X-Git-Url: http://git.plutz.net/?a=blobdiff_plain;f=markdown.awk;h=ae1cb934132d9d54c66755a850707d77db9a5217;hb=887a68a142e6eb59aecd1757b591ab7ab37cd279;hp=fbfc455355c14c6df2d34cd44a42daa8dee09794;hpb=e6442fb6aad465f6b0696b2ed1477d2dbc05bfc7;p=cgilite

diff --git a/markdown.awk b/markdown.awk
index fbfc455..ae1cb93 100755
--- a/markdown.awk
+++ b/markdown.awk
@@ -38,7 +38,9 @@
 #
 # Extensions - Block elements:
 # ----------------------------
+# - [x] Automatic <section>-wrapping (custom)
 # -  ?  Heading identifiers (php md, pandoc)
+#   - [x] Heading attributes (custom)
 # - [x] Automatic heading identifiers (custom)
 # - [x] Fenced code blocks (php md, pandoc)
 #   - [x] Fenced code attributes
@@ -52,7 +54,7 @@
 #   - [x] Pipe table (php md, pandoc)
 # - [x] Line blocks (pandoc)
 # - [x] Task lists (pandoc, custom)
-# - [ ] Definition lists (php md, pandoc)
+# - [x] Definition lists (php md, pandoc)
 # - [-] Numbered example lists (pandoc)
 # - [-] Metadata blocks (pandoc)
 # - [x] Metadata blocks (custom)
@@ -89,13 +91,13 @@ function HTML ( text ) {
   return text;
 }
 
-function URL ( text ) {
+function URL ( text, sharp ) {
   gsub( /&/,  "%26",  text );
   gsub( /"/,  "%22", text );
   gsub( /'/,  "%27", text );
   gsub( /`/,  "%60", text );
   gsub( /\?/,  "%3F", text );
-  gsub( /#/,  "%23", text );
+  if (sharp) gsub( /#/,  "%23", text );
   gsub( /\[/,  "%5B", text );
   gsub( /\]/,  "%5D", text );
   gsub( / /,  "%20", text );
@@ -195,16 +197,16 @@ function inline( line, LOCAL, len, code, href, guard ) {
     title  = gensub(/^!\[([^]]*)\]\(([^"\)]+)([ \t]+"([^"]+)")?\)(\{([a-zA-Z \t-]*)\})?/, "\\4", "g", substr(line, 1, len) );
     attrib = gensub(/^!\[([^]]*)\]\(([^"\)]+)([ \t]+"([^"]+)")?\)(\{([a-zA-Z \t-]*)\})?/, "\\6", "g", substr(line, 1, len) );
     if ( title && attrib ) {
-      return "<img src=\"" URL(href) "\" alt=\"" HTML(text) "\" title=\"" HTML(title) "\" class=\"" HTML(attrib) "\">" \
+      return "<img src=\"" URL(href, 1) "\" alt=\"" HTML(text) "\" title=\"" HTML(title) "\" class=\"" HTML(attrib) "\">" \
              inline( substr( line, len + 1) );
     } else if ( title ) {
-      return "<img src=\"" URL(href) "\" alt=\"" HTML(text) "\" title=\"" HTML(title) "\">" \
+      return "<img src=\"" URL(href, 1) "\" alt=\"" HTML(text) "\" title=\"" HTML(title) "\">" \
              inline( substr( line, len + 1) );
     } else if ( attrib ) {
-      return "<img src=\"" URL(href) "\" alt=\"" HTML(text) "\" class=\"" HTML(attrib) "\">" \
+      return "<img src=\"" URL(href, 1) "\" alt=\"" HTML(text) "\" class=\"" HTML(attrib) "\">" \
              inline( substr( line, len + 1) );
     } else {
-      return "<img src=\"" URL(href) "\" alt=\"" HTML(text) "\">" \
+      return "<img src=\"" URL(href, 1) "\" alt=\"" HTML(text) "\">" \
              inline( substr( line, len + 1) );
     }
 
@@ -215,10 +217,10 @@ function inline( line, LOCAL, len, code, href, guard ) {
       id = gensub(/^!\[([^\n]*)\] ?\[([^\n]*)\].*/, "\\2", 1, substr(line, 1, len) );
     if ( ! id ) id = text;
     if ( rl_href[id] && rl_title[id] ) {
-      return "<img src=\"" URL(rl_href[id]) "\" alt=\"" HTML(text) "\" title=\"" HTML(rl_title[id]) "\">" \
+      return "<img src=\"" URL(rl_href[id], 1) "\" alt=\"" HTML(text) "\" title=\"" HTML(rl_title[id]) "\">" \
              inline( substr( line, len + 1) );
     } else if ( rl_href[id] ) {
-      return "<img src=\"" URL(rl_href[id]) "\" alt=\"" HTML(text) "\">" \
+      return "<img src=\"" URL(rl_href[id], 1) "\" alt=\"" HTML(text) "\">" \
              inline( substr( line, len + 1) );
     } else {
       return "" HTML(substr(line, 1, len)) inline( substr(line, len + 1) );
@@ -306,18 +308,41 @@ function inline( line, LOCAL, len, code, href, guard ) {
   }
 }
 
+function headline( hlvl, htxt, attrib, LOCAL, sec, n, HL) {
+  split( gensub( /^(.* )?([0-9]+( [0-9]+){5})$/, "\\2" ,"1", hstack),  HL);
+
+  for ( n = hlvl; n <= 6; n++ ) { sec = sec (HL[n]?"</section>":""); }
+  HL[hlvl]++; for ( n = hlvl + 1; n <= 6; n++) { HL[n] = 0;}
+  hid = ""; for ( n = 2; n <= blvl; n++) { hid = hid BL[n] "/"; }
+  hid = hid HL[1]; for ( n = 2; n <= hlvl; n++) { hid = hid "." HL[n] ; }
+  hid = hid ":" URL(htxt, 1);
+
+  hstack = gensub( /^(.* )?([0-9]+( [0-9]+){5})$/, "\\1" ,"1", hstack) \
+           HL[1] " " HL[2] " " HL[3] " " HL[4] " " HL[5] " " HL[6];
+
+  return sec "<section class=\"" (attrib ? "h" hlvl " " attrib : "h" hlvl)  "\" id=\"" hid "\">" \
+         "<h" hlvl (attrib ? " class=\"" attrib "\"" : "") ">" inline( htxt ) \
+         "<a class=\"anchor\" href=\"#" hid "\"></a>" \
+         "</h" hlvl ">\n";
+}
+
 # Nested Block, resets heading counters
-function _nblock( block, LOCAL, st, len, hlvl, htxt, hlsav, sec, guard, code, indent, attrib ) {
-  hlsav = HL[1] " " HL[2] " " HL[3] " " HL[4] " " HL[5] " " HL[6];
-  HL[1] = 0; HL[2] = 0; HL[3] = 0; HL[4] = 0; HL[5] = 0; HL[6] = 0;
+function _nblock( block, LOCAL, hlsav, sec, n ) {
+  hlsav = hstack;
+  hstack = hstack " 0 0 0 0 0 0";
+
+  # Block Level
+  blvl++; BL[blvl]++;
+  for ( n = blvl + 1; n in BL; n++) { delete BL[n]; }
 
   block = _block( block );
+  split( gensub( /^(.* )?([0-9]+( [0-9]+){5})$/, "\\2" ,"1", hstack),  HL);
   sec = ""; for ( n = 1; n <= 6; n++ ) { sec = sec (HL[n]?"</section>":""); }
-  split(hlsav, HL);
+  hstack = hlsav; blvl--;
   return block sec;
 }
 
-function _block( block, LOCAL, st, len, hlvl, htxt, sec, guard, code, indent, attrib ) {
+function _block( block, LOCAL, st, len, text, title, attrib, href, guard, code, indent ) {
   gsub( /^\n+|\n+$/, "", block );
 
   if ( block == "" ) {
@@ -355,7 +380,7 @@ function _block( block, LOCAL, st, len, hlvl, htxt, sec, guard, code, indent, at
   } else if ( match( block, /^> /) ) {
     match( block, /(^|\n)[[:space:]]*(\n|$)/ ) || match(block, /$/);
     len = RLENGTH; st = RSTART;
-    return "<blockquote>\n" _nblock( gensub( /(^|\n)> /, "\n", "g", substr(block, 1, st - 1) ) ) "</blockquote>\n\n" \
+    return "<blockquote>" gensub( /^\n|\n$/, "", "g",  _nblock( gensub( /(^|\n)> /, "\n", "g", substr(block, 1, st - 1) ) ) ) "</blockquote>\n\n" \
            _block( substr(block, st + len) );
 
   # Pipe Tables (pandoc / php md / gfm )
@@ -489,11 +514,11 @@ function _block( block, LOCAL, st, len, hlvl, htxt, sec, guard, code, indent, at
   # Line Blocks (pandoc)
   } else if ( match(block, /^\| [^\n]*(\n|$)(\| [^\n]*(\n|$)|[ \t]+[^\n[:space:]][^\n]*(\n|$))*/) ) {
     len = RLENGTH; st = RSTART;
-    code = substr(block, 1, len);
-    gsub(/\n[[:space:]]+/, " ", code);
-    gsub(/\n\| /, "\n", code);
-    gsub(/^\| |\n$/, "", code);
-    return "<div class=\"line-block\">" gensub(/\n/, "<br>\n", "g", inline( code )) "</div>\n" \
+    text = substr(block, 1, len);
+    gsub(/\n[[:space:]]+/, " ", text);
+    gsub(/\n\| /, "\n", text);
+    gsub(/^\| |\n$/, "", text);
+    return "<div class=\"line-block\">" gensub(/\n/, "<br>\n", "g", inline( text )) "</div>\n" \
            _block( substr( block, len + 1) );
 
   # Indented Code Block
@@ -565,38 +590,71 @@ function _block( block, LOCAL, st, len, hlvl, htxt, sec, guard, code, indent, at
   gsub("(^|\n) {0," indent "}", "\n", list);
   return "\n<ol>\n" _list( substr(list, 2) ) "</ol>\n" _block( block );
 
+  # Definition list
+  } else if (match( block, "^(([ \t]*\n)*[^:\n \t][^\n]+\n" \
+                           "([ \t]*\n)* ? ? ?:[ \t][^\n]+(\n|$)" \
+                           "(([ \t]*\n)* ? ? ?:[ \t][^\n]+(\n|$)" \
+                           "|[^:\n \t][^\n]+(\n|$)" \
+                           "|( ? ? ?\t|  +)[^\n]+(\n|$)" \
+                           "|([ \t]*\n)+( ? ? ?\t|  +)[^\n]+(\n|$))*)+" \
+  )) {
+  list = substr( block, 1, RLENGTH); block = substr( block, RLENGTH + 1);
+  return "\n<dl>\n" _dlist( list ) "</dl>\n" _block( block );
+
+  # First Order Heading H1 + Attrib
+  } else if ( match( block, /^([^\n]+)([ \t]*\{([^\}\n]+)\})\n===+(\n|$)/ ) ) {
+    len = RLENGTH;
+    text   = gensub(/^([^\n]+)([ \t]*\{([^\}\n]+)\})\n===+(\n.*)?$/, "\\1", 1, block)
+    attrib = gensub(/^([^\n]+)([ \t]*\{([^\}\n]+)\})\n===+(\n.*)?$/, "\\3", 1, block)
+    gsub(/[^a-zA-Z0-9_-]+/, " ", attrib); gsub(/(^ | $)/, "", attrib);
+
+    return headline(1, text, attrib) \
+           _block( substr( block, len + 1 ) );
+
   # First Order Heading H1
-  } else if ( match( block, /^[^\n]+\n===+(\n|$)/ ) ) {
-    len = RLENGTH; sec = "";
-    for ( n = 1; n <= 6; n++ ) { sec = sec (HL[n]?"</section>":""); }
-    HL[1]++; HL[2] = 0; HL[3] = 0; HL[4] = 0; HL[5] = 0; HL[6] = 0;
-    return sec "<section id=\"" HL[1] ":" URL(gensub( /\n.*$/, "", "g", block )) "\"><h1>" \
-           inline( gensub( /\n.*$/, "", "g", block ) ) \
-           "<a class=\"anchor\" href=\"#" HL[1] ":" \
-           URL(gensub( /\n.*$/, "", "g", block )) "\"></a></h1>\n\n" \
+  } else if ( match( block, /^([^\n]+)\n===+(\n|$)/ ) ) {
+    len = RLENGTH;
+    text   = gensub(/^([^\n]+)\n===+(\n.*)?$/, "\\1", 1, block)
+
+    return headline(1, text, 0) \
            _block( substr( block, len + 1 ) );
 
+  # Second Order Heading H2 + Attrib
+  } else if ( match( block, /^([^\n]+)([ \t]*\{([^\}\n]+)\})\n---+(\n|$)/ ) ) {
+    len = RLENGTH;
+    text   = gensub(/^([^\n]+)([ \t]*\{([^\}\n]+)\})\n---+(\n.*)?$/, "\\1", 1, block)
+    attrib = gensub(/^([^\n]+)([ \t]*\{([^\}\n]+)\})\n---+(\n.*)?$/, "\\3", 1, block)
+    gsub(/[^a-zA-Z0-9_-]+/, " ", attrib); gsub(/(^ | $)/, "", attrib);
+
+    return headline(2, text, attrib) \
+           _block( substr( block, len + 1) );
+
   # Second Order Heading H2
-  } else if ( match( block, /^[^\n]+\n---+(\n|$)/ ) ) {
-    len = RLENGTH; sec = "";
-    for ( n = 2; n <= 6; n++ ) { sec = sec (HL[n]?"</section>":""); }
-    HL[2]++; HL[3] = 0; HL[4] = 0; HL[5] = 0; HL[6] = 0;
-    return sec "<section id=\"" HL[1] "." HL[2] ":" URL(gensub( /\n.*$/, "", "g", block )) "\">" \
-           "<h2>" inline( gensub( /\n.*$/, "", "g", block ) ) \
-           "<a class=\"anchor\" href=\"#" HL[1] "." HL[2] ":" \
-           URL(gensub( /\n.*$/, "", "g", block )) "\"></a></h2>\n\n" \
+  } else if ( match( block, /^([^\n]+)\n---+(\n|$)/ ) ) {
+    len = RLENGTH;
+    text   = gensub(/^([^\n]+)\n---+(\n.*)?$/, "\\1", 1, block)
+
+    return headline(2, text, 0) \
+           _block( substr( block, len + 1) );
+
+  # Nth Order Heading H1 H2 H3 H4 H5 H6 + Attrib
+  } else if ( match( block, /^(#{1,6})[ \t]*(([^ \t\n]+|[ \t]+[^ \t\n#]|[ \t]+#+[ \t]*[^ \t\n#])+)[ \t]*#*([ \t]*\{([a-zA-Z \t-]*)\})(\n|$)/ ) ) {
+    len = RLENGTH;
+    n      = gensub(/^(#{1,6})[ \t]*(([^ \t\n]+|[ \t]+[^ \t\n#]|[ \t]+#+[ \t]*[^ \t\n#])+)[ \t]*#*([ \t]*\{([a-zA-Z \t-]*)\})(\n.*)?$/, "\\1", 1, block);
+    text   = gensub(/^(#{1,6})[ \t]*(([^ \t\n]+|[ \t]+[^ \t\n#]|[ \t]+#+[ \t]*[^ \t\n#])+)[ \t]*#*([ \t]*\{([a-zA-Z \t-]*)\})(\n.*)?$/, "\\2", 1, block);
+    attrib = gensub(/^(#{1,6})[ \t]*(([^ \t\n]+|[ \t]+[^ \t\n#]|[ \t]+#+[ \t]*[^ \t\n#])+)[ \t]*#*([ \t]*\{([a-zA-Z \t-]*)\})(\n.*)?$/, "\\5", 1, block);
+    gsub(/[^a-zA-Z0-9_-]+/, " ", attrib); gsub(/(^ | $)/, "", attrib);
+
+    return headline( length(n), text, attrib ) \
            _block( substr( block, len + 1) );
 
   # Nth Order Heading H1 H2 H3 H4 H5 H6
-  } else if ( match( block, /^#{1,6}[ \t]*[^\n]+([ \t]*#*)(\n|$)/ ) ) {
-    len = RLENGTH; sec = "";
-    hlvl = length( gensub( /^(#{1,6}).*$/, "\\1", "g", block ) );
-    htxt = gensub(/^#{1,6}[ \t]*(([^ \t\n]+|[ \t]+[^ \t\n#]|[ \t]+#+[^\n#])+)([ \t]*#*)(\n.*)?$/, "\\1", 1, block);
-    for ( n = hlvl; n <= 6; n++ ) { sec = sec (HL[n]?"</section>":""); }
-    HL[hlvl]++; for ( n = hlvl + 1; n <= 6; n++) { HL[n] = 0;}
-    hid = HL[1]; for ( n = 2; n <= hlvl; n++) { hid = hid "." HL[n] ; }
-    return sec "<section id=\"" hid ":" URL(htxt) "\"><h" hlvl ">" inline( htxt ) \
-           "<a class=\"anchor\" href=\"#" hid "\"></a></h" hlvl ">\n\n" \
+  } else if ( match( block, /^(#{1,6})[ \t]*(([^ \t\n]+|[ \t]+[^ \t\n#]|[ \t]+#+[ \t]*[^ \t\n#])+)[ \t]*#*(\n|$)/ ) ) {
+    len = RLENGTH;
+    n      = gensub(/^(#{1,6})[ \t]*(([^ \t\n]+|[ \t]+[^ \t\n#]|[ \t]+#+[ \t]*[^ \t\n#])+)[ \t]*#*(\n.*)?$/, "\\1", 1, block);
+    text   = gensub(/^(#{1,6})[ \t]*(([^ \t\n]+|[ \t]+[^ \t\n#]|[ \t]+#+[ \t]*[^ \t\n#])+)[ \t]*#*(\n.*)?$/, "\\2", 1, block);
+
+    return headline( length(n), text, 0 ) \
            _block( substr( block, len + 1) );
 
   # block images (wrapped in <figure>)
@@ -607,25 +665,25 @@ function _block( block, LOCAL, st, len, hlvl, htxt, sec, guard, code, indent, at
     title  = gensub(/^!\[([^]]*)\]\(([^"\)]+)([ \t]+"([^"]+)")?\)(\{([a-zA-Z \t-]*)\})?(\n.*)?$/, "\\4", "g", block);
     attrib = gensub(/^!\[([^]]*)\]\(([^"\)]+)([ \t]+"([^"]+)")?\)(\{([a-zA-Z \t-]*)\})?(\n.*)?$/, "\\6", "g", block);
     if ( title && attrib ) {
-      return "<figure data-src=\"" URL(href) "\" class=\"" HTML(attrib) "\">" \
-               "<img src=\"" URL(href) "\" alt=\"" HTML(text) "\" class=\"" HTML(attrib) "\">" \
+      return "<figure data-src=\"" URL(href, 1) "\" class=\"" HTML(attrib) "\">" \
+               "<img src=\"" URL(href, 1) "\" alt=\"" HTML(text) "\" class=\"" HTML(attrib) "\">" \
                "<figcaption>" inline(title) "</figcaption>" \
              "</figure>\n\n" \
              _block( substr( block, len + 1) );
     } else if ( title ) {
-      return "<figure data-src=\"" URL(href) "\">" \
-               "<img src=\"" URL(href) "\" alt=\"" HTML(text) "\">" \
+      return "<figure data-src=\"" URL(href, 1) "\">" \
+               "<img src=\"" URL(href, 1) "\" alt=\"" HTML(text) "\">" \
                "<figcaption>" inline(title) "</figcaption>" \
              "</figure>\n\n" \
              _block( substr( block, len + 1) );
     } else if ( attrib ) {
-      return "<figure data-src=\"" URL(href) "\" class=\"" HTML(attrib) "\">" \
-               "<img src=\"" URL(href) "\" alt=\"" HTML(text) "\" class=\"" HTML(attrib) "\">" \
+      return "<figure data-src=\"" URL(href, 1) "\" class=\"" HTML(attrib) "\">" \
+               "<img src=\"" URL(href, 1) "\" alt=\"" HTML(text) "\" class=\"" HTML(attrib) "\">" \
              "</figure>\n\n" \
              _block( substr( block, len + 1) );
     } else {
-      return "<figure data-src=\"" URL(href) "\">" \
-               "<img src=\"" URL(href) "\" alt=\"" HTML(text) "\">" \
+      return "<figure data-src=\"" URL(href, 1) "\">" \
+               "<img src=\"" URL(href, 1) "\" alt=\"" HTML(text) "\">" \
              "</figure>\n\n" \
              _block( substr( block, len + 1) );
     }
@@ -637,14 +695,14 @@ function _block( block, LOCAL, st, len, hlvl, htxt, sec, guard, code, indent, at
       id = gensub(/^!\[([^\n]*)\] ?\[([^\n]*)\](\n.*)?$/, "\\2", 1, block);
     if ( ! id ) id = text;
     if ( rl_href[id] && rl_title[id] ) {
-      return "<figure data-src=\"" URL(rl_href[id]) "\">" \
-               "<img src=\"" URL(rl_href[id]) "\" alt=\"" HTML(text) "\">" \
+      return "<figure data-src=\"" URL(rl_href[id], 1) "\">" \
+               "<img src=\"" URL(rl_href[id], 1) "\" alt=\"" HTML(text) "\">" \
                "<figcaption>" inline(rl_title[id]) "</figcaption>" \
              "</figure>\n\n" \
              _block( substr( block, len + 1) );
     } else if ( rl_href[id] ) {
-      return "<figure data-src=\"" URL(rl_href[id]) "\">" \
-               "<img src=\"" URL(rl_href[id]) "\" alt=\"" HTML(text) "\">" \
+      return "<figure data-src=\"" URL(rl_href[id], 1) "\">" \
+               "<img src=\"" URL(rl_href[id], 1) "\" alt=\"" HTML(text) "\">" \
              "</figure>\n\n" \
              _block( substr( block, len + 1) );
     } else {
@@ -674,6 +732,31 @@ function _block( block, LOCAL, st, len, hlvl, htxt, sec, guard, code, indent, at
   }
 }
 
+function _dlist (block, LOCAL, len, st, text, indent, p) {
+  if (match( block, "^([ \t]*\n)*[^:\n \t][^\n]+\n" )) {
+    len = RLENGTH; text = substr(block, 1, len);
+    gsub( "(^\n*|\n*$)", "", text );
+    return "<dt>" inline( text ) "</dt>\n" _dlist( substr(block, len + 1) );
+  } else if (match( block, "^([ \t]*\n)* ? ? ?:[ \t][^\n]+(\n|$)" \
+                         "([^:\n \t][^\n]+(\n|$)" \
+                         "|( ? ? ?\t|  +)[^\n]+(\n|$)" \
+                         "|([ \t]*\n)+( ? ? ?\t|  +)[^\n]+(\n|$))*" \
+  )) {
+    len = RLENGTH; text = substr(block, 1, len);
+    if (match( text, "^ ? ? ?:[ \t][^\n]+(\n|$)" \
+                    "([^:\n \t][^\n]+(\n|$)" \
+                    "|( ? ? ?\t|  +)[^\t\n ][^\n]*(\n|$))*$" \
+    )) p = 1; else p = 0;
+    sub( "^([ \t]*\n)*", "", text);
+    match(text, "^ ? ? ?:(\t| +)"); indent = RLENGTH;
+    sub( "^ ? ? ?:(\t| +)", "", text);
+    gsub( "(^|\n) {0," indent "}", "\n", text );
+    text = _block(text);
+    if (p) gsub("(^\n*<p>|</p>\n*$)", "", text);
+    return "<dd>" text "</dd>\n" _dlist( substr(block, len + 1) );
+  }
+}
+
 function _list( block, last, LOCAL, p) {
   if ( ! length(block) ) return "";
   gsub(/^([-+*]|[0-9]+\.|#\.)(  ? ? ?|\t)/, "", block)