pip install aiocogeo

# With S3 filesystem
pip install aiocogeo[s3]

COGs are opened using the COGReader asynchronous context manager:

from aiocogeo import COGReader

async with COGReader("http://cog.tif") as cog:
    ...

Several filesystems are supported:

• HTTP/HTTPS (http://, https://)
• S3 (s3://)
• File (/)

Generating a rasterio-style profile for the COG:

async with COGReader("https://async-cog-reader-test-data.s3.amazonaws.com/lzw_cog.tif") as cog:
    print(cog.profile)

>>> {'driver': 'GTiff', 'width': 10280, 'height': 12190, 'count': 3, 'dtype': 'uint8', 'transform': Affine(0.6, 0.0, 367188.0,
       0.0, -0.6, 3777102.0), 'blockxsize': 512, 'blockysize': 512, 'compress': 'lzw', 'interleave': 'pixel', 'crs': 'EPSG:26911', 'tiled': True, 'photometric': 'rgb'}

A COG is composed of several IFDs, each with many TIFF tags:

from aiocogeo.ifd import IFD
from aiocogeo.tag import Tag

async with COGReader("https://async-cog-reader-test-data.s3.amazonaws.com/lzw_cog.tif") as cog:
    for ifd in cog:
        assert isinstance(ifd, IFD)
        for tag in ifd:
            assert isinstance(tag, Tag)

Each IFD contains more granular metadata about the image than what is included in the profile. For example, finding the tilesize for each IFD:

async with COGReader("https://async-cog-reader-test-data.s3.amazonaws.com/lzw_cog.tif") as cog:
    for ifd in cog:
        print(ifd.TileWidth.value, ifd.TileHeight.value)

>>> 512 512
    128 128
    128 128
    128 128
    128 128
    128 128

More advanced use cases may need access to tag-level metadata:

async with COGReader("https://async-cog-reader-test-data.s3.amazonaws.com/lzw_cog.tif") as cog:
    first_ifd = cog.ifds[0]
    assert first_ifd.tag_count == 24

    for tag in first_ifd:
        print(tag)

>>> Tag(code=258, name='BitsPerSample', tag_type=TagType(format='H', size=2), count=3, length=6, value=(8, 8, 8))
    Tag(code=259, name='Compression', tag_type=TagType(format='H', size=2), count=1, length=2, value=5)
    Tag(code=257, name='ImageHeight', tag_type=TagType(format='H', size=2), count=1, length=2, value=12190)
    Tag(code=256, name='ImageWidth', tag_type=TagType(format='H', size=2), count=1, length=2, value=10280)
    ...

The reader also has methods for reading internal image tiles and performing partial reads. Currently only jpeg, lzw, deflate, packbits, and webp compressions are supported.

Reading the top left tile of an image at native resolution:

async with COGReader("https://async-cog-reader-test-data.s3.amazonaws.com/webp_cog.tif") as cog:
    x = y = z = 0
    tile = await cog.get_tile(x, y, z)

    ifd = cog.ifds[z]
    assert tile.shape == (ifd.bands, ifd.TileHeight.value, ifd.TileWidth.value)

You can read a portion of the image by specifying a bounding box in the native crs of the image and an output shape:

async with COGReader("https://async-cog-reader-test-data.s3.amazonaws.com/webp_cog.tif") as cog:
    assert cog.epsg == 26911
    partial_data = await cog.read(bounds=(368461,3770591,368796,3770921), shape=(512,512))

If the COG has an internal mask, the returned array will be a masked array:

import numpy as np

async with COGReader("https://async-cog-reader-test-data.s3.amazonaws.com/naip_image_masked.tif") as cog:
    assert cog.is_masked

    tile = await cog.get_tile(0,0,0)
    assert np.ma.is_masked(tile)

Configuration options are exposed through environment variables:

• INGESTED_BYTES_AT_OPEN - defines the number of bytes in the first GET request at file opening (defaults to 16KB)
• HEADER_CHUNK_SIZE - chunk size used to read header (defaults to 16KB)
• ENABLE_BLOCK_CACHE - determines if image blocks are cached in memory (defaults to TRUE)
• ENABLE_HEADER_CACHE - determines if COG headers are cached in memory (defaults to TRUE)
• HTTP_MERGE_CONSECUTIVE_RANGES - determines if consecutive ranges are merged into a single request (defaults to FALSE)
• BOUNDLESS_READ - determines if internal tiles outside the bounds of the IFD are read (defaults to TRUE)
• BOUNDLESS_READ_FILL_VALUE - determines the value used to fill boundless reads (defaults to 0)
• LOG_LEVEL - determines the log level used by the package (defaults to ERROR)
• VERBOSE_LOGS - enables verbose logging, designed for use when LOG_LEVEL=DEBUG (defaults to FALSE)
• AWS_REQUEST_PAYER - set to requester to enable reading from S3 RequesterPays buckets.
• ZOOM_LEVEL_STRATEGY - mimics GDAL's ZOOM_LEVEL_STRATEGY creation option:
  • AUTO or 50 (default) upsamples or downsamples the zoom level whose resolution is closest to the desired resolution.
  • LOWER or 100 always upsamples the zoom level immediately below the desired resolution (requesting less data).
  • UPPER or 0 always downsamples the zoom level immediately above the desired resoluion (requesting more data).
  • Another integer from 0 through 100: if the desired resolution more this percentage of the way from the zoom level immediately below to the zoom level immediately above, then upsample the zoom level immediately below, else downsample the zoom level immediately above. For example, 1 is the same as UPPER unless the COG's resolution is very close to the zoom level below e.g. due to floating point imprecision.

Refer to aiocogeo/config.py for more details about configuration options.

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

Options:
  --install-completion [bash|zsh|fish|powershell|pwsh]
                                  Install completion for the specified shell.
  --show-completion [bash|zsh|fish|powershell|pwsh]
                                  Show completion for the specified shell, to
                                  copy it or customize the installation.

  --help                          Show this message and exit.

Commands:
  create-tms  Create OGC TileMatrixSet.
  info        Read COG metadata.