Quelle  rbbicst.pl   Sprache: Shell

#**************************************************************************
#   Copyright (C) 2016 and later: Unicode, Inc. and others.
#   License & terms of use: http://www.unicode.org/copyright.html
#**************************************************************************
#**************************************************************************
#   Copyright (C) 2002-2016 International Business Machines Corporation
#   and others. All rights reserved.
#**************************************************************************
#
#  rbbicst   Compile the RBBI rule parser state table data into initialized C data.
#            Usage:
#                   cd icu/source/common
#                   perl rbbicst.pl    < rbbirpt.txt > rbbirpt.h
#                   perl rbbicst.pl -j < rbbirpt.txt > RBBIRuleParseTable.java
#
#             The output file, rbbrpt.h, is included by some of the .cpp rbbi
#             implementation files.   This perl script is NOT run as part
#             of a normal ICU build.  It is run by hand when needed, and the
#             rbbirpt.h generated file is put back into cvs.
#
#             See rbbirpt.txt for a description of the input format for this script.
#

if ($ARGV[0] eq "-j") {
    $javaOutput = 1;
    shift @ARGV;
}


$num_states = 1;     # Always the state number for the line being compiled.
$line_num  = 0;      # The line number in the input file.

$states{"pop"} = 255;    # Add the "pop"  to the list of defined state names.
                         # This prevents any state from being labelled with "pop",
                         #  and resolves references to "pop" in the next state field.

line_loop: while (<>) {
    chomp();
    $line = $_;
    @fields = split();
    $line_num++;

    # Remove # comments, which are any fields beginning with a #, plus all
    #  that follow on the line.
    for ($i=0; $i<@fields; $i++) {
        if ($fields[$i] =~ /^#/) {
            @fields = @fields[0 .. $i-1];
            last;
        }
    }
    # ignore blank lines, and those with no fields left after stripping comments..
    if (@fields == 0) {
        next;
    }

    #
    # State Label:  handling.
    #    Does the first token end with a ":"?  If so, it's the name  of a state.
    #    Put in a hash, together with the current state number,
    #        so that we can later look up the number from the name.
    #
    if (@fields[0] =~ /.*:$/) {
        $state_name = @fields[0];
        $state_name =~ s/://;        # strip off the colon from the state name.

        if ($states{$state_name} != 0) {
            print " rbbicst: at line $line-num duplicate definition of state $state_name\n";
        }
        $states{$state_name} = $num_states;
        $stateNames[$num_states] = $state_name;

        # if the label was the only thing on this line, go on to the next line,
        # otherwise assume that a state definition is on the same line and fall through.
        if (@fields == 1) {
            next line_loop;
        }
        shift @fields;                       # shift off label field in preparation
                                             #  for handling the rest of the line.
    }

    #
    # State Transition line.
    #   syntax is this,
    #       character   [n]  target-state  [^push-state]  [function-name]
    #   where
    #      [something]   is an optional something
    #      character     is either a single quoted character e.g. '['
    #                       or a name of a character class, e.g. white_space
    #

    $state_line_num[$num_states] = $line_num;   # remember line number with each state
                                                #  so we can make better error messages later.
    #
    # First field, character class or literal character for this transition.
    #
    if ($fields[0] =~ /^'.'$/) {
        # We've got a quoted literal character.
        $state_literal_chars[$num_states] = $fields[0];
        $state_literal_chars[$num_states] =~ s/'//g;
    } else {
        # We've got the name of a character class.
        $state_char_class[$num_states] = $fields[0];
        if ($fields[0] =~ /[\W]/) {
            print " rbbicsts: at line $line_num, bad character literal or character class name.\n";
            print " scanning $fields[0]\n";
            exit(-1);
        }
    }
    shift @fields;

    #
    # do the 'n' flag
    #
    $state_flag[$num_states] = "false";
    if ($fields[0] eq "n") {
        $state_flag[$num_states] = "true";
        shift @fields;
    }

    #
    # do the destination state.
    #
    $state_dest_state[$num_states] = $fields[0];
    if ($fields[0] eq "") {
        print " rbbicsts: at line $line_num, destination state missing.\n";
        exit(-1);
    }
    shift @fields;

    #
    # do the push state, if present.
    #
    if ($fields[0] =~ /^\^/) {
        $fields[0] =~ s/^\^//;
        $state_push_state[$num_states] = $fields[0];
        if ($fields[0] eq "" ) {
            print " rbbicsts: at line $line_num, expected state after ^ (no spaces).\n";
            exit(-1);
        }
        shift @fields;
    }

    #
    # Lastly, do the optional action name.
    #
    if ($fields[0] ne "") {
        $state_func_name[$num_states] = $fields[0];
        shift @fields;
    }

    #
    #  There should be no fields left on the line at this point.
    #
    if (@fields > 0) {
       print " rbbicsts: at line $line_num, unexpected extra stuff on input line.\n";
       print " scanning $fields[0]\n";
   }
   $num_states++;
}

#
# We've read in the whole file, now go back and output the
#   C source code for the state transition table.
#
# We read all states first, before writing anything,  so that the state numbers
# for the destination states are all available to be written.
#

#
# Make hashes for the names of the character classes and
#      for the names of the actions that appeared.
#
for ($state=1; $state < $num_states; $state++) {
    if ($state_char_class[$state] ne "") {
        if ($charClasses{$state_char_class[$state]} == 0) {
            $charClasses{$state_char_class[$state]} = 1;
        }
    }
    if ($state_func_name[$state] eq "") {
        $state_func_name[$state] = "doNOP";
    }
    if ($actions{$state_action_name[$state]} == 0) {
        $actions{$state_func_name[$state]} = 1;
    }
}

#
# Check that all of the destination states have been defined
#
#
$states{"exit"} = 0;              # Predefined state name, terminates state machine.
for ($state=1; $state<$num_states; $state++) {
   if ($states{$state_dest_state[$state]} == 0 && $state_dest_state[$state] ne "exit") {
       print "Error at line $state_line_num[$state]: target state \"$state_dest_state[$state]\" is not defined.\n";
       $errors++;
   }
   if ($state_push_state[$state] ne "" && $states{$state_push_state[$state]} == 0) {
       print "Error at line $state_line_num[$state]: target state \"$state_push_state[$state]\" is not defined.\n";
       $errors++;
   }
}

die if ($errors>0);

#
# Assign numbers to each of the character classes classes  used.
#   Sets are numbered from 128 - 250
#   The values 0-127 in the state table are used for matching
#     individual ASCII characters (the only thing that can appear in the rules.)
#   The "set" names appearing in the code below (default, etc.)  need special
#     handling because they do not correspond to a normal set of characters,
#     but trigger special handling by code in the state machine.
#
$i = 128;
foreach $setName (sort keys %charClasses) {
    if ($setName eq "default") {
        $charClasses{$setName} = 255;}
    elsif ($setName eq "escaped") {
        $charClasses{$setName} = 254;}
    elsif ($setName eq "escapedP") {
        $charClasses{$setName} = 253;}
    elsif ($setName eq "eof") {
        $charClasses{$setName} = 252;}
    else {
        # Normal (single) character class.  Number them.
        $charClasses{$setName} = $i;
        $i++;
    }
}


my ($sec, $min, $hour, , $day, $mon, $year, $wday, $yday, $isdst) = localtime;
$year += 1900;

if ($javaOutput) {
    print "/*\n";
    print " *******************************************************************************\n";
    print " * Copyright (C) 2003-$year,\n";
    print " * International Business Machines Corporation and others. All Rights Reserved.\n";
    print " *******************************************************************************\n";
    print " */\n";
    print " \n";
    print "package com.ibm.icu.text;\n";
    print " \n";
    print "/**\n";
    print " * Generated Java File. Do not edit by hand.\n";
    print " * This file contains the state table for the ICU Rule Based Break Iterator\n";
    print " * rule parser.\n";
    print " * It is generated by the Perl script \"rbbicst.pl\" from\n";
    print " * the rule parser state definitions file \"rbbirpt.txt\".\n";
    print " * \@internal \n";
    print " *\n";
    print " */\n";

    print "class RBBIRuleParseTable\n";
    print "{\n";

     #
    # Emit the constants for the actions to be performed.
    #
    $n = 1;
    foreach $act (sort keys %actions) {
        print " static final short $act = $n;\n";
        $n++;
    }
    print " \n";
    
    #
    # Emit constants for char class names
    #
    foreach $setName (sort keys %charClasses) {
       print " static final short kRuleSet_$setName = $charClasses{$setName};\n";
    }
    print "\n\n";
    
    
    print " static class RBBIRuleTableElement { \n";
    print " short fAction; \n";
    print " short fCharClass; \n";
    print " short fNextState; \n";
    print " short fPushState; \n";
    print " boolean fNextChar; \n";
    print " String fStateName; \n";
    print " RBBIRuleTableElement(short a, int cc, int ns, int ps, boolean nc, String sn) { \n";
    print " fAction = a; \n";
    print " fCharClass = (short)cc; \n";
    print " fNextState = (short)ns; \n";
    print " fPushState = (short)ps; \n";
    print " fNextChar = nc; \n";
    print " fStateName = sn; \n";
    print " } \n";
    print " }; \n";
    print " \n";
    
    
    print " static RBBIRuleTableElement[] gRuleParseStateTable = { \n ";
    print " new RBBIRuleTableElement(doNOP, 0, 0,0, true, null ) // 0 \n";  #output the unused state 0. 
    for ($state=1; $state < $num_states; $state++) {
        print " , new RBBIRuleTableElement($state_func_name[$state],";
        if ($state_literal_chars[$state] ne "") {
            $c = $state_literal_chars[$state];
            print("'$c', "); 
        }else {
            print " $charClasses{$state_char_class[$state]},";
        }
        print " $states{$state_dest_state[$state]},";
 
        # The push-state field is optional.  If omitted, fill field with a zero, which flags
        #   the state machine that there is no push state.
        if ($state_push_state[$state] eq "") {
            print "0, ";
        } else {
            print " $states{$state_push_state[$state]},";
        }
        print " $state_flag[$state], ";
 
        # if this is the first row of the table for this state, put out the state name.
        if ($stateNames[$state] ne "") {
            print " \"$stateNames[$state]\") ";
        } else {
            print " null ) ";
        }
            
        # Put out a comment showing the number (index) of this state row,
        print " // $state ";
        print "\n";
    }
    print " };\n";

    print "}; \n";
    
}
else
{
    #
    #  C++ Output ...
    #


    print "//---------------------------------------------------------------------------------\n";
    print "//\n";
    print "// Generated Header File. Do not edit by hand.\n";
    print "// This file contains the state table for the ICU Rule Based Break Iterator\n";
    print "// rule parser.\n";
    print "// It is generated by the Perl script \"rbbicst.pl\" from\n";
    print "// the rule parser state definitions file \"rbbirpt.txt\".\n";
    print "//\n";
    print "// Copyright (C) 2002-$year International Business Machines Corporation \n";
    print "// and others. All rights reserved. \n";
    print "//\n";
    print "//---------------------------------------------------------------------------------\n";
    print "#ifndef RBBIRPT_H\n";
    print "#define RBBIRPT_H\n";
    print "\n";
    print "#include \"unicode/utypes.h\"\n";
    print "\n";
    print "U_NAMESPACE_BEGIN\n";

    #
    # Emit the constants for indices of Unicode Sets
    #   Define one constant for each of the character classes encountered.
    #   At the same time, store the index corresponding to the set name back into hash.
    #
    print "//\n";
    print "// Character classes for RBBI rule scanning.\n";
    print "//\n";
    foreach $setName (sort keys %charClasses) {
        if ($charClasses{$setName} < 250) {
           # Normal character class.
           print " static const uint8_t kRuleSet_$setName = $charClasses{$setName};\n";
        }
    }
    print "\n\n";

    #
    # Emit the enum for the actions to be performed.
    #
    print "enum RBBI_RuleParseAction {\n";
    foreach $act (sort keys %actions) {
        print " $act,\n";
    }
    print " rbbiLastAction};\n\n";

    #
    # Emit the struct definition for transition table elements.
    #
    print "//-------------------------------------------------------------------------------\n";
    print "//\n";
    print "// RBBIRuleTableEl represents the structure of a row in the transition table\n";
    print "// for the rule parser state machine.\n";
    print "//-------------------------------------------------------------------------------\n";
    print "struct RBBIRuleTableEl {\n";
    print " RBBI_RuleParseAction fAction;\n";
    print " uint8_t fCharClass; // 0-127: an individual ASCII character\n";
    print " // 128-255: character class index\n";
    print " uint8_t fNextState; // 0-250: normal next-stat numbers\n";
    print " // 255: pop next-state from stack.\n";
    print " uint8_t fPushState;\n";
    print " UBool fNextChar;\n";
    print "};\n\n";

    #
    # emit the state transition table
    #
    print "static const struct RBBIRuleTableEl gRuleParseStateTable[] = {\n";
    print " {doNOP, 0, 0, 0, true}\n";    # State 0 is a dummy.  Real states start with index = 1.
    for ($state=1; $state < $num_states; $state++) {
        print " , {$state_func_name[$state],";
        if ($state_literal_chars[$state] ne "") {
            $c = $state_literal_chars[$state];
            printf(" %d /* $c */,", ord($c));   #  use numeric value, so EBCDIC machines are ok.
        }else {
            print " $charClasses{$state_char_class[$state]},";
        }
        print " $states{$state_dest_state[$state]},";

        # The push-state field is optional.  If omitted, fill field with a zero, which flags
        #   the state machine that there is no push state.
        if ($state_push_state[$state] eq "") {
            print "0, ";
        } else {
            print " $states{$state_push_state[$state]},";
        }
        print " $state_flag[$state]} ";

        # Put out a C++ comment showing the number (index) of this state row,
        #   and, if this is the first row of the table for this state, the state name.
        print " // $state ";
        if ($stateNames[$state] ne "") {
            print " $stateNames[$state]";
        }
        print "\n";
    };
    print " };\n";


    #
    # emit a mapping array from state numbers to state names.
    #
    #    This array is used for producing debugging output from the rule parser.
    #
    print "#ifdef RBBI_DEBUG\n";
    print "static const char * const RBBIRuleStateNames[] = {";
    for ($state=0; $state<$num_states; $state++) {
        if ($stateNames[$state] ne "") {
            print " \"$stateNames[$state]\",\n";
        } else {
            print " 0,\n";
        }
    }
    print " 0};\n";
    print "#endif\n\n";

    print "U_NAMESPACE_END\n";
    print "#endif\n";
}