Allow building jq in a mode that doesn't use decnumber for benchmarking
purposes. decnumber support is enabled by default, and this option is
meant to be removed once we're happy with the performance.
Extend jv_number to use decNumber for storing number literals. Any math
operations on the numbers will truncate them to double precision.
Comparisons when both numbers are literal numbers will compare them
without truncation.
Delay conversion of numbers to doubles until a math operation is performed,
to preserve precision. A literal jv_number will only need conversion to
double once, and will reuse the resultant double on subsequent
conversions.
Outputting literal jv_numbers preserves the original precision.
Add strong pthread requirement to manage contexts/allocations for
converting numbers between their decNumber, string, and double formats.
Expressions of the form `path(EXPR) | select(GENERATOR)`, where `EXPR`
is a path expression and `GENERATOR` is a generator conditional
expression (e.g., `has("a"), has("b")`) cause an assertion if the
jq_state VM is torn down too soon. That assert() was only correct if
assuming that the conditional is not a generator.
If the conditional is generator, then what we see is that when
backtracking a SUBEXP_END is executed without a corresponding
SUBEXP_BEGIN because the entire conditional is bracketed with
SUBEXP_BEGIN and SUBEXP_END, and since it's resumed in the middle, in
between the brackets.
Rather than assert that the jq->path_len being restored has some
particular value, we can simply re-compute it from the restored
jq->path.
The behavior of memmem for an empty needle is inconsistent between
implementations of libc.
Our tests imply that we want an empty string needle to be true,
so check for an empty needle before calling memmem.
block_bind_incremental is block_bind_referenced in a loop backwards. For
an 1-inst block, it does the same thing and isn't too much more
expensive, so it's not really useful to keep both.
Also, block_bind_referenced was a better name for the function.
Since 605bfb3, block_drop_unreferenced actually reverses the order of
instructions in the block it's run against. This bug was hidden by the
fact that normally it's run *twice* against the main program, flipping
it back, and that order of function definitionss doesn't really matter
after symbol resolution.
Stop losing \0s from the end of files because we don't know how much
fgets() read into our buffer. If we prefill the buffer with anything
other than zeroes, we can actually tell the difference between the
terminator and unused space in the buffer.
This is possibly more expensive, but at least it should be correct.
Fixes#1504
Defaults the else clause to the identity filter "." if an else clause is
missing.
The following two jq programs are considered equivalent:
```
if .foo then
.foo = "bar"
else
.
end
```
```
if .foo then
.foo = "bar"
end
```
Avoid doing the internal binding of top-level symbols in the parser,
leaving that work to be done in a post-processing step. For builtins,
this lets us do a reference-aware bind step (block_bind_incremental)
*after* generating builtins/0.
Libraries are a bit trickier since they may be bound multiple times, so
instead of thinking through the implications I added (block_bind_self)
to resolve all internal symbols immediately.
Remove the special code which loads ~/.jq in builtin.c, and instead glue
an optional include which points to the same file onto the main program
in linker.c.
Fixes a minor bug where errors in ~/.jq would be labelled <builtin>.
Only the second and subsequent path components were being checked, which
I guess is theoretically security-relevant.
There's no apparent point to reconstructing the path after splitting it
by adding /s back in, either.
"Module path must be a string" is not a useful error message when the
reason the module path isn't a string is because the string it was got
replaced with an invalid with an error message for some other reason.
Also fixes a few memory leaks on early exits.
The mingw-w64 runtime expands wildcards, which causes a discrepancy
between `main`'s `argv` (multiple file names) and Windows' `wargv` (a
single wildcard). Use `wmain` to retrieve the wide character args.
Windows regards the `NUL` device as tty, which causes `WriteFile` to
fail (the bytes written pointer cannot be `NULL` in this case). Tweak
the color test to ensure tty is accurate.
Windows 10 supports color output, but it must be explicitly enabled;
ANSICON provides support for other versions. Enable color if either of
those are available. Resolves#1494.
The parser generates a call to error/0 on break. Having that exported
from C directly removes a language dependency on builtin.jq.
In any case, error/0 seems to be the more fundamental operation, seeing
as the old C impementation of error/1 did nothing useful with its input.
The distinction between input/0 and _input/0 has been unnecessary since
a9c613..7dc34b, which moved end-of-input handling into C and replaced
the jq definition of input with `def input: _input;`
If these values do not match exactly, it is because the UTF-8 is invalid
anyway and we counted codepoints differently than oniguruma did. Perhaps it
would be better to error out here, but at least one similar loop already
uses < vs != and since we're off the rails anyway this might be OK. It is
certainly better than overruning the buffer.
Resolves https://github.com/stedolan/jq/issues/1804
William Langford