Next: , Previous: SL syntax, Up: Compiler

4.14 Grammar record

The grammar record is the result of compilation of a grammar file, and represents all the information contained in the grammar file. The XDK makes use of two different kinds of grammar records:

  1. stateless grammar records
  2. stateful grammar records
Stateless grammar records can be saved to disk. Before the XDK solver can actually use the grammar for solving, it has to be transformed into a stateful grammar record (an extension of the stateless grammar record).

4.14.1 Stateless grammar record

Below, we display the type definition of the stateless grammar record: 

     grammar(
        usedDIDAs: DIDAs
        allDIDAs: DIDAs
        dIDADimension1Rec: DIDASLERec
        pnPrinciple1Rec: PnSLERec
        usedPns: Pns
        chosenPns: Pns
        pnCAPriITups: PnCAITups
        dIDAUsedOnsRec: DIDAOnsRec
        onOutput1Rec: OnSLERec
        chosenOns: Ons
        usedOns: Ons
        aEntriesRec: ASLERec
        as: As
        entriesI: I
        entry1Tn: Tn
        nodeTn: Tn
        node1Tn: Tn
        tnTypeRec: TnILTCoRec)

The value of the usedDIDAs feature is a list of dimension identifiers (DIDAs) which are the used dimensions.

The value of the allDIDAs feature is a list of dimension identifiers (DIDAs) which are all the dimensions defined in the grammar.

The value of the dIDADimension1Rec feature is a record mapping dimension identifiers (DIDA) to encoded dimension definitions (SLE).

The value of the pnPrinciple1Rec feature is a record mapping principle names (Pn) to encoded principle uses (SLE). A principle name is a unique name for a principle chosen on a dimension.1

The value of the usedPns feature is a list of principle names (Pns) which are the used principles.

The value of the chosenPns feature is a list of principle names (Pns) which are all the principles chosen in the grammar.2

The value of the pnCAPriITups feature is a list of tuples Pn#CA#I of a principle name (Pn), a principle constraint name CA, and a priority (I). These are all principle constraints.

The value of the dIDAUsedOnsRec feature is a record mapping dimension identifiers (DIDA) to the output names of the outputs used used on that dimension. An output name is a unique name for an output chosen on a dimension.3

The value of the onOutput1Rec feature is a record mapping output names (On) to encoded output definitions (SLE).

The value of the chosenOns feature is a list of the chosen output names.

The value of the usedOns feature is a list of the used output names.

The value of the aEntriesRec feature is a record mapping words (A) to their corresponding encoded lexical entries (SLE). This record is empty if the lexicon is stored into a database.

The value of the as feature is the list of all words (As) in the lexicon.

The value of the entriesI feature is an integer denoting the number of lexical entries of the grammar.

The value of the entry1Tn feature is the type name (Tn) corresponding to the type of a lexical entry.

The value of the nodeTn feature is the type name (Tn) corresponding to the type of a node record, including the features word, index, nodeSet and entryIndex, and excluding the subrecords corresponding to the dimensions.

The value of the node1Tn feature is the type name (Tn) corresponding to the type of a complete node record, including the features of the type nodeTn plus the attrs, entry and model features from the individual dimensions.

The value of the tnTypeRec feature is a record mapping type names (Tn) to their corresponding types (ILTCo).

Notice that ILTCo types are different form IL types in various respects:

4.14.2 Stateful grammar record

Below, we display the type definition of the features extending a stateless grammar record into a stateful grammar record: 

     grammar(
        dIDADimensionRec: DIDASLCRec
        dIDA2AttrsLat: DIDA2Lat
        dIDA2EntryLat: DIDA2Lat
        dIDA2LabelLat: DIDA2Lat
        pnPrincipleRec: PnSLCRec
        pn2Principle: Pn2SLC
        pn2ModelLat: Pn2Lat
        pn2DIDA: Pn2DIDA
        pnIsActive: PnIsActive
        procProcPnCAPriITups: ProcProcPnCAITups
        onOutputRec: OnSLCRec
        onOutputTups: OnSLCTups
        checkAsInEntries: As2U
        as2ABRec: As2ABRec
        as2AEntriesRec: As2AEntriesRec
        entry1Lat: Lat
        nodeLat: Lat
        node1Lat: Lat)

The value of the dIDADimensionRec feature is a record mapping dimension identifiers (DIDA) to compiled dimension definitions (SLC).

The value of the dIDA2AttrsLat feature is a function from dimension identifiers (DIDA) to the lattice corresponding to the attributes type on that dimension (Lat).

The value of the dIDA2EntryLat feature is a function from dimension identifiers (DIDA) to the lattice corresponding to the entry type on that dimension (Lat).

The value of the dIDA2LabelLat feature is a function from dimension identifiers (DIDA) to the lattice corresponding to the label type on that dimension (Lat).

The value of the pnPrincipleRec feature is a record mapping principle names (Pn) to compiled principle uses (SLC).

The value of the pn2Principle feature is a function from principle names (Pn) to compiled principle uses (SLC).

The value of the pn2ModelLat feature is a function from principle names (Pn) to the lattice corresponding to the type of the model record introduced by that principle (Lat).

The value of the pn2DIDA feature is a function from principle names (Pn) to dimension identifiers corresponding to the dimension on which the principle is used (DIDA).

The value of the pnIsActive feature is the function PnIsActive: Pn UsedDIDAs UsedPns -> B, returning for principle Pn, whether it is active given used dimension IDs UsedDIDAs and used principle names UsedPns.

The value of the procProcPnCAPriITups feature is a list of tuples Proc#Proc1#Pn#CA#I of a constraint procedure (Proc), a profile procedure (Proc1), a principle name (Pn), a principle constraint name (CA) and a priority (I). These are all principle constraints.

The value of the onOutputRec feature is a record mapping output names (On) to compiled output uses (SLC).

The value of the onOutputTups feature is a list of tuples On#Output of an output name (On) and an output (Output). These tuples represent the mapping already contained in the value of the onOutputRec feature. The difference is that they are ordered alphabetically (by their name and their dimension name) to match the order of the outputs in the Outputs menu in the GUI.

The value of the checkAsInEntries feature is a function from lists of words (As) to unit (U), raising an exception if any word A in As is not contained in the lexicon.

The value of the as2ABRec feature is a function from lists of words (As) to records mapping words (A) to bool (B). For all A in As, B is true if A is contained in the lexicon, and false otherwise.

The value of the as2Entries feature is a function from lists of words (As) to records mapping words (A) to lists of compiled lexical entries for that word (Entries).

The value of the entry1Lat feature is the lattice (Lat) corresponding to the type of a lexical entry.

The value of the nodeLat feature is the lattice (Lat) corresponding to the type of a node record, including the features word, index, nodeSet and entryIndex, and excluding the subrecords corresponding to the dimensions.

The value of the node1Lat feature is the lattice (Lat) corresponding to the type of a node record, including the features word, index, nodeSet and entryIndex, and the subrecords corresponding to the dimensions.

Footnotes

[1] Contrary to principle names, principle identifiers are not unique because the same principle can be used on several dimensions.

[2] The distinction between chosen and used principles makes sense for the graphical user interface: here, the Principles pull-down menu displays all chosen principles, of which only the selected principles are actually used for solving.

[3] Like principles, contrary to output names, output identifiers are not unique because the same output can be chosen on several dimensions.