/*++

   Copyright (c) Microsoft Corporation. All rights reserved.

*/

#ifndef __usp10__
#define __usp10__
#if _MSC_VER > 1000
#pragma once
#endif

#include <windows.h>
#ifdef __cplusplus
extern "C" {
#endif


///// Uniscribe build number

#define USPBUILD 0400



/////   USP - Unicode Complex Script processor
//
//      Copyright (c) Microsoft Corporation. All rights reserved.




/////   SCRIPT
//
//      The SCRIPT enum is an opaque type used internally to identify
//      which shaping engine functions are used to process a given run.
//
//
#define SCRIPT_UNDEFINED  0
//
//p     SCRIPT_UNDEFINED: This is the only public script ordinal. May be
//      forced into the eScript field of a SCRIPT_ANALYSIS to disable shaping.
//      SCRIPT_UNDEFINED is supported by all fonts - ScriptShape will display
//      whatever glyph is defined in the font CMAP table, or, if none, the
//      missing glyph.







/////   USP Status Codes
//
#define USP_E_SCRIPT_NOT_IN_FONT   \
        MAKE_HRESULT(SEVERITY_ERROR,FACILITY_ITF,0x200)    // Script doesn't exist in font






/////   SCRIPT_CACHE
//
//      Many script APIs take a combination of HDC and SCRIPT_CACHE parameter.
//
//      A SCRIPT_CACHE is an opaque pointer to a Uniscribe font metric cache
//      structure.


typedef void *SCRIPT_CACHE;


//      The client must allocate and retain one SCRIPT_CACHE variable for each
//      character style used. It must be initialised by the client to NULL.
//
//      APIs are passed an HDC and the address of a SCRIPT_CACHE variable.
//      Uniscribe will first attempt to access font data via the SCRIPT_CACHE
//      and will only inspect the HDC if the required data is not already
//      cached.
//
//      The HDC may be passed as NULL. If data required by Uniscribe is
//      already cached, the HDC won't be accessed and operation continues
//      normally.
//
//      If the HDC is passed as NULL, and Uniscribe needs to access it for
//      any reason, Uniscribe will return E_PENDING.
//
//      E_PENDING is returned quickly, allowing the client to avoid time
//      consuming SelectObject calls. The following example applies to all
//      APIs that take a SCRIPT_CACHE and an optional HDC.
//
//c     hr = ScriptShape(NULL, &sc, ..);
//c     if (hr == E_PENDING) {
//c         ... select font into hdc ...
//c         hr = ScriptShape(hdc, &sc, ...);
//c     }






/////   ScriptFreeCache
//
//      The client may free a SCRIPT_CACHE at any time. Uniscribe maintains
//      reference counts in it's font and shaper caches, and frees font data
//      only when all sizes of the font are free, and shaper data only when
//      all fonts it supports are freed.
//
//      The client should free the SCRIPT_CACHE for a style when it discards
//      that style.
//
//      ScriptFreeCache always sets it's parameter to NULL to help avoid
//      mis-referencing.


HRESULT WINAPI ScriptFreeCache(
    SCRIPT_CACHE   *psc);       //InOut  Cache handle






/////   SCRIPT_CONTROL
//
//      The SCRIPT_CONTROL structure provides itemization control flags to the
//      ScriptItemize function.
//
//
typedef struct tag_SCRIPT_CONTROL {
    DWORD   uDefaultLanguage    :16; // For NADS, also default for context
    DWORD   fContextDigits      :1;  // Means use previous script instead of uDefaultLanguage

    // The following flags provide legacy support for GetCharacterPlacement features
    DWORD   fInvertPreBoundDir  :1;  // Reading order of virtual item immediately prior to string
    DWORD   fInvertPostBoundDir :1;  // Reading order of virtual item immediately following string
    DWORD   fLinkStringBefore   :1;  // Equivalent to presence of ZWJ before string
    DWORD   fLinkStringAfter    :1;  // Equivalent to presence of ZWJ after string
    DWORD   fNeutralOverride    :1;  // Causes all neutrals to be strong in the current embedding direction
    DWORD   fNumericOverride    :1;  // Causes all numerals to be strong in the current embedding direction
    DWORD   fLegacyBidiClass    :1;  // Causes plus and minus to be reated as neutrals, slash as a common separator
    DWORD   fReserved           :8;
} SCRIPT_CONTROL;
//
//
//p     uDefaultLanguage: Language to use when Unicode values are ambiguous.
//              Used by numeric processing to select digit shape when
//              fDigitSubstitute (see SCRIPT_STATE) is in force.
//
//p     fContextDigits: Specifies that national digits are chosen according to
//              the nearest previous strong text, rather than using
//              uDefaultLanguage.
//
//p     fInvertPreBoundDir: By default text at the start of the string is
//              laid out as if it follows strong text of the same direction
//              as the base embedding level. Set fInvertPreBoundDir to change
//              the initial context to the opposite of the base embedding
//              level. This flag is for GetCharacterPlacement legacy support.
//
//p     fInvertPostBoundDir: By default text at the end of the string is
//              laid out as if it preceeds strong text of the same direction
//              as the base embedding level. Set fInvertPostBoundDir to change
//              the final context to the opposite of the base embedding
//              level. This flag is for GetCharacterPlacement legacy support.
//
//p     fLinkStringBefore: Causes the first character of the string to be
//              shaped as if were joined to a previous character.
//
//p     fLinkStringAfter: Causes the last character of the string to be
//              shaped as if were joined to a following character.
//
//p     fNeutralOverride: Causes all neutral characters in the string to be
//              treated as if they were strong characters of their enclosing
//              embedding level. This effectively locks neutrals in place,
//              reordering occuring only between neutrals.
//
//p     fNumericOverride: Causes all numeric characters in the string to be
//              treated as if they were strong characters of their enclosing
//              embedding level. This effectively locks numerics in place,
//              reordering occuring only between numerics.
//
//p     fReserved: Reserved. Always initialise to 0.






/////   SCRIPT_STATE
//
//      The SCRIPT_STATE structure is used both to initialise the unicode
//      algorithm state as an input parameter to ScriptItemize, and is also
//      a component of each item analysis returned by ScriptItemize.
//
//
typedef struct tag_SCRIPT_STATE {
    WORD    uBidiLevel         :5;  // Unicode Bidi algorithm embedding level (0-16)
    WORD    fOverrideDirection :1;  // Set when in LRO/RLO embedding
    WORD    fInhibitSymSwap    :1;  // Set by U+206A (ISS), cleared by U+206B (ASS)
    WORD    fCharShape         :1;  // Set by U+206D (AAFS), cleared by U+206C (IAFS)
    WORD    fDigitSubstitute   :1;  // Set by U+206E (NADS), cleared by U+206F (NODS)
    WORD    fInhibitLigate     :1;  // Equiv !GCP_Ligate, no Unicode control chars yet
    WORD    fDisplayZWG        :1;  // Equiv GCP_DisplayZWG, no Unicode control characters yet
    WORD    fArabicNumContext  :1;  // For EN->AN Unicode rule
    WORD    fGcpClusters       :1;  // For Generating Backward Compatible GCP Clusters (legacy Apps)
    WORD    fReserved          :1;
    WORD    fEngineReserved    :2;  // For use by shaping engine
} SCRIPT_STATE;
//
//
//p     uBidiLevel: The embedding level associated with all characters in this
//              run according to the Unicode bidi algorithm. When passed to
//              ScriptItemize, should be initialised to 0 for an LTR base
//              embedding level, or 1 for RTL.
//
//p     fOverrideDirection: TRUE if this level is an override level (LRO/RLO).
//              In an override level, characters are layed out purely
//              left to right, or purely right to left. No reordering of digits
//              or strong characters of opposing direction takes place.
//              Note that this initial value is reset by LRE, RLE, LRO or
//              RLO codes in the string.
//
//p     fInhibitSymSwap: TRUE if the shaping engine is to bypass mirroring of
//              Unicode Mirrored glyphs such as brackets. Set by Unicode
//              character ISS, cleared by ASS.
//
//p     fCharShape: TRUE if character codes in the Arabic Presentation Forms
//              areas of Unicode should be shaped. (Not implemented).
//
//p     fDigitSubstitute: TRUE if character codes U+0030 through U+0039
//              (European digits) are to be substituted by national digits.
//              Set by Unicode NADS, Cleared by NODS.
//
//p     fInhibitLigate: TRUE if ligatures are not to be used in the shaping
//              of Arabic or Hebrew characters.
//
//p     fDisplayZWG: TRUE if control characters are to be shaped as
//              representational glyphs. (Normally, control characters are
//              shaped to the blank glyph and given a width of zero).
//
//p     fArabicNumContext: TRUE indicates prior strong characters were Arabic
//              for the purposes of rule P0 on page 3-19 of 'The Unicode
//              Standard, version 2.0'. Should normally be set TRUE before
//              itemizing an RTL paragraph in an Arabic language, FALSE
//              otherwise.
//
//p     fGcpClusters: For GetCharaterPlacement legacy support only.
//              Initialise to TRUE to request ScriptShape to generate
//              the LogClust array the same way as GetCharacterPlacement
//              does in Arabic and Hebrew Windows95. Affects only Arabic
//              and Hebrew items.
//
//p     fReserved: Reserved. Always initialise to 0.
//
//p     fEngineReserved: Reserved. Always initialise to 0.






/////   SCRIPT_ANALYSIS
//
//      Each analysed item is described by a SCRIPT_ANALYSIS structure.
//      It also includes a copy of the Unicode algorithm state (SCRIPT_STATE).
//
//
typedef struct tag_SCRIPT_ANALYSIS {
    WORD    eScript         :10;    // Shaping engine
    WORD    fRTL            :1;     // Rendering direction
    WORD    fLayoutRTL      :1;     // Set for GCP classes ARABIC/HEBREW and LOCALNUMBER
    WORD    fLinkBefore     :1;     // Implies there was a ZWJ before this item
    WORD    fLinkAfter      :1;     // Implies there is a ZWJ following this item.
    WORD    fLogicalOrder   :1;     // Set by client as input to ScriptShape/Place
    WORD    fNoGlyphIndex   :1;     // Generated by ScriptShape/Place - this item does not use glyph indices
    SCRIPT_STATE s;
} SCRIPT_ANALYSIS;
//
//
//p     eScript: Opaque value identifying which engine Uniscribe will use to
//              Shape, Place and TextOut this item. The value of eScript is
//              undefined, and will change in future releases, but attributes
//              of eScript may be obtained by calling ScriptGetProperties.
//
//p     fRTL: Rendering direction. Normally identical to the parity of the
//              Unicode embedding level, but may differ if overridden by
//              GetCharacterPlacement legacy support.
//
//p     fLayoutRTL: Logical direction - whether conceptually part of a
//              left-to-right sequenece or a right-to-left sequence. Although
//              this is usually the same as fRTL, for a number in a
//              right-to-left run, fRTL is False (because digits are always
//              displayed LTR), but fLayoutRTL is True (because the number is
//              read as part of the right-to-left sequence).
//
//p     fLinkBefore: If set, the shaping engine will shape the first character
//              of this item as if it were joining with a previous character.
//              Set by ScriptItemize, may be overriden before calling ScriptShape.
//
//p     fLinkAfter: If set, the shaping engine will shape the last character
//              of this item as if it were joining with a subsequient character.
//              Set by ScriptItemize, may be overriden before calling ScriptShape.
//
//p     fLogicalOrder: If set, the shaping engine will generate all glyph
//              related arrays in logical order. By default glyph related
//              arrays are in visual order, the first array entry corresponding
//              to the leftmost glyph.
//              Set to FALSE by ScriptItemize, may be overriden before calling
//              ScriptShape.
//
//p     fNoGlyphIndex: May be set TRUE on input to ScriptShape to disable use
//              of glyphs for this item. Additionally, ScriptShape will set it
//              TRUE for hdcs containing symbolic, unrecognised and device fonts.
//              Disabling glyphing disables complex script shaping. When set,
//              shaping and placing for this item is implemented directly by
//              calls to GetTextExtentExPoint and ExtTextOut.
/////   SCRIPT_ITEM
//
//      The SCRIPT_ITEM structure includes a SCRIPT_ANALYSIS with the string
//      ofset of the first character of the item.
//
//
typedef struct tag_SCRIPT_ITEM {
    int              iCharPos;      // Logical offset to first character in this item
    SCRIPT_ANALYSIS  a;
} SCRIPT_ITEM;
//
//
//p     iCharPos: Offset from beginning of itemised string to first character
//              of this item, counted in Unicode codepoints (i.e. words).
//
//p     a: Script analysis structure containing analysis specific to this
//              item, to be passed to ScriptShape, ScriptPlace etc.






/////   ScriptItemize - break text into items
//
//      Breaks a run of unicode into individually shapeable items.
//      Items are delimited by
//
//      o Change of shaping engine
//      o Change of direction
//
//      The client may create multiple runs from each item returned by
//      ScriptItemize, but should not combine multiple items into a single run.
//
//      Later the client will call ScriptShape for each run (when measuring or
//      rendering), and must pass the SCRIPT_ANALYSIS that ScriptItemize
//      returned.


HRESULT WINAPI ScriptItemize(
    const WCHAR           *pwcInChars,  // In   Unicode string to be itemized
    int                    cInChars,    // In   Codepoint count to itemize
    int                    cMaxItems,   // In   Max length of itemization array
    const SCRIPT_CONTROL  *psControl,   // In   Analysis control (optional)
    const SCRIPT_STATE    *psState,     // In   Initial bidi algorithm state (optional)
    SCRIPT_ITEM           *pItems,      // Out  Array to receive itemization
    int                   *pcItems);    // Out  Count of items processed (optional)






/////
//
//
//      Returns E_INVALIDARG if pwcInChars == NULL or cInChars == 0
//          or pItems == NULL or cMaxItems < 2.
//
//      Returns E_OUTOFMEMORY if the output buffer length (cMaxItems) is
//          insufficient. Note that in this case, as in all error cases, no
//          items have been fully processed so no part of the output array
//          contains defined values.
//
//      If psControl and psState are NULL on entry, ScriptItemize
//      breaks the unicode string purely by character code.  If they are all
//      non-null, it performs a full Unicode bidi analysis.
//
//      ScriptItemize always adds a terminal item to the item analysis array
//      (pItems) such that the length of an item at pItem is always available as:
//
//c     pItem[1].iCharPos - pItem[0].iCharPos
//
//      For this reason, it is invalid to call ScriptItemize with a buffer
//      of less than two SCRIPT_ANALYSIS items.
//
//      To perform a correct Unicode Bidi analysis, the SCRIPT_STATE should
//      be initialised according to the paragraph reading order at paragraph
//      start, and ScriptItemize should be passed the whole paragraph.
//
//      fRTL and fNumeric together provide the same classification as
//      the lpClass output from GetCharacterPlacement.
//
//      European digits U+0030 through U+0039 may be rendered as national
//      digits as follows:
//
//t     fDigitSubstitute | FContextDigits | Digit shapes displayed for Unicode U+0030 through U+0039
//t     ---------------- | -------------- | ------------------------------------
//t     False            | Any            | Western (European / American) digits
//t     True             | False          | As specified in SCRIPT_CONTROL.uDefaultLanguage
//t     True             | True           | As prior strong text, defaulting to SCRIPT_CONTROL.uDefaultLanguage
//
//
//      For fContextDigits, any Western digits (U+0030 - U+0039) encountered
//      before the first strongly directed character are substituted by the
//      traditional digits of the SCRIPT_CONTROL.uDefaultLanguage when that
//      language is written in the same direction as SCRIPT_STATE.uBidiLevel.
//
//      Thus, in a right-to-left string, if SCRIPT_CONTROL.uDefaultLanguage is
//      1 (LANG_ARABIC), then leading Western digits will be substituted by
//      traditional Arabic digits.
//
//      However, also in a right-to-left string, if SCRIPT_CONTROL.uDefaultLanguage
//      is 0x1e (LANG_THAI), then no substitution occurs on leading Western
//      digits because the Thai language is written left-to-right.
//
//      Following strongly directed characters, digits are substituted
//      by the traditional digits associated with the closest prior strongly
//      directed character.
//
//      The left-to-right mark (LRM) and right-to-left mark (RLM) are strong
//      characters whose language depends on the SCRIPT_CONTROL.uDefaultLangauge.
//
//      If SCRIPT_CONTROL.uDefaultLangauge is a left-to-right langauge, then
//      LRM causes subsequent Western digits to be substituted by the
//      traditional digits associated with that language, while Western
//      digits following RLM are not substituted.
//
//      Conversly, if SCRIPT_CONTROL.uDefaultLangauge is a right-to-left
//      langauge, then Western digits following LRM are not substituted, while
//      Western digits following RLM are substituted by the traditional digits
//      associated with that language.
//
//
//
//      Effect of Unicode control characters on SCRIPT_STATE:
//
//t     SCRIPT_STATE flag | Set by | Cleared by
//t     ----------------- | ------   ----------
//t     fDigitSubstitute  |  NADS  |   NODS
//t     fInhibitSymSwap   |  ISS   |   ASS
//t     fCharShape        |  AAFS  |   IAFS
//
//      SCRIPT_STATE.fArabicNumContext controls the Unicode EN->AN rule.
//      It should normally be initialised to TRUE
//      before itemizing an RTL paragraph in an Arabic language, FALSE
//      otherwise.
/////   ScriptLayout
//
//      The ScriptLayout function converts an array of run embedding levels to
//      a map of visual to logical position, and/or logical to visual position.
//
//      pbLevel must contain the embedding levels for all runs on the line,
//      ordered logically.
//
//      On output, piVisualToLogical[0] is the logical index of the run to
//      display at the far left. Subsequent entries should be displayed
//      progressing from left to right.
//
//      piLogicalToVisual[0] is the relative visual position where the first
//      logical run should be displayed - the leftmost display position being zero.
//
//      The caller may request either piLogicalToVisual or piVisualToLogical
//      or both.
//
//      Note: No other input is required since the embedding levels give all
//      necessary information for layout.


HRESULT WINAPI ScriptLayout(
    int           cRuns,              // In   Number of runs to process
    const BYTE   *pbLevel,            // In   Array of run embedding levels
    int          *piVisualToLogical,  // Out  List of run indices in visual order
    int          *piLogicalToVisual); // Out  List of visual run positions






/////   SCRIPT_JUSTIFY
//
//      The script justification enumeration provides the client with the
//      glyph characteristic information it needs to implement justification.


typedef enum tag_SCRIPT_JUSTIFY {
    SCRIPT_JUSTIFY_NONE           = 0,   // Justification can't be applied at this glyph
    SCRIPT_JUSTIFY_ARABIC_BLANK   = 1,   // This glyph represents a blank in an Arabic run
    SCRIPT_JUSTIFY_CHARACTER      = 2,   // Inter-character justification point follows this glyph
    SCRIPT_JUSTIFY_RESERVED1      = 3,   // Reserved #1
    SCRIPT_JUSTIFY_BLANK          = 4,   // This glyph represents a blank outside an Arabic run
    SCRIPT_JUSTIFY_RESERVED2      = 5,   // Reserved #2
    SCRIPT_JUSTIFY_RESERVED3      = 6,   // Reserved #3
    SCRIPT_JUSTIFY_ARABIC_NORMAL  = 7,   // Normal Middle-Of-Word glyph that connects to the right (begin)
    SCRIPT_JUSTIFY_ARABIC_KASHIDA = 8,   // Kashida(U+640) in middle of word
    SCRIPT_JUSTIFY_ARABIC_ALEF    = 9,   // Final form of Alef-like (U+627, U+625, U+623, U+632)
    SCRIPT_JUSTIFY_ARABIC_HA      = 10,  // Final form of Ha (U+647)
    SCRIPT_JUSTIFY_ARABIC_RA      = 11,  // Final form of Ra (U+631)
    SCRIPT_JUSTIFY_ARABIC_BA      = 12,  // Middle-Of-Word form of Ba (U+628)
    SCRIPT_JUSTIFY_ARABIC_BARA    = 13,  // Ligature of alike (U+628,U+631)
    SCRIPT_JUSTIFY_ARABIC_SEEN    = 14,  // Highest priority: Initial shape of Seen(U+633) (end)
    SCRIPT_JUSTIFY_RESERVED4      = 15,  // Reserved #4
} SCRIPT_JUSTIFY;



/////   SCRIPT_VISATTR
//
//      The visual (glyph) attribute buffer generated by ScriptShape
//      identifies clusters and justification points:


typedef struct tag_SCRIPT_VISATTR {
    WORD           uJustification   :4;  // Justification class
    WORD           fClusterStart    :1;  // First glyph of representation of cluster
    WORD           fDiacritic       :1;  // Diacritic
    WORD           fZeroWidth       :1;  // Blank, ZWJ, ZWNJ etc, with no width
    WORD           fReserved        :1;  // General reserved
    WORD           fShapeReserved   :8;  // Reserved for use by shaping engines
} SCRIPT_VISATTR;
//
//
//p     uJustification: Justification class for this glyph. See SCRIPT_JUSTIFY.
//
//p     fClusterStart: Set for the logically first glyph in every cluster,
//          even for clusters containing just one glyph.
//
//p     fDiacritic: Set for glyphs that combine with base characters.
//
//p     fZeroWidth: Set by the shaping engine for some, but not all, zero
//          width characters.


/////   ScriptShape
//
//      The ScriptShape function takes a Unicode run and generates glyphs and
//      visual attributes.
//
//      The number of glyphs generated varies according to the script and the
//      font. Only for simple scripts and fonts does each Unicode code point
//      generates a single glyph.
//
//      There is no limit on the number of glyphs generated by a codepoint.
//      For example, a sophisticated complex script font might choose to
//      constuct characters from components, and so generate many times as
//      many glyphs as characters.
//
//      There are also special cases like invalid character representations,
//      where extra glyphs are added to represent the invalid sequence.
//
//      A reasonable guess might be to provide a glyph buffer 1.5 times the
//      length of the character buffer, plus a 16 glyph fixed addition for
//      rare cases like invalid sequenece representation.
//
//      If ScriptShape returns E_OUTOFMEMORY it will be necessary to recall
//      it, possibly more than once, until a large enough buffer is found.


HRESULT WINAPI ScriptShape(
    HDC                 hdc,            // In    Optional (see under caching)
    SCRIPT_CACHE       *psc,            // InOut Cache handle
    const WCHAR        *pwcChars,       // In    Logical unicode run
    int                 cChars,         // In    Length of unicode run
    int                 cMaxGlyphs,     // In    Max glyphs to generate
    SCRIPT_ANALYSIS    *psa,            // InOut Result of ScriptItemize (may have fNoGlyphIndex set)
    WORD               *pwOutGlyphs,    // Out   Output glyph buffer
    WORD               *pwLogClust,     // Out   Logical clusters
    SCRIPT_VISATTR     *psva,           // Out   Visual glyph attributes
    int                *pcGlyphs);      // Out   Count of glyphs generated






/////
//
//      Returns E_OUTOFMEMORY if the output buffer length (cMaxGlyphs) is
//          insufficient. Note that in this case, as in all error cases, the
//          content of all output parameters are undefined.
//
//p     psa: Pass the SCRIPT_ANALYSIS field of the SCRIPT_ITEM entry for this
//          item. (The SCRIPT_ITEM array is returned by ScriptItemize.)
//
//      Clusters are sequenced uniformly within the run, as are glyphs within
//      the cluster - the fRTL item flag (from ScriptItemize) identifies
//      whether left to right, or right to left.
//
//p     pwLogClust: has cChars elements - each entry in pwLogClust corresponds
//          to a character in the input string (pwcChars). The value in each
//          pwLogCLust entry is the offset of the first glyph in the cluster
//          that contains this character.
//
//      Example: In the following example, there are four clusters:
//      1st cluster: one character represented by one glyph
//      2nd cluster: one character represented by 3 glyphs
//      3rd cluster: three characters represented by one glyph
//      4th cluster: 2 characters represented by three glyphs
//
//      Glyph array: (c<n>g<m> means cluster n glyph m)
//c        0      1    2    3      4      5    6    7
//c     -------------------------------------------------
//c     | c1g1 | c2g1 c2g2 c2g3 | c3g1 | c4g1 c4g2 c4g3 |
//c     -------------------------------------------------
//
//      Character array: (c<n>u<m> means cluster n Unicode codepoint m)
//c        0      1      2    3    4      5    6
//c     --------------------------------------------
//c     | c1u1 | c2u1 | c3u1 c3u2 c3u3 | c4u1 c4u2 |
//c     --------------------------------------------
//
//      LogClust: (one entry per character gives 1st glyph in cluster
//c     --------------------------------------------
//c     |   0  |   1  |   4    4    4  |   5    5  |
//c     --------------------------------------------
//
//      Note that for an RTL run (SCRIPT_ANALYSIS.a.fRTL == TRUE) and when
//      fLogicalOrder == FALSE (the default), glyphs are generated in visual
//      order - the reverse of the codepoint order, and the values in the
//      LogClust array will be descending.
//
//
//p     psva: has one visual attribute per glyph and so has maxGlyphs entries.
//
//
//      ScriptShape may set the fNoGlyphIndex flag in psa if the font or
//      OS cannot support glyph indices.
//
//      If fLogicalOrder is requested in psa, glyphs will be always be
//      generated in the same order as the original Unicode characters.
//
//      If fLogicalOrder is not set, right to left items are generated in
//      reverse order, so ScriptTextOut does not need to reverse them before
//      calling ExtTextOut.
/////   ScriptPlace
//
//      The ScriptPlace function takes the output of a ScriptShape call and
//      generates glyph advance width and 2D offset information.
//
//      The composite ABC width for the whole item identifies how much the
//      glyphs overhang to the left of the start position and to the right of
//      the length implied by the sum of the advance widths.
//
//      The total advance width of the line is exactly abcA + abcB + abcC.
//
//      abcA and abcC are maintained internally by Uniscribe as proportions
//      of the cell height represented in 8 bits and are thus roughly +/- 1%.
//      The total width returned (as the sum of piAdvance, and as the sum of
//      abcA+abcB+abcC) is accurate to the resolution of the TrueType shaping
//      engine.
//
//      All glyph related arrays are in visual order unless the fLogicalOrder
//      flag is set in psa.


#ifndef LSDEFS_DEFINED
typedef struct tagGOFFSET {
    LONG  du;
    LONG  dv;
} GOFFSET;
#endif


HRESULT WINAPI ScriptPlace(
    HDC                     hdc,        // In    Optional (see under caching)
    SCRIPT_CACHE           *psc,        // InOut Cache handle
    const WORD             *pwGlyphs,   // In    Glyph buffer from prior ScriptShape call
    int                     cGlyphs,    // In    Number of glyphs
    const SCRIPT_VISATTR   *psva,       // In    Visual glyph attributes
    SCRIPT_ANALYSIS        *psa,        // InOut Result of ScriptItemize (may have fNoGlyphIndex set)
    int                    *piAdvance,  // Out   Advance wdiths
    GOFFSET                *pGoffset,   // Out   x,y offset for combining glyph
    ABC                    *pABC);      // Out   Composite ABC for the whole run (Optional)






/////   ScriptTextOut
//
//      The ScriptTextOut function takes the output of both ScriptShape and
//      ScriptPlace calls and calls the operating system ExtTextOut function
//      appropriately. If the last parameter is not null, GDI's ExtTextOutW calls
//      are routed to this function.
//
//      All arrays are in visual order unless the fLogicalOrder flag is set in
//      psa.


HRESULT WINAPI ScriptTextOut(
    const HDC                           hdc,            // In     OS handle to device context (required)
    SCRIPT_CACHE                        *psc,           // InOut  Cache handle
    int                                 x,              // In     x,y position for first glyph
    int                                 y,              // In
    UINT                                fuOptions,      // In     ExtTextOut options
    const RECT                          *lprc,          // In     optional clipping/opaquing rectangle
    const SCRIPT_ANALYSIS               *psa,           // In     Result of ScriptItemize
    const WCHAR                         *pwcReserved,   // In     Reserved (requires NULL)
    int                                 iReserved,      // In     Reserved (requires 0)
    const WORD                          *pwGlyphs,      // In     Glyph buffer from prior ScriptShape call
    int                                 cGlyphs,        // In     Number of glyphs
    const int                           *piAdvance,     // In     Advance widths from ScriptPlace
    const int                           *piJustify,     // In     Justified advance widths (optional)
    const GOFFSET                       *pGoffset);     // In     x,y offset for combining glyph






/////
//
//      The caller should normally use SetTextAlign(hdc, TA_RIGHT) before
//      calling ScriptTextOut with an RTL item inlogical order.
//
//      The piJustify array provides requested cell widths for each glyph.
//      When the piJustify width of a glyph differs from the unjustified
//      width (in PiAdvance), space is added to or removed from the glyph
//      cell at it's trailing edge. The glyph is always aligned with the
//      leading edge of it's cell. (This rule applies even in visual order.)
//
//      When a glyph cell is extended the extra space is uaually made up by
//      the addition of white space, however for Arabic scripts, the extra
//      space is made up by one or more kashida glyphs, unless the extra space
//      is insufficient for the shortest kashida glyph in the font. (The
//      width of the shortest kashida is available by calling
//      ScriptGetFontProperties.)
//
//      piJustify should only be passed if re-justification of the string is
//      required. Normally pass NULL to this parameter.
//
//      fuOptions may contain ETO_CLIPPED or ETO_OPAQUE (or neither or both).
//
//      Do not use ScriptTextOut to write to a metafile unless you are sure
//      that the metafile will eventually be played back without any font
//      substitution. ScriptTextOut record glyph numbers in the metafile.
//      Since glyph numbers vary considerably from one font to another
//      such a metafile is unlikely to play back correctly when differant
//      fonts are substituted.
//
//      For example when a metafile is played back at a different scale
//      CreateFont requests recorded in the metafile may resolve to bitmap
//      instead of truetype fonts, or if the metafile is played back on
//      a different machine requested fonts may not be installed.//
//
//      To write complex scripts in a metafile in a font independant manner,
//      use ExtTextOut to write the logical characters directly, so that
//      glyph generation and placement does not occur until the text is
//      played back.
/////   ScriptJustify
//
//      ScriptJustify provides a simple minded implementation of multilingual
//      justification.
//
//      Sophisticated text formatters may prefer to generate their own delta
//      dx array by combining their own features with the information returned
//      by ScriptShape in the SCRIPT_VISATTR array.
//
//      ScriptJustify establishes how much adjustment to make at each glyph
//      position on the line. It interprets the SCRIPT_VISATTR array generated
//      by a call to ScriptShape, and gives top priority to kashida, then uses
//      inter word spacing if there's no kashida points, then uses
//      intercharacter spacing if there are no inter-word points.
//
//      The justified advance widths generated in ScriptJustify should be
//      passed to ScriptTextOut in the piJustify paramter.
//
//      ScriptJustify creates a justify array containing updated advance
//      widths for each glyph. Where a glyphs advance width is increased, it
//      is expected that the extra width will be rendered to the right of the
//      glyph, with as white space or, for Arabic text, as kashida.
/////
HRESULT WINAPI ScriptJustify(
    const SCRIPT_VISATTR  *psva,        // In   Collected visual attributes for entire line
    const int             *piAdvance,   // In   Advance widths from ScriptPlace
    int                    cGlyphs,     // In   Size of all arrays
    int                    iDx,         // In   Desired width change, either increase or descrease
    int                    iMinKashida, // In   Minimum length of continuous kashida glyph to generate
    int                   *piJustify);  // Out  Updated advance widths to pass to ScriptTextOut






/////   SCRIPT_LOGATTR
//
//      The SCRIPT_LOGATTR structure describes attributes of logical
//      characters useful when editing and formatting text.
//
//      Note that for wordbreaking and linebreaking, if the first character of
//      the run passed in is not whitespace, the client needs to check whether
//      the last character of the previous run is whitespace to determine if
//      the first character of this run is the start of a word.
//
//
typedef struct tag_SCRIPT_LOGATTR {
    BYTE    fSoftBreak      :1;     // Potential linebreak point
    BYTE    fWhiteSpace     :1;     // A unicode whitespace character, except NBSP, ZWNBSP
    BYTE    fCharStop       :1;     // Valid cursor position (for left/right arrow)
    BYTE    fWordStop       :1;     // Valid cursor position (for ctrl + left/right arrow)
    BYTE    fInvalid        :1;     // Invalid character sequence
    BYTE    fReserved       :3;
} SCRIPT_LOGATTR;
//
//
//p     fSoftBreak: It would be valid to break the line in front of this
//              character. This flag is set on the first character of
//              South-East Asian words. Note that when linebreaking the
//              client would usually also treat any nonblank following a blank
//              as a softbreak position, by inspecting the fWhiteSPace flag
//              below.
//
//p     fWhiteSpace: This character is one of the many Unicode character
//              that are classified as breakable whitespace.
//
//p     fCharStop: Valid cursor position. Set on most characters, but not
//              on codepoints inside Indian and South East Asian character
//              clusters. May be used to implement left and right arrow
//              operation in editors.
//
//p     fWordStop: Valid position following word advance/retire commonly
//              implemented at ctrl/left-arrow and ctrl/right-arrow.
//              May be used to implement ctrl+left and ctrl+right arrow
//              operation in editors. As with fSoftBreak clients should
//              normally also inspect the fWhiteSpace flag and treat the
//              first character after a run of whitespace as the start of a
//              word.
//
//p     fInvalid: Marks characters which form an invalid or undisplayable
//              combination. Scripts which can set this flag have the flag
//              fInvalidLogAttr set in their SCRIPT_PROPERTIES.






/////   ScriptBreak
//
//      The ScriptBreak function returns cursor movement and formatting break
//      positions for an item as an array of SCRIPT_LOGATTRs. To support
//      mixed formatting within a single word correctly, ScriptBreak should
//      be passed whole items as returned by ScriptItemize.
//
//      ScriptBreak does not require an hdc and does not execute glyph shaping.
//
//      The fCharStop flag marks cluster boundaries for those scripts where
//      it is conventional to restrict from moving inside clusters. The same
//      boundaries could also be inferred by inspecting the pLogCLust array
//      returned by ScriptShape, however ScriptBreak is considerably faster in
//      implementation and does not require an hdc to be prepared.
//
//      The fWordStop, fSoftBreak and fWhiteSpace flags are only available
//      through ScriptBreak.
//
//      Most shaping engines that identify invalid sequences do so by setting
//      the fInvalid flag in ScriptBreak. The fInvalidLogAttr flag in
//      ScriptProperties identifies which scripts do this.


HRESULT WINAPI ScriptBreak(
    const WCHAR            *pwcChars,  // In   Logical unicode item
    int                     cChars,    // In   Length of unicode item
    const SCRIPT_ANALYSIS  *psa,       // In   Result of earlier ScriptItemize call
    SCRIPT_LOGATTR         *psla);     // Out  Logical character attributes






/////   ScriptCPtoX
//
//      The ScriptCPtoX function returns the x offset from the left end
//      (!fLogical) or leading edge (fLogical) of a run to either the leading
//      or the trailing edge of a logical character cluster.
//
//      iCP is the offset of any logical character in the cluster.
//
//      For scripts where the caret may conventionally be placed into the
//      middle of clusters (e.g. Arabic, Hebrew), the returned X may be
//      an interpolated position for any codepoint in the line.
//
//      For scripts where the caret is conventionally snapped to the boundaries
//      of clusters, (e.g. Thai, Indian), the resulting X position will be
//      snapped to the requested edge of the cluster containing CP.


HRESULT WINAPI ScriptCPtoX(
    int                     iCP,        // In   Logical character position in run
    BOOL                    fTrailing,  // In   Which edge (default - leading)
    int                     cChars,     // In   Count of logical codepoints in run
    int                     cGlyphs,    // In   Count of glyphs in run
    const WORD             *pwLogClust, // In   Logical clusters
    const SCRIPT_VISATTR   *psva,       // In   Visual glyph attributes array
    const int              *piAdvance,  // In   Advance widths
    const SCRIPT_ANALYSIS  *psa,        // In   Script analysis from item attributes
    int                    *piX);       // Out  Resulting X position






/////   ScriptXtoCP
//
//      The ScriptXtoCP function converts an x offset from the left end
//      (!fLogical) or leading edge (fLogical) of a run to a logical
//      character position and a flag that indicates whether the X position
//      fell in the leading or the trailing half of the character.
//
//      For scripts where the cursor may conventionally be placed into the
//      middle of clusters (e.g. Arabic, Hebrew), the returned CP may be
//      for any codepoint in the line, and fTrailing will be either zero
//      or one.
//
//      For scripts where the cursor is conventionally snapped to the
//      boundaries of a cluster, the returned CP is always the position of
//      the logically first codepoint in a cluster, and fTrailing is either
//      zero, or the number of codepoints in the cluster.
//
//      Thus the appropriate cursor position for a mouse hit is always the
//      returned CP plus the value of fTrailing.
//
//      If the X positition passed is not in the item at all, the resulting
//      position will be the trailing edge of character -1 (for X positions
//      before the item), or the leading edge of character 'cChars' (for
//      X positions following the item).


HRESULT WINAPI ScriptXtoCP(
    int                     iX,             // In   X offset from left of run
    int                     cChars,         // In   Count of logical codepoints in run
    int                     cGlyphs,        // In   Count of glyphs in run
    const WORD             *pwLogClust,     // In   Logical clusters
    const SCRIPT_VISATTR   *psva,           // In   Visual glyph attributes
    const int              *piAdvance,      // In   Advance widths
    const SCRIPT_ANALYSIS  *psa,            // In   Script analysis from item attributes
    int                    *piCP,           // Out  Resulting character position
    int                    *piTrailing);    // Out  Leading or trailing half flag






/////   Relationship between caret positions, justifications points and clusters
//
//
//t     Job                              | Uniscribe support
//t     -------------------------------- | --------------------------------------------------------
//t     Caret move by character cluster  | LogClust or VISATTR.fClusterStart or LOGATTR.fCharStop
//t     Line breaking between characters | LogClust or VISATTR.fClusterStart or LOGATTR.fCharStop
//t     Caret move by word               | LOGATTR.fWordStop
//t     Line breaking between words      | LOGATTR.fWordStop
//t     Justification                    | VISATTR.uJustification
//
//
//
/////   Character clusters
//
//      Character clusters are glyph sequences that cannot be split between
//      lines.
//
//      Some languages (e.g. Thai, Indic) restrict caret placement to points
//      betwen clusters. This applies both to keyboard initiated caret
//      movement (e.g. cursor keys) and pointing and clicking with the mouse
//      (hit testing).
//
//      Uniscribe provides cluster information in both the visual and logical
//      attributes. If you've called ScriptShape you'll find the cluster
//      information represented both by sequences of the same value in the
//      pwLogClust array, and by the fClusterStart flag in the psva
//      SCRIPT_VISATTR array.
//
//      ScriptBreak also returns the fCharStop flag in the SCRIPT_LOGATTR
//      array to identify cluster positions.
//
//
//
/////   Word break points
//
//      Valid positions for moving the caret when moving in whole words are
//      marked by the fWordStop flag returned by ScriptBreak.
//
//      Valid positions for breaking lines between words are marked by the
//      fSoftBreak flag returned by ScriptBreak.
//
//
//
/////   Justification
//
//      Justification space or kashida should be inserted where identified by
//      the uJustificaion field of the SCRIPT_VISATTR.
//
//      When performing inter-character justification, insert extra space
//      only after glyphs marked with uJustify == SCRIPT_JUSTIFY_CHARACTER.
//
//
//
/////   Script specific processing
//
//      Uniscribe provides information about special processing for each
//      script in the SCRIPT_PROPERTIES array.
//
//      Use the following code during initialisation to get a pointer to
//      the SCRIPT_PROPERTIES array:
//
//c     const SCRIPT_PROPERTIES **g_ppScriptProperties; // Array of pointers to properties
//c     int iMaxScript;
//c     HRESULT hr;
//
//c     hr = ScriptGetProperties(&g_ppScriptProperties, &g_iMaxScript);
//
//      Then inspect the properties of the script of an item 'iItem' as follows:
//
//c     hr = ScriptItemize( ... , pItems, ... );
//c     ...
//c     if (g_ppScriptProperties[pItems[iItem].a.eScript]->fNeedsCaretInfo) {
//c         // Use ScriptBreak to restrict the caret from entering clusters (for example).
//c     }
//
//
//      SCRIPT_PROPERTIES.fNeedsCaretInfo
//
//      Caret placement should be restricted to cluster
//      edges for scripts such as Thai and Indian. The fNeedsCaretInfo flag
//      in SCRIPT_PROPERTIES identifies such languages.
//
//      Note that ScriptXtoCP and ScriptCPtoX automatically apply caret
//      placement restictions.
//
//
//      SCRIPT_PROPERTIES.fNeedsWordBreaking
//
//      For most scripts, word break placement  may be
//      identified by scanning for characters marked as fWhiteSpace in
//      SCRIPT_LOGATTR, or for glyphs marked as uJustify ==
//      SCRIPT_JUSTIFY_BLANK or SCRIPT_JUSTIFY_ARABIC_BLANK in SCRIPT_VISATTR.
//
//      For languages such as Thai, it is also necessary to call ScriptBreak,
//      and include character positions marked as fWordStop in SCRIPT_LOGATTR.
//      Such scripts are marked as fNeedsWordbreaking in SCRIPT_PROPERTIES.
//
//
//      SCRIPT_PROPERTIES.fNeedsCharacterJustify
//
//      Languages such as Thai also require inter-character spacing when
//      justifying (where uJustify == SCRIPT_JUSTIFY_CHARACTER in the
//      SCRIPT_VISATTR). Such languages are marked as fNeedsCharacterJustify
//      in SCRIPT_PROPERTIES.
//
//
//      SCRIPT_PROPERTIES.fAmbiguousCharSet
//
//      Many Uniscribe scripts do not correspond directly to 8 bit character
//      sets. For example Unicode characters in the range U+100 through U+024F
//      represent extended latin shapes used for many languages, including
//      those supported by EASTEUROPE_CHARSET, TURKISH_CHARSET and
//      VIETNAMESE_CHARSET. However many of these characters are supported by
//      more han one of thsese charsets.
//      fAmbiguousCharset is set for any script token which could contain
//      characters from a number of these charsets. In these cases the bCharSet
//      field may contain ANSI_CHARSET or DEFAULT_CHARSET. The Uniscribe client
//      will generally need to apply futher processing to determine which charset
//      to use when requesting a font suitable for this run. For example it
//      determine that the run consists of multiple languages and split it up
//      to use a different font for each language.






/////   Notes on ScriptXtoCP and ScriptCPtoX
//
//      Both functions work only within runs and require the results of a
//      previous ScriptShape call.
//
//      The client must establish which run a given cursor offset or x
//      position is within before passing it to ScriptCPtoX or ScriptXtoCP.
//
//      Cluster information in the logical cluster array is used to share
//      the width of a cluster of glyphs equally among the logical characters
//      they represent.
//
//      For example, the lam alif glyph is divided into four areas: the
//      leading half of the lam, the trailing half of the lam, the leading
//      half of the alif and the trailing half of the alif.
//
//      ScriptXtoCP Understands the caret position conventions of each script.
//      For Indian and Thai, caret positions are snapped to cluster boundaries,
//      for Arabic and Hebrew, caret positions are interpolated within clusters.
//
//
/////   Translating mouse hit 'x' offset to caret position
//
//      Conventionally, caret position 'cp' may be selected by clicking either
//      on the trailing half of character 'cp-1' or on the leading half of
//      character 'cp'. This may easily be implemented as follows:
//
//c     int iCharPos;
//c     int iCaretPos
//c     int fTrailing;
//
//c     ScriptXtoCP(iMouseX, ..., &iCharPos, &fTrailing);
//c     iCaretPos = iCharPos + fTrailing;
//
//      For scripts that snap the caret to cluster boundaries, ScriptXtoCP
//      returns ftrailing set to either 0, or the width of the cluster in
//      codepoints. Thus the above code correctly returns only valid
//      caret positions.
//
//
/////   Displaying the caret in bidi strings
//
//      In unidirectional text, the leading edge of a character is at the same
//      place as the trailing edge of the previous character, so there is no
//      ambiguity in placing the caret between characters.
//
//      In bidirectional text, the caret position between runs of opposing
//      direction may be ambiguous.
//
//      For example in the left to right paragraph 'helloMAALAS', the last
//      letter of 'hello' immediately preceeds the first letter of 'salaam'.
//      The best position to display the caret depends on whether it is
//      considered to follow the 'o' of 'hello', or to preceed the 's' of
//      'salaam'.
//
/////   Commonly used caret positioning conventions
//
//t     Situation       | Visual caret placement
//t     ---------       | -------------------------------------------
//t     Typing          | Trailing edge of last character typed
//t     Pasting         | Trailing edge of last character pasted
//t     Caret advancing | Trailing edge of last character passed over
//t     Caret retiring  | Leading edge of last character passed over
//t     Home            | Leading edge of line
//t     End             | Trailing edge of line
//
//      The caret may be positioned as follows:
//
//c     if (advancing) {
//c         ScriptCPtoX(iCharPos-1, TRUE, ..., &iCaretX);
//c     } else {
//c         ScriptCPtoX(iCharPos, FALSE, ..., &iCaretX);
//c     }
//
//      Or, more simply, given an fAdvancing BOOL restricted to TRUE or FALSE:
//
//c     ScriptCPtoX(iCharPos-fAdvancing, fAdvancing, ..., &iCaretX);
//
//      ScriptCPtoX handles out of range positions logically: it returns the
//      leading edge of the run for iCharPos <0, and the trailing edge of the
//      run for iCharPos >=length.
/////   ScriptGetLogicalWidths
//
//      Converts visual withs in piAdvance into logical widths,
//      one per original character, in logical order.
//
//      Ligature glyphs widths are divided evenly amongst the characters
//      they represent.


HRESULT WINAPI ScriptGetLogicalWidths(
    const SCRIPT_ANALYSIS  *psa,            // In   Script analysis from item attributes
    int                     cChars,         // In   Count of logical codepoints in run
    int                     cGlyphs,        // In   Count of glyphs in run
    const int              *piGlyphWidth,   // In   Advance widths
    const WORD             *pwLogClust,     // In   Logical clusters
    const SCRIPT_VISATTR   *psva,           // In   Visual glyph attributes
    int                    *piDx);          // Out  Logical widths






/////
//      ScriptGetLogicalWidths is useful for recording widths in a
//      font independant manner. By passing the recorded logical widths
//      to ScriptApplyLogicalWidths, a block of text can be replayed in the
//      same boundaries with acceptable loss of quality even when the original
//      font is not available.
/////   ScriptApplyLogicalWidth
//
//      Accepts an array of advance widths in logical order, corresponding
//      one to one with codepoints, and generates an array of glyph widths
//      suitable for passing to the piJustify parameter of ScriptTextOut.
//
//      ScriptApplyLogicalWidth may be used to reapply logical widths
//      obtained with ScriptGetLogicalWidths. It may be useful in situations
//      such as metafiling, where it is necessary to record and reapply
//      advance width information in a font independant manner.



HRESULT WINAPI ScriptApplyLogicalWidth(
    const int              *piDx,        // In     Logical dx array to apply
    int                     cChars,      // In     Count of logical codepoints in run
    int                     cGlyphs,     // In     Glyph count
    const WORD             *pwLogClust,  // In     Logical clusters
    const SCRIPT_VISATTR   *psva,        // In     Visual attributes from ScriptShape/Place
    const int              *piAdvance,   // In     Glyph advance widths from ScriptPlace
    const SCRIPT_ANALYSIS  *psa,         // In     Script analysis from item attributes
    ABC                    *pABC,        // InOut  Updated item ABC width (optional)
    int                    *piJustify);  // Out    Resulting glyph advance widths for ScriptTextOut






/////
//p     piDx: Pointer to an array of dx widths in logical order, one per codepoint.
//
//p     cChars: Count of the logical codepoints in the run.
//
//p     cGlyphs: Glyph count.
//
//p     pwLogClust: Pointer to an array of logical clusters from ScriptShape
//
//p     psva: Pointer to an array of visual attributes from ScriptShape and
//          updated by ScriptPlace.
//
//p     piAdvance: Pointer to an array of glyph advance widths from ScriptPlace.
//
//p     psa: Pointer to a SCRIPT_ANALYSIS structure from ScriptItemize and
//          updated by ScriptShape and SriptPlace..
//
//p     pABC: Pointer to the run overall ABC width (optional). If present,
//          when the function is called, it should contain the run ABC width
//          returned by ScriptPlace; when the function returns, the ABC width
//          has been updated to match the new widths.
//
//p     piJustify:Pointer to an array of the resulting glyph advance widths.
//          This is suitable for passing to the piJustify parameter of ScriptTextOut.
/////   ScriptGetCMap
//
//      ScriptGetCMap may be used to determine which characters in a run
//      are supported by the selected font.
//
//      It returns glyph indices of Unicode characters according to Truetype
//      Cmap table, or standard Cmap implemented for old style fonts. The
//      glyph indices are returned in the same order as the input string.
//
//      The caller may scan the returned glyph buffer looking for the default
//      glyph to determine which characters are not available. (The default
//      glyph index for the selected font should be determined by calling
//      ScriptGetFontProperties).
//
//      The return value indicates the presence of any missing glyphs.


#define SGCM_RTL  0x00000001      // Return mirrored glyph for mirrorable Unicode codepoints


HRESULT WINAPI ScriptGetCMap(
    HDC             hdc,            // In    Optional (see notes on caching)
    SCRIPT_CACHE   *psc,            // InOut Address of Cache handle
    const WCHAR    *pwcInChars,     // In    Unicode codepoint(s) to look up
    int             cChars,         // In    Number of characters
    DWORD           dwFlags,        // In    Flags such as SGCM_RTL
    WORD           *pwOutGlyphs);   // Out   Array of glyphs, one per input character






/////
//  returns S_OK     - All unicode codepoints were present in the font
//          S_FALSE  - Some of the Unicode codepoints were mapped to the default glyph
//          E_HANDLE - font or system does not support glyph indices
/////   ScriptGetGlyphABCWidth
//
//      Returns ABC width of a given glyph.
//      May be useful for drawing glyph charts. Should not be used for
//      run of the mill complex script text formatting.


HRESULT WINAPI ScriptGetGlyphABCWidth(
    HDC             hdc,            // In    Optional (see notes on caching)
    SCRIPT_CACHE   *psc,            // InOut Address of Cache handle
    WORD            wGlyph,         // In    Glyph
    ABC            *pABC);          // Out   ABC width






/////
//  returns S_OK     - Glyph width returned
//          E_HANDLE - font or system does not support glyph indices
/////   SCRIPT_PROPERTIES
//
typedef struct {
    DWORD   langid                 :16; // Primary and sublanguage associated with script
    DWORD   fNumeric               :1;
    DWORD   fComplex               :1;  // Script requires special shaping or layout
    DWORD   fNeedsWordBreaking     :1;  // Requires ScriptBreak for word breaking information
    DWORD   fNeedsCaretInfo        :1;  // Requires caret restriction to cluster boundaries
    DWORD   bCharSet               :8;  // Charset to use when creating font
    DWORD   fControl               :1;  // Contains only control characters
    DWORD   fPrivateUseArea        :1;  // This item is from the Unicode range U+E000 through U+F8FF
    DWORD   fNeedsCharacterJustify :1;  // Requires inter-character justification
    DWORD   fInvalidGlyph          :1;  // Invalid combinations generate glyph wgInvalid in the glyph buffer
    DWORD   fInvalidLogAttr        :1;  // Invalid combinations are marked by fInvalid in the logical attributes
    DWORD   fCDM                   :1;  // Contains Combining Diacritical Marks
    DWORD   fAmbiguousCharSet      :1;  // Script does not correspond 1:1 with a charset
    DWORD   fClusterSizeVaries     :1;  // Measured cluster width depends on adjacent clusters
    DWORD   fRejectInvalid         :1;  // Invalid combinations should be rejected
} SCRIPT_PROPERTIES;
//
//p     langid: Language associated with this script. When a script is used for many languages,
//          langid id represents a default language. For example, Western script is represented
//          by LANG_ENGLISH although it is also used for French, German, Spanish etc.
//
//p     fNumeric: Script contains numerics and characters used in conjunction with numerics
//          by the rules of the Unicode bidirectional algorithm. For example
//          dollar sign and period are classified as numeric when adjacent to or in between
//          digits.
//
//p     fComplex: Indicates a script that requires complex script handling. If fComplex is false
//          the script contains no combining characters and requires no contextual shaping or reordering.
//
//p     fNeedsWordBreaking: A script, such as Thai, which requires algorithmic wordbreaking.
//          Use ScriptBreak to obtain a wordbreak points using the standard system wordbreaker.
//
//p     fNeedsCaretInfo: A script, such as Thai and Indian, where the caret may not be placed
//          inside a cluster. To determine valid caret positions inspect the fCharStop flag in the
//          logical attributes returned by ScriptBreak, or compare adjacent values in the pwLogClust
//          array returned by ScriptShape.
//
//p     bCharSet: Nominal charset associated with script. May be used in a logfont when creating
//          a font suitable for displaying this script. Note that for new scripts where there
//          is no charset defined, bCharSet may be innapropriate and DEFAULT_CHARSET should
//          be used instead - see the description of fAmbiguousCharSet below.
//
//p     fControl: contains control characters.
//
//p     fPrivateUseArea: The Unicode range U+E000 through U+F8FF.
//
//p     fNeedsCharacterJustify: A script, such as Thai, where justification is conventionally
//          achieved by increasing the space between all letters, not just between words.
//
//p     fInvalidGlyph: A script for which ScriptShape generates an invalid glyph
//          to represent invalid sequences. The glyph index of the invalid glyph for
//          a particular font may be obtained by calling ScriptGetFontProperties.
//
//p     fInvalidLogAttr: A script for which ScriptBreak sets the fInvalid flag
//          in the logical attributes to mark invalid sequences.
//
//p     fCDM: Implies that an item analysed by ScriptItemize included combining
//          diacritical marks (U+0300 through U+36F).
//
//p     fAmbiguousCharSet: No single legacy charset supports this script.
//          For example the extended Latin Extended-A Unicode range includes
//          characters from the EASTUROPE_CHARSET, the TURKISH_CHARSET and the
//          BALTIC_CHARSET. It also contains characters that are not available
//          in any legacy charset. Use DEFAULT_CHARSET when creating fonts to
//          display parts of this run.
//
//p     fClusterSizeVaries: A script, such as Arabic, where contextual shaping
//          may cause a string to increase in size when removing characters.
//
//p     fRejectInvalid: A script, such as Thai, where invalid sequences conventionally
//          cause an editor such as notepad to beep, and ignore keypresses.


/////   ScriptGetProperties
//
//      ScriptGetProperties returns the address of a table that maps a
//      script in a SCRIPT_ANALYSIS uScript field to properties including
//      the primary language associated with that script, whether it's
//      numeric and whether it's complex.


HRESULT WINAPI ScriptGetProperties(
    const SCRIPT_PROPERTIES ***ppSp,             // Out  Receives pointer to table of pointers to properties indexed by script
    int                       *piNumScripts);    // Out  Receives number of scripts (valid values are 0 through NumScripts-1)






/////   SCRIPT_FONTPROPERTIES
//
typedef struct {
    int     cBytes;         // Structure length
    WORD    wgBlank;        // Blank glyph
    WORD    wgDefault;      // Glyph used for Unicode values not present in the font
    WORD    wgInvalid;      // Glyph used for invalid character combinations (especially in Thai)
    WORD    wgKashida;      // Shortest continuous kashida glyph in the font, -1 if doesn't exist
    int     iKashidaWidth;  // Widths of shortest continuous kashida glyph in the font
} SCRIPT_FONTPROPERTIES;


/////   ScriptGetFontProperties
//
//      Returns information from the font cache


HRESULT WINAPI ScriptGetFontProperties(
    HDC                     hdc,    // In    Optional (see notes on caching)
    SCRIPT_CACHE           *psc,    // InOut Address of Cache handle
    SCRIPT_FONTPROPERTIES  *sfp);   // Out   Receives properties for this font






/////   ScriptCacheGetHeight
//
//


HRESULT WINAPI ScriptCacheGetHeight(
    HDC            hdc,         // In    Optional (see notes on caching)
    SCRIPT_CACHE  *psc,         // InOut Address of Cache handle
    long          *tmHeight);   // Out   Receives font height in pixels






/////   ScriptStringAnalyse
//
//
#define SSA_PASSWORD         0x00000001  // Input string contains a single character to be duplicated iLength times
#define SSA_TAB              0x00000002  // Expand tabs
#define SSA_CLIP             0x00000004  // Clip string at iReqWidth
#define SSA_FIT              0x00000008  // Justify string to iReqWidth
#define SSA_DZWG             0x00000010  // Provide representation glyphs for control characters
#define SSA_FALLBACK         0x00000020  // Use fallback fonts
#define SSA_BREAK            0x00000040  // Return break flags (character and word stops)
#define SSA_GLYPHS           0x00000080  // Generate glyphs, positions and attributes
#define SSA_RTL              0x00000100  // Base embedding level 1
#define SSA_GCP              0x00000200  // Return missing glyphs and LogCLust with GetCharacterPlacement conventions
#define SSA_HOTKEY           0x00000400  // Replace '&' with underline on subsequent codepoint
#define SSA_METAFILE         0x00000800  // Write items with ExtTextOutW Unicode calls, not glyphs
#define SSA_LINK             0x00001000  // Apply FE font linking/association to non-complex text
#define SSA_HIDEHOTKEY       0x00002000  // Remove first '&' from displayed string
#define SSA_HOTKEYONLY       0x00002400  // Display underline only.

#define SSA_FULLMEASURE      0x04000000  // Internal - calculate full width and out the number of chars can fit in iReqWidth.
#define SSA_LPKANSIFALLBACK  0x08000000  // Internal - enable FallBack for all LPK Ansi calls Except BiDi hDC calls
#define SSA_PIDX             0x10000000  // Internal
#define SSA_LAYOUTRTL        0x20000000  // Internal - Used when DC is mirrored
#define SSA_DONTGLYPH        0x40000000  // Internal - Used only by GDI during metafiling - Use ExtTextOutA for positioning
#define SSA_NOKASHIDA        0x80000000  // Internal - Used by GCP to justify the non Arabic glyphs only.
//
//
//p     SSA_HOTKEY: Note that SSA_HOTKEY and SSA_HIDEHOTKEY remove the
//          hotkey '&' character from further processing, so functions
//          such as ScriptString_pLogAttr return arrays based on a string
//          which excludes the '&'.




/////   SCRIPT_TABDEF
//
//      Defines tabstop positions for ScriptStringAnalyse (ignored unless SSA_TAB passed)
//
typedef struct tag_SCRIPT_TABDEF {
    int   cTabStops;        // Number of entries in pTabStops array
    int   iScale;           // Scale factor for pTabStops (see below)
    int  *pTabStops;        // Pointer to array of one or more tab stops
    int   iTabOrigin;       // Initial offset for tab stops (logical units)
} SCRIPT_TABDEF;
//
//
//p     cTabStops: Number of entries in the pTabStops array. If zero, tabstops
//          are every 8 average character widths. If one, all tabstops are
//          the length of the first entry in pTabStops. If more than one,
//          the first cTabStops are as specified in the pTabStops array,
//          subsequent tabstops are every 8 average characters from the last
//          tabstop in the array.
//
//p     iScale: Scale factor for iTabOrigin and pTabStops entries. Values are
//          converted to device coordinates by multiplying by iScale then
//          dividing by 4. If values are already in device units, set iScale to
//          4. If values are in dialog units, set iScale to the average char
//          width of the dialog font. If values are multiples of the average
//          character width for the selected font, set iScale to 0.
//
//p     pTabStops: Array of cTabStops entries. Each entry specifies a
//          tabstop position. Positive values give nearedge alignment,
//          negative values give faredge alignment.
//
//p     iTabOrigin: Tabs are considered to start iTabOrigin before the
//          beginning of the string. Helps with multiple tabbed
//          outputs on the same line.






/////   ScriptStringAnalyse
//
//      cString - Input string must contain at least one character
//
//      hdc - required if SSA_GLYPH requested. Optional for SSA_BREAK.
//      If present the current font in the hdc is inspected and if a symbolic
//      font the character string is treated as a single neutral SCRIPT_UNDEFINED item.
//
//      Note that the uBidiLevel field in the initial SCRIPT_STATE value
//      is ignored - the uBidiLevel used is derived from the SSA_RTL
//      flag in combination with the layout of the hdc.


typedef void* SCRIPT_STRING_ANALYSIS;


HRESULT WINAPI ScriptStringAnalyse(
    HDC                      hdc,       //In  Device context (required)
    const void              *pString,   //In  String in 8 or 16 bit characters
    int                      cString,   //In  Length in characters (Must be at least 1)
    int                      cGlyphs,   //In  Required glyph buffer size (default cString*1.5 + 16)
    int                      iCharset,  //In  Charset if an ANSI string, -1 for a Unicode string
    DWORD                    dwFlags,   //In  Analysis required
    int                      iReqWidth, //In  Required width for fit and/or clip
    SCRIPT_CONTROL          *psControl, //In  Analysis control (optional)
    SCRIPT_STATE            *psState,   //In  Analysis initial state (optional)
    const int               *piDx,      //In  Requested logical dx array
    SCRIPT_TABDEF           *pTabdef,   //In  Tab positions (optional)
    const BYTE              *pbInClass, //In  Legacy GetCharacterPlacement character classifications (deprecated)

    SCRIPT_STRING_ANALYSIS  *pssa);     //Out Analysis of string






/////   ScriptStringFree - free a string analysis
//
//


HRESULT WINAPI ScriptStringFree(
    SCRIPT_STRING_ANALYSIS *pssa);  //InOut Address of pointer to analysis






/////   ScriptStringSize
//
//      returns a pointer to the size (width and height) of an analysed string
//
//      Note that the SIZE pointer remains valid only until the
//      SCRIPT_STRING_ANALYSIS is passed to ScriptStringFree.


const SIZE* WINAPI ScriptString_pSize(
    SCRIPT_STRING_ANALYSIS   ssa); 






/////   ScriptString_pcOutChars
//
//      returns pointer to length of string after clipping (requires SSA_CLIP set)
//
//      Note that the int pointer remains valid only until the
//      SCRIPT_STRING_ANALYSIS is passed to ScriptStringFree.


const int* WINAPI ScriptString_pcOutChars(
    SCRIPT_STRING_ANALYSIS   ssa); 






/////   ScriptString_pLogAttr
//
//      returns pointer to logical attributes buffer in a SCRIPT_STRING_ANALYSIS
//
//      Note that the buffer pointer remains valid only until the
//      SCRIPT_STRING_ANALYSIS is passed to ScriptStringFree.
//
//      The logical attribute array contains *ScriptString_pcOutChars(ssa)
//      entries.


const SCRIPT_LOGATTR* WINAPI ScriptString_pLogAttr(
    SCRIPT_STRING_ANALYSIS   ssa); 






/////   ScriptStringGetOrder
//
//      Creates an array mapping original character position to glyph position.
//
//      Treats clusters as they were in legacy systems - Unless a cluster
//      contains more glyphs than codepoints, each glyph is referenced at
//      least once from the puOrder array.
//
//      Requires SSA_GLYPHS requested in original ScriptStringAnalyse call.
//
//      The puOrder parameter should address a buffer containing room for
//      at least *ScriptString_pcOutChars(ssa) ints.


HRESULT WINAPI ScriptStringGetOrder(
    SCRIPT_STRING_ANALYSIS  ssa,
    UINT                    *puOrder); 






/////   ScriptStringCPtoX
//
//      Return x coordinate for leading or trailing edge of character icp.


HRESULT WINAPI ScriptStringCPtoX(
    SCRIPT_STRING_ANALYSIS  ssa,        //In  String analysis
    int                     icp,        //In  Caret character position
    BOOL                    fTrailing,  //In  Which edge of icp
    int                    *pX);        //Out Corresponding x offset






/////   ScriptStringXtoCP
//
//


HRESULT WINAPI ScriptStringXtoCP(
    SCRIPT_STRING_ANALYSIS  ssa,            // In
    int                     iX,             // In
    int                    *piCh,           // Out
    int                    *piTrailing);    // Out






/////   ScriptStringGetLogicalWidths
//
//      Converts visual withs in psa->piAdvance into logical widths,
//      one per original character, in logical order.
//
//      Requires SSA_GLYPHS requested in original ScriptStringAnalyse call.
//
//      The piDx parameter should address a buffer containing room for
//      at least *ScriptString_pcOutChars(ssa) ints.


HRESULT WINAPI ScriptStringGetLogicalWidths(
    SCRIPT_STRING_ANALYSIS  ssa,
    int                    *piDx); 






/////   ScriptStringValidate
//
//      Scans the string analysis for invalid glyphs.
//
//      Only glyphs generated by scripts that can generate invalid glyphs
//      are scanned.
//
//      returns S_OK    - no invalid glyphs are present
//              S_FALSE - one or more invalid glyphs are present


HRESULT WINAPI ScriptStringValidate(
    SCRIPT_STRING_ANALYSIS ssa); 






/////   ScriptStringOut
//
//      Displays the string generated by a prior ScriptStringAnalyze call,
//      then optionally adds highlighting corresponding to a logical selection.
//
//      Requires SSA_GLYPHS requested in original ScriptStringAnalyse call.


HRESULT WINAPI ScriptStringOut(
    SCRIPT_STRING_ANALYSIS ssa,         //In  Analysis with glyphs
    int              iX,                //In
    int              iY,                //In
    UINT             uOptions,          //In  ExtTextOut options
    const RECT      *prc,               //In  Clipping rectangle (iff ETO_CLIPPED)
    int              iMinSel,           //In  Logical selection. Set iMinSel>=iMaxSel for no selection
    int              iMaxSel,           //In
    BOOL             fDisabled);        //In  If disabled, only the background is highlighted.






/////
//      uOptions may nclude only ETO_CLIPPED or ETO_OPAQUE.
/////   ScriptIsComplex
//
//      Determines whether a Unicode string requires complex script processing
//
//      The dwFlags parameter may include the following requests
//
#define SIC_COMPLEX     1   // Treat complex script letters as complex
#define SIC_ASCIIDIGIT  2   // Treat digits U+0030 through U+0039 as complex
#define SIC_NEUTRAL     4   // Treat neutrals as complex
//
//      SIC_COMPLEX: Should normally set. Causes complex script letters to
//      be treated as complex.
//
//      SIC_ASCIIDIGIT: Set this flag if the string would be displayed with
//      digit substitution enabled. If you are following the users NLS
//      settings using the ScriptRecordDigitSubstitution API, you can pass
//      scriptDigitSubstitute.DigitSubstitute != SCRIPT_DIGITSUBSTITUTE_NONE.
//
//      SIC_NEUTRAL: Set this flag if you may be displaying the string with
//      right-to-left reading order. When this flag is set, neutral characters
//      are considered as complex.
//
//
//      Returns S_OK     if string requires complex script processing,
//              S_FALSE  if string contains only characters laid out side by
//                       side from left to right.


HRESULT WINAPI ScriptIsComplex(
    const WCHAR     *pwcInChars,        //In  String to be tested
    int              cInChars,          //In  Length in characters
    DWORD            dwFlags);          //In  Flags (see above)






/////   ScriptRecordDigitSubstitution
//
//      Reads NLS native digit and digit substitution settings and records
//      them in the SCRIPT_DIGITSUBSTITUTE structure.
//
//
typedef struct tag_SCRIPT_DIGITSUBSTITUTE {
    DWORD  NationalDigitLanguage    :16;   // Language for native substitution
    DWORD  TraditionalDigitLanguage :16;   // Language for traditional substitution
    DWORD  DigitSubstitute          :8;    // Substitution type
    DWORD  dwReserved;                     // Reserved
} SCRIPT_DIGITSUBSTITUTE;
//
//
//p     NationalDigitLanguage: Standard digits for the selected locale as
//          defined by the countries standard setting authority.
//
//p     TraditionalDigitLangauge: Digits originally used with the locales
//          script.
//
//p     DigitSubstitute: Selects between None, Context, National and
//          Traditional. See ScriptApplyDigitSubstitution below for
//          constant definitions.
//
//      Although most complex scripts have their own associated digits, many
//      countries using those scripts use western (so called
//      'Arabic') digits as their standard. NationalDigitLanguage reflects the
//      digits used as standard, and is set from
//      the NLS data for the locale.
//      On Windows 2000 the national digit langauge can be
//      adjusted to any digit script with the control panel/regional
//      options/numbers/Standard digits listbox.
//
//      The TraditionalDigitLanguage for a locale is derived directly from the
//      script used by that locale.


HRESULT WINAPI ScriptRecordDigitSubstitution(
    LCID                     Locale,    // In   LOCALE_USER_DEFAULT or desired locale
    SCRIPT_DIGITSUBSTITUTE  *psds);     // Out  Digit substitution settings






/////
//p     Locale: NLS locale to be queried. Should usually be set to
//          LOCALE_USER_DEFAULT. Alternatively may be passed as a locale
//          combined with LOCALE_NOUSEROVERRIDE to obtain default settings
//          for a given locale. Note that context digit substitution is
//          supported only in ARABIC and FARSI locales. In other locales,
//          context digit is mapped to no substitution.
//
//p     psds: Pointer to SCRIPT_DIGITSUBSTITUTE. This structure may be passed
//          later to ScriptApplyDigitSubstitution.
//
//p     returns: E_INVALIDARG if Locale is invalid or not installed. E_POINTER
//          if psds is NULL. Otherwise S_OK.
//
//      For performance reasons, you should not call
//      ScriptRecordDigitSubstitution frequently. In particular it would be a
//      considerable overhead to call it every time you call ScriptItemize
//      or ScriptStringAnalyse.
//
//      Instead, you may choose to save the SCRIPT_DIGITSUBSTITUTE
//      structure, and update it only when you receive a
//      WM_SETTINGCHANGE message or when a RegNotifyChangeKeyValue
//      call in a dedicated thread indicates a change in the registry
//      under HKCU\Control Panel\\International.
//
//      The normal way to call this function is simply
//
//c     SCRIPT_DIGITSUBSTITUTE sds;
//c     ScriptRecordDigitSubstitution(LOCALE_USER_DEFAULT, &sds);
//
//      Then every time you itemize, you'd use the results like this:
//
//c     SCRIPT_CONTROL  sc = {0};
//c     SCRIPT_STATE    ss = {0};
//
//c     ScriptApplyDigitSubstitution(&sds, &sc, &ss);
//
//
/////   ScriptApplyDigitSubstitution
//
//      Aplies the digit substitution settings recorded in a
//      SCRIPT_DIGIT_SUBSTITUTE structure to the SCRIPT_CONTROL and
//      SCRIPT_STATE structures.
//
//      The DigitSubstitute field of the SCRIPT_DIGITSUBSTITUTE structure
//      is normally set by ScriptRecordDigitSubstitution, however it may
//      be replaced by any one of the following values:
//
//
#define SCRIPT_DIGITSUBSTITUTE_CONTEXT      0  // Substitute to match preceeding letters
#define SCRIPT_DIGITSUBSTITUTE_NONE         1  // No substitution
#define SCRIPT_DIGITSUBSTITUTE_NATIONAL     2  // Substitute with official national digits
#define SCRIPT_DIGITSUBSTITUTE_TRADITIONAL  3  // Substitute with traditional digits of the locale
//
//
//p     SCRIPT_DIGITSUBSTITUTE_CONTEXT: Digits U+0030 - U+0039 will be
//          substituted according to the language of prior letters. Before
//          any letters, digits will be substituted according to the
//          TraditionalDigitLangauge field of the SCRIPT_DIGIT_SUBSTITUTE
//          structure. This field is normally set to the primary language of
//          the Locale passed to ScriptRecordDigitSubstitution.
//
//p     SCRIPT_DIGITSUBSTITUTE_NONE: Digits will not be substituted. Unicode
//          values U+0030 to U+0039 will be displayed with Arabic (i.e.
//          Western) numerals.
//
//p     SCRIPT_DIGITSUBSTITUTE_NATIONAL: Digits U+0030 - U+0039 will be
//          substituted according to the NationalDigitLangauge field of
//          the SCRIPT_DIGIT_SUBSTITUTE structure. This field is normally
//          set to the national digits returned for the NLS LCTYPE
//          LOCALE_SNATIVEDIGITS by ScriptRecordDigitSubstitution.
//
//p     SCRIPT_DIGITSUBSTITUTE_TRADITIONAL: Digits U+0030 - U+0039 will be
//          substituted according to the TraditionalDigitLangauge field of
//          the SCRIPT_DIGIT_SUBSTITUTE structure. This field is normally
//          set to the primary language of the Locale passed to
//          ScriptRecordDigitSubstitution.


HRESULT WINAPI ScriptApplyDigitSubstitution(
    const SCRIPT_DIGITSUBSTITUTE  *psds,   // In   Digit substitution settings
    SCRIPT_CONTROL                *psc,    // Out  Script control structure
    SCRIPT_STATE                  *pss);   // Out  Script state structure






/////
//p     psds: Pointer to SCRIPT_DIGITSUBSTITUTE structure recorded earlier.
//          If NULL, ScriptApplyDigitSubstitution calls
//          ScriptRecordDigitSubstitution with LOCALE_USER_DEFAULT.
//
//p     psc: SCRIPT_CONTROL structure. The fContextDigits and uDefaultLanguage
//          fields will be updated.
//
//p     pss: SCRIPT_CONTROL structure. The fDigitSubstitute field will be
//          updated.
//
//p     returns: E_INVALIDARG if the DigitSubstitute field of the
//          SCRIPT_DIGITSUBSTITUTE structure is unrecognised, else S_OK;
#ifdef __cplusplus
}
#endif
#endif