Merge upstream/main into fix/cql2-json-bbox-axis-order

Resolves the conflict in tests/parsers/cql2_text/test_parser.py (upstream's
NOT/paren precedence tests vs this stack's single-letter attribute tests);
both sets are kept. No change to the bbox axis-order fix itself.

Author: C1-BA-B1-F3

pygeofilter/parsers/cql2_text/grammar.lark

@@ -27,13 +27,17 @@
 
 ?start: condition
 
-?condition: condition_1
-          | condition "AND"i condition_1                                 -> and_
-          | condition "OR"i condition_1                                   -> or_
+// AND binds tighter than OR, and NOT tighter than both, as in booleanExpression of the CQL2 BNF
+?condition: condition_and
+          | condition "OR"i condition_and                                 -> or_
 
-?condition_1: predicate
-            | "NOT"i predicate                                               -> not_
-            | "NOT"i "(" predicate ")"                                       -> not_
+?condition_and: condition_1
+              | condition_and "AND"i condition_1                          -> and_
+
+?condition_1: condition_2
+            | "NOT"i condition_2                                             -> not_
+
+?condition_2: predicate
             | "(" condition ")"
 
 ?predicate: expression "=" expression                                       -> eq
@@ -135,7 +139,8 @@ bbox: "BBOX"i "(" full_number "," full_number "," full_number "," full_number ")
 BOOLEAN.2: ( "TRUE"i | "FALSE"i)
 
 DOUBLE_QUOTED: "\"" /.*?/ "\""
-SINGLE_QUOTED: "'" /.*?/ "'"
+// CQL2 escapes a quote as '' or \'
+SINGLE_QUOTED: /'(?:[^'\\]|\\.|'')*'/
 
 DATE: /[0-9]{4}-?[0-1][0-9]-?[0-3][0-9]/
 DATETIME: /[0-9]{4}-?[0-1][0-9]-?[0-3][0-9][T ][0-2][0-9]:?[0-5][0-9]:?[0-5][0-9](\.[0-9]+)?(Z|[+-][0-9]{2}:[0-9]{2})?/

pygeofilter/parsers/cql2_text/parser.py

@@ -27,6 +27,7 @@
 
 import logging
 import os.path
+import re
 
 from lark import Lark, logger, v_args
 
@@ -185,7 +186,13 @@ def DOUBLE_QUOTED(self, token):
         return token[1:-1]
 
     def SINGLE_QUOTED(self, token):
-        return token[1:-1]
+        # '' and \' stand for a quote, the other backslashes are kept for LIKE
+        return re.sub(
+            r"''|\\(.)",
+            lambda m: "'" if m[0] == "''" or m[1] == "'" else m[0],
+            token[1:-1],
+            flags=re.S,
+        )
 
     def geometry(self, value):
         return values.Geometry(value)

tests/parsers/cql2_text/test_parser.py

@@ -175,6 +175,36 @@ def test_string_not_like():
     )
 
 
+def test_string_doubled_quote():
+    result = parse("attr = 'it''s'")
+    assert result == ast.Equal(ast.Attribute("attr"), "it's")
+
+
+def test_string_backslash_quote():
+    result = parse(r"attr = 'it\'s'")
+    assert result == ast.Equal(ast.Attribute("attr"), "it's")
+
+
+def test_string_only_quotes():
+    assert parse("attr = ''") == ast.Equal(ast.Attribute("attr"), "")
+    assert parse("attr = ''''") == ast.Equal(ast.Attribute("attr"), "'")
+
+
+def test_strings_end_at_their_quote():
+    result = parse("attr = 'a' OR attr = 'b'")
+    assert result == ast.Or(
+        ast.Equal(ast.Attribute("attr"), "a"),
+        ast.Equal(ast.Attribute("attr"), "b"),
+    )
+
+
+def test_string_like_keeps_backslashes():
+    result = parse(r"attr LIKE '100\%'")
+    assert result.pattern == r"100\%"
+    result = parse(r"attr LIKE 'a\\''%'")
+    assert result.pattern == r"a\\'%"
+
+
 def test_attribute_in_list():
     result = parse("attr IN (1, 2, 3, 4)")
     assert result == ast.In(
@@ -491,6 +521,62 @@ def test_not_eq():
     assert result == ast.Not(ast.Equal(ast.Attribute("attr"), 2))
 
 
+def test_not_double_parens():
+    result = parse("NOT ((attr = 2))")
+    assert result == ast.Not(ast.Equal(ast.Attribute("attr"), 2))
+
+
+def test_not_or():
+    result = parse("NOT (attr = 1 OR attr = 2)")
+    assert result == ast.Not(
+        ast.Or(
+            ast.Equal(ast.Attribute("attr"), 1),
+            ast.Equal(ast.Attribute("attr"), 2),
+        )
+    )
+
+
+def test_not_binds_tighter_than_and():
+    result = parse("NOT attr = 1 AND attr = 2")
+    assert result == ast.And(
+        ast.Not(ast.Equal(ast.Attribute("attr"), 1)),
+        ast.Equal(ast.Attribute("attr"), 2),
+    )
+
+
+def test_and_binds_tighter_than_or():
+    result = parse("attr = 1 OR attr = 2 AND attr = 3")
+    assert result == ast.Or(
+        ast.Equal(ast.Attribute("attr"), 1),
+        ast.And(
+            ast.Equal(ast.Attribute("attr"), 2),
+            ast.Equal(ast.Attribute("attr"), 3),
+        ),
+    )
+
+
+def test_and_then_or():
+    result = parse("attr = 1 AND attr = 2 OR attr = 3")
+    assert result == ast.Or(
+        ast.And(
+            ast.Equal(ast.Attribute("attr"), 1),
+            ast.Equal(ast.Attribute("attr"), 2),
+        ),
+        ast.Equal(ast.Attribute("attr"), 3),
+    )
+
+
+def test_parens_override_precedence():
+    result = parse("(attr = 1 OR attr = 2) AND attr = 3")
+    assert result == ast.And(
+        ast.Or(
+            ast.Equal(ast.Attribute("attr"), 1),
+            ast.Equal(ast.Attribute("attr"), 2),
+        ),
+        ast.Equal(ast.Attribute("attr"), 3),
+    )
+
+
 def test_single_letter_attribute():
     # https://github.com/geopython/pygeofilter/issues/165
     # single-letter property names must parse