X-Git-Url: https://git.plutz.net/?p=cgilite;a=blobdiff_plain;f=markdown.awk;h=6271996440dab3b37132f8eccdecbf6094a4db99;hp=65e0aefe2bd3ed4da3f4d3a543cc12511c31b4f9;hb=9bb22561275a9ce0d5403a12fe7e8aeeeb2792e9;hpb=882f37daf3eac1ee8589eb623ec3bc4cfc3c7e9f

diff --git a/markdown.awk b/markdown.awk
index 65e0aef..6271996 100755
--- a/markdown.awk
+++ b/markdown.awk
@@ -4,10 +4,6 @@
 # EXPERIMENTAL Markdown processor with minimal dependencies.
 # Meant to support all features of John Grubers basic Markdown
 # + a number of common extensions, mostly inspired by Pandoc Markdown
-#
-# ToDo:
-# - HTML processing / escaping (according to environment flag)
-# - em-dashes and arrows
 
 # Supported Features / TODO:
 # ==========================
@@ -45,18 +41,19 @@
 # -  ?  Heading identifiers (php md, pandoc)
 # - [x] Automatic heading identifiers (custom)
 # - [x] Fenced code blocks (php md, pandoc)
-#   - [-] Fenced code attributes
+#   - [x] Fenced code attributes
 # - [ ] Tables
 #   -  ?  Simple table (pandoc)
 #   -  ?  Multiline table (pandoc)
 #   -  ?  Grid table (pandoc)
-#   -  ?  Pipe table (php md pandoc)
+#   - [x] Pipe table (php md pandoc)
 # - [x] Line blocks (pandoc)
-# - [x] Task lists (pandoc)
+# - [x] Task lists (pandoc, custom)
 # - [ ] Definition lists (php md, pandoc)
 # - [-] Numbered example lists (pandoc)
 # - [-] Metadata blocks (pandoc)
-# - [-] Fenced Divs (pandoc)
+# - [x] Metadata blocks (custom)
+# - [x] Fenced Divs (pandoc)
 #
 # Extensions - Inline elements:
 # ----------------------------
@@ -126,12 +123,18 @@ function inline( line, LOCAL, len, code, href, guard ) {
     href = HTML( substr( line, 2, len - 2) );
     return "<a href=\"" href "\">" href "</a>" inline( substr( line, len + 1) );
 
+  # quick link email
+  } else if ( match( line, /^<[a-zA-Z0-9.!#$%&'\''*+\/=?^_`{|}~-]+@[a-zA-Z0-9]([a-zA-Z0-9-]{0,61}[a-zA-Z0-9])?(\.[a-zA-Z0-9]([a-zA-Z0-9-]{0,61}[a-zA-Z0-9])?)*>/ ) ) {
+    len = RLENGTH;
+    href = HTML( substr( line, 2, len - 2) );
+    return "<a href=\"mailto:" href "\">" href "</a>" inline( substr( line, len + 1) );
+
   # inline links
-  } else if ( match(line, /^\[([^]]+)\]\(([^"\)]+)([ \t]+"([^"]+)")?\)/) ) {
+  } else if ( match(line, /^\[([^]]+)\]\(([^"\)]+)([[:space:]]+"([^"]+)")?\)/) ) {
     len = RLENGTH;
-    text  = gensub(/^\[([^]]+)\]\(([^"\)]+)([ \t]+"([^"]+)")?\)/, "\\1", "g", line);
-    href  = gensub(/^\[([^]]+)\]\(([^"\)]+)([ \t]+"([^"]+)")?\)/, "\\2", "g", line);
-    title = gensub(/^\[([^]]+)\]\(([^"\)]+)([ \t]+"([^"]+)")?\)/, "\\4", "g", line);
+    text  = gensub(/^\[([^]]+)\]\(([^"\)]+)([[:space:]]+"([^"]+)")?\)/, "\\1", 1, substr(line, 1, len) );
+    href  = gensub(/^\[([^]]+)\]\(([^"\)]+)([[:space:]]+"([^"]+)")?\)/, "\\2", 1, substr(line, 1, len) );
+    title = gensub(/^\[([^]]+)\]\(([^"\)]+)([[:space:]]+"([^"]+)")?\)/, "\\4", 1, substr(line, 1, len) );
     if ( title ) {
       return "<a href=\"" HTML(href) "\" title=\"" HTML(title) "\">" inline( text ) "</a>" inline( substr( line, len + 1) );
     } else {
@@ -141,23 +144,23 @@ function inline( line, LOCAL, len, code, href, guard ) {
   # reference style links
   } else if ( match(line, /^\[([^]]+)\] ?\[([^]]*)\]/ ) ) {
     len = RLENGTH;
-    text = gensub(/^\[([^\n]+)\] ?\[([^\n]*)\].*/, "\\1", 1, line);
-      id = gensub(/^\[([^\n]+)\] ?\[([^\n]*)\].*/, "\\2", 1, line);
+    text = gensub(/^\[([^\n]+)\] ?\[([^\n]*)\].*/, "\\1", 1, substr(line, 1, len) );
+      id = gensub(/^\[([^\n]+)\] ?\[([^\n]*)\].*/, "\\2", 1, substr(line, 1, len) );
     if ( ! id ) id = text;
     if ( rl_href[id] && rl_title[id] ) {
-      return "<a href=\"" rl_href[id] "\" title=\"" rl_title[id] "\">" inline(text) "</a>" inline( substr( line, len + 1) );
+      return "<a href=\"" HTML(rl_href[id]) "\" title=\"" HTML(rl_title[id]) "\">" inline(text) "</a>" inline( substr( line, len + 1) );
     } else if ( rl_href[id] ) {
-      return "<a href=\"" rl_href[id] "\">" inline(text) "</a>" inline( substr( line, len + 1) );
+      return "<a href=\"" HTML(rl_href[id]) "\">" inline(text) "</a>" inline( substr( line, len + 1) );
     } else {
-      return "" substr(line, 1, len) inline( substr(line, len + 1) );
+      return "" HTML(substr(line, 1, len)) inline( substr(line, len + 1) );
     }
 
   # inline images
   } else if ( match(line, /^!\[([^]]+)\]\(([^"\)]+)([ \t]+"([^"]+)")?\)/) ) {
     len = RLENGTH;
-    text  = gensub(/^!\[([^]]+)\]\(([^"\)]+)([ \t]+"([^"]+)")?\)/, "\\1", "g", line);
-    href  = gensub(/^!\[([^]]+)\]\(([^"\)]+)([ \t]+"([^"]+)")?\)/, "\\2", "g", line);
-    title = gensub(/^!\[([^]]+)\]\(([^"\)]+)([ \t]+"([^"]+)")?\)/, "\\4", "g", line);
+    text  = gensub(/^!\[([^]]+)\]\(([^"\)]+)([ \t]+"([^"]+)")?\)/, "\\1", "g", substr(line, 1, len) );
+    href  = gensub(/^!\[([^]]+)\]\(([^"\)]+)([ \t]+"([^"]+)")?\)/, "\\2", "g", substr(line, 1, len) );
+    title = gensub(/^!\[([^]]+)\]\(([^"\)]+)([ \t]+"([^"]+)")?\)/, "\\4", "g", substr(line, 1, len) );
     if ( title ) {
       return "<img src=\"" HTML(href) "\" alt=\"" HTML(text) "\" title=\"" HTML(title) "\" />" inline( substr( line, len + 1) );
     } else {
@@ -167,15 +170,15 @@ function inline( line, LOCAL, len, code, href, guard ) {
   # reference style images
   } else if ( match(line, /^!\[([^]]+)\] ?\[([^]]*)\]/ ) ) {
     len = RLENGTH;
-    text = gensub(/^!\[([^\n]+)\] ?\[([^\n]*)\].*/, "\\1", 1, line);
-      id = gensub(/^!\[([^\n]+)\] ?\[([^\n]*)\].*/, "\\2", 1, line);
+    text = gensub(/^!\[([^\n]+)\] ?\[([^\n]*)\].*/, "\\1", 1, substr(line, 1, len) );
+      id = gensub(/^!\[([^\n]+)\] ?\[([^\n]*)\].*/, "\\2", 1, substr(line, 1, len) );
     if ( ! id ) id = text;
     if ( rl_href[id] && rl_title[id] ) {
-      return "<img src=\"" rl_href[id] "\" alt=\"" HTML(text) "\" title=\"" rl_title[id] "\" />" inline( substr( line, len + 1) );
+      return "<img src=\"" HTML(rl_href[id]) "\" alt=\"" HTML(text) "\" title=\"" HTML(rl_title[id]) "\" />" inline( substr( line, len + 1) );
     } else if ( rl_href[id] ) {
-      return "<img src=\"" rl_href[id] "\" alt=\"" HTML(text) "\" />" inline( substr( line, len + 1) );
+      return "<img src=\"" HTML(rl_href[id]) "\" alt=\"" HTML(text) "\" />" inline( substr( line, len + 1) );
     } else {
-      return "" substr(line, 1, len) inline( substr(line, len + 1) );
+      return "" HTML(substr(line, 1, len)) inline( substr(line, len + 1) );
     }
 
   #  ~~strikeout~~ (pandoc)
@@ -195,7 +198,7 @@ function inline( line, LOCAL, len, code, href, guard ) {
 
   # ignore embedded underscores (pandoc, php md)
   } else if ( match(line, "^[[:alnum:]](__|_)") ) {
-    return substr( line, 1, RLENGTH) inline( substr(line, RLENGTH + 1) );
+    return HTML(substr( line, 1, RLENGTH)) inline( substr(line, RLENGTH + 1) );
 
   #  __strong__$
   } else if ( match(line, "^__(([^_[:space:]]|" ieu ")|([^_[:space:]]|" ieu ")(" nu "|" ieu ")*([^_[:space:]]|" ieu "))__$") ) {
@@ -252,7 +255,7 @@ function inline( line, LOCAL, len, code, href, guard ) {
   }
 }
 
-function _block( block, LOCAL, st, len, hlvl, htxt, guard, code, indent ) {
+function _block( block, LOCAL, st, len, hlvl, htxt, guard, code, indent, attrib ) {
   gsub( /^\n+|\n+$/, "", block );
 
   if ( block == "" ) {
@@ -279,6 +282,12 @@ function _block( block, LOCAL, st, len, hlvl, htxt, guard, code, indent ) {
   } else if ( AllowHTML && match( block, /^ ? ? ?(<\/[A-Za-z][A-Za-z0-9-]*[[:space:]]*>|<[A-Za-z][A-Za-z0-9-]*([[:space:]]+[A-Za-z_:][A-Za-z0-9_\.:-]*([[:space:]]*=[[:space:]]*([[:space:]"'=<>`]+|"[^"]*"|'[^']*'))?)*[[:space:]]*\/?>)([[:space:]]*\n)([^\n]|\n[ \t]*[^\n])*(\n[[:space:]]*\n|$)/) ) {
     len = RLENGTH; st = RSTART;
     return substr(block, st, len) _block(substr(block, st + len));
+
+  # Metadata (custom, block starting with %something)
+  # Metadata is ignored but can be interpreted externally
+  } else if ( match(block, /^%[a-zA-Z]+([[:space:]][^\n]*)?(\n|$)(%[a-zA-Z]+([[:space:]][^\n]*)?(\n|$)|%([[:space:]][^\n]*)?(\n|$)|[ \t]+[^\n[:space:]][^\n]*(\n|$))*/) ) {
+    len = RLENGTH; st = RSTART;
+    return  _block( substr( block, len + 1) );
  
   # Blockquote (leading >)
   } else if ( match( block, /^> /) ) {
@@ -287,6 +296,53 @@ function _block( block, LOCAL, st, len, hlvl, htxt, guard, code, indent ) {
     return "<blockquote>\n" _block( gensub( /(^|\n)> /, "\n", "g", substr(block, 1, st - 1) ) ) "</blockquote>\n\n" \
            _block( substr(block, st + len) );
 
+  # Pipe Tables (pandoc / php md / gfm )
+  } else if ( match(block, "^((\\|)?([^\n]+\\|)+[^\n]+(\\|)?)\n" \
+                           "((\\|)?:?(-+:?[\\|+])+:?-+:?(\\|)?)\n" \
+                           "((\\|)?([^\n]+\\|)+[^\n]+(\\|)?(\n|$))+" ) ) {
+    len = RLENGTH; st = RSTART;
+    #initialize empty arrays
+    split("", talign); split("", tarray);
+    cols = 0; ttext = ""; cnt=0;
+
+    # table header and alignment
+    split( gensub( /(^\||\|$)/, "", "g", \
+           gensub( /(^|[^\\])\\\|/, "\\1\\&#x7C;", "g", \
+           substr(block, 1, match(block, /(\n|$)/)) \
+    )), tarray, /\|/);
+    block = substr(block, match(block, /(\n|$)/) + 1 );
+    cols = split( \
+           gensub( /(^\||\|$)/, "", "g", \
+           substr(block, 1, match(block, /(\n|$)/)) \
+    ), talign, /[+\|]/);
+    block = substr(block, match(block, /(\n|$)/) + 1 );
+
+    for( cnt = 1; cnt < cols; cnt++ ) {
+           if (match(talign[cnt], /:-+:/)) talign[cnt]="center";
+      else if (match(talign[cnt],  /-+:/)) talign[cnt]="right";
+      else if (match(talign[cnt],  /:-+/)) talign[cnt]="left";
+      else talign[cnt]="";
+    }
+
+    ttext = "<thead>\n<tr>"
+    for (cnt = 1; cnt < cols; cnt++)
+      ttext = ttext "<th align=\"" talign[cnt] "\">" inline(tarray[cnt]) "</th>"
+    ttext = ttext "</tr>\n</thead><tbody>\n"
+
+    while ( match(block, "^((\\|)?([^\n]+\\|)+[^\n]+(\\|)?(\n|$))+" ) ){
+      split( gensub( /(^\||\|$)/, "", "g", \
+             gensub( /(^|[^\\])\\\|/, "\\1\\&#x7C;", "g", \
+             substr(block, 1, match(block, /(\n|$)/)) \
+      )), tarray, /\|/);
+      block = substr(block, match(block, /(\n|$)/) + 1 );
+
+      ttext = ttext "<tr>"
+      for (cnt = 1; cnt < cols; cnt++)
+        ttext = ttext "<td align=\"" talign[cnt] "\">" inline(tarray[cnt]) "</td>"
+      ttext = ttext "</tr>\n"
+    }
+    return "<table>" ttext "</tbody></table>\n" _block( substr(block, len + 1) );
+
   # Line Blocks (pandoc)
   } else if ( match(block, /^\| [^\n]*(\n|$)(\| [^\n]*(\n|$)|[ \t]+[^\n[:space:]][^\n]*(\n|$))*/) ) {
     len = RLENGTH; st = RSTART;
@@ -306,13 +362,34 @@ function _block( block, LOCAL, st, len, hlvl, htxt, guard, code, indent ) {
     return "<pre><code>" HTML( code ) "</code></pre>\n" \
            _block( substr( block, len + 1 ) );
 
+  # Fenced Divs (pandoc, custom)
+  } else if ( match( block, /^(:::+)/ ) ) {
+    guard = substr( block, 1, RLENGTH );
+    code = gensub(/^[^\n]+\n/, "", 1, block);
+    attrib = gensub(/^:::+[ \t]*\{?[ \t]*([^\}\n]*)\}?[ \t]*\n.*$/, "\\1", 1, block);
+    gsub(/[^a-zA-Z0-9_-]+/, " ", attrib);
+    gsub(/(^ | $)/, "", attrib);
+    if ( match(code, "(^|\n)" guard "+(\n|$)" ) ) {
+      len = RLENGTH; st = RSTART;
+      return "<div class=\"" attrib "\">" _block( substr(code, 1, st - 1) ) "</div>\n" \
+             _block( substr( code, st + len ) );
+    } else {
+      match( block, /(^|\n)[[:space:]]*(\n|$)/ ) || match( block, /$/ );
+      len = RLENGTH; st = RSTART;
+      return "<p>" inline( substr(block, 1, st - 1) ) "</p>\n" \
+             _block( substr(block, st + len) );
+    }
+
   # Fenced Code Block (pandoc)
   } else if ( match( block, /^(~~~+|```+)/ ) ) {
     guard = substr( block, 1, RLENGTH );
     code = gensub(/^[^\n]+\n/, "", 1, block);
+    attrib = gensub(/^(~~~+|```+)[ \t]*\{?[ \t]*([^\}\n]*)\}?[ \t]*\n.*$/, "\\2", 1, block);
+    gsub(/[^a-zA-Z0-9_-]+/, " ", attrib);
+    gsub(/(^ | $)/, "", attrib);
     if ( match(code, "(^|\n)" guard "+(\n|$)" ) ) {
       len = RLENGTH; st = RSTART;
-      return "<pre><code>" HTML( substr(code, 1, st - 1) ) "</code></pre>\n" \
+      return "<pre><code class=\"" attrib "\">" HTML( substr(code, 1, st - 1) ) "</code></pre>\n" \
              _block( substr( code, st + len ) );
     } else {
       match( block, /(^|\n)[[:space:]]*(\n|$)/ ) || match( block, /$/ );
@@ -369,6 +446,12 @@ function _block( block, LOCAL, st, len, hlvl, htxt, guard, code, indent ) {
     return "<h" hlvl " id=\"" hid " - " HTML(htxt) "\">" inline( htxt ) "</h" hlvl ">\n\n" \
            _block( substr( block, len + 1) );
 
+  # Split paragraphs
+  } else if ( match( block, /(^|\n)[[:space:]]*(\n|$)/) ) {
+    len = RLENGTH; st = RSTART;
+    return _block( substr(block, 1, st - 1) ) "\n" \
+           _block( substr(block, st + len) );
+
   # Horizontal rule
   } else if ( match( block, /(^|\n) ? ? ?((\* *){3,}|(- *){3,}|(_ *){3,})($|\n)/) ) {
     len = RLENGTH; st = RSTART;
@@ -376,10 +459,7 @@ function _block( block, LOCAL, st, len, hlvl, htxt, guard, code, indent ) {
 
   # Plain paragraph
   } else {
-    match( block, /(^|\n)[[:space:]]*(\n|$)/ ) || match( block, /$/ );
-    len = RLENGTH; st = RSTART;
-    return "<p>" inline( substr(block, 1, st - 1) ) "</p>\n" \
-           _block( substr(block, st + len) );
+    return "<p>" inline(block) "</p>\n";
   }
 }
 
@@ -407,12 +487,28 @@ function _list( block, last, LOCAL, p) {
   }
   sub( /\n$/, "", p );
 
-  # Task List (pandoc)
-       if ( p ~ /^\[ \].*/ )       { p = "<input type=checkbox disabled />" substr(p, 4); }
-  else if ( p ~ /^\[[xX]\].*/ )    { p = "<input type=checkbox disabled checked />" substr(p, 4); }
-  else if ( p ~ /^<p>\[ \].*/ )    { p = "<p><input type=checkbox disabled />" substr(p, 7); }
-  else if ( p ~ /^<p>\[[xX]\].*/ ) { p = "<p><input type=checkbox disabled checked />" substr(p, 7); }
-  return "<li>" p "</li>\n" _list( block, last );
+  # Task List (pandoc, custom)
+         if ( p ~ /^\[ \].*/ )       { return "<li class=\"task pending\"><input type=checkbox disabled />" \
+                                              substr(p, 4) "</li>\n" _list( block, last );
+  } else if ( p ~ /^\[-\].*/ )       { return "<li class=\"task negative\"><input type=checkbox disabled />" \
+                                              substr(p, 4) "</li>\n" _list( block, last );
+  } else if ( p ~ /^\[\?\].*/ )      { return "<li class=\"task unsure\"><input type=checkbox disabled />" \
+                                              substr(p, 4) "</li>\n" _list( block, last );
+  } else if ( p ~ /^\[\/\].*/ )      { return "<li class=\"task partial\"><input type=checkbox disabled />" \
+                                              substr(p, 4) "</li>\n" _list( block, last );
+  } else if ( p ~ /^\[[xX]\].*/ )    { return "<li class=\"task done\"><input type=checkbox disabled checked />" \
+                                            substr(p, 4) "</li>\n" _list( block, last );
+  } else if ( p ~ /^<p>\[ \].*/ )    { return "<li class=\"task pending\"><p><input type=checkbox disabled />" \
+                                              substr(p, 7) "</li>\n" _list( block, last );
+  } else if ( p ~ /^<p>\[-\].*/ )    { return "<li class=\"task negative\"><p><input type=checkbox disabled />" \
+                                              substr(p, 7) "</li>\n" _list( block, last );
+  } else if ( p ~ /^<p>\[\?\].*/ )   { return "<li class=\"task unsure\"><p><input type=checkbox disabled />" \
+                                              substr(p, 7) "</li>\n" _list( block, last );
+  } else if ( p ~ /^<p>\[\/\].*/ )   { return "<li class=\"task partial\"><p><input type=checkbox disabled />" \
+                                              substr(p, 7) "</li>\n" _list( block, last );
+  } else if ( p ~ /^<p>\[[xX]\].*/ ) { return "<li class=\"task done\"><p><input type=checkbox disabled checked />" \
+                                              substr(p, 7) "</li>\n" _list( block, last );
+  } else { return "<li>" p "</li>\n" _list( block, last ); }
 }
 
 BEGIN {