ESS_Bibliography_backend/classes/citation.php

<?php

/*
 * To change this license header, choose License Headers in Project Properties.
 * To change this template file, choose Tools | Templates
 * and open the template in the editor.
 */

/**
 * Description of citation
 *
 * @author may
 */
class citation {
    // entry point to parse citations. Type is needed.
    
    public static function parse ($citation, $type) {
        $record = -1;

        // check for DOI first
        // skip for speed

        $record = citation::_checkDOI($citation);
        
        // we don't have DOI
        if ($record == -1) {
            
            // while type is hardcoded here due to obvious reasons, 
            // friendly names can indeed be changed in the DB
            switch ($type) {

                case 1:
                    // Journal article 
                    $record = citation::_journal2arr($citation);
                    break;

                case 2:
                    // Book
                    $record = citation::_book2arr($citation);
                    break;

                case 3:
                    // Work paper
                    $record = citation::_workpaper2arr($citation);
                    break;
                
                case 4:
                    // Conference paper
                    $record = citation::_conference2arr($citation);
                    break;

                case 5:
                    // Report
                    $record = citation::_report2arr($citation);
                    break;
                
                case 6:
                    // Thesis
                    $record = citation::_thesis2arr($citation);
                    break;
                
                case 8:
                    // chapter
                    $record = citation::_chapter2arr($citation);
                    break;
                    
                                
                default:
                case 7: // project
                case 9: // ESRA paper
                case 10: // newspaper, magazine articles
                case 11: // manuscripts
                    return -2;
                    break;    
            }
        }
    
        return $record;
    }

    private static function _checkDOI ($citation) {
        $_d = -1;
        $doi = -1;
        // check for DOI (must be at the end of citation!)
        preg_match ('/(.*)doi:(.*)/i', $citation, $hits);

        if (count($hits) == 3) {
            $doi = trim($hits[2], " \n\r\t\v\0.");	    
        }
	
        preg_match ('/(.*)doi.org\/(.*)/i', $citation, $hits);
        if (count($hits) == 3) {
            $doi = trim($hits[2], " \n\r\t\v\0.");
        }
        
        if ($doi != -1) {
            $_d = doi::fetchData($doi);

        } 
        
        return $_d;
        
    }
    
    
    private static function _book2arr ($citation) {
        $back = -1;
        /*
         * Authors (YEAR). TITLE. City: Publisher.
         */
         preg_match ('/(.*?[^(]+)'   // authors  - before (
                    . '\('          //  (
                .'([0-9]?[^)]+)'    // year - digits before )
                . '\)\.'            // ).
                .'(.*[^.]+|.*[^?]+)'  // title in form 'title.' or 'title?' OR 'title (series)'. Two dots are a problem, sugest changing it to : .
                .'[\.?]'       // 
                .'(.*?[^\:]+)'      // City
                .'[\:]'       // series OR dot.
                .'(.*?[^\.]+)'    // publisher
                .'.*'             // junk
                .'/', $citation, $hits);       
        
        // If everything's OK, we get size of 6
        if (count($hits)==6) {
            $hits = citation::_trim ($hits);
            
            $back = array (
                'author' => $hits[1],
                'year' => $hits[2],
                'title' => $hits[3],
                'city' => $hits[4],
                'publisher' => $hits[5]);
            
            if (strpos ($hits[3], '(')!==false) {
                
                $tmp = explode ('(', $hits[3]);
            
                $back['title'] = trim($tmp[0]);
                $back['series'] = trim($tmp[1], " \(\)\n\r\t\{\}\,\.");
            }            
        }

        return $back;
    }
    
    private static function _trim ($arr) {
        $back = array();
        
        foreach ($arr as $offset=>$line) {
            $line =  trim($line, " \n\r\t\{\}\,\."); 
            $line =  rtrim($line, "("); 
            $line =  rtrim($line, " "); 
            $line =  rtrim($line, ","); 
            $line =  ltrim($line, ")"); 
            $line =  ltrim($line, " "); 
            $back[$offset] =  ltrim($line, ","); 
        }
        return $back;
    }
    
    private static function _journal2arr ($citation) {
        
        /*
        // Authors (year). title. journal, i(v), pages.
         * 
         */
        if (substr ($citation, -1)!='.') {
            $citation .= '.';
        }
        
        preg_match ('/(.*?[^(]+)'   // authors  - before (
                    . '\('          //  (
                .'([0-9]?[^)]+)'    // year - digits before )
                . '\)\.'            // ).
                .'(.*[^.]+|.*[^?]+)'  // title in form 'title.' or 'title?'
                .'[\.?]'
                .'(.*?[^0-9]+)'    // journal (some have commas, so until ( or digit (pages, volume,...)
                .'(.*?[^,]+)'      // vol/issue (match until comma)
                .'\,'
                .'(.*?[^\.\,]+)'    // pages (match against , or .)
                .'(\.|,)'          // check for size  
                .'(.*)'             // check for size
                .'/', $citation, $hits);

        if (count($hits) == 9) {
            
            $hits = citation::_trim($hits);

            return  array (
                        'author' => $hits[1],
                        'year' => $hits[2],
                        'title' => $hits[3],
                        'journal' => $hits[4],
                        'issue' => $hits[5],
                        'pages' => $hits[6]
                    ); 
        }
        
        else  {
            
            // plan B:
            // retrieve authors, year, title, journal and leave the rest.
            preg_match ('/(.*?[^(]+)'   // authors  - before (
                        . '\('          //  (
                    .'([0-9]?[^)]+)'    // year - digits before )
                    . '\)\.'            // ).
                    .'(.*?[^.]+|.*?[^?]+)'  // title in form 'title.' or 'title?'
                    .'[\.?]'
                    .'(.*?[^.]+|.*?[^,]+)'  // journal in form 'title.' or 'title?'
                    .'[\.,]'
                    .'(.*)'             // check for size
                    .'/', $citation, $hits);

            if (count($hits) == 6 && trim($hits[4]!="")) {

                $hits = citation::_trim($hits);

                return  array (
                            'author' => $hits[1],
                            'year' => $hits[2],
                            'title' => $hits[3],
                            'journal' => $hits[4]
                        ); 
            }
            else {            
                echo "\n***Journal article not recognised***\n";
                print_r ($hits);

                return -1;
            }
        }
    }
    
    
    private static function _workpaper2arr ($citation) {
        
        /*
         *  This is problematic, as there are many different forms (citation styles).
         *  We're taking Authors (Year). Title. Series. publisher
         */
        
        
        // let's just take out fields 1-3 (authors, year, title (whatever it is)
        preg_match ('/(.*?[^(]+)'   // authors  - before (
                    . '\('          //  (
                .'([0-9]?[^)]+)'    // year - digits before )
                . '\)\.'            // ).
                .'(.*[^.]+|.*[^?]+)'  // title in form 'title.' or 'title?'
                .'[\.?]'
                .'(.*?[^\.]+)'      // series
                .'\.'
/*                .'(.*?[^\.]+)'    // publisher
                .'[\.]'          // check for size */
                .'(.*)'             // check for size
                .'/', $citation, $hits);

            $hits = citation::_trim($hits);
            
            // field 3 can be broken. Manual fix needed.
            return  array (
                    'author' => $hits[1],
                    'year' => $hits[2],
                    'title' => $hits[3]
            );
    }
    
    
    
    private static function _conference2arr ($citation) {
        
        /*
         *  This is problematic, as there are many different forms (citation styles).
         *  We're taking Authors (Year). Title.  - same as working paper
         */
        
        
        
        // let's just take out fields 1-3 (authors, year, title (whatever it is)
        preg_match ('/(.*?[^(]+)'   // authors  - before (
                    . '\('          //  (
                .'([0-9]?[^)]+)'    // year - digits before )
                . '\)\.'            // ).
                .'(.*[^.]+|.*[^?]+)'  // title in form 'title.' or 'title?'
                .'[\.?]'
                .'(.*) for the'      // 
                .'(.*?[^,]+)'
                .','
                .'(.*)'             // check for size
                .'/', $citation, $hits);

            $hits = citation::_trim($hits);
            
            // field 3 can be broken. Manual fix needed.
            return  array (
                    'author' => $hits[1],
                    'year' => $hits[2],
                    'title' => $hits[3],
                    'conference' => $hits[5],
                    'city' => $hits[6]
            );
    }    
    
    
    
    private static function _report2arr ($citation) {
        // this is a mess.
        // it is most often AUTHORS (YEAR).
        // but what's next is... random.
        // title. title. publisher.
        // or
        // title. series. publisher.
        // or
        // title. title. series. publisher.
        // or title in series ....
        // let's just take authors, year, [*], publisher
        
        $back = -1;
        /*
         * Authors (YEAR). TITLE. REPORT_TYPE. Publisher.
         */
         preg_match ('/(.*?[^(]+)'   // authors  - before (
                    . '\('          //  (
                .'([0-9]?[^)]+)'    // year - digits before )
                . '\)\.'            // ).
                .'(.*[^.])'  // title in form 'title.' or 'title?'
                .'\.' 
                .'(?:(.*?[^\.]+)\.(.*?[^\.]+))|(.*?[^\.]+)'  // title in form 'title.' or 'title?' OR 'title (series)'. Two dots are a problem, sugest changing it to : .
                .'\.'
                .'(.*)'             // rest
                .'/', $citation, $hits);       

         
        // If everything's OK, we get size of 6
        if (count($hits)==6) {
            $hits = citation::_trim ($hits);
            
            $back = array (
                'author' => $hits[1],
                'year' => $hits[2],
                'title' => $hits[3],
                'publisher' => $hits[4]);
            
        }

        return $back;
    }    

    private static function _thesis2arr ($citation) {
       
        
        $back = -1;
        /*
         * 
         * Authors (YEAR). TITLE. TYPE. PUBLISHER.
         */
         preg_match ('/(.*?[^(]+)'   // authors  - before (
                    . '\('          //  (
                .'([0-9]?[^)]+)'    // year - digits before )
                . '\).'            // ).
                .'(.*?[^.]+|.*?[^?]+)'  // title in form 'title.' or 'title?'
                .'[\.?]'
                .'(.*?[^.]+|.*?[^,]+)'  // type
                .'[\.\,]'
                .'(.*?[^.]+)'  // location
                .'\.'
                .'(.*)'             // rest
                .'/', $citation, $hits);       
         
        // If everything's OK, we get size of 6
        if (count($hits)==7) {
            $hits = citation::_trim ($hits);
            
            $back = array (
                'author' => $hits[1],
                'year' => $hits[2],
                'title' => $hits[3],
                'thesis_type' => $hits[4],
                'publisher' => $hits[5]);
            
        }

        return $back;
    }    
    

    private static function _chapter2arr ($citation) {
       
        
        $back = -1;
        /*


         * Authors (YEAR). TITLE. WHERE (pp) publisher) 
         */
         preg_match ('/(.*?[^(]+)'   // authors  - before (
                    . '\('          //  (
                .'([0-9]?[^)]+)'    // year - digits before )
                . '\)\.'            // ).
                .'(.*[^.?]+)\bin '  // title in form 'title. IN' or 'title? IN'
//                .'([\.?] in )'
                .'(.*?[^)]+)'    // book author before )
                . '\)'            // ).
//                .'(.*)\(pp'  // book
                .'(.*)\bpp'  // book
                .'(.*?[^). ]+)'    // pps
                . '\b[.,]'            // ).
                .'(.*)'  // publisher
                .'/i', $citation, $hits);       

         // If everything's OK, we get size of 6
        if (count($hits)==8) {
            $hits = citation::_trim ($hits);
            
            $back = array (
                'author' => $hits[1],
                'year' => $hits[2],
                'title' => $hits[3],
                'book_author' => trim(preg_replace ('/(.*?[^(]+)(.*)/', '$1',$hits[4])),
                'booktitle' => trim ($hits[5], " \n\r\t\v\0("),
                'pages' => $hits[6],
                'location' => $hits[7],
                'publisher' => $hits[8]);
            
        }
        else {
            print_r ($hits);
        }

        return $back;
    }        
}