/*--------------------------------------------------------------------------
** File: inkstore.h
**
** Copyright 1993, Slate Corporation, All Rights Reserved.
**
** This document is part of the Jot specification for the storage and
** interchange of electronic ink data.  This specification is the joint work
** of representatives of Slate Corporation, Lotus Development Corporation,
** GO, Microsoft, Apple, General Magic, and others.
**
** This document and the accompanying code samples on disk comprise Version 
** 1.0 of the Jot specification for the storage and interchange of electronic 
** ink data.  Permission is granted to incorporate and otherwise use any 
** portion of the specification.  You may make copies of the specification 
** for distribution to others, provided you include the notice "Copyright 
** 1993, Slate Corporation.  All Rights Reserved" on both the document and
** the disk label.  You may not modify this specification without written 
** permission from Slate Corporation.
**
** The specification is provided "as is" without warranty of any kind.  Slate 
** further disclaims all implied warranties of merchantability or of fitness 
** for a particular purpose.  The entire risk arising out of the use or 
** performance of the specification remains with you.
**
**--------------------------------------------------------------------------
**
** This is the main body of definitions for the ink storage specification.
** See reference section 1.0 for revision history.
**
**------------------------------------------------------------------------*/

#ifndef INKSTORE_INCLUDED
#define INKSTORE_INCLUDED




/*************************/
/* REFERENCE SECTION 0.0 */
/*************************/

/*--------------------------------------------------------------------------
** "Rationale for the ink specification"
**
** This document defines a storage and interchange format for embedded ink
** data.  The format is device- and platform-independent.  The goal is to
** provide application programs on the same and different platforms and
** operating systems a way to store and exchange ink data.  Thus a PenPoint
** user might scribble a note and send the untranslated ink as part of an
** e-mail message to a colleague's pen computer running Windows for Pen
** Computing using Magic Mail.
**
** This specification is for a publicly-defined, external format for
** electronic ink data interchange, and neither assumes nor dictates the
** nature of how the application deals with ink data internally.  The format
** is not intended to be the "internal" ink format of an application, though
** there is no reason why it could not serve such a purpose. 
**
** The scope and goals of this format design are limited to the represent-
** ation of electronic ink data embedded in some other electronic document, 
** not to the larger document itself (such as an e-mail or enhanced word-
** processing data file).
**
** The approach taken is to capture the complete user input for the
** electronic ink, including not just X/Y coordinates, but also a large set
** of current drawing attributes such as nib type and ink color.  This
** differs from other possible approaches, such as those based on certain
** recognition models for handwritten text, which require decomposing the
** handwritten ink data first into a set of pre-defined approximation curves
** or sub-strokes, and then storing a list of encodings of these sub-strokes.
** In other words, Jot preserves all information about the original input as
** opposed attempting any sort of abstract characterization of the input.
** 
** The storage format has a number of properties:
** 
** * Simple.  Typical operations on the ink data are easy.  If you only wish
**   to read stroke coordinates and bounding information from the data,
**   complex information that might be present will not hinder the process.
**   Likewise, it is easy to write out just simple information.  The
**   complex information is all optional.
** 
** * Compact.  The storage format is intended to be as compact as possible 
**   without sacrificing simplicity or fidelity.  Optional information such
**   as time stamps or color specifications occupy space only when they are
**   present.  Specifications that apply to many strokes (such as line width
**   or color) are represented just once.
** 
** * Compression.  The stroke information that describes the ink can
**   optionally be represented in a compressed format.  Compression
**   techniques include both compression and reduction of the ink data.
** 
** * Inclusive.  The format is capable of storing every property of ink
**   conceivable as of today.  
** 
** * Expandable and Compatible.  The format is expandable, so as developers
**   discover new information that should be recorded in an ink storage
**   format, these new features can be added without changing the behavior of
**   existing application programs working with an older version of the
**   format.  In general, new features can generally be ignored by
**   applications reading older versions of the format.  Likewise, new
**   application programs can handle previous versions of the format without
**   special work.
** 
** The format is not designed to easily support lots of in-memory
** manipulation of the ink data, such as deleting strokes, changing line
** widths, and so on.  A format supporting these types of manipulations would
** be at odds with the above goals.  All the information needed to perform
** these manipulations is present in this data format, so an application
** might augment this format to facilitate manipulation of the ink data.
** 
** Applications are likely to use some other format internally for real-time
** ink manipulation.  Many operating environments provide some internal means
** for storing and manipulating ink data, the details of which may be hidden
** to some extent from the application designer.  Many such real-time data
** structures store fewer types of and/or less information (such as not
** preserving information about the tablet point data rate) than are covered
** in this definition.  
**
**------------------------------------------------------------------------*/




/*************************/
/* REFERENCE SECTION 1.0 */
/*************************/

/*-------------------------------------------------------------------------
** Jot Ink Specification
** ---------------------
**
** Revision History:
**
** March    16, 1992 - First public draft.
** July     13, 1992 - Major rewrite to put data into a series of records.
** July     19, 1992 - Inclusion of ink-compacting definitions.
** July     30, 1992 - Change of target rect to offset.
** December 28, 1992 - Changes incorporated from August 1992 review meeting.
** February 12, 1993 - Incremental fixes due to coding experience.
** March    13, 1993 - Revised definition of a "group".
** April    12, 1993 - Release of version 0.99 of the specification.  Moved
**                     reference sections 28 and 29 to a separate file called
**                     sample.h
** May      01, 1993 - Release of version 1.00 of the specification.  
**                     Changed INK_OFFSET_RECORD units from twips to pen 
**                     units for consistency and ease of implementation.
**                     Fixed a typo in reference section 26.0 in the diagram.
**                     The text accompanying the diagram was correct.
**                     Fixed a typo in reference section 27.0.  The old text
**                     "delta-X == 0 or 1" was replaced with the correct text
**                     "delta-X == 2".  The accompanying diagram was correct.
**                     Removed all sizeof() constructs and replaced with 
**                     appropriate #defines to reduce compiler dependencies.
**                     Tagged all struct definitions with tag_ prefix.
**                     Added comments and reordered some existing comments.
** May      17, 1993 - Added a few more _SIZE #defines, clarified reserved
**                     values.
**
**
** GENERAL NOTES
** -------------
**
**
** Record Structure
** ----------------
**
** If not otherwise specified, all words are stored in Intel order: low-order
** word first, then high-order word, and inside of a word,  low-order byte,
** then high-order byte.  For example, a 32 bit quantity 0x12345678 would be
** written to the file as 0x78 0x56 0x34 0x12.  The notable exception is the
** storage of point data in "standard compression" format.  Sign bits are
** used to indicate item types, so the bytes are stored high-order to low-
** order (exactly opposite).  See the sample code and reference section 23.0
** for more information on the compressed format.  Uncompressed data is 
** written in Intel order.
**
** All structures are packed for the purposes of writing to a stream.
** 
** Signed integer values are two's-complement.  Rectangles are stored 
** x,y,w,h.
**
** These definitions are intended to insulate the sample ink compaction and
** storage code from any possible variation in item alignment or structure
** packing across architectures.  The only possible area of portability
** concern lies in the use of unions in colors (see 11.0) and pen tips (see
** 14.0).
**
** Any use of units of mass to denote units of force ("grams of force"), or
** similar common misuses of physical units, are noted here with an apology
** to any purists, and should be interpreted in the common way by assuming
** one standard gravity.
**
** Record Sequence
** ---------------
**
** In this document, one piece of ink data is called an ink bundle. 
** Typically this might correspond to the strokes that make up the ink from
** the time when the pen touches down until the user finishes writing
** (usually determined by a timeout or the pen leaving proximity).  Thus an
** ink bundle usually contains many ink strokes, and the strokes do not have
** to describe a continuous line of ink.
** 
** As stated in reference section 5.0, all data conforming to this
** specification appears as a stream of ink bundles each of which must begin
** with an INK_BUNDLE_RECORD and end with an INK_END_RECORD.  There may be
** more than one INK_BUNDLE_RECORD/INK_END_RECORD pair in a given stream.
** A record stream might look something like this:
**
** INK_BUNDLE_RECORD       required    // for bundle number one
** INK_SCALE_RECORD        optional    // sets the scale for rendering
** INK_OFFSET_RECORD       optional    // sets the offset for rendering
** INK_COLOR_RECORD        optional    // sets the color for rendering
** INK_START_TIME_RECORD   optional    // sets the relative start time
** INK_PENTIP_RECORD       optional    // sets the pentip for rendering
** INK_GROUP_RECORD        optional    // tags the following PENDATA
** INK_PENDATA_RECORD      recommended // actual points
** INK_GROUP_RECORD        optional    // tags the following PENDATA
** INK_PENDATA_RECORD      recommended // actual points 
** INK_PENDATA_RECORD      recommended // more points in same group
** INK_SCALE_RESET_RECORD  optional    // resets to default scaling/offset
** INK_PENDATA_RECORD      recommended // actual points
** INK_END_TIME_RECORD     optional    // relative time inking ended
** INK_END_RECORD          required    // end of bundle number one
**
** It is perfectly reasonable to write out only the following (though doing
** so will cause the ink to be rendered in a completely default manner -- 
** black hairline width at 1:1 scaling with offset 0):
**
** INK_BUNDLE_RECORD
** INK_PENDATA_RECORD
** INK_END_RECORD
** 
**
** Specification Revisions
** -----------------------
** 
** Future enhancements to this specification may modify certain record types.
** It is guaranteed that any record modified in a subsequent revision of the 
** specification will be a strict superset of that record's definition in any
** previous revision of the specification.  That is, modified record types 
** will only be lengthened, not shortened.  If a particular record type must
** be extended such that it would not be a superset of the original, a new 
** record type would be added to cover that particular extension.
** 
** This extension strategy has two important ramifications: 
**
** 1) A reading application should *ALWAYS* use the size of a record as 
**    recorded in the record structure itself (i.e., the recordLength field
**    of the INK_RECORD_HEADERx structure) rather than the sizeof() or any
**    other size determined at compile time to determine how may bytes to 
**    read as the data structures are parsed.  This is due to the fact that 
**    a record may grow in a future revision of the standard.  The only
**    exception to this rule is the INK_BUNDLE_RECORD which contains a
**    version number that will be modified with each change to that record.
**    If an INK_BUNDLE_RECORD is encountered and its version matches the
**    version used at compile time, the size of the record should exactly
**    match the #define of inkRecordBundleSize.
**
** 2) Any particular record may be read into a target data structure up to 
**    the size of the target data structure and the rest may be ignored.  
**    This is due to the 'strict superset' rule which means that any 
**    extension of any record type must leave the meaning, content, and size
**    of any existing fields as is.  So, for example, if an INK_SCALE_RECORD
**    was modified by adding 2 bytes, the reading application can safely read
**    the data into the INK_SCALE_RECORD known at compile time and throw
**    away the extra two bytes: the header, x, and y will be in the same 
**    place and will have the same meaning.
**
** 
** Files of Ink
** ------------
**
** It is a recommended practice on DOS and UNIX style file systems to use the
** extension ".JOT" for files consisting solely of ink recorded according to
** this specification.  The specification is designed such that ink data can
** be embedded inside any file format and if such a file contains more than
** strictly ink data, it should not use the .JOT extension.
**
**------------------------------------------------------------------------*/




/*************************/
/* REFERENCE SECTION 2.0 */
/*************************/

/*-------------------------------------------------------------------------
** Definitions used in this header.
**
** These definitions must be defined appropriately to the target environment
** and compiler.
**
** For example, on some compilers for environments using segmented addressing
** and 64K segments sizes, the correct definition of FAR would be "_huge",
** rather than "_far", because the objects pointed to may be larger than 64K.
**
** In particular, check the definitions of FAR, U32, and S32 for
** compatibility with your compiler, environment, and memory model.
**
**------------------------------------------------------------------------*/

#ifndef FAR
#define FAR
#endif

/* useful constants */

#define flag0           (0x0001)
#define flag1           (0x0002)
#define flag2           (0x0004)
#define flag3           (0x0008)
#define flag4           (0x0010)
#define flag5           (0x0020)
#define flag6           (0x0040)
#define flag7           (0x0080)
#define flag8           (0x0100)
#define flag9           (0x0200)
#define flag10          (0x0400)
#define flag11          (0x0800)
#define flag12          (0x1000)
#define flag13          (0x2000)
#define flag14          (0x4000)
#define flag15          (0x8000)

#define flag16          (0x00010000L)
#define flag17          (0x00020000L)
#define flag18          (0x00040000L)
#define flag19          (0x00080000L)
#define flag20          (0x00100000L)
#define flag21          (0x00200000L)
#define flag22          (0x00400000L)
#define flag23          (0x00800000L)
#define flag24          (0x01000000L)
#define flag25          (0x02000000L)
#define flag26          (0x04000000L)
#define flag27          (0x08000000L)
#define flag28          (0x10000000L)
#define flag29          (0x20000000L)
#define flag30          (0x40000000L)
#define flag31          (0x80000000L)

#define TRUE    1
#define FALSE   0

/* void pointers */
typedef void        FAR *P_UNKNOWN;
typedef P_UNKNOWN   FAR *PP_UNKNOWN;

#define pNull       ((P_UNKNOWN)0)

/*  Unsigned integers */
typedef unsigned char   U8,  FAR *P_U8;
#define U8_SIZE         1
typedef unsigned short  U16, FAR *P_U16;
#define U16_SIZE        2
typedef unsigned long   U32, FAR *P_U32;
#define U32_SIZE        4

/*  Signed integers */
typedef char     S8,  FAR *P_S8;
#define S8_SIZE         1
typedef short    S16, FAR *P_S16;
#define S16_SIZE        2
typedef long     S32, FAR *P_S32;
#define S32_SIZE        4
/*
typedef signed char     S8,  FAR *P_S8;
#define S8_SIZE         1
typedef signed short    S16, FAR *P_S16;
#define S16_SIZE        2
typedef signed long     S32, FAR *P_S32;
#define S32_SIZE        4
*/

/* geometry structures */
typedef struct tag_XY32 {
    S32     x;
    S32     y;
} XY32, FAR *P_XY32;
#define XY32_SIZE (S32_SIZE+S32_SIZE)

typedef struct tag_XY16 {
    S16     x;
    S16     y;
} XY16, FAR *P_XY16;

/*-------------------------------------------------------------------------
** Note: 
** Angles from vertical can exceed +-90 degrees: in this case, the "back" end
** of the stylus is nearer the tablet surface than the "front" end.
**-------------------------------------------------------------------------*/

/*-------------------------------------------------------------------------
** Note: 
** Standard compaction will normally store angles in nibbles, or single 
** bytes, rather than in four-byte records.
**-------------------------------------------------------------------------*/

typedef struct tag_ANGLE16 {
    S16     theta;  /* "X" angle of the stylus, degrees from vertical, */
                    /* increasing in the positive "X" direction. */
    S16     phi;    /* "Y" angle of the stylus. */
} ANGLE16, FAR *P_ANGLE16;
#define ANGLE16_SIZE (S16_SIZE+S16_SIZE)

typedef struct tag_SIZE32 {
    S32     w;
    S32     h;
} SIZE32, FAR *P_SIZE32;
#define SIZE32_SIZE (S32_SIZE+S32_SIZE)

typedef struct tag_SIZE16 {
    S16     w;
    S16     h;
} SIZE16, FAR *P_SIZE16;
#define SIZE16_SIZE (S16_SIZE+S16_SIZE)

/*-------------------------------------------------------------------------
** Note:
** A rect where xmin==xmax and ymin==ymax has a size of zero:
** size.w and size.h are both zero.
**-------------------------------------------------------------------------*/

typedef struct tag_RECT32 {
    XY32    origin;
    SIZE32  size;
} RECT32, FAR *P_RECT32;
#define RECT32_SIZE (XY32_SIZE+SIZE32_SIZE)

typedef U32 FIXED_FRACTION; /* fixed point value, unity = 0x00010000 */
#define FIXED_FRACTION_SIZE U32_SIZE

#define INK_UNITY_SCALE ((U32) 0x00010000L)




/*************************/
/* REFERENCE SECTION 3.0 */
/*************************/

/*-------------------------------------------------------------------------
** A block of ink data is called an ink bundle.  Each ink bundle consists of
** a series of n records.  Each record has a common header that indicates the
** record type and the record length.  An ink bundle always starts with an
** INK_BUNDLE_RECORD and always ends with an INK_END_RECORD.
**
** Any records of unknown type can be skipped by simply reading the length of
** the record.
**
** Note:
** Within an ink bundle, time increases.  This implies a drawing order of
** back-to-front.  Between adjacent sequential bundles, the implicit drawing
** is also back-to-front.
**
** A number of record types are defined.  The most common is the
** inkRecordPenData which contains the actual pen data.  Other records are
** mostly attributes of the pen data and are optional.  They will, in
** general, only be present when a given attribute changes to something
** different than the default value for that attribute.
**
** In order to have the most compact format and also allow large records,
** several different record headers are defined, each with a different
** length.
**
** The top two bits of the record type indicate what kind of record length
** follows:
**
** The record length can be:
**
**  -   non-existent (the entire record consists of just the recordType)
**  -   An 8 bit length (one byte) for records up to 255 bytes
**  -   A 16 bit length (two bytes) for records up to 64k
**  -   A 32 bit length (four bytes) for really big records
**
**------------------------------------------------------------------------*/

#define inkRecordNoLength                  0    /* no length, just recordType */
#define inkRecordLength8              flag14    /* 8 bit length */
#define inkRecordLength16    flag15             /* 16 bit length */
#define inkRecordLength32   (flag15 | flag14)   /* 32 bit length */

/* useful defines for isolating or clearing the length type bits */
#define inkRecordLengthMask (flag15 | flag14)   /* mask for length bits */
#define inkRecordLengthClearMask (~inkRecordLengthMask)

/* some useful macros for declaring the various types of record types */
#define MakeRec0(recType)   (recType | inkRecordNoLength)   /* no rec length */
#define MakeRec8(recType)   (recType | inkRecordLength8)    /* 8 bit length */
#define MakeRec16(recType)  (recType | inkRecordLength16)   /* 16 bit length */
#define MakeRec32(recType)  (recType | inkRecordLength32)   /* 32 bit length */

typedef U16 INK_RECORD_TYPE, FAR *P_INK_RECORD_TYPE;
#define INK_RECORD_TYPE_SIZE U16_SIZE

#define inkRecordHeaderLength(record_type)  \
    (   (((record_type) & inkRecordLength32) == inkRecordNoLength) ?\
                    INK_RECORD_TYPE_SIZE                            \
     :  (((record_type) & inkRecordLength32) == inkRecordLength8)  ?\
                    INK_RECORD_TYPE_SIZE+U8_SIZE                    \
     :  (((record_type) & inkRecordLength32) == inkRecordLength16) ?\
                    INK_RECORD_TYPE_SIZE+U16_SIZE                   \
     :              INK_RECORD_TYPE_SIZE+U32_SIZE                   \
    )

/*-------------------------------------------------------------------------
** Note: most compilers will not generate code for the above macro but will 
** determine the proper value at compile time.
**-------------------------------------------------------------------------*/


/*************************/
/* REFERENCE SECTION 4.0 */
/*************************/

/*-------------------------------------------------------------------------
** These are all the currently defined record types.  The macro MakeRecX()
** encodes the right bits in with the record Id to define its recordLength.
**
** For simplicity, recType values may not be repeated for different
** INK_RECORD_TYPEs.  Use of a record type defined as MakeRec32(63) thus
** forbids the use of a record type defined as MakeRec16(63), MakeRec8(63),
** or MakeRec0(63).
**
** Record type 63 is reserved explicitly for possible future extension beyond
** 63 record types.
**
**------------------------------------------------------------------------*/

#define inkRecordEnd             MakeRec0(  0) /* end of bundle */
#define inkRecordBundle          MakeRec8(  1)
#define inkRecordPenData         MakeRec32( 2)
#define inkRecordScale           MakeRec8(  3)
#define inkRecordScaleReset      MakeRec0(  4)
#define inkRecordColor           MakeRec8(  5)
#define inkRecordTip             MakeRec8(  6)
#define inkRecordGroup           MakeRec8(  7)
#define inkRecordOffset          MakeRec8(  8)
#define inkRecordStartTime       MakeRec8(  9)
#define inkRecordEndTime         MakeRec8( 10)
#define inkRecordPointsPerSecond MakeRec8( 11)
#define inkRecordUnitsPerZ       MakeRec8( 12)
#define inkRecordUnitsPerForce   MakeRec8( 13)

/* Record types 14 .. 61 are reserved for future definition. */

#define inkRecordApp             MakeRec32(62) /* application-specific records*/
#define inkRecordExt             MakeRec32(63) /* reserved for extension */

/*-------------------------------------------------------------------------
** Every record starts with a header that contains the recordType and the
** recordLength.  The recordType indicates the type of data here.  The
** recordLength indicates the total length of all the data for the record
** (including the size of the header).
**-------------------------------------------------------------------------*/

/* no recordLength */
typedef struct tag_INK_RECORD_HEADER0 {
    INK_RECORD_TYPE     recordType;
} INK_RECORD_HEADER0, FAR *P_INK_RECORD_HEADER0;

/* 8 bit recordLength */
typedef struct tag_INK_RECORD_HEADER8 {
    INK_RECORD_TYPE     recordType;
    U8                  recordLength;
} INK_RECORD_HEADER8, FAR *P_INK_RECORD_HEADER8;

/* 16 bit recordLength */
typedef struct tag_INK_RECORD_HEADER16 {
    INK_RECORD_TYPE     recordType;
    U16                 recordLength;
} INK_RECORD_HEADER16, FAR *P_INK_RECORD_HEADER16;

/* 32 bit recordLength */
typedef struct tag_INK_RECORD_HEADER32 {
    INK_RECORD_TYPE     recordType;
    U32                 recordLength;
} INK_RECORD_HEADER32, FAR *P_INK_RECORD_HEADER32;




/*************************/
/* REFERENCE SECTION 5.0 */
/*************************/

/*-------------------------------------------------------------------------
** A bundle of ink consists of an INK_BUNDLE_RECORD, a series of records,
** terminated with an INK_END_RECORD.
**
** An INK_BUNDLE_RECORD, along with a matching INK_END_RECORD, are the
** mandatory records in the format.  The ink data must start with an
** INK_BUNDLE_RECORD.
**
** It is suggested that anyone reading this format do a number of validity
** checks on the first record in any ink data.  The first record should meet
** the following minimum requirements:
**
** 1)   header.recordType == INK_RECORD_BUNDLE
** 2)   header.recordLength >= inkRecordBundleSize (See general notes in 
**      reference section 1.0 for important information about record sizes.)
** 3)   compactionType is an expected and supported value
** 4)   penUnitsPerX and penUnitsPerY seem reasonable and expected:
**      greater than, say, 1000 units per meter (25.4/inch), less than, say,
**      400,000 (~10,000 units per inch)
**
**------------------------------------------------------------------------*/

typedef struct tag_INK_END_RECORD {
    INK_RECORD_HEADER0  header;         /* value is inkRecordEnd */
} INK_END_RECORD, FAR *P_INK_END_RECORD;
#define inkRecordEndSize     (inkRecordHeaderLength(inkRecordEnd))




/*************************/
/* REFERENCE SECTION 6.0 */
/*************************/

/*-------------------------------------------------------------------------
** The terms compression and compaction are used somewhat interchangeably
** in this specification but they actually have slightly different meanings
** and are both supported to a certain extent by Jot.
**
** Compression refers to a technique of encoding data such that the resuling
** data, while smaller, is still whole.  That is, compression under Jot is
** loss-less.  Compaction refers to a process where certain pieces of less
** important data are actually omitted from the stream and are possibly
** reconstructed by the reader of the data.  
**
** Using Jot, a writing application may choose to compress only, compact only
** or use some combination.  The standard compression mechanism defined here
** and implemented in the sample code supports both notions.
**
**------------------------------------------------------------------------*/

typedef U8 INK_COMPACTION_TYPE, FAR *P_INK_COMPACTION_TYPE;
#define INK_COMPACTION_TYPE_SIZE U8_SIZE
#define inkNoCompression            (0)
#define inkStdCompression           (1)

/*-------------------------------------------------------------------------
** Other compression schemes may be adopted in future revisions of this
** specification.
**-------------------------------------------------------------------------*/




/*************************/
/* REFERENCE SECTION 7.0 */
/*************************/

/*-------------------------------------------------------------------------
** The INK_BUNDLE_FLAGS contain some flags that apply to an entire bundle. 
** If you wanted to store several pieces of ink that had different
** INK_BUNDLE_FLAGS, you would do it by storing several different bundles.
**
** Advisory flags:
**
** inkPointsRemoved         
**      Indicates whether all original points are still present or whether
**      some points were removed to save space.  For applications that are
**      only interested in the visual aspects of ink, many points can be
**      removed that do not affect the appearance (i.e. duplicate points,
**      collinear points, points which deviate less than some screen
**      resolution, etc..).  Some other types of applications must know that
**      points are present at some consistent sampling rate (i.e. some forms
**      of handwriting translation).  This flag indicates whether all
**      original points are still there.
**
**      Note:
**      The purpose of "inkPointsRemoved" is to indicate that the timing
**      information cannot be accurately derived by counting points:
**      replacing individual points with an "elided point" item does not
**      constitute removing points. ("Elided" means omitted or skipped).
**
** inkProxDataRemoved   
**      Indicates that the original points between strokes (proximity) were
**      removed to save space.  An out-of-prox point should be stored between
**      strokes to delimit them.  Some applications depend on knowing the
**      time between strokes or at the ends of strokes for certain
**      functions. 
**
**      Note:
**      "Proximity" is defined as the stylus being close enough to the tablet
**      for the tablet to report the stylus position, although perhaps at
**      lower accuracy and perhaps at a lower number of points per second.  A
**      recommended practice is to include "out of proximity" points in the
**      recorded ink data when they are used as part of determining the
**      amount of time a stylus was out of contact with the tablet, or for
**      triggering the completion of an action such as a "gesture". 
**
** inkStrokeLimitsPresent   
**      Indicates that INK_BUTTONS items are also present, and that they
**      indicate what the storing app decided the stroke start/end points
**      were. (Note: the reading application may otherwise use a different
**      algorithm for using tip force values to delimit strokes.) 
**
**      Note:
**      If inkStrokeLimitsPresent is set, then inkButtonDataPresent must also
**      be set. 
**
** Data flags:
**
**  inkAngleDataPresent     indicates angle data is present.
**  inkForceDataPresent     indicates force data is present.
**  inkProxDataPresent      indicates points are present when pen is lifted
**                          up (i.e. the force drops below some threshold).
**  inkRotationDataPresent  indicates pen rotation data is present.
**  inkHeightDataPresent    indicates pen height data is present.
**  inkButtonDataPresent    indicates "button state" information is present.
**  inkPreMultiplyScale     indicates that scaling should be applied before
**                          the offset value is added ("pre-multiply")
**                          rather than after ("post-multiply")
**
** Note: 
** A previous draft version included a provision for compacting data to an
** approximation based on Bezier curves.  Initial results did not show
** promise in terms of efficiency and performance.
**
** "inkBezierRepresentation" would have indicated that the X/Y ordinates
** reflected a Bezier approximation to the original tablet data.  This would
** have meant that the ordinate data represented aggregates of anchor points
** and control points for each piece wise approximation, and therefore could
** not be used directly to render the data.  The definition of these anchor
** and control points, and the example code for the approximation and
** regeneration of the "true" coordinates could not be worked out at this
** time. 
**
** Some standard values for pen units per meter follow:
**
** 1000 points per inch digitizer == 39370 pen units per meter
**  500 points per inch digitizer == 19685 pen units per meter
**  200 points per inch digitizer ==  7874 pen units per meter
**  254 points per inch (1/10 mm) == 10000 pen units per meter
**
** 1000 pen units per meter is a reasonable minimum; 400,000 is a reasonable
** maximum value.
**
** The specific format for each of these types of data is described in the
** INK_PENDATA_RECORD documentation (reference section 8.0).
** 
** Note:
** The order in which these flags are defined has nothing to do with the
** order in which the data appears in the INK_POINT structure when reading
** or writing point data.  For more information, see reference section 21.0.
**
**------------------------------------------------------------------------*/

typedef U16 INK_BUNDLE_FLAGS, FAR *P_INK_BUNDLE_FLAGS;
#define INK_BUNDLE_FLAGS_SIZE U16_SIZE
#define inkPointsRemoved            (flag0)
#define inkProxDataRemoved          (flag1)
#define inkAngleDataPresent         (flag2)
#define inkForceDataPresent         (flag3)
#define inkRotationDataPresent      (flag4)
#define inkHeightDataPresent        (flag5)
#define inkButtonDataPresent        (flag6)
#define inkStrokeLimitsPresent      (flag7)
#define inkPreMultiplyScale         (flag8)

/*-------------------------------------------------------------------------
** Reserved: flag9, flag10, flag11, flag12, flag13, flag14, flag15.
** More flags beyond flag15 can be added in a new record type
** in a later revision to this specification.
**-------------------------------------------------------------------------*/


typedef struct tag_INK_BUNDLE_RECORD {
    INK_RECORD_HEADER8      header;         /* value is inkRecordBundle */
    U8                      version;        /* Value for release 1.0 is 1 */
    INK_COMPACTION_TYPE     compactionType;
    INK_BUNDLE_FLAGS        flags;          /* flags for the whole bundle */
    U32                     penUnitsPerX;   /* pen units per meter (x dir) */
    U32                     penUnitsPerY;   /* pen units per meter (y dir) */
} INK_BUNDLE_RECORD, FAR *P_INK_BUNDLE_RECORD;

#define inkPointDefaultVersion          (1)
#define inkPointDefaultCompactionType   (inkStdCompression)
#define inkPointDefaultBundleFlags      (0)
#define inkPointDefaultPenUnitsPerX     (1000)
#define inkPointDefaultPenUnitsPerY     (1000)

#define inkRecordBundleSize \
    (inkRecordHeaderLength(inkRecordBundle) +  U8_SIZE + \
     INK_COMPACTION_TYPE_SIZE + INK_BUNDLE_FLAGS_SIZE +       \
     U32_SIZE + U32_SIZE)




/*************************/
/* REFERENCE SECTION 8.0 */
/*************************/

/*-------------------------------------------------------------------------
** A penData record contains the actual pen data for one or more pen strokes.
** The bounds applies to all the strokes contained within this record. 
** Multiple strokes are typically grouped into one record to increase the
** efficiency of the compression algorithm, though strokes may be stored
** individually, if desired.
**
** The bounds is the pure mathematical bounds of the raw pen points and does
** not take into account any rendering information such as the pen tip or the
** line width.  All points in the INK_PENDATA have been normalized relative
** to the lower bounds in the INK_PENDATA header.
**
** Some applications will prefer to know the bounds of individual strokes.
** This can be accomplished in two ways.
**
** 1) The bounds for a given stroke can be computed when reading the file
**    by decompressing an INK_PENDATA_RECORD into its strokes and then
**    traversing the points in each stroke to build the bounds for each
**    stroke.
**
** 2) An application can decide to store only one stroke per
**    INK_PENDATA_RECORD (and thus the bounds of the PENDATA_RECORD is
**    already the bounds of one stroke).  The sacrifice here is in
**    compression efficiency and the need to still support reading files
**    written by other applications that might group multiple strokes
**    into a single INK_PENDATA_RECORD.
**
** Note: 
** In practice, our experience is that unpacking the data in order to compute
** the bounds for each stroke to check for strokes that intrude into a given
** region is not an excessive burden.  The checks that would have been done
** on the bounds of each stroke can be done on the builds for each penData
** group, and not all strokes must be checked individually.
**
** The format of the pen data is determined by the settings for
** compactionType and flags in the INK_BUNDLE_RECORD structure, and
** is described later in this file.  Two formats are currently defined:
** an uncompacted format and a delta-encoded compacted format, both with
** optional components present or absent depending on the state of the flags
** in the INK_BUNDLE_RECORD.
**
**------------------------------------------------------------------------*/

typedef struct tag_INK_PENDATA_RECORD {
    INK_RECORD_HEADER32 header;     /* value is inkRecordPenData */
    RECT32              bounds;
    U8                  inkData[1]; /* ink data goes here: definitions */
                                    /* follow later in this file. */
} INK_PENDATA_RECORD, FAR *P_INK_PENDATA_RECORD;
#define inkRecordPenDataSize(data_length)   \
   (inkRecordHeaderLength(inkRecordPenData) +  RECT32_SIZE + (data_length))




/*************************/
/* REFERENCE SECTION 9.0 */
/*************************/

/*-------------------------------------------------------------------------
** Ink scale is recorded in two fixed point values.  A unity scale (scale
** of one) is represented as 0x00010000, a scale of 0.5 as 0x00008000.
**
** Note:
** All ink is located relative to the lower-left (0,0) corner of a logical 
** page or window.  Scale and offset operations are cumulative, much in the 
** same way as in PostScript.  One begins with a normalized graphics state
** and sequentially applies the scale and offset operations to that matrix. 
** The INK_SCALE_RESET record returns the graphics state to its default state
** (i.e., the transformation matrix is set to an identity matrix and the
** offset is reset to the default of 0).  By default, scaling is applied
** after adding in any offset specified in an INK_OFFSET_RECORD.  If the ink
** bundle has the inkPreMultiplyScale bit set, for all ink in that bundle
** scaling is applied before adding in any offset.
** 
** As used in this format, ink scale and offset values are set by the storing
** application, to be applied by the rendering application.  If the storing
** application collected the ink at scales of (2.0,2.0), the storing
** application should insert an INK_SCALE_RECORD with a scale of (0.5,0.5)
** for the rendering application to multiply all ink X and Y coordinates by.
**
** It is the responsibility of the storing application to deal with any 
** effects from round-off or truncation error due to the limits of precision
** in the FIXED_FRACTION values used in INK_SCALE_RECORDs.
**
** An ink scale record indicates a scale change that stays in effect until
** another ink scale record is encountered.  Ink scale values compound: if
** the current scale is (2.0,2.0) and an INK_SCALE_RECORD is encountered with
** scale of (2.0,3.0), the scale to be applied to ink then becomes(4.0,6.0). 
** In absence of any ink scale record, the default ink scale is unity.  In
** general, a typical usage pattern for an application that supports drawing
** ink while zoomed at scale is to record a number of strokes at a given
** scale, reset the scale with an INK_SCALE_RESET_RECORD (which resets both
** the scale and the offset to the default values), then switch to another
** scale, then record a number more strokes, and so on.
**
** Note:
** The extension scaling and offset to the Z ordinate value is not defined in
** this version of the specification.  The extension to Z scaling and offset
** in a "standard" record type (i.e. not an application-specific record) may
** be addressed in the future.
**
**------------------------------------------------------------------------*/

typedef struct tag_INK_SCALE {
    FIXED_FRACTION      x;      /* scale in the x direction */
    FIXED_FRACTION      y;      /* scale in the y direction */
} INK_SCALE, FAR *P_INK_SCALE;
#define INK_SCALE_SIZE (FIXED_FRACTION_SIZE+FIXED_FRACTION_SIZE)

#define inkPointDefaultScale    (INK_UNITY_SCALE)       /* Unity. */

typedef struct tag_INK_SCALE_RECORD {
    INK_RECORD_HEADER8  header;         /* value is inkRecordScale */
    INK_SCALE           scale;
} INK_SCALE_RECORD, FAR *P_INK_SCALE_RECORD;
#define inkRecordScaleSize  \
       (inkRecordHeaderLength(inkRecordScale) + \
        FIXED_FRACTION_SIZE + FIXED_FRACTION_SIZE)




/**************************/
/* REFERENCE SECTION 10.0 */
/**************************/

/*-------------------------------------------------------------------------
** The offset position is used to relocate ink data, after scaling.  For
** example, in a forms application, ink in a sketch field is drawn relative
** to a given sketch field in the form.  The location of this original field
** is important to know so we know how the ink in this bundle relates to its
** original field.  If we wanted to move this ink to another field (i.e.
** cut/paste or move), we would need to know the location of the original
** field so we could render the ink in the new field in a manner consistent
** with how it was drawn relative to its original field (i.e. a similar
** baseline for a hand-written signature).
**
** This record is optional.  If it exists, it will then apply to all
** following pen data in the file.  If it is not present it is assumed that
** no information of this type is relevant.  For example, while field ink
** would have an offset position, markup ink over an entire form would not
** have a offset position (or would have an offset position of (0,0) and a
** scale of (1,1)) because it is relative to the entire form coordinate
** system, not relative to some piece in the form. 
**
** Note:
** This approach allows a reader to "blindly" apply the scale and offset 
** values specified to ink data, and puts the burden for computing
** compounding of multiple zoom levels, etc., on the writing application.
**
**------------------------------------------------------------------------*/

typedef struct tag_INK_OFFSET_RECORD {
    INK_RECORD_HEADER8  header;         /* value is inkRecordOffset */
    XY32                positionOffset; /* values are in pen units */
} INK_OFFSET_RECORD, FAR *P_INK_OFFSET_RECORD;

#define inkRecordOffsetSize \
       (inkRecordHeaderLength(inkRecordOffset) + XY32_SIZE)

#define inkPointDefaultOffset   ((S32) 0)   /* No offset. */


typedef struct tag_INK_SCALE_RESET_RECORD {
    INK_RECORD_HEADER0  header;         /* value is inkRecordScaleReset */
} INK_SCALE_RESET_RECORD, FAR *P_INK_SCALE_RESET_RECORD;

#define inkRecordScaleResetSize  (inkRecordHeaderLength(inkRecordScaleReset))




/**************************/
/* REFERENCE SECTION 11.0 */
/**************************/

/*-------------------------------------------------------------------------
** Ink color is represented as an rgb value, plus opacity.
**
** The default color is black (r,g,b,o) = (0,0,0,255).  A color change
** present in the file remains in effect until another color change.
** Typically, the color will stay the same for many ink strokes and thus
** a color record will only be used occasionally when the color changes.
**
** "Opacity" is rather vaguely understood, especially in color environments.
** In this context, opacity means the degree to which the display underneath
** the ink shows through.  An opacity value of 255 means that nothing under
** the ink shows through; 0 means that everything shows through (the ink
** is transparent).  It is up to the reading application to define the
** implementation of opacity on the reading platform.
**
** The color/opacity value of (255,255,255,0), or "transparent white" is
** defined as an "erase" color.  In inking applications that support a true
** "erase" function, such as the ability to erase annotation ink on an
** "original" document (perhaps a FAX image) the "erase" color restores the
** background image where painted.  The "background image" is defined by the
** rendering application.  
**
** Applications which do not support a true "erase" function may interpret
** this as some other drawing function, such as drawing the "background"      
** color.
**
**------------------------------------------------------------------------*/

typedef union {
  U32    all;
  struct {
    U8   red,
         green,
         blue,
         opacity;   /* opaqueness: see defines below */
  } rgb;
} INK_COLOR, FAR *P_INK_COLOR;
#define INK_COLOR_SIZE (U32_SIZE)

typedef struct tag_INK_COLOR_RECORD {
    INK_RECORD_HEADER8  header;         /* value is inkRecordColor */
    INK_COLOR           color;
} INK_COLOR_RECORD, FAR *P_INK_COLOR_RECORD;
#define inkRecordColorSize  \
       (inkRecordHeaderLength(inkRecordColor) + U32_SIZE)

/*-------------------------------------------------------------------------
** Standardized opacity values:
** A recommended practice is that an opacity value of 128 (midway between
** 0 and 255) be used for "highlighter" colors.  A recommended practice is 
** that grey values as defined below be used for "standard grey" 
** highlighters.
**-------------------------------------------------------------------------*/

#define inkOpacityTransparent   0x00
#define inkOpacityHighlight     0x80
#define inkOpacityOpaque        0xFF

/* Standard solid colors: */

#define inkColorErase           {0xFF,0xFF,0xFF,0x00}
#define inkColorWhite           {0xFF,0xFF,0xFF,0xFF}
#define inkColorLtGrey          {0x80,0x80,0x80,0xFF}
#define inkColorDkGrey          {0x40,0x40,0x40,0xFF}
#define inkColorBlack           {0x00,0x00,0x00,0xFF}

/* Standard highlighter (transparent) colors: */

#define inkColorLtGreyHighlight {0x80,0x80,0x80,0x80}
#define inkColorDkGreyHighlight {0x40,0x40,0x40,0x80}

#define inkDefaultColor ((INK_COLOR) inkColorBlack)




/**************************/
/* REFERENCE SECTION 12.0 */
/**************************/

/*-------------------------------------------------------------------------
** Time is measured in milliseconds.
** 
** Note:
** Because of the difficulty synchronizing clocks on different machines
** at the time granularity of digitizing tablets, and because the "editing"
** of ink at a later time makes the definition of the absolute time for each
** ink point ambiguous, the base for the time is arbitrary.  All times in
** strokes are just relative to each other with no absolute time
** relationship.
**
** These records, when encountered in the file, apply to the next stroke data
** in the file (so this record comes before the penData that it applies to).
** End time records are not required.  The interpretation of an end time
** which is in conflict with the end time inferred from the assumed data rate
** of points and the number of points (including elided points) is not
** defined.  
**
** Start time is the time for the first point in the following penData record
** and end time is the time of the last point in the following penData
** record, because if you are recording tip force, the exact definition of
** pen up and pen down may be fuzzy and/or application dependent.  
**
**------------------------------------------------------------------------*/

typedef U32 INK_TIME, FAR *P_INK_TIME;      /* milliseconds */
#define INK_TIME_SIZE U32_SIZE

#define inkDefaultTime      ((INK_TIME) 0)

typedef struct tag_INK_START_TIME_RECORD {
    INK_RECORD_HEADER8  header;         /* value is inkRecordStartTime */
    INK_TIME            startTime;
} INK_START_TIME_RECORD, FAR *P_INK_START_TIME_RECORD;
#define inkRecordStartTimeSize  \
       (inkRecordHeaderLength(inkRecordStartTime) + INK_TIME_SIZE)

typedef struct tag_INK_END_TIME_RECORD {
    INK_RECORD_HEADER8  header;         /* value is inkRecordEndTime */
    INK_TIME            endTime;
} INK_END_TIME_RECORD, FAR *P_INK_END_TIME_RECORD;
#define inkRecordEndTimeSize    \
       (inkRecordHeaderLength(inkRecordEndTime) + INK_TIME_SIZE)




/**************************/
/* REFERENCE SECTION 13.0 */
/**************************/

/*-------------------------------------------------------------------------
** INK_PENDATA_RECORDs can be grouped.  If they are grouped, each
** INK_PENDATA_RECORD can be assigned a group number.  All
** INK_PENDATA_RECORDs with the same group number belong to the same group. 
**
** The exact interpretation of grouping is up the applications involved.
** Writing applications may group ink data, but not all reading applications
** that read the data may interpret grouping in the same way.
**
** For example, grouping could be used in the traditional fashion as in
** drawing programs so the user moves or copies an entire group of
** INK_PENDATA_RECORDs together.  A group could also be used to signify a
** series of INK_PENDATA_RECORDs entered by the user all within some criteria
** (i.e. all during one proximity session or all within some time frame).
**
** Group numbers are simply signed 16 bit values and can be anything.  They
** do not need to be contiguous (i.e. they do not need to be 0,1,2).  They
** can be 12,49,-12345 if that is useful.
**
** This record can also be used as a simple marker for starting a new group
** when the groupId is not really used: Group numbers of 0,0,0,0 ... are
** thus permitted.
**
** INK_GROUPs are nestable.  Group 0 is reserved as the end-of-group marker
** for disjoint groups.  If no end-of-group marker is encountered before the
** end of the file or the end of all ink data (as indicated by an
** INK_END_RECORD), all current (and possibly nested) groups are terminated
** as if end-of-groups markers for them had been encountered.
**
**------------------------------------------------------------------------*/

typedef S32 INK_GROUP, FAR *P_INK_GROUP;
#define INK_GROUP_SIZE S32_SIZE

typedef struct tag_INK_GROUP_RECORD {
    INK_RECORD_HEADER8  header;         /* value is inkRecordGroup */
    INK_GROUP           groupId;    /* application-specific interpretation */
} INK_GROUP_RECORD, FAR *P_INK_GROUP_RECORD;

#define inkDefaultGroup ((INK_GROUP) 0)

#define inkRecordGroupSize  \
       (inkRecordHeaderLength(inkRecordGroup) + INK_GROUP_SIZE)




/**************************/
/* REFERENCE SECTION 14.0 */
/**************************/

/*-------------------------------------------------------------------------
** Some applications may support the idea of rendering ink as if it were 
** drawn by a certain shaped pen tip.  The most common pen tips would be
** round or rectangular.  The exact details of how to render a given pen
** tip will be application specific, but this record states what pen tip
** parameters were used by the storing app.
**
** Pen tips determine, in part, how ink is rendered.  For pen tip types
** defined in future versions of this format which require additional
** parameters (such as the X and Y rectangle for a simulated nib pen, or
** brush dimensions for a simulated brush), additional data is included
** at the end of the structure.
**
** The writing application should be aware that the reading application will
** only do "the best possible" job of rendering and that fully compliant 
** reading applications may not be able to render certain nib types and/or
** colors.  Both reading and writing applications should pay particular
** attention to the following notes regarding defaults and ink drawn at a
** width of zero.
** 
** A pen tip which is drawing ink in zero width renders at the minimum
** visible width the reading application will support.
** 
** A recommended practice is that ink which should not render (should this
** be called for) be drawn with a color value of (0,0,0, 0), i.e., black,
** completely transparent.
**
** Pen tip size should scale when an INK_SCALE_RECORD is encountered.  The
** writing application should write a new INK_PENTIP_RECORD after an
** INK_SCALE_RECORD if the writing application does not want the pen tip
** size to scale along with the ink.  If the pen tip scales to zero width, 
** it should be rendered by the reading application according to the comment
** above.
**
** The default pen tip if no pentip record exists is INK_PENTIP_ROUND, with a
** width of one twip.  The dimensions of a round nib specify diameter, not
** radius: the X/Y coordinate is the center of this diameter.  Similarly, for
** for rectangular nibs, the X/Y coordinate is the center of the rectangle.  
**
** Note:
** This specification does not specify information for an algorithmic
** variation in nib width, ink color, or other "brush" effects as a function
** of tip force, speed or any other factor.  An example would be for an
** application to draw wider ink as the user presses down harder with the
** stylus.  Applications wishing to implement such features may do so using
** application-specific record types for this revision of the specification.
** 
**------------------------------------------------------------------------*/

typedef S16 INK_PENTIP, FAR *P_INK_PENTIP;
#define INK_PENTIP_SIZE S16_SIZE
#define INK_PENTIP_ROUND            (0) /* Diameter in twips */
#define INK_PENTIP_RECTANGLE        (1) /* Dimensions in twips */
#define INK_PENTIP_SLANT_RECTANGLE  (2)
#define INK_PENTIP_ROUND_FLAT_END   (3)
/* ... more to be filled in here if needed */

#define inkDefaultPentip        INK_PENTIP_ROUND
#define inkDefaultPentipData    ((U16) 1)

typedef struct tag_INK_PENTIP_SLANT {
    SIZE16  rectangle_size; /* INK_PENTIP_SLANTRECTANGLE */
    U16     angle;          /* Whole degrees from vertical, counter-clockwise */
} INK_PENTIP_SLANT, FAR *P_INK_PENTIP_SLANT;

typedef union {
  U16               round_width;        /* INK_PENTIP_ROUND */
  SIZE16            rectangle_size;     /* INK_PENTIP_RECTANGLE */
  INK_PENTIP_SLANT  slant;              /* INK_PENTIP_SLANT_RECTANGLE */
  U16               round_flat_width;   /* INK_PENTIP_ROUND_FLAT_END */
} INK_PENTIP_DATA, FAR *P_INK_PENTIP_DATA;

/* Size of the union is determined by INK_PENTIP_SLANT */
#define INK_PENTIP_DATA_SIZE (SIZE16_SIZE+U16_SIZE)

typedef struct tag_INK_PENTIP_RECORD {
    INK_RECORD_HEADER8  header;         /* value is inkRecordTip */
    INK_PENTIP          tip;
    INK_PENTIP_DATA     tip_data;
} INK_PENTIP_RECORD, FAR *P_INK_PENTIP_RECORD;
#define inkRecordTipSize    \
       (inkRecordHeaderLength(inkRecordTip) + INK_PENTIP_SIZE + \
        SIZE16_SIZE + U16_SIZE)




/**************************/
/* REFERENCE SECTION 15.0 */
/**************************/

/*-------------------------------------------------------------------------
** For some applications, it will be important to know the sampling rate of
** the pen digitizer. 
**
** This record would likely be present once in a bundle and would typically
** be after the INK_BUNDLE_RECORD, but before the first pen data.
**
** Note:
** Writing applications are not required to report the "true" sampling rate
** of the digitizer, nor are rendering applications required to play back the
** ink at the specified rate.  It is likely that most types of rendering
** applications will render ink as rapidly as possible to construct a display
** in minimum time, and that some types of animation applications will
** intentionally set an arbitrary sampling rate to vary the display rate.  
**
** Note:
** For hardware which supports a highly variable sampling rate, the writing
** application can simulate a very high sampling rate (say, 1000 points/
** second), and use skip records for "elided" points to achieve an exact time
** value (at 1-millisecond resolution) for each point.
**
** A default value for sampling rate has been arbitrarily defined below.
**
**------------------------------------------------------------------------*/

typedef struct tag_INK_POINTS_PER_SECOND_RECORD {
    INK_RECORD_HEADER8  header;         /* value is inkRecordPointsPerSecond */
    U16                 pointsPerSecond;
} INK_POINTS_PER_SECOND_RECORD, FAR *P_INK_POINTS_PER_SECOND_RECORD;

#define inkPointDefaultPointsPerSecond  (100)

#define inkRecordPointsPerSecondSize    \
       (inkRecordHeaderLength(inkRecordPointsPerSecond) + U16_SIZE)




/**************************/
/* REFERENCE SECTION 16.0 */
/**************************/

/*-------------------------------------------------------------------------
** Units for Z height of stylus above the tablet.
**
** This record would only be present once in a bundle and would typically be
** after the INK_BUNDLE_RECORD, but before the first pen data.
**
**------------------------------------------------------------------------*/

typedef struct tag_INK_UNITS_PER_Z_RECORD {
    INK_RECORD_HEADER8  header;         /* value is inkRecordUnitsPerZ */
    U32                 unitsPerZ;      /* pen units per meter (Z height) */
} INK_UNITS_PER_Z_RECORD, FAR *P_INK_UNITS_PER_Z_RECORD;

#define inkPointDefaultUnitsPerZ  (10000)   /* 0.1 mm units */

#define inkRecordUnitsPerZSize  \
       (inkRecordHeaderLength(inkRecordUnitsPerZ) + U32_SIZE)




/**************************/
/* REFERENCE SECTION 17.0 */
/**************************/

/*-------------------------------------------------------------------------
** Units for stylus tip force.
**
** This record would only be present once in a bundle and would typically be
** after the INK_BUNDLE_RECORD, but before the first pen data.
**
** Note:
** This specification assumes some level of accuracy and linearity for tip
** force data, if such data is present.  However, since tip force sensors in
** current digitizer tablet and stylus designs may well vary in accuracy and
** linearity from one unit to the next even for hardware of the same design
** and model, and since algorithms for using tip force to determine stroke
** start and end are likely to differ, a recommended practice for writing
** applications that use the tip force value to determine the "touch" points
** in a stroke is to mark those points using the touch bit in the INK_BUTTONS
** structure.  
**
** It is also recommended that vendors supporting tip force sensing in their
** hardware linearize their transducers to the greatest extent possible.  
**
** Because of the likelihood that tip force transducers may not be accurately
** linearized, negative tip force values, while perhaps somewhat absurd
** are possible and are permitted in this specification.  
** 
**------------------------------------------------------------------------*/

typedef struct tag_INK_UNITS_PER_FORCE_RECORD {
    INK_RECORD_HEADER8  header;         /* value is inkRecordUnitsPerForce */
    U32                 unitsPerForce;  /* tip-force units per k-gram of force*/
} INK_UNITS_PER_FORCE_RECORD, FAR *P_INK_UNITS_PER_FORCE_RECORD;

#define inkPointDefaultUnitsPerForce  (1000)    /* grams of force */

#define inkRecordUnitsPerForceSize  \
       (inkRecordHeaderLength(inkRecordUnitsPerForce) + U32_SIZE)




/**************************/
/* REFERENCE SECTION 18.0 */
/**************************/

/*-------------------------------------------------------------------------
** The INK_APP_RECORD record is a universal record to be used by individual
** applications to put data into the file that is not supported by an
** additional publicly defined record type.  The basic idea is that an
** application puts its own unique application signature into the appData
** bytes in the INK_APP_RECORD.  This identifies the data as originating with
** a particular application.  Then, an application defines a set of
** subRecordTypes that they wish to use.  Then, using these subRecordTypes
** they can put a wide variety of information into the file.  By examining
** the appData signature and comparing it to theirs, an application can
** decide whether it knows how to interpret the various subRecordtypes.  
** 
**------------------------------------------------------------------------*/

typedef struct tag_INK_APP_RECORD {
    INK_RECORD_HEADER32 header;         /* value is inkRecordApp */
    U8                  appSignature[8];/* reserved for possible unique */
                                        /* application signature */
    U16                 subRecordType;
    /* data here appropriate to the subRecordType and appData signature */
} INK_APP_RECORD, FAR *P_INK_APP_RECORD;
#define inkRecordAppSize(data_length)   \
    (inkRecordHeaderLength(inkRecordApp) + 8 + U16_SIZE + (data_length))




/**************************/
/* REFERENCE SECTION 19.0 */
/**************************/

/*-------------------------------------------------------------------------
** Definition of the inkData components of an INK_PENDATA_RECORD:
**
** Uncompacted point format:
** -------------------------
**
** This structure immediately follows the rest of the INK_PENDATA_RECORD. 
** The structure has several optional components, present or not present as
** indicated by the INK_BUNDLE_FLAGS in the INK_BUNDLE_RECORD.  The format is
** a sequence of "point values", each containing all the scalar data for each
** sampled tablet point.  
** 
** In the uncompacted format, there is a single structure that contains all
** of the state information for each point from the tablet.  Components not
** present (as indicated by the INK_BUNDLE_FLAGS) are just that: not present,
** do not exist, do not occupy space.  
**
** Compacted point format:
** -----------------------
**
** In the compacted format, "State values", such as the stylus state of
** touch/no-touch/out-of-prox or the on/off state of the barrel switches, are
** stored in a compacted "INK_BUTTONS" item (represented using reserved
** encodings in the INK_POINT coordinate values) interjected when their state
** changes.  The initial state is assumed to be "not touching", "out of
** proximity", all barrel switches "off".  It is possible to have both tip
** force data, and explicit starts and ends of strokes: the starts and ends
** of the strokes are then points that were considered to be such by the
** original application storing the data.    The INK_BUTTONS record reflects
** the state of the next X/Y point following.  
**
**------------------------------------------------------------------------*/




/**************************/
/* REFERENCE SECTION 20.0 */
/**************************/

/*-------------------------------------------------------------------------
** "INK_BUTTONS" items may most often be used only to indicate stylus touch
** and out-of-prox state, and perhaps a single barrel button.  The format is
** optimized for this case.  The format extends to a total of 28 stylus/puck
** buttons.
**
**------------------------------------------------------------------------*/

/*-------------------------------------------------------------------------
** The lowest-order bit (flag0) is "0" when the stylus is out of
**   proximity, "1" when it is in proximity.
** Second lowest-order bit (flag1) is "1" to indicate the next inkPoints are
**   when the stylus is touching (the start of a stroke: tip-switch "on"),
**   "0" to indicate that the stylus is not touching (end of a stroke).
** Third bit (flag2) indicates state of first (or only) barrel switch,
**   etc.
** 31'st bit (flag30) is normally "0", "1" is indicates there are more
**   than 29 barrel/puck buttons with state, and the rest are in the next
**   four-byte word.
**-------------------------------------------------------------------------*/

typedef U32 INK_BUTTONS, FAR * P_INK_BUTTONS;

/*-------------------------------------------------------------------------
** These definitions hold the maximum and minimum values that
** can be used with the S15 and S31 representations described in
** this document:
**-------------------------------------------------------------------------*/

#define MAX_S31 ((S32) 0x3FFFFFFF)
#define MIN_S31 ((S32) 0xC0000000)

#define MAX_S15 ((S16) 0x3FFF)
#define MIN_S15 ((S16) 0xC000)

#define MAX_S7 ((S16) 0x003F)
#define MIN_S7 ((S16) 0xFFC0)

#define MAX_S3 ((S16) 0x0003)
#define MIN_S3 ((S16) 0xFFFC)

/* SignExtend4/8/16/32: Sign-extend an S3, S7, S15, S31 to an S32: */

#define SignExtend4(value)  ((S32)                      \
    (((value)&0x00000004l)== 0 ?  ((value)&0x00000007l)     \
                               : (((value)&0x00000007l) | 0xFFFFFFF8l)))

#define SignExtend8(value)  ((S32)                      \
    (((value)&0x00000040l)== 0 ?  ((value)&0x0000007Fl)     \
                               : (((value)&0x0000007Fl) | 0xFFFFFF80l)))

#define SignExtend16(value) ((S32)                      \
    (((value)&0x00004000l)== 0 ?  ((value)&0x00007FFFl)     \
                               : (((value)&0x00007FFFl) | 0xFFFF8000l)))

#define SignExtend32(value) ((S32)                      \
    (((value)&0x40000000l)== 0 ?  ((value)&0x7FFFFFFFl) \
                               : (((value)&0x7FFFFFFFl) | 0x80000000l)))



/**************************/
/* REFERENCE SECTION 21.0 */
/**************************/

/*-------------------------------------------------------------------------
** INK_POINT data.  The INK_POINT structure varies in size depending on 
** flags set in the bundle header.  The XY32 position is always present, but
** the force, rho, height, angle, and buttons members are only present when
** indicated by the corresponding flag in the bundle header.  When optional
** data is present, it is present in the order defined by this structure;
** that is, position, force, height, rho, angle, and finally buttons.  
**
** The INK_POINT structure has the following elements:
**
** position - required and always present
**    Positions are measured with (0,0) at the lower-left, X increasing to
**    the right, Y increasing upwards.  Values are actually S31, not S32.
**    The high bit in X and Y must be zero.
** force    - optional, present if inkForceDataPresent is asserted
**    Units are in pen force units, zero is no contact.
** height   - optional, present if inkHeightDataPresent is asserted
**    Units are in pen unitsPerZ as specified by inkPointDefaultUnitsPerZ or
**    by an INK_UNITS_PER_Z_RECORD, whichever is appropriate.  Units increase
**    as the stylus is taken away from the tablet.  Zero means "just in 
**    contact".  Negative values could possibly result from spring action if
**    the stylus is pressed hard, or if the tablet is not perfectly accurate.
** rho      - optional, present if inkRotationDataPresent is asserted
**    Angles are measured in degrees from some nominal orientation of 
**    "stylus button on top" (somewhat arbitrary).  Angles increase with 
**    clockwise rotation as seen from the rear end of the stylus.
** angle    - optional, present if inkAngleDataPresent is asserted
**    Angles are measured in pen angle units from the vertical.  Theta 
**    increases in the positive-X direction, phi in the positive-Y.
** buttons  - optional, present if inkButtonDataPresent is asserted
**    
** When the INK_BUNDLE_RECORD member compactionType is inkStdCompression,
** all data in this structure is compressed according to the methods 
** described in reference section 23.0.  For more details on how to interpret
** the compressed data stream, see the sample code.  The bundle flags which
** indicate whether a particular piece of data is present are used regardless
** of whether the data is compressed or not.  Note that when data is written
** in compressed format, it is NOT written in Intel order but rather most
** significant byte first.  In compressed form, some of the eight bit delta
** values are reserved for button data and elided (skipped) point counts.
** This has two important ramifications.  1) When expecting a point, 
** compacted button data or elided point data may be encountered instead, and
** 2) when the inkButtonDataPresent flag is asserted in the bundle header,
** button data will appear in the place of a point and not in addition to a
** point.  If inkButtonDataPresent is not asserted, the reader need not check
** the point data for the special case of button data; however, the point 
** data must still be checked to see if it is a count of elided points rather
** than an actual point.
**
**------------------------------------------------------------------------*/

typedef struct tag_INK_POINT {
    XY32        position;  /* required x/y point data */
    S16         force;     /* optional force data */
    S16         height;    /* optional z height data */
    S16         rho;       /* optional rotational data */
    ANGLE16     angle;     /* optional theta and phi data */
    INK_BUTTONS buttons;   /* optional proximity, contact, button data */
} INK_POINT, FAR *P_INK_POINT;




/**************************/
/* REFERENCE SECTION 22.0 */
/**************************/


/*-------------------------------------------------------------------------
** The following default values are assumed before the start of any 
** INK_BUNDLE:
**
**------------------------------------------------------------------------*/


#define inkPointDefaultXYPosition   ((S32) 0)
#define inkPointDefaultForce        ((S16) 0)
#define inkPointDefaultHeight       ((S16) 0)
#define inkPointDefaultRho          ((S16) 0)
#define inkPointDefaultAngle        ((S16) 0)
#define inkPointDefaultButtons      ((U32) 0)




/**************************/
/* REFERENCE SECTION 23.0 */
/**************************/

/*-------------------------------------------------------------------------
** Compacted point format:
** ---------------------
**
** A recommended practice is always to use the compacted point format, not
** the uncompacted point format.  Sample code for reading and writing the
** compacted format is included in an appendix.  
**
** This structure also immediately follows the rest of the
** INK_PENDATA_RECORD.
**
** The uncompacted values above are stored in sequential bytes in a more
** compact, delta-oriented format.  Deltas are all signed values, a value to
** add to the previous value.  The first point in an INK_PENDATA_RECORD is
** always relative to the defined default values for each component of the
** point.
**
** The storing application, as an alternative to eliminating points, can
** specify a "skip" record for elided points.  The skipRecord indicates that
** a number of points were skipped, and the reading application is free to
** insert values for the elided points (interpolating where appropriate). 
** The intent is to allow for accurate time information to be maintained
** between time stamps for synchronization with recorded voice, press-hold
** gesture recognition, etc.  
**
** Compacted data is written most significant byte first so that reading
** applications can read the first byte and determine (from the top two bits)
** how large the encoded delta is.
**
** Note:
** "Reserved encodings" are those encodings that, if real points, would fit
** into the next smaller delta size.  16 bit deltas and 8 bit deltas have
** reserved encodings.  The reserved encodings for 16 bit deltas are all 16
** bit delta pairs where both X and Y are within the inclusive range MIN_S7
** and MAX_S7.  Similarly, the reserved encoding for 8 bit deltas are all 8
** bit delta pairs where both X and Y are within the inclusive range MIN_S3
** and MAX_S3.  In revision 1.0 of Jot, three of the reserved encodings for 8
** bit deltas are used for special cases: skip counts (reference section 
** 27.0) and button changes (reference section 26.0).
** 
** x/y position:
** ------------
**
** 32-bit absolute X/Y:
**
**      Two 32 bit long words: Data is actually two S31s:
**
**      |0|0|     (30 low-order bits of X)                             |
**              ... Sign bit is taken from first bit of next word.
**      ----------------------------------------------------------------
**      |X|       (sign bit of X plus 31 bits of Y)                     |
**      ----------------------------------------------------------------
**
** 16-bit short delta X/Y:
**
**      Short words: two 16 bit words: Deltas are actually two S15s:
**      Values that would fit into an 8-bit byte delta are reserved.
**
**      |0|1|     (14 low-order bits of delta-X)         |
**              ... Sign bit is taken from first bit of next word.
**      --------------------------------------------------
**      |X|       (sign bit of X plus 15 bits of delta Y |
**      --------------------------------------------------
**
** 8-bit byte delta X/Y:
**
**      Bytes: two bytes: Deltas are actually two S7s:
**      Values that would fit into a 4-bit nibble delta are reserved.
**
**      |1|0|     (6 low-order bits of delta-X)  |
**              ... Sign bit is taken from first bit of next word.
**      ------------------------------------------
**      |X|       (7 bits of delta-Y)            |
**      ------------------------------------------
**
** 4-bit nibble delta X/Y:
**
**      Nibbles: one byte: Deltas are actually S3:
**
**      |1|1| (S3 delta-X) | (S3 delta-Y) |
**      -----------------------------------
**
**------------------------------------------------------------------------*/




/**************************/
/* REFERENCE SECTION 24.0 */
/**************************/

/*-------------------------------------------------------------------------
** Tip force:
** ----------
**
** 16-bit absolute force:
**
**      Short word: one word: Value is actually S15:
**      Values that would fit into an 8-bit byte delta are reserved.
**
**      |0|       (15 bits of force)                |
**      ---------------------------------------------
**
** 8-bit byte delta force:
**
**      Byte: one byte: Deltas are actually S7:
**
**      |1|  (S7 delta-force)    |
**      --------------------------
**
**
** Height:
** ------
**
** (Same encoding as tip force)
**
** Rho:
** ---
**
** (Same encoding as tip force)
**
**
** Stylus theta-phi:
** ----------------
**
** 16-bit absolute theta-phi:
**
**      Short words: two words: Data is actually S15:
**
**      |0|0|     (14 low-order bits of theta)    |
**      -------------------------------------------
**              ... Sign bit is taken from first bit of next word.
**      |X|       (15 bits of phi)                |
**      -------------------------------------------
**
**------------------------------------------------------------------------*/




/**************************/
/* REFERENCE SECTION 25.0 */
/**************************/

/*-------------------------------------------------------------------------
** 8-bit byte delta theta-phi
**
**      Bytes: two bytes: Deltas are actually S7:
**      Values that would fit into a 4-bit nibble delta are reserved.
**
**      |0|1|     (6 low-order bits of delta-theta)|
**      --------------------------------------------
**              ... Sign bit is taken from first bit of next word.
**      |X|       (7 bits of delta-phi)            |
**      --------------------------------------------
**
** 4-bit nibble delta theta-phi
**
**      Nibbles: one byte: Deltas are actually S3:
**
**      |1|0|(S3 delta-theta)|(S3 delta-phi)|
**      -------------------------------------
**
** Note: 
** Leading bit values of |1|1| are reserved
**
**------------------------------------------------------------------------*/




/**************************/
/* REFERENCE SECTION 26.0 */
/**************************/

/*-------------------------------------------------------------------------
** Since the X/Y data is always present, we use some of the reserved delta
** encodings to encode button states and elided (skipped) points.  We use the
** 8-bit delta encodings that are unused: the values that can fit into the
** smaller 4-bit delta encodings. 
**
** Button/tip records:
** ------------------
**
** It is assumed that the state of barrel buttons and the touching sensor on
** the stylus change infrequently.  A compacted button/tip record is only
** included when the state changes in one of the switches.  The button state 
** value applies to the X/Y point immediately following the button state 
** record. 
**
**
** (Taken from 8-bit byte delta X/Y: two bytes total)
**
**      |1|0| 0|0|0|0|0|0/1|   0|X|.|.|.|.|X|X|
**      ---------------------------------------
**              (delta-X)           (delta-Y)
**
** An eight-bit delta with delta-X == 0 or 1, and delta-Y in the range
** (-4..3) indicates a button state encoding. 
**
** It is likely to be the case that many hardware platforms have only one
** barrel button.
**
** The three delta-Y bits indicate the "touch", "out-of-prox", and "first
** barrel button" state as follows:
**
** low-order         delta-Y bit: 1 --> in proximity,    0 --> out of prox
** next              delta-Y bit: 1 --> touching tablet, 0 --> not touching
** high-order (sign) delta-Y bit: 1 --> first button closed, 0 --> open
**
**
** The lowest order bit of the delta-X bits is used to indicate that
** additional bytes follow: "1" indicates that the next byte is used for the
** next 7 barrel buttons with state.  The high order bit of each sequential 
** byte in the series is "1" if an additional byte must be fetched, "0"
** otherwise.  In these additional bytes, the additional buttons are 
** associated in order starting with the low-order bit.   
**
**------------------------------------------------------------------------*/



/**************************/
/* REFERENCE SECTION 27.0 */
/**************************/

/*-------------------------------------------------------------------------
** Skipped-point records:
** ---------------------
**
** (Taken from 8-bit byte delta X/Y: two bytes total)
**
**
**      |1|0| 0|0|0|0|1|0|   0|X|.|.|.|.|X|X|
**      -------------------------------------
**              (delta-X)           (delta-Y)
**
** An eight-bit delta with delta-X == 2, and delta-Y in the range
** (-4..3) indicates a count of elided points.  The delta-Y values in the
** range (-4..-1) are used to represent skip counts of (4..7).  If the
** delta-Y value is zero "0", the next two bytes are fetched to get a U16
** skip count value.
**
** The elided points are points removed between the point immediately prior
** to the skipped-point record and the point immediately afterward.  This 
** implies that at least one point must follow every skip record (though 
** the point may not appear next in the stream if there are intervening
** button state transitions).  Reading applications that are interested in 
** recovering elided points will typically interpolate.  Skip counts of zero
** are meaningless and not permitted.
**
** Reserved:
** --------
**
** The remaining encodings from the 8-bit byte delta X/Y are reserved:
**
** delta-X of -4, -3, -2, -1, 3 AND ((delta-Y >= -4) & ((delta-Y <= 3))
**
**------------------------------------------------------------------------*/

#endif      /* end of INKSTORE_INCLUDED */