X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=src%2Fcharconv.tcl;h=d49f23855946f8a1bc105a660ed46f63ed37b779;hb=60a702f390f7e2addfdab79f2328db3ba2897c8b;hp=5ad4be18d95a2597d97b5820fdc4d82717aeab4e;hpb=e3678b79cc201f61c67f46f558cfdbcb4a999ba3;p=yaz-moved-to-github.git

diff --git a/src/charconv.tcl b/src/charconv.tcl
index 5ad4be1..d49f238 100755
--- a/src/charconv.tcl
+++ b/src/charconv.tcl
@@ -2,7 +2,7 @@
 # the next line restats using tclsh \
 exec tclsh "$0" "$@"
 #
-# $Id: charconv.tcl,v 1.4 2004-03-16 13:13:34 adam Exp $
+# $Id: charconv.tcl,v 1.7 2004-08-07 08:06:57 adam Exp $
 
 proc usage {} {
     puts {charconv.tcl: [-p prefix] [-s split] [-o ofile] file ... }
@@ -17,11 +17,13 @@ proc preamble_trie {ofilehandle} {
     puts $f "\#include <string.h>"
     puts $f "
         struct yaz_iconv_trie_flat {
-            char *from;
+            char from\[6\];
+            unsigned combining : 1;
             $totype to;
         };
         struct yaz_iconv_trie_dir {
-            struct yaz_iconv_trie *ptr;
+            short ptr : 15;
+            unsigned combining : 1;
             $totype to;
         };
         
@@ -31,16 +33,17 @@ proc preamble_trie {ofilehandle} {
         };
     "
     puts $f {
-        static unsigned long lookup(struct yaz_iconv_trie *t, unsigned char *inp,
-                                    size_t inbytesleft, size_t *no_read)
+        static unsigned long lookup(struct yaz_iconv_trie **ptrs, int ptr, unsigned char *inp,
+                                    size_t inbytesleft, size_t *no_read, int *combining)
         {
+            struct yaz_iconv_trie *t = (ptr >= 0) ? ptrs[ptr] : 0;
             if (!t || inbytesleft < 1)
-            return 0;
+                return 0;
             if (t->dir)
             {
                 size_t ch = inp[0] & 0xff;
                 unsigned long code =
-                lookup(t->dir[ch].ptr, inp+1, inbytesleft-1, no_read);
+                lookup(ptrs, t->dir[ch].ptr, inp+1, inbytesleft-1, no_read, combining);
                 if (code)
                 {
                     (*no_read)++;
@@ -49,6 +52,7 @@ proc preamble_trie {ofilehandle} {
                 if (t->dir[ch].to)
                 {
                     code = t->dir[ch].to;
+		    *combining = t->dir[ch].combining;
                     *no_read = 1;
                     return code;
                 }
@@ -56,7 +60,7 @@ proc preamble_trie {ofilehandle} {
             else
             {
                 struct yaz_iconv_trie_flat *flat = t->flat;
-                while (flat->from)
+                while (flat->to)
                 {
                     size_t len = strlen(flat->from);
                     if (len <= inbytesleft)
@@ -64,6 +68,7 @@ proc preamble_trie {ofilehandle} {
                         if (memcmp(flat->from, inp, len) == 0)
                         {
                             *no_read = len;
+			    *combining = flat->combining;
                             return flat->to;
                         }
                     }
@@ -72,7 +77,6 @@ proc preamble_trie {ofilehandle} {
             }
             return 0;
         }
-        
     }
 }
 
@@ -86,11 +90,11 @@ proc reset_trie {} {
     set trie(no) 1
     set trie(size) 0
     set trie(max) 0
-    set trie(split) 40
+    set trie(split) 50
     set trie(prefix) {}
 }
 
-proc ins_trie {from to} {
+proc ins_trie {from to combining} {
     global trie
     if {![info exists trie(no)]} {
         set trie(no) 1
@@ -101,7 +105,7 @@ proc ins_trie {from to} {
 	set trie(max) $to
     }
     incr trie(size)
-    ins_trie_r [split $from] $to 0
+    ins_trie_r [split $from] $to $combining 0
 }
 
 proc split_trie {this} {
@@ -110,6 +114,7 @@ proc split_trie {this} {
     foreach e $trie($this,content) {
         set from [lindex $e 0]
         set to [lindex $e 1]
+	set combining [lindex $e 2]
         
         set ch [lindex $from 0]
         set rest [lrange $from 1 end]
@@ -119,27 +124,28 @@ proc split_trie {this} {
                 set trie($this,ptr,$ch) $trie(no)
                 incr trie(no)
             }
-            ins_trie_r $rest $to $trie($this,ptr,$ch)
+            ins_trie_r $rest $to $combining $trie($this,ptr,$ch)
         } else {
             set trie($this,to,$ch) $to
+            set trie($this,combining,$ch) $combining
         }
     }
     set trie($this,content) missing
 }
 
-proc ins_trie_r {from to this} {
+proc ins_trie_r {from to combining this} {
     global trie
 
     if {![info exist trie($this,type)]} {
         set trie($this,type) f
     }
     if {$trie($this,type) == "f"} {
-        lappend trie($this,content) [list $from $to]
+        lappend trie($this,content) [list $from $to $combining]
         
         # split ?
         if {[llength $trie($this,content)] > $trie(split)} {
             split_trie $this
-            return [ins_trie_r $from $to $this]
+            return [ins_trie_r $from $to $combining $this]
         }
     } else {
         set ch [lindex $from 0]
@@ -150,9 +156,10 @@ proc ins_trie_r {from to this} {
                 set trie($this,ptr,$ch) $trie(no)
                 incr trie(no)
             }
-            ins_trie_r $rest $to $trie($this,ptr,$ch)
+            ins_trie_r $rest $to $combining $trie($this,ptr,$ch)
         } else {
             set trie($this,to,$ch) $to
+            set trie($this,combining,$ch) $combining
         }
     }
 }
@@ -174,10 +181,10 @@ proc dump_trie {ofilehandle} {
                 foreach d [lindex $m 0] {
                     puts -nonewline $f "\\x$d"
                 }
-                puts -nonewline $f "\", 0x[lindex $m 1]"
+                puts -nonewline $f "\", [lindex $m 2], 0x[lindex $m 1]"
                 puts $f "\},"
             }
-            puts $f "  \{0, 0\}"
+            puts $f "  \{\"\", 0\}"
             puts $f "\};"
             puts $f "struct yaz_iconv_trie $trie(prefix)page${this} = \{"
             puts $f "  $trie(prefix)page${this}_flat, 0"
@@ -189,9 +196,14 @@ proc dump_trie {ofilehandle} {
                 set ch [format %02X $i]
                 set null 1
                 if {[info exist trie($this,ptr,$ch)]} {
-                    puts -nonewline $f "&$trie(prefix)page$trie($this,ptr,$ch), "
+                    puts -nonewline $f "$trie($this,ptr,$ch), "
                     set null 0
                 } else {
+                    puts -nonewline $f "-1, "
+                }
+                if {[info exist trie($this,combining,$ch)]} {
+                    puts -nonewline $f "$trie($this,combining,$ch), "
+                } else {
                     puts -nonewline $f "0, "
                 }
                 if {[info exist trie($this,to,$ch)]} {
@@ -215,12 +227,20 @@ proc dump_trie {ofilehandle} {
             puts $f "\};"
         }
     }
+
+    puts $f "struct yaz_iconv_trie *$trie(prefix)ptrs \[\] = {"
+    for {set this 0} {$this < $trie(no)} {incr this} {
+	puts $f " &$trie(prefix)page$this,"
+    }
+    puts $f "0, };"
+    puts $f ""
+
     puts $f "unsigned long yaz_$trie(prefix)_conv
-            (unsigned char *inp, size_t inbytesleft, size_t *no_read)
+            (unsigned char *inp, size_t inbytesleft, size_t *no_read, int *combining)
         {
             unsigned long code;
             
-            code = lookup(&$trie(prefix)page0, inp, inbytesleft, no_read);
+            code = lookup($trie(prefix)ptrs, 0, inp, inbytesleft, no_read, combining);
             if (!code)
             {
                 *no_read = 1;
@@ -239,6 +259,7 @@ proc readfile {fname ofilehandle prefix omits} {
     set lineno 0
     set f [open $fname r]
     set tablenumber x
+    set combining 0
     while {1} {
         incr lineno
         set cnt [gets $f line]
@@ -251,11 +272,12 @@ proc readfile {fname ofilehandle prefix omits} {
 	} elseif {[regexp {</entitymap>} $line s]} {
 	    dump_trie $ofilehandle
 	} elseif {[regexp {<character hex="([^\"]*)".*<unientity>([0-9A-Fa-f]*)</unientity>} $line s hex ucs]} {
-	    ins_trie $hex $ucs
+	    ins_trie $hex $ucs $combining
 	    unset hex
 	} elseif {[regexp {<codeTable number="([0-9]+)"} $line s tablenumber]} {
 	    reset_trie
 	    set trie(prefix) "${prefix}_$tablenumber"
+	    set combining 0
 	} elseif {[regexp {</codeTable>} $line s]} {
 	    if {[lsearch $omits $tablenumber] == -1} {
 		dump_trie $ofilehandle
@@ -266,13 +288,16 @@ proc readfile {fname ofilehandle prefix omits} {
 		    lappend hex [string range $marc $i [expr $i+1]]
 		}
 		# puts "ins_trie $hex $ucs"
-		ins_trie $hex $ucs
+		ins_trie $hex $ucs $combining
 		unset hex
 	    }
 	    set marc {}
 	    set uni {}
+	    set combining 0
 	} elseif {[regexp {<marc>([0-9A-Fa-f]*)</marc>} $line s marc]} {
 	    incr marc_lines
+	} elseif {[regexp {<isCombining>true</isCombining>} $line s]} {
+	    set combining 1
 	} elseif {[regexp {<ucs>([0-9A-Fa-f]*)</ucs>} $line s ucs]} {
 	    incr ucs_lines
 	}
@@ -283,7 +308,6 @@ proc readfile {fname ofilehandle prefix omits} {
 set verbose 0
 set ifile {}
 set ofile out.c
-set trie(split) 40
 set prefix {c}
 # Parse command line
 set l [llength $argv]