X-Git-Url: https://git.plutz.net/?p=cgilite;a=blobdiff_plain;f=markdown.awk;h=361e6005c3fc42fa5e1f81859d69ec59d0c0a138;hp=249f5074b213b68511f94009b5d708db99b0b51a;hb=HEAD;hpb=1e12e80fbfed62596f992d0939418059174c4928

diff --git a/markdown.awk b/markdown.awk
index 249f507..34879d2 100755
--- a/markdown.awk
+++ b/markdown.awk
@@ -158,18 +158,18 @@ function inline( line, LOCAL, len, text, code, href, guard ) {
     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) );
+    return "<a href=\"" HTML(href) "\">" HTML(text) "</a>" inline( substr( line, len + 1) );
 
   #  quick links ("automatic links" in md doc)
   } else if ( match( line, /^<[a-zA-Z]+:\/\/([-\.[:alnum:]]+)(:[0-9]*)?(\/[^>]*)?>/ ) ) {
     len = RLENGTH;
-    href = URL( substr( line, 2, len - 2) );
+    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 = URL( substr( line, 2, len - 2) );
+    href = HTML( substr( line, 2, len - 2) );
     return "<a href=\"mailto:" href "\">" href "</a>" inline( substr( line, len + 1) );
 
   # Verbatim inline HTML
@@ -178,12 +178,12 @@ function inline( line, LOCAL, len, text, code, href, guard ) {
     return substr( line, 1, len) inline(substr(line, len + 1));
 
   # inline links
-  } else if ( match(line, "^" lii "\\([\\n\\t ]*" lid "([\\n\\t ]+" lit ")?[\\n\\t ]*\\)") ) {
+  } else if ( match(line, "^" lii "\\([\n\t ]*" lid "([\n\t ]+" lit ")?[\n\t ]*\\)") ) {
     len = RLENGTH;
     text = href = title = substr( line, 1, len);
-    sub("^\\[", "", text); sub("\\]\\([\\n\\t ]*" lid "([\\n\\t ]+" lit ")?[\\n\\t ]*\\)$", "", text);
-    sub("^" lii "\\([\\n\\t ]*", "", href); sub("([\\n\\t ]+" lit ")?[\\n\\t ]*\\)$", "", href);
-    sub("^" lii "\\([\\n\\t ]*" lid, "", title); sub("[\\n\\t ]*\\)$", "", title); sub("^[\\n\\t ]+", "", title);
+    sub("^\\[", "", text); sub("\\]\\([\n\t ]*" lid "([\n\t ]+" lit ")?[\n\t ]*\\)$", "", text);
+    sub("^" lii "\\([\n\t ]*", "", href); sub("([\n\t ]+" lit ")?[\n\t ]*\\)$", "", href);
+    sub("^" lii "\\([\n\t ]*" lid, "", title); sub("[\n\t ]*\\)$", "", title); sub("^[\n\t ]+", "", title);
 
     if ( match(href, /^<.*>$/) ) { sub(/^</, "", href); sub(/>$/, "", href); }
          if ( match(title, /^".*"$/) ) { sub(/^"/, "", title); sub(/"$/, "", title); }
@@ -192,7 +192,7 @@ function inline( line, LOCAL, len, text, code, href, guard ) {
 
     gsub(/\\/, "", href); gsub(/\\/, "", title); gsub(/[\n\t]+/, " ", title);
 
-    return "<a href=\"" URL(href) "\"" (title?" title=\"" HTML(title) "\"":"") ">" \
+    return "<a href=\"" HTML(href) "\"" (title?" title=\"" HTML(title) "\"":"") ">" \
            inline( text ) "</a>" inline( substr( line, len + 1) );
 
   # reference style links
@@ -202,28 +202,28 @@ function inline( line, LOCAL, len, text, 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 "<a href=\"" URL(rl_href[id]) "\" title=\"" HTML(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=\"" URL(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 "" HTML(substr(line, 1, len)) inline( substr(line, len + 1) );
     }
 
   # inline images
-  } else if ( match(line, "^!" lix "\\([\\n\\t ]*" lid "([\\n\\t ]+" lit ")?[\\n\\t ]*\\)(\\{[a-zA-Z \\t-]*\\})?") ) {
+  } else if ( match(line, "^!" lix "\\([\n\t ]*" lid "([\n\t ]+" lit ")?[\n\t ]*\\)(\\{[a-zA-Z \t-]*\\})?") ) {
     len = RLENGTH; text = href = title = attrib = substr( line, 1, len);
 
     sub("^!\\[", "", text);
-    sub("\\]\\([\\n\\t ]*" lid "([\\n\\t ]+" lit ")?[\\n\\t ]*\\)(\\{[a-zA-Z \\t-]*\\})?$", "", text);
+    sub("\\]\\([\n\t ]*" lid "([\n\t ]+" lit ")?[\n\t ]*\\)(\\{[a-zA-Z \t-]*\\})?$", "", text);
 
-    sub("^!" lix "\\([\\n\\t ]*", "", href);
-    sub("([\\n\\t ]+" lit ")?[\\n\\t ]*\\)(\\{[a-zA-Z \\t-]*\\})?$", "", href);
+    sub("^!" lix "\\([\n\t ]*", "", href);
+    sub("([\n\t ]+" lit ")?[\n\t ]*\\)(\\{[a-zA-Z \t-]*\\})?$", "", href);
 
-    sub("^!" lix "\\([\\n\\t ]*" lid, "", title);
-    sub("[\\n\\t ]*\\)(\\{[a-zA-Z \\t-]*\\})?$", "", title);
-    sub("^[\\n\\t ]+", "", title);
+    sub("^!" lix "\\([\n\t ]*" lid, "", title);
+    sub("[\n\t ]*\\)(\\{[a-zA-Z \t-]*\\})?$", "", title);
+    sub("^[\n\t ]+", "", title);
 
-    sub("^!" lix "\\([\\n\\t ]*" lid "([\\n\\t ]+" lit ")?[\\n\\t ]*\\)", "", attrib);
+    sub("^!" lix "\\([\n\t ]*" lid "([\n\t ]+" lit ")?[\n\t ]*\\)", "", attrib);
     sub(/^\{[ \t]*/, "", attrib); sub(/[ \t]*\}$/, "", attrib); gsub(/[ \t]+/, " ", attrib);
 
     if ( match(href, /^<.*>$/) ) { sub(/^</, "", href); sub(/>$/, "", href); }
@@ -234,7 +234,7 @@ function inline( line, LOCAL, len, text, code, href, guard ) {
     gsub(/^[\t ]+$/, "", text); gsub(/\\/, "", href);
     gsub(/\\/, "", title); gsub(/[\n\t]+/, " ", title);
 
-    return "<img src=\"" URL(href, 1) "\" alt=\"" HTML(text?text:title?title:href) "\"" \
+    return "<img src=\"" HTML(href, 1) "\" alt=\"" HTML(text?text:title?title:href) "\"" \
            (title?" title=\"" HTML(title) "\"":"") (attrib?" class=\"" HTML(attrib) "\"":"") \
            ">" inline( substr( line, len + 1) );
 
@@ -245,10 +245,10 @@ function inline( line, LOCAL, len, text, 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], 1) "\" alt=\"" HTML(text) "\" title=\"" HTML(rl_title[id]) "\">" \
+      return "<img src=\"" HTML(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], 1) "\" alt=\"" HTML(text) "\">" \
+      return "<img src=\"" HTML(rl_href[id], 1) "\" alt=\"" HTML(text) "\">" \
              inline( substr( line, len + 1) );
     } else {
       return "" HTML(substr(line, 1, len)) inline( substr(line, len + 1) );
@@ -391,7 +391,7 @@ function _block( block, LOCAL, st, len, text, title, attrib, href, guard, code,
 
   # 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|$)|[ \t]+[^\n[:space:]][^\n]*(\n|$))*/) ) {
+  } 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) );
  
@@ -639,21 +639,21 @@ function _block( block, LOCAL, st, len, text, title, attrib, href, guard, code,
     return headline( n, text, 0 ) _block( substr( block, len + 1) );
 
   # block images (wrapped in <figure>)
-  } else if ( match(block, "^!" lix "\\([\\n\\t ]*" lid "([\\n\\t ]+" lit ")?[\\n\\t ]*\\)(\\{[a-zA-Z \\t-]*\\})?(\\n|$)") ) {
+  } else if ( match(block, "^!" lix "\\([\n\t ]*" lid "([\n\t ]+" lit ")?[\n\t ]*\\)(\\{[a-zA-Z \t-]*\\})?(\n|$)") ) {
     len = RLENGTH; text = href = title = attrib = substr( block, 1, len);
 
     sub("^!\\[", "", text);
-    sub("\\]\\([\\n\\t ]*" lid "([\\n\\t ]+" lit ")?[\\n\\t ]*\\)(\\{[a-zA-Z \\t-]*\\})?(\\n.*)?$", "", text);
+    sub("\\]\\([\n\t ]*" lid "([\n\t ]+" lit ")?[\n\t ]*\\)(\\{[a-zA-Z \t-]*\\})?(\n.*)?$", "", text);
 
-    sub("^!" lix "\\([\\n\\t ]*", "", href);
-    sub("([\\n\\t ]+" lit ")?[\\n\\t ]*\\)(\\{[a-zA-Z \\t-]*\\})?(\\n.*)?$", "", href);
+    sub("^!" lix "\\([\n\t ]*", "", href);
+    sub("([\n\t ]+" lit ")?[\n\t ]*\\)(\\{[a-zA-Z \t-]*\\})?(\n.*)?$", "", href);
 
-    sub("^!" lix "\\([\\n\\t ]*" lid, "", title);
-    sub("[\\n\\t ]*\\)(\\{[a-zA-Z \\t-]*\\})?(\\n.*)?$", "", title);
-    sub("^[\\n\\t ]+", "", title);
+    sub("^!" lix "\\([\n\t ]*" lid, "", title);
+    sub("[\n\t ]*\\)(\\{[a-zA-Z \t-]*\\})?(\n.*)?$", "", title);
+    sub("^[\n\t ]+", "", title);
 
-    sub("^!" lix "\\([\\n\\t ]*" lid "([\\n\\t ]+" lit ")?[\\n\\t ]*\\)", "", attrib);
-    sub("(\\n.*)?$", "", attrib);
+    sub("^!" lix "\\([\n\t ]*" lid "([\n\t ]+" lit ")?[\n\t ]*\\)", "", attrib);
+    sub("(\n.*)?$", "", attrib);
     sub(/^\{[ \t]*/, "", attrib); sub(/[ \t]*\}$/, "", attrib); gsub(/[ \t]+/, " ", attrib);
 
     if ( match(href, /^<.*>$/) ) { sub(/^</, "", href); sub(/>$/, "", href); }
@@ -663,8 +663,8 @@ function _block( block, LOCAL, st, len, text, title, attrib, href, guard, code,
 
     gsub(/^[\t ]+$/, "", text); gsub(/\\/, "", href);
 
-    return "<figure data-src=\"" URL(href, 1) "\"" (attrib?" class=\"" HTML(attrib) "\"":"") ">" \
-           "<img src=\"" URL(href, 1) "\" alt=\"" HTML(text?text:title?title:href) "\"" \
+    return "<figure data-src=\"" HTML(href, 1) "\"" (attrib?" class=\"" HTML(attrib) "\"":"") ">" \
+           "<img src=\"" HTML(href, 1) "\" alt=\"" HTML(text?text:title?title:href) "\"" \
            (attrib?" class=\"" HTML(attrib) "\"":"") ">" \
            (title?"<figcaption>" inline(title) "</figcaption>":"") \
            "</figure>\n\n" \
@@ -677,14 +677,14 @@ function _block( block, LOCAL, st, len, text, title, attrib, href, guard, code,
       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], 1) "\">" \
-               "<img src=\"" URL(rl_href[id], 1) "\" alt=\"" HTML(text) "\">" \
+      return "<figure data-src=\"" HTML(rl_href[id], 1) "\">" \
+               "<img src=\"" HTML(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], 1) "\">" \
-               "<img src=\"" URL(rl_href[id], 1) "\" alt=\"" HTML(text) "\">" \
+      return "<figure data-src=\"" HTML(rl_href[id], 1) "\">" \
+               "<img src=\"" HTML(rl_href[id], 1) "\" alt=\"" HTML(text) "\">" \
              "</figure>\n\n" \
              _block( substr( block, len + 1) );
     } else {
@@ -834,11 +834,11 @@ BEGIN {
   iea =    "\\*([^\\*[:space:]]|[^\\*[:space:]]" na "[^\\*[:space:]])\\*"     # inner <em> (asterisk)
   isa = "\\*\\*([^\\*[:space:]]|[^\\*[:space:]]" na "[^\\*[:space:]])\\*\\*"  # inner <strong> (asterisk)
 
-  lix="\\[(\\\\[^\\n]|[^]\\n\\\\[])*\\]"  # link text
-  lid="(<(\\\\[^\\n]|[^\\n<>\\\\])*>|([^<\\n\\t ()\\\\]|\\\\[^\\n])(\\\\[\\n]|[^\\n\\t ()\\\\])*)"  # link dest
-  lit="(\"(\\\\.|[^\"\\\\])*\"|'(\\\\.|[^'\\\\])*'|\\((\\\\.|[^()\\\\])*\\))"  # link text
+  lix="\\[(\\\\[^\n]|[^]\n\\\\[])*\\]"  # link text
+  lid="(<(\\\\[^\n]|[^\n<>\\\\])*>|(\\\\.|[^()\"'\\\\])+|([^<\n\t ()\\\\]|\\\\[^\n])(\\\\[\n]|[^\n\t \\(\\)\\\\])*)"  # link dest
+  lit="(\"(\\\\.|[^\"\\\\])*\"|'(\\\\.|[^'\\\\])*'|\\((\\\\.|[^\\(\\)\\\\])*\\))"  # link text
   # link text with image def
-  lii="\\[(\\\\[^\\n]|[^]\\n\\\\[])*(!" lix "\\([\\n\\t ]*" lid "([\\n\\t ]+" lit ")?[\\n\\t ]*\\))?(\\\\[^\\n]|[^]\\n\\\\[])*\\]"
+  lii="\\[(\\\\[^\n]|[^]\n\\\\[])*(!" lix "\\([\n\t ]*" lid "([\n\t ]+" lit ")?[\n\t ]*\\))?(\\\\[^\n]|[^]\n\\\\[])*\\]"
 
   # Buffering of full file ist necessary, e.g. to find reference links
   while (getline) { file = file $0 "\n"; }