IS NULL filter operator

[jamesfisher-geo]

Describe the bug
IS NULL is included in the filter extension logic. However, I cannot get it to work with cql2-json or cql2-text.

stac-fastapi-elasticsearch-opensearch/stac_fastapi/core/stac_fastapi/core/extensions/filter.py

Line 74 in 2d6cb4d

IS_NULL = "isNull"

To Reproduce
Steps to reproduce the behavior:

1. docker compose up app-opensearch or docker compose up app-elasticsearch
2. python3 data_loader.py --base-url http://localhost:8082
3. Try the following:
   POST http://localhost:8082/search
   body

{
  "filter-lang": "cql2-json",
  "filter": {
            "op": "isNull",
            "args": [ { "property": "sentinel:data_coverage" } ]
          }

}

{
    "detail": "Error with cql2_json filter: Q() can only accept dict with a single query ({\"match\": {...}}). Instead it got ({})"
}

GET http://localhost:8082/search?filter=sentinel:data_coverage > 50 OR landsat:coverage_percent < 10 OR (sentinel:data_coverage IS NULL AND landsat:coverage_percent IS NULL)

{
    "detail": "Error with cql2_json filter: Q() can only accept dict with a single query ({\"match\": {...}}). Instead it got ({})"
}

Expected behavior
valid ItemCollection

[jamesfisher-geo]

For full implementation we are waiting on a new version of pygeofilter

[jamesfisher-geo]

It is possible that we could migrate the cql2 parsing logic to use cql-rs Python bindings

[louisstuart96]

In

stac-fastapi-elasticsearch-opensearch/stac_fastapi/core/stac_fastapi/core/extensions/filter.py

Lines 135 to 142 in e45cfea

	elif query["op"] in [
	ComparisonOp.EQ,
	ComparisonOp.NEQ,
	ComparisonOp.LT,
	ComparisonOp.LTE,
	ComparisonOp.GT,
	ComparisonOp.GTE,
	]:

:
It seems ComparisonOp.IS_NULL is absent in the list, so that the following if query["op"] == ComparisonOp.IS_NULL: fails. Could this be the problem?

[louisstuart96]

Possible quick fix:

def to_es(query: Dict[str, Any]) -> Dict[str, Any]:
    """
    Transform a simplified CQL2 query structure to an Elasticsearch compatible query DSL.

    Args:
        query (Dict[str, Any]): The query dictionary containing 'op' and 'args'.

    Returns:
        Dict[str, Any]: The corresponding Elasticsearch query in the form of a dictionary.
    """
    if query["op"] in [LogicalOp.AND, LogicalOp.OR, LogicalOp.NOT]:
        bool_type = {
            LogicalOp.AND: "must",
            LogicalOp.OR: "should",
            LogicalOp.NOT: "must_not",
        }[query["op"]]
        return {"bool": {bool_type: [to_es(sub_query) for sub_query in query["args"]]}}

    elif query["op"] == ComparisonOp.IS_NULL:

        field = to_es_field(query["args"][0]["property"])
        return {"bool": {"must_not": {"exists": {"field": field}}}}

    elif query["op"] in [
        ComparisonOp.EQ,
        ComparisonOp.NEQ,
        ComparisonOp.LT,
        ComparisonOp.LTE,
        ComparisonOp.GT,
        ComparisonOp.GTE,
    ]:
        field = to_es_field(query["args"][0]["property"])
        value = query["args"][1]
        if isinstance(value, dict) and "timestamp" in value:
            # Handle timestamp fields specifically
            value = value["timestamp"]

        if query["op"] == ComparisonOp.EQ:
            return {"term": {field: value}}
        elif query["op"] == ComparisonOp.NEQ:
            return {"bool": {"must_not": [{"term": {field: value}}]}}
        else:
            range_op = {
                    ComparisonOp.LT: "lt",
                    ComparisonOp.LTE: "lte",
                    ComparisonOp.GT: "gt",
                    ComparisonOp.GTE: "gte",
                }[query["op"]]
            return {"range": {field: {range_op: value}}}

    elif query["op"] == AdvancedComparisonOp.BETWEEN:
        field = to_es_field(query["args"][0]["property"])
        gte, lte = query["args"][1], query["args"][2]
        if isinstance(gte, dict) and "timestamp" in gte:
            gte = gte["timestamp"]
        if isinstance(lte, dict) and "timestamp" in lte:
            lte = lte["timestamp"]
        return {"range": {field: {"gte": gte, "lte": lte}}}

    elif query["op"] == AdvancedComparisonOp.IN:
        field = to_es_field(query["args"][0]["property"])
        values = query["args"][1]
        if not isinstance(values, list):
            raise ValueError(f"Arg {values} is not a list")
        return {"terms": {field: values}}

    elif query["op"] == AdvancedComparisonOp.LIKE:
        field = to_es_field(query["args"][0]["property"])
        pattern = cql2_like_to_es(query["args"][1])
        return {"wildcard": {field: {"value": pattern, "case_insensitive": True}}}

    elif query["op"] == SpatialIntersectsOp.S_INTERSECTS:
        field = to_es_field(query["args"][0]["property"])
        geometry = query["args"][1]
        return {"geo_shape": {field: {"shape": geometry, "relation": "intersects"}}}

    return {}