read CaseMappingSpecials

jruby · Dec 22, 2017 · 6775574 · 6775574
1 parent d0de1b3
commit 6775574
Showing 4 changed files with 36 additions and 7 deletions.
diff --git a/resources/tables/CaseMappingSpecials.bin b/resources/tables/CaseMappingSpecials.bin
diff --git a/scripts/generate.rb b/scripts/generate.rb
@@ -8,11 +8,11 @@
 INDENT = " " * 4
 
 def generate_data
-    # generate_encoding_list
-    # generate_transcoder_list
-    # generate_transoder_data
-    # generate_coderange_data
-    # generate_coderange_list
+    generate_encoding_list
+    generate_transcoder_list
+    generate_transoder_data
+    generate_coderange_data
+    generate_coderange_list
     generate_fold_data
 end
 
@@ -245,9 +245,9 @@ def generate_fold_data
 
         when /CaseMappingSpecials/
             open("#{DST_BIN_DIR}/CaseMappingSpecials.bin", "wb") do |f|
-                size = src[/CaseMappingSpecials\[\]\s+=\s+\{(.*?)\}\;/m, 1].split(',').size
+                size =  src[/CaseMappingSpecials\[\]\s+=\s+\{(.*?)\}\;/m, 1].scan(/0x\d+/).size
                 f << [size].pack("N")
-                address.step(address + (size * 4 - 1), 4).each do |adr|
+                address.step(address + (size * 4), 4).each do |adr|
                     f << binary[adr, 4].unpack("l").pack("N")
                 end
             end

diff --git a/src/org/jcodings/Config.java b/src/org/jcodings/Config.java
@@ -40,6 +40,8 @@ public interface Config {
     final int SpecialIndexWidth =                   10;
     final int SpecialIndexMask =                    ((1 << SpecialIndexWidth) - 1) << SpecialIndexShift;
 
+    final int SpecialsLengthOffset =                25;
+
     final int CASE_UPCASE =                         (1<<13); /* has/needs uppercase mapping */
     final int CASE_DOWNCASE =                       (1<<14); /* has/needs lowercase mapping */
     final int CASE_TITLECASE =                      (1<<15); /* has/needs (special) titlecase mapping */

diff --git a/src/org/jcodings/unicode/UnicodeEncoding.java b/src/org/jcodings/unicode/UnicodeEncoding.java
@@ -21,6 +21,7 @@
 
 import java.io.DataInputStream;
 import java.io.IOException;
+import java.util.ArrayList;
 
 import org.jcodings.ApplyAllCaseFoldFunction;
 import org.jcodings.CaseFoldCodeItem;
@@ -631,4 +632,30 @@ private static IntArrayHash<CodeList> initializeUnfold3Hash() {
 
         static final IntArrayHash<CodeList> Hash = initializeUnfold3Hash();
     }
+
+    private static class CaseMappingSpecials {
+        static ArrayList<int[]> read() {
+            try {
+                DataInputStream dis = ArrayReader.openStream("CaseMappingSpecials");
+                int size = dis.readInt();
+                ArrayList<int[]> values = new ArrayList<int[]>();
+                for (int i = 0; i < size; i++) {
+                    int packed = dis.readInt();
+                    int length = packed >>> Config.SpecialsLengthOffset;
+                    int[]codes = new int[length];
+                    codes[0] = packed & ((1 << Config.SpecialsLengthOffset) - 1);
+                    for (int j = 1; j < length; j++) {
+                        i++;
+                        codes[j] = dis.readInt();
+                    }
+                    values.add(codes);
+                }
+                return values;
+            } catch (IOException ioe) {
+                throw new RuntimeException(ioe);
+            }
+        }
+
+        private static ArrayList<int[]> Values = read();
+    }
 }