/*
* Copyright (c) 2003, 2010, Oracle and/or its affiliates. All rights reserved.
* DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
*
* This code is free software; you can redistribute it and/or modify it
* under the terms of the GNU General Public License version 2 only, as
* published by the Free Software Foundation. Oracle designates this
* particular file as subject to the "Classpath" exception as provided
* by Oracle in the LICENSE file that accompanied this code.
*
* This code is distributed in the hope that it will be useful, but WITHOUT
* ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
* FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
* version 2 for more details (a copy is included in the LICENSE file that
* accompanied this code).
*
* You should have received a copy of the GNU General Public License version
* 2 along with this work; if not, write to the Free Software Foundation,
* Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
*
* Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA
* or visit www.oracle.com if you need additional information or have any
* questions.
*/
package sun.font;
import java.awt.Font;
import java.awt.font.FontRenderContext;
import java.awt.geom.AffineTransform;
import java.lang.ref.Reference;
import java.lang.ref.SoftReference;
import java.util.concurrent.ConcurrentHashMap;
import java.util.Locale;
public abstract class Font2D {
/* Note: JRE and FONT_CONFIG ranks are identical. I don't know of a reason
* to distingish these. Possibly if a user adds fonts to the JRE font
* directory that are the same font as the ones specified in the font
* configuration but that is more likely to be the legitimate intention
* than a problem. One reason why these should be the same is that on
* Linux the JRE fonts ARE the font configuration fonts, and although I
* believe all are assigned FONT_CONFIG rank, it is conceivable that if
* this were not so, that some JRE font would not be allowed to joint the
* family of its siblings which were assigned FONT_CONFIG rank. Giving
* them the same rank is the easy solution for now at least.
*/
public static final int FONT_CONFIG_RANK = 2;
public static final int JRE_RANK = 2;
public static final int TTF_RANK = 3;
public static final int TYPE1_RANK = 4;
public static final int NATIVE_RANK = 5;
public static final int UNKNOWN_RANK = 6;
public static final int DEFAULT_RANK = 4;
private static final String[] boldNames = {
"bold", "demibold", "demi-bold", "demi bold", "negreta", "demi", };
private static final String[] italicNames = {
"italic", "cursiva", "oblique", "inclined", };
private static final String[] boldItalicNames = {
"bolditalic", "bold-italic", "bold italic",
"boldoblique", "bold-oblique", "bold oblique",
"demibold italic", "negreta cursiva","demi oblique", };
private static final FontRenderContext DEFAULT_FRC =
new FontRenderContext(null, false, false);
public Font2DHandle handle;
protected String familyName; /* Family font name (english) */
protected String fullName; /* Full font name (english) */
protected int style = Font.PLAIN;
protected FontFamily family;
protected int fontRank = DEFAULT_RANK;
/*
* A mapper can be independent of the strike.
* Perhaps the reference to the mapper ought to be held on the
* scaler, as it may be implemented via scaler functionality anyway
* and so the mapper would be useless if its native portion was
* freed when the scaler was GC'd.
*/
protected CharToGlyphMapper mapper;
/*
* The strike cache is maintained per "Font2D" as that is the
* principal object by which you look up fonts.
* It means more Hashmaps, but look ups can be quicker because
* the map will have fewer entries, and there's no need to try to
* make the Font2D part of the key.
*/
protected ConcurrentHashMap<FontStrikeDesc, Reference>
strikeCache = new ConcurrentHashMap<FontStrikeDesc, Reference>();
/* Store the last Strike in a Reference object.
* Similarly to the strike that was stored on a C++ font object,
* this is an optimisation which helps if multiple clients (ie
* typically SunGraphics2D instances) are using the same font, then
* as may be typical of many UIs, they are probably using it in the
* same style, so it can be a win to first quickly check if the last
* strike obtained from this Font2D satifies the needs of the next
* client too.
* This pre-supposes that a FontStrike is a shareable object, which
* it should.
*/
protected Reference lastFontStrike = new SoftReference(null);
/*
* POSSIBLE OPTIMISATION:
* Array of length 1024 elements of 64 bits indicating if a font
* contains these. This kind of information can be shared between
* all point sizes.
* if corresponding bit in knownBitmaskMap is set then canDisplayBitmaskMap
* is valid. This is 16Kbytes of data per composite font style.
* What about UTF-32 and surrogates?
* REMIND: This is too much storage. Probably can only cache this
* information for latin range, although possibly OK to store all
* for just the "logical" fonts.
* Or instead store arrays of subranges of 1024 bits (128 bytes) in
* the range below surrogate pairs.
*/
// protected long[] knownBitmaskMap;
// protected long[] canDisplayBitmaskMap;
/* Returns the "real" style of this Font2D. Eg the font face
* Lucida Sans Bold" has a real style of Font.BOLD, even though
* it may be able to used to simulate bold italic
*/
public int getStyle() {
return style;
}
protected void setStyle() {
String fName = fullName.toLowerCase();
for (int i=0; i < boldItalicNames.length; i++) {
if (fName.indexOf(boldItalicNames[i]) != -1) {
style = Font.BOLD|Font.ITALIC;
return;
}
}
for (int i=0; i < italicNames.length; i++) {
if (fName.indexOf(italicNames[i]) != -1) {
style = Font.ITALIC;
return;
}
}
for (int i=0; i < boldNames.length; i++) {
if (fName.indexOf(boldNames[i]) != -1 ) {
style = Font.BOLD;
return;
}
}
}
int getRank() {
return fontRank;
}
void setRank(int rank) {
fontRank = rank;
}
abstract CharToGlyphMapper getMapper();
/* This isn't very efficient but its infrequently used.
* StandardGlyphVector uses it when the client assigns the glyph codes.
* These may not be valid. This validates them substituting the missing
* glyph elsewhere.
*/
protected int getValidatedGlyphCode(int glyphCode) {
if (glyphCode < 0 || glyphCode >= getMapper().getNumGlyphs()) {
glyphCode = getMapper().getMissingGlyphCode();
}
return glyphCode;
}
/*
* Creates an appropriate strike for the Font2D subclass
*/
abstract FontStrike createStrike(FontStrikeDesc desc);
/* this may be useful for APIs like canDisplay where the answer
* is dependent on the font and its scaler, but not the strike.
* If no strike has ever been returned, then create a one that matches
* this font with the default FRC. It will become the lastStrike and
* there's a good chance that the next call will be to get exactly that
* strike.
*/
public FontStrike getStrike(Font font) {
FontStrike strike = (FontStrike)lastFontStrike.get();
if (strike != null) {
return strike;
} else {
return getStrike(font, DEFAULT_FRC);
}
}
/* SunGraphics2D has font, tx, aa and fm. From this info
* can get a Strike object from the cache, creating it if necessary.
* This code is designed for multi-threaded access.
* For that reason it creates a local FontStrikeDesc rather than filling
* in a shared one. Up to two AffineTransforms and one FontStrikeDesc will
* be created by every lookup. This appears to perform more than
* adequately. But it may make sense to expose FontStrikeDesc
* as a parameter so a caller can use its own.
* In such a case if a FontStrikeDesc is stored as a key then
* we would need to use a private copy.
*
* Note that this code doesn't prevent two threads from creating
* two different FontStrike instances and having one of the threads
* overwrite the other in the map. This is likely to be a rare
* occurrence and the only consequence is that these callers will have
* different instances of the strike, and there'd be some duplication of
* population of the strikes. However since users of these strikes are
* transient, then the one that was overwritten would soon be freed.
* If there is any problem then a small synchronized block would be
* required with its attendant consequences for MP scaleability.
*/
public FontStrike getStrike(Font font, AffineTransform devTx,
int aa, int fm) {
/* Create the descriptor which is used to identify a strike
* in the strike cache/map. A strike is fully described by
* the attributes of this descriptor.
*/
/* REMIND: generating garbage and doing computation here in order
* to include pt size in the tx just for a lookup! Figure out a
* better way.
*/
double ptSize = font.getSize2D();
AffineTransform glyphTx = (AffineTransform)devTx.clone();
glyphTx.scale(ptSize, ptSize);
if (font.isTransformed()) {
glyphTx.concatenate(font.getTransform());
}
if (glyphTx.getTranslateX() != 0 || glyphTx.getTranslateY() != 0) {
glyphTx.setTransform(glyphTx.getScaleX(),
glyphTx.getShearY(),
glyphTx.getShearX(),
glyphTx.getScaleY(),
0.0, 0.0);
}
FontStrikeDesc desc = new FontStrikeDesc(devTx, glyphTx,
font.getStyle(), aa, fm);
return getStrike(desc, false);
}
public FontStrike getStrike(Font font, AffineTransform devTx,
AffineTransform glyphTx,
int aa, int fm) {
/* Create the descriptor which is used to identify a strike
* in the strike cache/map. A strike is fully described by
* the attributes of this descriptor.
*/
FontStrikeDesc desc = new FontStrikeDesc(devTx, glyphTx,
font.getStyle(), aa, fm);
return getStrike(desc, false);
}
public FontStrike getStrike(Font font, FontRenderContext frc) {
AffineTransform at = frc.getTransform();
double ptSize = font.getSize2D();
at.scale(ptSize, ptSize);
if (font.isTransformed()) {
at.concatenate(font.getTransform());
if (at.getTranslateX() != 0 || at.getTranslateY() != 0) {
at.setTransform(at.getScaleX(),
at.getShearY(),
at.getShearX(),
at.getScaleY(),
0.0, 0.0);
}
}
int aa = FontStrikeDesc.getAAHintIntVal(this, font, frc);
int fm = FontStrikeDesc.getFMHintIntVal(frc.getFractionalMetricsHint());
FontStrikeDesc desc = new FontStrikeDesc(frc.getTransform(),
at, font.getStyle(),
aa, fm);
return getStrike(desc, false);
}
FontStrike getStrike(FontStrikeDesc desc) {
return getStrike(desc, true);
}
private FontStrike getStrike(FontStrikeDesc desc, boolean copy) {
/* Before looking in the map, see if the descriptor matches the
* last strike returned from this Font2D. This should often be a win
* since its common for the same font, in the same size to be
* used frequently, for example in many parts of a UI.
*
* If its not the same then we use the descriptor to locate a
* Reference to the strike. If it exists and points to a strike,
* then we update the last strike to refer to that and return it.
*
* If the key isn't in the map, or its reference object has been
* collected, then we create a new strike, put it in the map and
* set it to be the last strike.
*/
FontStrike strike = (FontStrike)lastFontStrike.get();
if (strike != null && desc.equals(strike.desc)) {
//strike.lastlookupTime = System.currentTimeMillis();
return strike;
} else {
Reference strikeRef = strikeCache.get(desc);
if (strikeRef != null) {
strike = (FontStrike)strikeRef.get();
if (strike != null) {
//strike.lastlookupTime = System.currentTimeMillis();
lastFontStrike = new SoftReference(strike);
StrikeCache.refStrike(strike);
return strike;
}
}
/* When we create a new FontStrike instance, we *must*
* ask the StrikeCache for a reference. We must then ensure
* this reference remains reachable, by storing it in the
* Font2D's strikeCache map.
* So long as the Reference is there (reachable) then if the
* reference is cleared, it will be enqueued for disposal.
* If for some reason we explicitly remove this reference, it
* must only be done when holding a strong reference to the
* referent (the FontStrike), or if the reference is cleared,
* then we must explicitly "dispose" of the native resources.
* The only place this currently happens is in this same method,
* where we find a cleared reference and need to overwrite it
* here with a new reference.
* Clearing the whilst holding a strong reference, should only
* be done if the
*/
if (copy) {
desc = new FontStrikeDesc(desc);
}
strike = createStrike(desc);
//StrikeCache.addStrike();
/* If we are creating many strikes on this font which
* involve non-quadrant rotations, or more general
* transforms which include shears, then force the use
* of weak references rather than soft references.
* This means that it won't live much beyond the next GC,
* which is what we want for what is likely a transient strike.
*/
int txType = desc.glyphTx.getType();
if (txType == AffineTransform.TYPE_GENERAL_TRANSFORM ||
(txType & AffineTransform.TYPE_GENERAL_ROTATION) != 0 &&
strikeCache.size() > 10) {
strikeRef = StrikeCache.getStrikeRef(strike, true);
} else {
strikeRef = StrikeCache.getStrikeRef(strike);
}
strikeCache.put(desc, strikeRef);
//strike.lastlookupTime = System.currentTimeMillis();
lastFontStrike = new SoftReference(strike);
StrikeCache.refStrike(strike);
return strike;
}
}
void removeFromCache(FontStrikeDesc desc) {
Reference ref = strikeCache.get(desc);
if (ref != null) {
Object o = ref.get();
if (o == null) {
strikeCache.remove(desc);
}
}
}
/**
* The length of the metrics array must be >= 8. This method will
* store the following elements in that array before returning:
* metrics[0]: ascent
* metrics[1]: descent
* metrics[2]: leading
* metrics[3]: max advance
* metrics[4]: strikethrough offset
* metrics[5]: strikethrough thickness
* metrics[6]: underline offset
* metrics[7]: underline thickness
*/
public void getFontMetrics(Font font, AffineTransform at,
Object aaHint, Object fmHint,
float metrics[]) {
/* This is called in just one place in Font with "at" == identity.
* Perhaps this can be eliminated.
*/
int aa = FontStrikeDesc.getAAHintIntVal(aaHint, this, font.getSize());
int fm = FontStrikeDesc.getFMHintIntVal(fmHint);
FontStrike strike = getStrike(font, at, aa, fm);
StrikeMetrics strikeMetrics = strike.getFontMetrics();
metrics[0] = strikeMetrics.getAscent();
metrics[1] = strikeMetrics.getDescent();
metrics[2] = strikeMetrics.getLeading();
metrics[3] = strikeMetrics.getMaxAdvance();
getStyleMetrics(font.getSize2D(), metrics, 4);
}
/**
* The length of the metrics array must be >= offset+4, and offset must be
* >= 0. Typically offset is 4. This method will
* store the following elements in that array before returning:
* metrics[off+0]: strikethrough offset
* metrics[off+1]: strikethrough thickness
* metrics[off+2]: underline offset
* metrics[off+3]: underline thickness
*
* Note that this implementation simply returns default values;
* subclasses can override this method to provide more accurate values.
*/
public void getStyleMetrics(float pointSize, float[] metrics, int offset) {
metrics[offset] = -metrics[0] / 2.5f;
metrics[offset+1] = pointSize / 12;
metrics[offset+2] = metrics[offset+1] / 1.5f;
metrics[offset+3] = metrics[offset+1];
}
/**
* The length of the metrics array must be >= 4. This method will
* store the following elements in that array before returning:
* metrics[0]: ascent
* metrics[1]: descent
* metrics[2]: leading
* metrics[3]: max advance
*/
public void getFontMetrics(Font font, FontRenderContext frc,
float metrics[]) {
StrikeMetrics strikeMetrics = getStrike(font, frc).getFontMetrics();
metrics[0] = strikeMetrics.getAscent();
metrics[1] = strikeMetrics.getDescent();
metrics[2] = strikeMetrics.getLeading();
metrics[3] = strikeMetrics.getMaxAdvance();
}
/* Currently the layout code calls this. May be better for layout code
* to check the font class before attempting to run, rather than needing
* to promote this method up from TrueTypeFont
*/
byte[] getTableBytes(int tag) {
return null;
}
/* for layout code */
protected long getUnitsPerEm() {
return 2048;
}
boolean supportsEncoding(String encoding) {
return false;
}
public boolean canDoStyle(int style) {
return (style == this.style);
}
/*
* All the important subclasses override this which is principally for
* the TrueType 'gasp' table.
*/
public boolean useAAForPtSize(int ptsize) {
return true;
}
public boolean hasSupplementaryChars() {
return false;
}
/* The following methods implement public methods on java.awt.Font */
public String getPostscriptName() {
return fullName;
}
public String getFontName(Locale l) {
return fullName;
}
public String getFamilyName(Locale l) {
return familyName;
}
public int getNumGlyphs() {
return getMapper().getNumGlyphs();
}
public int charToGlyph(int wchar) {
return getMapper().charToGlyph(wchar);
}
public int getMissingGlyphCode() {
return getMapper().getMissingGlyphCode();
}
public boolean canDisplay(char c) {
return getMapper().canDisplay(c);
}
public boolean canDisplay(int cp) {
return getMapper().canDisplay(cp);
}
public byte getBaselineFor(char c) {
return Font.ROMAN_BASELINE;
}
public float getItalicAngle(Font font, AffineTransform at,
Object aaHint, Object fmHint) {
/* hardwire psz=12 as that's typical and AA vs non-AA for 'gasp' mode
* isn't important for the caret slope of this rarely used API.
*/
int aa = FontStrikeDesc.getAAHintIntVal(aaHint, this, 12);
int fm = FontStrikeDesc.getFMHintIntVal(fmHint);
FontStrike strike = getStrike(font, at, aa, fm);
StrikeMetrics metrics = strike.getFontMetrics();
if (metrics.ascentY == 0 || metrics.ascentX == 0) {
return 0f;
} else {
/* ascent is "up" from the baseline so its typically
* a negative value, so we need to compensate
*/
return metrics.ascentX/-metrics.ascentY;
}
}
}