TUFLOW initialiser: The maintained initialiser is incorporated into
ryan-tools, retainingtuflow-initialiser init|build|catchments. #312 coordinates development and engineering acceptance; #247 tracks the authorised archive/private closeout. The separate July 2027 review tracks conditional deletion readiness.
ryan-tools is a collection of Python utilities for TUFLOW, RORB, 12D, GDAL, QGIS and general data-processing
workflows. Reusable code lives in ryan_library; files under ryan-scripts are human-facing wrappers and standalone
utilities.
The distribution is named ryan-tools, while maintained imports use ryan_library. The wheel also installs
ryan-scripts beside ryan_library, so MCP workflows do not require a Git checkout.
The PyPI and local-wheel installers automatically migrate the old distribution. When using pip directly, run
python -m pip uninstall ryan_functions before installing ryan-tools. The removed ryan_functions import alias must be replaced with ryan_library.functions.
See the migration policy.
The ryan_library.scripts import namespace and its six root-level forwarding aliases were retired
on 18 October 2026. Import maintained modules directly instead (breaking change).
Python 3.14 or newer remains the supported development, installation and CI baseline. The BAT installers can bypass pip's Python-version metadata for older Python 3 interpreters; this does not establish runtime compatibility. Maintained library code, orchestrators and unversioned wrappers follow the current repository standards; older standalone, versioned and compatibility files remain where they still support migration or narrow workflows.
Start with:
- Development guide for architecture, code categories, lifecycle terms and validation.
- Environment guide for Python, VS Code, installed-wheel and QGIS/OSGeo4W setup.
- Ryan Scripts guide for choosing and safely running scripts.
- Maintained wrapper standard when changing a library-backed wrapper.
- Project documentation index for the complete topic map, including specialised guides kept beside scripts, processors, examples and maintenance tools.
- Examples for direct library use when an existing wrapper is not the right fit.
ryan-tools/
|-- ryan_library/ # Maintained Python package
| |-- classes/ # Configuration, metadata and filename parsing
| |-- functions/ # Reusable algorithms and focused I/O helpers
| |-- orchestrators/ # Complete workflow controllers
| `-- processors/ # Stateful processors for supported result formats
|-- ryan-scripts/ # Human-facing wrappers and standalone utilities
|-- docs/ # Development guidance, plans and setup documentation
|-- examples/ # Example notebooks and supporting demonstrations
|-- repo-scripts/ # Build, environment and repository maintenance tools
|-- tests/ # Unit, integration and regression tests
| `-- test_data/ # Required synthetic test-data submodule
|-- excel-resources/ # Excel workbook resources submodule
|-- qgis-resources/ # QGIS resources submodule
|-- unsorted/ # Separate holding-area submodule
|-- pyproject.toml # Package metadata and tool configuration
`-- setup.py # Setuptools hook that stages QGIS styles into wheels
The repository uses Git submodules and Git LFS for optional workbook resources. Do not use recursive submodule checkout: the unsorted submodule is excluded from all automated work. Initialise only the required, maintained submodules.
git clone https://github.com/Chain-Frost/ryan-tools.git
cd ryan-toolsSource review, documentation and static checks need no submodules. Initialise the maintained fixtures only when the selected test or build needs them. The hosted test suite uses:
git submodule update --init tests/test_data qgis-resourcesInitialise excel-resources separately when working with its workbook resources. For source-only automation where Git
LFS is installed, set GIT_LFS_SKIP_SMUDGE=1 for checkout and submodule commands, then restore the prior environment
setting afterwards. This leaves LFS pointers available without downloading optional binaries.
Git LFS binary downloads are not required for routine source and CI checks. Install Git LFS when you need an LFS-tracked workbook, then retrieve only the resources required for that task (for example, git -C excel-resources lfs pull --include="*.xlsx"). Avoid unconditional git lfs pull in every checkout: repeated downloads consume the account's LFS bandwidth allowance.
The test suite requires tests/test_data; it does not download or substitute those fixtures automatically. Because
run_hy8 and culvert_solver are independent PyPI dependencies (run-hy8 and ryan-culverts), not Git submodules or bundled code. Install ryan-tools with pip to resolve supported dependencies automatically. Python 3.14 or newer is the supported runtime for ryan-tools and its published dependencies; bypassing Requires-Python during installation does not make older interpreters compatible.
Install the repository requirements into the user's normal Python 3.14 installation. ryan-tools does not require or
assume that users know how to create or activate a virtual environment:
py -3.14 -m pip install --upgrade pip
py -3.14 repo-scripts\install_latest_wheel.py --dependencies-only
py -3.14 repo-scripts\vendor_pytuflow_wheel.py
py -3.14 -m pip install --force-reinstall --no-deps -r repo-scripts\pytuflow-requirement.txt
py -3.14 -m pip install -r requirements.txtrequirements.txt installs the checkout and development tools into that Python installation. Installing the project is
important when running wrappers copied outside the repository because they import the shared implementation from the
installed ryan-tools distribution. The dependency bootstrap installs binary Fiona, Rasterio and GDAL packages
from the configured geospatial wheel index, avoiding a local source build on Windows.
The parser-only replacement handles incompatible PyTUFLOW builds with the same
version number; the following requirements install resolves dependencies normally.
See installation verification
for the dependency and public-API checks.
After changing ryan_library or package metadata, rebuild from the repository root:
python repo-scripts/build_library.pyBy default, the build script advances the normalized calendar version in pyproject.toml. Use --no-bump to rebuild
and verify the current version without changing it, --skip-pip when the build dependency is already installed, or
--skip-artifacts in an environment that cannot create or retain wheel artifacts. The wheel is built in temporary
storage, checked against the project metadata and bundled resource sources, and only then promoted into dist/; a
failed build or verification leaves the previous wheel and package version intact. Wheel builds require the QGIS
resource submodule because setup.py stages the pinned TUFLOW QML styles into the package.
To re-check the retained wheel without rebuilding it:
python repo-scripts/verify_wheel.pyWindows convenience entry points are:
.\install-ryan-tools.bat
.\build-and-install-local-wheel.bat
.\install-local-wheel.batinstall-ryan-tools.bat installs the latest published ryan-tools[mcp] release
directly from PyPI, along with upstream QGIS MCP and the locally vendored PyTuflow dependency wheel
built from exact contributor commit 4ca2dbe35c1d proposed in upstream PR #17. The verified
dependency wheel is distinct from the built ryan-tools wheel.
It keeps the binary GIS bootstrap, replaces incompatible same-version PyTUFLOW
builds, and checks installed dependencies and the public validation API before
reporting success. The local-wheel installer uses the same dependency setup and
verification, including the installed parser wheel's source path and SHA-256. Plain pip install ryan-tools
does not discover the wheel stored alongside these repository scripts: use the supplied installer or
the repository requirements when the accepted fork is needed. If Codex is detected, the PyPI installer
configures the QGIS and ryan-tools MCP servers only as needed, preserving other
settings and backing up any changed Codex configuration. install-from-pypi.bat is the explicit PyPI entry point;
install-ryan-tools.bat forwards to it.
The installers select the highest installed Python 3 release through the Windows launcher; override with RYAN_TOOLS_PYTHON to select a particular interpreter. All pip installation steps use --ignore-requires-python so Python 3.13 or another Python 3 version can attempt installation despite package metadata. Python versions below 3.14 are unsupported and may fail on dependencies, imports or runtime features. The PyPI installer also recognises the openai.chatgpt Codex extension in VS Code, even without the Codex CLI.
For developers, build-and-install-local-wheel.bat builds and installs a local wheel, including
installing or updating its dependencies and the QGIS MCP bridge.
build-and-force-reinstall-local-wheel.bat builds locally and forcibly reinstalls only that wheel
with --no-deps, leaving dependency installation to the normal installer.
Both use the local dist/ wheel for ryan-tools; dependencies may still be downloaded.
install-local-wheel.bat installs an existing local wheel without rebuilding.
force-reinstall-local-wheel.bat reinstalls that wheel with --no-deps.
build-local-wheel.bat builds and verifies a local wheel without installing it.
The ambiguous installer and packager filenames have been renamed; update saved shortcuts accordingly.
QGIS itself and its QGIS MCP desktop plugin must be installed separately.
This project is licensed under the Sustainable Use License v1.0 (SUL-1.0).
The software may be used and modified for personal, non-commercial and internal business purposes, including commercial professional and consulting work where the software itself is not provided as a commercial product or service. Commercial distribution, incorporation into software supplied commercially to third parties, or provision of the software or its functionality as a paid hosted service, SaaS or API is not permitted under the licence.
Use focused pytest commands for a bounded change. Keep temporary files under the repository on this Windows checkout:
python -m pytest tests\path\to\test_file.py --basetemp=.pytest_cache\basetempRun the complete suite through the repository runner:
cmd.exe /C repo-scripts\run_tests.batThe runner uses installed PyPI hydraulic dependencies and a repository-local base temporary directory, and generates terminal, HTML and XML coverage reports. See the development guide for proportional validation expectations.
| Need | Start here |
|---|---|
| Run or adapt a human-facing script | ryan-scripts/README.md |
| Process TUFLOW results with maintained wrappers | ryan-scripts/TUFLOW-python/README.md |
| Extend the TUFLOW processor framework | ryan_library/processors/tuflow/README.md |
| Use reusable Python APIs directly | examples/README.md |
| Run maintained GDAL workflows | ryan-scripts/gdal-python/README.md |
| Convert geospatial, point-cloud, CAD or model files | Supported file converters and formats |
| Map a remaining legacy GDAL BAT file to its replacement | ryan-scripts/gdal-bat/README.md |
| Connect the repository MCP server | docs/MCP_SETUP.md |
TuflowStringParser extracts the configured data type and run metadata from a result path:
from pathlib import Path
from ryan_library.classes.tuflow_string_classes import TuflowStringParser
parser = TuflowStringParser(Path("M11_01p_00120m_TP01_1d_Q.csv"))
print(parser.data_type)
print(parser.raw_run_code)
print(parser.clean_run_code)
print(parser.aep)
print(parser.duration)
print(parser.tp)The authoritative suffix registry is
ryan_library/classes/tuflow_results_validation_and_datatypes.json.
It contains both processor-backed tabular types and raster classifications used for discovery.
load_tuflow_data() handles discovery, logging, serial or parallel processing and optional location filtering:
from ryan_library.functions.tuflow.notebook_helpers import load_tuflow_data
collection = load_tuflow_data(
paths=["results"],
data_types=["Q", "V", "H", "Nmx", "Cmx", "Chan", "EOF"],
parallel=True,
locations=["Culvert_01"],
)
timeseries = collection.combine_1d_timeseries()
maximums = collection.combine_1d_maximums()Processor-backed types currently include POMM, PO, Cmx, Nmx, Chan, ccA, RLL_Qmx, Q, H, CF,
V, EOF and TLF. Use the processor development notes for combination behavior, caching and extension guidance.
Orchestrators are callable workflow controllers; wrappers add editable defaults, CLI handling, banners, exit codes and optional pauses:
from pathlib import Path
from ryan_library.orchestrators.tuflow.tuflow_culverts_merge import main_processing
main_processing(
paths_to_process=[Path("results")],
include_data_types=["Nmx", "Cmx", "Chan", "ccA", "RLL_Qmx", "EOF"],
locations_to_include=["Culvert_01"],
output_dir=Path("outputs"),
export_mode="both",
)Representative orchestrators cover culvert maximums and time series, PO/POMM combination, closure durations, peak and
stability checks, TUFLOW log summaries and result styling. The corresponding maintained wrappers are under
ryan-scripts/TUFLOW-python; use that folder's README and each wrapper's --help output as the current interface.
RORB hydrograph discovery and parsing:
from pathlib import Path
from ryan_library.functions.RORB.read_rorb_files import find_batch_files, parse_batch_output
batch_files = find_batch_files([Path("rorb_outputs")])
runs = [parse_batch_output(path) for path in batch_files]12D culvert export processing:
from pathlib import Path
from ryan_library.functions.process_12D_culverts import get_combined_df_from_files
culverts = get_combined_df_from_files(Path("12d_exports"))For GDAL raster conversion, mosaics, flood extents, footprints, metadata and point-cloud conversion, use the
maintained GDAL Python wrappers. The BAT files that remain under
ryan-scripts/gdal-bat are legacy migration references and should not be selected for new work.
The library provides Python-native and GDAL-backed conversion functions, batch orchestrators, and maintained wrappers across geospatial, point-cloud, CAD, hydrologic, and tabular formats:
| Domain | Source formats | Destination formats | Key library APIs and entry points | Notes |
|---|---|---|---|---|
| Raster translation & compression | Any GDAL raster (.flt, .asc, .rst, .xyz, .tif, .ecw, etc.) |
.tif (GeoTIFF), .ovr (pyramid overviews) |
translate_to_geotiff(), convert_rasters(), gdal_translate_TIF_ovr.py |
Lossless conversion with choice of tuflow (DEFLATE) or efficient (tiled ZSTD) profiles. |
| Raster mosaics & virtual datasets | Multiple raster tiles (.tif, .xyz, .asc, .flt) |
.tif (GeoTIFF mosaic), .vrt (VRT dataset) |
merge_directory(), create_grouped_mosaics(), gdal_merge.py, build_VRT.py |
Supports vector extent clipping, assigned CRS/NoData, and grouped TUFLOW result sets. |
| Vector dataset translation | .gpkg, .shp, .fgb, .geojson, .sqlite |
.gpkg (GeoPackage), .shp (ESRI Shapefile), .fgb (FlatGeobuf), .geojson (GeoJSON), .sqlite (SQLite) |
translate_vector_dataset(), require_vector_driver() |
Atomic vector translation via GDAL; multi-layer support for GPKG and SQLite. |
| File Geodatabase export | .gdb (ESRI File Geodatabase directories) |
.gpkg, .shp, .fgb, .geojson, .sqlite |
export_file_geodatabase(), discover_file_geodatabases() |
Batch GDB discovery; exports each layer to separate files or one multi-layer database. |
| Raster to vector classification | Depth, water level, or DEM rasters (.tif, .ecw, etc.) |
.gpkg, .shp, .geojson, .fgb, .sqlite (flood extent polygons, raster footprints) |
main_processing(), polygonize_flood_extent(), gdal_flood_extent.py, gdal_raster_footprint.py |
Thresholding, band selection, optional GDAL sieve filtering, and boundary vectorization. |
| LiDAR & point clouds | .laz (compressed LiDAR) |
.las (uncompressed ASPRS LAS point clouds) |
convert_laz_to_las(), convert_laz_directory(), laz_to_las.py |
Memory-efficient chunked streaming preserving headers, point formats, scales, offsets, and VLRs. |
| Elevation rasters to tables/points | .tif (GeoTIFF DEMs) |
Tabular X, Y, Z DataFrames, .csv, .xyz, .las |
read_geotiff(), tile_data(), thin-raster-terrain-for-12D_v5.py, tif-to-LAS-valid-only_v6.py |
Spatial tiling, NoData masking, and point thinning for 12D Model or tabular workflows. |
| CAD, mining & 12D geometry | .str (Surpac binary/text strings), .dtm (Surpac binary/text DTM meshes), .dxf (polyface meshes), 12D culvert export text |
.gpkg (3D lines & polygon meshes), .parquet (geometry tables), DataFrames |
get_combined_df_from_files(), dtm_str_converter_to_gpkg.py, extract_dxf_polyface.py, polyface_parquet_to_gpkg.py |
Translates mining CAD and 12D export formats into modern GIS/Parquet geometries. |
| Hydrologic & model results | TUFLOW 1D/2D CSVs (_POMM, _PO, _Cmx, _Nmx, _Chan, _ccA, _RLL_Qmx, _Q, _H, etc.), .tlf logs, RORB .out files |
.xlsx (multi-sheet styled workbooks), .parquet, .csv, DataFrames |
load_tuflow_data(), parse_batch_output(), ExcelExporter, save_to_excel() |
Ingests simulation outputs into consolidated tabular datasets, summary reports, and Excel exports. |
The resource submodules contain project templates and application assets:
excel-resources/workbooks/: Excel templates for hydrology, frequency and culvert workflows.qgis-resources/processing-models/tuflow/: QGIS models for common TUFLOW input layers.qgis-resources/processing-models/tuflow/supporting-workbooks/: formula-driven supporting workbooks used by relevant models.qgis-resources/styles/: QML styles, QPT layouts and supporting spatial assets.qgis-resources/scripts/: QGIS Python console and PyQGIS utilities.
Except for the TUFLOW QML files staged into wheel builds, treat these resources as project templates rather than files installed with the Python package. The parent repository pins each submodule to a reviewed commit.
- Agent instructions contain the repository-wide Code Review Rules used for Codex GitHub reviews, alongside mandatory automated-agent guidance.
- Pull request template records summaries, validation and review checklists.
- Development guide is the canonical architecture and validation reference.
- MCP setup documents focused inspection tools and staged CLI workflow discovery for AI clients.