admin/include/fuzzy.php

<?php
/**********************************************************
Function: newResPageHeader
Author: Paul Bramscher
Last Modified: 06.25.2003
***********************************************************
Draws the HTML header of the enter new resource page at the
point of checking for possible matches.  The header is
pulled out separately here, since we may not always want
it to display.  In the event of no fuzzy (or exact)
matches we redirect the user to formResource and, hence,
cannot output this header.
**********************************************************/
function newResPageHeader() {

        // Load globals
        include ("global_vars.php");

        // HTML header
        printf("<HTML>\n");
        printf("<HEAD>\n");
        printf("<title>LibData: Enter a New Resource</title>\n");
        printf("<link rel=\"stylesheet\" href=\"%s\" type=\"text/css\">\n", $GLOBAL_ADMIN_CSS);
        printf("</HEAD>\n");

        // Page header
        include($GLOBAL_ADMIN_HEADER);
        printf("<center>\n");
        printf("<h3>Adding New Resource..</h3>\n");

}


/**********************************************************
Function: resTitlePat
Original Author: Paul Bramscher <brams006@umn.edu>
Last Modified: 03.03.2004
***********************************************************
Comments:
This function takes $string as input and builds eight
permutations of it into an SQL query against resource
titles.  These permutations are meant as possible matches.
This is not an exact science, and this algorithm could
certainly be further tweaked.

Pass definitions:

pass1 = left 50% of string, remainder wildcarded
pass2 = middle 80%, both ends wildcarded
pass3 = compacted without spaces
pass4 = a/and/the articles removed
pass5 = every other starting with 0 position
pass6 = every other starting with 1 position
pass7 = replacing 'ies' with 'y'
pass8 = replacing 'y' with 'ies'
**********************************************************/
function resTitlePat($string) {

        // Save this for display purposes
        $title = $string;
        $title = stripslashes($string); 

        // Clean up the string -- we must remove single quotes and slashes
        $string = ereg_replace("'","",$string);
        $string = stripslashes($string);

        // Initialize
        $pass1 = "";
        $pass2 = "";
        $pass3 = "";
        $pass4 = "";
        $pass5 = "";
        $pass6 = "";
        $pass7 = "";
        $pass8 = "";

        // String sizes.
        $length = strlen($string);
        $max_length = ceil($length * 2);

        // Permute #1: Match left 50%
        $left50_pos = floor($length * .5);
        $pass1 = substr($string, 0, $left50_pos);

        // Permute #2: Cut-out and float match middle 80% substring
        $left20_pos = floor($length * .2);
        $pass2 = substr($string, $left20_pos, $length - ($left20_pos * 2));

        // Permute #3: Compacting, remove all whitespaces.
        $pass3 = $string;
        $pass3 = ereg_replace(" ","",$pass3);

        // Permute #4: Pulling out a/an/the articles
        $pass4 = $string;
        $pass4 = ereg_replace("the ","",$pass4);
        $pass4 = ereg_replace("a ","",$pass4);
        $pass4 = ereg_replace("an ","",$pass4);

        // Permute #5: Wildcard every other position, starting with 0
        $pass5 = $string;
        for ($x = 0; $x < strlen($pass5); $x++) {

                if ($x % 2 == 0) $pass5[$x] = "_";

        }

        // Permute #6: Wildcard every other position, starting with 1
        $pass6 = $string;
        for ($x = 0; $x < strlen($pass6); $x++) {

                if ($x % 2 == 1) $pass6[$x] = "_";

        }

        // Permute #7: Replacing 'ies' with 'y'
        $pass7 = $string;
        $pass7 = ereg_replace("ies","y", $pass7);

        // Permute #8: Replacing 'y' with 'ies'
        $pass8 = $string;
        $pass8 = ereg_replace("y","ies", $pass8);

        // Assemble the SQL
        $sql = "SELECT resource_id, title FROM resource WHERE (title LIKE '"
                . $pass1
                . "%' OR title LIKE '%"
                . $pass2
                . "%' OR title LIKE '%"
                . $pass3
                . "%' OR title LIKE '%"
                . $pass4
                . "%' OR title LIKE '%"
                . $pass5
                . "%' OR title LIKE '%"
                . $pass6
                . "%' OR title LIKE '%"
                . $pass7
                . "%' OR title LIKE '%"
                . $pass8                
                . "%') AND (LENGTH(title) >= "
                . $left50_pos
                . " AND LENGTH(title) <= "
                . $max_length
                . ")";

        // Debugging
        // printf("sql was: %s", $sql);

        $rs = xx_tryquery($sql);
        $hits = xx_num_rows($rs);
        if ($hits > 0) {

                // Draw page header
                newResPageHeader();

                // Table
                printf("<table width = \"60%%\" border = \"3\" cellpadding =\"4\" class=\"backLight\">\n");
                printf("<tr><td>\n");
                printf("<b>Messages:</b><br>\n");

                printf ("Possible duplicate resource titles found with title '%s'.  Are you sure you want to continue?<br><br>\n", $title);
                while ($row = xx_fetch_array ($rs, xx_ASSOC)) {
                        $title = $row["title"];
                        $resource_id = $row["resource_id"];

                        printf("<b>Resource ID:</b> %d ", $resource_id);
                        printf("<b>Title:</b> %s<BR>\n", $title);

                } // this result set

                // Close things
                printf("<br><br>\n");
                printf("</td></tr></table>\n");
                printf("</center>\n");                  

        }


        return $hits;

} // end of php function
?>
1	dpavlin	10	<?php
2			/**********************************************************
3			Function: newResPageHeader
4			Author: Paul Bramscher
5			Last Modified: 06.25.2003
6			***********************************************************
7			Draws the HTML header of the enter new resource page at the
8			point of checking for possible matches. The header is
9			pulled out separately here, since we may not always want
10			it to display. In the event of no fuzzy (or exact)
11			matches we redirect the user to formResource and, hence,
12			cannot output this header.
13			**********************************************************/
14			function newResPageHeader() {
15
16			// Load globals
17			include ("global_vars.php");
18
19			// HTML header
20			printf("<HTML>\n");
21			printf("<HEAD>\n");
22			printf("<title>LibData: Enter a New Resource</title>\n");
23			printf("<link rel=\"stylesheet\" href=\"%s\" type=\"text/css\">\n", $GLOBAL_ADMIN_CSS);
24			printf("</HEAD>\n");
25
26			// Page header
27			include($GLOBAL_ADMIN_HEADER);
28			printf("<center>\n");
29			printf("<h3>Adding New Resource..</h3>\n");
30
31			}
32
33
34			/**********************************************************
35			Function: resTitlePat
36	dpavlin	72	Original Author: Paul Bramscher <brams006@umn.edu>
37			Last Modified: 03.03.2004
38	dpavlin	10	***********************************************************
39			Comments:
40			This function takes $string as input and builds eight
41			permutations of it into an SQL query against resource
42			titles. These permutations are meant as possible matches.
43			This is not an exact science, and this algorithm could
44			certainly be further tweaked.
45
46			Pass definitions:
47
48			pass1 = left 50% of string, remainder wildcarded
49			pass2 = middle 80%, both ends wildcarded
50			pass3 = compacted without spaces
51			pass4 = a/and/the articles removed
52			pass5 = every other starting with 0 position
53			pass6 = every other starting with 1 position
54			pass7 = replacing 'ies' with 'y'
55			pass8 = replacing 'y' with 'ies'
56			**********************************************************/
57	dpavlin	72	function resTitlePat($string) {
58	dpavlin	10
59			// Save this for display purposes
60			$title = $string;
61			$title = stripslashes($string);
62
63			// Clean up the string -- we must remove single quotes and slashes
64			$string = ereg_replace("'","",$string);
65			$string = stripslashes($string);
66
67			// Initialize
68			$pass1 = "";
69			$pass2 = "";
70			$pass3 = "";
71			$pass4 = "";
72			$pass5 = "";
73			$pass6 = "";
74			$pass7 = "";
75			$pass8 = "";
76
77			// String sizes.
78			$length = strlen($string);
79			$max_length = ceil($length * 2);
80
81			// Permute #1: Match left 50%
82			$left50_pos = floor($length * .5);
83			$pass1 = substr($string, 0, $left50_pos);
84
85			// Permute #2: Cut-out and float match middle 80% substring
86			$left20_pos = floor($length * .2);
87			$pass2 = substr($string, $left20_pos, $length - ($left20_pos * 2));
88
89			// Permute #3: Compacting, remove all whitespaces.
90			$pass3 = $string;
91			$pass3 = ereg_replace(" ","",$pass3);
92
93			// Permute #4: Pulling out a/an/the articles
94			$pass4 = $string;
95			$pass4 = ereg_replace("the ","",$pass4);
96			$pass4 = ereg_replace("a ","",$pass4);
97			$pass4 = ereg_replace("an ","",$pass4);
98
99			// Permute #5: Wildcard every other position, starting with 0
100			$pass5 = $string;
101			for ($x = 0; $x < strlen($pass5); $x++) {
102
103			if ($x % 2 == 0) $pass5[$x] = "_";
104
105			}
106
107			// Permute #6: Wildcard every other position, starting with 1
108			$pass6 = $string;
109			for ($x = 0; $x < strlen($pass6); $x++) {
110
111			if ($x % 2 == 1) $pass6[$x] = "_";
112
113			}
114
115			// Permute #7: Replacing 'ies' with 'y'
116			$pass7 = $string;
117			$pass7 = ereg_replace("ies","y", $pass7);
118
119			// Permute #8: Replacing 'y' with 'ies'
120			$pass8 = $string;
121			$pass8 = ereg_replace("y","ies", $pass8);
122
123			// Assemble the SQL
124			$sql = "SELECT resource_id, title FROM resource WHERE (title LIKE '"
125			. $pass1
126			. "%' OR title LIKE '%"
127			. $pass2
128			. "%' OR title LIKE '%"
129			. $pass3
130			. "%' OR title LIKE '%"
131			. $pass4
132			. "%' OR title LIKE '%"
133			. $pass5
134			. "%' OR title LIKE '%"
135			. $pass6
136			. "%' OR title LIKE '%"
137			. $pass7
138			. "%' OR title LIKE '%"
139			. $pass8
140			. "%') AND (LENGTH(title) >= "
141			. $left50_pos
142			. " AND LENGTH(title) <= "
143			. $max_length
144			. ")";
145
146			// Debugging
147			// printf("sql was: %s", $sql);
148
149	dpavlin	72	$rs = xx_tryquery($sql);
150	dpavlin	42	$hits = xx_num_rows($rs);
151	dpavlin	10	if ($hits > 0) {
152
153			// Draw page header
154			newResPageHeader();
155
156			// Table
157			printf("<table width = \"60%%\" border = \"3\" cellpadding =\"4\" class=\"backLight\">\n");
158	dpavlin	72	printf("<tr><td>\n");
159			printf("<b>Messages:</b><br>\n");
160	dpavlin	10
161			printf ("Possible duplicate resource titles found with title '%s'. Are you sure you want to continue?<br><br>\n", $title);
162	dpavlin	72	while ($row = xx_fetch_array ($rs, xx_ASSOC)) {
163	dpavlin	10	$title = $row["title"];
164			$resource_id = $row["resource_id"];
165
166			printf("<b>Resource ID:</b> %d ", $resource_id);
167	dpavlin	72	printf("<b>Title:</b> %s<BR>\n", $title);
168	dpavlin	10
169			} // this result set
170
171			// Close things
172			printf("<br><br>\n");
173			printf("</td></tr></table>\n");
174			printf("</center>\n");
175
176			}
177
178
179			return $hits;
180
181			} // end of php function
182	dpavlin	72	?>