Redesign of ParserItem so that its sum type nature no longer mandates
indirection, but rather a tag check on all operations. Because of this,
ParserRecords can now store ParserItems directly, and iterators are no
longer iterators to pointers, but behave like normal vector iterators.
Replaces the internal inheritance + unique_ptr scheme to a flat sum type
similar scheme that uses a tag to determine which operations are legal,
rather than using the indirection itself as a tag.
Reuses the original records string_view rather than expanding to the
same std::string, we save some allocations, memory cache misses and
simplify the class slightly.
Additionally, the uninteresting add-multiple-identical-records logic
ParserItem did before has been moved into RawRecord and is now performed
by std::deque (which also means it can allocate better for itself). The
addition of prepend deprecates push_front.
The accumulated strings are moved into RawRecords, which reduces
execution time (rough measurements indicates 4-8%). To facilitate this,
RawRecords are stored directly in the vector in favour of via
shared_ptrs.
Instead of going via an immediately discarded temporary, feed the
readValueToken result value directly to push_back. This enables rvalue
optimisations to kick in, when available.
The boost provided lexical cast are inefficient and is shown to be a
slowdown in the inner loop. Replaces them with std::atoi/std::atof and
some simple correctness checking.
Modifies RawRecord to internally use string_view instead of copies of
the substrings. This *vastly* reduces copying in the processing of each
record and subsequently improves performance. Reduces total memory usage
in Deck construction.
Since the Deck* family of classes have changed their interfaces to no
longer use shared_ptr, a lot of code broke. This patch fixes all
problems in tests, other signatures and accesses to now use the new Deck
interfaces.
Every header is self-contained and includes only what it must to
function, relying on users include what they need in source files,
adopting a pay-what-you-use model (in particular for internal
dependencies).
This is an effort to improve build performance. Several includes
scattered across the project are either unused or partially used (i.e.
just used to import a type name, not depending on the actual contents of
the header file).
Replaces a lot of these includes with forward declarations.
With this commit the generation of built in keywords is completely
changed. The most important changes include:
1) We have autogenerated a class for each keyword in the new
ParserKeywords { ... } namespace.
2) The autogenerated classes derive from ParserKeyword, and the
default constructor will build of a fully initialized
ParserKeyword instance, i.e. the keyword used to parse the EQUIL
keyword can be instantiated as simple as:
ParserKeywords::EQUIL kw;
3) The generated keywords have built in static constants for keyword
and item names, and item default values. That way it should be
possible for the compiler to catch trivial errors like trying to
access the keyword "PoRO"; also the the access to default values
means that properties can be initialized without actually
insantiating a DeckKeyword.
4) Two new classes Generator/KeywordLoader and
Generator/KeywordGenerator have been created, with the help of
these classes the keyword generation code is significantly
simplified.
... but throw later when trying to access the data of the item in
question.
Note that this was demanded by [at] joakim-hove and that I do not want
to be held responsible for any issues which are caused by this
approach. (read: please direct your barks to Joakim if you fell on
your nose because of this...)
this is just the result of
```
find -iname "*.[ch]pp" | xargs sed -i "s/ *$//"
find opm/parser/share/keywords -type f | xargs sed -i "s/ *$//"
```
so if it causes conflicts with other patches, the others should get
priority. The rationale behind this patch is that some people tell
their editor to remove white space which leads to larger than
necessary patches...
This allows to arbitrary characters like stars into strings. e.g.
MYKEYWORD
'123*456' 2*'Hello, World! (*)' /
is now a valid record with three strings while it threw an exception
before.
This patch works by transferring the removal of the quotes from the
RawDeck class to the readValueToken<T>() function which now has a
specialization for strings that deals with quotes. One small
complication is that the RawDeck needs to be adapted in order not to
split tokens in the middle of strings.
Finally, the StarToken class does not deal with the conversion from
string to the value type of the item anymore which allows it to become
a normal class instead of a template...
i.e. remove the defaultSet() method and its friends. this is required
to be able to specify defaults in DATA items like grid properties or
saturation tables. e.g.
SGL
10*0.1 10* 10*0.2 /
would not be possible without this. If no meaningful default for an
item is defined, float and double items get NaN, int items get -1 and
string ones get an empty string. The hope is that if these values get
used in the simulation, they will make the result obviously
incorrect. (Whether a data point of an item was defaulted can be
queried using item->defaultApplied(index).)
This is the first of several large commits changing several aspects of
the handling of defaults. The overall main purpose of these changes is
to protect against using non sensible values in the case the values have
not been sensibly specified in the deck. The changes in this commit
include:
1. The "default default" values to be used when an item without an
explicit default are defaulted are removed completely.
2. If a ParserItem is queried for a default value when no default has
been assigned it will raise an exception.
The purpose of this flag is to keep track of whether a keyword is
supposed to have only one element, i.e. scalar, or several. The
defaultApplied method only makes sense in the case of scalar items, this
method will now throw if it is called on a non-scalar item.
