ili2ogc

A pure-Python INTERLIS 2 toolkit: parses .ili models into a graph of typed Python objects (the IlisMeta16 metamodel), validates .xtf data transfers against them, and converts both to JSON Schema, SQL DDL (PostgreSQL/ GeoPackage), OGC JSON-FG, CQL2-JSON filters, and back to .xtf. TRANSLATION OF models are supported throughout via --lang (renames identifiers/keys in the output, the transfer wire format itself is unaffected).

uv sync          # or: pip install .

The generated ANTLR lexer/parser are committed under src/interlis/antlr/, so no grammar build step is needed.

interlis build model.ili --repo models/                                 # parse + print the built model
interlis validate transfer.xtf --repo models/                           # check .xtf against its .ili schema
interlis convert model.ili -o model.schema.json                         # .ili -> JSON Schema
interlis convert-sql model.ili --dialect postgresql -o model.sql        # .ili -> SQL DDL
interlis convert-jsonfg transfer.xtf --repo models/ -o out.jsonfg.json   # .xtf -> OGC JSON-FG
interlis convert-cql2 model.ili -o model.cql2.json                      # CONSTRAINT -> CQL2-JSON filters
interlis write-xtf view-model.ili source.xtf -o view.xtf                # VIEW TOPIC data -> .xtf

--repo DIR (repeatable) resolves IMPORTS/schema references against a directory of .ili files. Run interlis <command> --help for every option.

from pathlib import Path
from interlis.runtime.parse import parse_file
from interlis.builder.model_builder import InterlisModelBuilder
from interlis.xtf.parse import parse_xtf
from interlis.xtf.validate import validate_transfer

tree, syntax_errors = parse_file(Path("model.ili"))
builder = InterlisModelBuilder(Path("mappings"), Path("spec/grammar/mapping"))
model = builder.build(tree)  # graph of Pydantic objects, e.g. model.Element[0].Name

transfer = parse_xtf(Path("transfer.xtf"))
issues = validate_transfer(transfer, symbol_table=builder.symbol_table)

interlis.convert.jsonschema/.sql/.jsonfg/.cql2/.xtf_writer expose the same conversions as the CLI's convert/convert-sql/convert-jsonfg/ convert-cql2/write-xtf. Pass a ModelRepository (interlis.builder.repository) to the builder to resolve IMPORTS against other local .ili models.

uv run pytest                 # test suite
uv run ruff check src/ tests/ # lint (also enforced in CI)
uv run black src/ tests/      # format

CI (.github/workflows/ci.yml) runs ruff, black --check and the test suite on Python 3.10/3.12/3.14.

spec/grammar/mapping/*.yml and mappings/ilismeta16-*.yml (the IlisMeta16 metamodel) are the executable source of truth: InterlisModelBuilder is a generic engine driven by this data, not hand-written parsing code per grammar rule. Regenerating the vendored grammar under vendor/interlis-antlr4 is a maintainer-only step, needed only when that submodule changes.