i18npool/source/localedata/data/currency-check.awk

   1 #!/usr/bin/gawk -f
   2 # Usage: gawk -f currency-check.awk *.xml
   3 # Check any
   4 # <FormatCode>...[$xxx-...]...</FormatCode>
   5 # against every
   6 # <CurrencySymbol>xxx</CurrencySymbol>
   7 # definition of the same XML file and output symbols if no match was found.
   8 # For formatindex="12" to formatindex="15" and for formatindex="17" it is
   9 # checked if the used currency symbol is the usedInCompatibleFormatCodes
  10 # currency symbol as it is needed by the number formatter.
  11 # Also generates output if the generic currency symbol (UTF8 string 0xC2A4)
  12 # is used instead of a real currency symbol.
  13 # Author: Eike Rathke <er@openoffice.org>
  14
  15 BEGIN {
  16     file = ""
  17 }
  18
  19
  20 file != FILENAME {
  21     if ( file )
  22         checkIt()
  23     file = FILENAME
  24     line = 0
  25     nFormats = 0
  26     nCurrencies = 0
  27     bFormatAuto = 0
  28     sReplaceFrom = ""
  29     sReplaceTo = ""
  30     sMatchReplace = ""
  31     sRefCurrencyFromLocale = ""
  32     crlf = 0
  33 }
  34
  35 {
  36     ++line
  37     # If run under Unix a CrLf spoils ...$ line end checks. DOS line endings
  38     # are boo anyways.
  39     if ( /\x0D$/ )
  40     {
  41         print "Error: not Unix line ending in line " line
  42         crlf = 1
  43         exit(1)
  44     }
  45     if ( $1 ~ /^<LC_FORMAT(>|$)/ )
  46     {
  47         if ( $0 ~ /replaceFrom="\[CURRENCY\]"/ )
  48         {
  49             sReplaceFrom = "\\[CURRENCY\\]"
  50             sMatchReplace = "^<FormatCode>.*" sReplaceFrom
  51         }
  52         for ( j=2; j<=NF; ++j )
  53         {
  54             if ( $j ~ /^replaceTo="/ )
  55             {
  56                 l = 12
  57                 if ( $j ~ />$/ )
  58                     ++l
  59                 if ( $j ~ /\/>$/ )
  60                     ++l
  61                 sReplaceTo = substr( $j, 12, length($j)-l )
  62             }
  63         }
  64     }
  65     else if ( $1 ~ /^<FormatElement(>|$)/ )
  66     {
  67         if ( $0 ~ /usage="CURRENCY"/ )
  68         {
  69             if ( $0 ~ /formatindex="1[23457]"/ )
  70                 bFormatAuto = 1
  71             else
  72                 bFormatAuto = 0
  73         }
  74     }
  75     else if ( $0 ~ /^[[:blank:]]*<FormatCode>.*\[\$.*-[0-9a-fA-F]+\]/ ||
  76             (sMatchReplace && $0 ~ sMatchReplace ) )
  77     {
  78         if ( sReplaceFrom )
  79             gsub( sReplaceFrom, sReplaceTo )
  80         split( $0, arr, /<|>/ )
  81         split( arr[3], code, /(\[\$)|(-[0-9a-fA-F]+\])/ )
  82         for ( j in code )
  83         {
  84             if ( code[j] && code[j] !~ /\#|0|\[NatNum/ )
  85             {
  86                 FormatLine[nFormats] = file " line " line
  87                 FormatAuto[nFormats] = bFormatAuto
  88                 Formats[nFormats++] = code[j]
  89             }
  90         }
  91         bFormatAuto = 0
  92     }
  93     else if ( $1 ~ /^<LC_CURRENCY(>|$)/ )
  94     {
  95         for ( j=2; j<=NF; ++j )
  96         {
  97             if ( $j ~ /^ref="/ )
  98             {
  99                 l = 6
 100                 if ( $j ~ />$/ )
 101                     ++l
 102                 if ( $j ~ /\/>$/ )
 103                     ++l
 104                 locale = substr( $j, 6, length($j)-l )
 105                 sRefCurrencyFromLocale = file
 106                 oldfile = file
 107                 oldline = line
 108                 file = locale ".xml"
 109                 line = 0
 110                 while ( (getline <file) > 0 )
 111                 {
 112                     ++line
 113                     getCurrencyParams()
 114                 }
 115                 close( file )
 116                 if ( !line )
 117                     print "ref locale not available: " file \
 118                         " (from " oldfile " line " oldline ")"
 119                 file = oldfile
 120                 line = oldline
 121                 sRefCurrencyFromLocale = ""
 122             }
 123         }
 124     }
 125     else
 126         getCurrencyParams()
 127 }
 128
 129
 130 END {
 131     if ( file && !crlf )
 132         checkIt()
 133 }
 134
 135
 136 function getCurrencyParams() {
 137     # Assumes that each element is on a line on its own!
 138     if ( $1 ~ /^<Currency(>|$)/ )
 139     {
 140         if ( $0 ~ /default="true"/ )
 141             SymbolDefault[nCurrencies] = 1
 142         else
 143             SymbolDefault[nCurrencies] = 0
 144         if ( $0 ~ /usedInCompatibleFormatCodes="true"/ )
 145             SymbolCompati[nCurrencies] = 1
 146         else
 147             SymbolCompati[nCurrencies] = 0
 148     }
 149     else if ( $0 ~ /^[[:blank:]]*<CurrencyID>/ )
 150     {
 151         split( $0, arr, /<|>/ )
 152         if ( sRefCurrencyFromLocale )
 153             IDLine[nCurrencies] = file " line " line \
 154                 " (referenced from " sRefCurrencyFromLocale ")"
 155         else
 156             IDLine[nCurrencies] = file " line " line
 157         IDs[nCurrencies] = arr[3]
 158     }
 159     else if ( $0 ~ /^[[:blank:]]*<CurrencySymbol>/ )
 160     {
 161         split( $0, arr, /<|>/ )
 162         if ( sRefCurrencyFromLocale )
 163             SymbolLine[nCurrencies] = file " line " line \
 164                 " (referenced from " sRefCurrencyFromLocale ")"
 165         else
 166             SymbolLine[nCurrencies] = file " line " line
 167         Symbols[nCurrencies] = arr[3]
 168     }
 169     else if ( $0 ~ /^[[:blank:]]*<BankSymbol>/ )
 170     {
 171         split( $0, arr, /<|>/ )
 172         if ( sRefCurrencyFromLocale )
 173             BankSymbolLine[nCurrencies] = file " line " line \
 174                 " (referenced from " sRefCurrencyFromLocale ")"
 175         else
 176             BankSymbolLine[nCurrencies] = file " line " line
 177         BankSymbols[nCurrencies] = arr[3]
 178     }
 179     else if ( $1 ~ /^<\/Currency>/ )
 180     {
 181         ++nCurrencies
 182     }
 183 }
 184
 185
 186 function checkIt() {
 187     bad = 0
 188     for ( j=0; j<nFormats; ++j )
 189     {
 190         state = FormatInSymbol( Formats[j] )
 191         if ( Formats[j] == "\xc2\xa4" )
 192         {
 193             bad = 1
 194             print "    bad: `" Formats[j] "'   (" FormatLine[j] ")"
 195         }
 196         else if ( state == 0 )
 197         {
 198             bad = 1
 199             print "unknown: `" Formats[j] "'   (" FormatLine[j] ")"
 200         }
 201         else if ( FormatAuto[j] && state < 2 )
 202         {
 203             bad = 1
 204             print "badauto: `" Formats[j] "'   (" FormatLine[j] ")"
 205         }
 206     }
 207     if ( bad )
 208     {
 209         for ( j=0; j<nCurrencies; ++j )
 210         {
 211             bDef = 0
 212             if ( Symbols[j] == "\xc2\xa4" )
 213                 print "def bad: `" Symbols[j] "'   (" SymbolLine[j] ")"
 214             if ( SymbolDefault[j] )
 215             {
 216                 bDef = 1
 217                 print "default: `" Symbols[j] "'   (" SymbolLine[j] ")"
 218             }
 219             if ( SymbolCompati[j] )
 220             {
 221                 bDef = 1
 222                 print "compati: `" Symbols[j] "'   (" SymbolLine[j] ")"
 223             }
 224             if ( !bDef )
 225                 print "defined: `" Symbols[j] "'   (" SymbolLine[j] ")"
 226         }
 227     }
 228     else
 229     {
 230         bHasDefault = 0
 231         bHasCompati = 0
 232         for ( j=0; j<nCurrencies; ++j )
 233         {
 234             if ( Symbols[j] == "\xc2\xa4" )
 235             {
 236                 bad = 1
 237                 print "def bad: `" Symbols[j] "'   (" SymbolLine[j] ")"
 238             }
 239             if ( SymbolDefault[j] )
 240             {
 241                 if ( !bHasDefault )
 242                     bHasDefault = 1
 243                 else
 244                 {
 245                     bad = 1
 246                     print "dupe default: `" Symbols[j] "'   (" SymbolLine[j] ")"
 247                 }
 248             }
 249             if ( SymbolCompati[j] )
 250             {
 251                 if ( !bHasCompati )
 252                     bHasCompati = 1
 253                 else
 254                 {
 255                     bad = 1
 256                     print "dupe compati: `" Symbols[j] "'   (" SymbolLine[j] ")"
 257                 }
 258             }
 259         }
 260         if ( !bHasDefault )
 261         {
 262             bad = 1
 263             print "  no default: (" file ")"
 264         }
 265         if ( !bHasCompati )
 266         {
 267             bad = 1
 268             print "  no compati: (" file ")"
 269         }
 270     }
 271     for ( j=0; j<nCurrencies; ++j )
 272     {
 273         # Check if CurrencyID at least resembles some ISO 4217 code.
 274         # The only exception is zh_MO that had an erroneous original data set
 275         # with BankSymbol="P" (stored as ISO code in documents, hence copied to
 276         # CurrencyID now) and needs that entry for legacy documents.
 277         # There is a strange bug in gawk 3.1.4 that does a match of [A-Z] on
 278         # lower case except 'a', regardless of IGNORECASE setting, hence this
 279         # ugly notation. [[:upper:]] wouldn't be correct since we want only
 280         # ASCII to match.
 281         if ( IDs[j] !~ /^[ABCDEFGHIJKLMNOPQRSTUVWXYZ][ABCDEFGHIJKLMNOPQRSTUVWXYZ][ABCDEFGHIJKLMNOPQRSTUVWXYZ]$/ \
 282               && !(file == "zh_MO.xml" && IDs[j] == "P") )
 283         {
 284             bad = 1
 285             print "no ISO 4217 code: `" IDs[j] "'   (" IDLine[j] ")"
 286         }
 287         # CurrencyID should equal BankSymbol for now.
 288         if ( IDs[j] != BankSymbols[j] )
 289         {
 290             bad = 1
 291             print "not equal: CurrencyID `" IDs[j] "' != BankSymbol `" BankSymbols[j] \
 292                   "'   (" IDLine[j] " and " BankSymbolLine[j] ")"
 293         }
 294     }
 295     if ( bad )
 296         print ""
 297 }
 298
 299
 300 function FormatInSymbol( format ) {
 301     state = 0
 302     for ( nSym=0; nSym<nCurrencies; ++nSym )
 303     {
 304         if ( format == Symbols[nSym] )
 305         {
 306             # Two currencies can have the same symbol (e.g. az_AZ.xml 'man.'
 307             # for AZM and AZN), continue to lookup if the match isn't the
 308             # compatible one.
 309             if ( SymbolCompati[nSym] )
 310                 return 2
 311             else
 312                 state = 1
 313         }
 314     }
 315     return state
 316 }
 317
 318 # vim: ts=4 sw=4 expandtab