| 1 | package ClassifyTreeModel;
|
|---|
| 2 |
|
|---|
| 3 | use ClassifyTreeNode;
|
|---|
| 4 | use strict;
|
|---|
| 5 |
|
|---|
| 6 |
|
|---|
| 7 | # /** Constructor
|
|---|
| 8 | # *
|
|---|
| 9 | # * @param $class The name of the class to bless as a string
|
|---|
| 10 | # * @param $collection The name of the collection whose info database we
|
|---|
| 11 | # * will be accessing as a string
|
|---|
| 12 | # * @param $root The oid of the root node of the classifier as a
|
|---|
| 13 | # * string
|
|---|
| 14 | # * @return A reference to the ClassifyTreeModel object
|
|---|
| 15 | # *
|
|---|
| 16 | # * @author John Thompson, DL Consulting Ltd.
|
|---|
| 17 | # */
|
|---|
| 18 | sub new()
|
|---|
| 19 | {
|
|---|
| 20 | my ($class, $collection, $infodbtype, $root) = @_;
|
|---|
| 21 | my $debug = 0;
|
|---|
| 22 | print STDERR "ClassifyTreeModel.new(\"$collection\", $infodbtype, \"$root\")\n" unless !$debug;
|
|---|
| 23 | # Store the variables
|
|---|
| 24 | my $self = {};
|
|---|
| 25 | $self->{'collection'} = $collection;
|
|---|
| 26 | $self->{'infodbtype'} = $infodbtype;
|
|---|
| 27 | $self->{'debug'} = $debug;
|
|---|
| 28 | $self->{'root'} = $root;
|
|---|
| 29 | # Bless me father for I have sinned
|
|---|
| 30 | bless $self, $class;
|
|---|
| 31 | return $self;
|
|---|
| 32 | }
|
|---|
| 33 | # /** new() **/
|
|---|
| 34 |
|
|---|
| 35 | # /** Given a path and a document id, add this document to the classifier tree
|
|---|
| 36 | # * creating any necessary tree nodes first.
|
|---|
| 37 | # *
|
|---|
| 38 | # * @param $value The path to store this document in
|
|---|
| 39 | # * @param $oid Unique identifier of a document
|
|---|
| 40 | # *
|
|---|
| 41 | # * @author John Thompson, DL Consulting Ltd.
|
|---|
| 42 | # */
|
|---|
| 43 | sub addDocument()
|
|---|
| 44 | {
|
|---|
| 45 | my ($self, $value, $oid) = @_;
|
|---|
| 46 | print STDERR "ClassifyTreeModel.addDocument(\"$value\", \"$oid\")\n" unless !$self->{'debug'};
|
|---|
| 47 | # Generate a treepath object from the metadata value, remembering to prefix
|
|---|
| 48 | # with the root nodes path
|
|---|
| 49 | my $root_node_obj = $self->getRootNode();
|
|---|
| 50 | my $path_obj = $root_node_obj->getPath();
|
|---|
| 51 | $path_obj->addPathComponent($value);
|
|---|
| 52 | # Ensure that this classifier node, and if necessary its ancestor nodes,
|
|---|
| 53 | # exist in our tree.
|
|---|
| 54 | my $node_obj = $self->getNodeByPath($path_obj);
|
|---|
| 55 | if (!$node_obj)
|
|---|
| 56 | {
|
|---|
| 57 | # The node doesn't exist, so we need to add it
|
|---|
| 58 | $node_obj = $self->addNode($path_obj);
|
|---|
| 59 | }
|
|---|
| 60 | # Add the document to the node.
|
|---|
| 61 | $node_obj->addDocument($oid);
|
|---|
| 62 | # Done.
|
|---|
| 63 | }
|
|---|
| 64 | # /** addDocument() **/
|
|---|
| 65 |
|
|---|
| 66 | # /** Add a node into the tree first ensuring all its parent nodes are inserted
|
|---|
| 67 | # * to.
|
|---|
| 68 | # *
|
|---|
| 69 | # * @param $path_obj The path to insert the new node at
|
|---|
| 70 | # *
|
|---|
| 71 | # * @author John Thompson, DL Consulting Ltd.
|
|---|
| 72 | # */
|
|---|
| 73 | sub addNode()
|
|---|
| 74 | {
|
|---|
| 75 | my ($self, $path_obj) = @_;
|
|---|
| 76 | print STDERR "ClassifyTreeModel.addNode(\"" . $path_obj->toString() . "\")\n" unless !$self->{'debug'};
|
|---|
| 77 | # Ensure the parent exists, assuming we aren't at the root
|
|---|
| 78 | my $parent_path_obj = $path_obj->getParentPath();
|
|---|
| 79 | #rint STDERR "* parent path: " . $parent_path_obj->toString() . "\n";
|
|---|
| 80 | my $parent_node_obj = $self->getNodeByPath($parent_path_obj);
|
|---|
| 81 | #rint STDERR "* does parent node already exist? " . $parent_node_obj . "\n";
|
|---|
| 82 | #rint STDERR "* are we at the root node yet? " . $parent_path_obj->isRootPath() . "\n";
|
|---|
| 83 | if (!$parent_node_obj && !$parent_path_obj->isRootPath())
|
|---|
| 84 | {
|
|---|
| 85 | #rint STDERR "* recursive call!\n";
|
|---|
| 86 | $parent_node_obj = $self->addNode($parent_path_obj);
|
|---|
| 87 | }
|
|---|
| 88 | # Insert this node into it's parent.
|
|---|
| 89 | return $parent_node_obj->insertNode($path_obj);
|
|---|
| 90 | }
|
|---|
| 91 | # /** addNode() **/
|
|---|
| 92 |
|
|---|
| 93 | # /** Retrieve the name of the collection this model is drawing from.
|
|---|
| 94 | # *
|
|---|
| 95 | # * @return The collection name as a string
|
|---|
| 96 | # *
|
|---|
| 97 | # * @author John Thompson, DL Consulting Ltd.
|
|---|
| 98 | # */
|
|---|
| 99 | sub getCollection()
|
|---|
| 100 | {
|
|---|
| 101 | my ($self) = @_;
|
|---|
| 102 | print STDERR "ClassifyTreeModel.getCollection()\n" unless !$self->{'debug'};
|
|---|
| 103 | return $self->{'collection'};
|
|---|
| 104 | }
|
|---|
| 105 | # /** getCollection() **/
|
|---|
| 106 |
|
|---|
| 107 | # /** Retrieve a node from this tree based upon its CLID (OID).
|
|---|
| 108 | # * @param $clid The CLID as a string
|
|---|
| 109 | # * @return The indicated ClassifyTreeNode or null
|
|---|
| 110 | sub getNodeByCLID()
|
|---|
| 111 | {
|
|---|
| 112 | my ($self, $clid) = @_;
|
|---|
| 113 | print STDERR "ClassifyTreeModel.getNodeByCLID(\"$clid\")\n" unless !$self->{'debug'};
|
|---|
| 114 | my $result = 0;
|
|---|
| 115 | # Test if this clid is even in our tree
|
|---|
| 116 | if($clid !~ /^$self->{'root'}/)
|
|---|
| 117 | {
|
|---|
| 118 | print STDERR "Requested node $clid, which isn't part of " . $self->{'root'} . "\n";
|
|---|
| 119 | return 0;
|
|---|
| 120 | }
|
|---|
| 121 | # Unfortunately I have to check that there is text to retrieve before I
|
|---|
| 122 | # create a new node.
|
|---|
| 123 |
|
|---|
| 124 | my $index_text_directory_path = &util::filename_cat($ENV{'GSDLHOME'}, "collect", $self->getCollection(), "index", "text");
|
|---|
| 125 | my $infodb_file_path = &dbutil::get_infodb_file_path($self->{'infodbtype'}, $self->getCollection(), $index_text_directory_path);
|
|---|
| 126 | if (&dbutil::read_infodb_entry($self->{'infodbtype'}, $infodb_file_path, $clid) =~ /\w+/)
|
|---|
| 127 | {
|
|---|
| 128 | # Since the CLID can directly reference the correct entry in the info database we
|
|---|
| 129 | # just create the node and return it
|
|---|
| 130 | $result = new ClassifyTreeNode($self, $clid);
|
|---|
| 131 | }
|
|---|
| 132 | return $result;
|
|---|
| 133 | }
|
|---|
| 134 |
|
|---|
| 135 | # /** Retrieve a node from this tree based upon a path
|
|---|
| 136 | # *
|
|---|
| 137 | # * @param $path The path to the node as a ClassifyTreePath
|
|---|
| 138 | # * @return The indicated ClassifyTreeNode or null
|
|---|
| 139 | # *
|
|---|
| 140 | # * @author John Thompson, DL Consulting Ltd.
|
|---|
| 141 | # */
|
|---|
| 142 | sub getNodeByPath()
|
|---|
| 143 | {
|
|---|
| 144 | my ($self, $path_obj) = @_;
|
|---|
| 145 | print STDERR "ClassifyTreeModel.getNodeByPath(\"" . $path_obj->toString() . "\")\n" unless !$self->{'debug'};
|
|---|
| 146 | # Starting at the ROOT of the tree, and with the first path component,
|
|---|
| 147 | # recursively descend through the tree looking for the node - we can assume
|
|---|
| 148 | # that we've found the root node (otherwise we won't be in a tree)
|
|---|
| 149 | my $cur_node_obj = $self->getRootNode();
|
|---|
| 150 | my $cur_path_obj = $cur_node_obj->getPath();
|
|---|
| 151 | my $depth = 1;
|
|---|
| 152 | # Continue till we either find the node we want, or run out a nodes
|
|---|
| 153 | while(!$cur_node_obj->getPath()->equals($path_obj))
|
|---|
| 154 | {
|
|---|
| 155 | # Append the path component at this depth to the current path we
|
|---|
| 156 | # are searching for
|
|---|
| 157 | $cur_path_obj->addPathComponent($path_obj->getPathComponent($depth));
|
|---|
| 158 | $depth++;
|
|---|
| 159 | #rint STDERR "Searching " . $cur_node_obj->getPath()->toString() . "'s children for: " . $cur_path_obj->toString() . "\n";
|
|---|
| 160 | # Search through the current nodes children, looking for one that
|
|---|
| 161 | # matches the current path
|
|---|
| 162 | my $found = 0;
|
|---|
| 163 | foreach my $child_node_obj ($cur_node_obj->getChildren())
|
|---|
| 164 | {
|
|---|
| 165 | #rint STDERR "* testing " . $child_node_obj->getPath()->toString() . "\n";
|
|---|
| 166 | if($child_node_obj->getPath()->equals($cur_path_obj))
|
|---|
| 167 | {
|
|---|
| 168 | $cur_node_obj = $child_node_obj;
|
|---|
| 169 | $found = 1;
|
|---|
| 170 | last;
|
|---|
| 171 | }
|
|---|
| 172 | }
|
|---|
| 173 | # Couldn't find any node with this path
|
|---|
| 174 | if(!$found)
|
|---|
| 175 | {
|
|---|
| 176 | #rint STDERR "* no such node exists!\n";
|
|---|
| 177 | return 0;
|
|---|
| 178 | }
|
|---|
| 179 | }
|
|---|
| 180 | return $cur_node_obj;
|
|---|
| 181 | }
|
|---|
| 182 | # /** getChild() **/
|
|---|
| 183 |
|
|---|
| 184 | # /** Retrieve the parent node of the given node.
|
|---|
| 185 | # *
|
|---|
| 186 | # * @param $child_node The node whose parent we want to retrieve
|
|---|
| 187 | # * @return The parent node, or 0 if this is the root
|
|---|
| 188 | # *
|
|---|
| 189 | # * @author John Thompson, DL Consulting Ltd.
|
|---|
| 190 | # */
|
|---|
| 191 | sub getParentNode()
|
|---|
| 192 | {
|
|---|
| 193 | my ($self, $child_node) = @_;
|
|---|
| 194 | print STDERR "ClassifyTreeModel.getParentNode()\n" unless !$self->{'debug'};
|
|---|
| 195 | return $child_node->getParentNode();
|
|---|
| 196 | }
|
|---|
| 197 | # /** getParentNode() **/
|
|---|
| 198 |
|
|---|
| 199 | sub getRootNode()
|
|---|
| 200 | {
|
|---|
| 201 | my ($self) = @_;
|
|---|
| 202 | print STDERR "ClassifyTreeModel.getRootNode()\n" unless !$self->{'debug'};
|
|---|
| 203 | return new ClassifyTreeNode($self, $self->{'root'});
|
|---|
| 204 | }
|
|---|
| 205 |
|
|---|
| 206 | # /** Remove the given document from the classifier tree, and then remove any
|
|---|
| 207 | # * empty nodes if required.
|
|---|
| 208 | # *
|
|---|
| 209 | # * @param $value The value which contains the path of the node to remove
|
|---|
| 210 | # * the document from
|
|---|
| 211 | # * @param $oid The unique identifier of the document to remove
|
|---|
| 212 | # * @param $remove_empty Sets whether empty nodes are removed
|
|---|
| 213 | # *
|
|---|
| 214 | # * @author John Thompson, DL Consulting Ltd.
|
|---|
| 215 | # */
|
|---|
| 216 | sub removeDocument()
|
|---|
| 217 | {
|
|---|
| 218 | my ($self, $path, $oid, $remove_empty) = @_;
|
|---|
| 219 | print STDERR "ClassifyTreeModel.removeDocument(\"$path\",\"$oid\",$remove_empty)\n" unless !$self->{'debug'};
|
|---|
| 220 | # Append to root path
|
|---|
| 221 | my $root_node_obj = $self->getRootNode();
|
|---|
| 222 | my $path_obj = $root_node_obj->getPath();
|
|---|
| 223 | $path_obj->addPathComponent($path);
|
|---|
| 224 | # Retrieve the node in question
|
|---|
| 225 | my $node_obj = $self->getNodeByPath($path_obj);
|
|---|
| 226 | # Check we retrieved a node
|
|---|
| 227 | if ($node_obj)
|
|---|
| 228 | {
|
|---|
| 229 | # Remove the document
|
|---|
| 230 | $node_obj->removeDocument($oid);
|
|---|
| 231 | # If we have been asked to remove empty nodes, do so now.
|
|---|
| 232 | if ($remove_empty)
|
|---|
| 233 | {
|
|---|
| 234 | my $cur_node_obj = $node_obj;
|
|---|
| 235 | my $empty_node_obj = 0;
|
|---|
| 236 | while ($cur_node_obj->getNumLeafDocs() == 0)
|
|---|
| 237 | {
|
|---|
| 238 | $empty_node_obj = $cur_node_obj;
|
|---|
| 239 | $cur_node_obj = $cur_node_obj->getParentNode();
|
|---|
| 240 | }
|
|---|
| 241 | if ($empty_node_obj)
|
|---|
| 242 | {
|
|---|
| 243 | # Try to retrieve the parent of this node
|
|---|
| 244 | my $parent_node_obj = $empty_node_obj->getParentNode();
|
|---|
| 245 | # As long as we have a parent (i.e. we aren't the root node) go
|
|---|
| 246 | # ahead and delete this subtree starting at empty node
|
|---|
| 247 | if ($parent_node_obj)
|
|---|
| 248 | {
|
|---|
| 249 | $parent_node_obj->removeNode($empty_node_obj);
|
|---|
| 250 | }
|
|---|
| 251 | }
|
|---|
| 252 | }
|
|---|
| 253 | }
|
|---|
| 254 | # If the node doesn't exist in this tree, then we can't very well remove
|
|---|
| 255 | # anything from it!
|
|---|
| 256 | }
|
|---|
| 257 | # /** removeDocument() **/
|
|---|
| 258 |
|
|---|
| 259 | 1;
|
|---|
