improved lists

diff --git a/markdown.awk b/markdown.awk
index ae1cb934132d9d54c66755a850707d77db9a5217..949ea9390f8450e8982b10945854f29076bde118 100755 (executable)
--- a/markdown.awk
+++ b/markdown.awk
@@ -342,8 +342,8 @@ function _nblock( block, LOCAL, hlsav, sec, n ) {
   return block sec;
 }
 
-function _block( block, LOCAL, st, len, text, title, attrib, href, guard, code, indent ) {
-  gsub( /^\n+|\n+$/, "", block );
+function _block( block, LOCAL, st, len, text, title, attrib, href, guard, code, indent, list ) {
+  gsub( "(^\n+|\n+$)", "", block );
 
   if ( block == "" ) {
     return "";
@@ -533,7 +533,7 @@ function _block( block, LOCAL, st, len, text, title, attrib, href, guard, code,
   # Fenced Divs (pandoc, custom)
   } else if ( match( block, /^(:::+)/ ) ) {
     guard = substr( block, 1, RLENGTH );
-    code = gensub(/^[^\n]+\n/, "", 1, block);
+    code = block; sub(/^[^\n]+\n/, "", code);
     attrib = gensub(/^:::+[ \t]*\{?[ \t]*([^\}\n]*)\}?[ \t]*\n.*$/, "\\1", 1, block);
     gsub(/[^a-zA-Z0-9_-]+/, " ", attrib);
     gsub(/(^ | $)/, "", attrib);
@@ -566,41 +566,6 @@ function _block( block, LOCAL, st, len, text, title, attrib, href, guard, code,
              _block( substr(block, st + len) );
     }
 
-  # Unordered list
-  } else if ( match( block, "^ ? ? ?[-+*][ \t]+[^\n]+(\n|$)" \
-                            "(([ \t]*\n)* ? ? ?[-+*][ \t]+[^\n]+(\n|$)" \
-                            "|([ \t]*\n)*( ? ? ?\t|  +)[^\n]+(\n|$)" \
-                            "|[^\n]+(\n|$))*" ) ) {
-  list = substr( block, 1, RLENGTH);
-  block = substr( block, RLENGTH + 1);
-  indent = length( gensub(/[-+*][ \t]+[^\n]+.*$/, "", 1, list) );
-
-  gsub("(^|\n) {0," indent "}", "\n", list);
-  return "\n<ul>\n" _list( substr(list, 2) ) "</ul>\n" _block( block );
-
-  # Ordered list
-  } else if ( match( block, "^ ? ? ?([0-9]+|#)\\.[ \t]+[^\n]+(\n|$)" \
-                            "(([ \t]*\n)* ? ? ?([0-9]+|#)\\.[ \t]+[^\n]+(\n|$)" \
-                            "|([ \t]*\n)*( ? ? ?\t|  +)[^\n]+(\n|$)" \
-                            "|[^\n]+(\n|$))*" ) ) {
-  list = substr( block, 1, RLENGTH);
-  block = substr( block, RLENGTH + 1);
-  indent = length( gensub(/([0-9]+|#)\.[ \t]+[^\n]+.*$/, "", 1, list) );
-
-  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;
@@ -715,23 +680,96 @@ function _block( block, LOCAL, st, len, text, title, attrib, href, guard, code,
     text = gensub(/^<<(([^>]|>[^>])+)>>(\n.*)?$/, "\\1", 1, block);
     return macro(text) _block(substr(block, len + 1) );
 
+  # Horizontal rule
+  } else if ( match( block, /(^|\n) ? ? ?((\* *){3,}|(- *){3,}|(_ *){3,})($|\n)/) ) {
+    len = RLENGTH; st = RSTART;
+    return _block(substr(block, 1, st - 1)) "<hr>\n" _block(substr(block, st + len));
+
+  # Definition list
+  } else if (match( block, "^(([ \t]*\n)*[^:\n \t][^\n]+\n" \
+                           "([ \t]*\n)* ? ? ?:[ \t][^\n]+(\n|$)" \
+                          "(([ \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 );
+
+  # Unordered list
+  } else if ( match( block, "(^|\n) ? ? ?[-+*][ \t][^\n]+(\n|$)" \
+                 "(([ \t]*\n)* ? ? ?[-+*][ \t][^\n]+(\n|$)" \
+                 "|([ \t]*\n)*( ? ? ?\t|  +)[^\n]+(\n|$)" \
+                 "|[^\n \t][^\n]+(\n|$))*" ) ) {
+    st = RSTART; len = RLENGTH; list = substr( block, RSTART, RLENGTH);
+    sub("^\n", "", list); match(list, "^ ? ? ?[-+*]"); indent = RLENGTH;
+    gsub( "(^|\n) {0," indent - 1 "}", "\n", list); sub("^\n", "", list);
+
+    text = substr(block, 1, st - 1); block = substr(block, st + len);
+    if (match( list, "\n([0-9]+\\.|#\\.)[ \t]" )) {
+      block = substr(list, RSTART + 1) block;
+      list = substr(list, 1, RSTART);
+    }
+
+    return _block( text ) "<ul>\n" _list( list, "[-+*]" ) "</ul>\n" _block( block );
+
+  # Ordered list
+  } else if ( match( block, "(^|\n) ? ? ?([0-9]+\\.|#\\.)[ \t][^\n]+(\n|$)" \
+                 "(([ \t]*\n)* ? ? ?([0-9]+\\.|#\\.)[ \t][^\n]+(\n|$)" \
+                 "|([ \t]*\n)*( ? ? ?\t|  +)[^\n]+(\n|$)" \
+                 "|[^\n \t][^\n]+(\n|$))*" ) ) {
+    st = RSTART; len = RLENGTH; list = substr( block, RSTART, RLENGTH);
+    sub("^\n", "", list); match(list, "^ ? ? ?[0-9#]"); indent = RLENGTH;
+    gsub( "(^|\n) {0," indent - 1 "}", "\n", list); sub("^\n", "", list);
+
+    text = substr(block, 1, st - 1); block = substr(block, st + len);
+    if (match( list, "\n[-+*][ \t]" )) {
+      block = substr(list, RSTART + 1) block;
+      list = substr(list, 1, RSTART);
+    }
+
+    return _block( text ) "<ol>\n" _list( list, "([0-9]+\\.|#\\.)" ) "</ol>\n" _block( block );
+
   # 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;
-    return _block(substr(block, 1, st - 1)) "<hr>\n" _block(substr(block, st + len));
-
   # Plain paragraph
   } else {
     return "<p>" inline(block) "</p>\n";
   }
 }
 
+function _list (block, mark, LOCAL, len, st, text, indent, task) {
+  if ( match(block, "^([ \t]*\n)*$")) return;
+  match(block, "^" mark "[ \t]"); indent = RLENGTH;
+  sub("^" mark "[ \t]", "", block);
+
+  match( block, "\n" mark "[ \t][^\n]+(\n|$)" \
+      "(([ \t]*\n)* ? ? ?" mark "[ \t][^\n]+(\n|$)" \
+      "|([ \t]*\n)*( ? ? ?\t|  +)[^\n]+(\n|$)" \
+      "|[^\n \t][^\n]+(\n|$))*");
+  (RLENGTH == -1) ? st = length(block) + 1 : st = RSTART;
+  text = substr(block, 1, st); block = substr(block, st + 1);
+
+  gsub("\n {0," indent "}", "\n", text);
+
+  task = match( text, /^\[ \]/   ) ? "<li class=\"task pending\"><input type=checkbox disabled>"      : \
+         match( text, /^\[-\]/   ) ? "<li class=\"task negative\"><input type=checkbox disabled>"     : \
+         match( text, /^\[\/\]/  ) ? "<li class=\"task partial\"><input type=checkbox disabled>"      : \
+         match( text, /^\[\?\]/  ) ? "<li class=\"task unsure\"><input type=checkbox disabled>"       : \
+         match( text, /^\[[xX]\]/) ? "<li class=\"task done\"><input type=checkbox disabled checked>" : "<li>";
+  sub(/^\[[- /xX]\]/, "", text);
+
+  text = _block( text );
+  if (match( text, "^<p>(</p[^>]|</[^p]|<[^/]|[^<])*</p>\n$" ))
+     gsub( "(^<p>|</p>\n$)", "", text);
+
+  return task text "</li>\n" _list(block, mark);
+}
+
 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);
@@ -743,66 +781,17 @@ function _dlist (block, LOCAL, len, st, text, indent, p) {
                          "|([ \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)
 
-  # slice next list item from input
-  if ( match( block, /\n([-+*]|[0-9]+\.|#\.)[ \t]+[^\n]+/) ) {
-    p = substr( block, 1, RSTART);
-    block = substr( block, RSTART + 1);
-  } else {
-    p = block; block = "";
-  }
-  sub( /\n +([-+*]|[0-9]+\.|#\.)/, "\n&", p );
+    text = _block(text);
+    if (match( text, "^<p>(</p[^>]|</[^p]|<[^/]|[^<])*</p>\n$" ))
+       gsub( "(^<p>|</p>\n$)", "", text);
 
-  # if this should be a paragraph item
-  # either previous item (last) or current item (p) contains blank lines
-  if (match(last, /\n[[:space:]]*\n/) || match(p, /\n[[:space:]]*\n/) ) {
-    last = p; p = _nblock(p);
-  } else {
-    last = p; p = _nblock(p);
-    sub( /^<p>/, "", p );
-    sub( /<\/p>\n/, "", p );
+    return "<dd>" text "</dd>\n" _dlist( substr(block, len + 1) );
   }
-  sub( /\n$/, "", p );
-
-  # 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 {