X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;f=bin%2Fdebbugs-loadsql;h=f452431d3bc8b34dc01f53dcd91dc1e8cc2029de;hb=5a7f0c77d7d30a0b9052f42796c8fead5202670a;hp=c91826677ffb5d666019ab85c3ef37ff0c7f8e98;hpb=d7bda9dac45d437e0f5297e0f09be5f70269c9c3;p=debbugs.git

diff --git a/bin/debbugs-loadsql b/bin/debbugs-loadsql
index c918266..f452431 100755
--- a/bin/debbugs-loadsql
+++ b/bin/debbugs-loadsql
@@ -17,10 +17,13 @@ debbugs-loadsql -- load debbugs sql database
 
 =head1 SYNOPSIS
 
-debbugs-loadsql [options]
+debbugs-loadsql [options] [subcommand]
 
+ Subcommands:
+  bugs help versions configuration
+  suites logs packages debinfo
  Options:
-  --quick, -q only load changed bugs
+  --quick, -q only load changed things
   --progress Show progress bar
   --service, -s service name
   --sysconfdir, -c postgresql service config dir
@@ -37,15 +40,45 @@ Display this manual
 
 =head2 bugs
 
-Add bugs
+Add bugs (subject, number, etc) to the database
+
+   --preload create all bugs first, then add information
 
 =head2 versions
 
-Add versions
+Add version descendant information (which version is based on which version) to
+the database
 
 =head2 maintainers
 
-Add source maintainers
+Add source maintainers to the BTS
+
+=head2 configuration
+
+Add debbugs configuration information (tags, severity, etc)
+
+=head2 suites
+
+Add suite information from ftp distribution
+
+  --ftpdists location of FTP mirror
+
+=head2 logs
+
+Add bug logs
+
+=head2 packages
+
+Add package information from the ftp archive
+
+  --ftpdists location of FTP mirror
+  --suites Suite to operate on
+
+=head2 debinfo
+
+Add package information from a debinfo file
+
+  --null -0 names of debinfo files are null separated
 
 =head1 OPTIONS
 
@@ -99,12 +132,12 @@ Display this manual.
 use vars qw($DEBUG);
 
 use Debbugs::Common (qw(checkpid lockpid get_hashname getparsedaddrs getbugcomponent make_list getsourcemaintainers),
-		     qw(hash_slice));
+		     qw(hash_slice open_compressed_file),);
 use Debbugs::Config qw(:config);
 use Debbugs::Status qw(read_bug split_status_fields);
 use Debbugs::Log;
 use Debbugs::DB;
-use Debbugs::DB::Load qw(load_bug handle_load_bug_queue :load_package :load_suite);
+use Debbugs::DB::Load qw(:load_bug :load_package :load_suite);
 use DateTime;
 use File::stat;
 use File::Basename;
@@ -113,6 +146,7 @@ use IO::Dir;
 use IO::File;
 use IO::Uncompress::AnyUncompress;
 use Encode qw(decode_utf8);
+use List::MoreUtils qw(natatime);
 
 my %options =
     (debug           => 0,
@@ -161,6 +195,8 @@ my %subcommands =
                  },
      'logs' => {function => \&add_logs,
                },
+     'bugs_and_logs' => {function => \&add_bugs_and_logs,
+			},
      'packages' => {function => \&add_packages,
 		    arguments => {'ftpdists=s' => 1,
 				  'suites=s@' => 0,
@@ -244,8 +280,10 @@ sub add_bugs {
 		  'summary',
 		  undef,
 		  sub {
-		      push @bugs,shift;
-		  });
+		    push @bugs,@_;
+		  },
+		  10000
+		 );
 	$s->resultset('Bug')->quick_insert_bugs(@bugs);
     }
     walk_bugs([(@{$argv}?@{$argv} : $initialdir)],
@@ -253,31 +291,30 @@ sub add_bugs {
               'summary',
               $verbose,
               sub {
-                  my $bug = shift;
-                  my $stat = stat(getbugcomponent($bug,'summary',$initialdir));
-                  if (not defined $stat) {
-                      print STDERR "Unable to stat $bug $!\n";
-                      next;
-                  }
-                  if ($options{quick}) {
-                      my $rs = $s->resultset('Bug')->search({bug=>$bug})->single();
-                      next if defined $rs and $stat->mtime < $rs->last_modified()->epoch();
-                  }
-                  my $data = read_bug(bug => $bug,
-                                      location => $initialdir);
-                  eval {
-                      load_bug(db => $s,
-                               data => split_status_fields($data),
-                               tags => \%tags,
-                               severities => \%severities,
-                               queue => \%queue);
-                  };
-                  if ($@) {
-                      use Data::Dumper;
-                      print STDERR Dumper($data) if $DEBUG;
-                      die "failure while trying to load bug $bug\n$@";
-                  }
-              }
+		my @bugs = @_;
+		my @bugs_to_update;
+		if ($options{quick}) {
+		    @bugs_to_update =
+			bugs_to_update($s,$initialdir,@bugs);
+		} else {
+		    @bugs_to_update = @bugs;
+		}
+		eval {
+		  $s->txn_do(sub {
+			       for my $bug (@bugs_to_update) {
+				 load_bug(db => $s,
+					  bug => $bug,
+					  tags => \%tags,
+					  severities => \%severities,
+					  queue => \%queue);
+			       }
+			     });
+		};
+		if ($@) {
+		  die "failure while trying to load bug: $@";
+		}
+              },
+	      50
              );
     handle_load_bug_queue(db => $s,
                           queue => \%queue);
@@ -308,19 +345,21 @@ sub add_versions {
             my $sp;
             if (not defined $src_pkgs{$versions[$i][0]}) {
                 $src_pkgs{$versions[$i][0]} =
-                    $s->resultset('SrcPkg')->find_or_create({pkg => $versions[$i][0]});
+                    $s->resultset('SrcPkg')->
+		    get_src_pkg_id($versions[$i][0]);
             }
             $sp = $src_pkgs{$versions[$i][0]};
             # There's probably something wrong if the source package
             # doesn't exist, but we'll skip it for now
-            next unless defined $sp;
-            my $sv = $s->resultset('SrcVer')->find({src_pkg=>$sp->id(),
+            last if not defined $sp;
+            my $sv = $s->resultset('SrcVer')->find({src_pkg=>$sp,
                                                     ver => $versions[$i][1],
                                                    });
+	    last if not defined $sv;
             if (defined $ancestor_sv and defined $sv and not defined $sv->based_on()) {
-                $sv->update({based_on => $ancestor_sv->id()})
+                $sv->update({based_on => $ancestor_sv})
             }
-            $ancestor_sv = $sv;
+            $ancestor_sv = $sv->id();
         }
         $p->update() if $p;
     }
@@ -333,63 +372,54 @@ sub add_debinfo {
     my @files = @{$argv};
     if (not @files) {
        {
-           if ($opts->{0}) {
-               local $/ = "\0";
-           }
+	   local $/ = "\n";
+           local $/ = "\0" if $opts->{0};
            while (<STDIN>) {
+	       s/\n$// unless $opts->{0};
+	       s/\0$// if $opts->{0};
                push @files, $_;
            }
        }
     }
     return unless @files;
     my $s = db_connect($options);
-    my %arch;
     $p->target(scalar @files) if $p;
-    for my $file (@files) {
-        my $fh = IO::File->new($file,'r') or
-            die "Unable to open $file for reading: $!";
-        my $f_stat = stat($file);
-        while (<$fh>) {
-            chomp;
-            next unless length $_;
-            my ($binname, $binver, $binarch, $srcname, $srcver) = split;
-            # if $srcver is not defined, this is probably a broken
-            # .debinfo file [they were causing #686106, see commit
-            # 49c85ab8 in dak.] Basically, $binarch didn't get put into
-            # the file, so we'll fudge it from the filename.
-            if (not defined $srcver) {
-                ($srcname,$srcver) = ($binarch,$srcname);
-                ($binarch) = $file =~ /_([^\.]+)\.debinfo/;
-            }
-            my $sp = $s->resultset('SrcPkg')->find_or_create({pkg => $srcname});
-            # update the creation date if the data we have is earlier
-            my $ct_date = DateTime->from_epoch(epoch => $f_stat->ctime);
-            if ($ct_date < $sp->creation) {
-                $sp->creation($ct_date);
-                $sp->last_modified(DateTime->now);
-                $sp->update;
-            }
-            my $sv = $s->resultset('SrcVer')->find_or_create({src_pkg =>$sp->id(),
-                                                              ver => $srcver});
-            if (not defined $sv->upload_date() or $ct_date < $sv->upload_date()) {
-                $sv->upload_date($ct_date);
-                $sv->update;
-            }
-            my $arch;
-            if (defined $arch{$binarch}) {
-                $arch = $arch{$binarch};
-            } else {
-                $arch = $s->resultset('Arch')->find_or_create({arch => $binarch});
-                $arch{$binarch} = $arch;
-            }
-            my $bp = $s->resultset('BinPkg')->find_or_create({pkg => $binname});
-            $s->resultset('BinVer')->find_or_create({bin_pkg => $bp->id(),
-                                                     src_ver => $sv->id(),
-                                                     arch    => $arch->id(),
-                                                     ver        => $binver,
-                                                    });
-        }
-        $p->update() if $p;
+    my $it = natatime 100, @files;
+    while (my @v = $it->()) {
+	my %cache;
+	my @debinfos;
+	for my $file (@v) {
+	    my $fh = IO::File->new($file,'r') or
+		die "Unable to open $file for reading: $!";
+	    my $f_stat = stat($file);
+	    my $ct_date = DateTime->from_epoch(epoch => $f_stat->ctime);
+	    while (<$fh>) {
+		chomp;
+		next unless length $_;
+		my ($binname, $binver, $binarch, $srcname, $srcver) = split;
+		# if $srcver is not defined, this is probably a broken
+		# .debinfo file [they were causing #686106, see commit
+		# 49c85ab8 in dak.] Basically, $binarch didn't get put into
+		# the file, so we'll fudge it from the filename.
+		if (not defined $srcver) {
+		    ($srcname,$srcver) = ($binarch,$srcname);
+		    ($binarch) = $file =~ /_([^\.]+)\.debinfo/;
+		}
+		if (not defined $srcver) {
+		    print STDERR "malformed debinfo (no srcver): $file\n";
+		    next;
+		}
+		push @debinfos,
+		    [$binname,$binver,$binarch,$srcname,$srcver,$ct_date];
+	    }
+	}
+	$s->txn_do(
+	    sub {
+		for my $di (@debinfos) {
+		    Debbugs::DB::Load::load_debinfo($s,@{$di}[0..5],\%cache);
+		}
+	    });
+	$p->update($p->last_update()+@v) if $p;
     }
     $p->remove() if $p;
 }
@@ -440,25 +470,35 @@ sub add_configuration {
 
     # tags
     # add all tags
+    my %tags;
+    for my $tag (@{$config{tags}}) {
+	$tags{$tag} = 1;
+	$s->resultset('Tag')->find_or_create({tag => $tag});
+    }
     # mark obsolete tags
+    for my $tag ($s->resultset('Tag')->search_rs()->all()) {
+	next if exists $tags{$tag->tag};
+	$tag->obsolete(1);
+	$tag->update;
+    }
 
     # severities
     my %sev_names;
-    my $order = 0;
-    for my $sev_name (@{$config{severities}}) {
+    my $order = -1;
+    for my $sev_name (($config{default_severity},@{$config{severity_list}})) {
         # add all severitites
         my $sev = $s->resultset('Severity')->find_or_create({severity => $sev_name});
         # mark strong severities
         if (grep {$_ eq $sev_name} @{$config{strong_severities}}) {
             $sev->strong(1);
         }
-        $sev->order($order);
+        $sev->ordering($order);
         $sev->update();
         $order++;
         $sev_names{$sev_name} = 1;
     }
     # mark obsolete severities
-    for my $sev ($s->resultset('Severity')->find()) {
+    for my $sev ($s->resultset('Severity')->search_rs()->all()) {
         next if exists $sev_names{$sev->severity()};
         $sev->obsolete(1);
         $sev->update();
@@ -509,7 +549,18 @@ sub add_logs {
               $verbose,
               sub {
                   my $bug = shift;
-                  eval { 
+		  my $stat = stat(getbugcomponent($bug,'log',$initialdir));
+                  if (not defined $stat) {
+                      print STDERR "Unable to stat $bug $!\n";
+                      next;
+                  }
+                  if ($options{quick}) {
+                      my $rs = $s->resultset('Bug')->
+			  search({id=>$bug})->single();
+                      return if defined $rs and
+			  $stat->mtime <= $rs->last_modified()->epoch();
+                  }
+                  eval {
                       load_bug_log(db => $s,
                                    bug => $bug);
                   };
@@ -519,6 +570,81 @@ sub add_logs {
               });
 }
 
+sub add_bugs_and_logs {
+    my ($options,$opts,$p,$config,$argv) = @_;
+
+    chdir($config->{spool_dir}) or
+        die "chdir $config->{spool_dir} failed: $!";
+
+    my $verbose = $options->{debug};
+
+    my $initialdir = "db-h";
+
+    if (defined $argv->[0] and $argv->[0] eq "archive") {
+        $initialdir = "archive";
+    }
+    my $s = db_connect($options);
+
+    my %tags;
+    my %severities;
+    my %queue;
+
+    walk_bugs([(@{$argv}?@{$argv} : $initialdir)],
+              $p,
+              'summary',
+              $verbose,
+              sub {
+                  my @bugs = @_;
+		  my @bugs_to_update;
+		  if ($options{quick}) {
+		      @bugs_to_update =
+			  bugs_to_update($s,$initialdir,@bugs);
+		  } else {
+		      @bugs_to_update = @bugs;
+		  }
+		  eval {
+		      $s->txn_do(sub {
+				     for my $bug (@bugs_to_update) {
+					 load_bug(db => $s,
+						  bug => $bug,
+						  tags => \%tags,
+						  severities => \%severities,
+						  queue => \%queue);
+				     }
+				 });
+		  };
+		  if ($@) {
+		      die "failure while trying to load bug: $@";
+		  }
+		  for my $bug (@bugs) {
+		      my $stat = stat(getbugcomponent($bug,'log',$initialdir));
+		      if (not defined $stat) {
+			  print STDERR "Unable to stat $bug $!\n";
+			  next;
+		      }
+		      if ($options{quick}) {
+			  my $rs = $s->resultset('Bug')->
+			      search({id=>$bug})->single();
+			  return if defined $rs and
+			      $stat->mtime <= $rs->last_modified()->epoch();
+		      }
+		      eval {
+			  load_bug_log(db => $s,
+				       bug => $bug);
+		      };
+		      if ($@) {
+			  die "failure while trying to load bug log $bug\n$@";
+		      }
+		  }
+              },
+	      50
+	     );
+    handle_load_bug_queue(db=>$s,
+			  queue => \%queue,
+			 );
+
+}
+
 sub add_packages {
     my ($options,$opts,$p,$config,$argv) = @_;
 
@@ -630,27 +756,6 @@ sub db_connect {
         die "Unable to connect to database: ";
 }
 
-sub open_compressed_file {
-    my ($file) = @_;
-    my $fh;
-    my $mode = '<:encoding(UTF-8)';
-    my @opts;
-    if ($file =~ /\.gz$/) {
-	$mode = '-|:encoding(UTF-8)';
-	push @opts,'gzip','-dc';
-    }
-    if ($file =~ /\.xz$/) {
-	$mode = '-|:encoding(UTF-8)';
-	push @opts,'xz','-dc';
-    }
-    if ($file =~ /\.bz2$/) {
-	$mode = '-|:encoding(UTF-8)';
-	push @opts,'bzip2','-dc';
-    }
-    open($fh,$mode,@opts,$file);
-    return $fh;
-}
-
 sub read_release_file {
     my ($file) = @_;
     # parse release
@@ -682,12 +787,13 @@ sub read_release_file {
 }
 
 sub walk_bugs {
-    my ($dirs,$p,$what,$verbose,$sub) = @_;
+    my ($dirs,$p,$what,$verbose,$sub,$n) = @_;
     my @dirs = @{$dirs};
     my $tot_dirs = @dirs;
     my $done_dirs = 0;
     my $avg_subfiles = 0;
     my $completed_files = 0;
+    $n //= 1;
     while (my $dir = shift @dirs) {
         printf "Doing dir %s ...\n", $dir if $verbose;
 
@@ -707,16 +813,34 @@ sub walk_bugs {
         $avg_subfiles = ($avg_subfiles * $done_dirs + @list) / ($done_dirs+1);
         $done_dirs += 1;
 
-        for my $bug (@list) {
-            $completed_files++;
-            $p->update($completed_files) if $p;
-            print "Up to $completed_files bugs...\n" if ($completed_files % 100 == 0 && $verbose);
-            $sub->($bug);
+	my $it = natatime $n,@list;
+	while (my @bugs = $it->()) {
+	  $sub->(@bugs);
+	  $completed_files += scalar @bugs;
+	  $p->update($completed_files) if $p;
+	  print "Up to $completed_files bugs...\n"
+	    if ($completed_files % 100 == 0 && $verbose);
         }
     }
     $p->remove() if $p;
 }
 
 
+sub bugs_to_update {
+    my ($s,$initialdir,@bugs) = @_;
+    my @bugs_to_update;
+    for my $bug (@bugs) {
+	my $stat = stat(getbugcomponent($bug,'summary',$initialdir));
+	if (not defined $stat) {
+	    print STDERR "Unable to stat $bug $!\n";
+	    next;
+	}
+	my $rs = $s->resultset('Bug')->search({id=>$bug})->single();
+	next if defined $rs and $stat->mtime <= $rs->last_modified()->epoch();
+	push @bugs_to_update, $bug;
+    }
+    @bugs_to_update;
+}
+
 
 __END__