article.inc.php

   1 <?php
   2 global $monthname;
   3 $monthname = ['?',
   4         'januari', 'februari', 'maart', 'april', 'mei', 'juni',
   5         'juli', 'augustus', 'september', 'oktober', 'november', 'december',
   6 ];
   7
   8 function showdate($parts)
   9 {
  10         global $monthname;
  11         return implode(' ', array_filter([
  12                 intval(@$parts[2]), $parts[1] > 0 ? $monthname[intval($parts[1])] : '', $parts[0],
  13                 count($parts) > 5 ? "$parts[3]:$parts[4]" : '',
  14         ]));
  15 }
  16
  17 class ArchiveArticle
  18 {
  19         public $raw, $preface, $title, $body;
  20         public $meta = [];
  21
  22         function __construct($path)
  23         {
  24                 $this->page = $path;
  25                 $this->link = preg_replace('{(?:/index)?\.html$}', '', $path);
  26                 if (file_exists($this->page)) {
  27                         $this->raw = file_get_contents($this->page);
  28
  29                         if (preg_match_all('{
  30                                 \G <meta \s+ property="( [^"]+ )" \s+ content="( [^"]* )" > \s*
  31                         }x', $this->raw, $meta)) {
  32                                 $matchlen = array_sum(array_map('strlen', $meta[0]));
  33                                 $this->raw = substr($this->raw, $matchlen); # delete matched contents
  34                                 $this->meta = array_combine($meta[1], $meta[2]); # [property => content]
  35                         }
  36
  37                         @list ($this->preface, $this->title, $this->body) =
  38                                 preg_split('{<h2>(.*?)</h2>\s*}', $this->raw, 2, PREG_SPLIT_DELIM_CAPTURE);
  39                 }
  40         }
  41
  42         function __get($col)
  43         {
  44                 return $this->$col = $this->$col();  # run method and cache
  45         }
  46
  47         function safetitle()
  48         {
  49                 return trim($this->meta['og:title'] ?? strip_tags($this->title));
  50         }
  51         function name()
  52         {
  53                 return $this->safetitle ?: $this->link;
  54         }
  55
  56         function last()
  57         {
  58                 return filemtime($this->page);
  59         }
  60         function lastiso()
  61         {
  62                 return date(DATE_ATOM, $this->last);
  63         }
  64
  65         function dateparts()
  66         {
  67                 preg_match('< / (\d{4}) [/-] (\d{2}) (?:- (\d{2}) )? - >x', $this->page, $ymd);
  68                 array_shift($ymd);
  69                 return $ymd;
  70         }
  71         function dateiso()
  72         {
  73                 return implode('-', $this->dateparts()) . 'T12:00:00+02:00';
  74         }
  75         function date()
  76         {
  77                 return showdate($this->dateparts);
  78         }
  79
  80         function story()
  81         {
  82                 if ( preg_match('{
  83                         \n (?: < (?: p | figure [^>]* ) >\s* )+ (<img\ [^>]*>) | \n <hr\ />
  84                 }x', $this->body, $img, PREG_OFFSET_CAPTURE) ) {
  85                         # strip part after matching divider (image)
  86                         if (isset($img[1])) {
  87                                 $this->img = $img[1][0];
  88                         }
  89                         return substr($this->body, 0, $img[0][1]);
  90                 }
  91                 return $this->body;
  92         }
  93
  94         function teaser()
  95         {
  96                 if ($override = @$this->meta['og:description']) {
  97                         # prefer specific page description if found in metadata
  98                         return $override;
  99                 }
 100
 101                 if (preg_match('{
 102                         </h2> (?: \s+ | <p\sclass="nav\b.*?</p> | <div[^>]*> )* <p> \s* (.*?) </p>
 103                 }sx', $this->raw, $bodyp, PREG_OFFSET_CAPTURE)) {
 104                         # fallback paragraph contents following the page header
 105                         if ($bodyp[1][1] < 256) {
 106                                 return $bodyp[1][0];
 107                         }
 108                 }
 109
 110                 # starting paragraph for documents without title (assumed simple/partial)
 111                 if (strpos($this->raw, '<h2') === FALSE and preg_match('{
 112                         \A <p> \s* (.*?) </p>
 113                 }sx', $this->raw, $bodyp)) {
 114                         return $bodyp[1];
 115                 }
 116         }
 117
 118         function img()
 119         {
 120                 $this->img = NULL;
 121                 $this->story;
 122                 return $this->img;
 123         }
 124         function image()
 125         {
 126                 if ($override = @$this->meta['og:image']) {
 127                         # prefer specific page image if found in metadata
 128                         return $override;
 129                 }
 130
 131                 if ( preg_match('/\bsrc="([^"]*)"/', $this->img, $src) ) {
 132                         return $src[1];
 133                 }
 134         }
 135         function thumb($size = '300x')
 136         {
 137                 if (!$this->image or $this->image[0] !== '/') return;
 138                 return preg_replace(
 139                         ['{^(?:/thumb/[^/]*)?}', '/\.groot(?=\.\w+$)/'], ["thumb/$size", ''],
 140                         $this->image
 141                 );
 142         }
 143 }