From fc47d5d7118fe570dad487482789e7447004f4f9 Mon Sep 17 00:00:00 2001 From: =?utf8?q?Paul=20H=C3=A4nsch?= Date: Wed, 30 Aug 2023 00:15:54 +0200 Subject: [PATCH] improved lists --- markdown.awk | 181 ++++++++++++++++++++++++--------------------------- 1 file changed, 85 insertions(+), 96 deletions(-) diff --git a/markdown.awk b/markdown.awk index ae1cb93..949ea93 100755 --- 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\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]+.*$/, "", 1, list) ); - - gsub("(^|\n) {0," indent "}", "\n", list); - return "\n
    \n" _list( substr(list, 2) ) "
\n" _block( block ); - - # Definition list - } else if (match( block, "^(([ \t]*\n)*[^:\n \t][^\n]+\n" \ - "([ \t]*\n)* ? ? ?:[ \t][^\n]+(\n|$)" \ - "(([ \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
\n" _dlist( list ) "
\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,})($|\n)/) ) { + len = RLENGTH; st = RSTART; + return _block(substr(block, 1, st - 1)) "
\n" _block(substr(block, st + len)); + + # Definition list + } else if (match( block, "^(([ \t]*\n)*[^:\n \t][^\n]+\n" \ + "([ \t]*\n)* ? ? ?:[ \t][^\n]+(\n|$)" \ + "(([ \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
\n" _dlist( list ) "
\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 ) "\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 ) "
    \n" _list( list, "([0-9]+\\.|#\\.)" ) "
\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)) "
\n" _block(substr(block, st + len)); - # Plain paragraph } else { return "

" inline(block) "

\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, /^\[ \]/ ) ? "
  • " : \ + match( text, /^\[-\]/ ) ? "
  • " : \ + match( text, /^\[\/\]/ ) ? "
  • " : \ + match( text, /^\[\?\]/ ) ? "
  • " : \ + match( text, /^\[[xX]\]/) ? "
  • " : "
  • "; + sub(/^\[[- /xX]\]/, "", text); + + text = _block( text ); + if (match( text, "^

    (]|\n$" )) + gsub( "(^

    |

    \n$)", "", text); + + return task text "
    • \n" _list(block, mark); +} + function _dlist (block, LOCAL, len, st, text, indent, p) { if (match( block, "^([ \t]*\n)*[^:\n \t][^\n]+\n" )) { len = RLENGTH; text = substr(block, 1, len); @@ -743,66 +781,17 @@ function _dlist (block, LOCAL, len, st, text, indent, p) { "|([ \t]*\n)+( ? ? ?\t| +)[^\n]+(\n|$))*" \ )) { 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*

    |

    \n*$)", "", text); - return "
    " text "
    \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, "^

    (]|\n$" )) + gsub( "(^

    |

    \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 ); - sub( /<\/p>\n/, "", p ); + return "

    " text "
    \n" _dlist( substr(block, len + 1) ); } - sub( /\n$/, "", p ); - - # Task List (pandoc, custom) - if ( p ~ /^\[ \].*/ ) { return "
  • " \ - substr(p, 4) "
    • \n" _list( block, last ); - } else if ( p ~ /^\[-\].*/ ) { return "
  • " \ - substr(p, 4) "
    • \n" _list( block, last ); - } else if ( p ~ /^\[\?\].*/ ) { return "
  • " \ - substr(p, 4) "
    • \n" _list( block, last ); - } else if ( p ~ /^\[\/\].*/ ) { return "
  • " \ - substr(p, 4) "
    • \n" _list( block, last ); - } else if ( p ~ /^\[[xX]\].*/ ) { return "
  • " \ - substr(p, 4) "
    • \n" _list( block, last ); - } else if ( p ~ /^

    \[ \].*/ ) { return "

  • " \ - substr(p, 7) "

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

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

  • " \ - substr(p, 7) "

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

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

  • " \ - substr(p, 7) "

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

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

  • " \ - substr(p, 7) "

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

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

  • " \ - substr(p, 7) "

    • \n" _list( block, last ); - } else { return "
  • " p "
    • \n" _list( block, last ); } } BEGIN { -- 2.39.2