From 1653669a6c2a769cc541893182436fa3a603e9c4 Mon Sep 17 00:00:00 2001 From: =?utf8?q?Paul=20H=C3=A4nsch?= Date: Fri, 30 Aug 2024 00:54:16 +0200 Subject: [PATCH] compatibility changes: no gensub function --- markdown.awk | 166 +++++++++++++++++++++++++++++++-------------------- 1 file changed, 100 insertions(+), 66 deletions(-) diff --git a/markdown.awk b/markdown.awk index 7e20ebb..fcd42cf 100755 --- a/markdown.awk +++ b/markdown.awk @@ -154,9 +154,9 @@ function inline( line, LOCAL, len, text, code, href, guard ) { # 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) ); + len = RLENGTH; href = text = substr(line, 1, len); + sub(/^\[\[([^]|]+)(\|([^]]+))?\]\]/, "\\1", href ); + sub(/^\[\[([^]|]+)(\|([^]]+))?\]\]/, "\\3", text ); if ( ! text ) text = href; return "" HTML(text) "" inline( substr( line, len + 1) ); @@ -197,9 +197,11 @@ function inline( line, LOCAL, len, text, code, href, guard ) { # reference style links } 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) ); + len = RLENGTH; text = id = substr(line, 1, len); + sub(/\n.*$/, "", text); sub(/^\[/, "", text); sub(/\] ?\[([^\n]*)\].*$/, "", text); + sub(/\n.*$/, "", id); sub(/^\[([^]]+)\] ?\[/, "", id); sub(/\].*$/, "", id); + # text = gensub(/^\[([^\n]+)\] ?\[([^\n]*)\].*/, "\\1", 1, text ); + # id = gensub(/^\[([^\n]+)\] ?\[([^\n]*)\].*/, "\\2", 1, id ); if ( ! id ) id = text; if ( rl_href[id] && rl_title[id] ) { return "" inline(text) "" inline( substr( line, len + 1) ); @@ -240,9 +242,11 @@ function inline( line, LOCAL, len, text, code, href, guard ) { # reference style images } 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) ); + len = RLENGTH; text = id = substr(line, 1, len); + sub(/\n.*$/, "", text); sub(/^!\[/, "", text); sub(/\] ?\[([^\n]*)\].*$/, "", text); + sub(/\n.*$/, "", id); sub(/^!\[([^]]+)\] ?\[/, "", id); sub(/\].*$/, "", id); + # 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 "\""" \ @@ -304,7 +308,9 @@ function inline( line, LOCAL, len, text, code, href, guard ) { return "" inline( substr( line, 2, len - 2 ) ) "" inline( substr( line, len + 1 ) ); # Literal HTML entities - } else if ( match( line, /^&([a-zA-Z]{2,32}|#[0-9]{1,7}|#[xX][0-9a-fA-F]{1,6});/) ) { + # } else if ( match( line, /^&([a-zA-Z]{2,32}|#[0-9]{1,7}|#[xX][0-9a-fA-F]{1,6});/) ) { + # mawk does not support repitition ranges + } else if ( match( line, /^&([a-zA-Z][a-zA-Z][a-zA-Z]?[a-zA-Z]?[a-zA-Z]?[a-zA-Z]?[a-zA-Z]?[a-zA-Z]?[a-zA-Z]?[a-zA-Z]?[a-zA-Z]?[a-zA-Z]?[a-zA-Z]?[a-zA-Z]?[a-zA-Z]?[a-zA-Z]?[a-zA-Z]?[a-zA-Z]?[a-zA-Z]?[a-zA-Z]?[a-zA-Z]?[a-zA-Z]?[a-zA-Z]?[a-zA-Z]?[a-zA-Z]?[a-zA-Z]?[a-zA-Z]?[a-zA-Z]?[a-zA-Z]?[a-zA-Z]?[a-zA-Z]?[a-zA-Z]?|#[0-9][0-9]?[0-9]?[0-9]?[0-9]?[0-9]?[0-9]?|#[xX][0-9a-fA-F][0-9a-fA-F]?[0-9a-fA-F]?[0-9a-fA-F]?[0-9a-fA-F]?[0-9a-fA-F]?);/) ) { len = RLENGTH; return substr( line, 1, len ) inline(substr(line, len + 1)); @@ -327,7 +333,8 @@ function inline( line, LOCAL, len, text, code, href, guard ) { } function headline( hlvl, htxt, attrib, LOCAL, sec, n, HL) { - match(hstack, /([0-9]+( [0-9]+){5})$/); split( substr(hstack, RSTART), HL); + # match(hstack, /([0-9]+( [0-9]+){5})$/); split( substr(hstack, RSTART), HL); + match(hstack, /([0-9]+( [0-9]+)( [0-9]+)( [0-9]+)( [0-9]+)( [0-9]+))$/); split( substr(hstack, RSTART), HL); for ( n = hlvl; n <= 6; n++ ) { sec = sec (HL[n]?"":""); } HL[hlvl]++; for ( n = hlvl + 1; n <= 6; n++) { HL[n] = 0;} @@ -336,7 +343,8 @@ function headline( hlvl, htxt, attrib, LOCAL, sec, n, HL) { hid = hid HL[1]; for ( n = 2; n <= hlvl; n++) { hid = hid "." HL[n] ; } hid = hid ":" URL(htxt, 1); - sub(/([0-9]+( [0-9]+){5})$/, "", hstack); + # sub(/([0-9]+( [0-9]+){5})$/, "", hstack); + sub(/([0-9]+( [0-9]+)( [0-9]+)( [0-9]+)( [0-9]+)( [0-9]+))$/, "", hstack); hstack = hstack HL[1] " " HL[2] " " HL[3] " " HL[4] " " HL[5] " " HL[6]; return sec "
" \ @@ -413,15 +421,14 @@ function _block( block, LOCAL, st, len, text, title, attrib, href, guard, code, cols = 0; cnt=0; ttext = ""; # table header and alignment - split( gensub( /(^\||\|$)/, "", "g", \ - gensub( /(^|[^\\])\\\|/, "\\1\\|", "g", \ - substr(block, 1, match(block, /(\n|$)/)) \ - )), tarray, /\|/); + tmp = substr(block, 1, match(block, /(\n|$)/)); + gsub( /(^|[^\\])\\\|/, "\\1\\|", tmp ); + gsub( /(^\||\|$)/, "", tmp) + split( tmp, tarray, /\|/); block = substr(block, match(block, /(\n|$)/) + 1 ); - cols = split( \ - gensub( /(^\||\|$)/, "", "g", \ - substr(block, 1, match(block, /(\n|$)/)) \ - ), talign, /[+\|]/); + tmp = substr(block, 1, match(block, /(\n|$)/)); + gsub( /(^\||\|$)/, "", tmp ); + cols = split( tmp , talign, /[+\|]/); block = substr(block, match(block, /(\n|$)/) + 1 ); for( cnt = 1; cnt < cols; cnt++ ) { @@ -437,10 +444,10 @@ function _block( block, LOCAL, st, len, text, title, attrib, href, guard, code, ttext = ttext "\n\n" while ( match(block, "^((\\|)?([^\n]+\\|)+[^\n]+(\\|)?(\n|$))+" ) ){ - split( gensub( /(^\||\|$)/, "", "g", \ - gensub( /(^|[^\\])\\\|/, "\\1\\|", "g", \ - substr(block, 1, match(block, /(\n|$)/)) \ - )), tarray, /\|/); + tmp = substr(block, 1, match(block, /(\n|$)/)); + gsub( /(^|[^\\])\\\|/, "\\1\\|", tmp ); + gsub( /(^\||\|$)/, "", tmp ); + split( tmp, tarray, /\|/); block = substr(block, match(block, /(\n|$)/) + 1 ); ttext = ttext "" @@ -469,11 +476,14 @@ function _block( block, LOCAL, st, len, text, title, attrib, href, guard, code, cols = 0; cnt=0; ttext = ""; # Column Count - cols = split( gensub( "^(\\+(:?-+:?\\+)+)(\n.*)*$", "\\1", 1, block), tread, /\+/) - 2; + tmp = block; sub( "(\n.*)*$", "", tmp); + cols = split( tmp, tread, /\+/) - 2; # debug(" Cols: " gensub( "^(\\+(:?-+:?\\+)+)(\n.*)*$", "\\1", 1, block )); # table alignment - split( gensub( "^(.*\n)?\\+((:?=+:?\\+|(:-+|-+:|:-+:)\\+)+)(\n.*)$", "\\2", "g", block ), talign, /\+/ ); + match(block, "((:?=+:?\\+|(:-+|-+:|:-+:)\\+)+)"); + split( substr(block, RSTART, RLENGTH) , talign, /\+/ ); + # split( gensub( "^(.*\n)?\\+((:?=+:?\\+|(:-+|-+:|:-+:)\\+)+)(\n.*)$", "\\2", "g", block ), talign, /\+/ ); # debug("Align: " gensub( "^(.*\n)?\\+((:?=+:?\\+|(:-+|-+:|:-+:)\\+)+)(\n.*)$", "\\2", "g", block )); for (cnt = 1; cnt <= cols; cnt++) { @@ -492,10 +502,10 @@ function _block( block, LOCAL, st, len, text, title, attrib, href, guard, code, # table header block = substr(block, match(block, /(\n|$)/) + 1 ); while ( match(block, "^\\|([^\n]+\\|)+\n") ) { - split( gensub( /(^\||\|$)/, "", "g", \ - gensub( /(^|[^\\])\\\|/, "\\1\\|", "g", \ - substr(block, 1, match(block, /(\n|$)/)) \ - )), tread, /\|/); + tmp = substr(block, 1, match(block, /(\n|$)/)); + gsub( /\\\\/, "\\\", tmp); gsub(/\\\|/, "\\|", tmp); + gsub( /(^\||\|$)/, "", tmp ); + split(tmp, tread, /\|/); block = substr(block, match(block, /(\n|$)/) + 1 ); for (cnt = 1; cnt <= cols; cnt++) tarray[cnt] = tarray[cnt] "\n" tread[cnt]; @@ -514,10 +524,10 @@ function _block( block, LOCAL, st, len, text, title, attrib, href, guard, code, 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|$)/)) \ - )), tread, /\|/); + tmp = substr(block, 1, match(block, /(\n|$)/)); + gsub( /\\\\/, "\\\", tmp); gsub(/\\\|/, "\\|", tmp); + gsub( /(^\||\|$)/, "", tmp); + split( tmp, tread, /\|/); block = substr(block, match(block, /(\n|$)/) + 1 ); for (cnt = 1; cnt <= cols; cnt++) tarray[cnt] = tarray[cnt] "\n" tread[cnt]; @@ -542,8 +552,9 @@ function _block( block, LOCAL, st, len, text, title, attrib, href, guard, code, return "
" text "
\n" _block( substr( block, len + 1) ); # Indented Code Block - } else if ( match(block, /^( |\t)( *\t*[^ \t\n]+ *\t*)+(\n|$)(( |\t)[^\n]+(\n|$)|[ \t]*(\n|$))*/) ) { + } else if ( match(block, /^(( |\t)[^\n]*[^\n\t ][^\n]*(\n|$))(( |\t)[^\n]*(\n|$)|[\t ]*(\n|$))*/) ) { len = RLENGTH; st = RSTART; + code = substr(block, 1, len); gsub(/(^|\n)( |\t)/, "\n", code); gsub(/^\n|\n+$/, "", code); @@ -552,9 +563,10 @@ function _block( block, LOCAL, st, len, text, title, attrib, href, guard, code, # Fenced Divs (pandoc, custom) } else if ( match( block, /^(:::+)/ ) ) { - guard = substr( block, 1, RLENGTH ); - code = block; sub(/^[^\n]+\n/, "", code); - attrib = gensub(/^:::+[ \t]*\{?[ \t]*([^\}\n]*)\}?[ \t]*\n.*$/, "\\1", 1, block); + guard = substr( block, 1, RLENGTH ); attrib = code = block; + sub(/^[^\n]+\n/, "", code); + sub(/^:::+[ \t]*\{?[ \t]*/, "", attrib); sub(/\}?[ \t]*\n.*$/, "", attrib); + # attrib = gensub(/^:::+[ \t]*\{?[ \t]*([^\}\n]*)\}?[ \t]*\n.*$/, "\\1", 1, attrib); gsub(/[^a-zA-Z0-9_-]+/, " ", attrib); gsub(/(^ | $)/, "", attrib); if ( match(code, "(^|\n)" guard "+(\n|$)" ) && attrib ) { @@ -574,9 +586,10 @@ function _block( block, LOCAL, st, len, text, title, attrib, href, guard, code, # Fenced Code Block (pandoc) } else if ( match( block, /^(~~~+|```+)/ ) ) { - guard = substr( block, 1, RLENGTH ); - code = gensub(/^[^\n]+\n/, "", 1, block); - attrib = gensub(/^(~~~+|```+)[ \t]*\{?[ \t]*([^\}\n]*)\}?[ \t]*\n.*$/, "\\2", 1, block); + guard = substr( block, 1, RLENGTH ); attrib = code = block; + sub(/^[^\n]+\n/, "", code); + sub(/^(~~~+|```+)[ \t]*\{?[ \t]*/, "", attrib); sub(/\}?[ \t]*\n.*$/, "", attrib); + # attrib = gensub(/^(~~~+|```+)[ \t]*\{?[ \t]*([^\}\n]*)\}?[ \t]*\n.*$/, "\\2", 1, attrib); gsub(/[^a-zA-Z0-9_-]+/, " ", attrib); gsub(/(^ | $)/, "", attrib); if ( match(code, "(^|\n)" guard "+(\n|$)" ) && attrib ) { @@ -627,22 +640,26 @@ function _block( block, LOCAL, st, len, text, title, attrib, href, guard, code, return headline(2, text, 0) _block( substr( block, len + 1) ); # Nth Order Heading H1 H2 H3 H4 H5 H6 + Attrib - } else if ( match( block, /^(#{1,6})[ \t]*(([^ \t\n]+|[ \t]+[^ \t\n#]|[ \t]+#+[ \t]*[^ \t\n#])+)[ \t]*#*([ \t]*\{([a-zA-Z \t-]*)\})(\n|$)/ ) ) { + } else if ( match( block, /^(##?#?#?#?#?)[ \t]*(([^ \t\n]+|[ \t]+[^ \t\n#]|[ \t]+#+[ \t]*[^ \t\n#])+)[ \t]*#*([ \t]*\{([a-zA-Z \t-]*)\})(\n|$)/ ) ) { len = RLENGTH; text = attrib = substr(block, 1, len); - match(block, /^#{1,6}/); n = RLENGTH; + match(block, /^##?#?#?#?#?[^#]/); n = RLENGTH - 1; - sub(/^(#{1,6})[ \t]*/, "", text); sub(/[ \t]*#*([ \t]*\{([a-zA-Z \t-]*)\})(\n.*)?$/, "", text); - sub(/^(#{1,6})[ \t]*(([^ \t\n]+|[ \t]+[^ \t\n#]|[ \t]+#+[ \t]*[^ \t\n#])+)[ \t]*#*[ \t]*\{/, "", attrib); + # sub(/^(##?#?#?#?#?)[ \t]*/, "", text); # not working in mawk + text = substr(text, n + 1); sub(/^[ \t]*/, "", text); + sub(/[ \t]*#*([ \t]*\{([a-zA-Z \t-]*)\})(\n.*)?$/, "", text); + sub(/^(##?#?#?#?#?)[ \t]*(([^ \t\n]+|[ \t]+[^ \t\n#]|[ \t]+#+[ \t]*[^ \t\n#])+)[ \t]*#*[ \t]*\{/, "", attrib); sub(/\})(\n.*)?$/, "", attrib); gsub(/[^a-zA-Z0-9_-]+/, " ", attrib); gsub(/(^ | $)/, "", attrib); return headline( n, text, attrib ) _block( substr( block, len + 1) ); # Nth Order Heading H1 H2 H3 H4 H5 H6 - } else if ( match( block, /^(#{1,6})[ \t]*(([^ \t\n]+|[ \t]+[^ \t\n#]|[ \t]+#+[ \t]*[^ \t\n#])+)[ \t]*#*(\n|$)/ ) ) { + } else if ( match( block, /^(##?#?#?#?#?)[ \t]*(([^ \t\n]+|[ \t]+[^ \t\n#]|[ \t]+#+[ \t]*[^ \t\n#])+)[ \t]*#*(\n|$)/ ) ) { len = RLENGTH; text = substr(block, 1, len); - match(block, /^#{1,6}/); n = RLENGTH; - sub(/^(#{1,6})[ \t]*/, "", text); sub(/[ \t]*#*(\n.*)?$/, "", text); + match(block, /^##?#?#?#?#?[^#]/); n = RLENGTH - 1; + # sub(/^(##?#?#?#?#?)[ \t]+/, "", text); # not working in mawk + text = substr(text, n + 1); sub(/^[ \t]*/, "", text); + sub(/[ \t]*#*(\n.*)?$/, "", text); return headline( n, text, 0 ) _block( substr( block, len + 1) ); @@ -678,11 +695,12 @@ function _block( block, LOCAL, st, len, text, title, attrib, href, guard, code, "\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); + } else if ( match(block, /^!\[([^]]*)\] ?\[([^]]*)\](\n|$)/ ) ) { + len = RLENGTH; text = id = block; + sub(/(\n.*)?$/, "", text); sub( /^!\[/, "", text); sub(/\] ?\[([^\n]*)\]$/, "", text); + sub(/(\n.*)?$/, "", id); sub( /^!\[([^\n]*)\] ?\[/, "", id); sub(/\]$/, "", id); + # text = gensub(/^!\[([^\n]*)\] ?\[([^\n]*)\](\n.*)?$/, "\\1", 1, block); + # id = gensub(/^!\[([^\n]*)\] ?\[([^\n]*)\](\n.*)?$/, "\\2", 1, block); if ( ! id ) id = text; if ( rl_href[id] && rl_title[id] ) { return "
" \ @@ -701,8 +719,9 @@ function _block( block, LOCAL, st, len, text, title, attrib, href, guard, code, # Macros (standalone <> calls handled as block, so they are not wrapped in paragraph) } else if ( match( block, /^<<(([^>]|>[^>])+)>>(\n|$)/ ) ) { - len = RLENGTH; - text = gensub(/^<<(([^>]|>[^>])+)>>(\n.*)?$/, "\\1", 1, block); + len = RLENGTH; text = block; + sub(/^<>(\n.*)?$/, "", text); + # text = gensub(/^<<(([^>]|>[^>])+)>>(\n.*)?$/, "\\1", 1, block); return "" HTML(text) "" _block(substr(block, len + 1) ); # Definition list @@ -743,7 +762,8 @@ function _block( block, LOCAL, st, len, text, title, attrib, href, guard, code, _block( substr(block, st + len) ); # Horizontal rule - } else if ( match( block, /(^|\n) ? ? ?((\* *){3,}|(- *){3,}|(_ *){3,})($|\n)/) ) { + # } else if ( match( block, /(^|\n) ? ? ?((\* *){3,}|(- *){3,}|(_ *){3,})($|\n)/) ) { + } else if ( match( block, /(^|\n) ? ? ?((\* *)(\* *)(\* *)(\* *)*|(- *)(- *)(- *)(- *)*|(_ *)(_ *)(_ *)(_ *)*)($|\n)/) ) { len = RLENGTH; st = RSTART; return _block(substr(block, 1, st - 1)) "
\n" _block(substr(block, st + len)); @@ -753,7 +773,7 @@ function _block( block, LOCAL, st, len, text, title, attrib, href, guard, code, } } -function _startlist(block, type, mark, exclude, LOCAL, st, len, list, indent, text) { +function _startlist(block, type, mark, exclude, LOCAL, st, len, list, indent, it, text) { if (match( block, "(^|\n) ? ? ?" mark "[ \t][^\n]+(\n|$)" \ "(([ \t]*\n)* ? ? ?" mark "[ \t][^\n]+(\n|$)" \ "|([ \t]*\n)*( ? ? ?\t| +)[^\n]+(\n|$)" \ @@ -761,7 +781,9 @@ function _startlist(block, type, mark, exclude, LOCAL, st, len, list, indent, te st = RSTART; len = RLENGTH; list = substr( block, st, len); sub("^\n", "", list); match(list, "^ ? ? ?"); indent = RLENGTH; - gsub( "(^|\n) {0," indent "}", "\n", list); sub("^\n", "", list); + it = ""; while ( indent > 0 ) { it = it " ?"; indent--; } + # gsub( "(^|\n) {0," indent "}", "\n", list); sub("^\n", "", list); + gsub( "(^|\n)" it, "\n", list); sub("^\n", "", list); text = substr(block, 1, st - 1); block = substr(block, st + len); if (match(text, /\n[[:space:]]*\n/)) return 0; @@ -775,10 +797,12 @@ function _startlist(block, type, mark, exclude, LOCAL, st, len, list, indent, te } else return 0; } -function _list (block, mark, p, LOCAL, len, st, text, indent, task) { +function _list (block, mark, p, LOCAL, len, st, text, indent, it, task) { if ( match(block, "^([ \t]*\n)*$")) return; match(block, "^" mark "[ \t]"); indent = RLENGTH; + it = ""; while ( indent > 0 ) { it = it " ?"; indent--; } + sub("^" mark "[ \t]", "", block); if (match(block, /\n[ \t]*\n/)) p = 1; @@ -787,7 +811,8 @@ function _list (block, mark, p, LOCAL, len, st, text, indent, task) { st = (RLENGTH == -1) ? length(block) + 1 : RSTART; text = substr(block, 1, st); block = substr(block, st + 1); - gsub("\n {0," indent "}", "\n", text); + # gsub("\n {0," indent "}", "\n", text); + gsub("\n" it, "\n", text); task = match( text, /^\[ \]/ ) ? "
  • " : \ match( text, /^\[-\]/ ) ? "
  • " : \ @@ -803,7 +828,7 @@ function _list (block, mark, p, LOCAL, len, st, text, indent, task) { return task text "
    • \n" _list(block, mark, p); } -function _dlist (block, LOCAL, len, st, text, indent, p) { +function _dlist (block, LOCAL, len, st, text, indent, it, p) { if (match( block, "^([ \t]*\n)*[^:\n \t][^\n]+\n" )) { len = RLENGTH; text = substr(block, 1, len); gsub( "(^\n*|\n*$)", "", text ); @@ -816,8 +841,10 @@ function _dlist (block, LOCAL, len, st, text, indent, p) { len = RLENGTH; text = substr(block, 1, len); sub( "^([ \t]*\n)*", "", text); match(text, "^ ? ? ?:(\t| +)"); indent = RLENGTH; + it = ""; while ( indent > 0 ) { it = it " ?"; indent--; } sub( "^ ? ? ?:(\t| +)", "", text); - gsub( "(^|\n) {0," indent "}", "\n", text ); + # gsub( "(^|\n) {0," indent "}", "\n", text ); + gsub( "(^|\n)" it, "\n", text ); text = _nblock(text); if (match( text, "^

    (]|\n$" )) @@ -858,10 +885,17 @@ BEGIN { re_reflink = "(^|\n) ? ? ?\\[([^]\n]+)\\]: ([^ \t\n]+)(\n?[ \t]+(\"([^\"]+)\"|'([^']+)'|\\(([^)]+)\\)))?(\n|$)"; # /(^|\n) ? ? ?\[([^]\n]+)\]: ([^ \t\n]+)(\n?[ \t]+("([^"]+)"|'([^']+)'|\(([^)]+)\)))?(\n|$)/ while ( match(f, re_reflink ) ) { - rl_id = gensub( re_reflink, "\\2", 1, substr(f, RSTART, RLENGTH) ); - rl_href[rl_id] = gensub( re_reflink, "\\3", 1, substr(f, RSTART, RLENGTH) ); - rl_title[rl_id] = gensub( re_reflink, "\\5", 1, substr(f, RSTART, RLENGTH) ); - f = substr(f, RSTART + RLENGTH); + tt = th = ti = substr(f, RSTART, RLENGTH); f = substr(f, RSTART + RLENGTH); + sub("(^|\n) ? ? ?\\[", "", ti); sub("\\]: ([^ \t\n]+)(\n?[ \t]+(\"([^\"]+)\"|'([^']+)'|\\(([^)]+)\\)))?(\n.*)?$", "", ti); + sub("(^|\n) ? ? ?\\[([^]\n]+)\\]: ", "", th); sub("(\n?[ \t]+(\"([^\"]+)\"|'([^']+)'|\\(([^)]+)\\)))?(\n.*)?$", "", th); + if (match(tt, "(^|\n) ? ? ?\\[([^]\n]+)\\]: ([^ \t\n]+)(\n?[ \t]+(\"([^\"]+)\"|'([^']+)'|\\(([^)]+)\\)))(\n|$)")) { + sub("(^|\n) ? ? ?\\[([^]\n]+)\\]: ([^ \t\n]+)", "", tt); sub("^\n?[ \t]+", "", tt); sub("(\n.*)?$", "", tt); + } else { tt = ""; } + rl_id = ti; rl_href[rl_id] = th; rl_title[rl_id] = tt; + # rl_id = gensub( re_reflink, "\\2", 1, substr(f, RSTART, RLENGTH) ); + # rl_href[rl_id] = gensub( re_reflink, "\\3", 1, substr(f, RSTART, RLENGTH) ); + # rl_title[rl_id] = gensub( re_reflink, "\\5", 1, substr(f, RSTART, RLENGTH) ); + # f = substr(f, RSTART + RLENGTH); rl_title[rl_id] = substr( rl_title[rl_id], 2, length(rl_title[rl_id]) - 2 ); if ( rl_href[rl_id] ~ /<.*>/ ) rl_href[rl_id] = substr( rl_href[rl_id], 2, length(rl_href[rl_id]) - 2 ); } -- 2.39.2