Tcl Source Code

Check-in [190272213a]
Login
Bounty program for improvements to Tcl and certain Tcl packages.
Tcl 2018 Conference, Houston/TX, US, Oct 15-19
Send your abstracts to tclconference@googlegroups.com
or submit via the online form by Aug 20.

Many hyperlinks are disabled.
Use anonymous login to enable hyperlinks.

Overview
Comment:New routine TclStringReplace() serves as a common implementation for both compiled and direct eval paths of [string replace]. One routine to debug, optimize, convert to new data structures, etc.

This routine will also make a good core engine for [string insert].

With this refactoring done, the value of the INST_STR_REPLACE instruction is no longer clear. Until now, the value of compiling to it was that it routed you through the "better" of the two implementations. An instruction that does nothing different from what invocation to the direct eval fallback would do does not have obvious value.

Downloads: Tarball | ZIP archive | SQL archive
Timelines: family | ancestors | descendants | both | core-8-branch
Files: files | file ages | folders
SHA3-256:190272213adb7b0f44c4248013a8cea1609c29c36b7ffac9367a7381fd044d75
User & Date: dgp 2018-03-14 15:17:27
Context
2018-03-14
17:20
unused variable fix check-in: 67e1ea3939 user: dgp tags: core-8-branch
16:17
merge 8.7 check-in: 961ccb69d2 user: dgp tags: trunk
15:17
New routine TclStringReplace() serves as a common implementation for both compiled and direct eval p... check-in: 190272213a user: dgp tags: core-8-branch
05:41
merge mark check-in: 97dd207476 user: dgp tags: core-8-branch
Changes
Hide Diffs Unified Diffs Ignore Whitespace Patch

Changes to generic/tclCmdMZ.c.

2335
2336
2337
2338
2339
2340
2341
2342
2343
2344
2345
2346
2347
2348
2349
2350
2351
2352


2353
2354
2355
2356
2357

2358
2359
2360
2361
2362
2363
2364
2365
2366
2367
2368
	 * have (first <= end < 0 <= last) and an empty string is permitted
	 * to be replaced.
	 */
	Tcl_SetObjResult(interp, objv[1]);
    } else {
	Tcl_Obj *resultPtr;

	/*
	 * We are re-fetching in case the string argument is same value as 
	 * an index argument, and shimmering cost us our ustring.
	 */

	ustring = Tcl_GetUnicodeFromObj(objv[1], &length);
	end = length-1;

	if (first < 0) {
	    first = 0;
	}



	resultPtr = Tcl_NewUnicodeObj(ustring, first);
	if (objc == 5) {
	    Tcl_AppendObjToObj(resultPtr, objv[4]);
	}

	if (last < end) {
	    Tcl_AppendUnicodeToObj(resultPtr, ustring + last + 1,
		    end - last);
	}
	Tcl_SetObjResult(interp, resultPtr);
    }
    return TCL_OK;
}
 
/*
 *----------------------------------------------------------------------







<
<
<
<
<
<
<
<



>
>
|
<
<
<
|
>
|
|
<
|







2335
2336
2337
2338
2339
2340
2341








2342
2343
2344
2345
2346
2347



2348
2349
2350
2351

2352
2353
2354
2355
2356
2357
2358
2359
	 * have (first <= end < 0 <= last) and an empty string is permitted
	 * to be replaced.
	 */
	Tcl_SetObjResult(interp, objv[1]);
    } else {
	Tcl_Obj *resultPtr;









	if (first < 0) {
	    first = 0;
	}
	if (last > end) {
	    last = end;
	}




	resultPtr = TclStringReplace(interp, objv[1], first,
		last + 1 - first, (objc == 5) ? objv[4] : NULL,
		TCL_STRING_IN_PLACE);


	Tcl_SetObjResult(interp, resultPtr);
    }
    return TCL_OK;
}
 
/*
 *----------------------------------------------------------------------

Changes to generic/tclExecute.c.

5463
5464
5465
5466
5467
5468
5469
5470
5471
5472
5473
5474
5475
5476
5477
5478
5479
5480
5481
5482
5483
5484
5485
5486
5487
5488
5489
5490
5491
5492
5493
5494
5495
5496
5497
5498
5499
5500
5501
5502
5503
5504
5505
5506
5507
5508
5509
5510
5511
5512
5513
5514
5515
5516
5517
5518
5519
5520
5521

5522
5523
5524
5525
5526
5527
5528
5529
5530
5531
5532
5533
5534
5535
5536
5537
5538
5539
5540
5541
5542
5543
5544
5545
5546
5547
5548
5549
5550
5551
5552
	if (fromIdx == 0 && toIdx == endIdx) {
	    TclDecrRefCount(OBJ_AT_TOS);
	    OBJ_AT_TOS = value3Ptr;
	    TRACE_APPEND(("\"%.30s\"\n", O2S(value3Ptr)));
	    NEXT_INST_F(1, 0, 0);
	}

	length3 = Tcl_GetCharLength(value3Ptr);

	/*
	 * See if we can splice in place. This happens when the number of
	 * characters being replaced is the same as the number of characters
	 * in the string to be inserted.
	 */

	if (length3 - 1 == toIdx - fromIdx) {
	    unsigned char *bytes1, *bytes2;

	    if (Tcl_IsShared(valuePtr)) {
		objResultPtr = Tcl_DuplicateObj(valuePtr);
	    } else {
		objResultPtr = valuePtr;
	    }
	    if (TclIsPureByteArray(objResultPtr)
		    && TclIsPureByteArray(value3Ptr)) {
		bytes1 = Tcl_GetByteArrayFromObj(objResultPtr, NULL);
		bytes2 = Tcl_GetByteArrayFromObj(value3Ptr, NULL);
		memcpy(bytes1 + fromIdx, bytes2, length3);
	    } else {
		ustring1 = Tcl_GetUnicodeFromObj(objResultPtr, NULL);
		ustring2 = Tcl_GetUnicodeFromObj(value3Ptr, NULL);
		memcpy(ustring1 + fromIdx, ustring2,
			length3 * sizeof(Tcl_UniChar));
	    }
	    Tcl_InvalidateStringRep(objResultPtr);
	    TclDecrRefCount(value3Ptr);
	    TRACE_APPEND(("\"%.30s\"\n", O2S(objResultPtr)));
	    if (objResultPtr == valuePtr) {
		NEXT_INST_F(1, 0, 0);
	    } else {
		NEXT_INST_F(1, 1, 1);
	    }
	}

	/*
	 * Get the unicode representation; this is where we guarantee to lose
	 * bytearrays.
	 */

	ustring1 = Tcl_GetUnicodeFromObj(valuePtr, &length);
	length--;

	/*
	 * Remove substring using copying.
	 */

	objResultPtr = NULL;
	if (fromIdx > 0) {
	    objResultPtr = Tcl_NewUnicodeObj(ustring1, fromIdx);

	}
	if (length3 > 0) {
	    if (objResultPtr) {
		Tcl_AppendObjToObj(objResultPtr, value3Ptr);
	    } else if (Tcl_IsShared(value3Ptr)) {
		objResultPtr = Tcl_DuplicateObj(value3Ptr);
	    } else {
		objResultPtr = value3Ptr;
	    }
	}
	if (toIdx < length) {
	    if (objResultPtr) {
		Tcl_AppendUnicodeToObj(objResultPtr, ustring1 + toIdx + 1,
			length - toIdx);
	    } else {
		objResultPtr = Tcl_NewUnicodeObj(ustring1 + toIdx + 1,
			length - toIdx);
	    }
	}
	if (objResultPtr == NULL) {
	    /* This has to be the case [string replace $s 0 end {}] */
	    /* which has result {} which is same as value3Ptr. */
	    objResultPtr = value3Ptr;
	}
	if (objResultPtr == value3Ptr) {
	    /* See [Bug 82e7f67325] */
	    TclDecrRefCount(OBJ_AT_TOS);
	    OBJ_AT_TOS = value3Ptr;
	    TRACE_APPEND(("\"%.30s\"\n", O2S(value3Ptr)));
	    NEXT_INST_F(1, 0, 0);
	}







<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
|
<
<
>
|
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<







5463
5464
5465
5466
5467
5468
5469

















































5470


5471
5472























5473
5474
5475
5476
5477
5478
5479
	if (fromIdx == 0 && toIdx == endIdx) {
	    TclDecrRefCount(OBJ_AT_TOS);
	    OBJ_AT_TOS = value3Ptr;
	    TRACE_APPEND(("\"%.30s\"\n", O2S(value3Ptr)));
	    NEXT_INST_F(1, 0, 0);
	}


















































	objResultPtr = TclStringReplace(interp, valuePtr, fromIdx,


		toIdx - fromIdx + 1, value3Ptr, TCL_STRING_IN_PLACE);
























	if (objResultPtr == value3Ptr) {
	    /* See [Bug 82e7f67325] */
	    TclDecrRefCount(OBJ_AT_TOS);
	    OBJ_AT_TOS = value3Ptr;
	    TRACE_APPEND(("\"%.30s\"\n", O2S(value3Ptr)));
	    NEXT_INST_F(1, 0, 0);
	}

Changes to generic/tclInt.h.

4002
4003
4004
4005
4006
4007
4008



4009
4010
4011
4012
4013
4014
4015
			    Tcl_Obj *const objv[], int flags);
MODULE_SCOPE int	TclStringFirst(Tcl_Obj *needle, Tcl_Obj *haystack,
			    int start);
MODULE_SCOPE int	TclStringLast(Tcl_Obj *needle, Tcl_Obj *haystack,
			    int last);
MODULE_SCOPE Tcl_Obj *	TclStringRepeat(Tcl_Interp *interp, Tcl_Obj *objPtr,
			    int count, int flags);



MODULE_SCOPE Tcl_Obj *	TclStringReverse(Tcl_Obj *objPtr, int flags);

/* Flag values for the [string] ensemble functions. */

#define TCL_STRING_MATCH_NOCASE TCL_MATCH_NOCASE /* (1<<0) in tcl.h */
#define TCL_STRING_IN_PLACE (1<<1)








>
>
>







4002
4003
4004
4005
4006
4007
4008
4009
4010
4011
4012
4013
4014
4015
4016
4017
4018
			    Tcl_Obj *const objv[], int flags);
MODULE_SCOPE int	TclStringFirst(Tcl_Obj *needle, Tcl_Obj *haystack,
			    int start);
MODULE_SCOPE int	TclStringLast(Tcl_Obj *needle, Tcl_Obj *haystack,
			    int last);
MODULE_SCOPE Tcl_Obj *	TclStringRepeat(Tcl_Interp *interp, Tcl_Obj *objPtr,
			    int count, int flags);
MODULE_SCOPE Tcl_Obj *	TclStringReplace(Tcl_Interp *interp, Tcl_Obj *objPtr,
			    int first, int count, Tcl_Obj *insertPtr,
			    int flags);
MODULE_SCOPE Tcl_Obj *	TclStringReverse(Tcl_Obj *objPtr, int flags);

/* Flag values for the [string] ensemble functions. */

#define TCL_STRING_MATCH_NOCASE TCL_MATCH_NOCASE /* (1<<0) in tcl.h */
#define TCL_STRING_IN_PLACE (1<<1)

Changes to generic/tclStringObj.c.

34
35
36
37
38
39
40

41
42
43
44
45
46
47
....
3523
3524
3525
3526
3527
3528
3529
















































































































































3530
3531
3532
3533
3534
3535
3536
 * this file, and for a DISCLAIMER OF ALL WARRANTIES.
 */

#include "tclInt.h"
#include "tommath.h"
#include "tclStringRep.h"


/*
 * Prototypes for functions defined later in this file:
 */

static void		AppendPrintfToObjVA(Tcl_Obj *objPtr,
			    const char *format, va_list argList);
static void		AppendUnicodeToUnicodeRep(Tcl_Obj *objPtr,
................................................................................
	}
	/* Pass 2. Reverse all the bytes. */
	ReverseBytes((unsigned char *)to, (unsigned char *)from, numBytes);
    }

    return objPtr;
}
















































































































































 
/*
 *---------------------------------------------------------------------------
 *
 * FillUnicodeRep --
 *
 *	Populate the Unicode internal rep with the Unicode form of its string







>







 







>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>







34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
....
3524
3525
3526
3527
3528
3529
3530
3531
3532
3533
3534
3535
3536
3537
3538
3539
3540
3541
3542
3543
3544
3545
3546
3547
3548
3549
3550
3551
3552
3553
3554
3555
3556
3557
3558
3559
3560
3561
3562
3563
3564
3565
3566
3567
3568
3569
3570
3571
3572
3573
3574
3575
3576
3577
3578
3579
3580
3581
3582
3583
3584
3585
3586
3587
3588
3589
3590
3591
3592
3593
3594
3595
3596
3597
3598
3599
3600
3601
3602
3603
3604
3605
3606
3607
3608
3609
3610
3611
3612
3613
3614
3615
3616
3617
3618
3619
3620
3621
3622
3623
3624
3625
3626
3627
3628
3629
3630
3631
3632
3633
3634
3635
3636
3637
3638
3639
3640
3641
3642
3643
3644
3645
3646
3647
3648
3649
3650
3651
3652
3653
3654
3655
3656
3657
3658
3659
3660
3661
3662
3663
3664
3665
3666
3667
3668
3669
3670
3671
3672
3673
3674
3675
3676
3677
3678
3679
3680
3681
 * this file, and for a DISCLAIMER OF ALL WARRANTIES.
 */

#include "tclInt.h"
#include "tommath.h"
#include "tclStringRep.h"

#include "assert.h"
/*
 * Prototypes for functions defined later in this file:
 */

static void		AppendPrintfToObjVA(Tcl_Obj *objPtr,
			    const char *format, va_list argList);
static void		AppendUnicodeToUnicodeRep(Tcl_Obj *objPtr,
................................................................................
	}
	/* Pass 2. Reverse all the bytes. */
	ReverseBytes((unsigned char *)to, (unsigned char *)from, numBytes);
    }

    return objPtr;
}
 
/*
 *---------------------------------------------------------------------------
 *
 * TclStringReplace --
 *
 *	Implements the inner engine of the [string replace] command.
 *
 *	The result is a concatenation of a prefix from objPtr, characters
 *	0 through first-1, the insertPtr string value, and a suffix from
 *	objPtr, characters from first + count to the end. The effect is
 *	as if the inner substring of characters first through first+count-1
 *	are removed and replaced with insertPtr.
 *	If insertPtr is NULL, it is treated as an empty string.
 *	When passed the flag TCL_STRING_IN_PLACE, this routine will try
 *	to do the work within objPtr, so long as no sharing forbids it.
 *	Without that request, or as needed, a new Tcl value will be allocated
 *	to be the result.
 *
 * Results:
 *	A Tcl value that is the result of the substring replacement.
 *	May return NULL in case of an error. When NULL is returned and
 *	interp is non-NULL, error information is left in interp
 *
 *---------------------------------------------------------------------------
 */

Tcl_Obj *
TclStringReplace(
    Tcl_Interp *interp,		/* For error reporting, may be NULL */
    Tcl_Obj *objPtr,		/* String to act upon */
    int first,			/* First index to replace */
    int count,			/* How many chars to replace */
    Tcl_Obj *insertPtr,		/* Replacement string, may be NULL */
    int flags)			/* TCL_STRING_IN_PLACE => attempt in-place */
{
    int inPlace = flags & TCL_STRING_IN_PLACE;
    Tcl_Obj *result;

    /* Caller is expected to pass sensible arguments */
    assert ( count >= 0 ) ;
    assert ( first >= 0 ) ;

    /* Replace nothing with nothing */
    if ((insertPtr == NULL) && (count == 0)) {
	if (inPlace) {
	    return objPtr;
	} else {
	    return Tcl_DuplicateObj(objPtr);
	}
    }

    /*
     * The caller very likely had to call Tcl_GetCharLength() or similar
     * to be able to process index values.  This means it is like that
     * objPtr is either a proper "bytearray" or a "string" or else it has
     * a known and short string rep.
     */

    if (TclIsPureByteArray(objPtr)) {
	int numBytes;
	unsigned char *bytes = Tcl_GetByteArrayFromObj(objPtr, &numBytes);

	if (insertPtr == NULL) {
	    /* Replace something with nothing. */

	    assert ( first <= numBytes ) ;
	    assert ( count <= numBytes ) ;
	    assert ( first + count <= numBytes ) ;

	    result = Tcl_NewByteArrayObj(NULL, numBytes - count);/* PANIC? */
	    TclAppendBytesToByteArray(result, bytes, first);
	    TclAppendBytesToByteArray(result, bytes + first + count,
		    numBytes - count - first);
	    return result;
	}

	/* Replace everything */
	if ((first == 0) && (count == numBytes)) {
	    return insertPtr;
	}

	if (TclIsPureByteArray(insertPtr)) {
	    int newBytes;
	    unsigned char *iBytes
		    = Tcl_GetByteArrayFromObj(insertPtr, &newBytes);

	    if (count == newBytes && inPlace && !Tcl_IsShared(objPtr)) {
		/*
		 * Removal count and replacement count are equal.
		 * Other conditions permit. Do in-place splice.
		 */

		memcpy(bytes + first, iBytes, count);
		Tcl_InvalidateStringRep(objPtr);
		return objPtr;
	    }

	    if (newBytes > INT_MAX - (numBytes - count)) {
		if (interp) {
		    Tcl_SetObjResult(interp, Tcl_ObjPrintf(
			    "max size for a Tcl value (%d bytes) exceeded",
			    INT_MAX));
		    Tcl_SetErrorCode(interp, "TCL", "MEMORY", NULL);
		}
		return NULL;
	    }
	    result = Tcl_NewByteArrayObj(NULL, numBytes - count + newBytes);
								/* PANIC? */
	    TclAppendBytesToByteArray(result, bytes, first);	
	    TclAppendBytesToByteArray(result, iBytes, newBytes);
	    TclAppendBytesToByteArray(result, bytes + first + count,
		    numBytes - count - first);
	    return result;
	}

	/* Flow through to try other approaches below */
    }

    /*
     * TODO: Figure out how not to generate a Tcl_UniChar array rep
     * when it can be determined objPtr->bytes points to a string of
     * all single-byte characters so we can index it directly.
     */

    /* The traditional implementation... */
    {
	int numChars;
	Tcl_UniChar *ustring = Tcl_GetUnicodeFromObj(objPtr, &numChars);

	/* TODO: Is there an in-place option worth pursuing here? */
	
	result = Tcl_NewUnicodeObj(ustring, first);
	if (insertPtr) {
	    Tcl_AppendObjToObj(result, insertPtr);
	}
	if (first + count < numChars) {
	    Tcl_AppendUnicodeToObj(result, ustring + first + count,
		    numChars - first - count);
	}

	return result;
    }
}
 
/*
 *---------------------------------------------------------------------------
 *
 * FillUnicodeRep --
 *
 *	Populate the Unicode internal rep with the Unicode form of its string