/* -*- Mode: C++; tab-width: 20; indent-tabs-mode: nil; c-basic-offset: 4 -*- * ***** BEGIN LICENSE BLOCK ***** * Version: MPL 1.1/GPL 2.0/LGPL 2.1 * * The contents of this file are subject to the Mozilla Public License Version * 1.1 (the "License"); you may not use this file except in compliance with * the License. You may obtain a copy of the License at * http://www.mozilla.org/MPL/ * * Software distributed under the License is distributed on an "AS IS" basis, * WITHOUT WARRANTY OF ANY KIND, either express or implied. See the License * for the specific language governing rights and limitations under the * License. * * The Original Code is Mozilla Foundation code. * * The Initial Developer of the Original Code is Mozilla Foundation. * Portions created by the Initial Developer are Copyright (C) 2005-2009 * the Initial Developer. All Rights Reserved. * * Contributor(s): * Stuart Parmenter * Masayuki Nakano * John Daggett * Jonathan Kew * * Alternatively, the contents of this file may be used under the terms of * either the GNU General Public License Version 2 or later (the "GPL"), or * the GNU Lesser General Public License Version 2.1 or later (the "LGPL"), * in which case the provisions of the GPL or the LGPL are applicable instead * of those above. If you wish to allow use of your version of this file only * under the terms of either the GPL or the LGPL, and not to allow others to * use your version of this file under the terms of the MPL, indicate your * decision by deleting the provisions above and replace them with the notice * and other provisions required by the GPL or the LGPL. If you do not delete * the provisions above, a recipient may use your version of this file under * the terms of any one of the MPL, the GPL or the LGPL. * * ***** END LICENSE BLOCK ***** */ #ifndef GFX_FONT_H #define GFX_FONT_H #include "prtypes.h" #include "nsAlgorithm.h" #include "gfxTypes.h" #include "nsString.h" #include "gfxPoint.h" #include "gfxFontUtils.h" #include "nsTArray.h" #include "nsTHashtable.h" #include "nsHashKeys.h" #include "gfxSkipChars.h" #include "gfxRect.h" #include "nsExpirationTracker.h" #include "gfxFontConstants.h" #include "gfxPlatform.h" #include "nsIAtom.h" #include "nsISupportsImpl.h" typedef struct _cairo_scaled_font cairo_scaled_font_t; #ifdef DEBUG #include #endif class gfxContext; class gfxTextRun; class gfxFont; class gfxFontFamily; class gfxFontGroup; class gfxUserFontSet; class gfxUserFontData; class gfxShapedWord; class nsILanguageAtomService; typedef struct _hb_blob_t hb_blob_t; // We should eliminate these synonyms when it won't cause many merge conflicts. #define FONT_STYLE_NORMAL NS_FONT_STYLE_NORMAL #define FONT_STYLE_ITALIC NS_FONT_STYLE_ITALIC #define FONT_STYLE_OBLIQUE NS_FONT_STYLE_OBLIQUE // We should eliminate these synonyms when it won't cause many merge conflicts. #define FONT_WEIGHT_NORMAL NS_FONT_WEIGHT_NORMAL #define FONT_WEIGHT_BOLD NS_FONT_WEIGHT_BOLD #define FONT_MAX_SIZE 2000.0 #define NO_FONT_LANGUAGE_OVERRIDE 0 // An OpenType feature tag and value pair struct THEBES_API gfxFontFeature { PRUint32 mTag; // see http://www.microsoft.com/typography/otspec/featuretags.htm PRUint32 mValue; // 0 = off, 1 = on, larger values may be used as parameters // to features that select among multiple alternatives }; inline bool operator<(const gfxFontFeature& a, const gfxFontFeature& b) { return (a.mTag < b.mTag) || ((a.mTag == b.mTag) && (a.mValue < b.mValue)); } inline bool operator==(const gfxFontFeature& a, const gfxFontFeature& b) { return (a.mTag == b.mTag) && (a.mValue == b.mValue); } struct THEBES_API gfxFontStyle { gfxFontStyle(); gfxFontStyle(PRUint8 aStyle, PRUint16 aWeight, PRInt16 aStretch, gfxFloat aSize, nsIAtom *aLanguage, float aSizeAdjust, bool aSystemFont, bool aPrinterFont, const nsString& aFeatureSettings, const nsString& aLanguageOverride); gfxFontStyle(const gfxFontStyle& aStyle); // The style of font (normal, italic, oblique) PRUint8 style : 7; // Say that this font is a system font and therefore does not // require certain fixup that we do for fonts from untrusted // sources. bool systemFont : 1; // Say that this font is used for print or print preview. bool printerFont : 1; // The weight of the font: 100, 200, ... 900. PRUint16 weight; // The stretch of the font (the sum of various NS_FONT_STRETCH_* // constants; see gfxFontConstants.h). PRInt16 stretch; // The logical size of the font, in pixels gfxFloat size; // The aspect-value (ie., the ratio actualsize:actualxheight) that any // actual physical font created from this font structure must have when // rendering or measuring a string. A value of 0 means no adjustment // needs to be done. float sizeAdjust; // the language (may be an internal langGroup code rather than an actual // language code) specified in the document or element's lang property, // or inferred from the charset nsRefPtr language; // Language system tag, to override document language; // an OpenType "language system" tag represented as a 32-bit integer // (see http://www.microsoft.com/typography/otspec/languagetags.htm). // Normally 0, so font rendering will use the document or element language // (see above) to control any language-specific rendering, but the author // can override this for cases where the options implemented in the font // do not directly match the actual language. (E.g. lang may be Macedonian, // but the font in use does not explicitly support this; the author can // use font-language-override to request the Serbian option in the font // in order to get correct glyph shapes.) PRUint32 languageOverride; // custom opentype feature settings nsTArray featureSettings; // Return the final adjusted font size for the given aspect ratio. // Not meant to be called when sizeAdjust = 0. gfxFloat GetAdjustedSize(gfxFloat aspect) const { NS_ASSERTION(sizeAdjust != 0.0, "Not meant to be called when sizeAdjust = 0"); gfxFloat adjustedSize = NS_MAX(NS_round(size*(sizeAdjust/aspect)), 1.0); return NS_MIN(adjustedSize, FONT_MAX_SIZE); } PLDHashNumber Hash() const { return ((style + (systemFont << 7) + (weight << 8)) + PRUint32(size*1000) + PRUint32(sizeAdjust*1000)) ^ nsISupportsHashKey::HashKey(language); } PRInt8 ComputeWeight() const; bool Equals(const gfxFontStyle& other) const { return (size == other.size) && (style == other.style) && (systemFont == other.systemFont) && (printerFont == other.printerFont) && (weight == other.weight) && (stretch == other.stretch) && (language == other.language) && (sizeAdjust == other.sizeAdjust) && (featureSettings == other.featureSettings) && (languageOverride == other.languageOverride); } static void ParseFontFeatureSettings(const nsString& aFeatureString, nsTArray& aFeatures); static PRUint32 ParseFontLanguageOverride(const nsString& aLangTag); }; class gfxFontEntry { public: NS_INLINE_DECL_REFCOUNTING(gfxFontEntry) gfxFontEntry(const nsAString& aName, gfxFontFamily *aFamily = nsnull, bool aIsStandardFace = false) : mName(aName), mItalic(false), mFixedPitch(false), mIsProxy(false), mIsValid(true), mIsBadUnderlineFont(false), mIsUserFont(false), mIsLocalUserFont(false), mStandardFace(aIsStandardFace), mSymbolFont(false), mIgnoreGDEF(false), mWeight(500), mStretch(NS_FONT_STRETCH_NORMAL), #ifdef MOZ_GRAPHITE mCheckedForGraphiteTables(false), #endif mHasCmapTable(false), mCmapInitialized(false), mUVSOffset(0), mUVSData(nsnull), mUserFontData(nsnull), mLanguageOverride(NO_FONT_LANGUAGE_OVERRIDE), mFamily(aFamily) { } virtual ~gfxFontEntry(); // unique name for the face, *not* the family; not necessarily the // "real" or user-friendly name, may be an internal identifier const nsString& Name() const { return mName; } // the "real" name of the face, if available from the font resource // (may be expensive); returns Name() if nothing better is available virtual nsString RealFaceName(); gfxFontFamily* Family() const { return mFamily; } PRUint16 Weight() const { return mWeight; } PRInt16 Stretch() const { return mStretch; } bool IsUserFont() const { return mIsUserFont; } bool IsLocalUserFont() const { return mIsLocalUserFont; } bool IsFixedPitch() const { return mFixedPitch; } bool IsItalic() const { return mItalic; } bool IsBold() const { return mWeight >= 600; } // bold == weights 600 and above bool IgnoreGDEF() const { return mIgnoreGDEF; } virtual bool IsSymbolFont(); #ifdef MOZ_GRAPHITE inline bool HasGraphiteTables() { if (!mCheckedForGraphiteTables) { CheckForGraphiteTables(); mCheckedForGraphiteTables = true; } return mHasGraphiteTables; } #endif inline bool HasCmapTable() { if (!mCmapInitialized) { ReadCMAP(); } return mHasCmapTable; } inline bool HasCharacter(PRUint32 ch) { if (mCharacterMap.test(ch)) return true; return TestCharacterMap(ch); } virtual bool SkipDuringSystemFallback() { return false; } virtual bool TestCharacterMap(PRUint32 aCh); nsresult InitializeUVSMap(); PRUint16 GetUVSGlyph(PRUint32 aCh, PRUint32 aVS); virtual nsresult ReadCMAP(); virtual bool MatchesGenericFamily(const nsACString& aGeneric) const { return true; } virtual bool SupportsLangGroup(nsIAtom *aLangGroup) const { return true; } virtual nsresult GetFontTable(PRUint32 aTableTag, FallibleTArray& aBuffer) { return NS_ERROR_FAILURE; // all platform subclasses should reimplement this! } void SetFamily(gfxFontFamily* aFamily) { mFamily = aFamily; } virtual nsString FamilyName() const; already_AddRefed FindOrMakeFont(const gfxFontStyle *aStyle, bool aNeedsBold); // Get an existing font table cache entry in aBlob if it has been // registered, or return false if not. Callers must call // hb_blob_destroy on aBlob if true is returned. // // Note that some gfxFont implementations may not call this at all, // if it is more efficient to get the table from the OS at that level. bool GetExistingFontTable(PRUint32 aTag, hb_blob_t** aBlob); // Elements of aTable are transferred (not copied) to and returned in a // new hb_blob_t which is registered on the gfxFontEntry, but the initial // reference is owned by the caller. Removing the last reference // unregisters the table from the font entry. // // Pass NULL for aBuffer to indicate that the table is not present and // NULL will be returned. Also returns NULL on OOM. hb_blob_t *ShareFontTableAndGetBlob(PRUint32 aTag, FallibleTArray* aTable); nsString mName; bool mItalic : 1; bool mFixedPitch : 1; bool mIsProxy : 1; bool mIsValid : 1; bool mIsBadUnderlineFont : 1; bool mIsUserFont : 1; bool mIsLocalUserFont : 1; bool mStandardFace : 1; bool mSymbolFont : 1; bool mIgnoreGDEF : 1; PRUint16 mWeight; PRInt16 mStretch; #ifdef MOZ_GRAPHITE bool mHasGraphiteTables; bool mCheckedForGraphiteTables; #endif bool mHasCmapTable; bool mCmapInitialized; gfxSparseBitSet mCharacterMap; PRUint32 mUVSOffset; nsAutoArrayPtr mUVSData; gfxUserFontData* mUserFontData; nsTArray mFeatureSettings; PRUint32 mLanguageOverride; protected: friend class gfxPlatformFontList; friend class gfxMacPlatformFontList; friend class gfxUserFcFontEntry; friend class gfxFontFamily; friend class gfxSingleFaceMacFontFamily; gfxFontEntry() : mItalic(false), mFixedPitch(false), mIsProxy(false), mIsValid(true), mIsBadUnderlineFont(false), mIsUserFont(false), mIsLocalUserFont(false), mStandardFace(false), mSymbolFont(false), mIgnoreGDEF(false), mWeight(500), mStretch(NS_FONT_STRETCH_NORMAL), #ifdef MOZ_GRAPHITE mCheckedForGraphiteTables(false), #endif mHasCmapTable(false), mCmapInitialized(false), mUVSOffset(0), mUVSData(nsnull), mUserFontData(nsnull), mLanguageOverride(NO_FONT_LANGUAGE_OVERRIDE), mFamily(nsnull) { } virtual gfxFont *CreateFontInstance(const gfxFontStyle *aFontStyle, bool aNeedsBold) { NS_NOTREACHED("oops, somebody didn't override CreateFontInstance"); return nsnull; } #ifdef MOZ_GRAPHITE virtual void CheckForGraphiteTables(); #endif gfxFontFamily *mFamily; private: /** * Font table hashtable, to support GetFontTable for harfbuzz. * * The harfbuzz shaper (and potentially other clients) needs access to raw * font table data. This needs to be cached so that it can be used * repeatedly (each time we construct a text run; in some cases, for * each character/glyph within the run) without re-fetching large tables * every time. * * Because we may instantiate many gfxFonts for the same physical font * file (at different sizes), we should ensure that they can share a * single cached copy of the font tables. To do this, we implement table * access and sharing on the fontEntry rather than the font itself. * * The default implementation uses GetFontTable() to read font table * data into byte arrays, and wraps them in blobs which are registered in * a hashtable. The hashtable can then return pre-existing blobs to * harfbuzz. * * Harfbuzz will "destroy" the blobs when it is finished with them. When * the last blob reference is removed, the FontTableBlobData user data * will remove the blob from the hashtable if still registered. */ class FontTableBlobData; /** * FontTableHashEntry manages the entries of hb_blob_ts for two * different situations: * * The common situation is to share font table across fonts with the same * font entry (but different sizes) for use by HarfBuzz. The hashtable * does not own a strong reference to the blob, but keeps a weak pointer, * managed by FontTableBlobData. Similarly FontTableBlobData keeps only a * weak pointer to the hashtable, managed by FontTableHashEntry. * * Some font tables are saved here before they would get stripped by OTS * sanitizing. These are retained for harfbuzz, which does its own * sanitizing. The hashtable owns a reference, so ownership is simple. */ class FontTableHashEntry : public nsUint32HashKey { public: // Declarations for nsTHashtable typedef nsUint32HashKey KeyClass; typedef KeyClass::KeyType KeyType; typedef KeyClass::KeyTypePointer KeyTypePointer; FontTableHashEntry(KeyTypePointer aTag) : KeyClass(aTag), mBlob() { }; // Copying transfers blob association. FontTableHashEntry(FontTableHashEntry& toCopy) : KeyClass(toCopy), mBlob(toCopy.mBlob) { toCopy.mBlob = nsnull; } ~FontTableHashEntry() { Clear(); } // FontTable/Blob API // Transfer (not copy) elements of aTable to a new hb_blob_t and // return ownership to the caller. A weak reference to the blob is // recorded in the hashtable entry so that others may use the same // table. hb_blob_t * ShareTableAndGetBlob(FallibleTArray& aTable, nsTHashtable *aHashtable); // Transfer (not copy) elements of aTable to a new hb_blob_t that is // owned by the hashtable entry. void SaveTable(FallibleTArray& aTable); // Return a strong reference to the blob. // Callers must hb_blob_destroy the returned blob. hb_blob_t *GetBlob() const; void Clear(); private: static void DeleteFontTableBlobData(void *aBlobData); // not implemented FontTableHashEntry& operator=(FontTableHashEntry& toCopy); FontTableBlobData *mSharedBlobData; hb_blob_t *mBlob; }; nsTHashtable mFontTableCache; gfxFontEntry(const gfxFontEntry&); gfxFontEntry& operator=(const gfxFontEntry&); }; // used when picking fallback font struct FontSearch { FontSearch(const PRUint32 aCharacter, gfxFont *aFont) : mCh(aCharacter), mFontToMatch(aFont), mMatchRank(0), mCount(0) { } const PRUint32 mCh; gfxFont* mFontToMatch; PRInt32 mMatchRank; nsRefPtr mBestMatch; PRUint32 mCount; }; class gfxFontFamily { public: NS_INLINE_DECL_REFCOUNTING(gfxFontFamily) gfxFontFamily(const nsAString& aName) : mName(aName), mOtherFamilyNamesInitialized(false), mHasOtherFamilyNames(false), mFaceNamesInitialized(false), mHasStyles(false), mIsSimpleFamily(false), mIsBadUnderlineFamily(false) { } virtual ~gfxFontFamily() { } const nsString& Name() { return mName; } virtual void LocalizedName(nsAString& aLocalizedName); virtual bool HasOtherFamilyNames(); nsTArray >& GetFontList() { return mAvailableFonts; } void AddFontEntry(nsRefPtr aFontEntry) { // bug 589682 - set the IgnoreGDEF flag on entries for Italic faces // of Times New Roman, because of buggy table in those fonts if (aFontEntry->IsItalic() && !aFontEntry->IsUserFont() && Name().EqualsLiteral("Times New Roman")) { aFontEntry->mIgnoreGDEF = true; } mAvailableFonts.AppendElement(aFontEntry); aFontEntry->SetFamily(this); } // note that the styles for this family have been added void SetHasStyles(bool aHasStyles) { mHasStyles = aHasStyles; } // choose a specific face to match a style using CSS font matching // rules (weight matching occurs here). may return a face that doesn't // precisely match (e.g. normal face when no italic face exists). // aNeedsSyntheticBold is set to true when synthetic bolding is // needed, false otherwise gfxFontEntry *FindFontForStyle(const gfxFontStyle& aFontStyle, bool& aNeedsSyntheticBold); // iterates over faces looking for a match with a given characters // used as part of the font fallback process void FindFontForChar(FontSearch *aMatchData); // read in other family names, if any, and use functor to add each into cache virtual void ReadOtherFamilyNames(gfxPlatformFontList *aPlatformFontList); // set when other family names have been read in void SetOtherFamilyNamesInitialized() { mOtherFamilyNamesInitialized = true; } // read in other localized family names, fullnames and Postscript names // for all faces and append to lookup tables virtual void ReadFaceNames(gfxPlatformFontList *aPlatformFontList, bool aNeedFullnamePostscriptNames); // find faces belonging to this family (platform implementations override this; // should be made pure virtual once all subclasses have been updated) virtual void FindStyleVariations() { } // search for a specific face using the Postscript name gfxFontEntry* FindFont(const nsAString& aPostscriptName); // read in cmaps for all the faces void ReadCMAP() { PRUint32 i, numFonts = mAvailableFonts.Length(); // called from RunLoader BEFORE CheckForSimpleFamily so that there cannot // be any NULL entries in mAvailableFonts for (i = 0; i < numFonts; i++) mAvailableFonts[i]->ReadCMAP(); } // mark this family as being in the "bad" underline offset blacklist void SetBadUnderlineFamily() { mIsBadUnderlineFamily = true; if (mHasStyles) { SetBadUnderlineFonts(); } } bool IsBadUnderlineFamily() const { return mIsBadUnderlineFamily; } // sort available fonts to put preferred (standard) faces towards the end void SortAvailableFonts(); // check whether the family fits into the simple 4-face model, // so we can use simplified style-matching; // if so set the mIsSimpleFamily flag (defaults to False before we've checked) void CheckForSimpleFamily(); protected: // fills in an array with weights of faces that match style, // returns whether any matching entries found virtual bool FindWeightsForStyle(gfxFontEntry* aFontsForWeights[], bool anItalic, PRInt16 aStretch); bool ReadOtherFamilyNamesForFace(gfxPlatformFontList *aPlatformFontList, FallibleTArray& aNameTable, bool useFullName = false); // set whether this font family is in "bad" underline offset blacklist. void SetBadUnderlineFonts() { PRUint32 i, numFonts = mAvailableFonts.Length(); for (i = 0; i < numFonts; i++) { if (mAvailableFonts[i]) { mAvailableFonts[i]->mIsBadUnderlineFont = true; } } } nsString mName; nsTArray > mAvailableFonts; bool mOtherFamilyNamesInitialized; bool mHasOtherFamilyNames; bool mFaceNamesInitialized; bool mHasStyles; bool mIsSimpleFamily; bool mIsBadUnderlineFamily; enum { // for "simple" families, the faces are stored in mAvailableFonts // with fixed positions: kRegularFaceIndex = 0, kBoldFaceIndex = 1, kItalicFaceIndex = 2, kBoldItalicFaceIndex = 3, // mask values for selecting face with bold and/or italic attributes kBoldMask = 0x01, kItalicMask = 0x02 }; }; struct gfxTextRange { enum { // flags for recording the kind of font-matching that was used kFontGroup = 0x0001, kPrefsFallback = 0x0002, kSystemFallback = 0x0004 }; gfxTextRange(PRUint32 aStart, PRUint32 aEnd, gfxFont* aFont, PRUint8 aMatchType) : start(aStart), end(aEnd), font(aFont), matchType(aMatchType) { } PRUint32 Length() const { return end - start; } PRUint32 start, end; nsRefPtr font; PRUint8 matchType; }; /** * Font cache design: * * The mFonts hashtable contains most fonts, indexed by (gfxFontEntry*, style). * It does not add a reference to the fonts it contains. * When a font's refcount decreases to zero, instead of deleting it we * add it to our expiration tracker. * The expiration tracker tracks fonts with zero refcount. After a certain * period of time, such fonts expire and are deleted. * * We're using 3 generations with a ten-second generation interval, so * zero-refcount fonts will be deleted 20-30 seconds after their refcount * goes to zero, if timer events fire in a timely manner. * * The font cache also handles timed expiration of cached ShapedWords * for "persistent" fonts: it has a repeating timer, and notifies * each cached font to "age" its shaped words. The words will be released * by the fonts if they get aged three times without being re-used in the * meantime. * * Note that the ShapedWord timeout is much larger than the font timeout, * so that in the case of a short-lived font, we'll discard the gfxFont * completely, with all its words, and avoid the cost of aging the words * individually. That only happens with longer-lived fonts. */ class THEBES_API gfxFontCache MOZ_FINAL : public nsExpirationTracker { public: enum { FONT_TIMEOUT_SECONDS = 10, SHAPED_WORD_TIMEOUT_SECONDS = 60 }; gfxFontCache(); ~gfxFontCache(); /* * Get the global gfxFontCache. You must call Init() before * calling this method --- the result will not be null. */ static gfxFontCache* GetCache() { return gGlobalCache; } static nsresult Init(); // It's OK to call this even if Init() has not been called. static void Shutdown(); // Look up a font in the cache. Returns an addrefed pointer, or null // if there's nothing matching in the cache already_AddRefed Lookup(const gfxFontEntry *aFontEntry, const gfxFontStyle *aStyle); // We created a new font (presumably because Lookup returned null); // put it in the cache. The font's refcount should be nonzero. It is // allowable to add a new font even if there is one already in the // cache with the same key; we'll forget about the old one. void AddNew(gfxFont *aFont); // The font's refcount has gone to zero; give ownership of it to // the cache. We delete it if it's not acquired again after a certain // amount of time. void NotifyReleased(gfxFont *aFont); // This gets called when the timeout has expired on a zero-refcount // font; we just delete it. virtual void NotifyExpired(gfxFont *aFont); // Cleans out the hashtable and removes expired fonts waiting for cleanup. // Other gfxFont objects may be still in use but they will be pushed // into the expiration queues and removed. void Flush() { mFonts.Clear(); AgeAllGenerations(); } void FlushShapedWordCaches() { mFonts.EnumerateEntries(ClearCachedWordsForFont, nsnull); } protected: void DestroyFont(gfxFont *aFont); static gfxFontCache *gGlobalCache; struct Key { const gfxFontEntry* mFontEntry; const gfxFontStyle* mStyle; Key(const gfxFontEntry* aFontEntry, const gfxFontStyle* aStyle) : mFontEntry(aFontEntry), mStyle(aStyle) {} }; class HashEntry : public PLDHashEntryHdr { public: typedef const Key& KeyType; typedef const Key* KeyTypePointer; // When constructing a new entry in the hashtable, we'll leave this // blank. The caller of Put() will fill this in. HashEntry(KeyTypePointer aStr) : mFont(nsnull) { } HashEntry(const HashEntry& toCopy) : mFont(toCopy.mFont) { } ~HashEntry() { } bool KeyEquals(const KeyTypePointer aKey) const; static KeyTypePointer KeyToPointer(KeyType aKey) { return &aKey; } static PLDHashNumber HashKey(const KeyTypePointer aKey) { return NS_PTR_TO_INT32(aKey->mFontEntry) ^ aKey->mStyle->Hash(); } enum { ALLOW_MEMMOVE = true }; gfxFont* mFont; }; nsTHashtable mFonts; static PLDHashOperator ClearCachedWordsForFont(HashEntry* aHashEntry, void*); static PLDHashOperator AgeCachedWordsForFont(HashEntry* aHashEntry, void*); static void WordCacheExpirationTimerCallback(nsITimer* aTimer, void* aCache); nsCOMPtr mWordCacheExpirationTimer; }; class THEBES_API gfxTextRunFactory { NS_INLINE_DECL_REFCOUNTING(gfxTextRunFactory) public: // Flags in the mask 0xFFFF0000 are reserved for textrun clients // Flags in the mask 0x0000F000 are reserved for per-platform fonts // Flags in the mask 0x00000FFF are set by the textrun creator. enum { CACHE_TEXT_FLAGS = 0xF0000000, USER_TEXT_FLAGS = 0x0FFF0000, PLATFORM_TEXT_FLAGS = 0x0000F000, TEXTRUN_TEXT_FLAGS = 0x00000FFF, SETTABLE_FLAGS = CACHE_TEXT_FLAGS | USER_TEXT_FLAGS, /** * When set, the text string pointer used to create the text run * is guaranteed to be available during the lifetime of the text run. */ TEXT_IS_PERSISTENT = 0x0001, /** * When set, the text is known to be all-ASCII (< 128). */ TEXT_IS_ASCII = 0x0002, /** * When set, the text is RTL. */ TEXT_IS_RTL = 0x0004, /** * When set, spacing is enabled and the textrun needs to call GetSpacing * on the spacing provider. */ TEXT_ENABLE_SPACING = 0x0008, /** * When set, GetHyphenationBreaks may return true for some character * positions, otherwise it will always return false for all characters. */ TEXT_ENABLE_HYPHEN_BREAKS = 0x0010, /** * When set, the text has no characters above 255 and it is stored * in the textrun in 8-bit format. */ TEXT_IS_8BIT = 0x0020, /** * When set, the RunMetrics::mBoundingBox field will be initialized * properly based on glyph extents, in particular, glyph extents that * overflow the standard font-box (the box defined by the ascent, descent * and advance width of the glyph). When not set, it may just be the * standard font-box even if glyphs overflow. */ TEXT_NEED_BOUNDING_BOX = 0x0040, /** * When set, optional ligatures are disabled. Ligatures that are * required for legible text should still be enabled. */ TEXT_DISABLE_OPTIONAL_LIGATURES = 0x0080, /** * When set, the textrun should favour speed of construction over * quality. This may involve disabling ligatures and/or kerning or * other effects. */ TEXT_OPTIMIZE_SPEED = 0x0100, /** * For internal use by the memory reporter when accounting for * storage used by textruns. * Because the reporter may visit each textrun multiple times while * walking the frame trees and textrun cache, it needs to mark * textruns that have been seen so as to avoid multiple-accounting. */ TEXT_RUN_SIZE_ACCOUNTED = 0x0200, /** * nsTextFrameThebes sets these, but they're defined here rather than * in nsTextFrameUtils.h because ShapedWord creation/caching also needs * to check the _INCOMING flag */ TEXT_TRAILING_ARABICCHAR = 0x20000000, /** * When set, the previous character for this textrun was an Arabic * character. This is used for the context detection necessary for * bidi.numeral implementation. */ TEXT_INCOMING_ARABICCHAR = 0x40000000, TEXT_UNUSED_FLAGS = 0x90000000 }; /** * This record contains all the parameters needed to initialize a textrun. */ struct Parameters { // A reference context suggesting where the textrun will be rendered gfxContext *mContext; // Pointer to arbitrary user data (which should outlive the textrun) void *mUserData; // A description of which characters have been stripped from the original // DOM string to produce the characters in the textrun. May be null // if that information is not relevant. gfxSkipChars *mSkipChars; // A list of where linebreaks are currently placed in the textrun. May // be null if mInitialBreakCount is zero. PRUint32 *mInitialBreaks; PRUint32 mInitialBreakCount; // The ratio to use to convert device pixels to application layout units PRUint32 mAppUnitsPerDevUnit; }; virtual ~gfxTextRunFactory() {} }; /** * This stores glyph bounds information for a particular gfxFont, at * a particular appunits-per-dev-pixel ratio (because the compressed glyph * width array is stored in appunits). * * We store a hashtable from glyph IDs to float bounding rects. For the * common case where the glyph has no horizontal left bearing, and no * y overflow above the font ascent or below the font descent, and tight * bounding boxes are not required, we avoid storing the glyph ID in the hashtable * and instead consult an array of 16-bit glyph XMost values (in appunits). * This array always has an entry for the font's space glyph --- the width is * assumed to be zero. */ class THEBES_API gfxGlyphExtents { public: gfxGlyphExtents(PRUint32 aAppUnitsPerDevUnit) : mAppUnitsPerDevUnit(aAppUnitsPerDevUnit) { MOZ_COUNT_CTOR(gfxGlyphExtents); mTightGlyphExtents.Init(); } ~gfxGlyphExtents(); enum { INVALID_WIDTH = 0xFFFF }; // returns INVALID_WIDTH => not a contained glyph // Otherwise the glyph has no before-bearing or vertical bearings, // and the result is its width measured from the baseline origin, in // appunits. PRUint16 GetContainedGlyphWidthAppUnits(PRUint32 aGlyphID) const { return mContainedGlyphWidths.Get(aGlyphID); } bool IsGlyphKnown(PRUint32 aGlyphID) const { return mContainedGlyphWidths.Get(aGlyphID) != INVALID_WIDTH || mTightGlyphExtents.GetEntry(aGlyphID) != nsnull; } bool IsGlyphKnownWithTightExtents(PRUint32 aGlyphID) const { return mTightGlyphExtents.GetEntry(aGlyphID) != nsnull; } // Get glyph extents; a rectangle relative to the left baseline origin // Returns true on success. Can fail on OOM or when aContext is null // and extents were not (successfully) prefetched. bool GetTightGlyphExtentsAppUnits(gfxFont *aFont, gfxContext *aContext, PRUint32 aGlyphID, gfxRect *aExtents); void SetContainedGlyphWidthAppUnits(PRUint32 aGlyphID, PRUint16 aWidth) { mContainedGlyphWidths.Set(aGlyphID, aWidth); } void SetTightGlyphExtents(PRUint32 aGlyphID, const gfxRect& aExtentsAppUnits); PRUint32 GetAppUnitsPerDevUnit() { return mAppUnitsPerDevUnit; } private: class HashEntry : public nsUint32HashKey { public: // When constructing a new entry in the hashtable, we'll leave this // blank. The caller of Put() will fill this in. HashEntry(KeyTypePointer aPtr) : nsUint32HashKey(aPtr) {} HashEntry(const HashEntry& toCopy) : nsUint32HashKey(toCopy) { x = toCopy.x; y = toCopy.y; width = toCopy.width; height = toCopy.height; } float x, y, width, height; }; typedef PRUptrdiff PtrBits; enum { BLOCK_SIZE_BITS = 7, BLOCK_SIZE = 1 << BLOCK_SIZE_BITS }; // 128-glyph blocks class GlyphWidths { public: void Set(PRUint32 aIndex, PRUint16 aValue); PRUint16 Get(PRUint32 aIndex) const { PRUint32 block = aIndex >> BLOCK_SIZE_BITS; if (block >= mBlocks.Length()) return INVALID_WIDTH; PtrBits bits = mBlocks[block]; if (!bits) return INVALID_WIDTH; PRUint32 indexInBlock = aIndex & (BLOCK_SIZE - 1); if (bits & 0x1) { if (GetGlyphOffset(bits) != indexInBlock) return INVALID_WIDTH; return GetWidth(bits); } PRUint16 *widths = reinterpret_cast(bits); return widths[indexInBlock]; } #ifdef DEBUG PRUint32 ComputeSize(); #endif ~GlyphWidths(); private: static PRUint32 GetGlyphOffset(PtrBits aBits) { NS_ASSERTION(aBits & 0x1, "This is really a pointer..."); return (aBits >> 1) & ((1 << BLOCK_SIZE_BITS) - 1); } static PRUint32 GetWidth(PtrBits aBits) { NS_ASSERTION(aBits & 0x1, "This is really a pointer..."); return aBits >> (1 + BLOCK_SIZE_BITS); } static PtrBits MakeSingle(PRUint32 aGlyphOffset, PRUint16 aWidth) { return (aWidth << (1 + BLOCK_SIZE_BITS)) + (aGlyphOffset << 1) + 1; } nsTArray mBlocks; }; GlyphWidths mContainedGlyphWidths; nsTHashtable mTightGlyphExtents; PRUint32 mAppUnitsPerDevUnit; }; /** * gfxFontShaper * * This class implements text shaping (character to glyph mapping and * glyph layout). There is a gfxFontShaper subclass for each text layout * technology (uniscribe, core text, harfbuzz,....) we support. * * The shaper is responsible for setting up glyph data in gfxTextRuns. * * A generic, platform-independent shaper relies only on the standard * gfxFont interface and can work with any concrete subclass of gfxFont. * * Platform-specific implementations designed to interface to platform * shaping APIs such as Uniscribe or CoreText may rely on features of a * specific font subclass to access native font references * (such as CTFont, HFONT, DWriteFont, etc). */ class gfxFontShaper { public: gfxFontShaper(gfxFont *aFont) : mFont(aFont) { NS_ASSERTION(aFont, "shaper requires a valid font!"); } virtual ~gfxFontShaper() { } virtual bool ShapeWord(gfxContext *aContext, gfxShapedWord *aShapedWord, const PRUnichar *aText) = 0; gfxFont *GetFont() const { return mFont; } protected: // the font this shaper is working with gfxFont * mFont; }; /* a SPECIFIC single font family */ class THEBES_API gfxFont { public: nsrefcnt AddRef(void) { NS_PRECONDITION(PRInt32(mRefCnt) >= 0, "illegal refcnt"); if (mExpirationState.IsTracked()) { gfxFontCache::GetCache()->RemoveObject(this); } ++mRefCnt; NS_LOG_ADDREF(this, mRefCnt, "gfxFont", sizeof(*this)); return mRefCnt; } nsrefcnt Release(void) { NS_PRECONDITION(0 != mRefCnt, "dup release"); --mRefCnt; NS_LOG_RELEASE(this, mRefCnt, "gfxFont"); if (mRefCnt == 0) { NotifyReleased(); // |this| may have been deleted. return 0; } return mRefCnt; } PRInt32 GetRefCount() { return mRefCnt; } // options to specify the kind of AA to be used when creating a font typedef enum { kAntialiasDefault, kAntialiasNone, kAntialiasGrayscale, kAntialiasSubpixel } AntialiasOption; // Options for how the text should be drawn typedef enum { // GLYPH_FILL and GLYPH_STROKE draw into the current context // and may be used together with bitwise OR. GLYPH_FILL = 1, // Note: using GLYPH_STROKE will destroy the current path. GLYPH_STROKE = 2, // Appends glyphs to the current path. Can NOT be used with // GLYPH_FILL or GLYPH_STROKE. GLYPH_PATH = 4 } DrawMode; protected: nsAutoRefCnt mRefCnt; cairo_scaled_font_t *mScaledFont; void NotifyReleased() { gfxFontCache *cache = gfxFontCache::GetCache(); if (cache) { // Don't delete just yet; return the object to the cache for // possibly recycling within some time limit cache->NotifyReleased(this); } else { // The cache may have already been shut down. delete this; } } gfxFont(gfxFontEntry *aFontEntry, const gfxFontStyle *aFontStyle, AntialiasOption anAAOption = kAntialiasDefault, cairo_scaled_font_t *aScaledFont = nsnull); public: virtual ~gfxFont(); bool Valid() const { return mIsValid; } // options for the kind of bounding box to return from measurement typedef enum { LOOSE_INK_EXTENTS, // A box that encloses all the painted pixels, and may // include sidebearings and/or additional ascent/descent // within the glyph cell even if the ink is smaller. TIGHT_INK_EXTENTS, // A box that tightly encloses all the painted pixels // (although actually on Windows, at least, it may be // slightly larger than strictly necessary because // we can't get precise extents with ClearType). TIGHT_HINTED_OUTLINE_EXTENTS // A box that tightly encloses the glyph outline, // ignoring possible antialiasing pixels that extend // beyond this. // NOTE: The default implementation of gfxFont::Measure(), // which works with the glyph extents cache, does not // differentiate between this and TIGHT_INK_EXTENTS. // Whether the distinction is important depends on the // antialiasing behavior of the platform; currently the // distinction is only implemented in the gfxWindowsFont // subclass, because of ClearType's tendency to paint // outside the hinted outline. // Also NOTE: it is relatively expensive to request this, // as it does not use cached glyph extents in the font. } BoundingBoxType; const nsString& GetName() const { return mFontEntry->Name(); } const gfxFontStyle *GetStyle() const { return &mStyle; } cairo_scaled_font_t* GetCairoScaledFont() { return mScaledFont; } virtual gfxFont* CopyWithAntialiasOption(AntialiasOption anAAOption) { // platforms where this actually matters should override return nsnull; } virtual gfxFloat GetAdjustedSize() { return mAdjustedSize > 0.0 ? mAdjustedSize : mStyle.size; } float FUnitsToDevUnitsFactor() const { // check this was set up during font initialization NS_ASSERTION(mFUnitsConvFactor > 0.0f, "mFUnitsConvFactor not valid"); return mFUnitsConvFactor; } // check whether this is an sfnt we can potentially use with harfbuzz bool FontCanSupportHarfBuzz() { return mFontEntry->HasCmapTable(); } #ifdef MOZ_GRAPHITE // check whether this is an sfnt we can potentially use with Graphite bool FontCanSupportGraphite() { return mFontEntry->HasGraphiteTables(); } #endif // Access to raw font table data (needed for Harfbuzz): // returns a pointer to data owned by the fontEntry or the OS, // which will remain valid until released. // // Default implementations forward to the font entry, // and maintain a shared table. // // Subclasses should override this if they can provide more efficient // access than getting tables with mFontEntry->GetFontTable() and sharing // them via the entry. // // Get pointer to a specific font table, or NULL if // the table doesn't exist in the font virtual hb_blob_t *GetFontTable(PRUint32 aTag); // Subclasses may choose to look up glyph ids for characters. // If they do not override this, gfxHarfBuzzShaper will fetch the cmap // table and use that. virtual bool ProvidesGetGlyph() const { return false; } // Map unicode character to glyph ID. // Only used if ProvidesGetGlyph() returns true. virtual PRUint32 GetGlyph(PRUint32 unicode, PRUint32 variation_selector) { return 0; } // subclasses may provide (possibly hinted) glyph widths (in font units); // if they do not override this, harfbuzz will use unhinted widths // derived from the font tables virtual bool ProvidesGlyphWidths() { return false; } // The return value is interpreted as a horizontal advance in 16.16 fixed // point format. virtual PRInt32 GetGlyphWidth(gfxContext *aCtx, PRUint16 aGID) { return -1; } gfxFloat SynthesizeSpaceWidth(PRUint32 aCh); // Font metrics struct Metrics { gfxFloat xHeight; gfxFloat superscriptOffset; gfxFloat subscriptOffset; gfxFloat strikeoutSize; gfxFloat strikeoutOffset; gfxFloat underlineSize; gfxFloat underlineOffset; gfxFloat internalLeading; gfxFloat externalLeading; gfxFloat emHeight; gfxFloat emAscent; gfxFloat emDescent; gfxFloat maxHeight; gfxFloat maxAscent; gfxFloat maxDescent; gfxFloat maxAdvance; gfxFloat aveCharWidth; gfxFloat spaceWidth; gfxFloat zeroOrAveCharWidth; // width of '0', or if there is // no '0' glyph in this font, // equal to .aveCharWidth }; virtual const gfxFont::Metrics& GetMetrics() = 0; /** * We let layout specify spacing on either side of any * character. We need to specify both before and after * spacing so that substring measurement can do the right things. * These values are in appunits. They're always an integral number of * appunits, but we specify them in floats in case very large spacing * values are required. */ struct Spacing { gfxFloat mBefore; gfxFloat mAfter; }; /** * Metrics for a particular string */ struct THEBES_API RunMetrics { RunMetrics() { mAdvanceWidth = mAscent = mDescent = 0.0; mBoundingBox = gfxRect(0,0,0,0); } void CombineWith(const RunMetrics& aOther, bool aOtherIsOnLeft); // can be negative (partly due to negative spacing). // Advance widths should be additive: the advance width of the // (offset1, length1) plus the advance width of (offset1 + length1, // length2) should be the advance width of (offset1, length1 + length2) gfxFloat mAdvanceWidth; // For zero-width substrings, these must be zero! gfxFloat mAscent; // always non-negative gfxFloat mDescent; // always non-negative // Bounding box that is guaranteed to include everything drawn. // If a tight boundingBox was requested when these metrics were // generated, this will tightly wrap the glyphs, otherwise it is // "loose" and may be larger than the true bounding box. // Coordinates are relative to the baseline left origin, so typically // mBoundingBox.y == -mAscent gfxRect mBoundingBox; }; /** * Draw a series of glyphs to aContext. The direction of aTextRun must * be honoured. * @param aStart the first character to draw * @param aEnd draw characters up to here * @param aBaselineOrigin the baseline origin; the left end of the baseline * for LTR textruns, the right end of the baseline for RTL textruns. On return, * this should be updated to the other end of the baseline. In application * units, really! * @param aSpacing spacing to insert before and after characters (for RTL * glyphs, before-spacing is inserted to the right of characters). There * are aEnd - aStart elements in this array, unless it's null to indicate * that there is no spacing. * @param aDrawMode specifies whether the fill or stroke of the glyph should be * drawn, or if it should be drawn into the current path * * Callers guarantee: * -- aStart and aEnd are aligned to cluster and ligature boundaries * -- all glyphs use this font * * The default implementation builds a cairo glyph array and * calls cairo_show_glyphs or cairo_glyph_path. */ virtual void Draw(gfxTextRun *aTextRun, PRUint32 aStart, PRUint32 aEnd, gfxContext *aContext, DrawMode aDrawMode, gfxPoint *aBaselineOrigin, Spacing *aSpacing); /** * Measure a run of characters. See gfxTextRun::Metrics. * @param aTight if false, then return the union of the glyph extents * with the font-box for the characters (the rectangle with x=0,width= * the advance width for the character run,y=-(font ascent), and height= * font ascent + font descent). Otherwise, we must return as tight as possible * an approximation to the area actually painted by glyphs. * @param aContextForTightBoundingBox when aTight is true, this must * be non-null. * @param aSpacing spacing to insert before and after glyphs. The bounding box * need not include the spacing itself, but the spacing affects the glyph * positions. null if there is no spacing. * * Callers guarantee: * -- aStart and aEnd are aligned to cluster and ligature boundaries * -- all glyphs use this font * * The default implementation just uses font metrics and aTextRun's * advances, and assumes no characters fall outside the font box. In * general this is insufficient, because that assumption is not always true. */ virtual RunMetrics Measure(gfxTextRun *aTextRun, PRUint32 aStart, PRUint32 aEnd, BoundingBoxType aBoundingBoxType, gfxContext *aContextForTightBoundingBox, Spacing *aSpacing); /** * Line breaks have been changed at the beginning and/or end of a substring * of the text. Reshaping may be required; glyph updating is permitted. * @return true if anything was changed, false otherwise */ bool NotifyLineBreaksChanged(gfxTextRun *aTextRun, PRUint32 aStart, PRUint32 aLength) { return false; } // Expiration tracking nsExpirationState *GetExpirationState() { return &mExpirationState; } // Get the glyphID of a space virtual PRUint32 GetSpaceGlyph() = 0; gfxGlyphExtents *GetOrCreateGlyphExtents(PRUint32 aAppUnitsPerDevUnit); // You need to call SetupCairoFont on the aCR just before calling this virtual void SetupGlyphExtents(gfxContext *aContext, PRUint32 aGlyphID, bool aNeedTight, gfxGlyphExtents *aExtents); // This is called by the default Draw() implementation above. virtual bool SetupCairoFont(gfxContext *aContext) = 0; bool IsSyntheticBold() { return mApplySyntheticBold; } // Amount by which synthetic bold "fattens" the glyphs: 1/16 of the em-size gfxFloat GetSyntheticBoldOffset() { return GetAdjustedSize() * (1.0 / 16.0); } gfxFontEntry *GetFontEntry() { return mFontEntry.get(); } bool HasCharacter(PRUint32 ch) { if (!mIsValid) return false; return mFontEntry->HasCharacter(ch); } PRUint16 GetUVSGlyph(PRUint32 aCh, PRUint32 aVS) { if (!mIsValid) { return 0; } return mFontEntry->GetUVSGlyph(aCh, aVS); } // call the (virtual) InitTextRun method to do glyph generation/shaping, // limiting the length of text passed by processing the run in multiple // segments if necessary template bool SplitAndInitTextRun(gfxContext *aContext, gfxTextRun *aTextRun, const T *aString, PRUint32 aRunStart, PRUint32 aRunLength, PRInt32 aRunScript); // Get a ShapedWord representing the given text (either 8- or 16-bit) // for use in setting up a gfxTextRun. template gfxShapedWord* GetShapedWord(gfxContext *aContext, const T *aText, PRUint32 aLength, PRUint32 aHash, PRInt32 aRunScript, PRInt32 aAppUnitsPerDevUnit, PRUint32 aFlags); // Ensure the ShapedWord cache is initialized. This MUST be called before // any attempt to use GetShapedWord(). void InitWordCache() { if (!mWordCache.IsInitialized()) { mWordCache.Init(); } } // Called by the gfxFontCache timer to increment the age of all the words, // so that they'll expire after a sufficient period of non-use void AgeCachedWords() { if (mWordCache.IsInitialized()) { (void)mWordCache.EnumerateEntries(AgeCacheEntry, this); } } // Discard all cached word records; called on memory-pressure notification. void ClearCachedWords() { if (mWordCache.IsInitialized()) { mWordCache.Clear(); } } protected: // Call the appropriate shaper to generate glyphs for aText and store // them into aShapedWord. // The length of the text is aShapedWord->Length(). virtual bool ShapeWord(gfxContext *aContext, gfxShapedWord *aShapedWord, const PRUnichar *aText, bool aPreferPlatformShaping = false); nsRefPtr mFontEntry; struct CacheHashKey { union { const PRUint8 *mSingle; const PRUnichar *mDouble; } mText; PRUint32 mLength; PRUint32 mFlags; PRInt32 mScript; PRInt32 mAppUnitsPerDevUnit; PLDHashNumber mHashKey; bool mTextIs8Bit; CacheHashKey(const PRUint8 *aText, PRUint32 aLength, PRUint32 aStringHash, PRInt32 aScriptCode, PRInt32 aAppUnitsPerDevUnit, PRUint32 aFlags) : mLength(aLength), mFlags(aFlags), mScript(aScriptCode), mAppUnitsPerDevUnit(aAppUnitsPerDevUnit), mHashKey(aStringHash + aScriptCode + aAppUnitsPerDevUnit * 0x100 + aFlags * 0x10000), mTextIs8Bit(true) { NS_ASSERTION(aFlags & gfxTextRunFactory::TEXT_IS_8BIT, "8-bit flag should have been set"); mText.mSingle = aText; } CacheHashKey(const PRUnichar *aText, PRUint32 aLength, PRUint32 aStringHash, PRInt32 aScriptCode, PRInt32 aAppUnitsPerDevUnit, PRUint32 aFlags) : mLength(aLength), mFlags(aFlags), mScript(aScriptCode), mAppUnitsPerDevUnit(aAppUnitsPerDevUnit), mHashKey(aStringHash + aScriptCode + aAppUnitsPerDevUnit * 0x100 + aFlags * 0x10000), mTextIs8Bit(false) { // We can NOT assert that TEXT_IS_8BIT is false in aFlags here, // because this might be an 8bit-only word from a 16-bit textrun, // in which case the text we're passed is still in 16-bit form, // and we'll have to use an 8-to-16bit comparison in KeyEquals. mText.mDouble = aText; } }; class CacheHashEntry : public PLDHashEntryHdr { public: typedef const CacheHashKey &KeyType; typedef const CacheHashKey *KeyTypePointer; // When constructing a new entry in the hashtable, the caller of Put() // will fill us in. CacheHashEntry(KeyTypePointer aKey) { } CacheHashEntry(const CacheHashEntry& toCopy) { NS_ERROR("Should not be called"); } ~CacheHashEntry() { } bool KeyEquals(const KeyTypePointer aKey) const; static KeyTypePointer KeyToPointer(KeyType aKey) { return &aKey; } static PLDHashNumber HashKey(const KeyTypePointer aKey) { return aKey->mHashKey; } enum { ALLOW_MEMMOVE = true }; nsAutoPtr mShapedWord; }; nsTHashtable mWordCache; static PLDHashOperator AgeCacheEntry(CacheHashEntry *aEntry, void *aUserData); static const PRUint32 kShapedWordCacheMaxAge = 3; bool mIsValid; // use synthetic bolding for environments where this is not supported // by the platform bool mApplySyntheticBold; nsExpirationState mExpirationState; gfxFontStyle mStyle; nsAutoTArray mGlyphExtentsArray; gfxFloat mAdjustedSize; float mFUnitsConvFactor; // conversion factor from font units to dev units // the AA setting requested for this font - may affect glyph bounds AntialiasOption mAntialiasOption; // a copy of the font without antialiasing, if needed for separate // measurement by mathml code nsAutoPtr mNonAAFont; // we may switch between these shapers on the fly, based on the script // of the text run being shaped nsAutoPtr mPlatformShaper; nsAutoPtr mHarfBuzzShaper; #ifdef MOZ_GRAPHITE nsAutoPtr mGraphiteShaper; #endif // Create a default platform text shaper for this font. // (TODO: This should become pure virtual once all font backends have // been updated.) virtual void CreatePlatformShaper() { } // Helper for subclasses that want to initialize standard metrics from the // tables of sfnt (TrueType/OpenType) fonts. // This will use mFUnitsConvFactor if it is already set, else compute it // from mAdjustedSize and the unitsPerEm in the font's 'head' table. // Returns TRUE and sets mIsValid=TRUE if successful; // Returns TRUE but leaves mIsValid=FALSE if the font seems to be broken. // Returns FALSE if the font does not appear to be an sfnt at all, // and should be handled (if possible) using other APIs. bool InitMetricsFromSfntTables(Metrics& aMetrics); // Helper to calculate various derived metrics from the results of // InitMetricsFromSfntTables or equivalent platform code void CalculateDerivedMetrics(Metrics& aMetrics); // some fonts have bad metrics, this method sanitize them. // if this font has bad underline offset, aIsBadUnderlineFont should be true. void SanitizeMetrics(gfxFont::Metrics *aMetrics, bool aIsBadUnderlineFont); // Bug 674909. When synthetic bolding text by drawing twice, need to // render using a pixel offset in device pixels, otherwise text // doesn't appear bolded, it appears as if a bad text shadow exists // when a non-identity transform exists. Use an offset factor so that // the second draw occurs at a constant offset in device pixels. // This helper calculates the scale factor we need to apply to the // synthetic-bold offset. static double CalcXScale(gfxContext *aContext); }; // proportion of ascent used for x-height, if unable to read value from font #define DEFAULT_XHEIGHT_FACTOR 0.56f /* * gfxShapedWord stores a list of zero or more glyphs for each character. For each * glyph we store the glyph ID, the advance, and possibly an xoffset and yoffset. * The idea is that a string is rendered by a loop that draws each glyph * at its designated offset from the current point, then advances the current * point by the glyph's advance in the direction of the textrun (LTR or RTL). * Each glyph advance is always rounded to the nearest appunit; this ensures * consistent results when dividing the text in a textrun into multiple text * frames (frame boundaries are always aligned to appunits). We optimize * for the case where a character has a single glyph and zero xoffset and yoffset, * and the glyph ID and advance are in a reasonable range so we can pack all * necessary data into 32 bits. * * This glyph data is copied into gfxTextRuns as needed from the cache of * ShapedWords associated with each gfxFont instance. * * gfxTextRun methods that measure or draw substrings will associate all the * glyphs in a cluster with the first character of the cluster; if that character * is in the substring, the glyphs will be measured or drawn, otherwise they * won't. */ class gfxShapedWord { public: static const PRUint32 kMaxLength = 0x7fff; // Create a ShapedWord that can hold glyphs for aLength characters, // with mCharacterGlyphs sized appropriately. // // Returns null on allocation failure (does NOT use infallible alloc) // so caller must check for success. // // This does NOT perform shaping, so the returned word contains no // glyph data; the caller must call gfxFont::Shape() with appropriate // parameters to set up the glyphs. static gfxShapedWord* Create(const PRUint8 *aText, PRUint32 aLength, PRInt32 aRunScript, PRInt32 aAppUnitsPerDevUnit, PRUint32 aFlags) { NS_ASSERTION(aLength <= kMaxLength, "excessive length for gfxShapedWord!"); // Compute size needed including the mCharacterGlyphs array // and a copy of the original text PRUint32 size = offsetof(gfxShapedWord, mCharacterGlyphs) + aLength * (sizeof(CompressedGlyph) + sizeof(PRUint8)); void *storage = moz_malloc(size); if (!storage) { return nsnull; } // Construct in the pre-allocated storage, using placement new return new (storage) gfxShapedWord(aText, aLength, aRunScript, aAppUnitsPerDevUnit, aFlags); } static gfxShapedWord* Create(const PRUnichar *aText, PRUint32 aLength, PRInt32 aRunScript, PRInt32 aAppUnitsPerDevUnit, PRUint32 aFlags) { NS_ASSERTION(aLength <= kMaxLength, "excessive length for gfxShapedWord!"); // In the 16-bit version of Create, if the TEXT_IS_8BIT flag is set, // then we convert the text to an 8-bit version and call the 8-bit // Create function instead. if (aFlags & gfxTextRunFactory::TEXT_IS_8BIT) { nsCAutoString narrowText; LossyAppendUTF16toASCII(nsDependentSubstring(aText, aLength), narrowText); return Create((const PRUint8*)(narrowText.BeginReading()), aLength, aRunScript, aAppUnitsPerDevUnit, aFlags); } PRUint32 size = offsetof(gfxShapedWord, mCharacterGlyphs) + aLength * (sizeof(CompressedGlyph) + sizeof(PRUnichar)); void *storage = moz_malloc(size); if (!storage) { return nsnull; } return new (storage) gfxShapedWord(aText, aLength, aRunScript, aAppUnitsPerDevUnit, aFlags); } // Override operator delete to properly free the object that was // allocated via moz_malloc. void operator delete(void* p) { moz_free(p); } /** * This class records the information associated with a character in the * input string. It's optimized for the case where there is one glyph * representing that character alone. * * A character can have zero or more associated glyphs. Each glyph * has an advance width and an x and y offset. * A character may be the start of a cluster. * A character may be the start of a ligature group. * A character can be "missing", indicating that the system is unable * to render the character. * * All characters in a ligature group conceptually share all the glyphs * associated with the characters in a group. */ class CompressedGlyph { public: CompressedGlyph() { mValue = 0; } enum { // Indicates that a cluster and ligature group starts at this // character; this character has a single glyph with a reasonable // advance and zero offsets. A "reasonable" advance // is one that fits in the available bits (currently 12) (specified // in appunits). FLAG_IS_SIMPLE_GLYPH = 0x80000000U, // Indicates whether a linebreak is allowed before this character; // this is a two-bit field that holds a FLAG_BREAK_TYPE_xxx value // indicating the kind of linebreak (if any) allowed here. FLAGS_CAN_BREAK_BEFORE = 0x60000000U, FLAGS_CAN_BREAK_SHIFT = 29, FLAG_BREAK_TYPE_NONE = 0, FLAG_BREAK_TYPE_NORMAL = 1, FLAG_BREAK_TYPE_HYPHEN = 2, FLAG_CHAR_IS_SPACE = 0x10000000U, // The advance is stored in appunits ADVANCE_MASK = 0x0FFF0000U, ADVANCE_SHIFT = 16, GLYPH_MASK = 0x0000FFFFU, // Non-simple glyphs may or may not have glyph data in the // corresponding mDetailedGlyphs entry. They have the following // flag bits: // When NOT set, indicates that this character corresponds to a // missing glyph and should be skipped (or possibly, render the character // Unicode value in some special way). If there are glyphs, // the mGlyphID is actually the UTF16 character code. The bit is // inverted so we can memset the array to zero to indicate all missing. FLAG_NOT_MISSING = 0x01, FLAG_NOT_CLUSTER_START = 0x02, FLAG_NOT_LIGATURE_GROUP_START = 0x04, FLAG_CHAR_IS_TAB = 0x08, FLAG_CHAR_IS_NEWLINE = 0x10, FLAG_CHAR_IS_LOW_SURROGATE = 0x20, GLYPH_COUNT_MASK = 0x00FFFF00U, GLYPH_COUNT_SHIFT = 8 }; // "Simple glyphs" have a simple glyph ID, simple advance and their // x and y offsets are zero. Also the glyph extents do not overflow // the font-box defined by the font ascent, descent and glyph advance width. // These case is optimized to avoid storing DetailedGlyphs. // Returns true if the glyph ID aGlyph fits into the compressed representation static bool IsSimpleGlyphID(PRUint32 aGlyph) { return (aGlyph & GLYPH_MASK) == aGlyph; } // Returns true if the advance aAdvance fits into the compressed representation. // aAdvance is in appunits. static bool IsSimpleAdvance(PRUint32 aAdvance) { return (aAdvance & (ADVANCE_MASK >> ADVANCE_SHIFT)) == aAdvance; } bool IsSimpleGlyph() const { return (mValue & FLAG_IS_SIMPLE_GLYPH) != 0; } PRUint32 GetSimpleAdvance() const { return (mValue & ADVANCE_MASK) >> ADVANCE_SHIFT; } PRUint32 GetSimpleGlyph() const { return mValue & GLYPH_MASK; } bool IsMissing() const { return (mValue & (FLAG_NOT_MISSING|FLAG_IS_SIMPLE_GLYPH)) == 0; } bool IsClusterStart() const { return (mValue & FLAG_IS_SIMPLE_GLYPH) || !(mValue & FLAG_NOT_CLUSTER_START); } bool IsLigatureGroupStart() const { return (mValue & FLAG_IS_SIMPLE_GLYPH) || !(mValue & FLAG_NOT_LIGATURE_GROUP_START); } bool IsLigatureContinuation() const { return (mValue & FLAG_IS_SIMPLE_GLYPH) == 0 && (mValue & (FLAG_NOT_LIGATURE_GROUP_START | FLAG_NOT_MISSING)) == (FLAG_NOT_LIGATURE_GROUP_START | FLAG_NOT_MISSING); } // Return true if the original character was a normal (breakable, // trimmable) space (U+0020). Not true for other characters that // may happen to map to the space glyph (U+00A0). bool CharIsSpace() const { return (mValue & FLAG_CHAR_IS_SPACE) != 0; } bool CharIsTab() const { return !IsSimpleGlyph() && (mValue & FLAG_CHAR_IS_TAB) != 0; } bool CharIsNewline() const { return !IsSimpleGlyph() && (mValue & FLAG_CHAR_IS_NEWLINE) != 0; } bool CharIsLowSurrogate() const { return !IsSimpleGlyph() && (mValue & FLAG_CHAR_IS_LOW_SURROGATE) != 0; } void SetClusterStart(bool aIsClusterStart) { NS_ASSERTION(!IsSimpleGlyph(), "can't call SetClusterStart on simple glyphs"); if (aIsClusterStart) { mValue &= ~FLAG_NOT_CLUSTER_START; } else { mValue |= FLAG_NOT_CLUSTER_START; } } PRUint8 CanBreakBefore() const { return (mValue & FLAGS_CAN_BREAK_BEFORE) >> FLAGS_CAN_BREAK_SHIFT; } // Returns FLAGS_CAN_BREAK_BEFORE if the setting changed, 0 otherwise PRUint32 SetCanBreakBefore(PRUint8 aCanBreakBefore) { NS_ASSERTION(aCanBreakBefore <= 2, "Bogus break-before value!"); PRUint32 breakMask = (PRUint32(aCanBreakBefore) << FLAGS_CAN_BREAK_SHIFT); PRUint32 toggle = breakMask ^ (mValue & FLAGS_CAN_BREAK_BEFORE); mValue ^= toggle; return toggle; } CompressedGlyph& SetSimpleGlyph(PRUint32 aAdvanceAppUnits, PRUint32 aGlyph) { NS_ASSERTION(IsSimpleAdvance(aAdvanceAppUnits), "Advance overflow"); NS_ASSERTION(IsSimpleGlyphID(aGlyph), "Glyph overflow"); mValue = (mValue & FLAGS_CAN_BREAK_BEFORE) | FLAG_IS_SIMPLE_GLYPH | (aAdvanceAppUnits << ADVANCE_SHIFT) | aGlyph; return *this; } CompressedGlyph& SetComplex(bool aClusterStart, bool aLigatureStart, PRUint32 aGlyphCount) { mValue = (mValue & FLAGS_CAN_BREAK_BEFORE) | FLAG_NOT_MISSING | (aClusterStart ? 0 : FLAG_NOT_CLUSTER_START) | (aLigatureStart ? 0 : FLAG_NOT_LIGATURE_GROUP_START) | (aGlyphCount << GLYPH_COUNT_SHIFT); return *this; } /** * Missing glyphs are treated as ligature group starts; don't mess with * the cluster-start flag (see bugs 618870 and 619286). */ CompressedGlyph& SetMissing(PRUint32 aGlyphCount) { mValue = (mValue & (FLAGS_CAN_BREAK_BEFORE | FLAG_NOT_CLUSTER_START)) | (aGlyphCount << GLYPH_COUNT_SHIFT); return *this; } PRUint32 GetGlyphCount() const { NS_ASSERTION(!IsSimpleGlyph(), "Expected non-simple-glyph"); return (mValue & GLYPH_COUNT_MASK) >> GLYPH_COUNT_SHIFT; } void SetIsSpace() { mValue |= FLAG_CHAR_IS_SPACE; } void SetIsTab() { NS_ASSERTION(!IsSimpleGlyph(), "Expected non-simple-glyph"); mValue |= FLAG_CHAR_IS_TAB; } void SetIsNewline() { NS_ASSERTION(!IsSimpleGlyph(), "Expected non-simple-glyph"); mValue |= FLAG_CHAR_IS_NEWLINE; } void SetIsLowSurrogate() { NS_ASSERTION(!IsSimpleGlyph(), "Expected non-simple-glyph"); mValue |= FLAG_CHAR_IS_LOW_SURROGATE; } private: PRUint32 mValue; }; /** * When the glyphs for a character don't fit into a CompressedGlyph record * in SimpleGlyph format, we use an array of DetailedGlyphs instead. */ struct DetailedGlyph { /** The glyphID, or the Unicode character * if this is a missing glyph */ PRUint32 mGlyphID; /** The advance, x-offset and y-offset of the glyph, in appunits * mAdvance is in the text direction (RTL or LTR) * mXOffset is always from left to right * mYOffset is always from top to bottom */ PRInt32 mAdvance; float mXOffset, mYOffset; }; bool IsClusterStart(PRUint32 aPos) { NS_ASSERTION(aPos < Length(), "aPos out of range"); return mCharacterGlyphs[aPos].IsClusterStart(); } bool IsLigatureGroupStart(PRUint32 aPos) { NS_ASSERTION(aPos < Length(), "aPos out of range"); return mCharacterGlyphs[aPos].IsLigatureGroupStart(); } PRUint32 Length() const { return mLength; } const PRUint8* Text8Bit() const { NS_ASSERTION(TextIs8Bit(), "invalid use of Text8Bit()"); return reinterpret_cast(&mCharacterGlyphs[Length()]); } const PRUnichar* TextUnicode() const { NS_ASSERTION(!TextIs8Bit(), "invalid use of TextUnicode()"); return reinterpret_cast(&mCharacterGlyphs[Length()]); } PRUnichar GetCharAt(PRUint32 aOffset) const { NS_ASSERTION(aOffset < Length(), "aOffset out of range"); return TextIs8Bit() ? PRUnichar(Text8Bit()[aOffset]) : TextUnicode()[aOffset]; } PRUint32 Flags() const { return mFlags; } bool IsRightToLeft() const { return (Flags() & gfxTextRunFactory::TEXT_IS_RTL) != 0; } float GetDirection() const { return IsRightToLeft() ? -1.0 : 1.0; } bool DisableLigatures() const { return (Flags() & gfxTextRunFactory::TEXT_DISABLE_OPTIONAL_LIGATURES) != 0; } bool TextIs8Bit() const { return (Flags() & gfxTextRunFactory::TEXT_IS_8BIT) != 0; } PRInt32 Script() const { return mScript; } PRInt32 AppUnitsPerDevUnit() const { return mAppUnitsPerDevUnit; } void ResetAge() { mAgeCounter = 0; } PRUint32 IncrementAge() { return ++mAgeCounter; } void SetSimpleGlyph(PRUint32 aCharIndex, CompressedGlyph aGlyph) { NS_ASSERTION(aGlyph.IsSimpleGlyph(), "Should be a simple glyph here"); NS_ASSERTION(mCharacterGlyphs, "mCharacterGlyphs pointer is null!"); mCharacterGlyphs[aCharIndex] = aGlyph; } void SetGlyphs(PRUint32 aCharIndex, CompressedGlyph aGlyph, const DetailedGlyph *aGlyphs); void SetMissingGlyph(PRUint32 aIndex, PRUint32 aChar, gfxFont *aFont); void SetIsSpace(PRUint32 aIndex) { mCharacterGlyphs[aIndex].SetIsSpace(); } void SetIsLowSurrogate(PRUint32 aIndex) { SetGlyphs(aIndex, CompressedGlyph().SetComplex(false, false, 0), nsnull); mCharacterGlyphs[aIndex].SetIsLowSurrogate(); } bool FilterIfIgnorable(PRUint32 aIndex); const CompressedGlyph *GetCharacterGlyphs() const { return &mCharacterGlyphs[0]; } bool HasDetailedGlyphs() const { return mDetailedGlyphs != nsnull; } // NOTE that this must not be called for a character offset that does // not have any DetailedGlyph records; callers must have verified that // mCharacterGlyphs[aCharIndex].GetGlyphCount() is greater than zero. DetailedGlyph *GetDetailedGlyphs(PRUint32 aCharIndex) const { NS_ASSERTION(HasDetailedGlyphs() && !mCharacterGlyphs[aCharIndex].IsSimpleGlyph() && mCharacterGlyphs[aCharIndex].GetGlyphCount() > 0, "invalid use of GetDetailedGlyphs; check the caller!"); return mDetailedGlyphs->Get(aCharIndex); } void AdjustAdvancesForSyntheticBold(float aSynBoldOffset); // this is a public static method in order to make it available // for gfxTextRun to use directly on its own CompressedGlyph array, // in addition to the use within ShapedWord static void SetupClusterBoundaries(CompressedGlyph *aGlyphs, const PRUnichar *aString, PRUint32 aLength); private: // so that gfxTextRun can share our DetailedGlyphStore class friend class gfxTextRun; // Construct storage for a ShapedWord, ready to receive glyph data gfxShapedWord(const PRUint8 *aText, PRUint32 aLength, PRInt32 aRunScript, PRInt32 aAppUnitsPerDevUnit, PRUint32 aFlags) : mLength(aLength) , mFlags(aFlags | gfxTextRunFactory::TEXT_IS_8BIT) , mAppUnitsPerDevUnit(aAppUnitsPerDevUnit) , mScript(aRunScript) , mAgeCounter(0) { memset(mCharacterGlyphs, 0, aLength * sizeof(CompressedGlyph)); PRUint8 *text = reinterpret_cast(&mCharacterGlyphs[aLength]); memcpy(text, aText, aLength * sizeof(PRUint8)); } gfxShapedWord(const PRUnichar *aText, PRUint32 aLength, PRInt32 aRunScript, PRInt32 aAppUnitsPerDevUnit, PRUint32 aFlags) : mLength(aLength) , mFlags(aFlags) , mAppUnitsPerDevUnit(aAppUnitsPerDevUnit) , mScript(aRunScript) , mAgeCounter(0) { memset(mCharacterGlyphs, 0, aLength * sizeof(CompressedGlyph)); PRUnichar *text = reinterpret_cast(&mCharacterGlyphs[aLength]); memcpy(text, aText, aLength * sizeof(PRUnichar)); SetupClusterBoundaries(&mCharacterGlyphs[0], aText, aLength); } // Allocate aCount DetailedGlyphs for the given index DetailedGlyph *AllocateDetailedGlyphs(PRUint32 aCharIndex, PRUint32 aCount); // For characters whose glyph data does not fit the "simple" glyph criteria // in CompressedGlyph, we use a sorted array to store the association // between the source character offset and an index into an array // DetailedGlyphs. The CompressedGlyph record includes a count of // the number of DetailedGlyph records that belong to the character, // starting at the given index. class DetailedGlyphStore { public: DetailedGlyphStore() : mLastUsed(0) { } // This is optimized for the most common calling patterns: // we rarely need random access to the records, access is most commonly // sequential through the textRun, so we record the last-used index // and check whether the caller wants the same record again, or the // next; if not, it's most likely we're starting over from the start // of the run, so we check the first entry before resorting to binary // search as a last resort. // NOTE that this must not be called for a character offset that does // not have any DetailedGlyph records; callers must have verified that // mCharacterGlyphs[aOffset].GetGlyphCount() is greater than zero // before calling this, otherwise the assertions here will fire (in a // debug build), and we'll probably crash. DetailedGlyph* Get(PRUint32 aOffset) { NS_ASSERTION(mOffsetToIndex.Length() > 0, "no detailed glyph records!"); DetailedGlyph* details = mDetails.Elements(); // check common cases (fwd iteration, initial entry, etc) first if (mLastUsed < mOffsetToIndex.Length() - 1 && aOffset == mOffsetToIndex[mLastUsed + 1].mOffset) { ++mLastUsed; } else if (aOffset == mOffsetToIndex[0].mOffset) { mLastUsed = 0; } else if (aOffset == mOffsetToIndex[mLastUsed].mOffset) { // do nothing } else if (mLastUsed > 0 && aOffset == mOffsetToIndex[mLastUsed - 1].mOffset) { --mLastUsed; } else { mLastUsed = mOffsetToIndex.BinaryIndexOf(aOffset, CompareToOffset()); } NS_ASSERTION(mLastUsed != nsTArray::NoIndex, "detailed glyph record missing!"); return details + mOffsetToIndex[mLastUsed].mIndex; } DetailedGlyph* Allocate(PRUint32 aOffset, PRUint32 aCount) { PRUint32 detailIndex = mDetails.Length(); DetailedGlyph *details = mDetails.AppendElements(aCount); if (!details) { return nsnull; } // We normally set up glyph records sequentially, so the common case // here is to append new records to the mOffsetToIndex array; // test for that before falling back to the InsertElementSorted // method. if (mOffsetToIndex.Length() == 0 || aOffset > mOffsetToIndex[mOffsetToIndex.Length() - 1].mOffset) { if (!mOffsetToIndex.AppendElement(DGRec(aOffset, detailIndex))) { return nsnull; } } else { if (!mOffsetToIndex.InsertElementSorted(DGRec(aOffset, detailIndex), CompareRecordOffsets())) { return nsnull; } } return details; } size_t SizeOfIncludingThis(nsMallocSizeOfFun aMallocSizeOf) { return aMallocSizeOf(this, sizeof(DetailedGlyphStore)) + mDetails.SizeOfExcludingThis(aMallocSizeOf) + mOffsetToIndex.SizeOfExcludingThis(aMallocSizeOf); } private: struct DGRec { DGRec(const PRUint32& aOffset, const PRUint32& aIndex) : mOffset(aOffset), mIndex(aIndex) { } PRUint32 mOffset; // source character offset in the textrun PRUint32 mIndex; // index where this char's DetailedGlyphs begin }; struct CompareToOffset { bool Equals(const DGRec& a, const PRUint32& b) const { return a.mOffset == b; } bool LessThan(const DGRec& a, const PRUint32& b) const { return a.mOffset < b; } }; struct CompareRecordOffsets { bool Equals(const DGRec& a, const DGRec& b) const { return a.mOffset == b.mOffset; } bool LessThan(const DGRec& a, const DGRec& b) const { return a.mOffset < b.mOffset; } }; // Concatenated array of all the DetailedGlyph records needed for the // textRun; individual character offsets are associated with indexes // into this array via the mOffsetToIndex table. nsTArray mDetails; // For each character offset that needs DetailedGlyphs, we record the // index in mDetails where the list of glyphs begins. This array is // sorted by mOffset. nsTArray mOffsetToIndex; // Records the most recently used index into mOffsetToIndex, so that // we can support sequential access more quickly than just doing // a binary search each time. nsTArray::index_type mLastUsed; }; nsAutoPtr mDetailedGlyphs; // Number of PRUnichar characters and CompressedGlyph glyph records; // note that gfx font code will never attempt to create a ShapedWord // with a huge number of characters, so we could limit this to 16 bits // to minimize memory usage for large numbers of cached words. PRUint32 mLength; PRUint32 mFlags; PRInt32 mAppUnitsPerDevUnit; PRInt32 mScript; PRUint32 mAgeCounter; // The mCharacterGlyphs array is actually a variable-size member; // when the ShapedWord is created, its size will be increased as necessary // to allow the proper number of glyphs to be stored. // The original text, in either 8-bit or 16-bit form, will be stored // immediately following the CompressedGlyphs. CompressedGlyph mCharacterGlyphs[1]; }; /** * gfxTextRun is an abstraction for drawing and measuring substrings of a run * of text. It stores runs of positioned glyph data, each run having a single * gfxFont. The glyphs are associated with a string of source text, and the * gfxTextRun APIs take parameters that are offsets into that source text. * * gfxTextRuns are not refcounted. They should be deleted when no longer required. * * gfxTextRuns are mostly immutable. The only things that can change are * inter-cluster spacing and line break placement. Spacing is always obtained * lazily by methods that need it, it is not cached. Line breaks are stored * persistently (insofar as they affect the shaping of glyphs; gfxTextRun does * not actually do anything to explicitly account for line breaks). Initially * there are no line breaks. The textrun can record line breaks before or after * any given cluster. (Line breaks specified inside clusters are ignored.) * * It is important that zero-length substrings are handled correctly. This will * be on the test! */ class THEBES_API gfxTextRun { public: // we use the same glyph storage as gfxShapedWord, to facilitate copying // glyph data from shaped words into text runs as needed typedef gfxShapedWord::CompressedGlyph CompressedGlyph; typedef gfxShapedWord::DetailedGlyph DetailedGlyph; typedef gfxShapedWord::DetailedGlyphStore DetailedGlyphStore; // Override operator delete to properly free the object that was // allocated via moz_malloc. void operator delete(void* p) { moz_free(p); } virtual ~gfxTextRun(); typedef gfxFont::RunMetrics Metrics; // Public textrun API for general use bool IsClusterStart(PRUint32 aPos) { NS_ASSERTION(aPos < mCharacterCount, "aPos out of range"); return mCharacterGlyphs[aPos].IsClusterStart(); } bool IsLigatureGroupStart(PRUint32 aPos) { NS_ASSERTION(aPos < mCharacterCount, "aPos out of range"); return mCharacterGlyphs[aPos].IsLigatureGroupStart(); } bool CanBreakLineBefore(PRUint32 aPos) { NS_ASSERTION(aPos < mCharacterCount, "aPos out of range"); return mCharacterGlyphs[aPos].CanBreakBefore() == CompressedGlyph::FLAG_BREAK_TYPE_NORMAL; } bool CanHyphenateBefore(PRUint32 aPos) { NS_ASSERTION(aPos < mCharacterCount, "aPos out of range"); return mCharacterGlyphs[aPos].CanBreakBefore() == CompressedGlyph::FLAG_BREAK_TYPE_HYPHEN; } bool CharIsSpace(PRUint32 aPos) { NS_ASSERTION(0 <= aPos && aPos < mCharacterCount, "aPos out of range"); return mCharacterGlyphs[aPos].CharIsSpace(); } bool CharIsTab(PRUint32 aPos) { NS_ASSERTION(0 <= aPos && aPos < mCharacterCount, "aPos out of range"); return mCharacterGlyphs[aPos].CharIsTab(); } bool CharIsNewline(PRUint32 aPos) { NS_ASSERTION(0 <= aPos && aPos < mCharacterCount, "aPos out of range"); return mCharacterGlyphs[aPos].CharIsNewline(); } bool CharIsLowSurrogate(PRUint32 aPos) { NS_ASSERTION(0 <= aPos && aPos < mCharacterCount, "aPos out of range"); return mCharacterGlyphs[aPos].CharIsLowSurrogate(); } PRUint32 GetLength() { return mCharacterCount; } // All PRUint32 aStart, PRUint32 aLength ranges below are restricted to // grapheme cluster boundaries! All offsets are in terms of the string // passed into MakeTextRun. // All coordinates are in layout/app units /** * Set the potential linebreaks for a substring of the textrun. These are * the "allow break before" points. Initially, there are no potential * linebreaks. * * This can change glyphs and/or geometry! Some textruns' shapes * depend on potential line breaks (e.g., title-case-converting textruns). * This function is virtual so that those textruns can reshape themselves. * * @return true if this changed the linebreaks, false if the new line * breaks are the same as the old */ virtual bool SetPotentialLineBreaks(PRUint32 aStart, PRUint32 aLength, PRUint8 *aBreakBefore, gfxContext *aRefContext); /** * Layout provides PropertyProvider objects. These allow detection of * potential line break points and computation of spacing. We pass the data * this way to allow lazy data acquisition; for example BreakAndMeasureText * will want to only ask for properties of text it's actually looking at. * * NOTE that requested spacing may not actually be applied, if the textrun * is unable to apply it in some context. Exception: spacing around a * whitespace character MUST always be applied. */ class PropertyProvider { public: // Detect hyphenation break opportunities in the given range; breaks // not at cluster boundaries will be ignored. virtual void GetHyphenationBreaks(PRUint32 aStart, PRUint32 aLength, bool *aBreakBefore) = 0; // Returns the provider's hyphenation setting, so callers can decide // whether it is necessary to call GetHyphenationBreaks. // Result is an NS_STYLE_HYPHENS_* value. virtual PRInt8 GetHyphensOption() = 0; // Returns the extra width that will be consumed by a hyphen. This should // be constant for a given textrun. virtual gfxFloat GetHyphenWidth() = 0; typedef gfxFont::Spacing Spacing; /** * Get the spacing around the indicated characters. Spacing must be zero * inside clusters. In other words, if character i is not * CLUSTER_START, then character i-1 must have zero after-spacing and * character i must have zero before-spacing. */ virtual void GetSpacing(PRUint32 aStart, PRUint32 aLength, Spacing *aSpacing) = 0; }; class ClusterIterator { public: ClusterIterator(gfxTextRun *aTextRun); void Reset(); bool NextCluster(); PRUint32 Position() const { return mCurrentChar; } PRUint32 ClusterLength() const; gfxFloat ClusterAdvance(PropertyProvider *aProvider) const; private: gfxTextRun *mTextRun; PRUint32 mCurrentChar; }; /** * Draws a substring. Uses only GetSpacing from aBreakProvider. * The provided point is the baseline origin on the left of the string * for LTR, on the right of the string for RTL. * @param aAdvanceWidth if non-null, the advance width of the substring * is returned here. * * Drawing should respect advance widths in the sense that for LTR runs, * Draw(ctx, pt, offset1, length1, dirty, &provider, &advance) followed by * Draw(ctx, gfxPoint(pt.x + advance, pt.y), offset1 + length1, length2, * dirty, &provider, nsnull) should have the same effect as * Draw(ctx, pt, offset1, length1+length2, dirty, &provider, nsnull). * For RTL runs the rule is: * Draw(ctx, pt, offset1 + length1, length2, dirty, &provider, &advance) followed by * Draw(ctx, gfxPoint(pt.x + advance, pt.y), offset1, length1, * dirty, &provider, nsnull) should have the same effect as * Draw(ctx, pt, offset1, length1+length2, dirty, &provider, nsnull). * * Glyphs should be drawn in logical content order, which can be significant * if they overlap (perhaps due to negative spacing). */ void Draw(gfxContext *aContext, gfxPoint aPt, gfxFont::DrawMode aDrawMode, PRUint32 aStart, PRUint32 aLength, PropertyProvider *aProvider, gfxFloat *aAdvanceWidth); /** * Computes the ReflowMetrics for a substring. * Uses GetSpacing from aBreakProvider. * @param aBoundingBoxType which kind of bounding box (loose/tight) */ Metrics MeasureText(PRUint32 aStart, PRUint32 aLength, gfxFont::BoundingBoxType aBoundingBoxType, gfxContext *aRefContextForTightBoundingBox, PropertyProvider *aProvider); /** * Computes just the advance width for a substring. * Uses GetSpacing from aBreakProvider. */ gfxFloat GetAdvanceWidth(PRUint32 aStart, PRUint32 aLength, PropertyProvider *aProvider); /** * Clear all stored line breaks for the given range (both before and after), * and then set the line-break state before aStart to aBreakBefore and * after the last cluster to aBreakAfter. * * We require that before and after line breaks be consistent. For clusters * i and i+1, we require that if there is a break after cluster i, a break * will be specified before cluster i+1. This may be temporarily violated * (e.g. after reflowing line L and before reflowing line L+1); to handle * these temporary violations, we say that there is a break betwen i and i+1 * if a break is specified after i OR a break is specified before i+1. * * This can change textrun geometry! The existence of a linebreak can affect * the advance width of the cluster before the break (when kerning) or the * geometry of one cluster before the break or any number of clusters * after the break. (The one-cluster-before-the-break limit is somewhat * arbitrary; if some scripts require breaking it, then we need to * alter nsTextFrame::TrimTrailingWhitespace, perhaps drastically becase * it could affect the layout of frames before it...) * * We return true if glyphs or geometry changed, false otherwise. This * function is virtual so that gfxTextRun subclasses can reshape * properly. * * @param aAdvanceWidthDelta if non-null, returns the change in advance * width of the given range. */ virtual bool SetLineBreaks(PRUint32 aStart, PRUint32 aLength, bool aLineBreakBefore, bool aLineBreakAfter, gfxFloat *aAdvanceWidthDelta, gfxContext *aRefContext); /** * Finds the longest substring that will fit into the given width. * Uses GetHyphenationBreaks and GetSpacing from aBreakProvider. * Guarantees the following: * -- 0 <= result <= aMaxLength * -- result is the maximal value of N such that either * N < aMaxLength && line break at N && GetAdvanceWidth(aStart, N) <= aWidth * OR N < aMaxLength && hyphen break at N && GetAdvanceWidth(aStart, N) + GetHyphenWidth() <= aWidth * OR N == aMaxLength && GetAdvanceWidth(aStart, N) <= aWidth * where GetAdvanceWidth assumes the effect of * SetLineBreaks(aStart, N, aLineBreakBefore, N < aMaxLength, aProvider) * -- if no such N exists, then result is the smallest N such that * N < aMaxLength && line break at N * OR N < aMaxLength && hyphen break at N * OR N == aMaxLength * * The call has the effect of * SetLineBreaks(aStart, result, aLineBreakBefore, result < aMaxLength, aProvider) * and the returned metrics and the invariants above reflect this. * * @param aMaxLength this can be PR_UINT32_MAX, in which case the length used * is up to the end of the string * @param aLineBreakBefore set to true if and only if there is an actual * line break at the start of this string. * @param aSuppressInitialBreak if true, then we assume there is no possible * linebreak before aStart. If false, then we will check the internal * line break opportunity state before deciding whether to return 0 as the * character to break before. * @param aTrimWhitespace if non-null, then we allow a trailing run of * spaces to be trimmed; the width of the space(s) will not be included in * the measured string width for comparison with the limit aWidth, and * trimmed spaces will not be included in returned metrics. The width * of the trimmed spaces will be returned in aTrimWhitespace. * Trimmed spaces are still counted in the "characters fit" result. * @param aMetrics if non-null, we fill this in for the returned substring. * If a hyphenation break was used, the hyphen is NOT included in the returned metrics. * @param aBoundingBoxType whether to make the bounding box in aMetrics tight * @param aRefContextForTightBoundingBox a reference context to get the * tight bounding box, if requested * @param aUsedHyphenation if non-null, records if we selected a hyphenation break * @param aLastBreak if non-null and result is aMaxLength, we set this to * the maximal N such that * N < aMaxLength && line break at N && GetAdvanceWidth(aStart, N) <= aWidth * OR N < aMaxLength && hyphen break at N && GetAdvanceWidth(aStart, N) + GetHyphenWidth() <= aWidth * or PR_UINT32_MAX if no such N exists, where GetAdvanceWidth assumes * the effect of * SetLineBreaks(aStart, N, aLineBreakBefore, N < aMaxLength, aProvider) * * @param aCanWordWrap true if we can break between any two grapheme * clusters. This is set by word-wrap: break-word * * @param aBreakPriority in/out the priority of the break opportunity * saved in the line. If we are prioritizing break opportunities, we will * not set a break with a lower priority. @see gfxBreakPriority. * * Note that negative advance widths are possible especially if negative * spacing is provided. */ PRUint32 BreakAndMeasureText(PRUint32 aStart, PRUint32 aMaxLength, bool aLineBreakBefore, gfxFloat aWidth, PropertyProvider *aProvider, bool aSuppressInitialBreak, gfxFloat *aTrimWhitespace, Metrics *aMetrics, gfxFont::BoundingBoxType aBoundingBoxType, gfxContext *aRefContextForTightBoundingBox, bool *aUsedHyphenation, PRUint32 *aLastBreak, bool aCanWordWrap, gfxBreakPriority *aBreakPriority); /** * Update the reference context. * XXX this is a hack. New text frame does not call this. Use only * temporarily for old text frame. */ void SetContext(gfxContext *aContext) {} // Utility getters bool IsRightToLeft() const { return (mFlags & gfxTextRunFactory::TEXT_IS_RTL) != 0; } gfxFloat GetDirection() const { return (mFlags & gfxTextRunFactory::TEXT_IS_RTL) ? -1.0 : 1.0; } void *GetUserData() const { return mUserData; } void SetUserData(void *aUserData) { mUserData = aUserData; } PRUint32 GetFlags() const { return mFlags; } void SetFlagBits(PRUint32 aFlags) { NS_ASSERTION(!(aFlags & ~gfxTextRunFactory::SETTABLE_FLAGS), "Only user flags should be mutable"); mFlags |= aFlags; } void ClearFlagBits(PRUint32 aFlags) { NS_ASSERTION(!(aFlags & ~gfxTextRunFactory::SETTABLE_FLAGS), "Only user flags should be mutable"); mFlags &= ~aFlags; } const gfxSkipChars& GetSkipChars() const { return mSkipChars; } PRUint32 GetAppUnitsPerDevUnit() const { return mAppUnitsPerDevUnit; } gfxFontGroup *GetFontGroup() const { return mFontGroup; } // Call this, don't call "new gfxTextRun" directly. This does custom // allocation and initialization static gfxTextRun *Create(const gfxTextRunFactory::Parameters *aParams, const void *aText, PRUint32 aLength, gfxFontGroup *aFontGroup, PRUint32 aFlags); // The text is divided into GlyphRuns as necessary struct GlyphRun { nsRefPtr mFont; // never null PRUint32 mCharacterOffset; // into original UTF16 string PRUint8 mMatchType; }; class THEBES_API GlyphRunIterator { public: GlyphRunIterator(gfxTextRun *aTextRun, PRUint32 aStart, PRUint32 aLength) : mTextRun(aTextRun), mStartOffset(aStart), mEndOffset(aStart + aLength) { mNextIndex = mTextRun->FindFirstGlyphRunContaining(aStart); } bool NextRun(); GlyphRun *GetGlyphRun() { return mGlyphRun; } PRUint32 GetStringStart() { return mStringStart; } PRUint32 GetStringEnd() { return mStringEnd; } private: gfxTextRun *mTextRun; GlyphRun *mGlyphRun; PRUint32 mStringStart; PRUint32 mStringEnd; PRUint32 mNextIndex; PRUint32 mStartOffset; PRUint32 mEndOffset; }; class GlyphRunOffsetComparator { public: bool Equals(const GlyphRun& a, const GlyphRun& b) const { return a.mCharacterOffset == b.mCharacterOffset; } bool LessThan(const GlyphRun& a, const GlyphRun& b) const { return a.mCharacterOffset < b.mCharacterOffset; } }; friend class GlyphRunIterator; friend class FontSelector; // API for setting up the textrun glyphs. Should only be called by // things that construct textruns. /** * We've found a run of text that should use a particular font. Call this * only during initialization when font substitution has been computed. * Call it before setting up the glyphs for the characters in this run; * SetMissingGlyph requires that the correct glyphrun be installed. * * If aForceNewRun, a new glyph run will be added, even if the * previously added run uses the same font. If glyph runs are * added out of strictly increasing aStartCharIndex order (via * force), then SortGlyphRuns must be called after all glyph runs * are added before any further operations are performed with this * TextRun. */ nsresult AddGlyphRun(gfxFont *aFont, PRUint8 aMatchType, PRUint32 aStartCharIndex, bool aForceNewRun); void ResetGlyphRuns() { mGlyphRuns.Clear(); } void SortGlyphRuns(); void SanitizeGlyphRuns(); // Call the following glyph-setters during initialization or during reshaping // only. It is OK to overwrite existing data for a character. void SetSimpleGlyph(PRUint32 aCharIndex, CompressedGlyph aGlyph) { NS_ASSERTION(aGlyph.IsSimpleGlyph(), "Should be a simple glyph here"); mCharacterGlyphs[aCharIndex] = aGlyph; } /** * Set the glyph data for a character. aGlyphs may be null if aGlyph is a * simple glyph or has no associated glyphs. If non-null the data is copied, * the caller retains ownership. */ void SetGlyphs(PRUint32 aCharIndex, CompressedGlyph aGlyph, const DetailedGlyph *aGlyphs); void SetMissingGlyph(PRUint32 aCharIndex, PRUint32 aUnicodeChar); void SetSpaceGlyph(gfxFont *aFont, gfxContext *aContext, PRUint32 aCharIndex); // Set the glyph data for the given character index to the font's // space glyph, IF this can be done as a "simple" glyph record // (not requiring a DetailedGlyph entry). This avoids the need to call // the font shaper and go through the shaped-word cache for most spaces. // // The parameter aSpaceChar is the original character code for which // this space glyph is being used; if this is U+0020, we need to record // that it could be trimmed at a run edge, whereas other kinds of space // (currently just U+00A0) would not be trimmable/breakable. // // Returns true if it was able to set simple glyph data for the space; // if it returns false, the caller needs to fall back to some other // means to create the necessary (detailed) glyph data. bool SetSpaceGlyphIfSimple(gfxFont *aFont, gfxContext *aContext, PRUint32 aCharIndex, PRUnichar aSpaceChar); // Record the positions of specific characters that layout may need to // detect in the textrun, even though it doesn't have an explicit copy // of the original text. These are recorded using flag bits in the // CompressedGlyph record; if necessary, we convert "simple" glyph records // to "complex" ones as the Tab and Newline flags are not present in // simple CompressedGlyph records. void SetIsTab(PRUint32 aIndex) { CompressedGlyph *g = &mCharacterGlyphs[aIndex]; if (g->IsSimpleGlyph()) { DetailedGlyph *details = AllocateDetailedGlyphs(aIndex, 1); details->mGlyphID = g->GetSimpleGlyph(); details->mAdvance = g->GetSimpleAdvance(); details->mXOffset = details->mYOffset = 0; SetGlyphs(aIndex, CompressedGlyph().SetComplex(true, true, 1), details); } g->SetIsTab(); } void SetIsNewline(PRUint32 aIndex) { CompressedGlyph *g = &mCharacterGlyphs[aIndex]; if (g->IsSimpleGlyph()) { DetailedGlyph *details = AllocateDetailedGlyphs(aIndex, 1); details->mGlyphID = g->GetSimpleGlyph(); details->mAdvance = g->GetSimpleAdvance(); details->mXOffset = details->mYOffset = 0; SetGlyphs(aIndex, CompressedGlyph().SetComplex(true, true, 1), details); } g->SetIsNewline(); } void SetIsLowSurrogate(PRUint32 aIndex) { SetGlyphs(aIndex, CompressedGlyph().SetComplex(false, false, 0), nsnull); mCharacterGlyphs[aIndex].SetIsLowSurrogate(); } /** * Prefetch all the glyph extents needed to ensure that Measure calls * on this textrun not requesting tight boundingBoxes will succeed. Note * that some glyph extents might not be fetched due to OOM or other * errors. */ void FetchGlyphExtents(gfxContext *aRefContext); // API for access to the raw glyph data, needed by gfxFont::Draw // and gfxFont::GetBoundingBox CompressedGlyph *GetCharacterGlyphs() { return mCharacterGlyphs; } // NOTE that this must not be called for a character offset that does // not have any DetailedGlyph records; callers must have verified that // mCharacterGlyphs[aCharIndex].GetGlyphCount() is greater than zero. DetailedGlyph *GetDetailedGlyphs(PRUint32 aCharIndex) { NS_ASSERTION(mDetailedGlyphs != nsnull && !mCharacterGlyphs[aCharIndex].IsSimpleGlyph() && mCharacterGlyphs[aCharIndex].GetGlyphCount() > 0, "invalid use of GetDetailedGlyphs; check the caller!"); return mDetailedGlyphs->Get(aCharIndex); } bool HasDetailedGlyphs() { return mDetailedGlyphs != nsnull; } PRUint32 CountMissingGlyphs(); const GlyphRun *GetGlyphRuns(PRUint32 *aNumGlyphRuns) { *aNumGlyphRuns = mGlyphRuns.Length(); return mGlyphRuns.Elements(); } // Returns the index of the GlyphRun containing the given offset. // Returns mGlyphRuns.Length() when aOffset is mCharacterCount. PRUint32 FindFirstGlyphRunContaining(PRUint32 aOffset); // Copy glyph data from a ShapedWord into this textrun. void CopyGlyphDataFrom(const gfxShapedWord *aSource, PRUint32 aStart); // Copy glyph data for a range of characters from aSource to this // textrun. void CopyGlyphDataFrom(gfxTextRun *aSource, PRUint32 aStart, PRUint32 aLength, PRUint32 aDest); nsExpirationState *GetExpirationState() { return &mExpirationState; } struct LigatureData { // textrun offsets of the start and end of the containing ligature PRUint32 mLigatureStart; PRUint32 mLigatureEnd; // appunits advance to the start of the ligature part within the ligature; // never includes any spacing gfxFloat mPartAdvance; // appunits width of the ligature part; includes before-spacing // when the part is at the start of the ligature, and after-spacing // when the part is as the end of the ligature gfxFloat mPartWidth; bool mClipBeforePart; bool mClipAfterPart; }; // return storage used by this run, for memory reporter; // nsTransformedTextRun needs to override this as it holds additional data virtual NS_MUST_OVERRIDE size_t SizeOfExcludingThis(nsMallocSizeOfFun aMallocSizeOf); virtual NS_MUST_OVERRIDE size_t SizeOfIncludingThis(nsMallocSizeOfFun aMallocSizeOf); // Get the size, if it hasn't already been gotten, marking as it goes. size_t MaybeSizeOfIncludingThis(nsMallocSizeOfFun aMallocSizeOf) { if (mFlags & gfxTextRunFactory::TEXT_RUN_SIZE_ACCOUNTED) { return 0; } mFlags |= gfxTextRunFactory::TEXT_RUN_SIZE_ACCOUNTED; return SizeOfIncludingThis(aMallocSizeOf); } void ResetSizeOfAccountingFlags() { mFlags &= ~gfxTextRunFactory::TEXT_RUN_SIZE_ACCOUNTED; } #ifdef DEBUG void Dump(FILE* aOutput); #endif protected: /** * Create a textrun, and set its mCharacterGlyphs to point immediately * after the base object; this is ONLY used in conjunction with placement * new, after allocating a block large enough for the glyph records to * follow the base textrun object. */ gfxTextRun(const gfxTextRunFactory::Parameters *aParams, const void *aText, PRUint32 aLength, gfxFontGroup *aFontGroup, PRUint32 aFlags); /** * Helper for the Create() factory method to allocate the required * glyph storage for a textrun object with the basic size aSize, * plus room for aLength glyph records. */ static void* AllocateStorageForTextRun(size_t aSize, PRUint32 aLength); // All our glyph data is in logical order, not visual. // Space for mCharacterGlyphs is allocated fused with the textrun object, // and then the constructor sets the pointer to the beginning of this // storage area. Thus, this pointer must NOT be freed! CompressedGlyph *mCharacterGlyphs; private: // **** general helpers **** // Allocate aCount DetailedGlyphs for the given index DetailedGlyph *AllocateDetailedGlyphs(PRUint32 aCharIndex, PRUint32 aCount); // Get the total advance for a range of glyphs. PRInt32 GetAdvanceForGlyphs(PRUint32 aStart, PRUint32 aEnd); // Spacing for characters outside the range aSpacingStart/aSpacingEnd // is assumed to be zero; such characters are not passed to aProvider. // This is useful to protect aProvider from being passed character indices // it is not currently able to handle. bool GetAdjustedSpacingArray(PRUint32 aStart, PRUint32 aEnd, PropertyProvider *aProvider, PRUint32 aSpacingStart, PRUint32 aSpacingEnd, nsTArray *aSpacing); // **** ligature helpers **** // (Platforms do the actual ligaturization, but we need to do a bunch of stuff // to handle requests that begin or end inside a ligature) // if aProvider is null then mBeforeSpacing and mAfterSpacing are set to zero LigatureData ComputeLigatureData(PRUint32 aPartStart, PRUint32 aPartEnd, PropertyProvider *aProvider); gfxFloat ComputePartialLigatureWidth(PRUint32 aPartStart, PRUint32 aPartEnd, PropertyProvider *aProvider); void DrawPartialLigature(gfxFont *aFont, gfxContext *aCtx, PRUint32 aStart, PRUint32 aEnd, gfxPoint *aPt, PropertyProvider *aProvider); // Advance aStart to the start of the nearest ligature; back up aEnd // to the nearest ligature end; may result in *aStart == *aEnd void ShrinkToLigatureBoundaries(PRUint32 *aStart, PRUint32 *aEnd); // result in appunits gfxFloat GetPartialLigatureWidth(PRUint32 aStart, PRUint32 aEnd, PropertyProvider *aProvider); void AccumulatePartialLigatureMetrics(gfxFont *aFont, PRUint32 aStart, PRUint32 aEnd, gfxFont::BoundingBoxType aBoundingBoxType, gfxContext *aRefContext, PropertyProvider *aProvider, Metrics *aMetrics); // **** measurement helper **** void AccumulateMetricsForRun(gfxFont *aFont, PRUint32 aStart, PRUint32 aEnd, gfxFont::BoundingBoxType aBoundingBoxType, gfxContext *aRefContext, PropertyProvider *aProvider, PRUint32 aSpacingStart, PRUint32 aSpacingEnd, Metrics *aMetrics); // **** drawing helper **** void DrawGlyphs(gfxFont *aFont, gfxContext *aContext, gfxFont::DrawMode aDrawMode, gfxPoint *aPt, PRUint32 aStart, PRUint32 aEnd, PropertyProvider *aProvider, PRUint32 aSpacingStart, PRUint32 aSpacingEnd); nsAutoPtr mDetailedGlyphs; // XXX this should be changed to a GlyphRun plus a maybe-null GlyphRun*, // for smaller size especially in the super-common one-glyphrun case nsAutoTArray mGlyphRuns; void *mUserData; gfxFontGroup *mFontGroup; // addrefed gfxSkipChars mSkipChars; nsExpirationState mExpirationState; PRUint32 mAppUnitsPerDevUnit; PRUint32 mFlags; PRUint32 mCharacterCount; bool mSkipDrawing; // true if the font group we used had a user font // download that's in progress, so we should hide text // until the download completes (or timeout fires) }; class THEBES_API gfxFontGroup : public gfxTextRunFactory { public: static void Shutdown(); // platform must call this to release the languageAtomService gfxFontGroup(const nsAString& aFamilies, const gfxFontStyle *aStyle, gfxUserFontSet *aUserFontSet = nsnull); virtual ~gfxFontGroup(); virtual gfxFont *GetFontAt(PRInt32 i) { // If it turns out to be hard for all clients that cache font // groups to call UpdateFontList at appropriate times, we could // instead consider just calling UpdateFontList from someplace // more central (such as here). NS_ASSERTION(!mUserFontSet || mCurrGeneration == GetGeneration(), "Whoever was caching this font group should have " "called UpdateFontList on it"); NS_ASSERTION(mFonts.Length() > PRUint32(i), "Requesting a font index that doesn't exist"); return static_cast(mFonts[i]); } virtual PRUint32 FontListLength() const { return mFonts.Length(); } bool Equals(const gfxFontGroup& other) const { return mFamilies.Equals(other.mFamilies) && mStyle.Equals(other.mStyle); } const gfxFontStyle *GetStyle() const { return &mStyle; } virtual gfxFontGroup *Copy(const gfxFontStyle *aStyle); /** * The listed characters should be treated as invisible and zero-width * when creating textruns. */ static bool IsInvalidChar(PRUint8 ch); static bool IsInvalidChar(PRUnichar ch); /** * Make a textrun for a given string. * If aText is not persistent (aFlags & TEXT_IS_PERSISTENT), the * textrun will copy it. * This calls FetchGlyphExtents on the textrun. */ virtual gfxTextRun *MakeTextRun(const PRUnichar *aString, PRUint32 aLength, const Parameters *aParams, PRUint32 aFlags); /** * Make a textrun for a given string. * If aText is not persistent (aFlags & TEXT_IS_PERSISTENT), the * textrun will copy it. * This calls FetchGlyphExtents on the textrun. */ virtual gfxTextRun *MakeTextRun(const PRUint8 *aString, PRUint32 aLength, const Parameters *aParams, PRUint32 aFlags); /** * Textrun creation helper for clients that don't want to pass * a full Parameters record. */ template gfxTextRun *MakeTextRun(const T *aString, PRUint32 aLength, gfxContext *aRefContext, PRUint32 aAppUnitsPerDevUnit, PRUint32 aFlags) { gfxTextRunFactory::Parameters params = { aRefContext, nsnull, nsnull, nsnull, 0, aAppUnitsPerDevUnit }; return MakeTextRun(aString, aLength, ¶ms, aFlags); } /* helper function for splitting font families on commas and * calling a function for each family to fill the mFonts array */ typedef bool (*FontCreationCallback) (const nsAString& aName, const nsACString& aGenericName, bool aUseFontSet, void *closure); bool ForEachFont(const nsAString& aFamilies, nsIAtom *aLanguage, FontCreationCallback fc, void *closure); bool ForEachFont(FontCreationCallback fc, void *closure); /** * Check whether a given font (specified by its gfxFontEntry) * is already in the fontgroup's list of actual fonts */ bool HasFont(const gfxFontEntry *aFontEntry); const nsString& GetFamilies() { return mFamilies; } // This returns the preferred underline for this font group. // Some CJK fonts have wrong underline offset in its metrics. // If this group has such "bad" font, each platform's gfxFontGroup initialized mUnderlineOffset. // The value should be lower value of first font's metrics and the bad font's metrics. // Otherwise, this returns from first font's metrics. enum { UNDERLINE_OFFSET_NOT_SET = PR_INT16_MAX }; virtual gfxFloat GetUnderlineOffset() { if (mUnderlineOffset == UNDERLINE_OFFSET_NOT_SET) mUnderlineOffset = GetFontAt(0)->GetMetrics().underlineOffset; return mUnderlineOffset; } virtual already_AddRefed FindFontForChar(PRUint32 ch, PRUint32 prevCh, PRInt32 aRunScript, gfxFont *aPrevMatchedFont, PRUint8 *aMatchType); // search through pref fonts for a character, return nsnull if no matching pref font virtual already_AddRefed WhichPrefFontSupportsChar(PRUint32 aCh); virtual already_AddRefed WhichSystemFontSupportsChar(PRUint32 aCh); template void ComputeRanges(nsTArray& mRanges, const T *aString, PRUint32 aLength, PRInt32 aRunScript); gfxUserFontSet* GetUserFontSet(); // With downloadable fonts, the composition of the font group can change as fonts are downloaded // for each change in state of the user font set, the generation value is bumped to avoid picking up // previously created text runs in the text run word cache. For font groups based on stylesheets // with no @font-face rule, this always returns 0. PRUint64 GetGeneration(); // If there is a user font set, check to see whether the font list or any // caches need updating. virtual void UpdateFontList(); bool ShouldSkipDrawing() const { return mSkipDrawing; } protected: nsString mFamilies; gfxFontStyle mStyle; nsTArray< nsRefPtr > mFonts; gfxFloat mUnderlineOffset; gfxUserFontSet* mUserFontSet; PRUint64 mCurrGeneration; // track the current user font set generation, rebuild font list if needed // cache the most recent pref font to avoid general pref font lookup nsRefPtr mLastPrefFamily; nsRefPtr mLastPrefFont; eFontPrefLang mLastPrefLang; // lang group for last pref font eFontPrefLang mPageLang; bool mLastPrefFirstFont; // is this the first font in the list of pref fonts for this lang group? bool mSkipDrawing; // hide text while waiting for a font // download to complete (or fallback // timer to fire) /** * Textrun creation short-cuts for special cases where we don't need to * call a font shaper to generate glyphs. */ gfxTextRun *MakeEmptyTextRun(const Parameters *aParams, PRUint32 aFlags); gfxTextRun *MakeSpaceTextRun(const Parameters *aParams, PRUint32 aFlags); gfxTextRun *MakeBlankTextRun(const void* aText, PRUint32 aLength, const Parameters *aParams, PRUint32 aFlags); // Used for construction/destruction. Not intended to change the font set // as invalidation of font lists and caches is not considered. void SetUserFontSet(gfxUserFontSet *aUserFontSet); // Initialize the list of fonts void BuildFontList(); // Init this font group's font metrics. If there no bad fonts, you don't need to call this. // But if there are one or more bad fonts which have bad underline offset, // you should call this with the *first* bad font. void InitMetricsForBadFont(gfxFont* aBadFont); // Set up the textrun glyphs for an entire text run: // find script runs, and then call InitScriptRun for each template void InitTextRun(gfxContext *aContext, gfxTextRun *aTextRun, const T *aString, PRUint32 aLength); // InitTextRun helper to handle a single script run, by finding font ranges // and calling each font's InitTextRun() as appropriate template void InitScriptRun(gfxContext *aContext, gfxTextRun *aTextRun, const T *aString, PRUint32 aScriptRunStart, PRUint32 aScriptRunEnd, PRInt32 aRunScript); /* If aResolveGeneric is true, then CSS/Gecko generic family names are * replaced with preferred fonts. * * If aResolveFontName is true then fc() is called only for existing fonts * and with actual font names. If false then fc() is called with each * family name in aFamilies (after resolving CSS/Gecko generic family names * if aResolveGeneric). * If aUseFontSet is true, the fontgroup's user font set is checked; * if false then it is skipped. */ bool ForEachFontInternal(const nsAString& aFamilies, nsIAtom *aLanguage, bool aResolveGeneric, bool aResolveFontName, bool aUseFontSet, FontCreationCallback fc, void *closure); static bool FontResolverProc(const nsAString& aName, void *aClosure); static bool FindPlatformFont(const nsAString& aName, const nsACString& aGenericName, bool aUseFontSet, void *closure); static NS_HIDDEN_(nsILanguageAtomService*) gLangService; }; #endif