[apple/icu.git] / icuSources / common / rbbicst.pl

#**************************************************************************
#   Copyright (C) 2002-2016 International Business Machines Corporation
#   and others. All rights reserved.
#**************************************************************************
#
#  rbbicst   Compile the RBBI rule paser 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] = $javaOutput? "false" : "FALSE";
    if ($fields[0] eq "n") {
        $state_flag[$num_states] = $javaOutput? "true": "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 indicies 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 transtion 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";
}
Commit	Line	Data
46f4442e	1	#**************************************************************************
2ca993e8 A	2	# Copyright (C) 2002-2016 International Business Machines Corporation
2ca993e8 A	3	# and others. All rights reserved.
46f4442e A	4	#**************************************************************************
	5	#
	6	# rbbicst Compile the RBBI rule paser state table data into initialized C data.
	7	# Usage:
	8	# cd icu/source/common
	9	# perl rbbicst.pl < rbbirpt.txt > rbbirpt.h
	10	# perl rbbicst.pl -j < rbbirpt.txt > RBBIRuleParseTable.java
	11	#
	12	# The output file, rbbrpt.h, is included by some of the .cpp rbbi
	13	# implementation files. This perl script is NOT run as part
	14	# of a normal ICU build. It is run by hand when needed, and the
	15	# rbbirpt.h generated file is put back into cvs.
	16	#
	17	# See rbbirpt.txt for a description of the input format for this script.
	18	#
	19
	20	if ($ARGV[0] eq "-j") {
	21	$javaOutput = 1;
	22	shift @ARGV;
	23	}
	24
	25
	26	$num_states = 1; # Always the state number for the line being compiled.
	27	$line_num = 0; # The line number in the input file.
	28
	29	$states{"pop"} = 255; # Add the "pop" to the list of defined state names.
	30	# This prevents any state from being labelled with "pop",
	31	# and resolves references to "pop" in the next state field.
	32
	33	line_loop: while (<>) {
	34	chomp();
	35	$line = $_;
	36	@fields = split();
	37	$line_num++;
	38
	39	# Remove # comments, which are any fields beginning with a #, plus all
	40	# that follow on the line.
	41	for ($i=0; $i<@fields; $i++) {
	42	if ($fields[$i] =~ /^#/) {
	43	@fields = @fields[0 .. $i-1];
	44	last;
	45	}
	46	}
	47	# ignore blank lines, and those with no fields left after stripping comments..
	48	if (@fields == 0) {
	49	next;
	50	}
	51
	52	#
	53	# State Label: handling.
	54	# Does the first token end with a ":"? If so, it's the name of a state.
	55	# Put in a hash, together with the current state number,
	56	# so that we can later look up the number from the name.
	57	#
	58	if (@fields[0] =~ /.*:$/) {
	59	$state_name = @fields[0];
	60	$state_name =~ s/://; # strip off the colon from the state name.
	61
	62	if ($states{$state_name} != 0) {
	63	print " rbbicst: at line $line-num duplicate definition of state $state_name\n";
	64	}
	65	$states{$state_name} = $num_states;
	66	$stateNames[$num_states] = $state_name;
	67
68	# if the label was the only thing on this line, go on to the next line,
69	# otherwise assume that a state definition is on the same line and fall through.
70	if (@fields == 1) {
71	next line_loop;
72	}
73	shift @fields; # shift off label field in preparation
74	# for handling the rest of the line.
75	}
76
77	#
78	# State Transition line.
79	# syntax is this,
80	# character [n] target-state [^push-state] [function-name]
81	# where
82	# [something] is an optional something
83	# character is either a single quoted character e.g. '['
84	# or a name of a character class, e.g. white_space
85	#
86
87	$state_line_num[$num_states] = $line_num; # remember line number with each state
88	# so we can make better error messages later.
89	#
90	# First field, character class or literal character for this transition.
91	#
92	if ($fields[0] =~ /^'.'$/) {
93	# We've got a quoted literal character.
94	$state_literal_chars[$num_states] = $fields[0];
95	$state_literal_chars[$num_states] =~ s/'//g;
96	} else {
97	# We've got the name of a character class.
98	$state_char_class[$num_states] = $fields[0];
99	if ($fields[0] =~ /[\W]/) {
100	print " rbbicsts: at line $line_num, bad character literal or character class name.\n";
101	print " scanning $fields[0]\n";
102	exit(-1);
103	}
104	}
105	shift @fields;
106
107	#
108	# do the 'n' flag
109	#
110	$state_flag[$num_states] = $javaOutput? "false" : "FALSE";
111	if ($fields[0] eq "n") {
112	$state_flag[$num_states] = $javaOutput? "true": "TRUE";
113	shift @fields;
114	}
115
116	#
117	# do the destination state.
118	#
119	$state_dest_state[$num_states] = $fields[0];
120	if ($fields[0] eq "") {
121	print " rbbicsts: at line $line_num, destination state missing.\n";
122	exit(-1);
123	}
124	shift @fields;
125
126	#
127	# do the push state, if present.
128	#
129	if ($fields[0] =~ /^\^/) {
130	$fields[0] =~ s/^\^//;
131	$state_push_state[$num_states] = $fields[0];
132	if ($fields[0] eq "" ) {
133	print " rbbicsts: at line $line_num, expected state after ^ (no spaces).\n";
134	exit(-1);
135	}
136	shift @fields;
137	}
138
139	#
140	# Lastly, do the optional action name.
141	#
142	if ($fields[0] ne "") {
143	$state_func_name[$num_states] = $fields[0];
144	shift @fields;
145	}
146
147	#
148	# There should be no fields left on the line at this point.
149	#
150	if (@fields > 0) {
151	print " rbbicsts: at line $line_num, unexpected extra stuff on input line.\n";
152	print " scanning $fields[0]\n";
153	}
154	$num_states++;
155	}
156
157	#
158	# We've read in the whole file, now go back and output the
159	# C source code for the state transition table.
160	#
161	# We read all states first, before writing anything, so that the state numbers
162	# for the destination states are all available to be written.
163	#
164
165	#
166	# Make hashes for the names of the character classes and
167	# for the names of the actions that appeared.
168	#
169	for ($state=1; $state < $num_states; $state++) {
170	if ($state_char_class[$state] ne "") {
171	if ($charClasses{$state_char_class[$state]} == 0) {
172	$charClasses{$state_char_class[$state]} = 1;
173	}
174	}
175	if ($state_func_name[$state] eq "") {
176	$state_func_name[$state] = "doNOP";
177	}
178	if ($actions{$state_action_name[$state]} == 0) {
179	$actions{$state_func_name[$state]} = 1;
180	}
181	}
182
183	#
184	# Check that all of the destination states have been defined
185	#
186	#
187	$states{"exit"} = 0; # Predefined state name, terminates state machine.
188	for ($state=1; $state<$num_states; $state++) {
189	if ($states{$state_dest_state[$state]} == 0 && $state_dest_state[$state] ne "exit") {
190	print "Error at line $state_line_num[$state]: target state \"$state_dest_state[$state]\" is not defined.\n";
191	$errors++;
192	}
193	if ($state_push_state[$state] ne "" && $states{$state_push_state[$state]} == 0) {
194	print "Error at line $state_line_num[$state]: target state \"$state_push_state[$state]\" is not defined.\n";
195	$errors++;
196	}
197	}
198
199	die if ($errors>0);
200
201	#
202	# Assign numbers to each of the character classes classes used.
203	# Sets are numbered from 128 - 250
204	# The values 0-127 in the state table are used for matching
205	# individual ASCII characters (the only thing that can appear in the rules.)
206	# The "set" names appearing in the code below (default, etc.) need special
207	# handling because they do not correspond to a normal set of characters,
208	# but trigger special handling by code in the state machine.
209	#
210	$i = 128;
211	foreach $setName (sort keys %charClasses) {
212	if ($setName eq "default") {
213	$charClasses{$setName} = 255;}
214	elsif ($setName eq "escaped") {
215	$charClasses{$setName} = 254;}
216	elsif ($setName eq "escapedP") {
217	$charClasses{$setName} = 253;}
218	elsif ($setName eq "eof") {
219	$charClasses{$setName} = 252;}
220	else {
221	# Normal (single) character class. Number them.
222	$charClasses{$setName} = $i;
223	$i++;
224	}
225	}
226
227
228	my ($sec, $min, $hour, , $day, $mon, $year, $wday, $yday, $isdst) = localtime;
229	$year += 1900;
230
231	if ($javaOutput) {
232	print "/*\n";
233	print " *******************************************************************************\n";
234	print " * Copyright (C) 2003-$year,\n";
235	print " * International Business Machines Corporation and others. All Rights Reserved.\n";
236	print " *******************************************************************************\n";
237	print " */\n";
238	print " \n";
239	print "package com.ibm.icu.text;\n";
240	print " \n";
241	print "/**\n";
242	print " * Generated Java File. Do not edit by hand.\n";
243	print " * This file contains the state table for the ICU Rule Based Break Iterator\n";
244	print " * rule parser.\n";
245	print " * It is generated by the Perl script \"rbbicst.pl\" from\n";
246	print " * the rule parser state definitions file \"rbbirpt.txt\".\n";
247	print " * \@internal \n";
248	print " *\n";
249	print " */\n";
250
251	print "class RBBIRuleParseTable\n";
252	print "{\n";
253
254	#
255	# Emit the constants for the actions to be performed.
256	#
257	$n = 1;
258	foreach $act (sort keys %actions) {
259	print " static final short $act = $n;\n";
260	$n++;
261	}
262	print " \n";
263
264	#
265	# Emit constants for char class names
266	#
267	foreach $setName (sort keys %charClasses) {
268	print " static final short kRuleSet_$setName = $charClasses{$setName};\n";
269	}
270	print "\n\n";
271
272
273	print " static class RBBIRuleTableElement { \n";
274	print " short fAction; \n";
275	print " short fCharClass; \n";
276	print " short fNextState; \n";
277	print " short fPushState; \n";
278	print " boolean fNextChar; \n";
279	print " String fStateName; \n";
280	print " RBBIRuleTableElement(short a, int cc, int ns, int ps, boolean nc, String sn) { \n";
281	print " fAction = a; \n";
282	print " fCharClass = (short)cc; \n";
283	print " fNextState = (short)ns; \n";
284	print " fPushState = (short)ps; \n";
285	print " fNextChar = nc; \n";
286	print " fStateName = sn; \n";
287	print " } \n";
288	print " }; \n";
289	print " \n";
290
291
292	print " static RBBIRuleTableElement[] gRuleParseStateTable = { \n ";
293	print " new RBBIRuleTableElement(doNOP, 0, 0,0, true, null ) // 0 \n"; #output the unused state 0.
294	for ($state=1; $state < $num_states; $state++) {
295	print " , new RBBIRuleTableElement($state_func_name[$state],";
296	if ($state_literal_chars[$state] ne "") {
297	$c = $state_literal_chars[$state];
298	print("'$c', ");
299	}else {
300	print " $charClasses{$state_char_class[$state]},";
301	}
302	print " $states{$state_dest_state[$state]},";
303
304	# The push-state field is optional. If omitted, fill field with a zero, which flags
305	# the state machine that there is no push state.
306	if ($state_push_state[$state] eq "") {
307	print "0, ";
308	} else {
309	print " $states{$state_push_state[$state]},";
310	}
311	print " $state_flag[$state], ";
312
313	# if this is the first row of the table for this state, put out the state name.
314	if ($stateNames[$state] ne "") {
315	print " \"$stateNames[$state]\") ";
316	} else {
317	print " null ) ";
318	}
319
320	# Put out a comment showing the number (index) of this state row,
321	print " // $state ";
322	print "\n";
323	}
324	print " };\n";
325
326	print "}; \n";
327
328	}
329	else
330	{
331	#
332	# C++ Output ...
333	#
334
335
336	print "//---------------------------------------------------------------------------------\n";
337	print "//\n";
338	print "// Generated Header File. Do not edit by hand.\n";
339	print "// This file contains the state table for the ICU Rule Based Break Iterator\n";
340	print "// rule parser.\n";
341	print "// It is generated by the Perl script \"rbbicst.pl\" from\n";
342	print "// the rule parser state definitions file \"rbbirpt.txt\".\n";
343	print "//\n";
344	print "// Copyright (C) 2002-$year International Business Machines Corporation \n";
345	print "// and others. All rights reserved. \n";
346	print "//\n";
347	print "//---------------------------------------------------------------------------------\n";
348	print "#ifndef RBBIRPT_H\n";
349	print "#define RBBIRPT_H\n";
350	print "\n";
2ca993e8 A	351	print "#include \"unicode/utypes.h\"\n";
2ca993e8 A	352	print "\n";
46f4442e A	353	print "U_NAMESPACE_BEGIN\n";
	354
	355	#
	356	# Emit the constants for indicies of Unicode Sets
	357	# Define one constant for each of the character classes encountered.
	358	# At the same time, store the index corresponding to the set name back into hash.
	359	#
	360	print "//\n";
	361	print "// Character classes for RBBI rule scanning.\n";
	362	print "//\n";
	363	foreach $setName (sort keys %charClasses) {
	364	if ($charClasses{$setName} < 250) {
	365	# Normal character class.
	366	print " static const uint8_t kRuleSet_$setName = $charClasses{$setName};\n";
	367	}
	368	}
	369	print "\n\n";
	370
	371	#
	372	# Emit the enum for the actions to be performed.
	373	#
	374	print "enum RBBI_RuleParseAction {\n";
	375	foreach $act (sort keys %actions) {
	376	print " $act,\n";
	377	}
	378	print " rbbiLastAction};\n\n";
	379
	380	#
	381	# Emit the struct definition for transtion table elements.
	382	#
	383	print "//-------------------------------------------------------------------------------\n";
	384	print "//\n";
	385	print "// RBBIRuleTableEl represents the structure of a row in the transition table\n";
	386	print "// for the rule parser state machine.\n";
	387	print "//-------------------------------------------------------------------------------\n";
	388	print "struct RBBIRuleTableEl {\n";
	389	print " RBBI_RuleParseAction fAction;\n";
	390	print " uint8_t fCharClass; // 0-127: an individual ASCII character\n";
	391	print " // 128-255: character class index\n";
	392	print " uint8_t fNextState; // 0-250: normal next-stat numbers\n";
	393	print " // 255: pop next-state from stack.\n";
	394	print " uint8_t fPushState;\n";
	395	print " UBool fNextChar;\n";
	396	print "};\n\n";
	397
	398	#
	399	# emit the state transition table
	400	#
	401	print "static const struct RBBIRuleTableEl gRuleParseStateTable[] = {\n";
	402	print " {doNOP, 0, 0, 0, TRUE}\n"; # State 0 is a dummy. Real states start with index = 1.
	403	for ($state=1; $state < $num_states; $state++) {
	404	print " , {$state_func_name[$state],";
	405	if ($state_literal_chars[$state] ne "") {
	406	$c = $state_literal_chars[$state];
	407	printf(" %d /* $c */,", ord($c)); # use numeric value, so EBCDIC machines are ok.
	408	}else {
	409	print " $charClasses{$state_char_class[$state]},";
	410	}
	411	print " $states{$state_dest_state[$state]},";
	412
	413	# The push-state field is optional. If omitted, fill field with a zero, which flags
	414	# the state machine that there is no push state.
	415	if ($state_push_state[$state] eq "") {
	416	print "0, ";
417	} else {
418	print " $states{$state_push_state[$state]},";
419	}
420	print " $state_flag[$state]} ";
421
422	# Put out a C++ comment showing the number (index) of this state row,
423	# and, if this is the first row of the table for this state, the state name.
424	print " // $state ";
425	if ($stateNames[$state] ne "") {
426	print " $stateNames[$state]";
427	}
428	print "\n";
429	};
430	print " };\n";
431
432
433	#
434	# emit a mapping array from state numbers to state names.
435	#
436	# This array is used for producing debugging output from the rule parser.
437	#
438	print "#ifdef RBBI_DEBUG\n";
439	print "static const char * const RBBIRuleStateNames[] = {";
440	for ($state=0; $state<$num_states; $state++) {
441	if ($stateNames[$state] ne "") {
442	print " \"$stateNames[$state]\",\n";
443	} else {
444	print " 0,\n";
445	}
446	}
447	print " 0};\n";
448	print "#endif\n\n";
449
450	print "U_NAMESPACE_END\n";
451	print "#endif\n";
452	}
453
454
455