X-Git-Url: http://git.plutz.net/?a=blobdiff_plain;f=markdown.awk;h=4f18b9ba499685017a6fcef4ab25c5a5a581c392;hb=a8a5ea58ca40e57165c37d62b61fdf1c60832c7d;hp=540292c5cfbedd31875114e7d9cca073f3d629ff;hpb=d1bb79cc914d707553de40a9b146265ac0f6b577;p=cgilite diff --git a/markdown.awk b/markdown.awk index 540292c..4f18b9b 100755 --- a/markdown.awk +++ b/markdown.awk @@ -42,11 +42,11 @@ # - [x] Automatic heading identifiers (custom) # - [x] Fenced code blocks (php md, pandoc) # - [x] Fenced code attributes -# - [ ] Tables +# - [/] Tables # - ? Simple table (pandoc) # - ? Multiline table (pandoc) -# - ? Grid table (pandoc) -# - ? Pipe table (php md pandoc) +# - [x] Grid table (pandoc) +# - [x] Pipe table (php md pandoc) # - [x] Line blocks (pandoc) # - [x] Task lists (pandoc, custom) # - [ ] Definition lists (php md, pandoc) @@ -62,6 +62,8 @@ # - [x] ^Superscript^ ~Subscript~ (pandoc) # - [-] Bracketed spans (pandoc) # - [-] Inline attributes (pandoc) +# - [x] Image attributes (custom, pandoc inspired, inline only) +# - [x] Wiki style links [[PageName]] / [[PageName|Link Text]] # - [-] TEX-Math (pandoc) # - ? Footnotes (php md) # - ? Abbreviations (php md) @@ -84,6 +86,20 @@ function HTML ( text ) { return text; } +function URL ( text ) { + gsub( /&/, "%26", text ); + gsub( /"/, "%22", text ); + gsub( /'/, "%27", text ); + gsub( /\?/, "%3F", text ); + gsub( /#/, "%23", text ); + gsub( /\[/, "%5B", text ); + gsub( /\]/, "%5D", text ); + gsub( / /, "%20", text ); + gsub( / /, "%09", text ); + gsub( /\\/, "%5C", text ); + return text; +} + function inline( line, LOCAL, len, code, href, guard ) { nu = "(\\\\\\\\|\\\\[^\\\\]|[^\\\\_]|_[[:alnum:]])*" # not underline (except when escaped) na = "(\\\\\\\\|\\\\[^\\\\]|[^\\\\\\*])*" # not asterisk (except when escaped) @@ -101,7 +117,7 @@ function inline( line, LOCAL, len, code, href, guard ) { # hard brakes } else if ( match(line, /^ \n/) ) { - return "
\n" inline( substr(line, RLENGTH + 1) ); + return "
\n" inline( substr(line, RLENGTH + 1) ); # ``code spans`` } else if ( match( line, /^`+/) ) { @@ -117,28 +133,40 @@ function inline( line, LOCAL, len, code, href, guard ) { return "" code "" inline( substr( line, len + 1 ) ) } + # Wiki style links + } else if ( match( line, /^\[\[([^\]\|]+)(\|([^\]]+))?\]\]/) ) { + len = RLENGTH; + href = gensub(/^\[\[([^\]\|]+)(\|([^\]]+))?\]\]/, "\\1", 1, substr(line, 1, len) ); + text = gensub(/^\[\[([^\]\|]+)(\|([^\]]+))?\]\]/, "\\3", 1, substr(line, 1, len) ); + if ( ! text ) text = href; + return "" HTML(text) "" 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 = HTML( substr( line, 2, len - 2) ); + href = URL( substr( line, 2, len - 2) ); return "" href "" 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) ); + href = URL( substr( line, 2, len - 2) ); return "" href "" 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 "" inline( text ) "" inline( substr( line, len + 1) ); + return "" inline( text ) "" inline( substr( line, len + 1) ); } else { - return "" inline( text ) "" inline( substr( line, len + 1) ); + return "" inline( text ) "" inline( substr( line, len + 1) ); } # reference style links @@ -148,23 +176,32 @@ 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 "" inline(text) "" inline( substr( line, len + 1) ); + return "" inline(text) "" inline( substr( line, len + 1) ); } else if ( rl_href[id] ) { - return "" inline(text) "" inline( substr( line, len + 1) ); + return "" inline(text) "" inline( substr( line, len + 1) ); } else { return "" HTML(substr(line, 1, len)) inline( substr(line, len + 1) ); } # inline images - } else if ( match(line, /^!\[([^]]+)\]\(([^"\)]+)([ \t]+"([^"]+)")?\)/) ) { + } else if ( match(line, /^!\[([^]]+)\]\(([^"\)]+)([ \t]+"([^"]+)")?\)(\{([a-zA-Z \t-]*)\})?/) ) { len = RLENGTH; - 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 "\""" inline( substr( line, len + 1) ); + 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 "\""" \ + inline( substr( line, len + 1) ); + } else if ( title ) { + return "\""" \ + inline( substr( line, len + 1) ); + } else if ( attrib ) { + return "\""" \ + inline( substr( line, len + 1) ); } else { - return "\""" inline( substr( line, len + 1) ); + return "\""" \ + inline( substr( line, len + 1) ); } # reference style images @@ -174,9 +211,11 @@ 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 "\""" inline( substr( line, len + 1) ); + return "\""" \ + inline( substr( line, len + 1) ); } else if ( rl_href[id] ) { - return "\""" inline( substr( line, len + 1) ); + return "\""" \ + inline( substr( line, len + 1) ); } else { return "" HTML(substr(line, 1, len)) inline( substr(line, len + 1) ); } @@ -296,6 +335,115 @@ function _block( block, LOCAL, st, len, hlvl, htxt, guard, code, indent, attrib return "
\n" _block( gensub( /(^|\n)> /, "\n", "g", substr(block, 1, st - 1) ) ) "
\n\n" \ _block( substr(block, st + len) ); + # Pipe Tables (pandoc / php md / gfm ) + } else if ( match(block, "^((\\|)?([^\n]+\\|)+[^\n]+(\\|)?)\n" \ + "((\\|)?:?(-+:?[\\|+])+:?-+:?(\\|)?)\n" \ + "((\\|)?([^\n]+\\|)+[^\n]+(\\|)?(\n|$))+" ) ) { + len = RLENGTH; st = RSTART; + #initialize empty arrays + split("", talign); split("", tarray); + cols = 0; cnt=0; ttext = ""; + + # table header and alignment + split( gensub( /(^\||\|$)/, "", "g", \ + gensub( /(^|[^\\])\\\|/, "\\1\\|", "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|$)/) + 1 ); + + for( cnt = 1; cnt < cols; cnt++ ) { + if (match(talign[cnt], /:-+:/)) talign[cnt]="center"; + else if (match(talign[cnt], /-+:/)) talign[cnt]="right"; + else if (match(talign[cnt], /:-+/)) talign[cnt]="left"; + else talign[cnt]=""; + } + + ttext = "\n" + for (cnt = 1; cnt < cols; cnt++) + ttext = ttext "" inline(tarray[cnt]) "" + ttext = ttext "\n\n" + + while ( match(block, "^((\\|)?([^\n]+\\|)+[^\n]+(\\|)?(\n|$))+" ) ){ + split( gensub( /(^\||\|$)/, "", "g", \ + gensub( /(^|[^\\])\\\|/, "\\1\\|", "g", \ + substr(block, 1, match(block, /(\n|$)/)) \ + )), tarray, /\|/); + block = substr(block, match(block, /(\n|$)/) + 1 ); + + ttext = ttext "" + for (cnt = 1; cnt < cols; cnt++) + ttext = ttext "" inline(tarray[cnt]) "" + ttext = ttext "\n" + } + return "" ttext "
\n" _block(block); + + # Grid Tables (pandoc) + } else if ( match(block, "^\\+(-+\\+)+\n" \ + "(\\|([^\n]+\\|)+\n)+" \ + "\\+(:?=+:?\\+)+\n" \ + "((\\|([^\n]+\\|)+\n)+" \ + "\\+(-+\\+)+(\n|$))+" \ + ) ) { + 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\\|", "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]; + } + + cols = split( \ + gensub( /(^\+|\+$)/, "", "g", \ + substr(block, 1, match(block, /(\n|$)/)) \ + ), talign, /\+/); + block = substr(block, match(block, /(\n|$)/) + 1 ); + + for (cnt = 1; cnt < cols; cnt++) { + if (match(talign[cnt], /:=+:/)) talign[cnt]="center"; + else if (match(talign[cnt], /=+:/)) talign[cnt]="right"; + else if (match(talign[cnt], /:=+/ )) talign[cnt]="left"; + else talign[cnt]=""; + } + + ttext = "\n" + for (cnt = 1; cnt < cols; cnt++) + ttext = ttext "" _block(tarray[cnt]) "" + ttext = ttext "\n\n" + + while ( match(block, /^((\|([^\n]+\|)+\n)+\+(-+\+)+(\n|$))+/ ) ){ + split("", tarray); + while ( match(block, /^\|([^\n]+\|)+\n/) ) { + split( gensub( /(^\||\|$)/, "", "g", \ + gensub( /(^|[^\\])\\\|/, "\\1\\|", "g", \ + substr(block, 1, match(block, /(\n|$)/)) \ + )), tread, /\|/); + block = substr(block, match(block, /(\n|$)/) + 1 ); + for (cnt = 1; cnt < cols; cnt++) + tarray[cnt] = tarray[cnt] "\n" tread[cnt]; + } + block = substr(block, match(block, /(\n|$)/) + 1 ); + + ttext = ttext "" + for (cnt = 1; cnt < cols; cnt++) + ttext = ttext "" _block(tarray[cnt]) "" + ttext = ttext "\n" + } + return "" ttext "
\n" _block(block); + # Line Blocks (pandoc) } else if ( match(block, /^\| [^\n]*(\n|$)(\| [^\n]*(\n|$)|[ \t]+[^\n[:space:]][^\n]*(\n|$))*/) ) { len = RLENGTH; st = RSTART; @@ -303,11 +451,11 @@ function _block( block, LOCAL, st, len, hlvl, htxt, guard, code, indent, attrib gsub(/\n[[:space:]]+/, " ", code); gsub(/\n\| /, "\n", code); gsub(/^\| |\n$/, "", code); - return "
" gensub(/\n/, "
\n", "g", inline( code )) "
\n" \ + return "
" gensub(/\n/, "
\n", "g", inline( code )) "
\n" \ _block( substr( block, len + 1) ); # Indented Code Block - } else if ( match(block, /^( |\t)[^\n]+(\n|$)(( |\t)[^\n]+(\n|$)|[ \t]*(\n|$))*/) ) { + } else if ( match(block, /^( |\t)( *\t*[^ \t\n]+ *\t*)+(\n|$)(( |\t)[^\n]+(\n|$)|[ \t]*(\n|$))*/) ) { len = RLENGTH; st = RSTART; code = substr(block, 1, len); gsub(/(^|\n)( |\t)/, "\n", code); @@ -379,14 +527,20 @@ function _block( block, LOCAL, st, len, hlvl, htxt, guard, code, indent, attrib } else if ( match( block, /^[^\n]+\n===+(\n|$)/ ) ) { len = RLENGTH; HL[1]++; HL[2] = 0; HL[3] = 0; HL[4] = 0; HL[5] = 0; HL[6] = 0; - return "

" inline( gensub( /\n.*$/, "", "g", block ) ) "

\n\n" \ + return "

" \ + inline( gensub( /\n.*$/, "", "g", block ) ) \ + "

\n\n" \ _block( substr( block, len + 1 ) ); # Second Order Heading } else if ( match( block, /^[^\n]+\n---+(\n|$)/ ) ) { len = RLENGTH; HL[2]++; HL[3] = 0; HL[4] = 0; HL[5] = 0; HL[6] = 0; - return "

" inline( gensub( /\n.*$/, "", "g", block ) ) "

\n\n" \ + return "

" \ + inline( gensub( /\n.*$/, "", "g", block ) ) \ + "

\n\n" \ _block( substr( block, len + 1) ); # Nth Order Heading @@ -396,7 +550,8 @@ function _block( block, LOCAL, st, len, hlvl, htxt, guard, code, indent, attrib htxt = gensub(/^#{1,6}[ \t]*(([^ \t\n]+|[ \t]+[^ \t\n#]|[ \t]+#+[^\n#])+)([ \t]*#*)(\n.*)?$/, "\\1", 1, block); HL[hlvl]++; for ( n = hlvl + 1; n < 7; n++) { HL[n] = 0;} hid = HL[1]; for ( n = 2; n <= hlvl; n++) { hid = hid "." HL[n] ; } - return "" inline( htxt ) "\n\n" \ + return "" inline( htxt ) \ + "\n\n" \ _block( substr( block, len + 1) ); # Split paragraphs @@ -408,7 +563,7 @@ function _block( block, LOCAL, st, len, hlvl, htxt, guard, code, indent, attrib # Horizontal rule } else if ( match( block, /(^|\n) ? ? ?((\* *){3,}|(- *){3,}|(_ *){3,})($|\n)/) ) { len = RLENGTH; st = RSTART; - return _block(substr(block, 1, st - 1)) "
\n" _block(substr(block, st + len)); + return _block(substr(block, 1, st - 1)) "
\n" _block(substr(block, st + len)); # Plain paragraph } else { @@ -441,25 +596,25 @@ function _list( block, last, LOCAL, p) { sub( /\n$/, "", p ); # Task List (pandoc, custom) - if ( p ~ /^\[ \].*/ ) { return "
  • " \ + if ( p ~ /^\[ \].*/ ) { return "
  • " \ substr(p, 4) "
  • \n" _list( block, last ); - } else if ( p ~ /^\[-\].*/ ) { return "
  • " \ + } else if ( p ~ /^\[-\].*/ ) { return "
  • " \ substr(p, 4) "
  • \n" _list( block, last ); - } else if ( p ~ /^\[\?\].*/ ) { return "
  • " \ + } else if ( p ~ /^\[\?\].*/ ) { return "
  • " \ substr(p, 4) "
  • \n" _list( block, last ); - } else if ( p ~ /^\[\/\].*/ ) { return "
  • " \ + } else if ( p ~ /^\[\/\].*/ ) { return "
  • " \ substr(p, 4) "
  • \n" _list( block, last ); - } else if ( p ~ /^\[[xX]\].*/ ) { return "
  • " \ + } else if ( p ~ /^\[[xX]\].*/ ) { return "
  • " \ substr(p, 4) "
  • \n" _list( block, last ); - } else if ( p ~ /^

    \[ \].*/ ) { return "

  • " \ + } else if ( p ~ /^

    \[ \].*/ ) { return "

  • " \ substr(p, 7) "

  • \n" _list( block, last ); - } else if ( p ~ /^

    \[-\].*/ ) { return "

  • " \ + } else if ( p ~ /^

    \[-\].*/ ) { return "

  • " \ substr(p, 7) "

  • \n" _list( block, last ); - } else if ( p ~ /^

    \[\?\].*/ ) { return "

  • " \ + } else if ( p ~ /^

    \[\?\].*/ ) { return "

  • " \ substr(p, 7) "

  • \n" _list( block, last ); - } else if ( p ~ /^

    \[\/\].*/ ) { return "

  • " \ + } else if ( p ~ /^

    \[\/\].*/ ) { return "

  • " \ substr(p, 7) "

  • \n" _list( block, last ); - } else if ( p ~ /^

    \[[xX]\].*/ ) { return "

  • " \ + } else if ( p ~ /^

    \[[xX]\].*/ ) { return "

  • " \ substr(p, 7) "

  • \n" _list( block, last ); } else { return "
  • " p "
  • \n" _list( block, last ); } }