110
   433
          case _ => Meta_Info.empty
wenzelm@64110
   434
        }
wenzelm@64110
   435
wenzelm@64341
   436
      case line :: _ if line.startsWith("\u0000") => Meta_Info.empty
wenzelm@64109
   437
      case List(Isatest.End(_)) => Meta_Info.empty
wenzelm@64109
   438
      case _ :: AFP_Test.Bad_Init() :: _ => Meta_Info.empty
wenzelm@64105
   439
      case Nil => Meta_Info.empty
wenzelm@64104
   440
wenzelm@64110
   441
      case _ => log_file.err("cannot detect log file format")
wenzelm@64061
   442
    }
wenzelm@64061
   443
  }
wenzelm@64061
   444
wenzelm@64085
   445
wenzelm@64098
   446
wenzelm@65276
   447
  /** build info: toplevel output of isabelle build or Admin/build_history **/
wenzelm@64119
   448
wenzelm@64119
   449
  val ML_STATISTICS_MARKER = "\fML_statistics = "
wenzelm@65937
   450
  val ERROR_MESSAGE_MARKER = "\ferror_message = "
wenzelm@64119
   451
  val SESSION_NAME = "session_name"
wenzelm@64085
   452
wenzelm@64061
   453
  object Session_Status extends Enumeration
wenzelm@64061
   454
  {
wenzelm@65633
   455
    val existing, finished, failed, cancelled = Value
wenzelm@64061
   456
  }
wenzelm@64061
   457
wenzelm@65643
   458
  object Session_Entry
wenzelm@65643
   459
  {
wenzelm@65643
   460
    val empty: Session_Entry = Session_Entry()
wenzelm@65643
   461
  }
wenzelm@65643
   462
wenzelm@64085
   463
  sealed case class Session_Entry(
wenzelm@65643
   464
    chapter: String = "",
wenzelm@65643
   465
    groups: List[String] = Nil,
wenzelm@65643
   466
    threads: Option[Int] = None,
wenzelm@65643
   467
    timing: Timing = Timing.zero,
wenzelm@65643
   468
    ml_timing: Timing = Timing.zero,
wenzelm@65643
   469
    heap_size: Option[Long] = None,
wenzelm@65643
   470
    status: Option[Session_Status.Value] = None,
wenzelm@65937
   471
    errors: List[String] = Nil,
wenzelm@65643
   472
    ml_statistics: List[Properties.T] = Nil)
wenzelm@64085
   473
  {
wenzelm@65631
   474
    def proper_groups: Option[String] = if (groups.isEmpty) None else Some(cat_lines(groups))
wenzelm@65643
   475
    def finished: Boolean = status == Some(Session_Status.finished)
wenzelm@65937
   476
    def failed: Boolean = status == Some(Session_Status.failed)
wenzelm@64085
   477
  }
wenzelm@64054
   478
wenzelm@64085
   479
  sealed case class Build_Info(sessions: Map[String, Session_Entry])
wenzelm@64054
   480
  {
wenzelm@65937
   481
    def finished_sessions: List[String] = for ((a, b) <- sessions.toList if b.finished) yield a
wenzelm@65937
   482
    def failed_sessions: List[String] = for ((a, b) <- sessions.toList if b.failed) yield a
wenzelm@64054
   483
  }
wenzelm@64054
   484
wenzelm@65646
   485
  private def parse_build_info(log_file: Log_File, parse_ml_statistics: Boolean): Build_Info =
wenzelm@64085
   486
  {
wenzelm@64085
   487
    object Chapter_Name
wenzelm@64085
   488
    {
wenzelm@64085
   489
      def unapply(s: String): Some[(String, String)] =
wenzelm@64085
   490
        space_explode('/', s) match {
wenzelm@64085
   491
          case List(chapter, name) => Some((chapter, name))
wenzelm@64085
   492
          case _ => Some(("", s))
wenzelm@64085
   493
        }
wenzelm@64085
   494
    }
wenzelm@64054
   495
wenzelm@64085
   496
    val Session_No_Groups = new Regex("""^Session (\S+)$""")
wenzelm@64085
   497
    val Session_Groups = new Regex("""^Session (\S+) \((.*)\)$""")
wenzelm@64085
   498
    val Session_Finished1 =
wenzelm@64085
   499
      new Regex("""^Finished (\S+) \((\d+):(\d+):(\d+) elapsed time, (\d+):(\d+):(\d+) cpu time.*$""")
wenzelm@64085
   500
    val Session_Finished2 =
wenzelm@64085
   501
      new Regex("""^Finished (\S+) \((\d+):(\d+):(\d+) elapsed time.*$""")
wenzelm@64085
   502
    val Session_Timing =
wenzelm@65679
   503
      new Regex("""^Timing (\S+) \((\d+) threads, (\d+\.\d+)s elapsed time, (\d+\.\d+)s cpu time, (\d+\.\d+)s GC time.*$""")
wenzelm@64086
   504
    val Session_Started = new Regex("""^(?:Running|Building) (\S+) \.\.\.$""")
wenzelm@64085
   505
    val Session_Failed = new Regex("""^(\S+) FAILED""")
wenzelm@64085
   506
    val Session_Cancelled = new Regex("""^(\S+) CANCELLED""")
wenzelm@64120
   507
    val Heap = new Regex("""^Heap (\S+) \((\d+) bytes\)$""")
wenzelm@64085
   508
wenzelm@64085
   509
    var chapter = Map.empty[String, String]
wenzelm@64085
   510
    var groups = Map.empty[String, List[String]]
wenzelm@64085
   511
    var threads = Map.empty[String, Int]
wenzelm@64054
   512
    var timing = Map.empty[String, Timing]
wenzelm@64085
   513
    var ml_timing = Map.empty[String, Timing]
wenzelm@64086
   514
    var started = Set.empty[String]
wenzelm@64085
   515
    var failed = Set.empty[String]
wenzelm@64085
   516
    var cancelled = Set.empty[String]
wenzelm@65627
   517
    var heap_sizes = Map.empty[String, Long]
wenzelm@64119
   518
    var ml_statistics = Map.empty[String, List[Properties.T]]
wenzelm@65937
   519
    var errors = Map.empty[String, List[String]]
wenzelm@64119
   520
wenzelm@64085
   521
    def all_sessions: Set[String] =
wenzelm@64120
   522
      chapter.keySet ++ groups.keySet ++ threads.keySet ++ timing.keySet ++ ml_timing.keySet ++
wenzelm@65627
   523
      failed ++ cancelled ++ started ++ heap_sizes.keySet ++ ml_statistics.keySet
wenzelm@64085
   524
wenzelm@64054
   525
wenzelm@64062
   526
    for (line <- log_file.lines) {
wenzelm@64054
   527
      line match {
wenzelm@64085
   528
        case Session_No_Groups(Chapter_Name(chapt, name)) =>
wenzelm@64085
   529
          chapter += (name -> chapt)
wenzelm@64085
   530
          groups += (name -> Nil)
wenzelm@64119
   531
wenzelm@64085
   532
        case Session_Groups(Chapter_Name(chapt, name), grps) =>
wenzelm@64085
   533
          chapter += (name -> chapt)
wenzelm@64085
   534
          groups += (name -> Word.explode(grps))
wenzelm@64119
   535
wenzelm@64086
   536
        case Session_Started(name) =>
wenzelm@64086
   537
          started += name
wenzelm@64119
   538
wenzelm@64054
   539
        case Session_Finished1(name,
wenzelm@64054
   540
            Value.Int(e1), Value.Int(e2), Value.Int(e3),
wenzelm@64054
   541
            Value.Int(c1), Value.Int(c2), Value.Int(c3)) =>
wenzelm@64054
   542
          val elapsed = Time.hms(e1, e2, e3)
wenzelm@64054
   543
          val cpu = Time.hms(c1, c2, c3)
wenzelm@64085
   544
          timing += (name -> Timing(elapsed, cpu, Time.zero))
wenzelm@64119
   545
wenzelm@64054
   546
        case Session_Finished2(name,
wenzelm@64054
   547
            Value.Int(e1), Value.Int(e2), Value.Int(e3)) =>
wenzelm@64054
   548
          val elapsed = Time.hms(e1, e2, e3)
wenzelm@64085
   549
          timing += (name -> Timing(elapsed, Time.zero, Time.zero))
wenzelm@64119
   550
wenzelm@64054
   551
        case Session_Timing(name,
wenzelm@64054
   552
            Value.Int(t), Value.Double(e), Value.Double(c), Value.Double(g)) =>
wenzelm@64054
   553
          val elapsed = Time.seconds(e)
wenzelm@64054
   554
          val cpu = Time.seconds(c)
wenzelm@64054
   555
          val gc = Time.seconds(g)
wenzelm@64085
   556
          ml_timing += (name -> Timing(elapsed, cpu, gc))
wenzelm@64054
   557
          threads += (name -> t)
wenzelm@64119
   558
wenzelm@64120
   559
        case Heap(name, Value.Long(size)) =>
wenzelm@64120
   560
          heap_sizes += (name -> size)
wenzelm@64120
   561
wenzelm@65937
   562
        case _ if parse_ml_statistics && line.startsWith(ML_STATISTICS_MARKER) && YXML.detect(line) =>
wenzelm@64119
   563
          val (name, props) =
wenzelm@64119
   564
            Library.try_unprefix(ML_STATISTICS_MARKER, line).map(log_file.parse_props(_)) match {
wenzelm@65937
   565
              case Some((SESSION_NAME, name) :: props) => (name, props)
wenzelm@64119
   566
              case _ => log_file.err("malformed ML_statistics " + quote(line))
wenzelm@64119
   567
            }
wenzelm@64120
   568
          ml_statistics += (name -> (props :: ml_statistics.getOrElse(name, Nil)))
wenzelm@64119
   569
wenzelm@65937
   570
        case _ if line.startsWith(ERROR_MESSAGE_MARKER) && YXML.detect(line) =>
wenzelm@65937
   571
          val (name, msg) =
wenzelm@65937
   572
            Library.try_unprefix(ERROR_MESSAGE_MARKER, line).map(log_file.parse_props(_)) match {
wenzelm@65937
   573
              case Some(List((SESSION_NAME, name), (Markup.CONTENT, msg))) => (name, msg)
wenzelm@65937
   574
              case _ => log_file.err("malformed error message " + quote(line))
wenzelm@65937
   575
            }
wenzelm@65937
   576
          errors += (name -> (Library.decode_lines(msg) :: errors.getOrElse(name, Nil)))
wenzelm@65937
   577
wenzelm@64054
   578
        case _ =>
wenzelm@64054
   579
      }
wenzelm@64054
   580
    }
wenzelm@64054
   581
wenzelm@64085
   582
    val sessions =
wenzelm@64085
   583
      Map(
wenzelm@64085
   584
        (for (name <- all_sessions.toList) yield {
wenzelm@64085
   585
          val status =
wenzelm@65633
   586
            if (failed(name)) Session_Status.failed
wenzelm@65633
   587
            else if (cancelled(name)) Session_Status.cancelled
wenzelm@64086
   588
            else if (timing.isDefinedAt(name) || ml_timing.isDefinedAt(name))
wenzelm@65633
   589
              Session_Status.finished
wenzelm@65633
   590
            else if (started(name)) Session_Status.failed
wenzelm@65633
   591
            else Session_Status.existing
wenzelm@64085
   592
          val entry =
wenzelm@64085
   593
            Session_Entry(
wenzelm@65643
   594
              chapter = chapter.getOrElse(name, ""),
wenzelm@65643
   595
              groups = groups.getOrElse(name, Nil),
wenzelm@65643
   596
              threads = threads.get(name),
wenzelm@65643
   597
              timing = timing.getOrElse(name, Timing.zero),
wenzelm@65643
   598
              ml_timing = ml_timing.getOrElse(name, Timing.zero),
wenzelm@65643
   599
              heap_size = heap_sizes.get(name),
wenzelm@65643
   600
              status = Some(status),
wenzelm@65937
   601
              errors = errors.getOrElse(name, Nil).reverse,
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@65934
   616
    task_statistics: List[Properties.T],
wenzelm@65934
   617
    errors: List[String])
wenzelm@64099
   618
wenzelm@64099
   619
  private def parse_session_info(
wenzelm@64099
   620
    log_file: Log_File,
wenzelm@64099
   621
    command_timings: Boolean,
wenzelm@64099
   622
    ml_statistics: Boolean,
wenzelm@64099
   623
    task_statistics: Boolean): Session_Info =
wenzelm@64099
   624
  {
wenzelm@65290
   625
    Session_Info(
wenzelm@65290
   626
      session_timing = log_file.find_props("\fTiming = ") getOrElse Nil,
wenzelm@65290
   627
      command_timings = if (command_timings) log_file.filter_props("\fcommand_timing = ") else Nil,
wenzelm@65290
   628
      ml_statistics = if (ml_statistics) log_file.filter_props(ML_STATISTICS_MARKER) else Nil,
wenzelm@65934
   629
      task_statistics = if (task_statistics) log_file.filter_props("\ftask_statistics = ") else Nil,
wenzelm@65937
   630
      errors = log_file.filter_lines(ERROR_MESSAGE_MARKER).map(Library.decode_lines(_)))
wenzelm@64099
   631
  }
wenzelm@65595
   632
wenzelm@65937
   633
  def compress_errors(errors: List[String]): Option[Bytes] =
wenzelm@65937
   634
    if (errors.isEmpty) None
wenzelm@65937
   635
    else Some(Bytes(YXML.string_of_body(XML.Encode.list(XML.Encode.string)(errors))).compress())
wenzelm@65937
   636
wenzelm@65937
   637
  def uncompress_errors(bytes: Bytes): List[String] =
wenzelm@65937
   638
    if (bytes.isEmpty) Nil
wenzelm@65937
   639
    else XML.Decode.list(YXML.string_of_body(_))(YXML.parse_body(bytes.uncompress().text))
wenzelm@65937
   640
wenzelm@65595
   641
wenzelm@65595
   642
wenzelm@65595
   643
  /** persistent store **/
wenzelm@65595
   644
wenzelm@65694
   645
  /* SQL data model */
wenzelm@65694
   646
wenzelm@65694
   647
  object Data
wenzelm@65694
   648
  {
wenzelm@65702
   649
    def build_log_table(name: String, columns: List[SQL.Column], body: String = ""): SQL.Table =
wenzelm@65700
   650
      SQL.Table("isabelle_build_log_" + name, columns, body)
wenzelm@65700
   651
wenzelm@65700
   652
wenzelm@65694
   653
    /* main content */
wenzelm@65694
   654
wenzelm@66857
   655
    val log_name = SQL.Column.string("log_name").make_primary_key
wenzelm@66857
   656
    val session_name = SQL.Column.string("session_name").make_primary_key
wenzelm@65694
   657
    val chapter = SQL.Column.string("chapter")
wenzelm@65694
   658
    val groups = SQL.Column.string("groups")
wenzelm@65694
   659
    val threads = SQL.Column.int("threads")
wenzelm@65694
   660
    val timing_elapsed = SQL.Column.long("timing_elapsed")
wenzelm@65694
   661
    val timing_cpu = SQL.Column.long("timing_cpu")
wenzelm@65694
   662
    val timing_gc = SQL.Column.long("timing_gc")
wenzelm@65694
   663
    val timing_factor = SQL.Column.double("timing_factor")
wenzelm@65694
   664
    val ml_timing_elapsed = SQL.Column.long("ml_timing_elapsed")
wenzelm@65694
   665
    val ml_timing_cpu = SQL.Column.long("ml_timing_cpu")
wenzelm@65694
   666
    val ml_timing_gc = SQL.Column.long("ml_timing_gc")
wenzelm@65694
   667
    val ml_timing_factor = SQL.Column.double("ml_timing_factor")
wenzelm@65694
   668
    val heap_size = SQL.Column.long("heap_size")
wenzelm@65694
   669
    val status = SQL.Column.string("status")
wenzelm@65937
   670
    val errors = SQL.Column.bytes("errors")
wenzelm@65694
   671
    val ml_statistics = SQL.Column.bytes("ml_statistics")
wenzelm@65783
   672
    val known = SQL.Column.bool("known")
wenzelm@65694
   673
wenzelm@65694
   674
    val meta_info_table =
wenzelm@65702
   675
      build_log_table("meta_info", log_name :: Prop.all_props ::: Settings.all_settings)
wenzelm@65694
   676
wenzelm@65694
   677
    val sessions_table =
wenzelm@65702
   678
      build_log_table("sessions",
wenzelm@65694
   679
        List(log_name, session_name, chapter, groups, threads, timing_elapsed, timing_cpu,
wenzelm@65694
   680
          timing_gc, timing_factor, ml_timing_elapsed, ml_timing_cpu, ml_timing_gc, ml_timing_factor,
wenzelm@65937
   681
          heap_size, status, errors))
wenzelm@65694
   682
wenzelm@65694
   683
    val ml_statistics_table =
wenzelm@65702
   684
      build_log_table("ml_statistics", List(log_name, session_name, ml_statistics))
wenzelm@65694
   685
wenzelm@65694
   686
wenzelm@66855
   687
    /* AFP versions */
wenzelm@66855
   688
wenzelm@66855
   689
    val isabelle_afp_versions_table: SQL.Table =
wenzelm@66855
   690
    {
wenzelm@66855
   691
      val version1 = Prop.isabelle_version
wenzelm@66855
   692
      val version2 = Prop.afp_version
wenzelm@66857
   693
      build_log_table("isabelle_afp_versions", List(version1.make_primary_key, version2),
wenzelm@66855
   694
        SQL.select(List(version1, version2), distinct = true) + meta_info_table +
wenzelm@66856
   695
        " WHERE " + version1.defined + " AND " + version2.defined)
wenzelm@66855
   696
    }
wenzelm@66855
   697
wenzelm@66855
   698
wenzelm@65705
   699
    /* earliest pull date for repository version (PostgreSQL queries) */
wenzelm@65694
   700
wenzelm@65694
   701
    val pull_date = SQL.Column.date("pull_date")
wenzelm@65694
   702
wenzelm@66863
   703
    def pull_date_table(afp: Boolean = false): SQL.Table =
wenzelm@65709
   704
    {
wenzelm@66863
   705
      val (name, versions) =
wenzelm@66863
   706
        if (afp) ("afp_pull_date", List(Prop.isabelle_version, Prop.afp_version))
wenzelm@66863
   707
        else ("pull_date", List(Prop.isabelle_version))
wenzelm@65694
   708
wenzelm@66863
   709
      build_log_table(name, versions.map(_.make_primary_key) ::: List(pull_date),
wenzelm@66863
   710
        "SELECT " + versions.mkString(", ") + ", min(" + Prop.build_start + ") AS " + pull_date +
wenzelm@66863
   711
        " FROM " + meta_info_table +
wenzelm@66863
   712
        " WHERE " + (versions ::: List(Prop.build_start)).map(_.defined).mkString(" AND ") +
wenzelm@66863
   713
        " GROUP BY " + versions.mkString(", "))
wenzelm@66855
   714
    }
wenzelm@66855
   715
wenzelm@66855
   716
wenzelm@66855
   717
    /* recent entries */
wenzelm@66855
   718
wenzelm@65736
   719
    def recent_time(days: Int): SQL.Source =
wenzelm@65736
   720
      "now() - INTERVAL '" + days.max(0) + " days'"
wenzelm@65736
   721
wenzelm@66863
   722
    def recent_pull_date_table(
wenzelm@66863
   723
      days: Int, rev: String = "", afp_rev: Option[String] = None): SQL.Table =
wenzelm@65702
   724
    {
wenzelm@66863
   725
      val afp = afp_rev.isDefined
wenzelm@66863
   726
      val rev2 = afp_rev.getOrElse("")
wenzelm@66863
   727
      val table = pull_date_table(afp)
wenzelm@66863
   728
wenzelm@66863
   729
      val version1 = Prop.isabelle_version
wenzelm@66863
   730
      val version2 = Prop.afp_version
wenzelm@66863
   731
      val eq1 = version1(table) + " = " + SQL.string(rev)
wenzelm@66863
   732
      val eq2 = version2(table) + " = " + SQL.string(rev2)
wenzelm@66863
   733
wenzelm@65777
   734
      SQL.Table("recent_pull_date", table.columns,
wenzelm@65783
   735
        table.select(table.columns,
wenzelm@65783
   736
          "WHERE " + pull_date(table) + " > " + recent_time(days) +
wenzelm@66863
   737
          (if (rev != "" && rev2 == "") " OR " + eq1
wenzelm@66863
   738
           else if (rev == "" && rev2 != "") " OR " + eq2
wenzelm@66863
   739
           else if (rev != "" && rev2 != "") " OR (" + eq1 + " AND " + eq2 + ")"
wenzelm@66863
   740
           else "")))
wenzelm@65702
   741
    }
wenzelm@65694
   742
wenzelm@65781
   743
    def select_recent_log_names(days: Int): SQL.Source =
wenzelm@65781
   744
    {
wenzelm@65781
   745
      val table1 = meta_info_table
wenzelm@65781
   746
      val table2 = recent_pull_date_table(days)
wenzelm@65781
   747
      table1.select(List(log_name), distinct = true) + SQL.join_inner + table2.query_named +
wenzelm@65781
   748
        " ON " + Prop.isabelle_version(table1) + " = " + Prop.isabelle_version(table2)
wenzelm@65781
   749
    }
wenzelm@65781
   750
wenzelm@66863
   751
    def select_recent_versions(days: Int,
wenzelm@66863
   752
      rev: String = "", afp_rev: Option[String] = None, sql: SQL.Source = ""): SQL.Source =
wenzelm@65783
   753
    {
wenzelm@66858
   754
      val version = Prop.isabelle_version
wenzelm@66863
   755
      val table1 = recent_pull_date_table(days, rev = rev, afp_rev = afp_rev)
wenzelm@65783
   756
      val table2 = meta_info_table
wenzelm@65783
   757
      val aux_table = SQL.Table("aux", table2.columns, table2.select(sql = sql))
wenzelm@65783
   758
wenzelm@65783
   759
      val columns =
wenzelm@66858
   760
        table1.columns.map(c => c(table1)) :::
wenzelm@66858
   761
          List(known.copy(expr = log_name(aux_table).defined))
wenzelm@65783
   762
      SQL.select(columns, distinct = true) +
wenzelm@65783
   763
        table1.query_named + SQL.join_outer + aux_table.query_named +
wenzelm@66858
   764
        " ON " + version(table1) + " = " + version(aux_table) +
wenzelm@65783
   765
        " ORDER BY " + pull_date(table1) + " DESC"
wenzelm@65783
   766
    }
wenzelm@65783
   767
wenzelm@65724
   768
wenzelm@65724
   769
    /* universal view on main data */
wenzelm@65724
   770
wenzelm@65724
   771
    val universal_table: SQL.Table =
wenzelm@65724
   772
    {
wenzelm@65724
   773
      val table1 = meta_info_table
wenzelm@66863
   774
      val table2 = pull_date_table()
wenzelm@65724
   775
      val table3 = sessions_table
wenzelm@65850
   776
      val table4 = ml_statistics_table
wenzelm@65724
   777
wenzelm@65850
   778
      val a_columns = log_name :: pull_date :: meta_info_table.columns.tail
wenzelm@65850
   779
      val a_table =
wenzelm@65850
   780
        SQL.Table("a", a_columns,
wenzelm@65850
   781
          SQL.select(a_columns.take(2) ::: a_columns.drop(2).map(_.apply(table1))) +
wenzelm@65850
   782
            table1 + SQL.join_outer + table2 + " ON " +
wenzelm@65850
   783
            Prop.isabelle_version(table1) + " = " + Prop.isabelle_version(table2))
wenzelm@65724
   784
wenzelm@65850
   785
      val b_columns = a_columns ::: sessions_table.columns.tail
wenzelm@65850
   786
      val b_table =
wenzelm@65850
   787
        SQL.Table("b", b_columns,
wenzelm@65850
   788
          SQL.select(log_name(a_table) :: b_columns.tail) + a_table.query_named +
wenzelm@65850
   789
          SQL.join_inner + table3 + " ON " + log_name(a_table) + " = " + log_name(table3))
wenzelm@65850
   790
wenzelm@65850
   791
      SQL.Table("isabelle_build_log", b_columns ::: List(ml_statistics),
wenzelm@65724
   792
        {
wenzelm@65850
   793
          SQL.select(b_columns.map(_.apply(b_table)) ::: List(ml_statistics)) +
wenzelm@65850
   794
            b_table.query_named + SQL.join_outer + table4 + " ON " +
wenzelm@65850
   795
            log_name(b_table) + " = " + log_name(table4) + " AND " +
wenzelm@65850
   796
            session_name(b_table) + " = " + session_name(table4)
wenzelm@65724
   797
        })
wenzelm@65724
   798
    }
wenzelm@65694
   799
  }
wenzelm@65694
   800
wenzelm@65694
   801
wenzelm@65694
   802
  /* database access */
wenzelm@65694
   803
wenzelm@65595
   804
  def store(options: Options): Store = new Store(options)
wenzelm@65595
   805
wenzelm@65857
   806
  class Store private[Build_Log](options: Options)
wenzelm@65595
   807
  {
wenzelm@65857
   808
    val xml_cache: XML.Cache = new XML.Cache()
wenzelm@65857
   809
wenzelm@65595
   810
    def open_database(
wenzelm@65595
   811
      user: String = options.string("build_log_database_user"),
wenzelm@65595
   812
      password: String = options.string("build_log_database_password"),
wenzelm@65595
   813
      database: String = options.string("build_log_database_name"),
wenzelm@65595
   814
      host: String = options.string("build_log_database_host"),
wenzelm@65595
   815
      port: Int = options.int("build_log_database_port"),
wenzelm@65595
   816
      ssh_host: String = options.string("build_log_ssh_host"),
wenzelm@65595
   817
      ssh_user: String = options.string("build_log_ssh_user"),
wenzelm@65595
   818
      ssh_port: Int = options.int("build_log_ssh_port")): PostgreSQL.Database =
wenzelm@65595
   819
    {
wenzelm@65595
   820
      PostgreSQL.open_database(
wenzelm@65595
   821
        user = user, password = password, database = database, host = host, port = port,
wenzelm@65595
   822
        ssh =
wenzelm@65595
   823
          if (ssh_host == "") None
wenzelm@65636
   824
          else Some(SSH.init_context(options).open_session(ssh_host, ssh_user, port)),
wenzelm@65636
   825
        ssh_close = true)
wenzelm@65595
   826
    }
wenzelm@65599
   827
wenzelm@65723
   828
    def update_database(db: PostgreSQL.Database, dirs: List[Path], ml_statistics: Boolean = false)
wenzelm@65694
   829
    {
wenzelm@65694
   830
      write_info(db, Log_File.find_files(dirs), ml_statistics = ml_statistics)
wenzelm@65694
   831
wenzelm@66863
   832
      db.create_view(Data.pull_date_table())
wenzelm@66863
   833
      db.create_view(Data.pull_date_table(afp = true))
wenzelm@65724
   834
      db.create_view(Data.universal_table)
wenzelm@65694
   835
    }
wenzelm@65694
   836
wenzelm@65723
   837
    def snapshot_database(db: PostgreSQL.Database, sqlite_database: Path,
wenzelm@65705
   838
      days: Int = 100, ml_statistics: Boolean = false)
wenzelm@65694
   839
    {
wenzelm@65694
   840
      Isabelle_System.mkdirs(sqlite_database.dir)
wenzelm@65694
   841
      sqlite_database.file.delete
wenzelm@65694
   842
wenzelm@65694
   843
      using(SQLite.open_database(sqlite_database))(db2 =>
wenzelm@65694
   844
      {
wenzelm@65694
   845
        db.transaction {
wenzelm@65694
   846
          db2.transaction {
wenzelm@65705
   847
            // main content
wenzelm@65705
   848
            db2.create_table(Data.meta_info_table)
wenzelm@65705
   849
            db2.create_table(Data.sessions_table)
wenzelm@65705
   850
            db2.create_table(Data.ml_statistics_table)
wenzelm@65705
   851
wenzelm@65705
   852
            val recent_log_names =
wenzelm@65781
   853
              db.using_statement(Data.select_recent_log_names(days))(stmt =>
wenzelm@65779
   854
                stmt.execute_query().iterator(_.string(Data.log_name)).toList)
wenzelm@65705
   855
wenzelm@65705
   856
            for (log_name <- recent_log_names) {
wenzelm@65705
   857
              read_meta_info(db, log_name).foreach(meta_info =>
wenzelm@65705
   858
                update_meta_info(db2, log_name, meta_info))
wenzelm@65705
   859
wenzelm@65705
   860
              update_sessions(db2, log_name, read_build_info(db, log_name))
wenzelm@65705
   861
wenzelm@65856
   862
              if (ml_statistics) {
wenzelm@65856
   863
                update_ml_statistics(db2, log_name,
wenzelm@65856
   864
                  read_build_info(db, log_name, ml_statistics = true))
wenzelm@65856
   865
              }
wenzelm@65705
   866
            }
wenzelm@65705
   867
wenzelm@65705
   868
            // pull_date
wenzelm@65694
   869
            {
wenzelm@66863
   870
              val table = Data.pull_date_table()
wenzelm@65694
   871
              db2.create_table(table)
wenzelm@65698
   872
              db2.using_statement(table.insert())(stmt2 =>
wenzelm@65694
   873
              {
wenzelm@65777
   874
                db.using_statement(Data.recent_pull_date_table(days).query)(stmt =>
wenzelm@65694
   875
                {
wenzelm@65740
   876
                  val res = stmt.execute_query()
wenzelm@65740
   877
                  while (res.next()) {
wenzelm@65748
   878
                    for ((c, i) <- table.columns.zipWithIndex) {
wenzelm@65748
   879
                      stmt2.string(i + 1) = res.get_string(c)
wenzelm@65748
   880
                    }
wenzelm@65740
   881
                    stmt2.execute()
wenzelm@65694
   882
                  }
wenzelm@65694
   883
                })
wenzelm@65694
   884
              })
wenzelm@65709
   885
            }
wenzelm@65705
   886
wenzelm@65705
   887
            // full view
wenzelm@65724
   888
            db2.create_view(Data.universal_table)
wenzelm@65694
   889
          }
wenzelm@65694
   890
        }
wenzelm@65708
   891
        db2.rebuild
wenzelm@65694
   892
      })
wenzelm@65694
   893
    }
wenzelm@65694
   894
wenzelm@65688
   895
    def domain(db: SQL.Database, table: SQL.Table, column: SQL.Column): Set[String] =
wenzelm@65698
   896
      db.using_statement(table.select(List(column), distinct = true))(stmt =>
wenzelm@65740
   897
        stmt.execute_query().iterator(_.string(column)).toSet)
wenzelm@65688
   898
wenzelm@65705
   899
    def update_meta_info(db: SQL.Database, log_name: String, meta_info: Meta_Info)
wenzelm@65632
   900
    {
wenzelm@65694
   901
      val table = Data.meta_info_table
wenzelm@65703
   902
      db.using_statement(db.insert_permissive(table))(stmt =>
wenzelm@65703
   903
      {
wenzelm@65748
   904
        stmt.string(1) = log_name
wenzelm@65703
   905
        for ((c, i) <- table.columns.tail.zipWithIndex) {
wenzelm@65703
   906
          if (c.T == SQL.Type.Date)
wenzelm@65748
   907
            stmt.date(i + 2) = meta_info.get_date(c)
wenzelm@65703
   908
          else
wenzelm@65748
   909
            stmt.string(i + 2) = meta_info.get(c)
wenzelm@65703
   910
        }
wenzelm@65703
   911
        stmt.execute()
wenzelm@65703
   912
      })
wenzelm@65600
   913
    }
wenzelm@65600
   914
wenzelm@65705
   915
    def update_sessions(db: SQL.Database, log_name: String, build_info: Build_Info)
wenzelm@65599
   916
    {
wenzelm@65694
   917
      val table = Data.sessions_table
wenzelm@65703
   918
      db.using_statement(db.insert_permissive(table))(stmt =>
wenzelm@65703
   919
      {
wenzelm@65703
   920
        val entries_iterator =
wenzelm@65703
   921
          if (build_info.sessions.isEmpty) Iterator("" -> Session_Entry.empty)
wenzelm@65703
   922
          else build_info.sessions.iterator
wenzelm@65703
   923
        for ((session_name, session) <- entries_iterator) {
wenzelm@65748
   924
          stmt.string(1) = log_name
wenzelm@65748
   925
          stmt.string(2) = session_name
wenzelm@65853
   926
          stmt.string(3) = proper_string(session.chapter)
wenzelm@65748
   927
          stmt.string(4) = session.proper_groups
wenzelm@65748
   928
          stmt.int(5) = session.threads
wenzelm@65748
   929
          stmt.long(6) = session.timing.elapsed.proper_ms
wenzelm@65748
   930
          stmt.long(7) = session.timing.cpu.proper_ms
wenzelm@65748
   931
          stmt.long(8) = session.timing.gc.proper_ms
wenzelm@65748
   932
          stmt.double(9) = session.timing.factor
wenzelm@65748
   933
          stmt.long(10) = session.ml_timing.elapsed.proper_ms
wenzelm@65748
   934
          stmt.long(11) = session.ml_timing.cpu.proper_ms
wenzelm@65748
   935
          stmt.long(12) = session.ml_timing.gc.proper_ms
wenzelm@65748
   936
          stmt.double(13) = session.ml_timing.factor
wenzelm@65748
   937
          stmt.long(14) = session.heap_size
wenzelm@65748
   938
          stmt.string(15) = session.status.map(_.toString)
wenzelm@65937
   939
          stmt.bytes(16) = compress_errors(session.errors)
wenzelm@65703
   940
          stmt.execute()
wenzelm@65703
   941
        }
wenzelm@65703
   942
      })
wenzelm@65642
   943
    }
wenzelm@65642
   944
wenzelm@65705
   945
    def update_ml_statistics(db: SQL.Database, log_name: String, build_info: Build_Info)
wenzelm@65645
   946
    {
wenzelm@65694
   947
      val table = Data.ml_statistics_table
wenzelm@65703
   948
      db.using_statement(db.insert_permissive(table))(stmt =>
wenzelm@65703
   949
      {
wenzelm@65703
   950
        val ml_stats: List[(String, Option[Bytes])] =
wenzelm@65703
   951
          Par_List.map[(String, Session_Entry), (String, Option[Bytes])](
wenzelm@65857
   952
            { case (a, b) => (a, Properties.compress(b.ml_statistics).proper) },
wenzelm@65703
   953
            build_info.sessions.iterator.filter(p => p._2.ml_statistics.nonEmpty).toList)
wenzelm@65703
   954
        val entries = if (ml_stats.nonEmpty) ml_stats else List("" -> None)
wenzelm@65703
   955
        for ((session_name, ml_statistics) <- entries) {
wenzelm@65748
   956
          stmt.string(1) = log_name
wenzelm@65748
   957
          stmt.string(2) = session_name
wenzelm@65748
   958
          stmt.bytes(3) = ml_statistics
wenzelm@65703
   959
          stmt.execute()
wenzelm@65703
   960
        }
wenzelm@65703
   961
      })
wenzelm@65645
   962
    }
wenzelm@65645
   963
wenzelm@65645
   964
    def write_info(db: SQL.Database, files: List[JFile], ml_statistics: Boolean = false)
wenzelm@65642
   965
    {
wenzelm@65705
   966
      abstract class Table_Status(table: SQL.Table)
wenzelm@65642
   967
      {
wenzelm@65688
   968
        db.create_table(table)
wenzelm@65694
   969
        private var known: Set[String] = domain(db, table, Data.log_name)
wenzelm@65688
   970
wenzelm@65642
   971
        def required(file: JFile): Boolean = !known(Log_File.plain_name(file.getName))
wenzelm@65705
   972
wenzelm@65705
   973
        def update_db(db: SQL.Database, log_file: Log_File): Unit
wenzelm@65642
   974
        def update(log_file: Log_File)
wenzelm@65642
   975
        {
wenzelm@65642
   976
          if (!known(log_file.name)) {
wenzelm@65642
   977
            update_db(db, log_file)
wenzelm@65642
   978
            known += log_file.name
wenzelm@65618
   979
          }
wenzelm@65614
   980
        }
wenzelm@65605
   981
      }
wenzelm@65642
   982
      val status =
wenzelm@65642
   983
        List(
wenzelm@65705
   984
          new Table_Status(Data.meta_info_table) {
wenzelm@65705
   985
            override def update_db(db: SQL.Database, log_file: Log_File): Unit =
wenzelm@65705
   986
              update_meta_info(db, log_file.name, log_file.parse_meta_info())
wenzelm@65705
   987
          },
wenzelm@65705
   988
          new Table_Status(Data.sessions_table) {
wenzelm@65705
   989
            override def update_db(db: SQL.Database, log_file: Log_File): Unit =
wenzelm@65705
   990
              update_sessions(db, log_file.name, log_file.parse_build_info())
wenzelm@65705
   991
          },
wenzelm@65705
   992
          new Table_Status(Data.ml_statistics_table) {
wenzelm@65705
   993
            override def update_db(db: SQL.Database, log_file: Log_File): Unit =
wenzelm@65705
   994
            if (ml_statistics) {
wenzelm@65705
   995
              update_ml_statistics(db, log_file.name,
wenzelm@65705
   996
                log_file.parse_build_info(ml_statistics = true))
wenzelm@65705
   997
            }
wenzelm@65705
   998
          })
wenzelm@65642
   999
wenzelm@65645
  1000
      for (file_group <- files.filter(file => status.exists(_.required(file))).grouped(100)) {
wenzelm@65642
  1001
        val log_files = Par_List.map[JFile, Log_File](Log_File.apply _, file_group)
wenzelm@65642
  1002
        db.transaction { log_files.foreach(log_file => status.foreach(_.update(log_file))) }
wenzelm@65642
  1003
      }
wenzelm@65605
  1004
    }
wenzelm@65605
  1005
wenzelm@65621
  1006
    def read_meta_info(db: SQL.Database, log_name: String): Option[Meta_Info] =
wenzelm@65621
  1007
    {
wenzelm@65694
  1008
      val table = Data.meta_info_table
wenzelm@65642
  1009
      val columns = table.columns.tail
wenzelm@65699
  1010
      db.using_statement(table.select(columns, Data.log_name.where_equal(log_name)))(stmt =>
wenzelm@65621
  1011
      {
wenzelm@65740
  1012
        val res = stmt.execute_query()
wenzelm@65740
  1013
        if (!res.next) None
wenzelm@65621
  1014
        else {
wenzelm@65621
  1015
          val results =
wenzelm@65642
  1016
            columns.map(c => c.name ->
wenzelm@65621
  1017
              (if (c.T == SQL.Type.Date)
wenzelm@65740
  1018
                res.get_date(c).map(Log_File.Date_Format(_))
wenzelm@65621
  1019
               else
wenzelm@65740
  1020
                res.get_string(c)))
wenzelm@65621
  1021
          val n = Prop.all_props.length
wenzelm@65621
  1022
          val props = for ((x, Some(y)) <- results.take(n)) yield (x, y)
wenzelm@65621
  1023
          val settings = for ((x, Some(y)) <- results.drop(n)) yield (x, y)
wenzelm@65621
  1024
          Some(Meta_Info(props, settings))
wenzelm@65621
  1025
        }
wenzelm@65621
  1026
      })
wenzelm@65621
  1027
    }
wenzelm@65621
  1028
wenzelm@65621
  1029
    def read_build_info(
wenzelm@65629
  1030
      db: SQL.Database,
wenzelm@65629
  1031
      log_name: String,
wenzelm@65629
  1032
      session_names: List[String] = Nil,
wenzelm@65629
  1033
      ml_statistics: Boolean = false): Build_Info =
wenzelm@65621
  1034
    {
wenzelm@65694
  1035
      val table1 = Data.sessions_table
wenzelm@65694
  1036
      val table2 = Data.ml_statistics_table
wenzelm@65629
  1037
wenzelm@65645
  1038
      val where_log_name =
wenzelm@65699
  1039
        Data.log_name(table1).where_equal(log_name) + " AND " +
wenzelm@65701
  1040
        Data.session_name(table1) + " <> ''"
wenzelm@65621
  1041
      val where =
wenzelm@65645
  1042
        if (session_names.isEmpty) where_log_name
wenzelm@65804
  1043
        else where_log_name + " AND " + SQL.member(Data.session_name(table1).ident, session_names)
wenzelm@65629
  1044
wenzelm@65645
  1045
      val columns1 = table1.columns.tail.map(_.apply(table1))
wenzelm@65645
  1046
      val (columns, from) =
wenzelm@65645
  1047
        if (ml_statistics) {
wenzelm@65694
  1048
          val columns = columns1 ::: List(Data.ml_statistics(table2))
wenzelm@65668
  1049
          val join =
wenzelm@65775
  1050
            table1 + SQL.join_outer + table2 + " ON " +
wenzelm@65738
  1051
            Data.log_name(table1) + " = " + Data.log_name(table2) + " AND " +
wenzelm@65738
  1052
            Data.session_name(table1) + " = " + Data.session_name(table2)
wenzelm@65668
  1053
          (columns, SQL.enclose(join))
wenzelm@65645
  1054
        }
wenzelm@65695
  1055
        else (columns1, table1.ident)
wenzelm@65645
  1056
wenzelm@65621
  1057
      val sessions =
wenzelm@65698
  1058
        db.using_statement(SQL.select(columns) + from + " " + where)(stmt =>
wenzelm@65621
  1059
        {
wenzelm@65740
  1060
          stmt.execute_query().iterator(res =>
wenzelm@65621
  1061
          {
wenzelm@65740
  1062
            val session_name = res.string(Data.session_name)
wenzelm@65626
  1063
            val session_entry =
wenzelm@65626
  1064
              Session_Entry(
wenzelm@65740
  1065
                chapter = res.string(Data.chapter),
wenzelm@65740
  1066
                groups = split_lines(res.string(Data.groups)),
wenzelm@65740
  1067
                threads = res.get_int(Data.threads),
wenzelm@65741
  1068
                timing = res.timing(Data.timing_elapsed, Data.timing_cpu, Data.timing_gc),
wenzelm@65626
  1069
                ml_timing =
wenzelm@65741
  1070
                  res.timing(Data.ml_timing_elapsed, Data.ml_timing_cpu, Data.ml_timing_gc),
wenzelm@65740
  1071
                heap_size = res.get_long(Data.heap_size),
wenzelm@65740
  1072
                status = res.get_string(Data.status).map(Session_Status.withName(_)),
wenzelm@65937
  1073
                errors = uncompress_errors(res.bytes(Data.errors)),
wenzelm@65629
  1074
                ml_statistics =
wenzelm@65857
  1075
                  if (ml_statistics)
wenzelm@65857
  1076
                    Properties.uncompress(res.bytes(Data.ml_statistics), Some(xml_cache))
wenzelm@65629
  1077
                  else Nil)
wenzelm@65626
  1078
            session_name -> session_entry
wenzelm@65621
  1079
          }).toMap
wenzelm@65621
  1080
        })
wenzelm@65621
  1081
      Build_Info(sessions)
wenzelm@65621
  1082
    }
wenzelm@65595
  1083
  }
wenzelm@64045
  1084
}