diff --git a/tools/unicodetools/com/ibm/text/UCD/ConvertUCD.java b/tools/unicodetools/com/ibm/text/UCD/ConvertUCD.java index 67abe786fb1..44b59a229eb 100644 --- a/tools/unicodetools/com/ibm/text/UCD/ConvertUCD.java +++ b/tools/unicodetools/com/ibm/text/UCD/ConvertUCD.java @@ -5,8 +5,8 @@ ******************************************************************************* * * $Source: /xsrl/Nsvn/icu/unicodetools/com/ibm/text/UCD/ConvertUCD.java,v $ -* $Date: 2004/03/11 19:03:18 $ -* $Revision: 1.15 $ +* $Date: 2004/11/12 23:17:15 $ +* $Revision: 1.16 $ * ******************************************************************************* */ @@ -864,7 +864,9 @@ public final class ConvertUCD implements UCD_Types { } else if (fieldName.equals("jt")) { uData.joiningType = Utility.lookup(fieldValue, UCD_Names.JOINING_TYPE, true); } else if (fieldName.equals("jg")) { - uData.joiningGroup = Utility.lookup(fieldValue, UCD_Names.OLD_JOINING_GROUP, true); + byte temp = (byte)Utility.find(fieldValue, UCD_Names.OLD_JOINING_GROUP, true); + if (temp != -1) uData.joiningGroup = temp; + else uData.joiningGroup = Utility.lookup(fieldValue, UCD_Names.JOINING_GROUP, true); } else if (fieldName.equals("nv")) { if (major < 2) { diff --git a/tools/unicodetools/com/ibm/text/UCD/MakeUnicodeFiles.java b/tools/unicodetools/com/ibm/text/UCD/MakeUnicodeFiles.java index ab7c535d54d..94b31ab1b75 100644 --- a/tools/unicodetools/com/ibm/text/UCD/MakeUnicodeFiles.java +++ b/tools/unicodetools/com/ibm/text/UCD/MakeUnicodeFiles.java @@ -39,7 +39,7 @@ import com.ibm.text.utility.Utility; import com.ibm.icu.text.Collator; public class MakeUnicodeFiles { - public static int dVersion = 8; // change to fix the generated file D version. If less than zero, no "d" + public static int dVersion = 6; // change to fix the generated file D version. If less than zero, no "d" /*static String[] hackNameList = { "noBreak", "Arabic_Presentation_Forms-A", "Arabic_Presentation_Forms-B", @@ -476,6 +476,9 @@ public class MakeUnicodeFiles { } public static void generateFile() throws IOException { + String[] lines = new String[2]; + Utility.filesAreIdentical("C:\\DATA\\UCD\\4.0.1-Update\\CaseFolding-4.0.1.txt", + "C:\\DATA\\GEN\\DerivedData\\CaseFolding-4.1.0d13.txt", lines); for (int i = 0; i < Format.theFormat.filesToDo.length; ++i) { String fileName = Format.theFormat.filesToDo[i].trim().toLowerCase( diff --git a/tools/unicodetools/com/ibm/text/UCD/MakeUnicodeFiles.txt b/tools/unicodetools/com/ibm/text/UCD/MakeUnicodeFiles.txt index 3fe1faaa967..24219f5ee85 100644 --- a/tools/unicodetools/com/ibm/text/UCD/MakeUnicodeFiles.txt +++ b/tools/unicodetools/com/ibm/text/UCD/MakeUnicodeFiles.txt @@ -1,5 +1,5 @@ Generate: -DeltaVersion: 13 +DeltaVersion: 7 File: Blocks Property: Block diff --git a/tools/unicodetools/com/ibm/text/UCD/UCD.java b/tools/unicodetools/com/ibm/text/UCD/UCD.java index 9393579e472..d9884bc8738 100644 --- a/tools/unicodetools/com/ibm/text/UCD/UCD.java +++ b/tools/unicodetools/com/ibm/text/UCD/UCD.java @@ -5,8 +5,8 @@ ******************************************************************************* * * $Source: /xsrl/Nsvn/icu/unicodetools/com/ibm/text/UCD/UCD.java,v $ -* $Date: 2004/06/26 00:26:16 $ -* $Revision: 1.34 $ +* $Date: 2004/11/12 23:17:15 $ +* $Revision: 1.35 $ * ******************************************************************************* */ @@ -42,7 +42,7 @@ public final class UCD implements UCD_Types { /** * Used for the default version. */ - public static final String latestVersion = "4.0.1"; + public static final String latestVersion = "4.1.0"; /** * Create singleton instance for default (latest) version diff --git a/tools/unicodetools/com/ibm/text/UCD/UCD_Names.java b/tools/unicodetools/com/ibm/text/UCD/UCD_Names.java index cba0feb14d4..d4142a89df9 100644 --- a/tools/unicodetools/com/ibm/text/UCD/UCD_Names.java +++ b/tools/unicodetools/com/ibm/text/UCD/UCD_Names.java @@ -5,8 +5,8 @@ ******************************************************************************* * * $Source: /xsrl/Nsvn/icu/unicodetools/com/ibm/text/UCD/UCD_Names.java,v $ -* $Date: 2004/03/11 19:03:16 $ -* $Revision: 1.27 $ +* $Date: 2004/11/12 23:17:15 $ +* $Revision: 1.28 $ * ******************************************************************************* */ @@ -827,63 +827,61 @@ final class UCD_Names implements UCD_Types { } public static String[] OLD_JOINING_GROUP = { - "", - "AIN", - "ALAPH", - "ALEF", - "BEH", - "BETH", - "DAL", - "DALATH_RISH", - "E", - "FEH", - "FINAL_SEMKATH", - "GAF", - "GAMAL", - "HAH", - "HAMZA_ON_HEH_GOAL", - "HE", - "HEH", - "HEH_GOAL", - "HETH", - "KAF", - "KAPH", - "KNOTTED_HEH", - "LAM", - "LAMADH", - "MEEM", - "MIM", - "NOON", - "NUN", - "PE", - "QAF", - "QAPH", - "REH", - "REVERSED_PE", - "SAD", - "SADHE", - "SEEN", - "SEMKATH", - "SHIN", - "SWASH_KAF", - "TAH", - "TAW", - "TEH_MARBUTA", - "TETH", - "WAW", - "SYRIAC WAW", - "YEH", - "YEH_BARREE", - "YEH_WITH_TAIL", - "YUDH", - "YUDH_HE", - "ZAIN", - "ZHAIN", - "KHAPH", - "FE", - }; - - + "", + "AIN", + "ALAPH", + "ALEF", + "BAA", + "BETH", + "DAL", + "DALATH_RISH", + "E", + "FA", + "FINAL_SEMKATH", + "GAF", + "GAMAL", + "HAA", + "HAMZAH_ON_HA_GOAL", + "HE", + "HA", + "HA_GOAL", + "HETH", + "CAF", + "KAPH", + "KNOTTED_HA", + "LAM", + "LAMADH", + "MEEM", + "MIM", + "NOON", + "NUN", + "PE", + "QAF", + "QAPH", + "RA", + "REVERSED_PE", + "SAD", + "SADHE", + "SEEN", + "SEMKATH", + "SHIN", + "SWASH_CAF", + "TAH", + "TAW", + "TAA_MARBUTAH", + "TETH", + "WAW", + "SYRIAC WAW", + "YA", + "YA_BARREE", + "ALEF_MAQSURAH", + "YUDH", + "YUDH_HE", + "ZAIN", + "ZHAIN", + "KHAPH", + "FE", + }; static String[] JAMO_L_TABLE = { // Value; Short Name; Unicode Name diff --git a/tools/unicodetools/com/ibm/text/UCD/UnicodeInvariants.txt b/tools/unicodetools/com/ibm/text/UCD/UnicodeInvariants.txt index 6aeb535bb51..3f44524e5be 100644 --- a/tools/unicodetools/com/ibm/text/UCD/UnicodeInvariants.txt +++ b/tools/unicodetools/com/ibm/text/UCD/UnicodeInvariants.txt @@ -110,6 +110,7 @@ Let $ZWJ = [\u200D] # [\N{ZERO WIDTH JOINER}] Let $gcAllPunctuation = [$gc:Open_Punctuation $gc:Close_Punctuation $gc:Dash_Punctuation $gc:Connector_Punctuation $gc:Other_Punctuation $gc:Initial_Punctuation $gc:Final_Punctuation] Let $gcAllSymbols = [$gc:Currency_Symbol $gc:Modifier_Symbol $gc:Math_Symbol $gc:Other_Symbol] Let $gcAllMarks = [$gc:Nonspacing_Mark $gc:Enclosing_Mark $gc:Spacing_Mark] +Let $strange = [\u24B6-\u24CF] # Unassigned, Control, Format, Private_Use, Surrogate, # Uppercase_Letter, Lowercase_Letter, Titlecase_Letter, Modifier_Letter, Other_Letter, @@ -121,9 +122,9 @@ Let $gcAllMarks = [$gc:Nonspacing_Mark $gc:Enclosing_Mark $gc:Spacing_Mark] # UTS Rules -Let $alpha = [$Alphabetic $Lowercase $Uppercase $ZWNJ $ZWJ] +Let $alpha = [$Alphabetic $Lowercase] # $Uppercase $ZWNJ $ZWJ] Let $lower = $Lowercase -Let $upper = $Uppercase +Let $upper = [$Uppercase - $strange] Let $punct = [$gcAllPunctuation $gcAllSymbols - $alpha] Let $digit = $gc:Decimal_Number Let $xdigit = [$gc:Decimal_Number $Hex_Digit] # in both! @@ -131,7 +132,7 @@ Let $alnum = [$alpha $digit] Let $space = $Whitespace Let $blank = [$Whitespace - [$LF $VTAB $FF $CR $NEL $gc:Line_Separator $gc:Paragraph_Separator]] Let $cntrl = $gc:Control -Let $graph = [[^$space $gc:Control $gc:Format $gc:Surrogate $gc:Unassigned] $ZWNJ $ZWJ] +Let $graph = [^$space $gc:Control $gc:Format $gc:Surrogate $gc:Unassigned] # $ZWNJ $ZWJ] Let $print = [$graph $blank - $cntrl] Let $word = [$alpha $gcAllMarks $digit $gc:Connector_Punctuation] diff --git a/tools/unicodetools/com/ibm/text/utility/UnicodeDataFile.java b/tools/unicodetools/com/ibm/text/utility/UnicodeDataFile.java index b0a1494ef9a..fd47781d715 100644 --- a/tools/unicodetools/com/ibm/text/utility/UnicodeDataFile.java +++ b/tools/unicodetools/com/ibm/text/utility/UnicodeDataFile.java @@ -50,8 +50,8 @@ public class UnicodeDataFile { } public void close() throws IOException { - Utility.renameIdentical(mostRecent, Utility.getOutputName(newFile), batName); out.close(); + Utility.renameIdentical(mostRecent, Utility.getOutputName(newFile), batName); } public static String generateDateLine() { diff --git a/tools/unicodetools/com/ibm/text/utility/Utility.java b/tools/unicodetools/com/ibm/text/utility/Utility.java index cfc11c097a7..4dfc23fa620 100644 --- a/tools/unicodetools/com/ibm/text/utility/Utility.java +++ b/tools/unicodetools/com/ibm/text/utility/Utility.java @@ -5,8 +5,8 @@ ******************************************************************************* * * $Source: /xsrl/Nsvn/icu/unicodetools/com/ibm/text/utility/Utility.java,v $ -* $Date: 2004/10/14 17:54:56 $ -* $Revision: 1.44 $ +* $Date: 2004/11/12 23:17:15 $ +* $Revision: 1.45 $ * ******************************************************************************* */ @@ -914,8 +914,23 @@ public final class Utility implements UCD_Types { // COMMON UTILITIES System.out.println("Null file"); return false; } - - boolean identical = false; + String lines[] = new String[2]; + boolean identical = filesAreIdentical(file1, file2, lines); + if (identical) { + renameIdentical(file2); + if (batFile != null) renameIdentical(batFile); + return true; + } else { + fixDot(); + System.out.println("Found difference in : " + file1 + ", " + file2); + int diff = compare(lines[0], lines[1]); + System.out.println(" File1: '" + lines[0].substring(0,diff) + "', '" + lines[0].substring(diff) + "'"); + System.out.println(" File2: '" + lines[1].substring(0,diff) + "', '" + lines[1].substring(diff) + "'"); + return false; + } + } + + public static boolean filesAreIdentical(String file1, String file2, String[] lines) throws IOException { BufferedReader br1 = new BufferedReader(new FileReader(file1), 32*1024); BufferedReader br2 = new BufferedReader(new FileReader(file2), 32*1024); String line1 = ""; @@ -925,31 +940,22 @@ public final class Utility implements UCD_Types { // COMMON UTILITIES line1 = getLineWithoutFluff(br1, lineCount == 0); line2 = getLineWithoutFluff(br2, lineCount == 0); if (line1 == null) { - if (line2 == null) identical = true; + if (line2 == null) return true; break; } if (!line1.equals(line2)) { break; } } + lines[0] = line1; + lines[1] = line2; + if (lines[0] == null) lines[0] = ""; + if (lines[1] == null) lines[1] = ""; + return false; } finally { br1.close(); br2.close(); } - if (identical) { - renameIdentical(file2); - if (batFile != null) renameIdentical(batFile); - return true; - } else { - if (line1 == null) line1 = ""; - if (line2 == null) line2 = ""; - fixDot(); - System.out.println("Found difference in : " + file1 + ", " + file2); - int diff = compare(line1, line2); - System.out.println(" Line1: '" + line1.substring(0,diff) + "', '" + line1.substring(diff)); - System.out.println(" Line2: '" + line2.substring(0,diff) + "', '" + line2.substring(diff)); - return false; - } } static void renameIdentical(String file2) { @@ -975,6 +981,7 @@ public final class Utility implements UCD_Types { // COMMON UTILITIES if (line1.equals("#")) continue; if (line1.startsWith("# Generated")) continue; if (line1.startsWith("# Date")) continue; + if (line1.equals("# ================================================")) continue; if (first && line1.startsWith("#")) { first = false; continue;