X-Git-Url: http://git.indexdata.com/?p=yaz-moved-to-github.git;a=blobdiff_plain;f=src%2Fcharconv.tcl;h=8f8cc9e94b49621254f234a772a7c95836c0261a;hp=7d1fd07474d42834b49e5fddd9dd43b4be6fd529;hb=8cb8947e3a7bff4dbf8f124871cb4905df1adce7;hpb=d30e034fc508f9d901f0fa60113eb160e2d7406f

diff --git a/src/charconv.tcl b/src/charconv.tcl
old mode 100755
new mode 100644
index 7d1fd07..8f8cc9e
--- a/src/charconv.tcl
+++ b/src/charconv.tcl
@@ -1,28 +1,36 @@
-#!/bin/sh
-# the next line restats using tclsh \
-exec tclsh "$0" "$@"
-#
-# $Id: charconv.tcl,v 1.5 2004-03-17 11:00:04 adam Exp $
+#!/usr/bin/tclsh
 
 proc usage {} {
     puts {charconv.tcl: [-p prefix] [-s split] [-o ofile] file ... }
     exit 1
 }
 
-proc preamble_trie {ofilehandle} {
+proc preamble_trie {ofilehandle ifiles ofile} {
     set f $ofilehandle
 
-    set totype {unsigned short}
+    set totype {unsigned }
 
+    puts $f "/** \\file $ofile"
+    puts $f "    \\brief Character conversion, generated from [lindex $ifiles 0]"
+    puts $f ""
+    puts $f "    Generated automatically by charconv.tcl"
+    puts $f "*/"
     puts $f "\#include <string.h>"
+
+    puts $f "\#if HAVE_CONFIG_H"
+    puts $f "\#include <config.h>"
+    puts $f "\#endif"
+
     puts $f "
         struct yaz_iconv_trie_flat {
             char from\[6\];
-            $totype to;
+            unsigned combining : 1;
+            $totype to : 24;
         };
         struct yaz_iconv_trie_dir {
-            short ptr;
-            $totype to;
+            int ptr : 15;
+            unsigned combining : 1;
+            $totype to : 24;
         };
         
         struct yaz_iconv_trie {
@@ -32,24 +40,28 @@ proc preamble_trie {ofilehandle} {
     "
     puts $f {
         static unsigned long lookup(struct yaz_iconv_trie **ptrs, int ptr, unsigned char *inp,
-                                    size_t inbytesleft, size_t *no_read)
+                                    size_t inbytesleft, size_t *no_read, int *combining, unsigned mask, unsigned int boffset)
         {
-            struct yaz_iconv_trie *t = (ptr >= 0) ? ptrs[ptr] : 0;
-            if (!t || inbytesleft < 1)
+            struct yaz_iconv_trie *t = ptrs[ptr-1];
+            if (inbytesleft < 1)
                 return 0;
             if (t->dir)
             {
-                size_t ch = inp[0] & 0xff;
-                unsigned long code =
-                lookup(ptrs, t->dir[ch].ptr, inp+1, inbytesleft-1, no_read);
-                if (code)
+                size_t ch = (inp[0] & mask) + boffset;
+                unsigned long code;
+                if (t->dir[ch].ptr)
                 {
-                    (*no_read)++;
-                    return code;
-                }
+                    code = lookup(ptrs, t->dir[ch].ptr, inp+1, inbytesleft-1, no_read, combining, mask, boffset);
+		    if (code)
+		    {
+			(*no_read)++;
+			return code;
+		    }
+		}
                 if (t->dir[ch].to)
                 {
                     code = t->dir[ch].to;
+		    *combining = t->dir[ch].combining;
                     *no_read = 1;
                     return code;
                 }
@@ -57,14 +69,21 @@ 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)
                     {
-                        if (memcmp(flat->from, inp, len) == 0)
+                        size_t i;
+			for (i = 0; i < len; i++)
+			{
+			    if (((unsigned char *) flat->from)[i] != (inp[i] & mask) + boffset)
+			        break;
+		        }
+                        if (i == len)
                         {
                             *no_read = len;
+			    *combining = flat->combining;
                             return flat->to;
                         }
                     }
@@ -90,7 +109,7 @@ proc reset_trie {} {
     set trie(prefix) {}
 }
 
-proc ins_trie {from to} {
+proc ins_trie {from to combining codename} {
     global trie
     if {![info exists trie(no)]} {
         set trie(no) 1
@@ -101,7 +120,7 @@ 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 $codename 0
 }
 
 proc split_trie {this} {
@@ -110,6 +129,8 @@ 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 codename [lindex $e 3]
         
         set ch [lindex $from 0]
         set rest [lrange $from 1 end]
@@ -119,27 +140,40 @@ proc split_trie {this} {
                 set trie($this,ptr,$ch) $trie(no)
                 incr trie(no)
             }
-            ins_trie_r $rest $to $trie($this,ptr,$ch)
+            ins_trie_r $rest $to $combining $codename $trie($this,ptr,$ch)
         } else {
             set trie($this,to,$ch) $to
+            set trie($this,combining,$ch) $combining
+            set trie($this,codename,$ch) $codename
         }
     }
     set trie($this,content) missing
 }
 
-proc ins_trie_r {from to this} {
+proc ins_trie_r {from to combining codename 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 $to]
+	set dup 0
+	if {[info exists trie($this,content)]} {
+	    foreach e $trie($this,content) {
+		set efrom [lindex $e 0]
+		if { $efrom == $from } {
+		    set dup 1
+		}
+	    }
+	}
+	if { $dup == 0 } {
+	    lappend trie($this,content) [list $from $to $combining $codename]
+	}
         
         # 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 $codename $this]
         }
     } else {
         set ch [lindex $from 0]
@@ -150,9 +184,13 @@ proc ins_trie_r {from to this} {
                 set trie($this,ptr,$ch) $trie(no)
                 incr trie(no)
             }
-            ins_trie_r $rest $to $trie($this,ptr,$ch)
+            ins_trie_r $rest $to $combining $codename $trie($this,ptr,$ch)
         } else {
-            set trie($this,to,$ch) $to
+	    if {![info exist trie($this,to,$ch)]} {
+                set trie($this,to,$ch) $to
+                set trie($this,combining,$ch) $combining
+                set trie($this,codename,$ch) $codename
+	    }
         }
     }
 }
@@ -174,10 +212,11 @@ proc dump_trie {ofilehandle} {
                 foreach d [lindex $m 0] {
                     puts -nonewline $f "\\x$d"
                 }
-                puts -nonewline $f "\", 0x[lindex $m 1]"
-                puts $f "\},"
+                puts -nonewline $f "\", [lindex $m 2], 0x[lindex $m 1]"
+		set v [lindex $m 3]
+                puts $f "\}, /* $v */"
             }
-            puts $f "  \{0, 0\}"
+            puts $f "  \{\"\", 0, 0\}"
             puts $f "\};"
             puts $f "struct yaz_iconv_trie $trie(prefix)page${this} = \{"
             puts $f "  $trie(prefix)page${this}_flat, 0"
@@ -189,10 +228,15 @@ proc dump_trie {ofilehandle} {
                 set ch [format %02X $i]
                 set null 1
                 if {[info exist trie($this,ptr,$ch)]} {
-                    puts -nonewline $f "$trie($this,ptr,$ch), "
+                    puts -nonewline $f "[expr $trie($this,ptr,$ch)+1], "
                     set null 0
                 } else {
-                    puts -nonewline $f "-1, "
+                    puts -nonewline $f "0, "
+                }
+                if {[info exist trie($this,combining,$ch)]} {
+                    puts -nonewline $f "$trie($this,combining,$ch), "
+                } else {
+                    puts -nonewline $f "0, "
                 }
                 if {[info exist trie($this,to,$ch)]} {
                     puts -nonewline $f "0x$trie($this,to,$ch)\}"
@@ -200,8 +244,9 @@ proc dump_trie {ofilehandle} {
                 } else {
                     puts -nonewline $f "0\}"
                 }
-                if {!$null} {
-                    puts -nonewline $f " /* $ch */"
+		if {[info exist trie($this,codename,$ch)]} {
+		    set v $trie($this,codename,$ch)
+                    puts -nonewline $f " /* $v */"
                 }
                 if {$i < 255} {
                     puts $f ","
@@ -224,65 +269,101 @@ proc dump_trie {ofilehandle} {
     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 mask, unsigned int boffset)
         {
             unsigned long code;
             
-            code = lookup($trie(prefix)ptrs, 0, inp, inbytesleft, no_read);
+            code = lookup($trie(prefix)ptrs, 1, inp, inbytesleft, no_read, combining, mask, boffset);
             if (!code)
             {
                 *no_read = 1;
-                code = *inp;
             }
             return code;
         }
     "
 }
 
-proc readfile {fname ofilehandle prefix omits} {
+proc readfile {fname ofilehandle prefix omits reverse} {
     global trie
 
     set marc_lines 0
     set ucs_lines 0
+    set utf_lines 0
+    set altutf_lines 0
+    set codename_lines 0
     set lineno 0
     set f [open $fname r]
     set tablenumber x
+    set combining 0
+    set codename {}
+    set altutf {}
     while {1} {
         incr lineno
         set cnt [gets $f line]
         if {$cnt < 0} {
             break
         }
-	if {[regexp {<entitymap>} $line s]} {
-	    reset_trie
-	    set trie(prefix) "${prefix}"
-	} elseif {[regexp {</entitymap>} $line s]} {
+	if {[regexp {</characterSet>} $line s]} {
 	    dump_trie $ofilehandle
-	} elseif {[regexp {<character hex="([^\"]*)".*<unientity>([0-9A-Fa-f]*)</unientity>} $line s hex ucs]} {
-	    ins_trie $hex $ucs
-	    unset hex
-	} elseif {[regexp {<codeTable number="([0-9]+)"} $line s tablenumber]} {
+	} elseif {[regexp {<characterSet .*ISOcode="([0-9A-Fa-f]+)"} $line s tablenumber]} {
 	    reset_trie
 	    set trie(prefix) "${prefix}_$tablenumber"
-	} elseif {[regexp {</codeTable>} $line s]} {
-	    if {[lsearch $omits $tablenumber] == -1} {
-		dump_trie $ofilehandle
-	    }
+	    set combining 0
 	} elseif {[regexp {</code>} $line s]} {
 	    if {[string length $ucs]} {
-		for {set i 0} {$i < [string length $marc]} {incr i 2} {
-		    lappend hex [string range $marc $i [expr $i+1]]
+		if {$reverse} {
+		    for {set i 0} {$i < [string length $utf]} {incr i 2} {
+			lappend hex [string range $utf $i [expr $i+1]]
+		    }
+		    # puts "ins_trie $hex $marc
+		    ins_trie $hex $marc $combining $codename
+		    unset hex
+
+		} else {
+		    for {set i 0} {$i < [string length $marc]} {incr i 2} {
+			lappend hex [string range $marc $i [expr $i+1]]
+		    }
+		    # puts "ins_trie $hex $ucs"
+		    ins_trie $hex $ucs $combining $codename
+		    unset hex
+		}
+	    }
+	    if {$reverse && [string length $marc]} {
+		for {set i 0} {$i < [string length $altutf]} {incr i 2} {
+		    lappend hex [string range $altutf $i [expr $i+1]]
+		}
+		if {[info exists hex]} {
+		    ins_trie $hex $marc $combining $codename
+		    unset hex
 		}
-		# puts "ins_trie $hex $ucs"
-		ins_trie $hex $ucs
-		unset hex
 	    }
 	    set marc {}
 	    set uni {}
+	    set codename {}
+	    set combining 0
+	    set altutf {}
 	} elseif {[regexp {<marc>([0-9A-Fa-f]*)</marc>} $line s marc]} {
 	    incr marc_lines
+	} elseif {[regexp {<name>(.*)</name>} $line s codename]} {
+	    incr codename_lines
+	} elseif {[regexp {<name>(.*)} $line s codename]} {
+	    incr codename_lines
+	    incr lineno
+	    set cnt [gets $f line]
+	    if {$cnt < 0} {
+		break
+	    }
+	    if {[regexp {(.*)</name>} $line s codename_ex]} {
+		set codename "${codename} ${codename_ex}"
+	    }
+	} elseif {[regexp {<isCombining>true</isCombining>} $line s]} {
+	    set combining 1
 	} elseif {[regexp {<ucs>([0-9A-Fa-f]*)</ucs>} $line s ucs]} {
 	    incr ucs_lines
+	} elseif {[regexp {<utf-8>([0-9A-Fa-f]*)</utf-8>} $line s utf]} {
+	    incr utf_lines
+	} elseif {[regexp {<altutf-8>([0-9A-Fa-f]*)</altutf-8>} $line s altutf]} {
+	    incr altutf_lines
 	}
     }
     close $f
@@ -292,6 +373,7 @@ set verbose 0
 set ifile {}
 set ofile out.c
 set prefix {c}
+set reverse_map 0
 # Parse command line
 set l [llength $argv]
 set i 0
@@ -326,6 +408,9 @@ while {$i < $l} {
             }
             lappend omits $arg
 	}
+	-r {
+	    set reverse_map 1
+	}
         default {
 	    lappend ifiles $arg
         }
@@ -337,12 +422,14 @@ if {![info exists ifiles]} {
     usage
 }
 
-set ofilehandle [open $ofile w]
-preamble_trie $ofilehandle
+set ofilehandle [open ${ofile}.tmp w]
+preamble_trie $ofilehandle $ifiles $ofile
 
 foreach ifile $ifiles {
-    readfile $ifile $ofilehandle $prefix $omits
+    readfile $ifile $ofilehandle $prefix $omits $reverse_map
 }
 close $ofilehandle
 
+file rename -force ${ofile}.tmp ${ofile}
+