thdltools
/
Jskad
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

Updated installers to include unicode to wylie and acip to wylie.

This commit is contained in:
amontano 2007-05-01 02:38:45 +00:00
parent c5c448f0c6
commit d6d2eeb004
4 changed files with 103 additions and 72 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

17
source/org/thdl/tib/input/ConvertDialog.java
View File

@ -238,12 +238,15 @@ class ConvertDialog extends JDialog
|| cmd.equals(BROWSENEW))
{
JButton src = (JButton)ae.getSource();
String choice = (String)choices.getSelectedItem();
if (src == browseOld) {
jfc.setFileFilter((ACIP_TO_UNI_TEXT.equals((String)choices.getSelectedItem())
|| WYLIE_TO_UNI_TEXT.equals((String)choices.getSelectedItem())
|| UNI_TO_WYLIE_TEXT.equals((String)choices.getSelectedItem())
|| ACIP_TO_TMW.equals((String)choices.getSelectedItem())
|| WYLIE_TO_TMW.equals((String)choices.getSelectedItem()))
jfc.setFileFilter((ACIP_TO_UNI_TEXT.equals(choice)
|| WYLIE_TO_UNI_TEXT.equals(choice)
|| UNI_TO_WYLIE_TEXT.equals(choice)
|| ACIP_TO_TMW.equals(choice)
|| WYLIE_TO_TMW.equals(choice)
|| ACIP_TO_WYLIE_TEXT.equals(choice)
|| WYLIE_TO_ACIP_TEXT.equals(choice))
? textFileFilter : rtfFileFilter);
} else {
jfc.setFileFilter((ACIP_TO_UNI_TEXT.equals((String)choices.getSelectedItem())
@ -500,12 +503,12 @@ class ConvertDialog extends JDialog
newFileNamePrefix = suggested_WYLIE_prefix;
if (UNI_TO_WYLIE_TEXT == ct)
newFileNameExtension = ".TXT";
} else if (TMW_TO_WYLIE_TEXT == ct) {
} else if (TMW_TO_WYLIE_TEXT == ct || ACIP_TO_WYLIE_TEXT == ct) {
newFileNamePrefix = suggested_WYLIE_prefix;
newFileNameExtension = ".TXT";
} else if (TMW_TO_ACIP == ct) {
newFileNamePrefix = suggested_ACIP_prefix;
} else if (TMW_TO_ACIP_TEXT == ct) {
} else if (TMW_TO_ACIP_TEXT == ct || WYLIE_TO_ACIP_TEXT == ct) {
newFileNamePrefix = suggested_ACIP_prefix;
newFileNameExtension = ".TXT";
} else if (TMW_TO_UNI == ct || ACIP_TO_UNI_TEXT == ct

15
source/org/thdl/tib/input/FontConverterConstants.java
View File

@ -22,13 +22,15 @@ package org.thdl.tib.input;
/** Constants used by ConvertDialog.
@author Nathaniel Garson, Tibetan and Himalayan Digital Library */
interface FontConverterConstants
public interface FontConverterConstants
{
final String UNI_TO_WYLIE_TEXT = "Unicode to Wylie (UTF-8 Text->Text)";
final String UNI_TO_WYLIE_TEXT = "Unicode to Wylie (UTF-16 Text->Text)";
final String WYLIE_TO_UNI_TEXT = "Wylie to Unicode (Text->Text)";
final String WYLIE_TO_TMW = "Wylie to TMW (Text->RTF)";
final String WYLIE_TO_ACIP_TEXT = "Wylie to ACIP (Text->Text)";
final String TMW_TO_SAME_TMW = "TMW to the same TMW (for testing only) (RTF->RTF)";
final String ACIP_TO_UNI_TEXT = "ACIP to Unicode (Text->Text)";
final String ACIP_TO_WYLIE_TEXT = "ACIP to Wylie (Text->Text)";
final String ACIP_TO_TMW = "ACIP to TMW (Text->RTF)";
final String TMW_TO_ACIP = "TMW to ACIP (RTF->RTF)";
final String TMW_TO_ACIP_TEXT = "TMW to ACIP (RTF->Text)";
@ -45,7 +47,9 @@ interface FontConverterConstants
final String[] CHOICES = new String[] {
WYLIE_TO_UNI_TEXT,
WYLIE_TO_TMW,
//WYLIE_TO_ACIP_TEXT,
ACIP_TO_UNI_TEXT,
ACIP_TO_WYLIE_TEXT,
ACIP_TO_TMW,
TMW_TO_ACIP,
TMW_TO_ACIP_TEXT,
@ -54,6 +58,7 @@ interface FontConverterConstants
TMW_TO_UNI,
TMW_TO_WYLIE,
TMW_TO_WYLIE_TEXT,
UNI_TO_WYLIE_TEXT,
FIND_SOME_NON_TMW, // TODO(dchandler): should this be in DEBUG_CHOICES only?
FIND_SOME_NON_TM, // TODO(dchandler): should this be in DEBUG_CHOICES only?
FIND_ALL_NON_TMW, // TODO(dchandler): should this be in DEBUG_CHOICES only?
@ -61,11 +66,11 @@ interface FontConverterConstants
};
final String[] DEBUG_CHOICES = new String[] {
UNI_TO_WYLIE_TEXT,
TMW_TO_SAME_TMW,
WYLIE_TO_UNI_TEXT,
WYLIE_TO_TMW,
ACIP_TO_UNI_TEXT,
ACIP_TO_WYLIE_TEXT,
ACIP_TO_TMW,
TMW_TO_ACIP,
TMW_TO_ACIP_TEXT,
@ -74,6 +79,7 @@ interface FontConverterConstants
TMW_TO_UNI,
TMW_TO_WYLIE,
TMW_TO_WYLIE_TEXT,
UNI_TO_WYLIE_TEXT,
FIND_SOME_NON_TMW,
FIND_SOME_NON_TM,
FIND_ALL_NON_TMW,
@ -88,5 +94,4 @@ interface FontConverterConstants
// String Constants
public final String PROGRAM_TITLE = "THDL Tibetan Converters -- featuring Jskad Technology";
}
}

13
source/org/thdl/tib/input/TibetanConverter.java
View File

@ -26,6 +26,7 @@ import java.io.InputStream;
import java.io.InputStreamReader;
import java.io.OutputStreamWriter;
import java.io.PrintStream;
import java.io.PrintWriter;
import java.util.ArrayList;
import javax.swing.text.SimpleAttributeSet;
@ -43,6 +44,7 @@ import org.thdl.util.ThdlDebug;
import org.thdl.util.ThdlLazyException;
import org.thdl.util.ThdlOptions;
import org.thdl.util.ThdlVersion;
import org.thdl.tib.scanner.*;
/** TibetanConverter is a command-line utility for converting to and
* from Tibetan Machine Web (TMW). It converts TMW to Wylie, ACIP,
@ -329,9 +331,9 @@ public class TibetanConverter implements FontConverterConstants {
static int reallyConvert(InputStream in, PrintStream out, String ct,
String warningLevel, boolean shortMessages,
boolean colors) {
if (UNI_TO_WYLIE_TEXT == ct) {
if (UNI_TO_WYLIE_TEXT == ct || WYLIE_TO_ACIP_TEXT == ct || ACIP_TO_WYLIE_TEXT == ct) {
try {
String uniText;
/*String uniText;
{
// TODO(dchandler): use, here and elsewhere in the
// codebase,
@ -355,7 +357,12 @@ public class TibetanConverter implements FontConverterConstants {
String ewtsText = Converter.convertToEwtsForComputers(uniText,
errors);
// TODO(dchandler): is 51 the right choice?
return (errors.length() > 0) ? 51 : 0;
return (errors.length() > 0) ? 51 : 0;*/
BasicTibetanTranscriptionConverter bc = null;
if (UNI_TO_WYLIE_TEXT == ct) bc = new BasicTibetanTranscriptionConverter(new BufferedReader(new InputStreamReader(in, "UTF16")), new PrintWriter(out));
else bc = new BasicTibetanTranscriptionConverter(new BufferedReader(new InputStreamReader(in)), new PrintWriter(out));
bc.run(ct);
return 0;
} catch (IOException e) {
// TODO(dchandler): print it? where to?
return 48;

130
source/org/thdl/tib/scanner/BasicTibetanTranscriptionConverter.java
View File

@ -23,6 +23,7 @@ import org.thdl.tib.text.TibTextUtils;
import org.thdl.tib.text.TibetanDocument;
import org.thdl.tib.text.reverter.Converter;
import org.thdl.tib.text.ttt.EwtsToUnicodeForXslt;
import org.thdl.tib.input.*;
import org.thdl.util.*;
import java.net.*;
import java.io.*;
@ -35,12 +36,12 @@ import java.io.*;
* @author Andres Montano
*
*/
public class BasicTibetanTranscriptionConverter
public class BasicTibetanTranscriptionConverter implements FontConverterConstants
{
private static BufferedReader in;
private static PrintWriter out;
private static int conversionType=0;
//private static int conversionType=0;
private static final int ACIP_TO_WYLIE=1;
private static final int WYLIE_TO_ACIP=2;
private static final int UNICODE_TO_WYLIE=3;
@ -81,39 +82,39 @@ public class BasicTibetanTranscriptionConverter
aa -> a, a'a -> A, ai->i, aee ->ai, au->u, aoo->au, ae->e,
ao->o, ee->ai, oo->au, 'I->-I I->-i, a'i->I, a'u->U, a'e->E, a'o->O,
a'i->I, a'u->U, a'e->E, a'o->O, ,->/, # -> @##, * -> @#, \ -> ?, ` -> !,
/-/ -> (-), ga-y -> g.y, g-y -> g.y, na-y -> n+y
/-/ -> (-), ga-y -> g.y, g-y -> g.y, na-y -> n+y */
nuevaLinea = replace(nuevaLinea, "ts", "tq");
nuevaLinea = replace(nuevaLinea, "tz", "ts");
nuevaLinea = replace(nuevaLinea, "tq", "tsh");
nuevaLinea = replace(nuevaLinea, "v", "w");
nuevaLinea = replace(nuevaLinea, "TH", "Th");
nuevaLinea = replace(nuevaLinea, "kSH", "k+Sh");
nuevaLinea = replace(nuevaLinea, "kaSH", "k+Sh");
nuevaLinea = replace(nuevaLinea, "SH", "Sh");
nuevaLinea = replace(nuevaLinea, ":", "H");
nuevaLinea = replace(nuevaLinea, "NH", "NaH");
nuevaLinea = replace(nuevaLinea, "dh", "d+h");
nuevaLinea = replace(nuevaLinea, "gh", "g+h");
nuevaLinea = replace(nuevaLinea, "bh", "b+h");
nuevaLinea = replace(nuevaLinea, "dzh", "dz+h");
nuevaLinea = replace(nuevaLinea, "aa", "a");
nuevaLinea = replace(nuevaLinea, "ai", "i");
nuevaLinea = replace(nuevaLinea, "aee", "ai");
nuevaLinea = replace(nuevaLinea, "au", "u");
nuevaLinea = replace(nuevaLinea, "aoo", "au");
nuevaLinea = replace(nuevaLinea, "ae", "e");
nuevaLinea = replace(nuevaLinea, "ao", "o");
nuevaLinea = replace(nuevaLinea, "ee", "ai");
nuevaLinea = replace(nuevaLinea, "oo", "au");
nuevaLinea = replace(nuevaLinea, "\'I", "\'q");
nuevaLinea = replace(nuevaLinea, "I", "-i");
nuevaLinea = replace(nuevaLinea, "\'q", "-I");
nuevaLinea = replace(nuevaLinea, "\\", "?");
nuevaLinea = replace(nuevaLinea, "`", "!");
nuevaLinea = replace(nuevaLinea, "ga-y", "g.y");
nuevaLinea = replace(nuevaLinea, "g-y", "g.y");
nuevaLinea = replace(nuevaLinea, "na-y", "n+y");
/* nuevaLinea = Manipulate.replace(nuevaLinea, "ts", "tq");
nuevaLinea = Manipulate.replace(nuevaLinea, "tz", "ts");
nuevaLinea = Manipulate.replace(nuevaLinea, "tq", "tsh");
nuevaLinea = Manipulate.replace(nuevaLinea, "v", "w");
nuevaLinea = Manipulate.replace(nuevaLinea, "TH", "Th");
nuevaLinea = Manipulate.replace(nuevaLinea, "kSH", "k+Sh");
nuevaLinea = Manipulate.replace(nuevaLinea, "kaSH", "k+Sh");
nuevaLinea = Manipulate.replace(nuevaLinea, "SH", "Sh");
nuevaLinea = Manipulate.replace(nuevaLinea, ":", "H");
nuevaLinea = Manipulate.replace(nuevaLinea, "NH", "NaH");
nuevaLinea = Manipulate.replace(nuevaLinea, "dh", "d+h");
nuevaLinea = Manipulate.replace(nuevaLinea, "gh", "g+h");
nuevaLinea = Manipulate.replace(nuevaLinea, "bh", "b+h");
nuevaLinea = Manipulate.replace(nuevaLinea, "dzh", "dz+h");
nuevaLinea = Manipulate.replace(nuevaLinea, "aa", "a");
nuevaLinea = Manipulate.replace(nuevaLinea, "ai", "i");
nuevaLinea = Manipulate.replace(nuevaLinea, "aee", "ai");
nuevaLinea = Manipulate.replace(nuevaLinea, "au", "u");
nuevaLinea = Manipulate.replace(nuevaLinea, "aoo", "au");
nuevaLinea = Manipulate.replace(nuevaLinea, "ae", "e");
nuevaLinea = Manipulate.replace(nuevaLinea, "ao", "o");
nuevaLinea = Manipulate.replace(nuevaLinea, "ee", "ai");
nuevaLinea = Manipulate.replace(nuevaLinea, "oo", "au");
nuevaLinea = Manipulate.replace(nuevaLinea, "\'I", "\'q");
nuevaLinea = Manipulate.replace(nuevaLinea, "I", "-i");
nuevaLinea = Manipulate.replace(nuevaLinea, "\'q", "-I");
nuevaLinea = Manipulate.replace(nuevaLinea, "\\", "?");
nuevaLinea = Manipulate.replace(nuevaLinea, "`", "!");
nuevaLinea = Manipulate.replace(nuevaLinea, "ga-y", "g.y");
nuevaLinea = Manipulate.replace(nuevaLinea, "g-y", "g.y");
nuevaLinea = Manipulate.replace(nuevaLinea, "na-y", "n+y");
len = nuevaLinea.length();
for (i=0; i<len; i++)
@ -136,9 +137,9 @@ public class BasicTibetanTranscriptionConverter
{
chP = nuevaLinea.charAt(i-1);
chN = nuevaLinea.charAt(i+1);
if (isVowel(chN))
if (Manipulate.isVowel(chN))
{
if (Character.isLetter(chP) && !isVowel(chP))
if (Character.isLetter(chP) && !Manipulate.isVowel(chP))
{
nuevaLinea = nuevaLinea.substring(0, i) + Character.toUpperCase(chN) + nuevaLinea.substring(i+2);
len--;
@ -172,7 +173,7 @@ public class BasicTibetanTranscriptionConverter
}
}
}
nuevaLinea = replace(nuevaLinea, ",", "/");
nuevaLinea = Manipulate.replace(nuevaLinea, ",", "/");
return nuevaLinea; */
}
@ -191,8 +192,8 @@ public class BasicTibetanTranscriptionConverter
}
return tibDoc.getACIP(new boolean[] { false });
/* DLC FIXME: for unknown things, return null.
if (wylie.equals("@##")) return "#";
/* DLC FIXME: for unknown things, return null. */
/* if (wylie.equals("@##")) return "#";
if (wylie.equals("@#")) return "*";
if (wylie.equals("!")) return "`";
if (wylie.equals("b+h")) return "BH";
@ -213,7 +214,7 @@ public class BasicTibetanTranscriptionConverter
for (j=0; j<len; j++)
{
i = j;
/*ciclo:
//ciclo:
while(true) // para manejar excepciones; que honda!
{
switch(caract[i])
@ -229,7 +230,7 @@ public class BasicTibetanTranscriptionConverter
caract[i] = Character.toUpperCase(caract[i]);
else if (Character.isUpperCase(caract[i]))
caract[i] = Character.toLowerCase(caract[i]);
/* break ciclo;
// break ciclo;
}
}
}
@ -238,18 +239,18 @@ public class BasicTibetanTranscriptionConverter
// ahora hacer los cambios de Michael Roach
nuevaPalabra = replace(nuevaPalabra, "TSH", "TQQ");
nuevaPalabra = replace(nuevaPalabra, "TS", "TZ");
nuevaPalabra = replace(nuevaPalabra, "TQQ", "TS");
nuevaPalabra = replace(nuevaPalabra, "a", "'A");
nuevaPalabra = replace(nuevaPalabra, "i", "'I");
nuevaPalabra = replace(nuevaPalabra, "u", "'U");
nuevaPalabra = replace(nuevaPalabra, "-I", "i");
nuevaPalabra = replace(nuevaPalabra, "/", ",");
nuevaPalabra = replace(nuevaPalabra, "_", " ");
nuevaPalabra = replace(nuevaPalabra, "|", ";");
nuevaPalabra = fixWazur(nuevaPalabra);
return nuevaPalabra; */
nuevaPalabra = Manipulate.replace(nuevaPalabra, "TSH", "TQQ");
nuevaPalabra = Manipulate.replace(nuevaPalabra, "TS", "TZ");
nuevaPalabra = Manipulate.replace(nuevaPalabra, "TQQ", "TS");
nuevaPalabra = Manipulate.replace(nuevaPalabra, "a", "'A");
nuevaPalabra = Manipulate.replace(nuevaPalabra, "i", "'I");
nuevaPalabra = Manipulate.replace(nuevaPalabra, "u", "'U");
nuevaPalabra = Manipulate.replace(nuevaPalabra, "-I", "i");
nuevaPalabra = Manipulate.replace(nuevaPalabra, "/", ",");
nuevaPalabra = Manipulate.replace(nuevaPalabra, "_", " ");
nuevaPalabra = Manipulate.replace(nuevaPalabra, "|", ";");
nuevaPalabra = Manipulate.fixWazur(nuevaPalabra);
return nuevaPalabra;*/
}
/** Converts Tibetan Unicode to EWTS. */
@ -291,7 +292,7 @@ public class BasicTibetanTranscriptionConverter
public static void printSyntax()
{
System.out.println("Syntax: NewBasicTibetanTranscriptionConverter [-format format-of-files | [-fi format-of-input-file] [-fo format-of-output-file]] [-it acip | wylie | utf8] [-ot acip | wylie | utf8] input-file [output-file]");
System.out.println("Syntax: BasicTibetanTranscriptionConverter [-format format-of-files | [-fi format-of-input-file] [-fo format-of-output-file]] [-it acip | wylie | UTF16] [-ot acip | wylie | UTF16] input-file [output-file]");
}
public BasicTibetanTranscriptionConverter(BufferedReader in, PrintWriter out)
@ -309,7 +310,7 @@ public class BasicTibetanTranscriptionConverter
String option;
String formatIn = null, formatOut = null, inputTransSyst="wylie", outputTransSyst="wylie";
boolean file = false;
int conversionType=0;
if (argNum<=currentArg)
{
@ -382,11 +383,26 @@ public class BasicTibetanTranscriptionConverter
in = getBufferedReader (args[currentArg], formatIn);
}
new BasicTibetanTranscriptionConverter(in, out).run();
new BasicTibetanTranscriptionConverter(in, out).run(conversionType);
}
/**
* This method was added for compatibility's sake with the FontConverterConstants interfase.
*
* @param conversionType
* @throws IOException
*/
public void run(String conversionType) throws IOException
{
int conversionTypeInt=0;
if (conversionType==ACIP_TO_WYLIE_TEXT) conversionTypeInt = ACIP_TO_WYLIE;
if (conversionType==WYLIE_TO_ACIP_TEXT) conversionTypeInt = WYLIE_TO_ACIP;
if (conversionType==UNI_TO_WYLIE_TEXT) conversionTypeInt = UNICODE_TO_WYLIE;
if (conversionType==WYLIE_TO_UNI_TEXT) conversionTypeInt = WYLIE_TO_UNICODE;
run(conversionTypeInt);
}
public void run() throws Exception
public void run(int conversionType) throws IOException
{
String linea, result;