tilebench

Inspect HEAD/LIST/GET requests withing Rasterio

Source Code: https://github.com/developmentseed/tilebench

Inspect HEAD/LIST/GET requests withing Rasterio.

Note: In GDAL 3.2, logging capabilities for /vsicurl, /vsis3 and the like was added (ref: https://github.com/OSGeo/gdal/pull/2742).

Install

You can install tilebench using pip

$ pip install -U pip
$ pip install -U tilebench

or install from source:

$ git clone https://github.com/developmentseed/tilebench.git
$ cd tilebench
$ pip install -U pip
$ pip install -e .

API

from tilebench import profile
import rasterio

[@profile](https://github.com/profile)()
def info(src_path: str):
    with rasterio.open(src_path) as src_dst:
        return src_dst.meta

meta = info("https://noaa-eri-pds.s3.amazonaws.com/2022_Hurricane_Ian/20221002a_RGB/20221002aC0795145w325100n.tif")

> 2022-10-25T00:20:24.215385+0200 | TILEBENCH | {"LIST": {"count": 0}, "HEAD": {"count": 1}, "GET": {"count": 1, "bytes": 32768, "ranges": ["0-32767"]}, "Timing": 0.8705799579620361}

from tilebench import profile
from rio_tiler.io import Reader

[@profile](https://github.com/profile)()
def _read_tile(src_path: str, x: int, y: int, z: int, tilesize: int = 256):
    with Reader(src_path) as cog:
        return cog.tile(x, y, z, tilesize=tilesize)

img = _read_tile(
    "https://noaa-eri-pds.s3.amazonaws.com/2022_Hurricane_Ian/20221002a_RGB/20221002aC0795145w325100n.tif",
    9114,
    13216,
    15,
)

> 2022-10-25T00:21:32.895752+0200 | TILEBENCH | {"LIST": {"count": 0}, "HEAD": {"count": 1}, "GET": {"count": 2, "bytes": 409600, "ranges": ["0-32767", "32768-409599"]}, "Timing": 1.2970409393310547}

Command Line Interface (CLI)

$ tilebench --help
Usage: tilebench [OPTIONS] COMMAND [ARGS]...

  Command line interface for the tilebench Python package.

Options:
  --help  Show this message and exit.

Commands:
  get-zooms  Get Mercator Zoom levels.
  profile    Profile COGReader Mercator Tile read.
  random     Get random tile.
  viz        WEB UI to visualize VSI statistics for a web mercator tile request

Examples

$ tilebench get-zooms https://noaa-eri-pds.s3.amazonaws.com/2022_Hurricane_Ian/20221002a_RGB/20221002aC0795145w325100n.tif | jq
{
  "minzoom": 14,
  "maxzoom": 19
}

$ tilebench random https://noaa-eri-pds.s3.amazonaws.com/2022_Hurricane_Ian/20221002a_RGB/20221002aC0795145w325100n.tif --zoom 15
15-9114-13215

$ tilebench profile https://noaa-eri-pds.s3.amazonaws.com/2022_Hurricane_Ian/20221002a_RGB/20221002aC0795145w325100n.tif --tile 15-9114-13215 --config GDAL_DISABLE_READDIR_ON_OPEN=EMPTY_DIR | jq
{
  "LIST": {
    "count": 0
  },
  "HEAD": {
    "count": 1
  },
  "GET": {
    "count": 2,
    "bytes": 409600,
    "ranges": [
      "0-32767",
      "32768-409599"
    ]
  },
  "Timing": 1.2364399433135986
}

$ tilebench profile https://noaa-eri-pds.s3.amazonaws.com/2022_Hurricane_Ian/20221002a_RGB/20221002aC0795145w325100n.tif --tile 15-9114-13215 --config GDAL_DISABLE_READDIR_ON_OPEN=FALSE | jq
{
  "LIST": {
    "count": 1
  },
  "HEAD": {
    "count": 8
  },
  "GET": {
    "count": 3,
    "bytes": 409600,
    "ranges": [
      "0-32767",
      "32768-409599"
    ]
  },
  "Timing": 2.2018940448760986
}

Starlette Middleware

Warning: This is highly experimental and should not be used in production (https://github.com/developmentseed/tilebench/issues/6)

In addition of the viz CLI we added a starlette middleware to easily integrate VSI statistics in your web services.

from fastapi import FastAPI

from tilebench.middleware import VSIStatsMiddleware

app = FastAPI()
app.add_middleware(VSIStatsMiddleware)

The middleware will add a vsi-stats entry in the response headers in form of:

vsi-stats: list;count=1, head;count=1, get;count=2;size=196608, ranges; values=0-65535|65536-196607

Some paths may be excluded from being handeld by the middleware by the exclude_paths argument:

app.add_middleware(VSIStatsMiddleware, exclude_paths=["/foo", "/bar"])

GDAL config options

• GDAL_DISABLE_READDIR_ON_OPEN: Allow or Disable listing of files in the directory (e.g external overview)
• GDAL_INGESTED_BYTES_AT_OPEN: Control how many bytes GDAL will ingest when opening a dataset (useful when a file has a big header)
• CPL_VSIL_CURL_ALLOWED_EXTENSIONS: Limit valid external files
• GDAL_CACHEMAX: Cache size
• GDAL_HTTP_MERGE_CONSECUTIVE_RANGES
• VSI_CACHE
• VSI_CACHE_SIZE

See the full list at https://gdal.org/user/configoptions.html

Internal tiles Vs Mercator grid

$ tilebench viz https://noaa-eri-pds.s3.amazonaws.com/2022_Hurricane_Ian/20221002a_RGB/20221002aC0795145w325100n.tif --config GDAL_DISABLE_READDIR_ON_OPEN=EMPTY_DIR

Blue lines represent the mercator grid for a specific zoom level and the red lines represent the internal tiles bounds

We can then click on a mercator tile and see how much requests GDAL/RASTERIO does.

Docker

Ready to use docker image can be found on Github registry.

• https://github.com/developmentseed/tilebench/pkgs/container/tilebench

docker run \
  --volume "$PWD":/data \
  --platform linux/amd64 \
  --rm -it -p 8080:8080 ghcr.io/developmentseed/tilebench:latest \
  tilebench viz --host 0.0.0.0 https://noaa-eri-pds.s3.us-east-1.amazonaws.com/2020_Nashville_Tornado/20200307a_RGB/20200307aC0865700w360900n.tif

Contribution & Development

Issues and pull requests are more than welcome.

dev install

$ git clone https://github.com/developmentseed/tilebench.git
$ cd tilebench
$ pip install -e .[dev]

pre-commit

This repo is set to use pre-commit to run isort, flake8, pydocstring, black ("uncompromising Python code formatter") and mypy when committing new code.

$ pre-commit install

$ git add .

$ git commit -m'my change'
isort....................................................................Passed
black....................................................................Passed
Flake8...................................................................Passed
Verifying PEP257 Compliance..............................................Passed
mypy.....................................................................Passed

$ git push origin

License

See LICENSE

Authors

See contributors for a listing of individual contributors.

Changes

See CHANGES.md.

Changelog

0.8.2 (2022-11-21)

• update hatch config

0.8.1 (2022-10-31)

• fix issue with min/max zoom when there is no overviews
• calculate windows from block_shapes

0.8.0 (2022-10-25)

• update rio-tiler/rasterio dependencies
• remove python 3.7 support
• add python 3.10 support
• add image endpoint to show the data footprint
• switch from mapbox to maplibre

0.7.0 (2022-06-14)

• add cProfile stats

0.6.1 (2022-04-19)

• Remove usage of VSIStatsMiddleware in tilebench viz

0.6.0 (2022-04-19)

• switch to pyproject.toml

0.5.1 (2022-03-04)

• make sure we don't cache previous request when using tilebench profile without --tile option

0.5.0 (2022-02-28)

• update rio-tiler requirement
• add reader option

0.4.1 (2022-02-14)

• update Fastapi requirement
• use WarpedVRT to get dataset bounds in epsg:4326

0.4.0 (2021-12-13)

• update rio-tiler's version requirement
• add more information about the raster in the Viz web page (author @drnextgis, https://github.com/developmentseed/tilebench/pull/14)
• fix bug for latest GDAL/rasterio version
• add default STAMEN basemap in viz and remove mapbox token/style options.
• update fastapi requirement

0.3.0 (2021-03-05)

• add exclude_paths options in VSIStatsMiddleware to exclude some endpoints (author @drnextgis, https://github.com/developmentseed/tilebench/pull/10)
• renamed ressources to resources

0.2.1 (2021-02-19)

• fix typo in UI

0.2.0 (2021-01-28)

• add warp-kernels in output in profile CLI
• add rasterio/curl stdout in output
• add dataread time in Viz

0.1.1 (2021-01-27)

• update requirements

0.1.0 (2021-01-04)

• add web UI for VSI stats visualization
• add starlette middleware

0.0.2 (2020-12-15)

• Update for rio-tiler==2.0.0rc3

0.1.0 (2020-07-13)

• Initial release