Node query processing steps

Steps in processing various model queries about nodes.

Of particular interest are cases where a LHSinferred node name is provided in a nodes argument to a step where it might be parsed and evaluated in a vars2... environment. Either LHSinferred names must be ensured safe or that processing path must be prevented.

Model methods and relevance

Notes include which other major methods, or modelDef or modelDef$maps data or methods, are used.

isEndNode

• Uses expandNodeNames(...returnType = 'ids') | modelDef$maps$isEndNode_byGID.

getNodeType

• Uses modelDef$nodeName2GraphIDs

getDeclID

• Uses modelDef$nodeName2GraphIDs

getDeclInfo

getUnrolledIndicesList

getDistribution

• Uses expandNodeNames | getDeclnfo

• Circuitous: expandNodeNames goes to graphIDs and back to names. getDeclInfo goes to getDeclID which goes back to graphIDs in order to declIDs

getParamExpr

getValueExpr

isMultivariate

• Uses expandNodeNames | getDistribution

• This and several others are circuitous (back and forth between graphIDs and names multiple times)

isDiscrete

isBinary

• Uses expandNodeNames | getDistribution, getParamExpr

isStoch

• Uses expandNodeNames | getNodeType

isDeterm

isTruncated

• Uses expandNodeNames | getDeclnfo

isUnivariate

• Uses expandNodeNames | getDistribution

getDimension

getVarNames

getNodeFunctions

• Uses modelDef$nodeName2GraphIDs | modelDef$graphIDs2indexedNodeInfo

getNodeNames

• Previous: Uses isDataFromGraphID(modelDef$maps$graphIDs) twice (note this is on all graphIDs)

• Previous: Uses expandNodeNames(modelDef$maps$graphID_2_nodeName[validValues],...)

• New (fixIssue340): filters out LHSinferred types from the begining. Uses new expandNodeNamesFromGraphIDs to avoid ID -> name -> ID -> name where the middle "name -> ID" could break from a LHSinferred node.

expandNodeNames

• Uses modelDef$nodeName2GraphIDs | modelDef$maps$elementNames[graphID], modelDef$maps$graphID_2_nodeName[graphID]

• Will not get vertexIDs, so will not return vertex names, hence will not return LHSinferred names.

topologicallySortNodes

• Uses expandNodeNames(..., returnType = 'ids') | expandNodeNames (not a typo)

getVarInfo

getSymbolTable

init_isDataEnv

resetData

setData

testDataFlags

• Uses getNodeNames() | eval(parse(...), envir = isDataEnv)

isData

• Uses modelDef$nodeName2GraphIDs | isDataFromGraphID

• This is safe because no vertexIDs (hence no LHSinferred nodes) are passed to isDataFromGraphID

isDataFromGraphID

• Uses modelDef$maps$graphID_2_nodeName | eval(parse(...), envir = isDataEnv), so cannot accept a vertex name such as a split LHSinferred name.

getDependenciesList provides access to modelDef$maps$edgesFrom2To, sorted and/or named

• Uses modelDef$maps$graphID_2_nodeName

• Should be safe.

• No longer used anywhere, so could be deprecated.

getDependencyPathCountOneNode

• Uses modelDef$nodeName2GraphIDs | modelDef$maps$nimbleGraph$getDependencyPathCountOneNode

• Should be safe because latter function dives into C++ getDependencyPathCountOneNode, which should handle any index.

• Used only from conjugacyRelationshipsClass::checkConjugacy

getDependencies

• Uses modelDef$nodeName2GraphIDs | modelDef$maps$elementID_2_vertexID | modelDef$maps$nimbleGraph$getDependencies | modelDef$maps$types, isDataFromGraphID, modelDef$nodeName2GraphIDs(modelDef$maps$graphID_2_nodeName[depIDs], modelDef$maps$elementNames, modelDef$maps$nodeNames

getDownstream

setInits

checkConjugacy

checkBasics

Underlying modelDef methods and data used by above

nodeName2GraphIDs returns either graphIDs (nodeFunctions and RHSonly) (if nodeFunctionID == TRUE) or elementIDs.

• Calls parseEval... steps so cannot take LHSinferred names as input.

• Uses maps$vars2GraphID_functions_and_RHSonly or maps$vars2ID_elements, so returns either graphIDs or elementIDs (not vertexIDs)