X-Git-Url: http://www.chiark.greenend.org.uk/ucgi/~yarrgweb/git?p=ypp-sc-tools.db-live.git;a=blobdiff_plain;f=yarrg%2Fyppedia-chart-parser;h=09e8fe8b1b82aa22e7140bf5b6ff3069310a499d;hp=2da027e14829255b308d8e2071e1976285a61c76;hb=0a4c9fe7e73cb5802471cbc0d975d3c7daa6513c;hpb=8fb8541655ff98c2d5cd80b1adb9effebed6e93c diff --git a/yarrg/yppedia-chart-parser b/yarrg/yppedia-chart-parser index 2da027e..09e8fe8 100755 --- a/yarrg/yppedia-chart-parser +++ b/yarrg/yppedia-chart-parser @@ -4,28 +4,38 @@ use strict (qw(vars)); use warnings; use Graph::Undirected; - +use Commods; use CommodsDatabase; +my $ocean= 'Midnight'; + + my $widists= Graph::Undirected->new(); my $wiarchs= Graph::Undirected->new(); my @wiarchlabels; my %wiisland2node; my %winode2island; -my %wiisland2arch; my %winode2lines; my %wiccix2arch; +my $wialldists; my $dbdists= Graph::Undirected->new(); my %dbisland2arch; -my %msgcount; -sub perr ($$) { print STDERR "$_[0]: $_[1]\n"; $msgcount{$_[0]}++; } -sub warning ($) { perr("warning",$_[0]); } -sub error ($) { perr("error", $_[0]); } -sub change ($) { perr("change", $_[0]); } +my %msgs; +sub pmsg ($$) { push @{ $msgs{$_[0]} }, "$_[0]: $_[1]\n"; } +sub warning ($) { pmsg("warning",$_[0]); } +sub error ($) { pmsg("error", $_[0]); } +sub change ($) { pmsg("change", $_[0]); } +sub print_messages () { + foreach my $k (qw(change warning error)) { + my $m= $msgs{$k}; + next unless $m; + print sort @$m or die $!; + } +} -if ($ARGV[0] eq '--debug') { +if (@ARGV && $ARGV[0] eq '--debug') { shift @ARGV; open DEBUG, ">&STDOUT" or die $!; select(DEBUG); $|=1; @@ -45,7 +55,7 @@ sub nn_xy ($$) { return $n; } -sub parse_yppedia_map () { +sub yppedia_chart_parse () { # We don't even bother with tag soup; instead we do line-oriented parsing. while (<>) { @@ -101,7 +111,7 @@ sub parse_yppedia_map () { } } -sub parse_database_map () { +sub database_fetch_ocean () { my ($row,$sth); $sth= $dbh->prepare('SELECT islandname, archipelago FROM islands'); $sth->execute(); @@ -120,22 +130,21 @@ sub parse_database_map () { } } -sub process_yppedia_graphs () { +sub yppedia_graphs_prune_boring () { # Prune the LP database by eliminating boring intermediate vertices foreach my $delete ($widists->vertices()) { next if exists $winode2island{$delete}; my @neigh= $widists->neighbours($delete); next unless @neigh==2; -# my @aneigh= $wiarchs->has_vertex($delete) -# ? $wiarchs->neighbours($delete) : (); -# next unless @aneigh==0 || @aneigh==2; my $weight= 0; map { $weight += $widists->get_edge_weight($delete, $_) } @neigh; $widists->add_weighted_edge(@neigh, $weight); $widists->delete_vertex($delete); printf DEBUG "%-5s elide %5s %-5s %2d\n", $delete, @neigh, $weight; } +} +sub yppedia_graphs_check () { # Check that it's connected. foreach my $cc ($widists->connected_components()) { next if 2*@$cc > $widists->vertices(); @@ -146,12 +155,30 @@ sub process_yppedia_graphs () { } warning($m); } +} - # Compute all-pairs-shortest-paths on dist, which is the - # actual distances between all LPs. - my $wialldists= $widists->APSP_Floyd_Warshall(); +sub yppedia_archs_sourceinfo () { + # Assign archipelagoes according to the source-info file + foreach my $arch (sort keys %{ $oceans{$ocean} }) { + foreach my $islename (sort keys %{ $oceans{$ocean}{$arch} }) { + my $islenode= $wiisland2node{$islename}; + defined $islenode or + error("island $islename in source-info but not in WP map"); + my $ccix= $wiarchs->connected_component_by_vertex($islenode); + my $oldarch= $wiccix2arch{$ccix}; + error("island $islename in $arch in source-info". + " connected to $oldarch as well") + if defined $oldarch && $oldarch ne $arch; + printf DEBUG "%-5s force-island-arch cc%-2d %-10s %s\n", + $islenode, $ccix, $arch, $islename; + $wiccix2arch{$ccix}= $arch; + } + } +} +sub yppedia_archs_chart_labels () { # Assign archipelago labels to groups of islands + # foreach my $label (@wiarchlabels) { my ($ax,$ay,$arch) = @$label; my $best_ccmulti= -1; @@ -168,8 +195,8 @@ sub process_yppedia_graphs () { my $d2= ($vx-$ax)*($vx-$ax) + ($vy-$ay)*($vy-$ay); my $cmp= $ccmulti <=> $best_ccmulti || $best_d2 <=> $d2; - printf DEBUG "%2d,%-2d arch-island-search %5s d2=%4d ccix=%-2d". - " cc=%2d ccmulti=%d cmp=%-2d %s\n", + printf DEBUG "%2d,%-2d arch-island-search %5s d2=%4d cc%-2d". + " #cc=%2d ccmulti=%d cmp=%2d %s\n", $ax,$ay, $vertex, $d2, $ccix, scalar(@cc), $ccmulti, $cmp, $winode2island{$vertex}; next unless $cmp > 0; @@ -178,15 +205,16 @@ sub process_yppedia_graphs () { $best_ccmulti= $ccmulti; } die 'no island vertices?!' unless defined $best_n; - printf DEBUG "%2d,%-2d arch-island-select %-5s d2=%-2d %-10s %s\n", - $ax,$ay, $best_n, $best_d2, $arch, $winode2island{$best_n}; my $ccix= $wiarchs->connected_component_by_vertex($best_n); + printf DEBUG + "%2d,%-2d arch-island-select %-5s d2=%4d cc%-2d %-10s %s\n", + $ax,$ay, $best_n, $ccix, $best_d2, $arch, $winode2island{$best_n}; my $desc= join "\n", map { my $in= $winode2island{$_}; " LP $_". (defined $in ? ", $in" : ""); } sort $wiarchs->connected_component_by_index($ccix); - if (exists $wiccix2arch{$ccix}) { + if (exists $wiccix2arch{$ccix} and $wiccix2arch{$ccix} ne $arch) { error("archipelago determination failed, wrongly merged:\n". " archipelago $arch\n". " archipelago $wiccix2arch{$ccix}\n". @@ -196,7 +224,9 @@ sub process_yppedia_graphs () { $wiccix2arch{$ccix}= $arch; # print "$ccix $arch ::\n$desc\n"; } +} +sub yppedia_archs_fillbynearest() { # Assign islands not labelled above to archipelagoes. # # We do this by, for each connected component (set of islands @@ -204,49 +234,58 @@ sub process_yppedia_graphs () { # connected component which has already been assigned an arch. # `Nearest' means shortest distance of unpurchaseable charts, in # leagues. - - # fixme need some hints - + # # we need only consider vertices which weren't `boring intermediate # vertices' (removed during optimisation as being of order 2) my @ccs_useful= map { [ grep { $widists->has_vertex($_) } @$_ ] } $wiarchs->connected_components(); + my @assignments; + foreach my $sourceccix (0..$#ccs_useful) { next if defined $wiccix2arch{$sourceccix}; + next unless $ccs_useful[$sourceccix]; - my $sourcecc= $ccs_useful[$sourceccix]; - my @islandnodes= grep { $winode2island{$_} } @$sourcecc; + my @sourcecc= $wiarchs->connected_component_by_index($sourceccix); + my @islandnodes= grep { $winode2island{$_} } @sourcecc; next unless @islandnodes; # don't care, then + foreach my $islandnode (@islandnodes) { + printf DEBUG "%-5s arch-join-need cc%-2d %s\n", + $islandnode, $sourceccix, $winode2island{$islandnode}; + } my $best_dist= 9999999; - my $best_target; + my ($best_target, $best_targetccix, $best_source); foreach my $targetccix (0..$#ccs_useful) { next unless defined $wiccix2arch{$targetccix}; # not helpful - my $targetcc= $ccs_useful[$targetccix]; - foreach my $target (@$targetcc) { - foreach my $source (@$sourcecc) { + next unless $ccs_useful[$targetccix]; + foreach my $target ($wiarchs-> + connected_component_by_index($targetccix)) { + foreach my $source (@sourcecc) { my $target_dist= $wialldists->path_length($target,$source); + next unless defined $target_dist; next if $target_dist >= $best_dist; $best_dist= $target_dist; + $best_source= $source; $best_target= $target; + $best_targetccix= $targetccix; } } } -# die "no possible target ?!" unless defined $best_target; -# -# printf DEBUG " -# -# foreach my $node (sort keys %winode2island) { -# my $island= $winode2island{$node}; -# my $arch= winode2arch($node); -# next if defined $arch; -# my $ccix= $wiarchs->connected_component_by_vertex($node); -# my @cc= $wiarchs->connected_component_by_index($ccix); -# @cc= grep { defined $winode2island{$_} } @cc; -# # We search for the best: -# # - member of this connected component + die "no possible target ?!" unless defined $best_target; + + my $arch= $wiccix2arch{$best_targetccix}; + my $best_island= $winode2island{$best_target}; + printf DEBUG "%-5s arch-join-to %-5s dist=%2d cc%-2d %-10s %s\n", + $best_source, $best_target, $best_dist, + $best_targetccix, $arch, + defined($best_island) ? $best_island : "-"; + + push @assignments, [ $sourceccix, $arch ]; + } + foreach my $assign (@assignments) { + $wiccix2arch{$assign->[0]}= $assign->[1]; } } @@ -294,9 +333,42 @@ sub compare_island_lists () { } } -db_setocean('Midnight'); +sub compare_distances () { + foreach my $ia (sort keys %dbisland2arch) { + my $na= $wiisland2node{$ia}; + next unless defined $na; + foreach my $ib (sort keys %dbisland2arch) { + next unless $ia le $ib; # do every pair only once + my $nb= $wiisland2node{$ib}; + next unless defined $nb; + my $dbdist= $dbdists->get_edge_weight($ia,$ib); + my $widist= $wialldists->path_length($na,$nb); + if (!defined $dbdist) { + change(sprintf "define distance %2d for %s..%s", + $widist, $ia,$ib); + } elsif ($dbdist != $widist) { + change(sprintf "change distance %2d to %2d for %s..%s", + $dbdist, $widist, $ia,$ib); + } + } + } +} + +parse_info_serverside(); + +db_setocean($ocean); db_connect(); -parse_yppedia_map(); -parse_database_map(); -process_yppedia_graphs(); +database_fetch_ocean(); + +yppedia_chart_parse(); +yppedia_graphs_prune_boring(); +yppedia_graphs_check(); +yppedia_archs_sourceinfo(); +$wialldists= $widists->APSP_Floyd_Warshall(); +yppedia_archs_chart_labels(); +yppedia_archs_fillbynearest(); + compare_island_lists(); +compare_distances(); + +print_messages();