This is mostly to have a general solution for matching region level
summary keywords which may reference a user-defined region set (FIP
keyword, e.g., FIPXYZ) through tags like
RPR__XYZ -- Average pressure in region, FIPXYZ region set
ROPR_XYZ -- Oil production rate in region, FIPXYZ region set
RODENXYZ -- Average oil density in region, FIPXYZ region set
The initial approach introduced in commit cfbafc236 was limited to
selected keywords.
To this end, add a new data member
std::string ParserKeyword::m_matchRegexSuffix
and introduce a new member function
bool ParserKeyword::matchesDeckNames()
which matches a candidate keyword string against the m_deckNames,
and, if applicable, as a regular expression against m_deckNames when
appended m_matchRegexSuffix.
The MULTREGT keyword has an independent way of defining the default
region--the default value of item 6--and does not need the default
GRIDOPTS-base region protocol of the other *REG keywords.
Mostly to group related functions and have the same order in the
declaration and the implementation files. While here, replace an
'enum' with a strong enum since the type does not need to support
arithmetic operations.
This commit adds a new, focused, member function
NumericalAquifers::allAquiferCellIds()
which returns a vector of those Cartesian/global cells that have
been marked as defining the model's numerical aquifers through the
AQUNUM keyword. We intend to use this to identify those NNCs that
go to numerical aquifers--or between numerical aquifer cells--as
those may need special treatment when processing the MULTREGT
keyword.
The 'base' checks that the input string looks like a valid keyword
and, if so, matches the string against the builtin set of known
keywords. The full keyword recognition process additionally
includes those keywords that match against keyword collections,
typically the SUMMARY section "meta" keywords in the *_PROBE files.
We will use the 'base' recognition separately to introduce support
for extension keywords with long--more than eight character--names.
This commit replaces the data members 'nx', 'ny', and 'nz' with a
copy of the GridDims object passed as an argument to the
constructor. In turn, this enables using GridDims::getIJK() to
compute the cell's IJK tuple from its global Cartesian index instead
of implementing the same calculation locally (and incorrectly).
While here, also include the layer index (K) in determining whether
or not a connection is an NNC, and do this just once instead of once
for each MULTREGT record. Cells (I,J,K) and (I,J,K+3) might be
connected across pinched-out layers and for the purposes of MULTREGT
that connection should be treated as an NNC.
In particular, ensure that we include all requisite headers, hide
type aliases that are not needed outside the class, consistently use
std::size_t, split a few long lines, and switch to range-for in one
location.
A numeric index serves the same purpose as a pointer in this very
specific instance and also has the benefit of being easily
transferable between processes. This, in turn, means we can remove
the 'getSearchMap()' and 'constructSearchMap()' member functions.
These are not directionally dependent, and they have a unit string
of 'Pressure'. This didn't use to matter, but upcoming changes
will depend on the unit string being correct.
