Implement CQL2-JSON AST for datetime index selection

[YuriZmytrakov]

This PR implements a CQL2 Abstract Syntax Tree (AST) structure for CQL2 Filters to enable better parameter access for index selection in search queries:

• Introduces a structured AST representation of CQL2 queries (replacing dict-based processing)
• Adds datetime extraction logic to easily identify and access datetime parameters
• Provides ability for selecting appropriate indexes such as datetime, start_datetime, end_datetime
  The AST-based approach makes it easier to parse CQL2 filters, particularly for identifying datetime constraints (datetime, start_datetime, end_datetime) to optimize search performance through proper index selection.

Examples of AST nodes:
AND:

LogicalNode(op=<LogicalOp.AND: 'and'>, children=[ComparisonNode(op=<ComparisonOp.EQ: '='>, field='collection', value='landsat-8-l1'), 
                                                 ComparisonNode(op=<ComparisonOp.GT: '>'>, field='properties.cloud_cover', value=50)])

OR:

LogicalNode(op=<LogicalOp.OR: 'or'>, children=[ComparisonNode(op=<ComparisonOp.EQ: '='>, field='properties.platform', value='landsat-8'), 
                                               ComparisonNode(op=<ComparisonOp.EQ: '='>, field='properties.platform', value='sentinel-2')])

NOT:

LogicalNode(op=<LogicalOp.NOT: 'not'>, children=[ComparisonNode(op=<ComparisonOp.EQ: '='>, field='properties.cloud_cover', value=0)])

EQ (=):

ComparisonNode(op=<ComparisonOp.EQ: '='>, field='properties.cloud_cover', value=20)

NEQ (<>)

ComparisonNode(op=<ComparisonOp.NEQ: '<>'>, field='properties.platform', value='landsat-8')

LT (<)

ComparisonNode(op=<ComparisonOp.LT: '<'>, field='properties.cloud_cover', value=30)

LTE (<=)

ComparisonNode(op=<ComparisonOp.LTE: '<='>, field='properties.cloud_cover', value=50)

GT (>)

ComparisonNode(op=<ComparisonOp.GT: '>'>, field='properties.gsd', value=10)

GTE (>=)

ComparisonNode(op=<ComparisonOp.GTE: '>='>, field='properties.cloud_cover', value=0)

IS_NULL

ComparisonNode(op=<ComparisonOp.IS_NULL: 'isNull'>, field='properties.cloud_cover', value=None)

LIKE

AdvancedComparisonNode(op=<AdvancedComparisonOp.LIKE: 'like'>, field='id', value='LC08%')

BETWEEN

AdvancedComparisonNode(op=<AdvancedComparisonOp.BETWEEN: 'between'>, field='properties.cloud_cover', value=(10, 50))

IN

AdvancedComparisonNode(op=<AdvancedComparisonOp.IN: 'in'>, field='properties.platform', value=['landsat-8', 'sentinel-2', 'modis'])

S_INTERSECTS

SpatialNode(op=<SpatialOp.S_INTERSECTS: 's_intersects'>, field='geometry', geometry={'type': 'Polygon', 'coordinates': [[[-110, 40], [-110, 41], [-109, 41], [-109, 40], [-110, 40]]]})

S_CONTAINS

SpatialNode(op=<SpatialOp.S_CONTAINS: 's_contains'>, field='geometry', geometry={'type': 'Point', 'coordinates': [-110.5, 40.5]})

S_WITHIN

SpatialNode(op=<SpatialOp.S_WITHIN: 's_within'>, field='geometry', geometry={'type': 'Polygon', 'coordinates': [[[-120, 35], [-120, 45], [-100, 45], [-100, 35], [-120, 35]]]})

S_DISJOINT

SpatialNode(op=<SpatialOp.S_DISJOINT: 's_disjoint'>, field='geometry', geometry={'type': 'Polygon', 'coordinates': [[[0, 0], [0, 10], [10, 10], [10, 0], [0, 0]]]})

DateTimeRange

LogicalNode(op=<LogicalOp.AND: 'and'>, children=[ComparisonNode(op=<ComparisonOp.GTE: '>='>, field='properties.datetime', value='2024-01-01T00:00:00Z'), 
                                                 ComparisonNode(op=<ComparisonOp.LTE: '<='>, field='properties.datetime', value='2024-12-31T23:59:59Z')])

DateTimeExactNode

ComparisonNode(op=<ComparisonOp.EQ: '='>, field='properties.datetime', value='2024-06-15T10:30:00Z')

PR Checklist:

• Code is formatted and linted (run pre-commit run --all-files)
• Tests pass (run make test)
• Documentation has been updated to reflect changes, if applicable
• Changes are added to the changelog

[Gomez324]

Hey, I ran one test with this query and body:
127.0.0.1:8080/v1/search

{
  "filter-lang": "cql2-json",
  "filter": {
    "op": "and",
    "args": [
      {
        "op": "=",
        "args": [{ "property": "collection" }, "sentinel-2-l2a"]
      },
      {
        "op": ">=",
        "args": [{ "property": "datetime" }, "2015-01-01T00:00:00Z"]
      },
      {
        "op": "<=",
        "args": [{ "property": "datetime" }, "2024-12-31T23:59:59Z"]
      }
    ]
  }
}

And I got that index_param has this value. it looks like it's not selecting temporal indices. Please add tests for temporal indices.

(Pdb) index_param
'items_*,-*kibana*,-collections*'

[YuriZmytrakov]

Hey, I ran one test with this query and body: 127.0.0.1:8080/v1/search

{
  "filter-lang": "cql2-json",
  "filter": {
    "op": "and",
    "args": [
      {
        "op": "=",
        "args": [{ "property": "collection" }, "sentinel-2-l2a"]
      },
      {
        "op": ">=",
        "args": [{ "property": "datetime" }, "2015-01-01T00:00:00Z"]
      },
      {
        "op": "<=",
        "args": [{ "property": "datetime" }, "2024-12-31T23:59:59Z"]
      }
    ]
  }
}

And I got that index_param has this value. it looks like it's not selecting temporal indices. Please add tests for temporal indices.

(Pdb) index_param
'items_*,-*kibana*,-collections*'

@Gomez324 I have fixed selection of collection IDs from the CQL2-JSON query, only then are the indexes selected correctly, it needs to have as input variables collection_ids + datetime. For the example query above, I can see that the following index is correctly selected:

(Pdb) index_param
'items_sentinel-2-l2a_2015-07-04-2017-04-16'

I’ve also added a test for the index_select function.

[Gomez324]

I created a somewhat more complex filter:

{
    "filter-lang": "cql2-json",
    "filter": {
        "op": "or",
        "args": [
            {
                "op": "and",
                "args": [
                    { "op": "=", "args": [{"property": "collection"}, "sentinel-2-l2a"] },
                    { "op": ">=", "args": [{"property": "datetime"}, "2010-01-01T00:00:00Z"] },
                    { "op": "<=", "args": [{"property": "datetime"}, "2025-12-31T23:59:59Z"] }
                ]
            },
            {
                "op": "and",
                "args": [
                    { "op": "=", "args": [{"property": "collection"}, "sentinel-2-l1c"] },
                    { "op": ">=", "args": [{"property": "datetime"}, "2012-01-01T00:00:00Z"] },
                    { "op": "<=", "args": [{"property": "datetime"}, "2025-12-31T23:59:59Z"] }
                ]
            }
        ]
    }
}

so sentinel-2-l2a from 2010-2025 and sentinel-2-l1c from 2012-2025.

But when I debug, I see:

(Pdb) datetime_search
{'gte': '2012-01-01T00:00:00.0Z', 'lte': '2025-12-31T23:59:59.0Z'}

So the filter is being parsed incorrectly. it's merging the datetime ranges instead of keeping them separate per collection.