Paginated query results after first page do not gracefully fallback to GET on HTTP-405 response #846
Description
I've come across an issue with trying to retrieve multi-page results (i.e. when using pystac-client.Client.search with limit=5 and the number of matched items is >5) and not sure if it's been covered before in past issues. Since the retry-on-405 strategy does seem to be covered in the pystac-client test suite I'm confused why I'm hitting it here.
Minimum working example:
from pystac_client import Client, __version__ as libver
print(f"Using pystac-client v{libver}")
stac_url = 'https://www.eodms-sgdot.nrcan-rncan.gc.ca/stac'
catalog = Client.open(stac_url)
print(f"{stac_url} contains 'rel' link with type 'search': {'search' in [link.rel for link in catalog.links]}")
print(f"{stac_url} conforms to ITEM_SEARCH: {catalog.conforms_to('ITEM_SEARCH')}")returns
Using pystac-client v0.9.0
https://www.eodms-sgdot.nrcan-rncan.gc.ca/stac contains 'rel' link with type 'search': True
https://www.eodms-sgdot.nrcan-rncan.gc.ca/stac conforms to ITEM_SEARCH: True
which to me means that the pystac_client.Client.search method will used as per docs.
Moving on to the actual searching, I expect the following query parameters to return 8 results.
bbox = [-105.40747286434971, 68.70958793236613, -97.90802872229544, 71.11275348080056]
time_window = "2025-10-27/2025-10-28"
limit = 5and since I do not specify the method= argument, the default request type is POST. The result of the following block fails with the subsequent APIError
query = catalog.search(
collections=['rcm'],
bbox=bbox,
datetime=time_window,
limit=limit
)
print(f"Query is using {query.method} method")
results = []
# I would normally use enumerate(query.pages()) here but I need to show in this issue that the first page is returned/parsed properly
page_num = 1
for page in query.pages():
print(f"Reading result page #{page_num}", end= " ", flush=True)
results.extend(page.items)
print("Done!")
page_num += 1
print(f"Query returned {len(results)} matches")Query is using POST method
Reading result page #1 Done!
<traceback snip>
APIError: {"detail":"Method Not Allowed"}
</traceback snip>
but what's interesting is that it DOES parse the first result page properly (i.e. properly retrying with GET) before failing on the second page. I would expect the query.pages() iterator to properly fallback to GET for all pages but it seems to me that it isn't?
The error can be completely avoided by using method="GET" in the pystac_client.Client.search call:
query = catalog.search(
collections=['rcm'],
bbox=bbox,
datetime=time_window,
limit=limit,
method="GET" # <---- explicit setting to GET
)
print(f"Query is using {query.method} method")
results = []
page_num = 1
for page in query.pages():
print(f"Reading result page #{page_num}", end= " ", flush=True)
results.extend(page.items)
print("Done!")
page_num += 1
print(f"Query returned {len(results)} matches")returns
Query is using GET method
Reading result page #1 Done!
Reading result page #2 Done!
Query returned 8 matches
but it doesn't seem like a good solution when there is existing logic in pystac-client to gracefully handle the HTTP-405 response from the endpoint.