poparse.tcl -- PO files (Portable Object) (GNU gettext format) parser

poparse.tcl

#**************************************************************************
# parse
#   Internally used proc to parse PO files (Portable Object) (GNU gettext
#   format).
# Parameters:
#   f_ --
#**************************************************************************
proc parse { f_ } {
  ## TODO [SB]: file should be read partially, possible with other RE (non greedy ".*?" - could be slow by big files) :
  set text_ [read $f_]
  set dcat [dict create]
  set st 0
  while { [regexp -start $st -indices {\nmsgid\M(.*?)\nmsgstr\M(.*?)\n(?:\n|$)} $text_ wh msgid msgstr] } {
    set st [lindex $wh 1]
    set msgid [string range $text_ [lindex $msgid 0] [lindex $msgid 1]]
    set msgstr [string range $text_ [lindex $msgstr 0] [lindex $msgstr 1]]
    set msgid [split [string trim $msgid " \n"] "\n"]
    set msgstr [split [string trim $msgstr " \n"] "\n"]
    set id ""
    set str ""
    foreach line $msgid {
      set line [string trim $line]
      set t {}
      regexp {"(.*)"} $line {} t
      append id $t
    }
    foreach line $msgstr {
      set line [string trim $line]
      set t {}
      regexp {"(.*)"} $line {} t
      append str $t
    }
    ## special case for escaped newline and quot :
    set id [string map [list \\n \n \\n\\r \n\r \\\" \"] $id]
    set str [string map [list \\n \n \\n\\r \n\r \\\" \"] $str]
    ## check group syntax :
    if { [regexp {^\[([^\[\]]+)\]\.(.*)$} $id {} grp id] } {
      ## save memory by equal id/str messages :
      if { $str == $id } {
        set str $id
      }
      ## add to group :
      dict set dcat $grp $id $str
      continue
    }
    ## save memory by equal id/str messages :
    if { $str == $id } {
      set str $id
    }
    dict set dcat $id $str
  }
  ## create one empty elm, cause dcat should be not empty - mark as already loaded :
  dict set dcat {} {}
  return $dcat
}