<?php if (!defined('PmWiki')) exit();
/* dash-pagenames.php, an extension for PmWiki 2.3, copyright Hans Bracker 2023. 

   This program is free software; you can redistribute it and/or modify
   it under the terms of the GNU General Public License as published
   by the Free Software Foundation; either version 2 of the License, or
   (at your option) any later version.

   Page-Names and URLs are created with dashes as word separators:
   [[Some Group/ Some other Name]] -> Some-Group/Some-other-Name
   Spaces in links get converted to dashes/hyphens in the target name, instead of WikiWords.
   Links in groups PmWiki, Site and SiteAdmin, as well as  WikiWord links keep their functionality.
   Cross-linking pages between groups needs no special attention.   
   UTF-8 encoded characters are allowed, lower and upper case letters are preserved. 
*/
$RecipeInfo['Dash-Pagenames']['Version'] = '2023-02-17';

// UTF-8 encoded characters in pagenames are allowed and preserved.
// Set to 1 will restrict pagenames to ASCII characters. 
// Any accented UTF-8 characters will be substituted for ASCII characters.  
SDV( $ForcePageNamesToASCII, 0); 

// Upper and lower case letters in pagenames are allowed and preserved.
// Set to 1 will force all characters in pagenames to lower case
SDV( $ForcePageNamesToLowerCase, 0); 

// groups excempt from dashed-name-patterns to preserve their CamelCase and link functiomalities
SDVA( $CamelCaseGroups, array('PmWiki'=>1,'Site'=>1,'SiteAdmin'=>1) );


if ($Charset == 'UTF-8') {
  // UTF-8 charset is set by /scripts/xlpage-utf-8.php, which needs to be included before this script. 
  // It also sets new $GroupPattern and $NamePattern which allow lower case letters at start of group or name
  $PageNameChars = '-[:alnum:]\\x80-\\xfe';
} else {
// $GroupPattern and $NamePattern need setting to allow lower case letters at start  
  $PageNameChars = '-[:alnum:]'; 
  $GroupPattern = '[\\w]*(?:-\\w+)*';
  $NamePattern = '[\\w]*(?:-\\w+)*';
}

// Replacement patterns for special URL characters to UTF-8 or ASCII text substitutions (rather than just to strip them)
if ($Charset == 'UTF-8' &&  $ForcePageNamesToASCII==0) {
  SDVA( $MPN_ReplacePatterns, array(
    "/'/" => "\xe2\x80\x99", //replace Apostrophe '  with Right Single Quotation Mark ’ u2019
    "/\:/" => "\xc2\xb7",    //replace Colon ':' with mid-dot '·'
    "/\&/" => ' and ',        //replace '&' with 'and' , for English
    "/\@/" => ' at ',         //replace '@' with 'at' , for English
    "/\=/" => ' equals ',     //replace '=' with 'equals' , for English
    "/\+/" => ' plus ',       //replace '+' with 'plus', for English
  ));
} else { //ASCII
  SDVA( $MPN_ReplacePatterns, array(
    "/(l)\'(?=\w+)/i" => '$1e ',  //replace 'l'mot' with 'le mot' , for French
    "/\&/" => ' and ',        //replace '&' with 'and' , for English
    "/\@/" => ' at ',         //replace '@' with 'at' , for English
    "/\=/" => ' equals ',     //replace '=' with 'equals' , for English
    "/\+/" => ' plus ',       //replace '+' with 'plus', for English
  ));  
}

// PmWiki standard MakePageNamePatterns. Used for groups exempt from dashed-pagename-patterns
$Pm_PNP = array(
   '/[?#].*$/' => '',                # strip everything after ? or #
   "/'/" => '',                      # strip single-quotes
   "/[^$PageNameChars]+/" => ' ',    # convert everything else to space
   '/((^|[^-\\w])\\w)/' => 'cb_toupper', # CamelCase, first letter to upper
   '/ /' => ''
);
// alternative MakePageNamePatterns. Dash-hyphen as word separator. No CamelCase. 
// letter cases will not be changed, unless $LowerCasePageNames = 1;
$Dash_PNP = array(
    '/[?#].*$/' => '',                # strip everything after ? or #
    "/'/" => '',                      # strip single-quotes
    "/[^$PageNameChars]+/" => '-',    # convert everything else to hyphen
    '/ /' => '',                      # strip spaces
    '/^-/' => '',                     # strip any dashes from start
    '/-$/' => '',                     # strip any dashes from end
    '/-+/' => '-',                    # strip extra dashes
);

// setting $MakePageNamePatterns and $AsSpacedFunction according to group
$group = PageVar($pagename,'$Group');
if (array_key_exists($group, $CamelCaseGroups))
    $MakePageNamePatterns =  $Pm_PNP;
else { //for all groups except PmWiki, Site, SiteAdmin, and any added to $CamelCaseGroups
    $MakePageNamePatterns =  $Dash_PNP;
    $FmtPV['$Title'] = 'FmtPageTitle(@$page["title"], $name, 1)'; //same as $Titlespaced (dashes to spaces) 
    $AsSpacedFunction = 'DashSpaced';
}  

// replace dashes in pagenames with spaces and space WikiWords
function DashSpaced($x) { 
  global $Charset;
  if ($Charset == 'UTF-8') return AsSpacedUTF8(str_replace('-',' ',$x));
  else   return AsSpaced(str_replace('-',' ',$x)); 
}

// helper function to set alternatives MakePageName patterns  
function SetPNP($group) { 
    global $Pm_PNP, $Dash_PNP, $CamelCaseGroups, $ForcePageNamesToLowerCase;
    if (array_key_exists($group, $CamelCaseGroups)) {
         return $Pm_PNP;
    } else {
        if ($ForcePageNamesToLowerCase==1) {
            $Dash_PNP['/(.*$)/'] = 'cb_tolower';
        }
        return $Dash_PNP;
    }
}

 // replaces function MakePageName so we can switch $MakePageNamePatterns according to target group
 $MakePageNameFunction = 'MakePageNameAlt';
 function MakePageNameAlt($basepage, $str) { #echo $str;
    global $PagePathFmt, $MPN_ReplacePatterns, $ForcePageNamesToASCII;
    $in = $str; //used for debug
    $str = preg_replace('~[#?].*$~', '', $str);    // strip anything from # or ?
    if ($ForcePageNamesToASCII==1 && preg_match('~[\\x80-\\xff]~', $str))
       $str = Dash_UnaccentUTF8($str);
    $str = htmlspecialchars_decode($str);  
    $str = preg_replace("/&#?[a-z0-9]+;/i", "", $str); //strip decoded html entities
    $str = PPRA($MPN_ReplacePatterns, $str);      // special language friendly replacements
    $m = preg_split('/[.\\/]/', $str);           //split pn by / or . into group and name parts
    if (count($m)<1 || count($m)>2 || $m[0]=='') return '';
    ##  handle "Group.Name" conversions according to link target group
    if (@$m[1] > '') {
      $pat = SetPNP($m[0]);
      $group = PPRA($pat, $m[0]);
      $name =  PPRA($pat, $m[1]); 
      return "$group.$name";
    } else
    $bp = preg_split('/[.\\/]/', $basepage); 
    $name = PPRA(SetPNP($bp[0]), $m[0]);  
    $isgrouphome = count($m) > 1;
    foreach((array)$PagePathFmt as $pg) {
      if ($isgrouphome && strncmp($pg, '$1.', 3) !== 0) continue;
      $pn = FmtPageName(str_replace('$1', $name, $pg), $basepage);
      if (PageExists($pn)) return $pn;
    }
    if ($isgrouphome) {
      foreach((array)$PagePathFmt as $pg) 
        if (strncmp($pg, '$1.', 3) == 0)
          return FmtPageName(str_replace('$1', $name, $pg), $basepage);
      return "$name.$name";
    }
    return preg_replace('/[^\\/.]+$/', $name, $basepage);
 }

 // converting accented characters to ASCII. PHP Intl module needs to be enabled 
 $DashTransliterator = Transliterator::createFromRules(
    ':: Latin-ASCII ; :: NFD; :: [:Nonspacing Mark:] Remove; :: NFC;', 
    Transliterator::FORWARD);
  function Dash_UnaccentUTF8($x) { 
    global $DashTransliterator, $Charset;
    if ($Charset!='UTF-8')
        $x = mb_convert_encoding($x, "UTF-8", $Charset);
    $x = preg_replace("/ä|ö|ü|ø|Ä|Ö|Ü|Ø/", '$0e', $x);
    $x = preg_replace("/å|Å/", '$0o', $x);
    $x = str_replace("Æ", 'Ae', $x);
    $x = str_replace("Œ", 'Oe', $x);
    $x = str_replace("Þ", 'Th', $x);
    $x = preg_replace("/\xcc\x88/", 'e', $x);
    $x = $DashTransliterator->transliterate($x);
    return $x;
  }

//EOF