X-Git-Url: http://git.plutz.net/?a=blobdiff_plain;f=markdown.awk;h=af3d7224657418caa809b5fa6223ce3cc1411224;hb=d61539cdd435eb812053104ab5a12700607732be;hp=63bb044e8a69c13a87c58c71194ecceaf9b381b8;hpb=ffe17caf6f15242bc1c0e2681d4aa5f46e15f378;p=cgilite

diff --git a/markdown.awk b/markdown.awk
index 63bb044..af3d722 100755
--- a/markdown.awk
+++ b/markdown.awk
@@ -42,11 +42,14 @@
 # - [x] Automatic heading identifiers (custom)
 # - [x] Fenced code blocks (php md, pandoc)
 #   - [x] Fenced code attributes
+# - [x] Images (as block elements, <figure>-wrapped) (custom)
+#   - [x] reference style block images
 # - [/] Tables
 #   -  ?  Simple table (pandoc)
 #   -  ?  Multiline table (pandoc)
 #   - [x] Grid table (pandoc)
-#   - [x] Pipe table (php md pandoc)
+#     - [x] Headerless
+#   - [x] Pipe table (php md, pandoc)
 # - [x] Line blocks (pandoc)
 # - [x] Task lists (pandoc, custom)
 # - [ ] Definition lists (php md, pandoc)
@@ -62,7 +65,7 @@
 # - [x] ^Superscript^ ~Subscript~ (pandoc)
 # - [-] Bracketed spans (pandoc)
 #   - [-] Inline attributes (pandoc)
-# - [x] Image attributes (custom, pandoc inspired, inline only)
+# - [x] Image attributes (custom, pandoc inspired, not for reference style)
 # - [x] Wiki style links [[PageName]] / [[PageName|Link Text]]
 # - [-] TEX-Math (pandoc)
 # -  ?  Footnotes (php md)
@@ -72,7 +75,7 @@
 # -  ?  ... three-dot ellipsis (smartypants)
 # - [-] en-dash (smartypants)
 # - [ ] Automatic em-dash / en-dash
-# - [ ] Automatic -> Arrows <-
+# - [x] Automatic -> Arrows <- (custom)
 
 function debug(text) { printf "\n---\n%s\n---\n", text > "/dev/stderr"; }
 
@@ -111,9 +114,9 @@ function inline( line, LOCAL, len, code, href, guard ) {
   if ( line ~ /^$/ ) {  # Recursion End
     return "";
 
-  #  omit processing of escaped characters
-  } else if ( line ~ /^\\[]\\`\*_\{\}\(\)#\+-\.![]/) {
-    return substr(line, 2, 1) inline( substr(line, 3) );
+  # omit processing of escaped characters
+  } else if ( line ~ /^\\./) {
+    return HTML(substr(line, 2, 1)) inline( substr(line, 3) );
 
   # hard brakes
   } else if ( match(line, /^  \n/) ) {
@@ -134,10 +137,10 @@ function inline( line, LOCAL, len, code, href, guard ) {
     }
 
   # Wiki style links
-  } else if ( match( line, /^\[\[([^\]\|]+)(\|([^\]]+))?\]\]/) ) {
+  } else if ( match( line, /^\[\[([^]|]+)(\|[^]]+)?\]\]/) ) {
     len = RLENGTH;
-    href = gensub(/^\[\[([^\]\|]+)(\|([^\]]+))?\]\]/, "\\1", 1, substr(line, 1, len) );
-    text = gensub(/^\[\[([^\]\|]+)(\|([^\]]+))?\]\]/, "\\3", 1, substr(line, 1, len) );
+    href = gensub(/^\[\[([^]|]+)(\|([^]]+))?\]\]/, "\\1", 1, substr(line, 1, len) );
+    text = gensub(/^\[\[([^]|]+)(\|([^]]+))?\]\]/, "\\3", 1, substr(line, 1, len) );
     if ( ! text ) text = href;
     return "<a href=\"" URL(href) "\">" HTML(text) "</a>" inline( substr( line, len + 1) );
 
@@ -154,11 +157,15 @@ function inline( line, LOCAL, len, code, href, guard ) {
     return "<a href=\"mailto:" href "\">" href "</a>" inline( substr( line, len + 1) );
 
   # inline links
-  } else if ( match(line, /^\[([^]]+)\]\(([^"\)]+)([[:space:]]+"([^"]+)")?\)/) ) {
+  #                                 ,_______________________Image____________________________,
+  } else if ( match(line, /^\[([^]]+|!\[[^]]*\]\([^"\)]+([ \t]+"[^"]+")?\)(\{[a-zA-Z \t-]*\})?)\]\(([^"\)]+)([[:space:]]+"([^"]+)")?\)/) ) {
     len = RLENGTH;
-    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) );
+    text  = gensub(/^\[([^]]+|!\[[^]]*\]\([^"\)]+([ \t]+"[^"]+")?\)(\{[a-zA-Z \t-]*\})?)\]\(([^"\)]+)([[:space:]]+"([^"]+)")?\)/, \
+                   "\\1", 1, substr(line, 1, len) );
+    href  = gensub(/^\[([^]]+|!\[[^]]*\]\([^"\)]+([ \t]+"[^"]+")?\)(\{[a-zA-Z \t-]*\})?)\]\(([^"\)]+)([[:space:]]+"([^"]+)")?\)/, \
+                   "\\4", 1, substr(line, 1, len) );
+    title = gensub(/^\[([^]]+|!\[[^]]*\]\([^"\)]+([ \t]+"[^"]+")?\)(\{[a-zA-Z \t-]*\})?)\]\(([^"\)]+)([[:space:]]+"([^"]+)")?\)/, \
+                   "\\6", 1, substr(line, 1, len) );
     if ( title ) {
       return "<a href=\"" URL(href) "\" title=\"" HTML(title) "\">" inline( text ) "</a>" inline( substr( line, len + 1) );
     } else {
@@ -180,12 +187,12 @@ function inline( line, LOCAL, len, code, href, guard ) {
     }
 
   # inline images
-  } else if ( match(line, /^!\[([^]]+)\]\(([^"\)]+)([ \t]+"([^"]+)")?\)(\{([a-zA-Z \t-]*)\})?/) ) {
+  } else if ( match(line, /^!\[([^]]*)\]\(([^"\)]+)([ \t]+"([^"]+)")?\)(\{([a-zA-Z \t-]*)\})?/) ) {
     len = RLENGTH;
-    text   = gensub(/^!\[([^]]+)\]\(([^"\)]+)([ \t]+"([^"]+)")?\)(\{([a-zA-Z \t-]*)\})?/, "\\1", "g", substr(line, 1, len) );
-    href   = gensub(/^!\[([^]]+)\]\(([^"\)]+)([ \t]+"([^"]+)")?\)(\{([a-zA-Z \t-]*)\})?/, "\\2", "g", substr(line, 1, len) );
-    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) );
+    text   = gensub(/^!\[([^]]*)\]\(([^"\)]+)([ \t]+"([^"]+)")?\)(\{([a-zA-Z \t-]*)\})?/, "\\1", "g", substr(line, 1, len) );
+    href   = gensub(/^!\[([^]]*)\]\(([^"\)]+)([ \t]+"([^"]+)")?\)(\{([a-zA-Z \t-]*)\})?/, "\\2", "g", substr(line, 1, len) );
+    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) "\">" \
              inline( substr( line, len + 1) );
@@ -201,10 +208,10 @@ function inline( line, LOCAL, len, code, href, guard ) {
     }
 
   # reference style images
-  } else if ( match(line, /^!\[([^]]+)\] ?\[([^]]*)\]/ ) ) {
+  } else if ( match(line, /^!\[([^]]*)\] ?\[([^]]*)\]/ ) ) {
     len = RLENGTH;
-    text = gensub(/^!\[([^\n]+)\] ?\[([^\n]*)\].*/, "\\1", 1, substr(line, 1, len) );
-      id = gensub(/^!\[([^\n]+)\] ?\[([^\n]*)\].*/, "\\2", 1, substr(line, 1, len) );
+    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=\"" URL(rl_href[id]) "\" alt=\"" HTML(text) "\" title=\"" HTML(rl_title[id]) "\">" \
@@ -280,6 +287,14 @@ function inline( line, LOCAL, len, code, href, guard ) {
     len = RLENGTH;
     return substr( line, 1, len ) inline(substr(line, len + 1));
 
+  # Arrows
+  } else if ( line ~ /^-->( |$)/) {  # ignore multidash-arrow
+    return "--&gt;" inline( substr(line, 4) );
+  } else if ( line ~ /^<-( |$)/) {
+    return "&larr;" inline( substr(line, 3) );
+  } else if ( line ~ /^->( |$)/) {
+    return "&rarr;" inline( substr(line, 3) );
+
   # Escape lone HTML character
   } else if ( match( line, /^[&<>"']/) ) {
     return HTML(substr(line, 1, 1)) inline(substr(line, 2));
@@ -379,46 +394,65 @@ function _block( block, LOCAL, st, len, hlvl, htxt, guard, code, indent, attrib
     return "<table>" ttext "</tbody></table>\n" _block(block);
 
   # Grid Tables (pandoc)
-  } else if ( match(block, "^\\+(-+\\+)+\n" \
-                       "(\\|([^\n]+\\|)+\n)+" \
-                        "\\+(:?=+:?\\+)+\n" \
-                      "((\\|([^\n]+\\|)+\n)+" \
-                            "\\+(-+\\+)+(\n|$))+" \
-            ) ) {
+  # (with, and without header)
+  } else if ( match( block, "^\\+(-+\\+)+\n" \
+                            "(\\|([^\n]+\\|)+\n)+" \
+                            "(\\+(:?=+:?\\+)+)\n" \
+                           "((\\|([^\n]+\\|)+\n)+" \
+                             "\\+(-+\\+)+(\n|$))+" \
+                   ) || \
+              match( block, "^()()()" \
+                            "(\\+(:?-+:?\\+)+)\n" \
+                           "((\\|([^\n]+\\|)+\n)+" \
+                             "\\+(-+\\+)+(\n|$))+" \
+  ) ) {
     len = RLENGTH; st = RSTART;
     #initialize empty arrays
     split("", talign); split("", tarray); split("", tread);
     cols = 0; cnt=0; ttext = "";
 
-    # table header and alignment
-    block = substr(block, match(block, /(\n|$)/) + 1 );
-    while ( match(block, "^\\|([^\n]+\\|)+\n") ) {
-      cols = split( gensub( /(^\||\|$)/, "", "g", \
-             gensub( /(^|[^\\])\\\|/, "\\1\\&#x7C;", "g", \
-             substr(block, 1, match(block, /(\n|$)/)) \
-      )), tread, /\|/);
-      block = substr(block, match(block, /(\n|$)/) + 1 );
-      for (cnt = 1; cnt < cols; cnt++)
-        tarray[cnt] = tarray[cnt] "\n" tread[cnt];
-    }
+    # Column Count
+    cols = split(   gensub( "^(\\+(:?-+:?\\+)+)(\n.*)*$", "\\1", 1, block), tread, /\+/) - 2;
+    # debug(" Cols: " gensub( "^(\\+(:?-+:?\\+)+)(\n.*)*$", "\\1", 1, block ));
 
-    cols = split( \
-           gensub( /(^\+|\+$)/, "", "g", \
-           substr(block, 1, match(block, /(\n|$)/)) \
-    ), talign, /\+/);
-    block = substr(block, match(block, /(\n|$)/) + 1 );
+    # table alignment
+    split( gensub( "^(.*\n)?\\+((:?=+:?\\+|(:-+|-+:|:-+:)\\+)+)(\n.*)$", "\\2", "g", block ), talign, /\+/ );
+    # debug("Align: " gensub( "^(.*\n)?\\+((:?=+:?\\+|(:-+|-+:|:-+:)\\+)+)(\n.*)$", "\\2", "g", block ));
 
-    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";
+    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] "\">" _block(tarray[cnt]) "</th>"
-    ttext = ttext "</tr>\n</thead><tbody>\n"
+    if ( match(block, "^\\+(-+\\+)+\n" \
+                      "(\\|([^\n]+\\|)+\n)+" \
+                       "\\+(:?=+:?\\+)+\n" \
+                     "((\\|([^\n]+\\|)+\n)+" \
+                       "\\+(-+\\+)+(\n|$))+" \
+    ) ) {
+      # table header
+      block = substr(block, match(block, /(\n|$)/) + 1 );
+      while ( match(block, "^\\|([^\n]+\\|)+\n") ) {
+        split( gensub( /(^\||\|$)/, "", "g", \
+                 gensub( /(^|[^\\])\\\|/, "\\1\\&#x7C;", "g", \
+                   substr(block, 1, match(block, /(\n|$)/)) \
+        )), tread, /\|/);
+        block = substr(block, match(block, /(\n|$)/) + 1 );
+        for (cnt = 1; cnt <= cols; cnt++)
+          tarray[cnt] = tarray[cnt] "\n" tread[cnt];
+      }
+
+      ttext = "<thead>\n<tr>"
+      for (cnt = 1; cnt <= cols; cnt++)
+        ttext = ttext "<th align=\"" talign[cnt] "\">" _block(tarray[cnt]) "</th>"
+      ttext = ttext "</tr>\n</thead>"
+    }
+
+    # table body
+    block = substr(block, match(block, /(\n|$)/) + 1 );
+    ttext = ttext "<tbody>\n"
 
     while ( match(block, /^((\|([^\n]+\|)+\n)+\+(-+\+)+(\n|$))+/ ) ){
       split("", tarray);
@@ -428,13 +462,13 @@ function _block( block, LOCAL, st, len, hlvl, htxt, guard, code, indent, attrib
                substr(block, 1, match(block, /(\n|$)/)) \
         )), tread, /\|/);
         block = substr(block, match(block, /(\n|$)/) + 1 );
-        for (cnt = 1; cnt < cols; cnt++)
+        for (cnt = 1; cnt <= cols; cnt++)
           tarray[cnt] = tarray[cnt] "\n" tread[cnt];
       }
       block = substr(block, match(block, /(\n|$)/) + 1 );
 
       ttext = ttext "<tr>"
-      for (cnt = 1; cnt < cols; cnt++)
+      for (cnt = 1; cnt <= cols; cnt++)
         ttext = ttext "<td align=\"" talign[cnt] "\">" _block(tarray[cnt]) "</td>"
       ttext = ttext "</tr>\n"
     }
@@ -550,6 +584,64 @@ function _block( block, LOCAL, st, len, hlvl, htxt, guard, code, indent, attrib
            "<a class=\"anchor\" href=\"#" hid "\"></a></h" hlvl ">\n\n" \
            _block( substr( block, len + 1) );
 
+  # block images (wrapped in <figure>)
+  } else if ( match(block, /^!\[([^]]*)\]\(([^"\)]+)([ \t]+"([^"]+)")?\)(\{([a-zA-Z \t-]*)\})?(\n|$)/) ) {
+    len = RLENGTH;
+    text   = gensub(/^!\[([^]]*)\]\(([^"\)]+)([ \t]+"([^"]+)")?\)(\{([a-zA-Z \t-]*)\})?(\n.*)?$/, "\\1", "g", block);
+    href   = gensub(/^!\[([^]]*)\]\(([^"\)]+)([ \t]+"([^"]+)")?\)(\{([a-zA-Z \t-]*)\})?(\n.*)?$/, "\\2", "g", block);
+    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) "\">" \
+               "<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) "\">" \
+               "<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) "\">" \
+             "</figure>\n\n" \
+             _block( substr( block, len + 1) );
+    } else {
+      return "<figure data-src=\"" URL(href) "\">" \
+               "<img src=\"" URL(href) "\" alt=\"" HTML(text) "\">" \
+             "</figure>\n\n" \
+             _block( substr( block, len + 1) );
+    }
+
+  # reference style images (block)
+  } else if ( match(line, /^!\[([^]]*)\] ?\[([^]]*)\](\n|$)/ ) ) {
+    len = RLENGTH;
+    text = gensub(/^!\[([^\n]*)\] ?\[([^\n]*)\](\n.*)?$/, "\\1", 1, block);
+      id = gensub(/^!\[([^\n]*)\] ?\[([^\n]*)\](\n.*)?$/, "\\2", 1, block);
+    if ( ! id ) id = text;
+    if ( rl_href[id] && rl_title[id] ) {
+      return "<figure data-src=\"" URL(rl_href[id]) "\">" \
+               "<img src=\"" URL(rl_href[id]) "\" alt=\"" HTML(text) "\">" \
+               "<figcaption>" inline(rl_title[id]) "</figcaption>" \
+             "</figure>\n\n" \
+             _block( substr( block, len + 1) );
+    } else if ( rl_href[id] ) {
+      return "<figure data-src=\"" URL(rl_href[id]) "\">" \
+               "<img src=\"" URL(rl_href[id]) "\" alt=\"" HTML(text) "\">" \
+             "</figure>\n\n" \
+             _block( substr( block, len + 1) );
+    } else {
+      return "<p>" HTML(substr(block, 1, len)) "</p>\n" _block( substr(block, len + 1) );
+    }
+
+  # Macros (standalone <<macro>> calls handled as block, so they are not wrapped in paragraph)
+  } else if ( AllowMacros && match( block, /^<<(([^>]|>[^>])+)>>(\n|$)/) ) {
+    len = RLENGTH;
+    text = gensub(/^<<(([^>]|>[^>])+)>>(\n.*)?$/, "\\1", 1, block);
+    return macro(text) _block(substr(block, len + 1));
+
   # Split paragraphs
   } else if ( match( block, /(^|\n)[[:space:]]*(\n|$)/) ) {
     len = RLENGTH; st = RSTART;