ann2html.sh

   1 #!/bin/sh
   2
   3 # todo: <sub>opt</sub>
   4
   5 export LC_ALL=C
   6 awk '
   7 BEGIN {
   8         noteid = 1
   9         sid = 1
  10         ss[sid] = "<pre>"
  11 }
  12
  13 {
  14         gsub(/\&/, "\\&amp;")
  15         gsub(/</, "\\&lt;")
  16         gsub(/>/, "\\&gt;")
  17 }
  18
  19 !title && /^[^@]/ {
  20         title = $0
  21         gsub(/  +/, "  ", title)
  22         gsub(/Committee Draft --/, "", title)
  23 }
  24
  25 /^@sect Contents/ {
  26         ss[sid] = ss[sid] "</pre>\n"
  27         seencontents = 1
  28         level = 0
  29 }
  30
  31 seencontents && !seenfore && /^[^@]/ {
  32         id = $1
  33         if (id ~ /Annex/)
  34                 id = $2
  35         sub(/\.$/, "", id)
  36
  37         s = $0
  38         if (!sub(/ +\. .*/, "", s)) {
  39                 getline
  40                 sub(/^ */, " ")
  41                 s = s $0
  42                 sub(/ +\. .*/, "", s)
  43         }
  44
  45         if (match(s, /&lt;[a-zA-Z0-9_]*\.h&gt;/)) {
  46                 h = substr($0,RSTART,RLENGTH)
  47                 if (!(h in header))
  48                         header[h] = id
  49         }
  50
  51         s = "<a href=\"#" id "\">" s "</a>\n"
  52
  53         s = "<li>" s
  54         n = split(id, a, /\./)
  55         while (n > level) {
  56                 s = "<ul>\n" s
  57                 level++
  58         }
  59         while (n < level) {
  60                 s = "</ul>\n" s
  61                 level--
  62         }
  63         ss[sid] = ss[sid] s
  64         next
  65 }
  66
  67 /^@sect Foreword/ {
  68         while (level--)
  69                 ss[sid] = ss[sid] "</ul>\n"
  70         seenfore = 1
  71 }
  72
  73 /^@sect Index/ {
  74         seenindex = 1
  75 }
  76
  77 /^@title/ {
  78         if (!seencontents) {
  79                 ss[sid] = ss[sid] "</pre>\n"
  80         }
  81         sid++
  82         getline
  83         ss[sid] = ss[sid] "<h1>" $0 "</h1>\n"
  84         if (!seencontents) {
  85                 ss[sid] = ss[sid] "<pre>\n"
  86         }
  87         next
  88 }
  89
  90 /^@sect 3\./ {
  91         markdef = 1
  92 }
  93
  94 /^@sect/ {
  95         sid++
  96         slevel = split($2,a,/\./)+1
  97         if (slevel > 5)
  98                 slevel = 5
  99         sect = $2
 100         getline
 101         # todo hX, back to top
 102         ss[sid] = sprintf("<h%s><a name=\"%s\" href=\"#%s\">%s</a></h%s>\n", slevel, sect, sect, $0, slevel)
 103         if ($0 == "Index")
 104                 ss[sid] = ss[sid] "<pre>\n"
 105         next
 106 }
 107
 108 /^@ul/ {
 109         ss[sid] = ss[sid] "<ul>\n"
 110         next
 111 }
 112 /^@end ul/ {
 113         ss[sid] = ss[sid] "</ul>\n"
 114         next
 115 }
 116 /^@ol/ {
 117         ss[sid] = ss[sid] "<ol>\n"
 118         next
 119 }
 120 /^@end ol/ {
 121         ss[sid] = ss[sid] "</ol>\n"
 122         next
 123 }
 124
 125 /^@li/ {
 126         ss[sid] = ss[sid] "<li>"
 127         next
 128 }
 129
 130 /^@pre/ {
 131         pre = "<pre>"
 132         next
 133 }
 134
 135 /^@end pre/ {
 136         if (!pre)
 137                 next
 138         pre = pre "</pre>\n"
 139         if (nn)
 140                 note[nn] = note[nn] "\n" pre
 141         else
 142                 ss[sid] = ss[sid] pre
 143         pre = ""
 144         next
 145 }
 146
 147 /^@note/ {
 148         nn = $2+0
 149         note[nn] = ""
 150         next
 151 }
 152
 153 /^@page/ {
 154         nn = 0
 155         p = $2
 156         getline
 157         ss[sid] = ss[sid] "<!--page " p " -->\n"
 158         next
 159 }
 160
 161 /^@para/ {
 162         ss[sid] = ss[sid] "<p><!--para " $2 " -->\n"
 163         next
 164 }
 165
 166 /^ ?(Syntax|Semantics|Description|Constraints|Synopsis|Returns|Recommended practice|Implementation limits|Environmental limits)$/ {
 167         ss[sid] = ss[sid] "<h6>" $0 "</h6>\n"
 168         next
 169 }
 170
 171 !seenfore {
 172         ss[sid] = ss[sid] $0 "\n"
 173         next
 174 }
 175
 176 {
 177         s = $0
 178         p = ""
 179         if (seenindex)
 180                 r = " [A-Z1-9][0-9.]*"
 181         else
 182                 r = "[ ([][A-Z1-9]\\.[0-9.]*[0-9]"
 183         # hack
 184         s = " " s
 185         while (match(s, r)) {
 186                 p = p substr(s,1,RSTART)
 187                 m = substr(s,RSTART+1,RLENGTH-1)
 188                 if (m ~ /\.0$/ || m ~ /[4-9][0-9]/ || m ~ /[0-3][0-9][0-9]/ ||
 189                     substr(s,RSTART+RLENGTH,1) ~ /[a-zA-Z_\-]/)
 190                         p = p m
 191                 else
 192                         p = p "<a href=\"#" m "\">" m "</a>"
 193                 s = substr(s,RSTART+RLENGTH)
 194         }
 195         s = p s
 196         p = ""
 197         while (match(s, /[Aa]nnex [A-Z]/)) {
 198                 p = p substr(s,1,RSTART-1)
 199                 m = substr(s,RSTART,RLENGTH)
 200                 p = p "<a href=\"#" substr(m,RLENGTH,1) "\">" m "</a>"
 201                 s = substr(s,RSTART+RLENGTH)
 202         }
 203         s = p s
 204         p = ""
 205         while (match(s, /&lt;[a-zA-Z0-9_]*\.h&gt;/)) {
 206                 p = p substr(s,1,RSTART-1)
 207                 m = substr(s,RSTART,RLENGTH)
 208                 if (m in header)
 209                         p = p "<a href=\"#" header[m] "\">" m "</a>"
 210                 else
 211                         p = p m
 212                 s = substr(s,RSTART+RLENGTH)
 213         }
 214         s = p s
 215         p = ""
 216         for (;;) {
 217                 while (match(s, noteid-1 "\\)")) {
 218                         p = p substr(s,1,RSTART-1)
 219                         p = p "<sup><a href=\"#note" noteid-1 "\"><b>" noteid-1 ")</b></a></sup>"
 220                         s = substr(s,RSTART+RLENGTH)
 221                 }
 222                 if (!match(s, noteid "\\)"))
 223                         break
 224                 if (noteid==1 && s !~ /\.1\)/)
 225                         break
 226                 p = p substr(s,1,RSTART-1)
 227                 p = p "<sup><a href=\"#note" noteid "\"><b>" noteid ")</b></a></sup>"
 228                 snote[sid] = snote[sid] " " noteid
 229                 noteid++
 230                 s = substr(s,RSTART+RLENGTH)
 231         }
 232         s = p s
 233         sub(/^ *Forward references/, "<p><b>&</b>", s)
 234         if (markdef) {
 235                 s = "<b>" s "</b><br>"
 236                 markdef = 0
 237         }
 238         if (pre)
 239                 pre = pre "\n" s
 240         else if (nn)
 241                 note[nn] = note[nn] s "\n"
 242         else
 243                 ss[sid] = ss[sid] s "\n"
 244 }
 245
 246 END {
 247         ss[sid] = ss[sid] "</pre>"
 248
 249         print "<html><head><title>" title "</title></head><body>"
 250
 251         for (i = 1; i <= sid; i++) {
 252                 print ss[i]
 253                 n = split(snote[i],a)
 254                 if (n > 0) {
 255                         s = "<h6>footnotes</h6>\n"
 256                         for (j = 1; j <= n; j++) {
 257                                 s = s "<p><small><a name=\"note" a[j] "\" href=\"#note" a[j] "\">" a[j] ")</a>" note[a[j]+0] "</small>\n"
 258                         }
 259                         print s
 260                 }
 261         }
 262
 263         print "</body></html>"
 264 }'