Now working full non-blocking (maybe leaky here and there)
diff --git a/lib/Korap.pm b/lib/Korap.pm
index 05135f2..30318bc 100644
--- a/lib/Korap.pm
+++ b/lib/Korap.pm
@@ -27,12 +27,9 @@
Notifications
Number::Commify
Search
- KorapInfo
KorapHelpers
KorapTagHelpers
/) {
- # Oro::Account
- # Oro::Account::ConfirmMail
$self->plugin($_);
};
@@ -62,10 +59,10 @@
$self->asset(
'korap.js' => (
- '/js/d3.v3.min.js',
+# '/js/d3.v3.min.js',
# '/js/dagre-d3.min.js',
- '/js/dagre-d3.js',
- '/js/translateTree.js',
+# '/js/dagre-d3.js',
+# '/js/translateTree.js',
'/js/jquery-2.0.0.min.js', # Temp
'/js/tutorialCookie.js',
'/js/translateTable.js',
@@ -86,26 +83,17 @@
# Routes
my $r = $self->routes;
- # User account management
- # $r->route('/login')->acct('login');
- # $r->route('/login/forgotten')->acct('forgotten');
- # $r->route('/login/remove')->acct('remove');
- # $r->route('/register')->acct('register');
- # $r->route('/logout')->acct('logout');
- # $r->route('/preferences')->acct('preferences');
-
# Base search route
- $r->get('/')->to('search#remote')->name('index');
+ $r->get('/')->to('search#query')->name('index');
+
+ # Get match information
+ my $corpus = $r->route('/corpus/:corpus_id');
+ my $doc = $corpus->route('/#doc_id');
+ my $match = $doc->route('/:match_id')->to('search#match_info')->name('match');
# Tutorial data
$r->get('/tutorial')->to('tutorial#page', tutorial => 'index');
$r->get('/tutorial/(*tutorial)')->to('tutorial#page')->name('tutorial');
-
- # Collection data
- my $collection = $r->bridge('/collection');
- $collection->to('info#about_collection');
- my $collection_id = $collection->bridge('/:collection_id');
- # stats
};
@@ -113,32 +101,3 @@
__END__
-
-
-
-# No shortcut!
-
-# $collection_id->search;
-
- # Corpus data
- my $corpus_res = $r->route('/corpus');
- my $corpus = $corpus_res->route('/:corpus_id');
-
- # Todo: Stats
-# $corpus->search->name('search_corpus');
-# my $doc = $corpus->route('/#doc_id');
-# $doc->search->name('search_document');
-
- # Match data
- my $match = $doc->route('/:match_id');
- $match->route->to('info#about_match')->name('match');
- # my $match_foundry = $match->route('/:foundry');
- # $match_foundry->route->to('info#about_match');
- # $match_foundry->route('/:layer')->to('info#about_match');
-
- # Utilities
- # $r->get('/util/query')->to('search#query');
-};
-
-
-1;
diff --git a/lib/Korap/API.pm b/lib/Korap/API.pm
index 59ea869..70ec277 100644
--- a/lib/Korap/API.pm
+++ b/lib/Korap/API.pm
@@ -8,7 +8,8 @@
# Todo: Add fixtures
# Todo: Support search in corpus and virtualcollection
-
+# Todo: Support caching everywhere!
+# Todo: Correct use of stash info everywhere!
# Register the plugin
sub register {
@@ -34,43 +35,30 @@
my $self = shift;
my $index = shift;
+ # Get controller
my $c = $index->controller;
- # No query defined
- return unless $index->query;
-
# If there is a callback, do async
my $cb = pop if ref $_[-1] && ref $_[-1] eq 'CODE';
- my %param = @_;
+ # No query defined
+ unless ($index->query) {
+ return $cb->($index) if $cb;
+ return;
+ };
- # Set cutoff from param
- $index->cutoff(delete $param{cutoff});
-
- # Set query language
- $index->query_language(delete $param{query_language} // 'poliqarp');
-
- # TODO: This should also be a query parameter
- $index->no_cache(1) if $param{no_cache} or $c->param('no_cache');
-
- my %query;
- $query{q} = $index->query;
- $query{ql} = $index->query_language;
- $query{page} = $index->start_page;
- $query{count} = $index->items_per_page;
- $query{cutoff} = 'true' if $index->cutoff;
-
- # Todo: support corpus and collection
- # Create query url
- my $url = Mojo::URL->new($index->api);
- $url->path('search');
- $url->query(\%query);
+ # Get query url
+ my $url = _query_url($index, @_);
# Cache based on URL
$index->_api_cache('total-' . $url->to_string);
+ my %param = @_;
# Set context based on parameter
- $url->query({ context => $c->param('context') // 'paragraph' });
+ $url->query({ context => $param{'context'} // 'paragraph' });
+
+ # Set path to search
+ $url->path('search');
# Check cache for total results
my $total_results;
@@ -90,43 +78,186 @@
$index->api_request($url->to_string);
# Create new user agent and set timeout to 2 minutes
- my $ua = Mojo::UserAgent->new;
+ my $ua = $c->ua; # Mojo::UserAgent->new;
$ua->inactivity_timeout(120);
- # Denugging
+ # Debugging
$c->app->log->debug('Search for ' . $index->api_request);
# Search non-blocking
if ($cb) {
- # Non-blocking request
$ua->get(
- $url->to_string => sub {
- my ($ua, $tx) = @_;
- unless ($tx->success) {
- if (my $e = $tx->error) {
- warn 'Problem: ' . $e->{message};
- return $cb->($index);
- };
- };
-
- $self->_process_response($index, pop);
+ $url => sub {
+ my $tx = pop;
+ $self->_process_response('matches', $index, $tx);
return $cb->($index);
});
-
- Mojo::IOLoop->wait unless Mojo::IOLoop->is_running;
}
-
# Search blocking
else {
my $tx = $ua->get($url);
- return $self->_process_response($index, $tx);
+ return $self->_process_response('matches', $index, $tx);
};
};
+# Trace query serialization
+sub trace {
+ my $self = shift;
+ my $index = shift;
+
+ # Get controller
+ my $c = $index->controller;
+
+ # If there is a callback, do async
+ my $cb = pop if ref $_[-1] && ref $_[-1] eq 'CODE';
+
+ my %param = @_;
+
+ # No query defined
+ unless ($index->query(delete $param{query})) {
+ return $cb->($index) if $cb;
+ return;
+ };
+
+ # Get query url
+ my $url = _query_url($index, @_);
+
+ $url->path('search');
+
+ # Create new user agent and set timeout to 30 seconds
+ my $ua = $c->ua; # Mojo::UserAgent->new;
+ $ua->inactivity_timeout(30);
+
+ # Build transaction
+ my $tx = $ua->build_tx(TRACE => $url);
+
+ # non-blocking
+ if ($cb) {
+
+ # Trace non-blocking
+ $ua->start(
+ $tx => sub {
+ $self->_process_response('trace', $index, pop);
+ return $cb->($index);
+ });
+ }
+ # Trace blocking
+ else {
+ my $tx = $ua->start($url);
+ return $self->_process_response('trace', $index, $tx);
+ };
+};
+
+
+# Get match info
+sub match {
+ my $self = shift;
+ my $index = shift;
+
+ # Get controller
+ my $c = $index->controller;
+
+ # If there is a callback, do async
+ my $cb = pop if ref $_[-1] && ref $_[-1] eq 'CODE';
+
+ my %param = @_;
+
+ my $url = Mojo::URL->new($index->api);
+
+ # Use hash slice to create path
+ $url->path(join('/', 'corpus', @param{qw/corpus_id doc_id match_id/}, 'matchInfo'));
+
+ # Build match id
+ # $match = 'match-' . $corpus . '!' . $corpus . '_' . $doc . '-' . $match;
+
+ my %query;
+ $query{foundry} = $param{foundry};
+ $query{layer} = $param{layer} if defined $param{layer};
+ $query{spans} = $param{spans} ? 'true' : 'false';
+
+ # Add query
+ $url->query(\%query);
+
+ $c->app->log->debug('Match info: ' . $url);
+
+ # Create new user agent and set timeout to 30 seconds
+ my $ua = $c->ua; # Mojo::UserAgent->new;
+ $ua->inactivity_timeout(30);
+
+ # non-blocking
+ if ($cb) {
+ $ua->get(
+ $url => sub {
+ my $tx = pop;
+ $self->_process_response('match', $index, $tx);
+ return $cb->($index);
+ });
+ }
+
+ # Match info blocking
+ else {
+ my $tx = $ua->get($url);
+ return $self->_process_response('match', $index, $tx);
+ };
+};
+
+
+# Trace query serialization
+sub resource {
+ my $self = shift;
+ my $index = shift;
+
+ # Get controller
+ my $c = $index->controller;
+
+ # If there is a callback, do async
+ my $cb = pop if ref $_[-1] && ref $_[-1] eq 'CODE';
+
+ my %param = @_;
+
+ # Rename info endpoints regarding resource
+ my $type = $param{type} // 'collection';
+ $type = 'virtualcollection' if $type eq 'collection';
+
+ my $url = Mojo::URL->new($index->api)->path($type);
+
+ $c->app->log->debug('Get resource info on '. $url);
+
+ # Check for cached information
+ if (my $json = $c->chi->get($url->to_string)) {
+
+ # TODO: That's unfortunate, as it prohibits multiple resources
+ $c->app->log->debug('Get resource info from cache');
+ $c->stash('search.resource' => $json);
+ return $cb->($index) if $cb;
+ return $json;
+ };
+
+ $c->stash('search._resource_cache' => $url->to_string);
+
+ # Create new user agent and set timeout to 30 seconds
+ my $ua = $c->ua; # Mojo::UserAgent->new;
+ $ua->inactivity_timeout(30);
+
+ if ($cb) {
+ $ua->get(
+ $url => sub {
+ $self->_process_response('resource', $index, pop);
+ return $cb->($index);
+ })
+ }
+ else {
+ my $tx = $ua->get($url);
+ $self->_process_response('resource', $index, $tx);
+ };
+};
+
+
+# Process response - especially error messages etc.
sub _process_response {
- my ($self, $index, $tx) = @_;
+ my ($self, $type, $index, $tx) = @_;
my $c = $index->controller;
# An error has occurded
@@ -152,33 +283,21 @@
return;
};
- # Reformat benchmark counter
- my $benchmark = $json->{benchmark};
- if ($benchmark && $benchmark =~ s/\s+(m)?s$//) {
- $benchmark = sprintf("%.2f", $benchmark) . ($1 ? $1 : '') . 's';
- };
- # Set benchmark
- $index->benchmark($benchmark);
-
- # Set time exceeded
- if ($json->{timeExceeded} && $json->{timeExceeded} eq Mojo::JSON::true) {
- $index->time_exceeded(1);
+ # expected response for matches
+ if ($type eq 'matches') {
+ $self->_process_response_matches($index, $json);
+ }
+ elsif ($type eq 'trace') {
+ $self->_process_response_trace($index, $json);
+ }
+ elsif ($type eq 'match') {
+ $self->_process_response_match($index, $json);
+ }
+ elsif ($type eq 'resource') {
+ $self->_process_response_resource($index, $json);
};
- # Set result values
- $index->items_per_page($json->{itemsPerPage});
- $index->query_jsonld($json->{request}->{query});
- $index->results(_map_matches($json->{matches}));
-
- # Total results not set by stash
- if ($index->total_results == -1) {
-
- if ($json->{totalResults} && $json->{totalResults} > -1) {
- $c->app->log->debug('Cache total result');
- $c->chi->set($index->_api_cache => $json->{totalResults}, '120min');
- $index->total_results($json->{totalResults});
- };
- };
+ return 1 if ref $json ne 'HASH';
# Add warnings (Legacy)
if ($json->{warning}) {
@@ -197,19 +316,79 @@
};
+sub _process_response_matches {
+ my ($self, $index, $json) = @_;
+ # Reformat benchmark counter
+ my $benchmark = $json->{benchmark};
+ if ($benchmark && $benchmark =~ s/\s+(m)?s$//) {
+ $benchmark = sprintf("%.2f", $benchmark) . ($1 ? $1 : '') . 's';
+ };
+
+ # Set benchmark
+ $index->benchmark($benchmark);
+
+ # Set time exceeded
+ if ($json->{timeExceeded} && $json->{timeExceeded} eq Mojo::JSON::true) {
+ $index->time_exceeded(1);
+ };
+
+ # Set result values
+ $index->items_per_page($json->{itemsPerPage});
+ $index->query_jsonld($json->{request}->{query});
+ $index->results(_map_matches($json->{matches}));
+
+ # Total results not set by stash
+ if ($index->total_results == -1) {
+
+ if ($json->{totalResults} && $json->{totalResults} > -1) {
+ my $c = $index->controller;
+
+ $c->app->log->debug('Cache total result');
+ $c->chi->set($index->_api_cache => $json->{totalResults}, '120min');
+ $index->total_results($json->{totalResults});
+ };
+ };
+};
+
+
+# Process query serialization response
+sub _process_response_match {
+ my ($self, $index, $json) = @_;
+ $index->results(_map_match($json));
+};
+
+
+# Process query serialization response
+sub _process_response_trace {
+ my ($self, $index, $json) = @_;
+ $index->query_jsonld($json);
+};
+
+sub _process_response_resource {
+ my ($self, $index, $json) = @_;
+ my $c = $index->controller;
+
+ # TODO: That's unfortunate, as it prohibits multiple resources
+ $c->stash('search.resource' => $json);
+ $c->app->log->debug('Cache resource info');
+ $c->chi->set($c->stash('search._resource_cache') => $json, '24 hours');
+};
+
+
+# Parse the error messages
sub _notify_on_error {
my ($self, $c, $failure, $res) = @_;
my $json = $res;
my $log = $c->app->log;
+ # Check if the response is already json
if (blessed $res) {
- if (blessed $res ne 'Mojo::JSON') {
- $json = $res->json;
- };
+ $json = $res->json if blessed $res ne 'Mojo::JSON';
};
+ # Chec json response error message
if ($json) {
if ($json->{error}) {
# Temp
@@ -233,6 +412,7 @@
};
};
+ # Doesn't matter what - there is a failure!
if ($failure) {
$c->notify(error => (
($res->{code} ? $res->{code} . ': ' : '') .
@@ -243,18 +423,53 @@
};
+# Cleanup array of matches
sub _map_matches {
return () unless $_[0];
- map {
- $_->{ID} =~ s/^match\-[^!]+![^-]+-//;
- $_->{docID} =~ s/^[^_]+_//;
- $_;
- } @{ shift() };
+ map { _map_match($_) } @{ shift() };
+};
+
+
+# Cleanup single match
+sub _map_match {
+ my $x = shift or return;
+ $x->{ID} =~ s/^match\-[^!]+![^-]+-//;
+ $x->{docID} =~ s/^[^_]+_//;
+ $x;
+};
+
+
+sub _query_url {
+ my ($index, %param) = @_;
+
+ # Set cutoff from param
+ $index->cutoff(delete $param{cutoff});
+
+ # Set query language
+ $index->query_language(delete $param{query_language} // 'poliqarp');
+
+ # Should results be cached? Defaults to "yes"
+ $index->no_cache(1) if $param{no_cache};
+
+ # Init the query with stuff coming from the index
+ my %query;
+ $query{q} = $index->query;
+ $query{ql} = $index->query_language;
+ $query{page} = $index->start_page if $index->start_page;
+ $query{count} = $index->items_per_page if $index->items_per_page;
+ $query{cutoff} = 'true' if $index->cutoff;
+
+ # Todo: support corpus and collection
+ # Create query url
+ my $url = Mojo::URL->new($index->api);
+ $url->query(\%query);
+ return $url;
};
1;
+
__END__
=pod
diff --git a/lib/Korap/Controller/Search.pm b/lib/Korap/Controller/Search.pm
new file mode 100644
index 0000000..a94f0de
--- /dev/null
+++ b/lib/Korap/Controller/Search.pm
@@ -0,0 +1,113 @@
+package Korap::Controller::Search;
+use Mojo::Base 'Mojolicious::Controller';
+
+# Add X-Forwarded-For to user agent call everywhere
+
+
+# Query the KorAP backends and render a template
+sub query {
+ my $c = shift;
+ $c->layout('default');
+ $c->title('KorAP');
+
+ my $query = $c->param('q');
+
+ # Base parameters for remote access
+ my %param = (
+ query_language => scalar $c->param('ql'),
+ query => $query,
+ );
+
+ my $inspect = (scalar $c->param('action') // '') eq 'inspect' ? 1 : 0;
+
+ # Just check the serialization non-blocking
+ if ($inspect) {
+ $c->search->trace(
+ %param => sub {
+ return $c->render(template => 'query_info');
+ }
+ );
+ return;
+ };
+
+ my $template = scalar $c->param('snippet') ? 'snippet' : 'search';
+
+ # Search non-blocking
+ $c->delay(
+ sub {
+ my $delay = shift;
+ $c->search(
+ cutoff => scalar $c->param('cutoff'),
+ count => scalar $c->param('count'),
+ start_page => scalar $c->param('page'),
+ cb => $delay->begin,
+ %param
+ ) if $query;
+
+ $c->search->resource(
+ type => 'collection',
+ $delay->begin
+ );
+ },
+ sub {
+ return $c->render(template => $template);
+ }
+ );
+};
+
+
+# Get informations about a match
+sub match_info {
+ my $c = shift;
+
+ my $foundry = '*';
+ my %query = (foundry => '*');
+ if ($c->param('foundry')) {
+ $query{foundry} = $c->param('foundry');
+ if ($c->param('layer')) {
+ $query{layer} = $c->param('layer');
+ };
+ if ($c->param('spans')) {
+ $query{spans} = 'true';
+ };
+ };
+
+ $c->render_later;
+
+ # Use the API for fetching matching information non-blocking
+ $c->search->match(
+ corpus_id => $c->stash('corpus_id'),
+ doc_id => $c->stash('doc_id'),
+ match_id => $c->stash('match_id'),
+ %query,
+
+ # Callback for async search
+ sub {
+ my $index = shift;
+
+ return $c->respond_to(
+
+ # Render json if requested
+ json => sub {
+ # Add notifications to the matching json
+ # TODO: There should be a special notification engine doing that!
+ my $notes = $c->notifications(json => $index->results->[0]);
+ $c->render(json => $notes);
+ },
+
+ # Render html if requested
+ html => sub {
+ return $c->render(
+ layout => 'default',
+ template => 'match_info'
+ )
+ }
+ );
+ }
+ );
+};
+
+1;
+
+
+__END__
diff --git a/lib/Korap/Controller/Tutorial.pm b/lib/Korap/Controller/Tutorial.pm
new file mode 100644
index 0000000..186439d
--- /dev/null
+++ b/lib/Korap/Controller/Tutorial.pm
@@ -0,0 +1,33 @@
+package Korap::Controller::Tutorial;
+use Mojo::Base 'Mojolicious::Controller';
+
+# Todo: Set title as defaults
+
+sub page {
+ my $c = shift;
+
+ # Show tutorial embedded
+ if ($c->param('embedded')) {
+ $c->layout('snippet');
+ $c->stash(embedded => 1);
+ }
+
+ # Show tutorial in full screen
+ else {
+ $c->layout('default');
+ };
+
+ # Title should be "KorAP"
+ $c->title('KorAP');
+
+ my $page = $c->stash('tutorial');
+ return $c->render(
+ template => 'tutorial/' . $page
+ );
+};
+
+
+1;
+
+
+__END__
diff --git a/lib/Korap/Info.pm b/lib/Korap/Info.pm
deleted file mode 100644
index bc8fc05..0000000
--- a/lib/Korap/Info.pm
+++ /dev/null
@@ -1,60 +0,0 @@
-package Korap::Info;
-use Mojo::Base 'Mojolicious::Controller';
-
-sub about_match {
- my $c = shift;
-
- my $corpus_id = $c->stash('corpus_id');
- my $doc_id = $c->stash('doc_id');
- my $match_id = $c->stash('match_id');
-
- my $foundry = '*';
- my %query = (foundry => '*');
- if ($c->param('foundry')) {
- $query{foundry} = $c->param('foundry');
- if ($c->param('layer')) {
- $query{layer} = $c->param('layer');
- };
- if ($c->param('spans')) {
- $query{spans} = 'true';
- };
- };
-
- return $c->respond_to(
- json => sub {
- $c->render(json => $c->notifications(
- json => $c->match_info($corpus_id, $doc_id, $match_id, %query))
- )
- },
- html => sub {
- my $match = $c->match_info($corpus_id, $doc_id, $match_id, %query);
- if ($match->{error}) {
- $c->notify(error => $match->{error});
- return $c->render_exception('error');
- };
-
- $c->render(
- layout => 'default',
- template => 'match_info',
- match => $match
- )
- }
- );
-};
-
-# Todo: Return info for all collections
-sub about_collection {
- my $c = shift;
- my $src = $c->stash('collection_id');
- if ($src) {
- $c->render(
- json => $c->notifications(
- json => $c->info_on($src)
- )
- );
- };
-};
-
-1;
-
-__END__
diff --git a/lib/Korap/Plugin/KorapInfo.pm b/lib/Korap/Plugin/KorapInfo.pm
deleted file mode 100644
index 4de0406..0000000
--- a/lib/Korap/Plugin/KorapInfo.pm
+++ /dev/null
@@ -1,194 +0,0 @@
-package Korap::Plugin::KorapInfo;
-use Mojo::Base 'Mojolicious::Plugin';
-use Mojo::JSON qw/decode_json/;
-use Mojo::ByteStream 'b';
-
-# Todo: Make the requests non-blocking
-
-sub map_match {
- my $x = shift or return;
- $x->{ID} =~ s/^match\-[^!]+![^-]+-// if $x->{ID};
- $x->{docID} =~ s/^[^_]+_// if $x->{docID};
- $x;
-};
-
-# Register plugin
-sub register {
- my ($plugin, $mojo, $param) = @_;
- $param ||= {};
-
- # Load parameter from Config file
- if (my $config_param = $mojo->config('KorAP')) {
- $param = { %$param, %$config_param };
- };
-
- # Temporary info data
- state $json = decode_json(b(join('', <DATA>))->encode);
-
- # Get the API endpoint
- my $api = $param->{'api-0.1'};
-
- # Todo: Make this recognize the user!
- $mojo->helper(
- resource_info => sub {
- my $c = shift;
- my $src = shift;
-
- # Dispatch for test mode
- if ($c->app->mode eq 'test') {
- if ($src eq 'collection') {
- return $json->{collections};
- };
- };
-
- # Rename info endpoints and build URL
- $src = 'virtualcollection' if $src eq 'collection';
- $src = 'corpus' if $src eq 'corpus';
- my $url = Mojo::URL->new($api)->path($src);
-
- $c->app->log->debug($url);
-
- # Check for cached information
- if (my $json = $c->chi->get($url->to_string)) {
- return $json;
- }
-
- # Retrieve and set the cache
- elsif (my $response = $c->ua->get($url)->success) {
- my $json = $response->json;
- $c->chi->set($url->to_string => $json, '24 hours');
- return $json;
- };
-
- # Something went wrong
- $c->notify(error => 'Unable to retrieve resource');
- return [];
- }
- );
-
- $mojo->helper(
- match_info => sub {
- my $c = shift;
-
- my $corpus = shift;
- my $doc = shift;
- my $match = shift;
- my %param = @_;
-
- if ($c->app->mode eq 'test') {
- return map_match($json->{treeInfo}) if $param{spans};
- return map_match($json->{matchInfo2});
- };
-
- # Build url
- my $temp_api = $c->config('KorAP')->{'api-0.1'};
- my $url = Mojo::URL->new($temp_api)->path(
- join('/', 'corpus', $corpus, $doc, $match, 'matchInfo')
- );
-
- # Build match id
- # $match = 'match-' . $corpus . '!' . $corpus . '_' . $doc . '-' . $match;
-
- my %query;
- $query{foundry} = $param{foundry} if defined $param{foundry};
- $query{layer} = $param{layer} if defined $param{layer};
- $query{spans} = $param{spans} ? 'true' : 'false';
-
- # Add query
- $url->query(\%query);
-
- $c->app->log->debug($url);
-
- if (my $response = $c->ua->get($url)->success) {
- return map_match($response->json);
- };
-
- $c->notify(error => 'Unable to retrieve resource');
- return {};
- }
- );
-};
-
-
-1;
-
-__DATA__
-{
- "treeInfo": {
- "author":"Filzstift,Alexander Sommer,TheK",
- "textClass":"freizeit-unterhaltung reisen",
- "corpusID":"WPD",
- "title":"Neuseeland",
- "foundries":"xip xip/morpho xip/constituency xip/dependency corenlp corenlp/namedentities corenlp/namedentities/ne_dewac_175m_600 corenlp/namedentities corenlp/namedentities/ne_hgc_175m_600 mate mate/morpho mate/dependency connexor connexor/morpho connexor/syntax connexor/phrase treetagger treetagger/morpho base base/sentences base/paragraphs opennlp opennlp/morpho",
- "tokenization":"opennlp#tokens",
- "field":"tokens",
- "startMore":false,
- "endMore":false,
- "docID":"WPD_NNN.02848",
- "snippet" : "<span class=\"context-left\"><span class=\"more\"></span></span><span class=\"match\"><span title=\"xip/c:VERB\">enthält</span>, <span title=\"xip/c:CONJ\">aber</span> <span title=\"xip/c:PTCL\">an</span> <span title=\"xip/c:CONJ\">und</span> <span title=\"xip/c:PP\"><span title=\"xip/c:PREP\">für</span> <span title=\"xip/c:NP\"><span title=\"xip/c:PRON\">sich</span></span></span> <span title=\"xip/c:NP\"><span title=\"xip/c:DET\">keinen</span> <span title=\"xip/c:NPA\"><span title=\"xip/c:AP\"><span title=\"xip/c:ADJ\">weiteren</span></span> <span title=\"xip/c:NOUN\">Sinn</span></span></span> <span title=\"xip/c:VERB\">enthält</span></span><span class=\"context-right\"><span class=\"more\"></span></span>",
- "ID":"match-WPD!WPD_NNN.02848-p1213-1224",
- "pubDate":"2005-03-28",
- "context":{
- "left":["token",0],
- "right":["token",0]
- }
- },
- "matchInfo": {
- "author":"Filzstift,Alexander Sommer,TheK",
- "textClass":"freizeit-unterhaltung reisen",
- "corpusID":"WPD",
- "title":"Neuseeland",
- "foundries":"xip xip/morpho xip/constituency xip/dependency corenlp corenlp/namedentities corenlp/namedentities/ne_dewac_175m_600 corenlp/namedentities corenlp/namedentities/ne_hgc_175m_600 mate mate/morpho mate/dependency connexor connexor/morpho connexor/syntax connexor/phrase treetagger treetagger/morpho base base/sentences base/paragraphs opennlp opennlp/morpho",
- "tokenization":"opennlp#tokens",
- "field":"tokens",
- "startMore":false,
- "endMore":false,
- "docID":"WPD_NNN.02848",
- "snippet":"<span class=\"context-left\"></span><span class=\"match\"><span title=\"mate/l:besonders\"><span title=\"mate/p:ADV\">Besonders</span></span> <span title=\"mate/l:auffällig\"><span title=\"mate/m:degree:pos\"><span title=\"mate/p:ADJD\">auffällig</span></span></span> <span title=\"mate/l:sein\"><span title=\"mate/m:mood:ind\"><span title=\"mate/m:number:pl\"><span title=\"mate/m:person:3\"><span title=\"mate/m:tense:pres\"><span title=\"mate/p:VAFIN\">sind</span></span></span></span></span></span> <span title=\"mate/l:schließlich\"><span title=\"mate/p:ADV\">schließlich</span></span> <span title=\"mate/l:noch\"><span title=\"mate/p:ADV\">noch</span></span> <span title=\"mate/l:der\"><span title=\"mate/m:case:nom\"><span title=\"mate/m:gender:masc\"><span title=\"mate/m:number:sg\"><span title=\"mate/p:ART\">der</span></span></span></span></span> <span title=\"mate/l:pohutukawa\"><span title=\"mate/m:case:nom\"><span title=\"mate/m:gender:masc\"><span title=\"mate/m:number:sg\"><span title=\"mate/p:NE\">Pohutukawa</span></span></span></span></span> <span title=\"mate/l:und\"><span title=\"mate/p:KON\">und</span></span> <span title=\"mate/l:der\"><span title=\"mate/m:case:nom\"><span title=\"mate/m:gender:masc\"><span title=\"mate/m:number:sg\"><span title=\"mate/p:ART\">der</span></span></span></span></span> <span title=\"mate/l:cabbage\"><span title=\"mate/m:case:nom\"><span title=\"mate/m:gender:masc\"><span title=\"mate/m:number:sg\"><span title=\"mate/p:NE\">Cabbage</span></span></span></span></span> <span title=\"mate/l:tree\"><span title=\"mate/m:gender:masc\"><span title=\"mate/m:number:sg\"><span title=\"mate/p:NE\"><span title=\"mate/m:case:nom\">Tree</span></span></span></span></span></span><span class=\"context-right\"></span>",
- "ID":"match-WPD!WPD_NNN.02848-p1213-1224",
- "pubDate":"2005-03-28",
- "context":{
- "left":["token",0],
- "right":["token",0]
- }
- },
- "collections" : [
- {
- "shared":false,
- "id":1,
- "managed":true,
- "created":1401193381119,
- "statistics":{
- "documents":196510,
- "tokens":51545081,
- "sentences":4116282,
- "paragraphs":2034752
- },
- "query":[
- {
- "@type":"korap:meta-filter",
- "@value":{
- "@type":"korap:term",
- "@field":"korap:field#corpusID",
- "@value":"WPD"
- }
- }
- ],
- "description":"Die freie Enzyklopädie",
- "name":"Wikipedia",
- "foundries":"base;corenlp;mate;mpt;opennlp;tt;xip"
- }
- ],
- "matchInfo2" : {
- "author":"AHZ,AkaBot,Zwobot",
- "textClass":"freizeit-unterhaltung reisen",
- "corpusID":"WPD",
- "title":"Okres Hodonín",
- "foundries":"opennlp opennlp/morpho base base/sentences base/paragraphs mate mate/morpho mate/dependency corenlp corenlp/namedentities corenlp/namedentities/ne_dewac_175m_600 corenlp/namedentities corenlp/namedentities/ne_hgc_175m_600 treetagger treetagger/morpho connexor connexor/morpho connexor/syntax connexor/phrase xip xip/morpho xip/constituency xip/dependency",
- "tokenization":"opennlp#tokens",
- "field":"tokens",
- "startMore":false,
- "endMore":false,
- "docID":"WPD_OOO.01534",
- "snippet":"<span class=\"context-left\"></span><span class=\"match\"><span title=\"mate/l:archlebov\"><span title=\"mate/p:NE\"><span title=\"tt/p:NE\"><span title=\"tt/p:NN\">Archlebov</span></span></span></span> - <span title=\"mate/l:blatnice\"><span title=\"mate/p:FM\"><span title=\"tt/p:NE\"><span title=\"tt/p:NN\">Blatnice</span></span></span></span> <span title=\"mate/l:pod\"><span title=\"mate/p:FM\"><span title=\"tt/l:pod\"><span title=\"tt/p:NE\">pod</span></span></span></span> <span title=\"mate/l:svatým\"><span title=\"mate/p:FM\"><span title=\"tt/p:NE\">Svatým</span></span></span> <span title=\"mate/l:antonínkem\"><span title=\"mate/p:FM\"><span title=\"tt/p:NE\"><span title=\"tt/p:NN\">Antonínkem</span></span></span></span> - <span title=\"mate/l:blatni\"><span title=\"mate/p:NE\"><span title=\"tt/p:NE\"><span title=\"tt/p:NN\">Blatni</span></span></span></span><span title=\"mate/l:?ka\"><span title=\"mate/p:NE\">čka</span></span> - <span title=\"mate/l:bukovany\"><span title=\"mate/p:NE\"><span title=\"tt/p:NE\"><span title=\"tt/p:NN\">Bukovany</span></span></span></span> - <span title=\"mate/l:bzenec\"><span title=\"mate/p:NE\"><span title=\"tt/p:NE\"><span title=\"tt/p:NN\">Bzenec</span></span></span></span> - <span title=\"mate/l:?ej\"><span title=\"mate/p:NE\">Čej</span></span>č - <span title=\"mate/l:?ejkovice\"><span title=\"mate/p:NE\">Čejkovice</span></span> - <span title=\"mate/l:?elo\"><span title=\"mate/p:NE\">Čelo</span></span><span title=\"mate/l:?nice\"><span title=\"mate/p:NE\">žnice</span></span> - <span title=\"mate/l:dambo\"><span title=\"mate/p:NE\"><span title=\"tt/p:NE\"><span title=\"tt/p:NN\">Dambo</span></span></span></span><span title=\"mate/l:?ice\"><span title=\"mate/p:NE\">řice</span></span> - <span title=\"mate/l:doln\"><span title=\"mate/p:NE\">Doln</span></span><span title=\"mate/l:í\"><span title=\"mate/p:NE\">í</span></span> <span title=\"mate/l:bojanovice\"><span title=\"mate/p:NE\"><span title=\"tt/p:NE\"><span title=\"tt/p:NN\">Bojanovice</span></span></span></span> - <span title=\"mate/l:domanín\"><span title=\"mate/p:NE\"><span title=\"tt/p:NE\"><span title=\"tt/p:NN\">Domanín</span></span></span></span> - <span title=\"mate/l:dra\"><span title=\"mate/p:NE\"><span title=\"tt/p:NE\">Dra</span></span></span><span title=\"mate/p:NE\">ž</span><span title=\"mate/l:?vky\"><span title=\"mate/p:NE\"><span title=\"tt/p:NE\"><span title=\"tt/p:NN\">ůvky</span></span></span></span> - <span title=\"mate/l:dub\"><span title=\"mate/p:NE\"><span title=\"tt/p:NE\">Dub</span></span></span><span title=\"mate/l:?any\"><span title=\"mate/p:NE\">ňany</span></span> - <span title=\"mate/l:hodonín\"><span title=\"mate/p:NE\"><span title=\"tt/p:NE\">Hodonín</span></span></span> - <span title=\"mate/l:hovorany\"><span title=\"mate/p:NE\"><span title=\"tt/p:NE\"><span title=\"tt/p:NN\">Hovorany</span></span></span></span> - <span title=\"mate/l:hroznová\"><span title=\"mate/p:NE\"><span title=\"tt/p:NE\">Hroznová</span></span></span> <span title=\"mate/l:lhota\"><span title=\"mate/p:NE\"><span title=\"tt/p:NE\">Lhota</span></span></span> - <span title=\"mate/l:hrubá\"><span title=\"mate/p:NE\"><span title=\"tt/p:NE\">Hrubá</span></span></span> <span title=\"mate/l:vrbka\"><span title=\"mate/p:NE\"><span title=\"tt/p:NE\">Vrbka</span></span></span> - <span title=\"mate/l:hýsly\"><span title=\"mate/p:NE\"><span title=\"tt/p:NE\"><span title=\"tt/p:NN\">Hýsly</span></span></span></span> - <span title=\"mate/l:javorník\"><span title=\"mate/p:NE\"><span title=\"tt/p:NE\"><span title=\"tt/p:NN\">Javorník</span></span></span></span> - <span title=\"mate/l:je\"><span title=\"mate/p:NE\"><span title=\"tt/l:Je\"><span title=\"tt/p:KOUS\">Je</span></span></span></span><span title=\"mate/l:?ov\"><span title=\"mate/p:NE\">žov</span></span> - <span title=\"mate/l:josefov\"><span title=\"mate/p:NE\"><span title=\"tt/p:NE\">Josefov</span></span></span> - <span title=\"mate/l:karlín\"><span title=\"mate/p:NE\"><span title=\"tt/p:NE\"><span title=\"tt/p:NN\">Karlín</span></span></span></span> - <span title=\"mate/l:kel\"><span title=\"mate/p:NE\"><span title=\"tt/p:NE\">Kel</span></span></span><span title=\"mate/l:?any\"><span title=\"mate/p:NE\">čany</span></span> - <span title=\"mate/l:kn\"><span title=\"mate/p:NN\"><span title=\"tt/p:NE\">Kn</span></span></span>ě<span title=\"mate/l:?dub\"><span title=\"mate/p:NE\"><span title=\"tt/p:NE\"><span title=\"tt/p:NN\">ždub</span></span></span></span> - <span title=\"mate/l:kostelec\"><span title=\"mate/p:NE\"><span title=\"tt/p:NE\">Kostelec</span></span></span> - <span title=\"mate/l:kozojídky\"><span title=\"mate/p:NE\"><span title=\"tt/p:NE\"><span title=\"tt/p:NN\">Kozojídky</span></span></span></span> - <span title=\"mate/l:ku\"><span title=\"mate/p:NE\"><span title=\"tt/p:NE\">Ku</span></span></span><span title=\"mate/l:?elov\"><span title=\"mate/p:NE\">želov</span></span> - <span title=\"mate/l:kyjov\"><span title=\"mate/p:NE\"><span title=\"tt/p:NE\">Kyjov</span></span></span> - <span title=\"mate/l:labuty\"><span title=\"mate/p:NE\"><span title=\"tt/p:NE\"><span title=\"tt/p:NN\">Labuty</span></span></span></span> - <span title=\"mate/l:lipov\"><span title=\"mate/p:NE\"><span title=\"tt/p:NE\">Lipov</span></span></span> - <span title=\"mate/l:louka\"><span title=\"mate/p:NE\"><span title=\"tt/p:NE\">Louka</span></span></span> - <span title=\"mate/l:lov\"><span title=\"mate/p:NE\"><span title=\"tt/p:NE\">Lov</span></span></span><span title=\"mate/l:?ice\"><span title=\"mate/p:NE\">čice</span></span> - <span title=\"mate/l:lu\"><span title=\"mate/p:NE\"><span title=\"tt/p:NE\"><span title=\"tt/l:Lu\">Lu</span></span></span></span><span title=\"mate/l:?ice\"><span title=\"mate/p:NE\">žice</span></span> - <span title=\"mate/l:malá\"><span title=\"mate/p:NE\"><span title=\"tt/p:NE\">Malá</span></span></span> <span title=\"mate/l:vrbka\"><span title=\"mate/p:NE\"><span title=\"tt/p:NE\">Vrbka</span></span></span> - <span title=\"mate/l:mikul\"><span title=\"mate/p:NE\"><span title=\"tt/p:NE\">Mikul</span></span></span><span title=\"mate/l:?ice\"><span title=\"mate/p:NE\">čice</span></span> - <span title=\"mate/l:milotice\"><span title=\"mate/p:NE\"><span title=\"tt/p:NE\"><span title=\"tt/p:NN\">Milotice</span></span></span></span> - <span title=\"mate/l:moravany\"><span title=\"mate/p:NE\"><span title=\"tt/p:NE\"><span title=\"tt/p:NN\">Moravany</span></span></span></span> - <span title=\"mate/l:moravský\"><span title=\"mate/p:NE\"><span title=\"tt/p:NE\">Moravský</span></span></span> <span title=\"mate/l:písek\"><span title=\"mate/p:NE\"><span title=\"tt/p:NE\">Písek</span></span></span> - <span title=\"mate/l:mut\"><span title=\"mate/p:NN\"><span title=\"tt/l:Mut\"><span title=\"tt/p:NN\">Mut</span></span></span></span><span title=\"mate/l:?nice\"><span title=\"mate/p:NE\">ěnice</span></span> - <span title=\"mate/l:násedlovice\"><span title=\"mate/p:NE\"><span title=\"tt/p:NE\"><span title=\"tt/p:NN\">Násedlovice</span></span></span></span> - <span title=\"mate/l:nechvalín\"><span title=\"mate/p:NE\"><span title=\"tt/p:NE\"><span title=\"tt/p:NN\">Nechvalín</span></span></span></span> - <span title=\"mate/l:nenkovice\"><span title=\"mate/p:NE\"><span title=\"tt/p:NE\"><span title=\"tt/p:NN\">Nenkovice</span></span></span></span> - <span title=\"mate/l:nová\"><span title=\"mate/p:NE\"><span title=\"tt/p:NE\">Nová</span></span></span> <span title=\"mate/l:lhota\"><span title=\"mate/p:NE\"><span title=\"tt/p:NE\">Lhota</span></span></span> - <span title=\"mate/l:nový\"><span title=\"mate/p:NE\"><span title=\"tt/p:NE\">Nový</span></span></span> <span title=\"mate/l:poddvorov\"><span title=\"mate/p:NE\"><span title=\"tt/p:NE\">Poddvorov</span></span></span> - <span title=\"mate/l:ostrovánky\"><span title=\"mate/p:NE\"><span title=\"tt/p:NE\"><span title=\"tt/p:NN\">Ostrovánky</span></span></span></span> - <span title=\"mate/l:petrov\"><span title=\"mate/p:NE\"><span title=\"tt/p:NE\">Petrov</span></span></span> - <span title=\"mate/l:pru\"><span title=\"mate/p:NE\"><span title=\"tt/p:NE\">Pru</span></span></span><span title=\"mate/l:?ánky\"><span title=\"mate/p:NE\">šánky</span></span> - <span title=\"mate/l:rad\"><span title=\"mate/p:NE\"><span title=\"tt/l:Rad\"><span title=\"tt/p:NN\">Rad</span></span></span></span><span title=\"mate/l:?jov\"><span title=\"mate/p:NE\">ějov</span></span> - <span title=\"mate/l:ratí\"><span title=\"mate/p:NE\"><span title=\"tt/p:NE\"><span title=\"tt/p:NN\">Ratí</span></span></span></span><span title=\"mate/l:?kovice\"><span title=\"mate/p:NE\">škovice</span></span> - <span title=\"mate/l:rohatec\"><span title=\"mate/p:NE\"><span title=\"tt/p:NE\"><span title=\"tt/p:NN\">Rohatec</span></span></span></span> - <span title=\"mate/l:skalka\"><span title=\"mate/p:NE\"><span title=\"tt/p:NE\">Skalka</span></span></span> - <span title=\"mate/l:skoronice\"><span title=\"mate/p:NE\"><span title=\"tt/p:NE\"><span title=\"tt/p:NN\">Skoronice</span></span></span></span> - <span title=\"mate/l:sob\"><span title=\"mate/p:NE\"><span title=\"tt/p:NE\">Sob</span></span></span><span title=\"mate/l:?lky\"><span title=\"mate/p:NE\">ůlky</span></span> - <span title=\"mate/l:starý\"><span title=\"mate/p:NE\"><span title=\"tt/p:NE\"><span title=\"tt/p:NN\">Starý</span></span></span></span> <span title=\"mate/l:poddvorov\"><span title=\"mate/p:NE\"><span title=\"tt/p:NE\">Poddvorov</span></span></span> - <span title=\"mate/l:stav\"><span title=\"mate/p:NE\"><span title=\"tt/p:NE\">Stav</span></span></span><span title=\"mate/p:NE\">ě</span><span title=\"mate/l:?ice\"><span title=\"mate/p:NE\"><span title=\"tt/p:NE\"><span title=\"tt/p:NN\">šice</span></span></span></span> - <span title=\"mate/l:strá\"><span title=\"mate/p:NE\"><span title=\"tt/p:NE\">Strá</span></span></span><span title=\"mate/l:?nice\"><span title=\"mate/p:NE\">žnice</span></span> - <span title=\"mate/l:strá\"><span title=\"mate/p:NE\"><span title=\"tt/p:NE\">Strá</span></span></span><span title=\"mate/l:?ovice\"><span title=\"mate/p:NE\">žovice</span></span> - <span title=\"mate/l:sudom\"><span title=\"mate/p:NE\"><span title=\"tt/p:NE\"><span title=\"tt/p:NN\">Sudom</span></span></span></span><span title=\"mate/p:NE\">ě</span><span title=\"mate/l:?ice\"><span title=\"mate/p:NE\"><span title=\"tt/p:NE\"><span title=\"tt/p:NN\">řice</span></span></span></span> - <span title=\"mate/l:suchov\"><span title=\"mate/p:NE\"><span title=\"tt/p:NE\">Suchov</span></span></span> - <span title=\"mate/l:svatobo\"><span title=\"mate/p:NE\"><span title=\"tt/p:NE\"><span title=\"tt/p:NN\">Svatobo</span></span></span></span><span title=\"mate/l:?ice-mist\"><span title=\"mate/p:NE\">řice-Mist</span></span><span title=\"mate/l:?ín\"><span title=\"mate/p:NE\">řín</span></span> - <span title=\"mate/l:syrovín\"><span title=\"mate/p:NE\"><span title=\"tt/p:NE\"><span title=\"tt/p:NN\">Syrovín</span></span></span></span> - <span title=\"mate/l:?ardice\"><span title=\"mate/p:NE\">Šardice</span></span> - <span title=\"mate/l:tasov\"><span title=\"mate/p:NE\"><span title=\"tt/p:NE\"><span title=\"tt/p:NN\">Tasov</span></span></span></span> - <span title=\"mate/l:t?mice\"><span title=\"mate/p:NE\">Těmice</span></span> - <span title=\"mate/l:terezín\"><span title=\"mate/p:NE\"><span title=\"tt/p:NE\">Terezín</span></span></span> - <span title=\"mate/l:tvaro\"><span title=\"mate/p:NE\"><span title=\"tt/p:NE\"><span title=\"tt/p:NN\">Tvaro</span></span></span></span><span title=\"mate/l:?ná\"><span title=\"mate/p:NE\">žná</span></span> <span title=\"mate/l:lhota\"><span title=\"mate/p:NE\"><span title=\"tt/p:NE\">Lhota</span></span></span> - <span title=\"mate/l:uh\"><span title=\"mate/p:NE\"><span title=\"tt/l:uh\"><span title=\"tt/p:ITJ\">Uh</span></span></span></span><span title=\"mate/l:?ice\"><span title=\"mate/p:NE\">řice</span></span> - <span title=\"mate/l:vacenovice\"><span title=\"mate/p:NE\"><span title=\"tt/p:NE\"><span title=\"tt/p:NN\">Vacenovice</span></span></span></span> - <span title=\"mate/l:velká\"><span title=\"mate/p:NE\"><span title=\"tt/p:NE\">Velká</span></span></span> <span title=\"mate/l:nad\"><span title=\"mate/p:NE\"><span title=\"tt/p:NE\"><span title=\"tt/l:nad\">nad</span></span></span></span> <span title=\"mate/l:veli\"><span title=\"mate/p:NE\"><span title=\"tt/p:NE\">Veli</span></span></span><span title=\"mate/l:?kou\"><span title=\"mate/p:NE\">čkou</span></span> - <span title=\"mate/l:veselí\"><span title=\"mate/p:NE\"><span title=\"tt/p:NE\">Veselí</span></span></span> <span title=\"mate/l:nad\"><span title=\"mate/p:NE\"><span title=\"tt/l:nad\"><span title=\"tt/p:NE\">nad</span></span></span></span> <span title=\"mate/l:moravou\"><span title=\"mate/p:NE\"><span title=\"tt/p:NE\">Moravou</span></span></span> - <span title=\"mate/l:v\"><span title=\"mate/p:CARD\"><span title=\"tt/l:V\"><span title=\"tt/p:NN\">V</span></span></span></span><span title=\"mate/l:?t\"><span title=\"mate/p:XY\">ěte</span></span><span title=\"mate/l:?ov\"><span title=\"mate/p:XY\">řov</span></span> - <span title=\"mate/p:FM\"><span title=\"mate/l:vlko\"><span title=\"tt/p:NE\"><span title=\"tt/p:NN\">Vlko</span></span></span></span>š - <span title=\"mate/l:vnorovy\"><span title=\"mate/p:NE\"><span title=\"tt/p:NE\"><span title=\"tt/p:NN\">Vnorovy</span></span></span></span> - <span title=\"mate/l:vracov\"><span title=\"mate/p:NE\"><span title=\"tt/p:NE\"><span title=\"tt/p:NN\">Vracov</span></span></span></span> - <span title=\"mate/l:v\"><span title=\"mate/p:NE\"><span title=\"tt/l:V\"><span title=\"tt/p:NN\">V</span></span></span></span><span title=\"mate/l:?esovice\"><span title=\"mate/p:NE\">řesovice</span></span> - <span title=\"mate/l:?ádovice\"><span title=\"mate/p:NE\">Žádovice</span></span> - <span title=\"mate/l:?aro\"><span title=\"mate/p:NE\">Žaro</span></span><span title=\"mate/l:?ice\"><span title=\"mate/p:NE\">šice</span></span> - <span title=\"mate/l:?dánice\"><span title=\"mate/p:NE\">Ždánice</span></span> - <span title=\"mate/l:?eletice\"><span title=\"mate/p:NE\">Želetice</span></span> - <span title=\"mate/l:?eravice\"><span title=\"mate/p:NE\">Žeravice</span></span> - <span title=\"mate/p:NE\"><span title=\"mate/l:?eraviny\">Žeraviny</span></span></span><span class=\"context-right\"></span>","ID":"match-WPD!WPD_OOO.01534-p328-462","pubDate":"2005-03-28","context":{"left":["token",0],"right":["token",0]}}
-}
diff --git a/lib/Korap/Plugin/KorapSearch.pm b/lib/Korap/Plugin/KorapSearch.pm
deleted file mode 100644
index 7aab520..0000000
--- a/lib/Korap/Plugin/KorapSearch.pm
+++ /dev/null
@@ -1,400 +0,0 @@
-package Korap::Plugin::KorapSearch;
-use Mojo::Base 'Mojolicious::Plugin';
-use Scalar::Util qw/blessed/;
-use Mojo::JSON qw/decode_json true/;
-use Mojo::ByteStream 'b';
-
-# TODO: This will probably be an engine for M::P::Search
-# cutoff has to be true or false
-# TODO: Write search snippet
-
-sub map_matches {
- return [] unless $_[0];
- [
- map {
- $_->{ID} =~ s/^match\-[^!]+![^-]+-//;
- $_->{docID} =~ s/^[^_]+_//;
- $_;
- } @{ shift() }
- ];
-};
-
-# Remove these attributes from stash after searching
-our @remove_from_stash =
- qw/hits totalResults benchmark itemsPerPage error query/;
-
-
-# Register plugin
-sub register {
- my ($plugin, $mojo, $param) = @_;
- $param ||= {};
-
- # Load parameter from Config file
- if (my $config_param = $mojo->config('KorAP')) {
- $param = { %$param, %$config_param };
- };
-
- my $api = $param->{'api-0.1'};
-
- # Create search endpoint
- $mojo->routes->add_shortcut(
- search => sub {
- # Todo: Support TRACE
- return shift->route('/search')->to('search#remote')
- }
- );
-
- # TODO: Create a hash and either make these stash values
- # or return the hash in case of non-blocking requests
- $mojo->helper(
- search => sub {
- my $c = shift;
-
- # If there is no callback, return the JSON object
- my $cb = pop if ref $_[-1] && ref $_[-1] eq 'CODE';
-
- my %param = @_;
-
- # Test environment / Fixtures - set via config!
- if ($c->app->mode eq 'test') {
- state $json = decode_json(join(' ', <DATA>));
-
- return $json unless $cb;
-
- $c->stash('search.count' => 10);
- $c->stash('search.startPage' => 1);
- $c->stash('search.totalResults' => 666);
- $c->stash('search.itemsPerPage' => 10);
- $c->stash('search.bm.hit' => 20);
- $c->stash('search.bm.result' => 10);
- $c->stash('search.query' => $json->{request}->{query});
- $c->stash('search.hits' => map_matches $json->{matches});
- return $cb->();
- };
-
- $c->stash(
- 'search.count' =>
- delete($param{count}) //
- scalar($c->param('count')) //
- $param->{count}
- );
-
- $c->stash('search.startPage' =>
- (delete($param{startPage}) //
- scalar $c->param('p') //
- 1
- ));
-
- my $query = $param{query} // scalar $c->param('q');
- my $cutoff = $param{cutoff} // scalar $c->param('cutoff') // undef;
-
- return '' unless $query;
-
- # Get stash information
- my $count = $c->stash('search.count');
- my $start_page = $c->stash('search.startPage');
-
- foreach ($start_page, $count) {
- $_ = undef if (!$_ || $_ !~ /^\d+$/);
- };
-
- my $url = Mojo::URL->new($api);
-
- # Search in corpus
- if ($c->stash('corpus_id')) {
- $url->path('corpus/' . $c->stash('corpus_id') . '/search');
- }
-
- # search in collection
- elsif ($c->stash('collection_id')) {
- $url->path('virtualcollection/' . $c->stash('collection_id') . '/search');
- }
-
- # Just search
- else {
- $url->path('search');
- };
-
- #if ($c->stash('resource')) {
- #$url->path($c->stash('resource'));
- #if ($c->stash('cid')) {
- #$url->path($c->stash('cid'));
- #};
- #};
-
- my %query = (q => $query);
- $query{ql} = $param{ql} // scalar $c->param('ql') // 'poliqarp';
- $query{count} = $count if $count;
- $query{cutoff} = 'true' if $cutoff;
-
- $url->query(\%query);
- my $cache_url = $url->to_string;
-
- $url->query({context => 'paragraph'});
-
- # Check cache for total results
- my $total_results;
- if (
- # TODO: This should also be a query parameter
- !$param{no_cache} &&
- defined ($total_results = $c->chi->get('total-' . $cache_url))) {
- $c->stash('search.totalResults' => $total_results);
- $c->app->log->debug('Get total result from cache');
-
- # Set cutoff unless already set
- $url->query({cutoff => 'true'}) unless defined $cutoff;
- }
-
- # No cache
- else {
- $c->stash('search.totalResults' => 0);
- };
-
- $url->query({page => $start_page});
-
- # This may be incorrect
- $c->stash('search.itemsPerPage' => $count);
-
- # Only set this when on test port
- if ($c->korap_test_port) {
- $c->stash('search.apirequest' => $url->to_string);
- };
-
- my $ua = Mojo::UserAgent->new;
-
- # Set timeout to 2 minutes
- $ua->inactivity_timeout(120);
-
- $c->app->log->debug('Search for ' . $url->to_string);
-
- # Blocking request
- # TODO: Make non-blocking
- my $tx = $ua->get($url);
-
- if (my $e = $tx->error) {
- $c->notify(
- error =>
- ($e->{code} ? $e->{code} . ': ' : '') .
- $e->{message} . ' (remote)'
- );
- return;
- };
-
- # Request successful
- if (my $res = $tx->success) {
-
- $c->stash('search.apiresponse' => $res->body);
-
- my $json = $res->json;
- return $json unless $cb;
-
- # Reformat benchmark counter
- my $benchmark = $json->{benchmark};
- if ($benchmark && $benchmark =~ s/\s+(m)?s$//) {
- $benchmark = sprintf("%.2f", $benchmark) . ($1 ? $1 : '') . 's';
- };
-
- for ($c->stash) {
- $_->{'search.benchmark'} = $benchmark;
- $_->{'search.itemsPerPage'} = $json->{itemsPerPage};
- $_->{'search.timeExceeded'} = ($json->{timeExceeded} &&
- $json->{timeExceeded}
- eq Mojo::JSON::true) ? 1 : 0;
- $_->{'search.query'} = $json->{request}->{query};
- $_->{'search.hits'} = map_matches($json->{matches});
- };
-
- # Learn results
- unless ($total_results) {
- if ($json->{totalResults} && $json->{totalResults} > -1) {
- $c->app->log->debug('Cache total result');
- $c->chi->set('total-' . $cache_url => $json->{totalResults}, '120min');
- };
- $c->stash('search.totalResults' => $json->{totalResults});
- };
-
- if ($json->{warning}) {
- # Temp
- $json->{warning} =~ s/;\s+null$//;
- $c->notify(warn => $json->{warning});
- };
-
- # Check for error
- $plugin->_notify_on_error($c, 0, $res, $json);
- }
-
- # Request failed
- else {
- return unless $cb;
- $plugin->_notify_on_error($c, 1, $tx->res);
- };
-
- # Run embedded template
- my $v = $cb->();
-
- # Delete useless stash keys
- foreach (@remove_from_stash) {
- delete $c->stash->{'search.' . $_};
- };
-
- return $v;
- }
- );
-
-
- # Establish 'search_hits' helper
- $mojo->helper(
- search_hits => sub {
- my $c = shift;
- my $cb = pop;
-
- if (!ref $cb || !(ref $cb eq 'CODE')) {
- $mojo->log->error("search_hits expects a code block");
- return '';
- };
-
- my $hits = delete $c->stash->{'search.hits'};
- my $string;
-
- # Iterate over all hits
- foreach (@$hits) {
- local $_ = $_;
- $c->stash('search.hit' => $_);
- $string .= $cb->($_);
- };
-
- # Delete unnecessary stash values
- delete $c->stash->{'search.hit'};
- return b($string || '');
- }
- );
-};
-
-sub _notify_on_error {
- my ($self, $c, $failure, $res) = @_;
- my $json = $res;
-
- my $log = $c->app->log;
-
- if (blessed $res) {
- if (blessed $res ne 'Mojo::JSON') {
- $json = $res->json;
- };
- }
- else {
- $json = undef;
- };
-
- if ($json) {
- if ($json->{error}) {
- # Temp
- $json->{error} =~ s/;\s+null$//;
- $c->notify(error => $json->{error});
- return;
- }
-
- # New error messages
- elsif ($json->{errstr}) {
- # Temp
- $json->{errstr} =~ s/;\s+null$//;
- $c->notify(error => $json->{errstr});
- return;
- }
-
- # policy service error messages
- elsif ($json->{status}) {
- $c->notify(error => 'Middleware error ' . $json->{status});
- return;
- };
- };
-
- if ($failure) {
- $c->notify(error => (
- ($res->{code} ? $res->{code} . ': ' : '') .
- ($res->{message} ? $res->{message} : 'Unknown error') .
- ' (remote)'
- ));
- };
-};
-
-
-1;
-
-__DATA__
-{
- "query" : "tokens:s:baum",
- "matches" : [ {
- "author" : "ErikDunsing,Ninjamask,Magnus",
- "textClass" : "freizeit-unterhaltung reisen",
- "corpusID" : "WPD",
- "title" : "Friedrich Wilhelm (Brandenburg)",
- "foundries" : "opennlp opennlp/morpho xip xip/morpho xip/constituency xip/dependency base base/sentences base/paragraphs mate mate/morpho mate/dependency corenlp corenlp/namedentities corenlp/namedentities/ne_dewac_175m_600 corenlp/namedentities corenlp/namedentities/ne_hgc_175m_600 connexor connexor/morpho connexor/syntax connexor/phrase treetagger treetagger/morpho",
- "tokenization" : "opennlp#tokens",
- "field" : "tokens",
- "startMore" : true,
- "endMore" : true,
- "docID" : "WPD_WWW.04738",
- "snippet" : "<span class=\"context-left\"><span class=\"more\"></span>wurde ihm der Mirabolanen frucht und </span><span class=\"match\">baum</span><span class=\"context-right\"> <Terminalia catappa L. oder Terminalia citrina<span class=\"more\"></span></span>",
- "ID" : "match-WPD!WPD_WWW.04738-p265-266",
- "pubDate" : "2005-03-28"
- }, {
- "author" : "Sk-Bot,Wiegels,Katharina",
- "textClass" : "freizeit-unterhaltung reisen",
- "corpusID" : "WPD",
- "title" : "Hans Heinrich von Wuthenau",
- "foundries" : "corenlp corenlp/namedentities corenlp/namedentities/ne_dewac_175m_600 corenlp/namedentities corenlp/namedentities/ne_hgc_175m_600 opennlp opennlp/morpho connexor connexor/morpho connexor/syntax connexor/phrase base base/sentences base/paragraphs xip xip/morpho xip/constituency xip/dependency mate mate/morpho mate/dependency treetagger treetagger/morpho",
- "tokenization" : "opennlp#tokens",
- "field" : "tokens",
- "startMore" : true,
- "endMore" : true,
- "docID" : "WPD_WWW.07095",
- "snippet" : "<span class=\"context-left\"><span class=\"more\"></span>:Jch heisse drumb Gerad', das dieser </span><span class=\"match\">baum</span><span class=\"context-right\"> uns zeigetDas man aufrichtig sey, nichts<span class=\"more\"></span></span>",
- "ID" : "match-WPD!WPD_WWW.07095-p154-155",
- "pubDate" : "2005-03-28"
- }, {
- "author" : "Zwobot,Holdbold,WiseWoman",
- "textClass" : "kultur literatur",
- "corpusID" : "WPD",
- "title" : "Johann Rist",
- "foundries" : "mate mate/morpho mate/dependency xip xip/morpho xip/constituency xip/dependency corenlp corenlp/namedentities corenlp/namedentities/ne_dewac_175m_600 corenlp/namedentities corenlp/namedentities/ne_hgc_175m_600 treetagger treetagger/morpho connexor connexor/morpho connexor/syntax connexor/phrase opennlp opennlp/morpho base base/sentences base/paragraphs",
- "tokenization" : "opennlp#tokens",
- "field" : "tokens",
- "startMore" : true,
- "endMore" : true,
- "docID" : "WPD_RRR.05363",
- "snippet" : "<span class=\"context-left\"><span class=\"more\"></span>Emblem wird das heilige holtz der </span><span class=\"match\">baum</span><span class=\"context-right\"> Guajacum genant, fremde Brasilianische landschaft, der<span class=\"more\"></span></span>",
- "ID" : "match-WPD!WPD_RRR.05363-p373-374",
- "pubDate" : "2005-03-28"
- } ],
- "totalResults" : 3,
- "startIndex" : 0,
- "itemsPerPage" : 25,
- "benchmarkSearchResults" : "1.638 ms",
- "benchmarkHitCounter" : "4.332 ms",
- "error" : null,
- "request" : {
- "@context" : "http://ids-mannheim.de/ns/KorAP/json-ld/v0.1/context.jsonld",
- "query" : {
- "@type" : "korap:token",
- "wrap" : {
- "@type" : "korap:term",
- "key" : "baum",
- "layer" : "orth",
- "match" : "match:eq"
- }
- },
- "collections" : [ {
- "@type" : "korap:meta-filter",
- "@value" : {
- "@type" : "korap:term",
- "@field" : "korap:field#corpusID",
- "@value" : "WPD"
- }
- } ],
- "meta" : { }
- },
- "context" : {
- "left" : [ "token", 6 ],
- "right" : [ "token", 6 ]
- }
-}
diff --git a/lib/Korap/Search.pm b/lib/Korap/Search.pm
deleted file mode 100644
index 5980b3f..0000000
--- a/lib/Korap/Search.pm
+++ /dev/null
@@ -1,58 +0,0 @@
-package Korap::Search;
-use Mojo::Base 'Mojolicious::Controller';
-
-# Add X-Forwarded-For to user agent call everywhere
-
-
-# This action will render a template
-sub remote {
- my $c = shift;
- $c->layout('default');
- $c->title('KorAP');
-
- my $api = $c->config('KorAP')->{'api-0.1'};
-
- # Todo: Support query building directly
- if ((scalar $c->param('action') // '') eq 'inspect') {
- my $url = Mojo::URL->new($api)->path('search');
-
- $url->query({
- q => scalar $c->param('q') // '',
- ql => scalar $c->param('ql') // 'poliqarp'
- });
-
- my $tx = $c->ua->build_tx(TRACE => $url);
-
- $c->app->log->debug('Trace URL: ' . $url);
- $c->app->log->debug('Trace Request: ' . $tx->req->url);
-
- if (my $response = $c->ua->start($tx)->success) {
- $c->stash('search.query' => $response->json);
- }
- else {
- $c->notify(error => 'Unable to trace query');
- };
-
- $c->param(cutoff => 1);
- return $c->render(template => 'query_info');
- }
-
- elsif ($c->param('snippet')) {
- $c->layout('snippet');
- };
-
- $c->render(template => 'search');
-};
-
-
-1;
-
-
-__END__
-
-[{"shared":false,"id":1,"managed":true,"created":1401193381119,"stats":{"documents":196510,"tokens":51545081,"sentences":4116282,"paragraphs":2034752},"query":[{"@type":"korap:meta-filter","@value":{"@type":"korap:term","@field":"korap:field#corpusID","@value":"WPD"}}],"description":"Die freie Enzyklopädie","name":"Wikipedia","foundries":"base;corenlp;mate;mpt;opennlp;tt;xip"}]
-
-
-http://10.0.10.13:8070/api/v1/resource/matchInfo?id=match-WPD!WPD_NNN.02848-p1223-1224&f=mate&l=
-
-{"author":"Filzstift,Alexander Sommer,TheK","textClass":"freizeit-unterhaltung reisen","corpusID":"WPD","title":"Neuseeland","foundries":"xip xip/morpho xip/constituency xip/dependency corenlp corenlp/namedentities corenlp/namedentities/ne_dewac_175m_600 corenlp/namedentities corenlp/namedentities/ne_hgc_175m_600 mate mate/morpho mate/dependency connexor connexor/morpho connexor/syntax connexor/phrase treetagger treetagger/morpho base base/sentences base/paragraphs opennlp opennlp/morpho","tokenization":"opennlp#tokens","field":"tokens","startMore":false,"endMore":false,"docID":"WPD_NNN.02848","snippet":"<span class=\"context-left\"></span><span class=\"match\"><span title=\"mate/l:besonders\"><span title=\"mate/p:ADV\">Besonders</span></span> <span title=\"mate/l:auffällig\"><span title=\"mate/m:degree:pos\"><span title=\"mate/p:ADJD\">auffällig</span></span></span> <span title=\"mate/l:sein\"><span title=\"mate/m:mood:ind\"><span title=\"mate/m:number:pl\"><span title=\"mate/m:person:3\"><span title=\"mate/m:tense:pres\"><span title=\"mate/p:VAFIN\">sind</span></span></span></span></span></span> <span title=\"mate/l:schließlich\"><span title=\"mate/p:ADV\">schließlich</span></span> <span title=\"mate/l:noch\"><span title=\"mate/p:ADV\">noch</span></span> <span title=\"mate/l:der\"><span title=\"mate/m:case:nom\"><span title=\"mate/m:gender:masc\"><span title=\"mate/m:number:sg\"><span title=\"mate/p:ART\">der</span></span></span></span></span> <span title=\"mate/l:pohutukawa\"><span title=\"mate/m:case:nom\"><span title=\"mate/m:gender:masc\"><span title=\"mate/m:number:sg\"><span title=\"mate/p:NE\">Pohutukawa</span></span></span></span></span> <span title=\"mate/l:und\"><span title=\"mate/p:KON\">und</span></span> <span title=\"mate/l:der\"><span title=\"mate/m:case:nom\"><span title=\"mate/m:gender:masc\"><span title=\"mate/m:number:sg\"><span title=\"mate/p:ART\">der</span></span></span></span></span> <span title=\"mate/l:cabbage\"><span title=\"mate/m:case:nom\"><span title=\"mate/m:gender:masc\"><span title=\"mate/m:number:sg\"><span title=\"mate/p:NE\">Cabbage</span></span></span></span></span> <span title=\"mate/l:tree\"><span title=\"mate/m:gender:masc\"><span title=\"mate/m:number:sg\"><span title=\"mate/p:NE\"><span title=\"mate/m:case:nom\">Tree</span></span></span></span></span></span><span class=\"context-right\"></span>","ID":"match-WPD!WPD_NNN.02848-p1213-1224","pubDate":"2005-03-28","context":{"left":["token",0],"right":["token",0]}}
diff --git a/lib/Korap/Tutorial.pm b/lib/Korap/Tutorial.pm
deleted file mode 100644
index f9d99da..0000000
--- a/lib/Korap/Tutorial.pm
+++ /dev/null
@@ -1,24 +0,0 @@
-package Korap::Tutorial;
-use Mojo::Base 'Mojolicious::Controller';
-
-sub page {
- my $c = shift;
-
- if ($c->param('embedded')) {
- $c->layout('snippet');
- $c->stash(embedded => 1);
- }
- else {
- $c->layout('default');
- };
- $c->title('KorAP');
-
- my $page = $c->stash('tutorial');
- return $c->render(template => 'tutorial/' . $page);
-};
-
-
-1;
-
-
-__END__