diff --git a/tools/unicodetools/com/ibm/text/UCA/Main.java b/tools/unicodetools/com/ibm/text/UCA/Main.java index 4f789cfa6be..2dbde2daf74 100644 --- a/tools/unicodetools/com/ibm/text/UCA/Main.java +++ b/tools/unicodetools/com/ibm/text/UCA/Main.java @@ -5,8 +5,8 @@ ******************************************************************************* * * $Source: /xsrl/Nsvn/icu/unicodetools/com/ibm/text/UCA/Main.java,v $ -* $Date: 2002/09/25 06:40:13 $ -* $Revision: 1.10 $ +* $Date: 2002/10/03 22:58:17 $ +* $Revision: 1.11 $ * ******************************************************************************* */ @@ -76,6 +76,10 @@ public class Main { else if (arg.equalsIgnoreCase("writeCaseFolding")) WriteCollationData.writeCaseFolding(); else if (arg.equalsIgnoreCase("javatest")) WriteCollationData.javatest(); else if (arg.equalsIgnoreCase("short")) shortPrint = true; + + else if (arg.equalsIgnoreCase("writeAllocation")) WriteCharts.writeAllocation(); + + else { System.out.println(); System.out.println("UNKNOWN OPTION (" + arg + "): must be one of the following (case-insensitive)"); diff --git a/tools/unicodetools/com/ibm/text/UCA/WriteCharts.java b/tools/unicodetools/com/ibm/text/UCA/WriteCharts.java index 9c8ad32a214..1637d1180e7 100644 --- a/tools/unicodetools/com/ibm/text/UCA/WriteCharts.java +++ b/tools/unicodetools/com/ibm/text/UCA/WriteCharts.java @@ -5,8 +5,8 @@ ******************************************************************************* * * $Source: /xsrl/Nsvn/icu/unicodetools/com/ibm/text/UCA/WriteCharts.java,v $ -* $Date: 2002/09/25 06:40:13 $ -* $Revision: 1.12 $ +* $Date: 2002/10/03 22:58:17 $ +* $Revision: 1.13 $ * ******************************************************************************* */ @@ -22,6 +22,7 @@ import com.ibm.icu.text.UTF16; import com.ibm.icu.text.UnicodeSetIterator; import com.ibm.icu.text.Transliterator; import com.ibm.icu.text.UnicodeSet; +import java.text.NumberFormat; import java.text.SimpleDateFormat; @@ -921,6 +922,100 @@ public class WriteCharts implements UCD_Types { } + public static void writeAllocation() throws IOException { + Default.setUCD(); + int counter = 0; + UnicodeSet[] values = new UnicodeSet[500]; + String[] names = new String[values.length]; + int[] starts = new int[values.length]; + int[] ends = new int[values.length]; + BufferedReader in = Utility.openUnicodeFile("Blocks", "", true, false); + try { + while (true) { + // 0000..007F; Basic Latin + String line = Utility.readDataLine(in); + if (line == null) break; + if (line.length() == 0) continue; + int pos1 = line.indexOf('.'); + int pos2 = line.indexOf(';', pos1); + int start = Integer.parseInt(line.substring(0, pos1), 16); + int end = Integer.parseInt(line.substring(pos1+2, pos2), 16); + String name = line.substring(pos2+1).trim(); // .replace(' ', '_'); + names[counter] = name; + values[counter] = new UnicodeSet(start, end); + starts[counter] = start; + ends[counter] = end; + //System.out.println(names[counter] + ", " + values[counter]); + ++counter; + } + } finally { + in.close(); + } + + PrintWriter out = Utility.openPrintWriter("Allocation.html", Utility.LATIN1_WINDOWS); + try { + out.println(""); + out.println("Unicode Allocation"); + out.println("

Unicode Allocation

"); + out.println(""); + out.println(""); + + UnicodeSetIterator it = new UnicodeSetIterator(); + + int lastEnd = -1; + for (int i = 0; i < counter; ++i) { + if (starts[i] != lastEnd + 1) { + drawAllocation(out, lastEnd + 1, "reserved", starts[i] - lastEnd + 1, 0); + } + int total = values[i].size(); + int alloc = 0; + it.reset(values[i]); + while (it.nextRange()) { + for (int j = it.codepoint; j <= it.codepointEnd; ++j) { + if (Default.ucd.isAllocated(j)) ++alloc; + } + } + System.out.println(names[i] + "\t" + alloc + "\t" + total); + drawAllocation(out, starts[i], names[i], total, alloc); + lastEnd = ends[i]; + } + out.println("
StartBlock NameSize
"); + out.println("

This chart lists all the Unicode blocks and their starting code points. " + + "The area of each bar is proportional to the total number of code points in each block, " + + "with green for the proportion of assigned code points. " + + "Tooltips on the bars show the total number of code points and the number assigned. " + + "(Remember that assigned code points are not necessarily assigned characters.)" + + "

"); + out.println(""); + } finally { + out.close(); + } + } + + static double longestBar = 1000; + static int longestBlock = 722402; + static NumberFormat nf = NumberFormat.getNumberInstance(Locale.US); + static {nf.setMaximumFractionDigits(0);} + + static void drawAllocation(PrintWriter out, int start, String title, int total, int alloc) { + int unalloc = total - alloc; + + double totalWidth = longestBar*(Math.sqrt(total) / Math.sqrt(longestBlock)); + double allocWidth = alloc * totalWidth / total; + double unallocWidth = totalWidth - allocWidth; + + out.println("" + Utility.hex(start) + + "" + title + + ""); + + if (alloc != 0) out.println(""); + if (unalloc != 0) out.println(""); + out.println("
  
"); + } + }