Our definition of `toString` for the internal tuple objects we create during the
points-to analysis may have been a _tad_ too ambitious. In particular, it can
easily lead to non-termination, e.g. using the following piece of code:
```python
x = ()
while True:
x = (x, x)
```
This commit cuts off the infinite recursion by replacing _nested_ tuples with
the string "...". In particular this means even non-recursive tuples will be cut
off at that point, so that the following tuples
```python
(1, "2")
((3, 4), [5, 6])
(1, 2, 3, 4, 5)
```
Get the following string representations.
```
"(int 1, '2', )"
"(..., List, )"
"(int 1, int 2, int 3, 2 more...)"
```
They are not tainted in assignment, only in use.
I also adopted an attempt at a better test-setup, where it's easy to see if
everything is the way you hoped for, instead of browsing through 100 of lines of
taint-step output :P
We designed the IR's `DataFlow::Node.asExpr` very carefully so that it's
suitable for taint tracking, but then we didn't use it in
`DefaultTaintTracking.qll`. This meant that the sources in
`ArithmeticWithExtremeValues.ql` didn't get associated with any
`Instruction` and thus didn't propagate anywhere.
With this commit, the mapping of `Expr`-based sources to IR data-flow
nodes uses `asExpr`.
This commit undoes the code sharing between `TranslatedAssignExpr` (`=`)
and `TranslatedAssignOperation` (`+=`, `<<=`, ...). In the next commit,
when we change how the `Load` works on the LHS of
`TranslatedAssignOperation`, these classes will become so different that
sharing is no longer helpful.
Previously, the `Load` would be associated with the `CrementOperation`
rather than its operand, which gave surprising results when mapping
taint sinks back to `Expr`.
The changes in `raw_ir.expected` are to add `Copy` operations on the
`x++` in code like `y = x++`. This is now needed because the result that
`x++` would otherwise have (the Load) no longer belongs to the `++`
expression. Copies are inserted to ensure that all expressions are
associated with an `Instruction` result.
The changes in `*aliased_ssa_ir.expected` appear to be just wobble.
