Merge pull request #257 from mwang87/formula-distribution

upadting formula distribution

Author: mwang87

massql/msql.ebnf

@@ -146,7 +146,7 @@ factor: floating
     | "formula(" moleculeformula ")"
     | "aminoaciddelta(" aminoacids ")"
     | peptidefunction
-    | "(" numericalexpression ")"
+    | "(" numericalexpression ")" -> factor_parens
 peptidefunction: "peptide(" peptide "," "charge=" peptidecharge "," "ion=" peptideion ")"
 multiply: "*"
 divide: "/"

massql/msql_parser.py

@@ -367,6 +367,9 @@ def divide(self, s):
    def factor(self, s):
       return s[0]
 
+   def factor_parens(self, items):
+      return f"({items[0]})"
+
    def term(self, items):
       if len(items) == 1:
          return items[0]

tests/reference_parses/QUERY_scaninfo_MS2DATA__WHERE_MS1MZ_X-2INTENSITYMA___61a2da302d53487ba89667bb7b7433a3.json

@@ -6,7 +6,7 @@
                 "qualifierintensitymatch": {
                     "comparator": "equal",
                     "name": "qualifierintensitymatch",
-                    "value": "Y*0.0608+2e-06*X"
+                    "value": "Y*(0.0608+(2e-06*X))"
                 },
                 "qualifierintensitytolpercent": {
                     "comparator": "equal",

tests/reference_parses/QUERY_scaninfo_MS2DATA__WHERE_MS2PROD_X_AND_MS2PRO___8ee2287663d6ad552d20963061530a3b.json

@@ -11,7 +11,7 @@
             "conditiontype": "where",
             "type": "ms2productcondition",
             "value": [
-                "2.0*X-55.9349375"
+                "2.0*(X-55.9349375)"
             ]
         }
     ],

tests/reference_parses/QUERY_scaninfo_MS2DATA__WHERE_MS2PROD_X_And_MS2PRO___4fe070795d41b0ea705398a6e5d940ab.json

@@ -11,7 +11,7 @@
             "conditiontype": "where",
             "type": "ms2productcondition",
             "value": [
-                "2.0*X-55.9349375"
+                "2.0*(X-55.9349375)"
             ]
         }
     ],

tests/reference_parses/QUERY_scaninfo_MS2DATA__WHERE_MS2PROD_X_and_MS2PRO___09ccf26b267546b40a6acd472bbea037.json

@@ -11,7 +11,7 @@
             "conditiontype": "where",
             "type": "ms2productcondition",
             "value": [
-                "2.0*X-55.9349375"
+                "2.0*(X-55.9349375)"
             ]
         }
     ],

tests/reference_parses/QUERY_scaninfo_MS2DATA__WHERE_X_range_min_100,_max___ce48d94145b8fe9dd12d4257ffda9ad1.json

@@ -12,7 +12,7 @@
                 "qualifierintensitymatch": {
                     "comparator": "equal",
                     "name": "qualifierintensitymatch",
-                    "value": "Y*0.0608+2e-06*X"
+                    "value": "Y*(0.0608+(2e-06*X))"
                 },
                 "qualifierintensitytolpercent": {
                     "comparator": "equal",

tests/test_parse.py

@@ -112,6 +112,25 @@ def test_variable_formula_parse2():
     parsed_output = msql_parser.parse_msql(query)
     print(json.dumps(parsed_output, indent=4))
 
+def test_formula_distribution():
+    """Verify that formula() is correctly distributed through multiplication in X-expressions.
+
+    Previously, 2*(X - formula(Fe)) was parsed as 2*X - 55.93 instead of 2*(X - 55.93),
+    which is a ~56 Da error when evaluated.
+    """
+    query = "QUERY scaninfo(MS2DATA) WHERE MS2PROD=X AND MS2PROD=2.0*(X - formula(Fe))"
+    parsed_output = msql_parser.parse_msql(query)
+
+    value = parsed_output["conditions"][1]["value"][0]
+    assert value == "2.0*(X-55.9349375)", f"Expected parenthesized form, got: {value}"
+
+    # Verify the expression evaluates correctly
+    from py_expression_eval import Parser
+    p = Parser()
+    result = p.parse(value).evaluate({"X": 100.0})
+    expected = 2.0 * (100.0 - 55.9349375)
+    assert abs(result - expected) < 1e-6, f"Expected {expected}, got {result}"
+
 def test_xrange_parse():
     query = "QUERY scaninfo(MS2DATA) WHERE MS2PROD=X AND MS2PROD=2.0*(X - formula(Fe)) AND X=range(min=5, max=100)"