/******************************************************************************\
* This is a part of the Microsoft Source Code Samples.
* Copyright 1993 - 1998 Microsoft Corporation.
* All rights reserved.
* This source code is only intended as a supplement to
* Microsoft Development Tools and/or WinHelp documentation.
* See these sources for detailed information regarding the
* Microsoft samples programs.
\******************************************************************************/
/*
* LIST.H
*/
/*------------------------------------------------------------------------
| Abstract data type LIST OF (*untyped*) object.
| Different lists can have different types of object in them
| Different items in a list can have different types of object in them.
| The price of this lack of typing is that you have a slightly more
| awkward syntax and you get no help from the compiler if you try to
| put the wrong type of data into the list.
|
| The list is implemented as a collection of items. Within the item
| somewhere is the object.
|
| Objects are stored UNALIGNED within items.
|
| Use:
|
| #include <list.h>
| . . .
| LIST MyList; (* or LIST liMyList for Hungarians *)
| . . .
| MyList = List_Create();
| List_AddLast(MyList,&MyObject,sizeof(OBJECT));
|
| In the abstract a LIST is a list of objects. The representation
| is a linked collection of items. The manner of the linking is
| implementation dependent (as I write this it's linear but when you
| read it it might be a tree (See Knuth for why a tree)).
|
| A LIST is a "handle" for a list which may be thought of as a POINTER
| (whether it is really a pointer or not is implementation dependent)
| so that it can be copied at the risk of creating an alias. e.g.
|
| L = List_Create();
| L1 = L; (* L and L1 are both healthy and empty *)
| List_AddFirst(L, &elem, sizeof(elem));
| (* L1 may also appear to have one object, there again it may be sick *)
| L1 = L; (* Now they both surely see the one element *)
| List_Destroy(&L1); (* L is almost certainly sick now too *)
| L1 = List_Create(); (* All bets off as to what L is like now
| but L1 is empty and healthy
| *)
|
| If two handles compare equal then the lists must be equal, but
| unequal handles could address two similar lists i.e. the same list
| of objects held in two different LISTs of items (like pointers).
|
| A LIST can be transferred from one variable to another like this:
|
| NewList = OldList; (* copy the handle *)
| OldList = List_Create(); (* kill the old alias *)
|
| and the Create statement can be omitted if OldList is never touched again.
|
| Items are identified by Cursors. A cursor is the address of an object
| within an item in the list. i.e. it is the address of the piece of your
| data that you had inserted. (It is probably NOT the address of the item).
| It is typed as pointer to void here, but you should declare it as a pointer
| to whatever sort of object you are putting in the LIST.
|
| The operations AddFirst, AddLast, AddAfter and AddBefore
| all copy elements by direct assignment. If an element is itself
| a complex structure (say a tree) then this will only copy a pointer
| or an anchor block or whatever and give all the usual problems of
| aliases. Clear will make the list empty, but will only free the
| storage that it can "see" directly. SplitBefore or Split After may
| also perform a Clear operation. To deal with fancy data structures
| use New rather than Add calls and copy the data yourself
| e.g. P = List_NewLast(MyList, sizeof(MyArray[14])*(23-14+1));
| CopyArraySlice(P, MyArray, 14, 23);
|
| The operations NewFirst, NewLast, NewAfter, NewBefore, First and Last
| all return pointers to elements and thus allow you to do any copying.
| This is how you might copy a whole list of fancy structures:
|
| void CopyFancyList(LIST * To, LIST From)
| (* Assumes that To has been Created and is empty *)
| { PELEMENT Cursor;
| PELEMENT P;
|
| List_TRAVERSE(From, Cursor);
| { P = List_NewLast(To, sizeof(element) );
| FancyCopy(P, Cursor); (* Copy so that *Cursor==*P afterwords *)
| }
| }
--------------------------------------------------------------------*/
typedef struct item_tag FAR * LIST;
typedef LIST FAR * PLIST;
void APIENTRY List_Init(void);
/* MUST BE CALLED BEFORE ANY OF THE OTHER FUNCTIONS. */
void APIENTRY List_Dump(LPSTR Header, LIST lst);
/* Dump the internals to current output stream -- debug only */
void APIENTRY List_Show(LIST lst);
/* Dump hex representation of handle to current out stream -- debug only */
LIST APIENTRY List_Create(void);
/* Create a list. It will be initially empty */
void APIENTRY List_Destroy(PLIST plst);
/* Destroy *plst. It does not need to be empty first.
| All storage directly in the list wil be freed.
*/
void APIENTRY List_AddFirst(LIST lst, LPVOID pObject, UINT uLen);
/* Add an item holding Object to the beginning of * plst */
LPVOID APIENTRY List_NewFirst(LIST lst, UINT uLen);
/* Return the address of the place for Len bytes of data in a new
| item at the start of *plst
*/
void APIENTRY List_DeleteFirst(LIST lst);
/* Delete the first item in lst. Error if lst is empty */
void APIENTRY List_AddLast(LIST lst, LPVOID pObject, UINT uLen);
/* Add an item holding Object to the end of lst */
LPVOID APIENTRY List_NewLast(LIST lst, UINT uLen);
/* Return the address of the place for uLen bytes of data in a new
| item at the end of lst
*/
void APIENTRY List_DeleteLast(LIST lst);
/* Delete the last item in lst. Error if lst is empty */
void APIENTRY List_AddAfter( LIST lst
, LPVOID Curs
, LPVOID pObject
, UINT uLen
);
/*--------------------------------------------------------------------
| Add an item holding *pObject to lst immediately after Curs.
| List_AddAfter(lst, NULL, pObject, Len) adds it to the start of the lst
---------------------------------------------------------------------*/
LPVOID APIENTRY List_NewAfter(LIST lst, LPVOID Curs, UINT uLen);
/*--------------------------------------------------------------------
| Return the address of the place for uLen bytes of data in a new
| item immediately after Curs.
| List_NewAfter(Lst, NULL, uLen) returns a pointer
| to space for uLen bytes in a new first element.
---------------------------------------------------------------------*/
void APIENTRY List_AddBefore( LIST lst
, LPVOID Curs
, LPVOID pObject
, UINT uLen
);
/*--------------------------------------------------------------------
| Add an item holding Object to lst immediately before Curs.
| List_AddBefore(Lst, NULL, Object, uLen) adds it to the end of the list
---------------------------------------------------------------------*/
LPVOID APIENTRY List_NewBefore(LIST lst, LPVOID Curs, UINT uLen );
/*--------------------------------------------------------------------
| Return the address of the place for uLen bytes of data in a new
| item immediately before Curs.
| List_NewBefore(Lst, NULL, uLen) returns a pointer
| to space for uLen bytes in a new last element.
---------------------------------------------------------------------*/
void APIENTRY List_Delete(LPVOID Curs);
/*------------------------------------------------------------------
| Delete the item that Curs identifies.
| This will be only a few (maybe as little as 3) machine instructions
| quicker than DeleteAndNext or DeleteAndPrev but leaves Curs dangling.
| It is therefore NOT usually to be preferred.
| It may be useful when you have a function which returns an LPVOID
| since the argument does not need to be a variable.
| Trivial example: List_Delete(List_First(L));
-------------------------------------------------------------------*/
int APIENTRY List_ItemLength(LPVOID Curs);
/* Return the length of the object identified by the cursor Curs */
/*------------------------------------------------------------------
| TRAVERSING THE ULIST
|
| LIST lst;
| object * Curs;
| . . .
| Curs = List_First(lst);
| while (Curs!=NULL)
| { DoSomething(*Curs); (* Curs points to YOUR data not to chain ptrs *)
| Curs = List_Next(Curs);
| }
|
| This is identically equal to
| List_TRAVERSE(lst, Curs) // note NO SEMI COLON!
| { DoSomething(*Curs); }
-------------------------------------------------------------------*/
#define List_TRAVERSE(lst, curs) for( curs=List_First(lst) \
; curs!=NULL \
; curs = List_Next((LPVOID)curs) \
)
LPVOID APIENTRY List_First(LIST lst);
/*------------------------------------------------------------------
| Return the address of the first object in lst
| If lst is empty then Return NULL.
--------------------------------------------------------------------*/
LPVOID APIENTRY List_Last(LIST lst);
/*------------------------------------------------------------------
| Return the address of the last object in lst
| If lst is empty then return NULL.
--------------------------------------------------------------------*/
LPVOID APIENTRY List_Next(LPVOID Curs);
/*------------------------------------------------------------------
| Return the address of the object after Curs^.
| List_Next(List_Last(lst)) == NULL; List_Next(NULL) is an error.
| List_Next(List_Prev(curs)) is illegal if curs identifies first el
--------------------------------------------------------------------*/
LPVOID APIENTRY List_Prev(LPVOID Curs);
/*------------------------------------------------------------------
| Return the address of the object after Curs^.
| List_Prev(List_First(L)) == NULL; List_Prev(NULL) is an error.
| List_Prev(List_Next(curs)) is illegal if curs identifies last el
--------------------------------------------------------------------*/
/*------------------------------------------------------------------
| Whole list operations
-----------------------------------------------------------------*/
void APIENTRY List_Clear(LIST lst);
/* arrange that lst is empty after this */
BOOL APIENTRY List_IsEmpty(LIST lst);
/* Return TRUE if and only if lst is empty */
void APIENTRY List_Join(LIST l1, LIST l2);
/*-----------------------------------------------------------------------
| l1 := l1||l2; l2 := empty
| The elements themselves are not moved, so pointers to them remain valid.
|
| l1 gets all the elements of l1 in their original order followed by
| all the elements of l2 in the order they were in in l2.
| l2 becomes empty.
------------------------------------------------------------------------*/
void APIENTRY List_InsertListAfter(LIST l1, LIST l2, LPVOID Curs);
/*-----------------------------------------------------------------------
| l1 := l1[...Curs] || l2 || l1[Curs+1...]; l2 := empty
| Curs=NULL means insert l2 at the start of l1
| The elements themselves are not moved, so pointers to them remain valid.
|
| l1 gets the elements of l1 from the start up to and including the element
| that Curs points at, in their original order,
| followed by all the elements that were in l2, in their original order,
| followed by the rest of l1
------------------------------------------------------------------------*/
void APIENTRY List_InsertListBefore(LIST l1, LIST l2, LPVOID Curs);
/*-----------------------------------------------------------------------
| l1 := l1[...Curs-1] || l2 || l1[Curs...]; l2 := empty
| Curs=NULL means insert l2 at the end of l1
| The elements themselves are not moved, so pointers to them remain valid.
|
| l1 gets the elements of l1 from the start up to but not including the
| element that Curs points at, in their original order,
| followed by all the elements that were in l2, in their original order,
| followed by the rest of l1.
------------------------------------------------------------------------*/
void APIENTRY List_SplitAfter(LIST l1, LIST l2, LPVOID Curs);
/*-----------------------------------------------------------------------
| Let l1 be l1 and l2 be l2
| Split l2 off from the front of l1: final l2,l1 = original l1
|
| Split l1 into l2: objects of l1 up to and including Curs object
| l1: objects of l1 after Curs
| Any original contents of l2 are freed.
| List_Spilt(l1, l2, NULL) splits l1 before the first object so l1 gets all.
| The elements themselves are not moved.
------------------------------------------------------------------------*/
void APIENTRY List_SplitBefore(LIST l1, LIST l2, LPVOID Curs);
/*----------------------------------------------------------------------
| Split l2 off from the back of l1: final l1,l2 = original l1
|
| Split l1 into l1: objects of l1 up to but not including Curs object
| l2: objects of l1 from Curs onwards
| Any original contants of l2 are freed.
| List_Spilt(l1, l2, NULL) splits l1 after the last object so l1 gets all.
| The elements themselves are not moved.
-----------------------------------------------------------------------*/
int APIENTRY List_Card(LIST lst);
/* Return the number of items in L */
/*------------------------------------------------------------------
| Error handling.
|
| Each list has within it a flag which indicates whether any illegal
| operation has been detected (e.g. DeleteFirst when empty).
| Rather than have a flag on every operation, there is a flag held
| within the list that can be queried when convenient. Many operations
| do not have enough redundancy to allow any meaningful check. This
| is a design compromise (for instance to allow P = List_Next(P);
| rather than P = List_Next(L, P); which is more awkward, especially
| if L is actually a lengthy phrase).
|
| List_IsOK tests this flag (so is a very simple, quick operation).
| MakeOK sets the flag to TRUE, in other words to accept the current
| state of the list.
|
| It is possible for a list to be damaged (whether or not the flag
| says OK) for instance by the storage being overwritten.
|
| List_Check attempts to verify that the list is sound (for instance where
| there are both forward and backward pointers they should agree).
|
| List_Recover attempts to make a sound list out of whatever debris is left.
| If the list is damaged, Recover may trap (e.g. address error) but
| if the list was damaged then ANY operation on it may trap.
| If Check succeeds without trapping then so will Recover.
-----------------------------------------------------------------*/
BOOL APIENTRY List_IsOK(LIST lst);
/* Check return code */
void APIENTRY List_MakeOK(LIST lst);
/* Set return code to good */
BOOL APIENTRY List_Check(LIST lst);
/* Attempt to validate the chains */
void APIENTRY List_Recover(PLIST plst);
/* Desperate stuff. Attempt to reconstruct something */
/*------------------------------------------------------------------
| It is designed to be as easy to USE as possible, consistent
| only with being an opaque type.
|
| In particular, the decision to use the address of an object a list cursor
| means that there is a small amount of extra arithmetic (in the
| IMPLEMENTATION) in cursor operations (e.g. Next and Prev).
| and spurious arguments are avoided whenever possible, even though
| it would allow greater error checking.
|
| Of the "whole list" operations, Clear is given because it seems to be
| a common operation, even though the caller can implement it with almost
| the same efficiency as the List implementation module.
| Join, Split and InsertListXxx cannot be implemented efficiently without
| knowing the representation.
--------------------------------------------------------------------*/