AuroraMiddleware/scuffed-code
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity
ca34222583
scuffed-code/tools/unicodetools/com/ibm/text/UCA/WriteCharts.java

761 lines
28 KiB
Java
Raw Normal View History

reorg X-SVN-Rev: 5824
2001-09-19 23:33:52 +00:00
/**
*******************************************************************************
* Copyright (C) 1996-2001, International Business Machines Corporation and *
* others. All Rights Reserved. *
*******************************************************************************
*
* $Source: /xsrl/Nsvn/icu/unicodetools/com/ibm/text/UCA/WriteCharts.java,v $
Updated with changes agreed with Ken for d6. Added line for rearrange (missing earlier); this changes conformance! X-SVN-Rev: 9022
2002-07-03 02:15:47 +00:00
* $Date: 2002/07/03 02:15:47 $
* $Revision: 1.11 $
reorg X-SVN-Rev: 5824
2001-09-19 23:33:52 +00:00
*
*******************************************************************************
*/
package com.ibm.text.UCA;
import java.util.*;
import java.io.*;
import com.ibm.text.UCD.*;
import com.ibm.text.utility.*;
ICU-0 updated to reorganized java directories X-SVN-Rev: 8039
2002-03-15 01:57:01 +00:00
import com.ibm.icu.text.UTF16;
chart fixes X-SVN-Rev: 8499
2002-04-23 01:59:16 +00:00
import java.text.SimpleDateFormat;
reorg X-SVN-Rev: 5824
2001-09-19 23:33:52 +00:00
public class WriteCharts implements UCD_Types {
chart fixes X-SVN-Rev: 8499
2002-04-23 01:59:16 +00:00
static boolean HACK_KANA = false;
Fixes for charts X-SVN-Rev: 6470
2001-10-26 23:33:48 +00:00
misc. updates X-SVN-Rev: 8714
2002-05-29 02:01:00 +00:00
static public void special() {
Default.setUCD();
for (int i = 0xE000; i < 0x10000; ++i) {
if (!Default.ucd.isRepresented(i)) continue;
Completed most of the upgrade to UCA 3.1.1; involved double CEs for Han; fixing back-mappings, etc. did a bit of code cleanup too. Remaining to do: backmap from UCA double CEs to original character codes, for constructing Fractional UCA. X-SVN-Rev: 8754
2002-05-31 01:41:04 +00:00
if (!Default.nfkc.isNormalized(i)) continue;
misc. updates X-SVN-Rev: 8714
2002-05-29 02:01:00 +00:00
System.out.println(Default.ucd.getCodeAndName(i));
}
}
fixed up charts X-SVN-Rev: 8501
2002-04-23 22:50:15 +00:00
static public void collationChart(UCA uca) throws IOException {
chart fixes X-SVN-Rev: 8499
2002-04-23 01:59:16 +00:00
Default.setUCD(uca.getUCDVersion());
HACK_KANA = true;
reorg X-SVN-Rev: 5824
2001-09-19 23:33:52 +00:00
uca.setAlternate(UCA.NON_IGNORABLE);
chart fixes X-SVN-Rev: 8499
2002-04-23 01:59:16 +00:00
//Normalizer nfd = new Normalizer(Normalizer.NFD);
//Normalizer nfc = new Normalizer(Normalizer.NFC);
reorg X-SVN-Rev: 5824
2001-09-19 23:33:52 +00:00
UCA.UCAContents cc = uca.getContents(UCA.FIXED_CE, null); // nfd instead of null if skipping decomps
Fixes for charts X-SVN-Rev: 6470
2001-10-26 23:33:48 +00:00
cc.enableSamples();
reorg X-SVN-Rev: 5824
2001-09-19 23:33:52 +00:00
Set set = new TreeSet();
while (true) {
String x = cc.next();
if (x == null) break;
chart fixes X-SVN-Rev: 8499
2002-04-23 01:59:16 +00:00
if (x.equals("\u2F00")) {
System.out.println("debug");
}
reorg X-SVN-Rev: 5824
2001-09-19 23:33:52 +00:00
set.add(new Pair(uca.getSortKey(x), x));
}
PrintWriter output = null;
Iterator it = set.iterator();
Fixes for charts X-SVN-Rev: 6470
2001-10-26 23:33:48 +00:00
byte oldScript = -127;
reorg X-SVN-Rev: 5824
2001-09-19 23:33:52 +00:00
Fixes for charts X-SVN-Rev: 6470
2001-10-26 23:33:48 +00:00
int[] scriptCount = new int[128];
reorg X-SVN-Rev: 5824
2001-09-19 23:33:52 +00:00
int counter = 0;
fixes for collation charts X-SVN-Rev: 6436
2001-10-25 20:35:42 +00:00
String lastSortKey = "\u0000";
reorg X-SVN-Rev: 5824
2001-09-19 23:33:52 +00:00
int high = uca.getSortKey("a").charAt(0);
int variable = UCA.getPrimary(uca.getVariableHigh());
int columnCount = 0;
fixed up charts X-SVN-Rev: 8501
2002-04-23 22:50:15 +00:00
String[] replacement = new String[] {"%%%", "Collation Charts"};
String folder = "charts\\uca\\";
Utility.copyTextFile("index.html", true, folder + "index.html", replacement);
Utility.copyTextFile("charts.css", false, folder + "charts.css");
Utility.copyTextFile("help.html", true, folder + "help.html");
fixes for collation charts X-SVN-Rev: 6436
2001-10-25 20:35:42 +00:00
fixed up charts X-SVN-Rev: 8501
2002-04-23 22:50:15 +00:00
indexFile = Utility.openPrintWriter(folder + "index_list.html", false, false);
Utility.appendFile("index_header.html", true, indexFile, replacement);
Fixes for charts X-SVN-Rev: 6470
2001-10-26 23:33:48 +00:00
/*
reorg X-SVN-Rev: 5824
2001-09-19 23:33:52 +00:00
indexFile.println("<html><head><meta http-equiv='Content-Type' content='text/html; charset=utf-8'>");
indexFile.println("<title>UCA Default Collation Table</title>");
indexFile.println("<base target='main'>");
fixes for collation charts X-SVN-Rev: 6436
2001-10-25 20:35:42 +00:00
indexFile.println("<style><!-- p { font-size: 90% } --></style>");
reorg X-SVN-Rev: 5824
2001-09-19 23:33:52 +00:00
indexFile.println("</head><body><h2 align='center'>UCA Default Collation Table</h2>");
indexFile.println("<p align='center'><a href = 'help.html'>Help</a>");
Fixes for charts X-SVN-Rev: 6470
2001-10-26 23:33:48 +00:00
*/
reorg X-SVN-Rev: 5824
2001-09-19 23:33:52 +00:00
while (it.hasNext()) {
Utility.dot(counter);
Pair p = (Pair) it.next();
String sortKey = (String) p.first;
String s = (String) p.second;
int cp = UTF16.charAt(s,0);
chart fixes X-SVN-Rev: 8499
2002-04-23 01:59:16 +00:00
byte script = Default.ucd.getScript(cp);
reorg X-SVN-Rev: 5824
2001-09-19 23:33:52 +00:00
// get first non-zero primary
chart fixes X-SVN-Rev: 8499
2002-04-23 01:59:16 +00:00
int currentPrimary = getFirstPrimary(sortKey);
int primary = currentPrimary >>> 16;
fixed up charts X-SVN-Rev: 8501
2002-04-23 22:50:15 +00:00
if (sortKey.length() < 4) script = NULL_ORDER;
else if (primary == 0) script = IGNORABLE_ORDER;
else if (primary < variable) script = VARIABLE_ORDER;
reorg X-SVN-Rev: 5824
2001-09-19 23:33:52 +00:00
else if (primary < high) script = COMMON_SCRIPT;
Updated with changes agreed with Ken for d6. Added line for rearrange (missing earlier); this changes conformance! X-SVN-Rev: 9022
2002-07-03 02:15:47 +00:00
else if (UCA.isImplicitLeadPrimary(primary)) {
if (primary < UCA_Types.UNSUPPORTED_CJK_AB_BASE) script = CJK;
else if (primary < UCA_Types.UNSUPPORTED_OTHER_BASE) script = CJK_AB;
else script = UNSUPPORTED;
}
reorg X-SVN-Rev: 5824
2001-09-19 23:33:52 +00:00
Fixes for charts X-SVN-Rev: 6470
2001-10-26 23:33:48 +00:00
if (script == KATAKANA_SCRIPT) script = HIRAGANA_SCRIPT;
else if ((script == INHERITED_SCRIPT || script == COMMON_SCRIPT) && oldScript >= 0) script = oldScript;
reorg X-SVN-Rev: 5824
2001-09-19 23:33:52 +00:00
if (script != oldScript
Fixes for charts X-SVN-Rev: 6470
2001-10-26 23:33:48 +00:00
// && (script != COMMON_SCRIPT && script != INHERITED_SCRIPT)
) {
reorg X-SVN-Rev: 5824
2001-09-19 23:33:52 +00:00
closeFile(output);
output = null;
Fixes for charts X-SVN-Rev: 6470
2001-10-26 23:33:48 +00:00
oldScript = script;
reorg X-SVN-Rev: 5824
2001-09-19 23:33:52 +00:00
}
Fixes for charts X-SVN-Rev: 6470
2001-10-26 23:33:48 +00:00
reorg X-SVN-Rev: 5824
2001-09-19 23:33:52 +00:00
if (output == null) {
++scriptCount[script+3];
if (scriptCount[script+3] > 1) {
System.out.println("\t\tFAIL: " + scriptCount[script+3] + ", " +
fixed up charts X-SVN-Rev: 8501
2002-04-23 22:50:15 +00:00
getChunkName(script, LONG) + ", " + Default.ucd.getCodeAndName(s));
reorg X-SVN-Rev: 5824
2001-09-19 23:33:52 +00:00
}
fixed up charts X-SVN-Rev: 8501
2002-04-23 22:50:15 +00:00
output = openFile(scriptCount[script+3], folder, script);
reorg X-SVN-Rev: 5824
2001-09-19 23:33:52 +00:00
}
chart fixes X-SVN-Rev: 8499
2002-04-23 01:59:16 +00:00
boolean firstPrimaryEquals = currentPrimary == getFirstPrimary(lastSortKey);
fixes for collation charts X-SVN-Rev: 6436
2001-10-25 20:35:42 +00:00
int strength = uca.strengthDifference(sortKey, lastSortKey);
if (strength < 0) strength = -strength;
reorg X-SVN-Rev: 5824
2001-09-19 23:33:52 +00:00
lastSortKey = sortKey;
fixes for collation charts X-SVN-Rev: 6436
2001-10-25 20:35:42 +00:00
// find out if this is an expansion: more than one primary weight
int primaryCount = 0;
for (int i = 0; i < sortKey.length(); ++i) {
char w = sortKey.charAt(i);
if (w == 0) break;
shifted the implicit range. X-SVN-Rev: 8767
2002-06-02 05:07:08 +00:00
if (UCA.isImplicitLeadPrimary(w)) {
chart fixes X-SVN-Rev: 8499
2002-04-23 01:59:16 +00:00
++i; // skip next
}
fixes for collation charts X-SVN-Rev: 6436
2001-10-25 20:35:42 +00:00
++ primaryCount;
}
reorg X-SVN-Rev: 5824
2001-09-19 23:33:52 +00:00
String breaker = "";
fixes for collation charts X-SVN-Rev: 6436
2001-10-25 20:35:42 +00:00
if (columnCount > 10 || !firstPrimaryEquals) {
reorg X-SVN-Rev: 5824
2001-09-19 23:33:52 +00:00
columnCount = 0;
chart fixes X-SVN-Rev: 8499
2002-04-23 01:59:16 +00:00
if (!firstPrimaryEquals || script == UNSUPPORTED) breaker = "</tr><tr>";
else {
breaker = "</tr><tr><td></td>"; // indent 1 cell
++columnCount;
}
reorg X-SVN-Rev: 5824
2001-09-19 23:33:52 +00:00
}
fixes for collation charts X-SVN-Rev: 6436
2001-10-25 20:35:42 +00:00
String classname = primaryCount > 1 ? XCLASSNAME[strength] : CLASSNAME[strength];
chart fixes X-SVN-Rev: 8499
2002-04-23 01:59:16 +00:00
String name = Default.ucd.getName(s);
if (s.equals("\u1eaf")) {
System.out.println("debug");
}
String comp = Default.nfc.normalize(s);
Updated with changes agreed with Ken for d6. Added line for rearrange (missing earlier); this changes conformance! X-SVN-Rev: 9022
2002-07-03 02:15:47 +00:00
String outline = breaker + classname
+ " title='"
+ (script != UNSUPPORTED
? Utility.quoteXML(name, true) + ": "
: "")
+ UCA.toString(sortKey) + "'>"
minor chart fixes X-SVN-Rev: 8504
2002-04-24 02:38:53 +00:00
+ Utility.quoteXML(comp, true)
Fixes for charts X-SVN-Rev: 6470
2001-10-26 23:33:48 +00:00
+ "<br><tt>"
+ Utility.hex(s)
reorg X-SVN-Rev: 5824
2001-09-19 23:33:52 +00:00
//+ "<br>" + script
Updated with changes agreed with Ken for d6. Added line for rearrange (missing earlier); this changes conformance! X-SVN-Rev: 9022
2002-07-03 02:15:47 +00:00
+ "</tt></td>"
+ (script == UNSUPPORTED
? "<td class='name'><tt>" + Utility.quoteXML(name, true) + "</td>"
: "")
;
chart fixes X-SVN-Rev: 8499
2002-04-23 01:59:16 +00:00
output.println(outline);
reorg X-SVN-Rev: 5824
2001-09-19 23:33:52 +00:00
++columnCount;
}
closeFile(output);
fixed up charts X-SVN-Rev: 8501
2002-04-23 22:50:15 +00:00
closeIndexFile(indexFile, "<br>UCA: " + uca.getDataVersion(), COLLATION);
reorg X-SVN-Rev: 5824
2001-09-19 23:33:52 +00:00
}
chart fixes X-SVN-Rev: 8499
2002-04-23 01:59:16 +00:00
static public void normalizationChart() throws IOException {
Default.setUCD();
HACK_KANA = false;
Set set = new TreeSet();
for (int i = 0; i <= 0x10FFFF; ++i) {
minor chart fixes X-SVN-Rev: 8504
2002-04-24 02:38:53 +00:00
if (!Default.ucd.isRepresented(i)) {
if (i < 0xAC00) continue;
if (i > 0xD7A3) continue;
if (i > 0xACFF && i < 0xD700) continue;
}
chart fixes X-SVN-Rev: 8499
2002-04-23 01:59:16 +00:00
byte cat = Default.ucd.getCategory(i);
if (cat == Cs || cat == Co) continue;
Completed most of the upgrade to UCA 3.1.1; involved double CEs for Han; fixing back-mappings, etc. did a bit of code cleanup too. Remaining to do: backmap from UCA double CEs to original character codes, for constructing Fractional UCA. X-SVN-Rev: 8754
2002-05-31 01:41:04 +00:00
if (Default.nfkd.isNormalized(i)) continue;
chart fixes X-SVN-Rev: 8499
2002-04-23 01:59:16 +00:00
String decomp = Default.nfkd.normalize(i);
byte script = getBestScript(decomp);
fixed up charts X-SVN-Rev: 8501
2002-04-23 22:50:15 +00:00
set.add(new Pair(new Integer(script == COMMON_SCRIPT ? cat + CAT_OFFSET : script),
minor chart fixes X-SVN-Rev: 8504
2002-04-24 02:38:53 +00:00
new Pair(Default.ucd.getCase(decomp, FULL, FOLD),
chart fixes X-SVN-Rev: 8499
2002-04-23 01:59:16 +00:00
new Integer(i))));
}
PrintWriter output = null;
Iterator it = set.iterator();
int oldScript = -127;
int counter = 0;
fixed up charts X-SVN-Rev: 8501
2002-04-23 22:50:15 +00:00
String[] replacement = new String[] {"%%%", "Normalization Charts"};
String folder = "charts\\normalization\\";
Utility.copyTextFile("index.html", true, folder + "index.html", replacement);
Utility.copyTextFile("charts.css", false, folder + "charts.css");
Utility.copyTextFile("norm_help.html", true, folder + "help.html");
chart fixes X-SVN-Rev: 8499
2002-04-23 01:59:16 +00:00
fixed up charts X-SVN-Rev: 8501
2002-04-23 22:50:15 +00:00
indexFile = Utility.openPrintWriter(folder + "index_list.html", false, false);
Utility.appendFile("index_header.html", true, indexFile, replacement);
chart fixes X-SVN-Rev: 8499
2002-04-23 01:59:16 +00:00
/*
indexFile.println("<html><head><meta http-equiv='Content-Type' content='text/html; charset=utf-8'>");
indexFile.println("<title>UCA Default Collation Table</title>");
indexFile.println("<base target='main'>");
indexFile.println("<style><!-- p { font-size: 90% } --></style>");
indexFile.println("</head><body><h2 align='center'>UCA Default Collation Table</h2>");
indexFile.println("<p align='center'><a href = 'help.html'>Help</a>");
*/
while (it.hasNext()) {
Utility.dot(counter);
Pair p = (Pair) it.next();
int script = ((Integer) p.first).intValue();
int cp = ((Integer)((Pair) p.second).second).intValue();
if (script != oldScript
// && (script != COMMON_SCRIPT && script != INHERITED_SCRIPT)
) {
closeFile(output);
output = null;
oldScript = script;
}
if (output == null) {
fixed up charts X-SVN-Rev: 8501
2002-04-23 22:50:15 +00:00
output = openFile(0, folder, script);
chart fixes X-SVN-Rev: 8499
2002-04-23 01:59:16 +00:00
output.println("<tr><td class='z'>Code</td><td class='z'>C</td><td class='z'>D</td><td class='z'>KC</td><td class='z'>KD</td></tr>");
}
output.println("<tr>");
String prefix;
String code = UTF16.valueOf(cp);
String c = Default.nfc.normalize(cp);
String d = Default.nfd.normalize(cp);
String kc = Default.nfkc.normalize(cp);
String kd = Default.nfkd.normalize(cp);
fixed up charts X-SVN-Rev: 8501
2002-04-23 22:50:15 +00:00
showCell(output, code, "<td class='z' ", "", false);
chart fixes X-SVN-Rev: 8499
2002-04-23 01:59:16 +00:00
prefix = c.equals(code) ? "<td class='g' " : "<td class='n' ";
fixed up charts X-SVN-Rev: 8501
2002-04-23 22:50:15 +00:00
showCell(output, c, prefix, "", c.equals(code));
chart fixes X-SVN-Rev: 8499
2002-04-23 01:59:16 +00:00
prefix = d.equals(c) ? "<td class='g' " : "<td class='n' ";
fixed up charts X-SVN-Rev: 8501
2002-04-23 22:50:15 +00:00
showCell(output, d, prefix, "", d.equals(c));
chart fixes X-SVN-Rev: 8499
2002-04-23 01:59:16 +00:00
prefix = kc.equals(c) ? "<td class='g' " : "<td class='n' ";
fixed up charts X-SVN-Rev: 8501
2002-04-23 22:50:15 +00:00
showCell(output, kc, prefix, "", kc.equals(c));
chart fixes X-SVN-Rev: 8499
2002-04-23 01:59:16 +00:00
prefix = (kd.equals(d) || kd.equals(kc)) ? "<td class='g' " : "<td class='n' ";
fixed up charts X-SVN-Rev: 8501
2002-04-23 22:50:15 +00:00
showCell(output, kd, prefix, "", (kd.equals(d) || kd.equals(kc)));
chart fixes X-SVN-Rev: 8499
2002-04-23 01:59:16 +00:00
output.println("</tr>");
}
closeFile(output);
fixed up charts X-SVN-Rev: 8501
2002-04-23 22:50:15 +00:00
closeIndexFile(indexFile, "", NORMALIZATION);
chart fixes X-SVN-Rev: 8499
2002-04-23 01:59:16 +00:00
}
static public void caseChart() throws IOException {
Default.setUCD();
HACK_KANA = false;
Set set = new TreeSet();
for (int i = 0; i <= 0x10FFFF; ++i) {
if (!Default.ucd.isRepresented(i)) continue;
byte cat = Default.ucd.getCategory(i);
if (cat == Cs || cat == Co) continue;
String code = UTF16.valueOf(i);
String lower = Default.ucd.getCase(i, FULL, LOWER);
String title = Default.ucd.getCase(i, FULL, TITLE);
String upper = Default.ucd.getCase(i, FULL, UPPER);
String fold = Default.ucd.getCase(i, FULL, FOLD);
String decomp = Default.nfkd.normalize(i);
fixed up charts X-SVN-Rev: 8501
2002-04-23 22:50:15 +00:00
int script = 0;
if (lower.equals(code) && upper.equals(code) && fold.equals(code) && title.equals(code)) {
if (!containsCase(decomp)) continue;
script = NO_CASE_MAPPING;
chart fixes X-SVN-Rev: 8499
2002-04-23 01:59:16 +00:00
}
if (script == 0) script = getBestScript(decomp);
fixed up charts X-SVN-Rev: 8501
2002-04-23 22:50:15 +00:00
set.add(new Pair(new Integer(script == COMMON_SCRIPT ? cat + CAT_OFFSET : script),
minor chart fixes X-SVN-Rev: 8504
2002-04-24 02:38:53 +00:00
new Pair(Default.ucd.getCase(decomp, FULL, FOLD),
chart fixes X-SVN-Rev: 8499
2002-04-23 01:59:16 +00:00
new Integer(i))));
}
PrintWriter output = null;
Iterator it = set.iterator();
int oldScript = -127;
int counter = 0;
fixed up charts X-SVN-Rev: 8501
2002-04-23 22:50:15 +00:00
String[] replacement = new String[] {"%%%", "Case Charts"};
String folder = "charts\\case\\";
chart fixes X-SVN-Rev: 8499
2002-04-23 01:59:16 +00:00
fixed up charts X-SVN-Rev: 8501
2002-04-23 22:50:15 +00:00
Utility.copyTextFile("index.html", true, folder + "index.html", replacement);
Utility.copyTextFile("charts.css", false, folder + "charts.css");
Utility.copyTextFile("case_help.html", true, folder + "help.html");
chart fixes X-SVN-Rev: 8499
2002-04-23 01:59:16 +00:00
fixed up charts X-SVN-Rev: 8501
2002-04-23 22:50:15 +00:00
indexFile = Utility.openPrintWriter(folder + "index_list.html", false, false);
Utility.appendFile("index_header.html", true, indexFile, replacement);
chart fixes X-SVN-Rev: 8499
2002-04-23 01:59:16 +00:00
/*
indexFile.println("<html><head><meta http-equiv='Content-Type' content='text/html; charset=utf-8'>");
indexFile.println("<title>UCA Default Collation Table</title>");
indexFile.println("<base target='main'>");
indexFile.println("<style><!-- p { font-size: 90% } --></style>");
indexFile.println("</head><body><h2 align='center'>UCA Default Collation Table</h2>");
indexFile.println("<p align='center'><a href = 'help.html'>Help</a>");
*/
fixed up charts X-SVN-Rev: 8501
2002-04-23 22:50:15 +00:00
int columnCount = 0;
chart fixes X-SVN-Rev: 8499
2002-04-23 01:59:16 +00:00
while (it.hasNext()) {
Utility.dot(counter);
Pair p = (Pair) it.next();
int script = ((Integer) p.first).intValue();
int cp = ((Integer)((Pair) p.second).second).intValue();
if (script != oldScript
// && (script != COMMON_SCRIPT && script != INHERITED_SCRIPT)
) {
closeFile(output);
output = null;
oldScript = script;
}
if (output == null) {
fixed up charts X-SVN-Rev: 8501
2002-04-23 22:50:15 +00:00
output = openFile(0, folder, script);
if (script == NO_CASE_MAPPING) output.println("<tr>");
else output.println("<tr><td class='z'>Code</td><td class='z'>Lower</td><td class='z'>Title</td>"
+"<td class='z'>Upper</td><td class='z'>Fold</td></tr>");
chart fixes X-SVN-Rev: 8499
2002-04-23 01:59:16 +00:00
}
fixed up charts X-SVN-Rev: 8501
2002-04-23 22:50:15 +00:00
if (script == NO_CASE_MAPPING) {
if (columnCount > 10) {
output.println("</tr><tr>");
columnCount = 0;
}
showCell(output, UTF16.valueOf(cp), "<td ", "", false);
++columnCount;
continue;
}
chart fixes X-SVN-Rev: 8499
2002-04-23 01:59:16 +00:00
output.println("<tr>");
String prefix;
String code = UTF16.valueOf(cp);
String lower = Default.ucd.getCase(cp, FULL, LOWER);
String title = Default.ucd.getCase(cp, FULL, TITLE);
String upper = Default.ucd.getCase(cp, FULL, UPPER);
String fold = Default.ucd.getCase(cp, FULL, FOLD);
fixed up charts X-SVN-Rev: 8501
2002-04-23 22:50:15 +00:00
showCell(output, code, "<td class='z' ", "", false);
chart fixes X-SVN-Rev: 8499
2002-04-23 01:59:16 +00:00
prefix = lower.equals(code) ? "<td class='g' " : "<td class='n' ";
fixed up charts X-SVN-Rev: 8501
2002-04-23 22:50:15 +00:00
showCell(output, lower, prefix, "", lower.equals(code));
chart fixes X-SVN-Rev: 8499
2002-04-23 01:59:16 +00:00
prefix = title.equals(upper) ? "<td class='g' " : "<td class='n' ";
fixed up charts X-SVN-Rev: 8501
2002-04-23 22:50:15 +00:00
showCell(output, title, prefix, "", title.equals(upper));
chart fixes X-SVN-Rev: 8499
2002-04-23 01:59:16 +00:00
prefix = upper.equals(code) ? "<td class='g' " : "<td class='n' ";
fixed up charts X-SVN-Rev: 8501
2002-04-23 22:50:15 +00:00
showCell(output, upper, prefix, "", upper.equals(code));
chart fixes X-SVN-Rev: 8499
2002-04-23 01:59:16 +00:00
fixed up charts X-SVN-Rev: 8501
2002-04-23 22:50:15 +00:00
prefix = fold.equals(lower) ? "<td class='g' " : "<td class='n' ";
showCell(output, fold, prefix, "", fold.equals(lower));
chart fixes X-SVN-Rev: 8499
2002-04-23 01:59:16 +00:00
output.println("</tr>");
}
fixed up charts X-SVN-Rev: 8501
2002-04-23 22:50:15 +00:00
closeFile(output);
closeIndexFile(indexFile, "", CASE);
}
misc. updates X-SVN-Rev: 8714
2002-05-29 02:01:00 +00:00
static public void addMapChar(Map m, Set stoplist, String key, String ch) {
if (stoplist.contains(key)) return;
for (int i = 0; i < key.length(); ++i) {
char c = key.charAt(i);
if ('0' <= c && c <= '9') return;
}
Set result = (Set)m.get(key);
if (result == null) {
result = new TreeSet();
m.put(key, result);
}
result.add(ch);
}
static public void indexChart() throws IOException {
Default.setUCD();
HACK_KANA = false;
Map map = new TreeMap();
Set stoplist = new TreeSet();
String[] stops = {"LETTER", "CHARACTER", "AND", "CAPITAL", "SMALL", "COMPATIBILITY", "WITH"};
stoplist.addAll(Arrays.asList(stops));
System.out.println("Stop-list: " + stoplist);
for (int i = 0; i < LIMIT_SCRIPT; ++i) {
stoplist.add(Default.ucd.getScriptID_fromIndex((byte)i));
}
System.out.println("Stop-list: " + stoplist);
for (int i = 0; i <= 0x10FFFF; ++i) {
if (!Default.ucd.isRepresented(i)) continue;
if (0xAC00 <= i && i <= 0xD7A3) continue;
if (Default.ucd.hasComputableName(i)) continue;
String s = Default.ucd.getName(i);
if (s == null) continue;
if (s.startsWith("<")) {
System.out.println("Wierd character at " + Default.ucd.getCodeAndName(i));
}
String ch = UTF16.valueOf(i);
int last = -1;
int j;
for (j = 0; j < s.length(); ++j) {
char c = s.charAt(j);
if ('A' <= c && c <= 'Z' || '0' <= c && c <= '9') {
if (last == -1) last = j;
} else {
if (last != -1) {
String word = s.substring(last, j);
addMapChar(map, stoplist, word, ch);
last = -1;
}
}
}
if (last != -1) {
String word = s.substring(last, j);
addMapChar(map, stoplist, word, ch);
}
}
PrintWriter output = null;
Iterator it = map.keySet().iterator();
int oldScript = -127;
int counter = 0;
String[] replacement = new String[] {"%%%", "Name Charts"};
String folder = "charts\\name\\";
Utility.copyTextFile("index.html", true, folder + "index.html", replacement);
Utility.copyTextFile("charts.css", false, folder + "charts.css");
Utility.copyTextFile("name_help.html", true, folder + "help.html");
indexFile = Utility.openPrintWriter(folder + "index_list.html", false, false);
Utility.appendFile("index_header.html", true, indexFile, replacement);
int columnCount = 0;
char lastInitial = 0;
while (it.hasNext()) {
Utility.dot(counter);
String key = (String) it.next();
Set chars = (Set) map.get(key);
char initial = key.charAt(0);
if (initial != lastInitial) {
closeFile(output);
output = null;
lastInitial = initial;
}
if (output == null) {
output = openFile2(0, folder, String.valueOf(initial));
}
output.println("<tr><td class='h'>" + key + "</td>");
columnCount = 1;
Iterator sublist = chars.iterator();
while (sublist.hasNext()) {
String ch = (String) sublist.next();
if (columnCount > 10) {
output.println("</tr><tr><td></td>");
columnCount = 1;
}
showCell(output, ch, "<td ", "", true);
++columnCount;
continue;
}
output.println("</tr>");
}
closeFile(output);
closeIndexFile(indexFile, "", CASE);
}
fixed up charts X-SVN-Rev: 8501
2002-04-23 22:50:15 +00:00
static void showCell(PrintWriter output, String s, String prefix, String extra, boolean skipName) {
String name = Default.ucd.getName(s);
String comp = Default.nfc.normalize(s);
String outline = prefix
minor chart fixes X-SVN-Rev: 8504
2002-04-24 02:38:53 +00:00
+ (skipName ? "" : " title='" + Utility.quoteXML(name, true) + "'")
fixed up charts X-SVN-Rev: 8501
2002-04-23 22:50:15 +00:00
+ extra + ">"
minor chart fixes X-SVN-Rev: 8504
2002-04-24 02:38:53 +00:00
+ Utility.quoteXML(comp, true)
fixed up charts X-SVN-Rev: 8501
2002-04-23 22:50:15 +00:00
+ "<br><tt>"
+ Utility.hex(s)
//+ "<br>" + script
+ "</tt></td>";
output.println(outline);
}
static byte getBestScript(String s) {
int cp;
byte result = COMMON_SCRIPT;
for (int i = 0; i < s.length(); i += UTF16.getCharCount(cp)) {
cp = UTF16.charAt(s, i);
result = Default.ucd.getScript(cp);
if (result != COMMON_SCRIPT && result != INHERITED_SCRIPT) return result;
}
return COMMON_SCRIPT;
}
static int getFirstPrimary(String sortKey) {
int result = sortKey.charAt(0);
shifted the implicit range. X-SVN-Rev: 8767
2002-06-02 05:07:08 +00:00
if (UCA.isImplicitLeadPrimary(result)) {
fixed up charts X-SVN-Rev: 8501
2002-04-23 22:50:15 +00:00
return (result << 16) | sortKey.charAt(1);
}
return (result << 16);
}
static final String[] CLASSNAME = {
"<td class='q'",
"<td class='q'",
"<td class='q'",
"<td class='t'",
"<td class='s'",
"<td class='p'"};
static final String[] XCLASSNAME = {
"<td class='eq'",
"<td class='eq'",
"<td class='eq'",
"<td class='et'",
"<td class='es'",
"<td class='ep'"};
static PrintWriter indexFile;
static PrintWriter openFile(int count, String directory, int script) throws IOException {
String scriptName = getChunkName(script, LONG);
String shortScriptName = getChunkName(script, SHORT);
String hover = scriptName.equals(shortScriptName) ? "" : "' title='" + shortScriptName;
String fileName = "chart_" + scriptName + (count > 1 ? count + "" : "") + ".html";
PrintWriter output = Utility.openPrintWriter(directory + fileName, false, false);
Utility.fixDot();
System.out.println("Writing: " + scriptName);
indexFile.println(" <a href = '" + fileName + hover + "'>" + scriptName + "</a>");
String title = "UCA: " + scriptName;
output.println("<html><head><meta http-equiv='Content-Type' content='text/html; charset=utf-8'>");
output.println("<title>" + title + "</title>");
output.println("<link rel='stylesheet' href='charts.css' type='text/css'>");
output.println("</head><body><h2>" + scriptName + "</h2>");
output.println("<table>");
return output;
}
misc. updates X-SVN-Rev: 8714
2002-05-29 02:01:00 +00:00
static PrintWriter openFile2(int count, String directory, String name) throws IOException {
String fileName = "chart_" + name + (count > 1 ? count + "" : "") + ".html";
PrintWriter output = Utility.openPrintWriter(directory + fileName, false, false);
Utility.fixDot();
System.out.println("Writing: " + name);
indexFile.println(" <a href = '" + fileName + "'>" + name + "</a>");
String title = name;
output.println("<html><head><meta http-equiv='Content-Type' content='text/html; charset=utf-8'>");
output.println("<title>" + title + "</title>");
output.println("<link rel='stylesheet' href='charts.css' type='text/css'>");
output.println("</head><body>");
output.println("<table>");
return output;
}
fixed up charts X-SVN-Rev: 8501
2002-04-23 22:50:15 +00:00
static final int
NULL_ORDER = -3,
IGNORABLE_ORDER = -2,
VARIABLE_ORDER = -1,
// scripts in here
Updated with changes agreed with Ken for d6. Added line for rearrange (missing earlier); this changes conformance! X-SVN-Rev: 9022
2002-07-03 02:15:47 +00:00
CJK = 120,
CJK_AB = 121,
UNSUPPORTED = 122,
fixed up charts X-SVN-Rev: 8501
2002-04-23 22:50:15 +00:00
CAT_OFFSET = 128,
// categories in here
NO_CASE_MAPPING = 200;
static String getChunkName(int script, byte length) {
switch(script) {
case NO_CASE_MAPPING: return "NoCaseMapping";
case NULL_ORDER: return "Null";
case IGNORABLE_ORDER: return "Ignorable";
case VARIABLE_ORDER: return "Variable";
Updated with changes agreed with Ken for d6. Added line for rearrange (missing earlier); this changes conformance! X-SVN-Rev: 9022
2002-07-03 02:15:47 +00:00
case CJK: return "CJK";
case CJK_AB: return "CJK-Extensions";
fixed up charts X-SVN-Rev: 8501
2002-04-23 22:50:15 +00:00
case UNSUPPORTED: return "Unsupported";
default:
if (script >= CAT_OFFSET) return Default.ucd.getCategoryID_fromIndex((byte)(script - CAT_OFFSET), length);
else if (script == HIRAGANA_SCRIPT && HACK_KANA) return length == SHORT ? "Kata-Hira" : "Katakana-Hiragana";
else return Default.ucd.getCase(Default.ucd.getScriptID_fromIndex((byte)script, length), FULL, TITLE);
}
}
static void closeFile(PrintWriter output) {
if (output == null) return;
output.println("</table></body></html>");
output.close();
}
static final byte COLLATION = 0, NORMALIZATION = 1, CASE = 2;
static void closeIndexFile(PrintWriter indexFile, String extra, byte choice) {
chart fixes X-SVN-Rev: 8499
2002-04-23 01:59:16 +00:00
SimpleDateFormat df = new SimpleDateFormat("yyyy.MM.dd HH:mm:ss");
fixed up charts X-SVN-Rev: 8501
2002-04-23 22:50:15 +00:00
df.setTimeZone(TimeZone.getTimeZone("GMT"));
chart fixes X-SVN-Rev: 8499
2002-04-23 01:59:16 +00:00
fixed up charts X-SVN-Rev: 8501
2002-04-23 22:50:15 +00:00
indexFile.println("</p><hr width='50%'><p>");
boolean gotOne = false;
if (choice != COLLATION) {
indexFile.println("<a href='..\\uca\\index.html' target='_top'>Collation&nbsp;Charts</a>");
gotOne = true;
}
if (choice != NORMALIZATION) {
if (gotOne) indexFile.println("<br>");
indexFile.println("<a href='..\\normalization\\index.html' target='_top'>Normalization&nbsp;Charts</a>");
gotOne = true;
}
if (choice != CASE) {
if (gotOne) indexFile.println("<br>");
indexFile.println("<a href='..\\case\\index.html' target='_top'>Case&nbsp;Charts</a>");
gotOne = true;
}
indexFile.println("</p><hr width='50%'><p style='font-size: 70%'>");
indexFile.println("UCD: " + Default.ucd.getVersion() + extra);
indexFile.println("<br>" + df.format(new Date()) + " <a href='http://www.macchiato.com/' target='_top'>MED</a>");
chart fixes X-SVN-Rev: 8499
2002-04-23 01:59:16 +00:00
indexFile.println("</p></body></html>");
indexFile.close();
}
fixed up charts X-SVN-Rev: 8501
2002-04-23 22:50:15 +00:00
static boolean containsCase(String s) {
int cp;
for (int i = 0; i < s.length(); i += UTF16.getCharCount(cp)) {
cp = UTF16.charAt(s, i);
// contains Lu, Lo, Lt, or Lowercase or Uppercase
byte cat = Default.ucd.getCategory(cp);
if (cat == Lu || cat == Ll || cat == Lt) return true;
if (Default.ucd.getBinaryProperty(cp, Other_Lowercase)) return true;
if (Default.ucd.getBinaryProperty(cp, Other_Uppercase)) return true;
}
return false;
}
reorg X-SVN-Rev: 5824
2001-09-19 23:33:52 +00:00
}
/*
static final IntStack p1 = new IntStack(30);
static final IntStack s1 = new IntStack(30);
static final IntStack t1 = new IntStack(30);
static final IntStack p2 = new IntStack(30);
static final IntStack s2 = new IntStack(30);
static final IntStack t2 = new IntStack(30);
static int getStrengthDifference(CEList ceList, CEList lastCEList) {
extractNonzeros(ceList, p1, s1, t1);
extractNonzeros(lastCEList, p2, s2, t2);
int temp = p1.compareTo(p2);
if (temp != 0) return 3;
temp = s1.compareTo(s2);
if (temp != 0) return 2;
temp = t1.compareTo(t2);
if (temp != 0) return 1;
return 0;
}
static void extractNonzeros(CEList ceList, IntStack primaries, IntStack secondaries, IntStack tertiaries) {
primaries.clear();
secondaries.clear();
tertiaries.clear();
for (int i = 0; i < ceList.length(); ++i) {
int ce = ceList.at(i);
int temp = UCA.getPrimary(ce);
if (temp != 0) primaries.push(temp);
temp = UCA.getSecondary(ce);
if (temp != 0) secondaries.push(temp);
temp = UCA.getTertiary(ce);
if (temp != 0) tertiaries.push(temp);
}
}
*/
Reference in New Issue Copy Permalink