annot.sh

   1 #!/bin/sh
   2
   3 export LC_ALL=C
   4 awk '
   5 function initpage() {
   6         if (innote)
   7                 endpre()
   8
   9         indent = 10
  10         para = 0
  11         innote = 0
  12         ok = 0
  13         empty = ""
  14         page = ""
  15 }
  16
  17 function endlist() {
  18         if (listindex) {
  19                 listindent = 0
  20                 listindex = 0
  21                 print "@end ol"
  22         } else if (listlevel) {
  23                 listlevel--
  24                 print "@end ul"
  25         }
  26 }
  27
  28 function donote(s) {
  29         if (match(s, /^[1-9][0-9]*\) +/)) {
  30                 # todo..
  31                 endpre()
  32                 print "@note " substr(s, 1, RLENGTH-2)
  33                 s = substr(s, RLENGTH+1)
  34                 innote = 1
  35                 noteindent = RLENGTH
  36         } else if (innote) {
  37                 if (match(s, /^ +/) && RLENGTH >= noteindent)
  38                         s = substr(s, noteindent+1)
  39                 else
  40                         innote = 0
  41         }
  42         return s
  43 }
  44
  45 function doul(s) {
  46         if (!listlevel) {
  47                 if (s ~ /^--/) {
  48                         s = dopre(s)
  49                         listlevel++
  50                         print "@ul"
  51                 } else
  52                         return s
  53         }
  54         if (listlevel == 1) {
  55                 if (s ~ /^   *o /) {
  56                         listlevel++
  57                         print "@ul"
  58                 } else if (s ~ /^--/) {
  59                         s = dopre(s)
  60                         s = substr(s , 3)
  61                         print "@li"
  62                 } else if (match(s, /^ +/) && RLENGTH > 1) {
  63                         sub(/^  /, "", s)
  64                 } else if (s !~ /^$/) {
  65                         endpre()
  66                         endlist()
  67                 }
  68         }
  69         if (listlevel == 2) {
  70                 if (s ~ /^   *o /) {
  71                         sub(/ *o /, "", s)
  72                         s = dopre(s)
  73                         print "@li"
  74                 } else if (match(s, /^ +/) && RLENGTH > 5) {
  75                         sub(/^ +/, "", s)
  76                 } else if (s !~ /^$/) {
  77                         sub(/^  /, "", s)
  78                         endlist()
  79                 }
  80         }
  81         return s
  82 }
  83
  84 function dool(s) {
  85         if (listindex == 0) {
  86                 if (s ~ /^   *1\. /)
  87                         print "@ol"
  88                 else
  89                         return s
  90         }
  91
  92         if (match(s, "^ +" (listindex+1) "\\. +")) {
  93                 listindex++
  94                 listindent = RLENGTH
  95                 s = substr(s, RLENGTH)
  96                 s = dopre(s)
  97                 print "@li " listindex
  98         } else if (match(s, /^ +/) && RLENGTH >= listindent) {
  99                 s = substr(s, listindent+1)
 100         } else if (s !~ /^$/) {
 101                 endpre()
 102                 endlist()
 103         }
 104         return s
 105 }
 106
 107 function endpre() {
 108         if (inpre) {
 109                 print "@end pre"
 110                 inpre = 0
 111         }
 112 }
 113
 114 function dopre(s) {
 115         if (seenindex != 1)
 116                 return s
 117         if (!inpre) {
 118                 if (s ~ /^    */) {
 119                         print "@pre"
 120                         inpre = 1
 121                 } else
 122                         return s
 123         }
 124         if (s !~ /^    */)
 125                 endpre()
 126         return s
 127 }
 128
 129 function dosect(s,   n,a) {
 130         if (seenindex > 1)
 131                 return s
 132         if (s ~ /^Programming languages/) {
 133                 print "@title " s
 134                 return s
 135         }
 136         if (s !~ /^([1-9]\.|[A-Z]\.[1-9]| *Annex |Contents|Index|Foreword|Introduction| *Bibliography)/)
 137                 return s
 138         if (s ~ /^[0-9.]+[^0-9. ]/)
 139                 return s
 140         n = split(s, a)
 141         id = a[1]
 142         if (id ~ /Annex/)
 143                 id = a[2]
 144         if (id ~ /^([A-Z]|[1-9]\.|[1-9A-Z]\.[0-9.]*[0-9]|Contents|Index|Foreword|Introduction|Bibliography)$/ &&
 145             (n==1 || a[2] ~ /^[A-Z.v]/)) {
 146                 sub(/^ +/,"",s)
 147                 if (id ~ /\.$/)
 148                         id = substr(id,1,length(id)-1)
 149                 if (seenindex || id == "Contents") {
 150                         endpre()
 151                         endlist()
 152                         print "@sect " id
 153                 }
 154                 if (id == "Index")
 155                         seenindex++
 156         }
 157         return s
 158 }
 159
 160 BEGIN {
 161         listlevel = 0
 162         listindex = 0
 163         listindent = 0
 164         noteindent = 0
 165         inpre = 0
 166         seenindex = 0
 167         pn = 1
 168         initpage()
 169 }
 170
 171 /^\[page/ {
 172         if(!para && indent && ok)
 173                 indent = -1
 174
 175         if (!ok)
 176                 indent = 0
 177
 178         n = split(page, a, /\n/)
 179         print "@page " pn
 180         print "@indent " indent
 181         if (indent < 0)
 182                 indent = 0
 183         for (i = 1; i < n; i++) {
 184                 if (a[i] ~ /^@/) {
 185                         if (a[i] ~ /^@para/)
 186                                 endlist()
 187                         print a[i]
 188                 } else {
 189                         s = substr(a[i], indent+1)
 190                         s = dosect(s)
 191                         s = donote(s)
 192                         if (!innote) {
 193                                 s = doul(s)
 194                                 s = dool(s)
 195                         }
 196                         s = dopre(s)
 197                         print s
 198                 }
 199         }
 200
 201         pn++
 202         initpage()
 203         next
 204 }
 205
 206 /^$/ {
 207         if (ok)
 208                 empty = empty "\n"
 209         next
 210 }
 211
 212 length(empty) > 0 {
 213         page = page empty
 214         empty = ""
 215 }
 216
 217 {
 218         ok = 1
 219         if (match($0, /^[0-9]+ +/)) {
 220                 para = 1
 221                 i = RLENGTH
 222                 match($0, /^[0-9]+/)
 223                 page = page "@para " substr($0,1,RLENGTH) "\n"
 224         } else if (match($0, /^ +/)) {
 225                 i = RLENGTH
 226         } else if ($0 !~ /^[0-9]*$/) {
 227                 i = 0
 228         } else
 229                 i = 10
 230         if (i < indent)
 231                 indent = i
 232         page = page $0 "\n"
 233 }'
 234