distrib/share/etc/xml.awk

   1 #!/usr/local/bin/gawk -f
   2
   3 # This script creates an .xml file from a .txt file
   4 # The .txt file should come from an MS-Word .doc file
   5 # The .txt file must be saved with the 'LF only' option
   6 # (Even with this option, ^M may remain in the .txt file. In this case,
   7 # it is necessary to remove them manually).
   8 # The .doc file must be well-formed
   9
  10 BEGIN {
  11     line = 0;
  12     chapter_opened = 0;
  13     section_opened = 0;
  14     list_opened = 0;
  15     level = 0;
  16 }
  17 {
  18     filexml = "YAG_introduction.xml";
  19
  20     if (line == 0)
  21         printf ("<?xml version=\"1.0\" encoding=\"ISO-8859-1\" ?>\n") > filexml;
  22
  23     #chapter or section
  24     if ((int ($1) >= 1) && (int ($1) <= 9)) {
  25
  26         #level
  27         level = split ($1, word, ".");
  28
  29         #chapter
  30         if (level == 1) {
  31             while (section_opened > 0) {
  32                 printf ("</section>\n\n") > filexml;
  33                 section_opened--;
  34             }
  35             if (chapter_opened == 1) {
  36                 printf ("</chapter>\n\n\n") > filexml;
  37                 chapter_opened = 0;
  38             }
  39             printf ("<chapter>\n") > filexml;
  40             chapter_opened = 1;
  41         }
  42
  43         #section
  44         if (list_opened == 1) {
  45             printf ("</list>\n") > filexml;
  46             list_opened = 0;
  47         }
  48         while (section_opened > level - 1) {
  49             printf ("</section>\n\n") > filexml;
  50             section_opened--;
  51         }
  52         printf ("<section niv='%d'><title>", level) > filexml;
  53         for (i = 2; i < NF; i++) printf ("%s ", $i) > filexml;
  54         printf ("%s</title>\n", $NF) > filexml;
  55         section_opened++;
  56         }
  57
  58     #body
  59     else {
  60         if (length ($0) > 0) {
  61
  62             #list
  63             if ($1 == "*") {
  64                 if (list_opened == 0) {
  65                     printf ("<list>\n") > filexml;
  66                     list_opened = 1;
  67                 }
  68                 printf ("<item>", level) > filexml;
  69                 for (i = 2; i < NF; i++) printf ("%s ", $i) > filexml;
  70                 printf ("%s</item>\n", $NF) > filexml;
  71             }
  72
  73             #other
  74             else {
  75                 if (list_opened == 1) {
  76                     printf ("</list>\n") > filexml;
  77                     list_opened = 0;
  78                 }
  79                 printf ("<p>%s</p>\n", $0) > filexml;
  80             }
  81         }
  82     }
  83
  84     line++;
  85 }
  86
  87
  88 END {
  89     while (section_opened > 0) {
  90         printf ("</section>\n\n") > filexml;
  91         section_opened--;
  92     }
  93     if (chapter_opened == 1) {
  94         printf ("</chapter>\n\n\n") > filexml;
  95         chapter_opened = 0;
  96     }
  97 }