/extras/unhtml.vim

http://txt2tags.googlecode.com/ · Vim Script · 95 lines · 54 code · 7 blank · 34 comment · 0 complexity · 5aefa158d798e1927c831221845785bf MD5 · raw file 

    

  1. " unhtml.vim - by Aurelio Jargas
    2. 

  2. " - Converts HTML tags into txt2tags marks
    3. 

  3. " - Part of the txt2tags <http://txt2tags.org> software
    4. 

  4. "
    5. 

  5. " INSTRUCTIONS
    6. 

  6. "   1. Open the HTML file on Vim and execute
    7. 

  7. "        :so /path/to/unhtml.vim
    8. 

  8. "
    9. 

  9. "   2. A new <yourfile>.html.t2t will be saved.
    10. 

  10. "
    11. 

  11. "   3. Check the new .t2t file and correct by hand what has left.
    12. 

  12. "
    13. 

    14. 

  14. """ [ preparing ]
    15. 

  15. " ignore case
    16. 

  16. set ic
    17. 

  17. " join multiline tags
    18. 

  18. g/<\s*\([ap]\|img\)\s*$/join
    19. 

  19. g/<\s*a\s[^>]*>[^<]*$/join
    20. 

  20. 

  21. 

  22. """ [ do it! ]
    23. 

  23. " link
    24. 

  24. %s,<\s*a\s[^>]*href="\(.\{-}\)"[^>]*>\(.\{-}\)<\/a>,[\2 \1],ge
    25. 

  25. %s,<\s*a\s[^>]*href=\([^ >]\+\)[^>]*>\(.\{-}\)<\/a>,[\2 \1],ge
    26. 

  26. " images
    27. 

  27. %s,<\s*img\s[^>]*src="\(.\{-}\)"[^>]*>,[\1],ge
    28. 

  28. %s,<\s*img\s[^>]*src=\([^ >]\+\)[^>]*>,[\1],ge
    29. 

  29. " anchor
    30. 

  30. %s,^<\s*a\s\+name=.\{-}>\(.*\)<\/a>,== \1 ==,ge
    31. 

  31. " comments
    32. 

  32. %s,\s*<!--\(.*\)-->,\% \1,ge
    33. 

  33. 

  34. /<!--/,/-->/s,^,\% ,e
    35. 

  35. " paragraph
    36. 

  36. %s,<\s*p\(\s[^>]*\)\=\s*>,
    37. 

  37. ,ge
    38. 

  38. " bar
    39. 

  39. %s,<\s*hr[^>]*>,-------------------------------------------------,ge
    40. 

  40. " title
    41. 

  41. %s,</\=\s*h1\s*>,=,ge
    42. 

  42. %s,</\=\s*h2\s*>,==,ge
    43. 

  43. %s,</\=\s*h3\s*>,===,ge
    44. 

  44. %s,</\=\s*h4\s*>,====,ge
    45. 

  45. %s,</\=\s*h5\s*>,=====,ge
    46. 

  46. %s,</\=\s*h6\s*>,=====,ge
    47. 

  47. " beautifiers
    48. 

  48. %s,</\=\s*code\s*>,``,ge
    49. 

  49. %s,</\=\s*\(b\|strong\)\s*>,**,ge
    50. 

  50. %s,</\=\s*\(i\|em\)\s*>,//,ge
    51. 

  51. %s,</\=\s*u\s*>,__,ge
    52. 

  52. %s,</\=\s*s\s*>,--,ge
    53. 

  53. " pre
    54. 

  54. %s,</\=\s*pre\s*>,
    55. 

  55. ```
    56. 

  56. ,ge
    57. 

  57. " bullet/numbered list
    58. 

  58. %s,<\s*li\s*>,- ,ge
    59. 

  59. %s,</\s*li\s*>,,ge
    60. 

  60. %s,<\s*[uo]l\s*>,,ge
    61. 

  61. %s,</\s*[uo]l\s*>,
    62. 

    63. 

  63. ,ge
    64. 

  64. " definition list
    65. 

  65. %s,<\s*dl\s*>,,ge
    66. 

  66. %s,</\s*dl\s*>,
    67. 

    68. 

  68. ,ge
    69. 

  69. %s,<\s*dt\s*>,: ,ge
    70. 

  70. %s,</\s*dt\s*>,
    71. 

  71. ,ge
    72. 

  72. %s,</\=\s*dd\s*>,,ge
    73. 

  73. " BR is ignored
    74. 

  74. %s,<\s*br\s*/*>,
    75. 

  75. ,ge
    76. 

  76. " trash
    77. 

  77. %s,</\s*font[^>]*\s*>,,ge
    78. 

  78. %s,</\s*p\s*>,,ge
    79. 

  79. %s,</\s*a\s*>,,ge
    80. 

  80. %s,</\=\s*blink\s*>,,ge
    81. 

  81. %s,<\s*a\s\+name=[^>]*>,,ge
    82. 

  82. %s,</\=\s*\(html\|body\|head\|title\)\(\s[^>]*\)\=\s*>,,ge
    83. 

  83. " mmmmm, dangerous! it removes all remaining HTML tags
    84. 

  84. "%s,<[^>]*>,,ge
    85. 

  85. " clear just-blanks lines
    86. 

  86. %s,^\s*$,,
    87. 

  87. " special entities
    88. 

  88. %s,&quot;,",ge
    89. 

  89. %s,&amp;,\&,ge
    90. 

  90. %s,&gt;,>,ge
    91. 

  91. %s,&lt;,<,ge
    92. 

  92. %s,&nbsp;, ,ge
    93. 

    94. 

  94. " save new .t2t file and turn on syntax
    95. 

  95. saveas! %.t2t | set ft=txt2tags
    96.