merge trunkrfe_3473670

expand all unicode tables and tools to eventually handle the
unicode chars > 0xffff (if UTF_MAX_CHAR > 3)

2 files changed, 49 insertions, 8 deletions

diff --git a/tools/uniClass.tcl b/tools/uniClass.tcl
index 55aa44c..f741434 100644
--- a/tools/uniClass.tcl
+++ b/tools/uniClass.tcl
@@ -13,22 +13,36 @@ exec tclsh "$0" ${1+"$@"}
 #
 
 proc emitRange {first last} {
-    global ranges numranges chars numchars
+    global ranges numranges chars numchars extchars extranges
 
     if {$first < ($last-1)} {
-	append ranges [format "{0x%04x, 0x%04x}, " \
+	if {!$extranges && ($first) > 0xffff} {
+	    set extranges 1
+	    set numranges 0
+	    set ranges [string trimright $ranges " \n\r\t,"]
+	    append ranges "\n#if MAX_UTF_CHAR > 4\n    ,"
+	}
+	append ranges [format "{0x%x, 0x%x}, " \
 		$first $last]
 	if {[incr numranges] % 4 == 0} {
+	    set ranges [string trimright $ranges]
 	    append ranges "\n    "
 	}
     } else {
-	append chars [format "0x%04x, " $first]
+	if {!$extchars && ($first) > 0xffff} {
+	    set extchars 1
+	    set numchars 0
+	    set chars [string trimright $chars " \n\r\t,"]
+	    append chars "\n#if MAX_UTF_CHAR > 4\n    ,"
+	}
+	append chars [format "0x%x, " $first]
 	incr numchars
 	if {$numchars % 9 == 0} {
+	    set chars [string trimright $chars]
 	    append chars "\n    "
 	}
 	if {$first != $last} {
-	    append chars [format "0x%04x, " $last]
+	    append chars [format "0x%x, " $last]
 	    incr numchars
 	    if {$numchars % 9 == 0} {
 		append chars "\n    "
@@ -38,7 +52,7 @@ proc emitRange {first last} {
 }
 
 proc genTable {type} {
-    global first last ranges numranges chars numchars
+    global first last ranges numranges chars numchars extchars extranges
     set first -2
     set last -2
 
@@ -46,8 +60,14 @@ proc genTable {type} {
     set numranges 0
     set chars "    "
     set numchars 0
+    set extchars 0
+    set extranges 0
 
-    for {set i 0} {$i <= 0xFFFF} {incr i} {
+    for {set i 0} {$i <= 0x10ffff} {incr i} {
+    if {$i == 0xd800} {
+	# Skip surrogates
+	set i 0xdc00
+    }
 	if {[string is $type [format %c $i]]} {
 	    if {$i == ($last + 1)} {
 		set last $i
@@ -63,7 +83,13 @@ proc genTable {type} {
     emitRange $first $last
 
     set ranges [string trimright $ranges "\t\n ,"]
+    if {$extranges} {
+	append ranges "\n#endif"
+    }
     set chars  [string trimright $chars "\t\n ,"]
+    if {$extchars} {
+	append chars "\n#endif"
+    }
     if {$ranges ne ""} {
 	puts "static CONST crange ${type}RangeTable\[\] = {\n$ranges\n};\n"
 	puts "#define NUM_[string toupper $type]_RANGE (sizeof(${type}RangeTable)/sizeof(crange))\n"
diff --git a/tools/uniParse.tcl b/tools/uniParse.tcl
index bddee3e..a7cc1ee 100644
--- a/tools/uniParse.tcl
+++ b/tools/uniParse.tcl
@@ -114,7 +114,7 @@ proc uni::buildTables {data} {
 	set items [split $line \;]
 
 	scan [lindex $items 0] %x index
-	if {$index > 0xffff} then {
+	if {$index > 0x2ffff} then {
 	    # Ignore non-BMP characters, as long as Tcl doesn't support them
 	    continue
 	}
@@ -173,7 +173,7 @@ proc uni::main {} {
 
     buildTables $data
     puts "X = [llength $pMap]  Y= [llength $pages]  A= [llength $groups]"
-    set size [expr {[llength $pMap]*2 + [llength $pages]*(1<<$shift)}]
+    set size [expr {[llength $pMap]*2 + ([llength $pages]<<$shift)}]
     puts "shift = $shift, space = $size"
 
     set f [open [file join [lindex $argv 1] tclUniData.c] w]
@@ -207,6 +207,14 @@ static CONST unsigned short pageMap\[\] = {"
     set line "    "
     set last [expr {[llength $pMap] - 1}]
     for {set i 0} {$i <= $last} {incr i} {
+	if {$i == [expr {0x10000 >> $shift}]} {
+	    set line [string trimright $line " \t,"]
+	    puts $f $line
+	    set lastpage [expr {[lindex $line end] >> $shift}]
+	    puts stdout "lastpage: $lastpage"
+	    puts $f "#if TCL_UTF_MAX > 3"
+	    set line "    ,"
+	}
 	append line [lindex $pMap $i]
 	if {$i != $last} {
 	    append line ", "
@@ -217,6 +225,7 @@ static CONST unsigned short pageMap\[\] = {"
 	}
     }
     puts $f $line
+    puts $f "#endif /* TCL_UTF_MAX > 3 */"
     puts $f "};
 
 /*
@@ -231,6 +240,11 @@ static CONST unsigned char groupMap\[\] = {"
     for {set i 0} {$i <= $lasti} {incr i} {
 	set page [lindex $pages $i]
 	set lastj [expr {[llength $page] - 1}]
+	if {$i == ($lastpage + 1)} {
+	    puts $f [string trimright $line " \t,"]
+	    puts $f "#if TCL_UTF_MAX > 3"
+	    set line "    ,"
+	}
 	for {set j 0} {$j <= $lastj} {incr j} {
 	    append line [lindex $page $j]
 	    if {$j != $lastj || $i != $lasti} {
@@ -243,6 +257,7 @@ static CONST unsigned char groupMap\[\] = {"
 	}
     }
     puts $f $line
+    puts $f "#endif /* TCL_UTF_MAX > 3 */"
     puts $f "};
 
 /*