creating tree traversal hierarchy from scratch using PHP/MySQL

Question

I am currently developing a category hierarchy, and I got the point of creating tree treversal i think. But I need to add a new node into this hierarchy usign PHP function.

The problem is rebuild_tree function would be good enough (in other words, efficient with large trees).

Sample query:

 CREATE TABLE `t_categories`(
   `id` INTEGER UNSIGNED NOT NULL AUTO_INCREMENT,
   `title` VARCHAR(45) NOT NULL,
   `lft` INTEGER UNSIGNED NOT NULL,
   `rght` INTEGER UNSIGNED NOT NULL,
   PRIMARY KEY (`id`)
 );

 INSERT INTO t_categories (title, lft, rght) VALUES ('Cat 1',1,16);
 INSERT INTO t_categories (title, lft, rght) VALUES ('Cat 2',2,3);
 INSERT INTO t_categories (title, lft, rght) VALUES ('Cat 3',4,7);
 INSERT INTO t_categories (title, lft, rght) VALUES ('Cat 4',5,6);
 INSERT INTO t_categories (title, lft, rght) VALUES ('Cat 5',8,13);
 INSERT INTO t_categories (title, lft, rght) VALUES ('Cat 6',9,12);
 INSERT INTO t_categories (title, lft, rght) VALUES ('Cat 7',10,11);
 INSERT INTO t_categories (title, lft, rght) VALUES ('Cat 8',14,15);

Table results look like that:

 ID    TITLE   LFT    RGHT
 1     Cat1    1      16
 2     Cat2    2      3
 3     Cat3    4      7
 4     Cat4    5      6
 5     Cat5    8      13
 6     Cat6    9      12
 7     Cat7    10     11
 8     Cat8    14     15

I gave sample data above, but I need to create completely new node from scratch as well.

So, how can I add a new node into this tree using PHP function that efficients with large trees?

Answer 1

This is a celko tree. Simpelst approach would be depth-first traversal of the tree and update only the left value and then in a recursive manner the right value. Insertition is much more costly.

Answer 2

I recommend that you add a "parent id" field to your table structure instead of the "left" and "right" fields. If its important to you have an order for the children items, use also a "localorder" int field.

With the current structure, each time you want to add an item, you have to check if there is a previous item, for the first item, and to check if there is a final item, for a last item.

CREATE TABLE `t_categories`(
   `keyid` INTEGER UNSIGNED NOT NULL,
   `title` VARCHAR(45) NOT NULL,
   `parentid` INTEGER UNSIGNED NOT NULL,
   `sortorder` INTEGER UNSIGNED NOT NULL,
   PRIMARY KEY (`id`)
 );

-- root item, no parent
 INSERT INTO t_categories (keyid, title, parentid, sortorder) VALUES (1, 'Root', 0, 0);

-- first level
 INSERT INTO t_categories (keyid, title, parentid, sortorder) VALUES (2, 'a:', 1, 1);
 INSERT INTO t_categories (keyid, title, parentid, sortorder) VALUES (3, 'b:', 1, 2);
 INSERT INTO t_categories (keyid, title, parentid, sortorder) VALUES (3, 'c:', 1, 3);
 INSERT INTO t_categories (keyid, title, parentid, sortorder) VALUES (4, 'd:', 1, 4);

-- second level

 INSERT INTO t_categories (keyid, title, parentid, sortorder) VALUES (5, 'a:\temp', 2, 1);
 INSERT INTO t_categories (keyid, title, parentid, sortorder) VALUES (6, 'a:\docs'', 2, 2);
 INSERT INTO t_categories (keyid, title, parentid, sortorder) VALUES (7, 'a:\music', 2, 3);

 INSERT INTO t_categories (keyid, title, parentid, sortorder) VALUES (8, 'c:\temp', 4, 1);
 INSERT INTO t_categories (keyid, title, parentid, sortorder) VALUES (9, 'c:\docs'', 4, 2);
 INSERT INTO t_categories (keyid, title, parentid, sortorder) VALUES (10, 'c:\music', 4, 3);

-- and so on

Answer 3

The same tree is used by the huffman compresssion to count the occurrence of a letter in the given document. I think to encode a string the algorithm then use also a depth-first traversal so that the letter with the most occurrence is encoded with the least bits possible. I don't know if it is useful here but the lowest entropy of a text is found using shannon theorm -log(x)+log(2) where x is the letter and log(2) is the base in bits it is always 2.

Answer 4

My previous answer is incomplete. It's a summary, the whole code is to long to be in a forum post. Forgot to ask, procedural PHP, or Object Oriented PHP ?

MySQL: CREATE TABLE t_categories ( keyid INTEGER UNSIGNED NOT NULL, title VARCHAR(45) NOT NULL, parentid INTEGER UNSIGNED NOT NULL, sortorder INTEGER UNSIGNED NOT NULL, PRIMARY KEY ( id ) );

PHP function headers:

// globar var, acts like a type /* typedef */ treeNodeType = array( "keyid" => 0, "title" => "", "parentid" => 0, "sortorder" => 0, )

// globar var, acts like a type /* typedef */ treeType = array( "root" => nil, "whatever" => "", )

/* treeNodeType / function insertNode(/ treeType / ATree, AParentId, ATitle) { ... } / void / function deleteNode(/ treeType */ ATree, AKeyId) { ... }

// --> MAIN main();

/* void */ function main() { // treeType myTree; myTree = treeType;

// insert root = 0 rootNode = insertNode(myTree, 0, '[PC]');

... }