uni_parse2.tcl

ejabberd-0.7.5 分布式Jabber服务器

    variable cclass_pmap
    variable cclass_pages
    variable decomp_shift
    variable decomp_pmap
    variable decomp_pages
    variable decomp_list
    variable comp_shift
    variable comp_map
    variable comp_pmap
    variable comp_pages
    variable comp_first_list
    variable comp_second_list
    variable comp_x_list
    variable comp_y_list
    variable pages
    variable groups {}
    variable titleCount

    if {$argc != 3} {
	puts stderr "\nusage: $argv0 <datafile> <exclusionsfile> <outdir>\n"
	exit 1
    }
    set f [open [lindex $argv 1] r]
    set data [read $f]
    close $f

    load_exclusions $data

    set f [open [lindex $argv 0] r]
    set data [read $f]
    close $f

    load_tables $data
    buildTables
    #puts "X = [llength $pMap]  Y= [llength $pages]  A= [llength $groups]"
    #set size [expr {[llength $pMap] + [llength $pages]*(1<<$shift)}]
    #puts "shift = 6, space = $size"
    #puts "title case count = $titleCount"

    set f [open [file join [lindex $argv 2] uni_norm.c] w]
    fconfigure $f -translation lf
    puts $f "/*
 * uni_norm.c --
 *
 *	Declarations of Unicode character information tables.  This file is
 *	automatically generated by the uni_parse2.tcl script.  Do not
 *	modify this file by hand.
 *
 * Copyright (c) 1998 by Scriptics Corporation.
 * All rights reserved.
 *
 * Modified for ejabberd by Alexey Shchepin
 *
 * RCS: @(#) \$Id\$
 */

/*
 * A 16-bit Unicode character is split into two parts in order to index
 * into the following tables.  The lower CCLASS_OFFSET_BITS comprise an offset
 * into a page of characters.  The upper bits comprise the page number.
 */

#define CCLASS_OFFSET_BITS $cclass_shift

/*
 * The pageMap is indexed by page number and returns an alternate page number
 * that identifies a unique page of characters.  Many Unicode characters map
 * to the same alternate page number.
 */

static unsigned char cclassPageMap\[\] = {"
    set line "    "
    set last [expr {[llength $cclass_pmap] - 1}]
    for {set i 0} {$i <= $last} {incr i} {
	append line [lindex $cclass_pmap $i]
	if {$i != $last} {
	    append line ", "
	}
	if {[string length $line] > 70} {
	    puts $f $line
	    set line "    "
	}
    }
    puts $f $line
    puts $f "};

/*
 * The cclassGroupMap is indexed by combining the alternate page number with
 * the page offset and returns a combining class number.
 */

static unsigned char cclassGroupMap\[\] = {"
    set line "    "
    set lasti [expr {[llength $cclass_pages] - 1}]
    for {set i 0} {$i <= $lasti} {incr i} {
	set page [lindex $cclass_pages $i]
	set lastj [expr {[llength $page] - 1}]
	for {set j 0} {$j <= $lastj} {incr j} {
	    append line [lindex $page $j]
	    if {$j != $lastj || $i != $lasti} {
		append line ", "
	    }
	    if {[string length $line] > 70} {
		puts $f $line
		set line "    "
	    }
	}
    }
    puts $f $line
    puts $f "};

#define GetUniCharCClass(ch) (cclassGroupMap\[(cclassPageMap\[(((int)(ch)) & 0xffff) >> CCLASS_OFFSET_BITS\] << CCLASS_OFFSET_BITS) | ((ch) & ((1 << CCLASS_OFFSET_BITS)-1))\])


#define DECOMP_OFFSET_BITS $decomp_shift

/*
 * The pageMap is indexed by page number and returns an alternate page number
 * that identifies a unique page of characters.  Many Unicode characters map
 * to the same alternate page number.
 */

static unsigned char decompPageMap\[\] = {"
    set line "    "
    set last [expr {[llength $decomp_pmap] - 1}]
    for {set i 0} {$i <= $last} {incr i} {
	append line [lindex $decomp_pmap $i]
	if {$i != $last} {
	    append line ", "
	}
	if {[string length $line] > 70} {
	    puts $f $line
	    set line "    "
	}
    }
    puts $f $line
    puts $f "};

/*
 * The decompGroupMap is indexed by combining the alternate page number with
 * the page offset and returns a group number that identifies a length and
 * shift of decomposition sequence in decompList
 */

static int decompGroupMap\[\] = {"
    set line "    "
    set lasti [expr {[llength $decomp_pages] - 1}]
    for {set i 0} {$i <= $lasti} {incr i} {
	set page [lindex $decomp_pages $i]
	set lastj [expr {[llength $page] - 1}]
	for {set j 0} {$j <= $lastj} {incr j} {
	    append line [lindex $page $j]
	    if {$j != $lastj || $i != $lasti} {
		append line ", "
	    }
	    if {[string length $line] > 70} {
		puts $f $line
		set line "    "
	    }
	}
    }
    puts $f $line
    puts $f "};

/*
 * List of decomposition sequences
 */

static int decompList\[\] = {"
    set line "    "
    set last [expr {[llength $decomp_list] - 1}]
    for {set i 0} {$i <= $last} {incr i} {
	set val [lindex $decomp_list $i]

	append line [format "%d" $val]
	if {$i != $last} {
	    append line ", "
	}
	if {[string length $line] > 70} {
	    puts $f $line
	    set line "    "
	}
    }
    puts $f $line
    puts $f "};


/*
 * This macro extracts the information about a character from the
 * Unicode character tables.
 */

#define GetUniCharDecompInfo(ch) (decompGroupMap\[(decompPageMap\[(((int)(ch)) & 0xffff) >> DECOMP_OFFSET_BITS\] << DECOMP_OFFSET_BITS) | ((ch) & ((1 << DECOMP_OFFSET_BITS)-1))\])

#define GetDecompShift(info) ((info) & 0xffff)
#define GetDecompLen(info) ((info) >> 16)


#define COMP_OFFSET_BITS $comp_shift

/*
 * The pageMap is indexed by page number and returns an alternate page number
 * that identifies a unique page of characters.  Many Unicode characters map
 * to the same alternate page number.
 */

static unsigned char compPageMap\[\] = {"
    set line "    "
    set last [expr {[llength $comp_pmap] - 1}]
    for {set i 0} {$i <= $last} {incr i} {
	append line [lindex $comp_pmap $i]
	if {$i != $last} {
	    append line ", "
	}
	if {[string length $line] > 70} {
	    puts $f $line
	    set line "    "
	}
    }
    puts $f $line
    puts $f "};

/*
 * The groupMap is indexed by combining the alternate page number with
 * the page offset and returns a group number that identifies a unique
 * set of character attributes.
 */

static int compGroupMap\[\] = {"
    set line "    "
    set lasti [expr {[llength $comp_pages] - 1}]
    for {set i 0} {$i <= $lasti} {incr i} {
	set page [lindex $comp_pages $i]
	set lastj [expr {[llength $page] - 1}]
	for {set j 0} {$j <= $lastj} {incr j} {
	    append line [lindex $page $j]
	    if {$j != $lastj || $i != $lasti} {
		append line ", "
	    }
	    if {[string length $line] > 70} {
		puts $f $line
		set line "    "
	    }
	}
    }
    puts $f $line
    puts $f "};

/*
 * Lists of compositions for characters that appears only in one composition
 */

static int compFirstList\[\]\[2\] = {"
    set line "    "
    set last [expr {[llength $comp_first_list] - 1}]
    for {set i 0} {$i <= $last} {incr i} {
	set val [lindex $comp_first_list $i]

	append line [format "{%d, %d}" [lindex $val 0] [lindex $val 1]]
	if {$i != $last} {
	    append line ", "
	}
	if {[string length $line] > 60} {
	    puts $f $line
	    set line "    "
	}
    }
    puts $f $line
    puts $f "};

static int compSecondList\[\]\[2\] = {"
    set line "    "
    set last [expr {[llength $comp_second_list] - 1}]
    for {set i 0} {$i <= $last} {incr i} {
	set val [lindex $comp_second_list $i]

	append line [format "{%d, %d}" [lindex $val 0] [lindex $val 1]]
	if {$i != $last} {
	    append line ", "
	}
	if {[string length $line] > 60} {
	    puts $f $line
	    set line "    "
	}
    }
    puts $f $line
    puts $f "};

/*
 * Compositions matrix
 */

static int compBothList\[[llength $comp_x_list]\]\[[llength $comp_y_list]\] = {"
    set lastx [expr {[llength $comp_x_list] - 1}]
    set lasty [expr {[llength $comp_y_list] - 1}]
    for {set i 0} {$i <= $lastx} {incr i} {
	puts $f "    \{"
	set line "        "
	for {set j 0} {$j <= $lasty} {incr j} {
	    set comp [list [lindex $comp_x_list $i] [lindex $comp_y_list $j]]
	    if {[info exists comp_map($comp)]} {
		set val $comp_map($comp)
	    } else {
		set val 0
	    }
	    
	    append line [format "%d" $val]
	    if {$j != $lasty} {
		append line ", "
	    }
	    if {[string length $line] > 70} {
		puts $f $line
		set line "        "
	    }
	}
	puts $f $line
	if {$j != $lasty} {
	    puts $f "    \},"
	} else {
	    puts $f "    \}"
	}
    }
    puts $f "};


#define GetUniCharCompInfo(ch) (compGroupMap\[(compPageMap\[(((int)(ch)) & 0xffff) >> COMP_OFFSET_BITS\] << COMP_OFFSET_BITS) | ((ch) & ((1 << COMP_OFFSET_BITS)-1))\])

#define CompSingleMask (1 << 16)
#define CompMask ((1 << 16) - 1)
"

    close $f
}

uni::main

return