generate component specific grammars

flopy4/mf6/codec/reader/__init__.py

@@ -2,7 +2,7 @@
 from pathlib import Path
 from typing import Any
 
-from flopy4.mf6.codec.reader.parser import make_basic_parser
+from flopy4.mf6.codec.reader.parser import get_basic_parser
 from flopy4.mf6.codec.reader.transformer import BasicTransformer
 
 
@@ -41,6 +41,6 @@ def loads(data: str) -> Any:
         Parsed MF6 input file structure
     """
 
-    parser = make_basic_parser()
+    parser = get_basic_parser()
     transformer = BasicTransformer()
     return transformer.transform(parser.parse(data))

flopy4/mf6/codec/reader/dfn2lark.py

@@ -0,0 +1,39 @@
+"""Convert (TOML/v2) DFNs to Lark grammars."""
+
+import argparse
+from os import PathLike
+from pathlib import Path
+
+from modflow_devtools.dfn import Dfn
+
+from flopy4.mf6.codec.reader.grammar import make_all_grammars
+
+_GRAMMAR_MODULE = Path(__file__).parent / "grammar"
+_GRAMMAR_GEN_DIR = _GRAMMAR_MODULE / "generated"
+
+
+def generate(dfndir: PathLike, outdir: PathLike):
+    """Generate lark grammars from DFNs."""
+    dfndir = Path(dfndir).expanduser().absolute()
+    outdir = Path(outdir).expanduser().absolute()
+    outdir.mkdir(exist_ok=True, parents=True)
+    dfns = Dfn.load_all(dfndir, version=2)
+    make_all_grammars(dfns, outdir)
+
+
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser(description="Generate lark grammars from DFNs.")
+    parser.add_argument(
+        "--dfndir",
+        "-d",
+        type=str,
+        help="Directory containing DFN files.",
+    )
+    parser.add_argument(
+        "--outdir",
+        "-o",
+        help="Output directory.",
+        default=_GRAMMAR_GEN_DIR,
+    )
+    args = parser.parse_args()
+    generate(args.dfndir, args.outdir)

flopy4/mf6/codec/reader/grammar/__init__.py

@@ -0,0 +1,46 @@
+from os import PathLike
+from pathlib import Path
+
+import jinja2
+from modflow_devtools.dfn import Dfn, get_blocks, get_fields
+
+from flopy4.mf6.codec.reader.grammar.filters import (
+    field_type,
+    keystring_children,
+    record_child_type,
+)
+
+
+def _get_template_env():
+    loader = jinja2.PackageLoader("flopy4", "mf6/codec/reader/grammar/templates/")
+    env = jinja2.Environment(
+        loader=loader,
+        trim_blocks=True,
+        lstrip_blocks=True,
+        keep_trailing_newline=True,
+    )
+    env.filters["field_type"] = field_type
+    env.filters["record_child_type"] = record_child_type
+    env.filters["keystring_children"] = keystring_children
+    return env
+
+
+def make_grammar(dfn: Dfn, outdir: PathLike):
+    """Generate a Lark grammar file for a single component."""
+    outdir = Path(outdir).expanduser().resolve().absolute()
+    env = _get_template_env()
+    template = env.get_template("component.lark.jinja")
+    target_path = outdir / f"{dfn['name']}.lark"
+    with open(target_path, "w") as f:
+        name = dfn["name"]
+        blocks = get_blocks(dfn)
+        fields = get_fields(dfn)
+        f.write(template.render(name=name, blocks=blocks, fields=fields))
+
+
+def make_all_grammars(dfns: dict[str, Dfn], outdir: PathLike):
+    """Generate grammars for all components."""
+    outdir = Path(outdir).expanduser().resolve().absolute()
+    outdir.mkdir(parents=True, exist_ok=True)
+    for dfn in dfns.values():
+        make_grammar(dfn, outdir)

flopy4/mf6/codec/reader/grammar/filters.py

@@ -0,0 +1,39 @@
+from modflow_devtools.dfn import Field
+
+
+def field_type(field: Field) -> str:
+    match field["type"]:
+        case t if t in ["string", "integer", "double precision"] and "shape" in field:
+            if "period" in field["block"]:
+                return "list"
+            return "array"
+        case "keyword":
+            return ""
+        case "keystring":
+            return ""  # keystrings generate their own union rules
+        case _:
+            return field["type"]
+
+
+def record_child_type(field: Field) -> str:
+    """Get the grammar type for a field within a record context."""
+    match field["type"]:
+        case "string":
+            return "string"
+        case "integer":
+            return "integer"
+        case "double precision":
+            return "double"
+        case "keyword":
+            return ""
+        case "keystring":
+            return ""  # keystrings generate their own union rules
+        case _:
+            return field["type"]
+
+
+def keystring_children(field: Field) -> dict:
+    """Get the children of a keystring field for union generation."""
+    if field["type"] != "keystring":
+        return {}
+    return field.get("children", {})

flopy4/mf6/codec/reader/grammar/generated/chf-cdb.lark

@@ -0,0 +1,22 @@
+// Auto-generated grammar for MF6 CHF-CDB
+%import "typed.lark"
+
+start: block*
+block: options_block | dimensions_block | period_block
+options_block: "begin"i "options"i options_fields "end"i "options"i
+dimensions_block: "begin"i "dimensions"i dimensions_fields "end"i "dimensions"i
+period_block: "begin"i "period"i period_fields "end"i "period"i
+options_fields: (auxiliary | boundnames | print_input | print_flows | save_flows | obs_filerecord)*
+dimensions_fields: (maxbound)*
+period_fields: (idcxs | width | aux | boundname)*
+auxiliary: "auxiliary"i array
+boundnames: "boundnames"i 
+print_input: "print_input"i 
+print_flows: "print_flows"i 
+save_flows: "save_flows"i 
+obs_filerecord: "obs6"i "filein"i string
+maxbound: "maxbound"i integer
+idcxs: "idcxs"i list
+width: "width"i list
+aux: "aux"i list
+boundname: "boundname"i list

flopy4/mf6/codec/reader/grammar/generated/chf-chd.lark

@@ -0,0 +1,23 @@
+// Auto-generated grammar for MF6 CHF-CHD
+%import "typed.lark"
+
+start: block*
+block: options_block | dimensions_block | period_block
+options_block: "begin"i "options"i options_fields "end"i "options"i
+dimensions_block: "begin"i "dimensions"i dimensions_fields "end"i "dimensions"i
+period_block: "begin"i "period"i period_fields "end"i "period"i
+options_fields: (auxiliary | auxmultname | boundnames | print_input | print_flows | save_flows | ts_filerecord | obs_filerecord)*
+dimensions_fields: (maxbound)*
+period_fields: (head | aux | boundname)*
+auxiliary: "auxiliary"i array
+auxmultname: "auxmultname"i string
+boundnames: "boundnames"i 
+print_input: "print_input"i 
+print_flows: "print_flows"i 
+save_flows: "save_flows"i 
+ts_filerecord: "ts6"i "filein"i string
+obs_filerecord: "obs6"i "filein"i string
+maxbound: "maxbound"i integer
+head: "head"i list
+aux: "aux"i list
+boundname: "boundname"i list

flopy4/mf6/codec/reader/grammar/generated/chf-cxs.lark

@@ -0,0 +1,18 @@
+// Auto-generated grammar for MF6 CHF-CXS
+%import "typed.lark"
+
+start: block*
+block: options_block | dimensions_block | packagedata_block | crosssectiondata_block
+options_block: "begin"i "options"i options_fields "end"i "options"i
+dimensions_block: "begin"i "dimensions"i dimensions_fields "end"i "dimensions"i
+packagedata_block: "begin"i "packagedata"i packagedata_fields "end"i "packagedata"i
+crosssectiondata_block: "begin"i "crosssectiondata"i crosssectiondata_fields "end"i "crosssectiondata"i
+options_fields: (print_input)*
+dimensions_fields: (nsections | npoints)*
+packagedata_fields: (packagedata)*
+crosssectiondata_fields: (crosssectiondata)*
+print_input: "print_input"i 
+nsections: "nsections"i integer
+npoints: "npoints"i integer
+packagedata: "packagedata"i recarray
+crosssectiondata: "crosssectiondata"i recarray

flopy4/mf6/codec/reader/grammar/generated/chf-dfw.lark

@@ -0,0 +1,20 @@
+// Auto-generated grammar for MF6 CHF-DFW
+%import "typed.lark"
+
+start: block*
+block: options_block | griddata_block
+options_block: "begin"i "options"i options_fields "end"i "options"i
+griddata_block: "begin"i "griddata"i griddata_fields "end"i "griddata"i
+options_fields: (central_in_space | length_conversion | time_conversion | save_flows | print_flows | save_velocity | obs_filerecord | export_array_ascii | dev_swr_conductance)*
+griddata_fields: (manningsn | idcxs)*
+central_in_space: "central_in_space"i 
+length_conversion: "length_conversion"i double precision
+time_conversion: "time_conversion"i double precision
+save_flows: "save_flows"i 
+print_flows: "print_flows"i 
+save_velocity: "save_velocity"i 
+obs_filerecord: "obs6"i "filein"i string
+export_array_ascii: "export_array_ascii"i 
+dev_swr_conductance: "dev_swr_conductance"i 
+manningsn: "manningsn"i array
+idcxs: "idcxs"i array

flopy4/mf6/codec/reader/grammar/generated/chf-disv1d.lark

@@ -0,0 +1,30 @@
+// Auto-generated grammar for MF6 CHF-DISV1D
+%import "typed.lark"
+
+start: block*
+block: options_block | dimensions_block | griddata_block | vertices_block | cell1d_block
+options_block: "begin"i "options"i options_fields "end"i "options"i
+dimensions_block: "begin"i "dimensions"i dimensions_fields "end"i "dimensions"i
+griddata_block: "begin"i "griddata"i griddata_fields "end"i "griddata"i
+vertices_block: "begin"i "vertices"i vertices_fields "end"i "vertices"i
+cell1d_block: "begin"i "cell1d"i cell1d_fields "end"i "cell1d"i
+options_fields: (length_units | nogrb | grb_filerecord | xorigin | yorigin | angrot | export_array_ascii | crs)*
+dimensions_fields: (nodes | nvert)*
+griddata_fields: (width | bottom | idomain)*
+vertices_fields: (vertices)*
+cell1d_fields: (cell1d)*
+length_units: "length_units"i string
+nogrb: "nogrb"i 
+grb_filerecord: "grb6"i "fileout"i string
+xorigin: "xorigin"i double precision
+yorigin: "yorigin"i double precision
+angrot: "angrot"i double precision
+export_array_ascii: "export_array_ascii"i 
+crs: "crs"i array
+nodes: "nodes"i integer
+nvert: "nvert"i integer
+width: "width"i array
+bottom: "bottom"i array
+idomain: "idomain"i array
+vertices: "vertices"i recarray
+cell1d: "cell1d"i recarray

flopy4/mf6/codec/reader/grammar/generated/chf-evp.lark

@@ -0,0 +1,23 @@
+// Auto-generated grammar for MF6 CHF-EVP
+%import "typed.lark"
+
+start: block*
+block: options_block | dimensions_block | period_block
+options_block: "begin"i "options"i options_fields "end"i "options"i
+dimensions_block: "begin"i "dimensions"i dimensions_fields "end"i "dimensions"i
+period_block: "begin"i "period"i period_fields "end"i "period"i
+options_fields: (auxiliary | auxmultname | boundnames | print_input | print_flows | save_flows | ts_filerecord | obs_filerecord)*
+dimensions_fields: (maxbound)*
+period_fields: (evaporation | aux | boundname)*
+auxiliary: "auxiliary"i array
+auxmultname: "auxmultname"i string
+boundnames: "boundnames"i 
+print_input: "print_input"i 
+print_flows: "print_flows"i 
+save_flows: "save_flows"i 
+ts_filerecord: "ts6"i "filein"i string
+obs_filerecord: "obs6"i "filein"i string
+maxbound: "maxbound"i integer
+evaporation: "evaporation"i list
+aux: "aux"i list
+boundname: "boundname"i list

flopy4/mf6/codec/reader/grammar/generated/chf-flw.lark

@@ -0,0 +1,23 @@
+// Auto-generated grammar for MF6 CHF-FLW
+%import "typed.lark"
+
+start: block*
+block: options_block | dimensions_block | period_block
+options_block: "begin"i "options"i options_fields "end"i "options"i
+dimensions_block: "begin"i "dimensions"i dimensions_fields "end"i "dimensions"i
+period_block: "begin"i "period"i period_fields "end"i "period"i
+options_fields: (auxiliary | auxmultname | boundnames | print_input | print_flows | save_flows | ts_filerecord | obs_filerecord)*
+dimensions_fields: (maxbound)*
+period_fields: (q | aux | boundname)*
+auxiliary: "auxiliary"i array
+auxmultname: "auxmultname"i string
+boundnames: "boundnames"i 
+print_input: "print_input"i 
+print_flows: "print_flows"i 
+save_flows: "save_flows"i 
+ts_filerecord: "ts6"i "filein"i string
+obs_filerecord: "obs6"i "filein"i string
+maxbound: "maxbound"i integer
+q: "q"i list
+aux: "aux"i list
+boundname: "boundname"i list

flopy4/mf6/codec/reader/grammar/generated/chf-ic.lark

@@ -0,0 +1,11 @@
+// Auto-generated grammar for MF6 CHF-IC
+%import "typed.lark"
+
+start: block*
+block: options_block | griddata_block
+options_block: "begin"i "options"i options_fields "end"i "options"i
+griddata_block: "begin"i "griddata"i griddata_fields "end"i "griddata"i
+options_fields: (export_array_ascii)*
+griddata_fields: (strt)*
+export_array_ascii: "export_array_ascii"i 
+strt: "strt"i array

flopy4/mf6/codec/reader/grammar/generated/chf-nam.lark

@@ -0,0 +1,15 @@
+// Auto-generated grammar for MF6 CHF-NAM
+%import "typed.lark"
+
+start: block*
+block: options_block | packages_block
+options_block: "begin"i "options"i options_fields "end"i "options"i
+packages_block: "begin"i "packages"i packages_fields "end"i "packages"i
+options_fields: (list | print_input | print_flows | save_flows | newtonoptions)*
+packages_fields: (packages)*
+list: "list"i string
+print_input: "print_input"i 
+print_flows: "print_flows"i 
+save_flows: "save_flows"i 
+newtonoptions: "newtonoptions"i record
+packages: "packages"i recarray

flopy4/mf6/codec/reader/grammar/generated/chf-oc.lark

@@ -0,0 +1,23 @@
+// Auto-generated grammar for MF6 CHF-OC
+%import "typed.lark"
+
+start: block*
+block: options_block | period_block
+options_block: "begin"i "options"i options_fields "end"i "options"i
+period_block: "begin"i "period"i period_fields "end"i "period"i
+options_fields: (budget_filerecord | budgetcsv_filerecord | qoutflow_filerecord | stage_filerecord | qoutflowprintrecord)*
+period_fields: (saverecord | printrecord)*
+budget_filerecord: "budget"i "fileout"i string
+budgetcsv_filerecord: "budgetcsv"i "fileout"i string
+qoutflow_filerecord: "qoutflow"i "fileout"i string
+stage_filerecord: "stage"i "fileout"i string
+qoutflowprintrecord: "qoutflow"i "print_format"i
+saverecord: "save"i string ocsetting
+printrecord: "print"i string ocsetting
+ocsetting: ocsetting_all | ocsetting_first | ocsetting_last | ocsetting_frequency | ocsetting_steps
+ocsetting_all: "all"i
+ocsetting_first: "first"i
+ocsetting_last: "last"i
+ocsetting_frequency: "frequency"i integer
+ocsetting_steps: "steps"i integer
+

flopy4/mf6/codec/reader/grammar/generated/chf-pcp.lark

@@ -0,0 +1,23 @@
+// Auto-generated grammar for MF6 CHF-PCP
+%import "typed.lark"
+
+start: block*
+block: options_block | dimensions_block | period_block
+options_block: "begin"i "options"i options_fields "end"i "options"i
+dimensions_block: "begin"i "dimensions"i dimensions_fields "end"i "dimensions"i
+period_block: "begin"i "period"i period_fields "end"i "period"i
+options_fields: (auxiliary | auxmultname | boundnames | print_input | print_flows | save_flows | ts_filerecord | obs_filerecord)*
+dimensions_fields: (maxbound)*
+period_fields: (precipitation | aux | boundname)*
+auxiliary: "auxiliary"i array
+auxmultname: "auxmultname"i string
+boundnames: "boundnames"i 
+print_input: "print_input"i 
+print_flows: "print_flows"i 
+save_flows: "save_flows"i 
+ts_filerecord: "ts6"i "filein"i string
+obs_filerecord: "obs6"i "filein"i string
+maxbound: "maxbound"i integer
+precipitation: "precipitation"i list
+aux: "aux"i list
+boundname: "boundname"i list

flopy4/mf6/codec/reader/grammar/generated/chf-sto.lark

@@ -0,0 +1,13 @@
+// Auto-generated grammar for MF6 CHF-STO
+%import "typed.lark"
+
+start: block*
+block: options_block | period_block
+options_block: "begin"i "options"i options_fields "end"i "options"i
+period_block: "begin"i "period"i period_fields "end"i "period"i
+options_fields: (save_flows | export_array_ascii)*
+period_fields: (steady-state | transient)*
+save_flows: "save_flows"i 
+export_array_ascii: "export_array_ascii"i 
+steady-state: "steady-state"i 
+transient: "transient"i