Modeling Algorithms - SIGSEGV in buildsweep

if either of the sections is omitted the sweep returns a result.

the following script reproduces the problem without relying on external BREPs

pload ALL
circle c1 0 0 0 40
mkedge e1 c1
wire w1 e1
tcopy w1 Sketch002
tcopy w1 Sketch001
renamevar w1 Sketch
trotate Sketch 0 0 0 0. 1. 0. 90
ttranslate Sketch 0. 0. 75.
trotate Sketch001 0 0 0 0. 1. 0. 90
ttranslate Sketch001 0. 0. -75.
ttranslate Sketch002 50. 0. 0.
mksweep Sketch002
setsweep -CF
addsweep Sketch001
addsweep Sketch
buildsweep Sweep

#An exception was caught 0x7f61f1538031 : OSD_SIGSEGV: SIGSEGV 'segmentation violation' detected. Address 10
#** Exception ** 0x7f61f1538031 : OSD_SIGSEGV: SIGSEGV 'segmentation violation' detected. Address 10

#additional example for test case creation

#generated by FreeCAD 0.15.4119 +12 (Git)
pload MODELING
#saved shape of Part::Part2DObjectPython::Draft._BezCurve Object
restore minihyd-draw-BezCurve.brep BezCurve
ttranslate BezCurve 0 -8 0
explode BezCurve E
wire Sweep001-0-spine BezCurve_1
mksweep Sweep001-0-spine
setsweep -CF
circle Circle-curve 0 0 0 18
mkedge Circle-edge Circle-curve 0 0
wire Circle Circle-edge
trotate Circle 0 0 0 0.577350269189625731 0.577350269189625731 0.577350269189625731 120
ttranslate Circle 0 -71 203.5
addsweep Circle
circle Circle001-curve 0 0 0 16
mkedge Circle001-edge Circle001-curve 0 0
wire Circle001 Circle001-edge
trotate Circle001 0 0 0 0.577350269189625731 0.577350269189625731 0.577350269189625731 120
ttranslate Circle001 0 -71 203.5
addsweep Circle001
buildsweep Sweep001 -C -S
donly Sweep001


#SIGSEGV, Segmentation fault.
#0x00007fff8aab9559 in GeomFill_Array1OfSectionLaw::Value (this=0x10, Index=1)
# at occt-eb46b12/inc/TCollection_Array1.lxx:119
#119 Standard_OutOfRange_Raise_if((Index < myLowerBound || Index > myUpperBound),NULL);

The test case Attachment 7 (DRAW) seems to be affected by #0025453 it does not longer create a sigsegv but a faulty shape
> checkshape Sweep
On Shape faulty_1 :
BRepCheck_UnorientableShape
Shape faulty_2 on shape faulty_1 :
BRepCheck_BadOrientationOfSubshape
On Shape faulty_3 :
BRepCheck_UnorientableShape
Shape faulty_4 on shape faulty_3 :
BRepCheck_BadOrientationOfSubshape

Faulty shapes in variables faulty_1 to faulty_4


But the other two examples (minihyd and from c29262) still create a SIGSEGV

I got a new user report that generates a similar stacktrace. (DRAWEXE example included in Attachment 2 (ZIP))

I got a new user report that generates a similar stacktrace. (DRAWEXE example included in Attachment 3 (ZIP))
it involves an ellipse with negative parameter range:
"mkedge Ellipse-edge Ellipse-curve 3.141592653589793 0"

Program received signal SIGSEGV, Segmentation fault.
0x00007f97c7135eb1 in GeomFill_Array1OfSectionLaw::Value (this=0x10, Index=1)
    at occt/inc/TCollection_Array1.lxx:119
119	  Standard_OutOfRange_Raise_if((Index < myLowerBound || Index > myUpperBound),NULL);

Problem described in issue is reproduced on current state of OCCT.

Also we have an exception on following script (trying to sweep line along circle if line starts from circle's center):

pload ALL
circle c1 0 0 0 40
mkedge e1 c1
wire w1 e1
line l1 0 0 0 0 0 20
mkedge e2 l1 0 10
wire w2 e2
mksweep w1
addsweep w2
buildsweep sw1

Branch [archived branch] has been created by Participant.

[revision removed]


Detailed log of new commits:

Author: asuraven
Date: Mon Jan 17 11:42:16 2022 +0300

    0024909: Modeling Algorithms - SIGSEGV in buildsweep
    
    Change NCollection_DataMap NCollection_IndexedDataMap to avoid undefined behavior in BRepFill_CompatibleWires::SameNumberByPolarMethod()
    
    Correction in BRepFill_PipeShell::Prepare()

Branch [archived branch] has been updated forcibly by Participant.

[revision removed]

Branch [archived branch] has been updated forcibly by Participant.

[revision removed]

Branch [archived branch] has been updated forcibly by Participant.

[revision removed]

Branch [archived branch] has been updated forcibly by Participant.

[revision removed]

Branch [archived branch] has been updated by Participant.

[revision removed]


Detailed log of new commits:

Author: akondrat
Date: Wed Aug 10 13:45:58 2022 +0300

    0024909: Modeling Algorithms - SIGSEGV in buildsweep
    
    Update algorithm.

-	    for (ii=1; ii<=Ext.NbExt(); ii++) {
+	    for (ii=1; ii<Ext.NbExt(); ii++) {

Why do you ignore the last solution?

Branch [archived branch] has been updated forcibly by Participant.

[revision removed]

Branch [archived branch] has been updated forcibly by Participant.

[revision removed]

Wrong indentation (when using tab size 8):

         (U2 >= Ut21 - RealEpsilon())  &&
         (U2 <= Ut22 + RealEpsilon())) {
       Val = AlgExt.SquareDistance(i);
+    if (mypoints.IsEmpty())
+      mySqDist.Clear();
       mySqDist.Append(Val);
           if (!theIsInverse)

The same is in the file BRepFill_CompatibleWires.cxx, line 1159.

src/TopTools/TopTools_IndexedDataMapOfShapeListOfShape.hxx
Changes in this file bring inconsistency with other TopTools_Indexed*.hxx files. I propose to revert it and use TopTools_IndexedDataMapOfShapeListOfShape::Iterator where it is needed.

Why the new version passes map parameter by value?

 static Standard_Boolean SearchRoot (const TopoDS_Vertex& V,
-				    const TopTools_DataMapOfShapeListOfShape& Map,
-				    TopoDS_Vertex& VRoot)
+                                    const TopTools_IndexedDataMapOfShapeListOfShape Map,
+                                    TopoDS_Vertex& VRoot)

This code (line 1000) looks overcomplicated, taking into account that Init is an empty list:

      TopTools_ListOfShape* aShapeList = RMap.ChangeSeek(Vi);
      if (!aShapeList)
        aShapeList = &RMap(RMap.Add(Vi, Init));
      aShapeList->Append(Init);

Here (line 1041), aShapeList may take duplicated shape Vi.

    TopTools_ListOfShape* aShapeList = MapVLV.ChangeSeek(Vi);
    if (!aShapeList)
      aShapeList = &MapVLV(MapVLV.Add(Vi, Init));
    aShapeList->Append(Init);