From 0d0a3666573461f0946b161dca1126a8e328449d Mon Sep 17 00:00:00 2001
From: Dominique Fournier <dominique@fournier38.fr>
Date: Tue, 9 Sep 2014 11:43:04 +0000
Subject: [PATCH] markdown : support of imbricated ul/ol entries with
 indentation

git-svn-id: https://svn.fournier38.fr/svn/ProgSVN/trunk@1798 bf3deb0d-5f1a-0410-827f-c0cc1f45334c
---
 markdown.php | 530 +++++++++++++++++++++------------------------------
 1 file changed, 214 insertions(+), 316 deletions(-)

diff --git a/markdown.php b/markdown.php
index 4d98117..2d53dd2 100644
--- a/markdown.php
+++ b/markdown.php
@@ -1,25 +1,17 @@
 <?php
-/** DomFramework
-    @package domframework
-    @author Dominique Fournier <dominique@fournier38.fr> */
-
-error_reporting (E_ALL);
-/** Markdown management */
 class markdown
 {
-
-  /** debug variable */
+  /** To debug the markdown analyzer, activate the option */
   public $debug = false;
 
-  /** Convert the markdown language to HTML
-      Return the HTML string
-      @param string $mark Message in markdown syntax to display */
-  public function html ($mark)
+  private $blockid = array ("<h1>","<h2>","<h3>","<h4>","<h5>","<h6>",
+                            "<hr/>");
+
+  /** Convert the markdown text to html */
+  public function html ($markdown)
   {
-    if ($this->debug && file_exists ("/tmp/debugMD"))
-      unlink ("/tmp/debugMD");
-    $res = "";
-    $mark = htmlentities ($mark, ENT_QUOTES);
+    $markdown = rtrim ($markdown);
+    $markdown = htmlentities ($markdown);
 
     // Here are the regexp on multilines
     $search = array ();
@@ -27,44 +19,34 @@ class markdown
     // Titles with underline (SeText)
     // Titre1
     // ======
-    $search[] = "/^(.+)\\n==+$\\n/Um";
-    $replace[] = "\n<h1>\\1</h1>\n";
+    $search[] = "/(.+)\\n==+$/Um";
+    $replace[] = "</p><h1>\\1</h1>\n<p>";
     // Titre2
     // ------
-    $search[] = "/^(.+)\\n--+$\\n/Um";
-    $replace[] = "\n<h2>\\1</h2>\n";
+    $search[] = "/(.+)\\n--+$/Um";
+    $replace[] = "</p><h2>\\1</h2>\n<p>";
 
     // SEPARATORS : *** --- ___ * * * - - - _ _ _
     // Must be placed before EMPHASIS
     $search[] = "/^[*_-] ?[*_-] ?[*_-]$/Um";
-    $replace[] = "\n<hr/>\n";
+    $replace[] = "</p><hr/>\n<p>";
 
-    // End of line with continuous on second line, without double spaces : add
-    // space
-    //$search[] = "/(\w+)\\n(\w+)/Um"; $replace[] = "\\1 \\2";
-    $mark = preg_replace ($search, $replace, $mark);
+    $markdown = preg_replace ($search, $replace, $markdown);
 
-    $res = $this->paragraph ($mark);
-
-    $res = str_replace ("\n<p></p>", "", $res);
-    $res = str_replace ("<p></p>", "", $res);
-    $res = str_replace ("<p> </p>", "", $res);
-    $res = str_replace ("<p>\n</p>\n", "", $res);
-    return $res;
+    $textArray = explode ("\n", $markdown);
+    $pos = 0;
+    $html = $this->detectBlock ($textArray, 0, $pos);
+    $html = str_replace ("<p></p>", "", $html);
+    $html = str_replace ("<p> </p>", "", $html);
+    $html = trim ($html);
+    return $html;
   }
 
-  /** Translate the Markdown paragraph in HTML
-      return the html */
-  private function paragraph ($mark)
+  /** Search and replace in the paragraph on one line */
+  private function searchReplace ($line)
   {
-    // Think thereis already htmlentities passed on $mark !!!
-    $timeStart = microtime (TRUE);
-    $timeregex = 0;
-    // Initialization of convertions
     $search = array ();
     $replace = array ();
-
-
     // Titles short
     // == TITRE1
     $search[] = "/^==+ (.+)( ==+)?$/Um";
@@ -122,294 +104,212 @@ class markdown
     // # Title1
     $search[] = "/^# (.+)( +#+)?$/Um";
     $replace[] = "</p><h1>\\1</h1><p>";
-    // End of line with double space : <br/>
-    $search[] = "/(  )$/Um";               $replace[] = "<br/>";
+    return preg_replace ($search, $replace, $line);
+  }
 
-    // Cleanning the markdown text
-    $mark = str_replace ("\t", "    ", $mark);
-    if (trim ($mark) === "")
-      return "";
-
-    $spacer = "    ";
-    $res = "";
-    // P, OL, UL (but not LI !)
-    // Use to found the changing of types
-    $typeStack = array ();
-    // Number of spaces
-    $indentStack = array (-1);
-    // All the HTML stack (with LI)
-    $htmlStack = array ();
-    $lines = explode ("\n", $mark);
-    $timeInit = microtime (TRUE) - $timeStart;
-    $blockLI = false;
-    foreach ($lines as $nb=>$line)
+  /** Return HTML code corresponding to the code block
+      @param array $text The Markdown text to translate split by \n
+      @param integer $depth The depth of current bloc (in number of space)
+      @param integer $pos The start line number of the bloc */
+  private function typeCode ($text, $depth, &$pos)
+  {
+    if ($this->debug) echo "CALL typeCode (\$text, $depth, $pos)\n";
+    $posStart = $pos;
+    $content = "";
+    // End of code block : end of markdown text / depth lighter than $depth
+    while (isset ($text[$pos]) &&
+           $this->depth($text[$pos]) >= $depth)
     {
-      $this->debugMKD ("DEBUT:$line");
-      if (substr (ltrim ($line), 0, 1) === "<")
+      // The Code blocks can't be imbricated
+      if ($pos > $posStart)
+        $content .= "\n";
+      $content .= substr ($text[$pos], $depth);
+      $pos++;
+    }
+    // Insert Geshi on $content
+    if ($this->debug) 
+      echo "RETURN typeCode : <pre><code>$content</code></pre>\n";
+    return "<pre><code>$content</code></pre>\n";
+  }
+
+  /** Return HTML code corresponding to the OL block
+      @param array $text The Markdown text to translate split by \n
+      @param integer $depth The depth of current bloc (in number of space)
+      @param integer $pos The start line number of the bloc */
+  private function typeOL ($text, $depth, &$pos)
+  {
+    if ($this->debug) echo "CALL typeOL (\$text, $depth, $pos)\n";
+    $content = $this->typeOLUL ($text, $depth, $pos, "ol");
+    if ($this->debug) echo "RETURN typeOL : $content\n";
+    return $content;
+  }
+
+  /** Return HTML code corresponding to the UL block
+      @param array $text The Markdown text to translate split by \n
+      @param integer $depth The depth of current bloc (in number of space)
+      @param integer $pos The start line number of the bloc */
+  private function typeUL ($text, $depth, &$pos)
+  {
+    if ($this->debug) echo "CALL typeUL (\$text, $depth, $pos)\n";
+    $content = $this->typeOLUL ($text, $depth, $pos, "ul");
+    if ($this->debug) echo "RETURN typeUL : $content\n";
+    return $content;
+  }
+
+  private function typeOLUL ($text, $depth, &$pos, $type)
+  {
+    if ($this->debug) echo "CALL typeOLUL (\$text, $depth, $pos, $type)\n";
+    $content = "";
+    // End of OL/UL block : end of markdown text / depth lighter than $depth /
+    //                   linetype changed
+    $blockStart = $pos;
+    $blockContent = "";
+    while (isset ($text[$pos]) &&
+           $this->depth($text[$pos]) >= $depth && 
+           $this->lineType ($text[$pos]) === $type)
+    {
+      if ($this->debug)
+        echo "Start while $pos\n";
+      if (1)
       {
-        $this->debugMKD ("HTML : Skipped");
-        $res .= $line;
-        continue;
+        $content .= str_repeat (" ", ($depth+2))."<li>";
+        $blockContent .= $text[$pos];
+        $pos++;
+        // Look at continuous lines
+        while (isset ($text[$pos]) &&
+               $this->lineType ($text[$pos]) !== "NONE" &&
+               $this->lineType ($text[$pos]) !== $type &&
+               $this->depth($text[$pos]) === $depth)
+        {
+          if ($this->debug)
+            echo "Continuous line : ".$pos."\n";
+          $blockContent .= " ".$text[$pos];
+          $pos++;
+          continue;
+        }
+        // Indent the li and remove the number and dot and space at start
+        if ($type === "ol")
+          preg_match ("/^( *)[0-9]+\. +(.*)/", $blockContent, $matches);
+        else
+          preg_match ("/^( *)[-+*] +(.*)/", $blockContent, $matches);
+        if (!isset ($matches[2]))
+          $lineTxt = $blockContent;
+        else
+          $lineTxt = $matches[2];
+        $lineTxt = $this->searchReplace ($lineTxt);
+        $content .= $lineTxt;
+        $blockStart = $pos;
+        $blockContent = "";
       }
-      $type = $this->paragraphType ($line);
-      $this->debugMKD ("DEBUT:  Type='$type'");
-      $matches = array ();
-      $indent = strspn ($line, " ");
-      if (end ($typeStack) === "code" && $indent >= end ($indentStack))
+      if  (isset ($text[$pos]) && $this->depth($text[$pos]) > $depth)
       {
-        $this->debugMKD ("DEBUT:  Force type=CODE");
-        $type = "code";
-        $lineTxt = $line;
-        $indent = end ($indentStack);
+        if ($this->debug)
+          echo "Detect Block\n";
+        $content .= "\n".
+                    $this->detectBlock ($text, $this->depth($text[$pos]), $pos).
+                    str_repeat (" ", ($depth+2))."</li>\n";
       }
       else
       {
-        switch ($type)
-        {
-        case "ol" :
-          preg_match ("/^( *)[0-9]+\. +(.*)/", $line, $matches);
-          if (!isset ($matches[2]))
-            $lineTxt = $line;
-          else
-            $lineTxt = $matches[2];
-          break ;
-        case "ul" :
-          preg_match ("/^( *)[-+*] +(.*)/", $line, $matches);
-          if (!isset ($matches[2]))
-            $lineTxt = $line;
-          else
-            $lineTxt = $matches[2];
-          break ;
-        default:
-          $lineTxt = $line;
-        }
+        $content .= "</li>\n";
       }
-
-      $this->debugMKD ("DEBUT:  Indent='$indent'");
-      $this->debugMKD ("DEBUT: indentStack=".print_r ($indentStack, TRUE));
-      $this->debugMKD ("DEBUT: typeStack=".print_r ($typeStack, TRUE));
-
-      // Spacing
-      if ($indent < end ($indentStack))
-      {
-        $this->debugMKD ("DEB1 :  Ending of block");
-        if (end ($htmlStack) === "li")
-        {
-          $this->debugMKD ("Pending <Li> : closing");
-          $this->debugMKD ("</li>");
-          $res .= "</li>\n";
-          array_pop ($htmlStack);
-        }
-        if (strlen ($res) && in_array ($type, array ("ul","ol","p")))
-        {
-          $this->debugMKD ("Add \\n because of new block and not at start");
-          $res .= "\n";
-        }
-
-
-        // Remove last \n to put closing tag at the end of line
-        $res = substr ($res, 0, -1);
-        $oldType = array_pop ($typeStack);
-        if ($oldType === "code")
-        {
-          $this->debugMKD ("</code></pre>");
-          $res .= "</code></pre>";
-          array_pop ($htmlStack);
-          array_pop ($indentStack);
-        }
-        else
-        {
-          $this->debugMKD (str_repeat (" ", end ($indentStack))."</$oldType>");
-          $res .= str_repeat (" ", end ($indentStack))."</$oldType>";
-        }
-        array_pop ($indentStack);
-        array_pop ($htmlStack);
-        if ($type === "ol" || $type === "ul")
-        {
-          $this->debugMKD ("DEB2 : Pending <Li> : closing");
-          $this->debugMKD ("</li>");
-          $res .= "</li>";
-          array_pop ($htmlStack);
-        }
-
-        if ($type === "")
-        {
-          $this->debugMKD ("DEB3 : End of block");
-          array_pop ($indentStack);
-          $res .= "\n";
-          continue;
-        }
-      }
-
-      if ($indent == end ($indentStack))
-      {
-        $this->debugMKD ("DEB1 : Continuous block $type/".end ($typeStack));
-        if (end ($htmlStack) === "li" && $type !== "p")
-        {
-          $this->debugMKD ("Pending <Li> : closing");
-          $this->debugMKD ("</li>");
-          $res .= "</li>";
-          array_pop ($htmlStack);
-          $blockLI = false;
-        }
-        elseif (end ($htmlStack) === "li" && $type === "p")
-        {
-          $this->debugMKD ("DEB2 : Continuous LI block");
-          // Suppress \n on last line
-//          $res = substr ($res, 0, -1);
-          $type = prev ($htmlStack);
-          $blockLI = true;
-          $lineTxt = " ".$lineTxt;
-        }
-        
-        if (end ($htmlStack) === "p" && $type === "p" && 
-            substr ($res, -5) !== "<br/>")
-        {
-          // Add a space to agregate two lines
-          $this->debugMKD ("DEB2 : Adding aggragate space");
-          $res .= " ";
-        }
-
-        if ($type !== end ($typeStack))
-        {
-          $this->debugMKD ("DEB2 : Continuous Block but type change");
-          if (end ($typeStack) !== FALSE)
-          {
-            // Remove last \n to put closing tag at the end of line
-            //$res = substr ($res, 0, -1);
-            $oldType = array_pop ($typeStack);
-            $this->debugMKD (str_repeat (" ", end ($indentStack)).
-                             "</$oldType>");
-            $res .= str_repeat (" ", end ($indentStack))."</$oldType>";
-            array_pop ($indentStack);
-            array_pop ($htmlStack);
-          }
-
-          if ($type !== "")
-          {
-            $this->debugMKD (str_repeat (" ", end ($indentStack))."<$type>");
-            $res .= str_repeat (" ", end ($indentStack))."<$type>";
-            $htmlStack[] = $type;
-            array_push ($indentStack, $indent);
-            array_push ($typeStack, $type);
-          }
-        }
-
-        if (($type === "ol" || $type === "ul") && $blockLI === false)
-        {
-          $this->debugMKD ("DEB2 : Adding li");
-          $htmlStack[] = "li";
-          $this->debugMKD (str_repeat (" ", $indent)."<li>");
-          $res .= "\n".str_repeat (" ", $indent)."<li>";
-        }
-
-      }
-
-      if ($indent > end ($indentStack) && end ($typeStack) !== "code")
-      {
-        // The code indentation should not be parsed as a new code : the
-        // <code></code> continue until the end of paragraph
-        $this->debugMKD ("DEB1 :  Starting a new block");
-        if ($type === "")
-        {
-          $this->debugMKD ("No type : skipped");
-          continue;
-        }
-
-        if (strlen ($res) && in_array ($type, array ("ul","ol","p")))
-        {
-          $this->debugMKD ("Add \\n because of new block");
-          $res .= "\n";
-        }
-
-        if (end ($indentStack))
-          array_pop ($indentStack);
-        if ($type === "code")
-        {
-          // Code need a pre before code
-          if (end ($typeStack))
-          {
-            $this->debugMKD ("DEB2 : CODE : Close older HTML");
-            $oldType = array_pop ($typeStack);
-            $this->debugMKD (str_repeat (" ", end ($indentStack)).
-                             "</$oldType>");
-            $res .= str_repeat (" ", end ($indentStack))."</$oldType>";
-            array_pop ($indentStack);
-            array_pop ($htmlStack);
-          }
-          $typetmp = "pre><code";
-          $htmlStack[] = "pre";
-          $htmlStack[] = "code";
-          array_push ($typeStack, "code");
-        }
-        else
-        {
-          $typetmp = $type;
-          $htmlStack[] = $type;
-          array_push ($typeStack, $type);
-        }
-        array_push ($indentStack, $indent);
-        $this->debugMKD (str_repeat (" ", $indent)."<$typetmp>");
-        $res .= str_repeat (" ", $indent)."<$typetmp>";
-        if ($type === "ol" || $type === "ul")
-        {
-          $this->debugMKD ("DEB2 : Adding li");
-          $htmlStack[] = "li";
-          $this->debugMKD (str_repeat (" ", $indent)."<li>");
-          $res .= "\n".str_repeat (" ", $indent)."<li>";
-        }
-      }
-
-      if ($type === "" &&  end ($indentStack))
-      {
-        $this->debugMKD ("DEB2 : Empty type");
-        // Remove last \n to put closing tag at the end of line
-        $res = substr ($res, 0, -1);
-        $oldType = array_pop ($typeStack);
-        $this->debugMKD (str_repeat (" ", end ($indentStack))."</$oldType>");
-        $res .= "\n".str_repeat (" ", end ($indentStack))."</$oldType>";
-        array_pop ($htmlStack);
-      }
-
-      // If code, there is no emphasis, email, and other conversions
-      if ($type !== "code")
-      {
-        $timetmp = microtime (TRUE);
-        $lineTxt = preg_replace ($search, $replace, $lineTxt);
-        $timeregex += (microtime (TRUE) - $timetmp);
-      }
-
-      $this->debugMKD (substr ($lineTxt, end ($indentStack))."");
-      $res .= substr ($lineTxt, end ($indentStack))."";
-      if ($type === "code")
-        $res .= "\n";
     }
+    if ($this->debug) echo "RETURN typeOLUL : <$type>\n$content</$type>\n";
+    return "<$type>\n$content".str_repeat (" ", $depth)."</$type>\n";
 
-    $this->debugMKD ("DEB1 : End of loop");
-    $htmlStack = array_reverse ($htmlStack);
-    foreach ($htmlStack as $i=>$type)
+  }
+
+  /** Return HTML code corresponding to the NONE block
+      The NONE type exists only on empty strings. Just skip the current and
+      empty line, and return an empty string */
+  private function typeNONE ($text, $depth, &$pos)
+  {
+    if ($this->debug) echo "CALL typeNONE (\$text, $depth, $pos)\n";
+    $pos++;
+    return "";
+  }
+
+  /** Return HTML code corresponding to the P block
+      @param array $text The Markdown text to translate split by \n
+      @param integer $depth The depth of current bloc (in number of space)
+      @param integer $pos The start line number of the bloc */
+  private function typeP ($text, $depth, &$pos)
+  {
+    if ($this->debug) echo "CALL typeP (\$text, $depth, $pos)\n";
+    $content = "";
+    // End of P block : end of markdown text / depth lighter than $depth /
+    //                   linetype changed
+    $Pinc = $pos;
+    while (isset ($text[$pos]) &&
+           $this->depth($text[$pos]) == $depth && 
+           $this->lineType ($text[$pos]) === "p")
     {
-      if ($type === "code")
-        $res = substr ($res, 0, -1);
-      $this->debugMKD ("FIN</$type>");
-      $res .= "</$type>";
-      if (($i+1) < count ($htmlStack) && $type !== "code")
-        $res .= "\n";
+      if (substr ($text[$pos], -2) === "  ")
+      {
+        // Two spaces at end of line : add <br/>
+        $content .= $this->searchReplace (substr ($text[$pos], 0, -2)) ."<br/>";
+      }
+      elseif ($pos > $Pinc && substr ($content, -5) !== "<br/>")
+      {
+        // Add a space between two lines from the same block, if this is not
+        // the continuity of the block
+        $content .= " ".$this->searchReplace ($text[$pos]);
+      }
+      else
+      {
+        $content .= $this->searchReplace ($text[$pos]);
+      }
+      $pos++;
     }
+    if ($this->debug) echo "RETURN typeP : <p>$content</p>\n";
+    return "<p>$content</p>\n";
+  }
 
-    $this->debugMKD ("TimeInit=".($timeInit*1000)."ms");
-    $this->debugMKD ("TimeRegex=".($timeregex*1000)."ms");
-    $this->debugMKD ("TimeAll=".((microtime (TRUE) - $timeStart)*1000)."ms");
-    $this->debugMKD ("-----------\n");
-    return $res;
+  /** Detect the type of the text and call the appropriate function *
+      @param array $text The Markdown text to translate split by \n
+      @param integer $depth The depth of current bloc (in number of space)
+      @param integer $pos The start line number of the bloc
+      @return the HTML code */
+  private function detectBlock ($text, $depth, &$pos)
+  {
+    if ($this->debug) echo "CALL detectBlock (\$text, $depth, $pos)\n";
+    $content = "";
+    $blockContent = "";
+    // detect the type and call the right type function
+    while (isset ($text[$pos]))
+    {
+      if ($this->depth ($text[$pos]) > $depth)
+      {
+        if ($this->debug)
+          echo "CALL DEPTH > MINDEPTH (".$this->depth ($text[$pos]).
+               " > $depth)\n";
+        $content .= $this->detectBlock ($text, $this->depth ($text[$pos]),
+                                        $pos);
+        continue;
+      }
+      elseif ($this->depth ($text[$pos]) < $depth)
+      {
+        if ($this->debug)
+          echo "CALL DEPTH > MINDEPTH (".$this->depth ($text[$pos]).
+               " < $depth)\n";
+        return $content;
+      }
+      
+      $type = $this->lineType ($text[$pos]);
+      $func = "type$type";
+      if ($this->debug)
+        echo "FROM DETECT : CALL $func (line=".$text[$pos].")\n";
+      $content .= str_repeat (" ", $depth). $this->$func ($text, $depth, $pos);
+    }
+    return $content;
   }
 
   /** Return the Type of object in the provided line
       p, ul, ol, code */
-  private function paragraphType ($line)
+  private function lineType ($line)
   {
     if (! isset ($line{0}))
-      return "";
+      return "NONE";
     if (preg_match ("/^[ \t]*[+*-] /", $line) === 1)
       return "ul";
     if (preg_match ("/^[ \t]*[0-9]+\. /", $line) === 1)
@@ -419,13 +319,11 @@ class markdown
     return "p";
   }
 
-  /** Function to display the MarkDown debug */
-  private function debugMKD ($msg)
+  /** Return the depth of the provided line
+      @param $line Line to analyze
+      @return the depth of the line */
+  private function depth ($line)
   {
-    if ($this->debug === false)  return;
-    $trace = debug_backtrace();
-    $back = reset ($trace);
-    file_put_contents ("/tmp/debugMD", "[".$back["line"]."] $msg\n",
-                       FILE_APPEND);
+    return strspn ($line, " ");
   }
 }