Merge branch 'master' of github.com:mwang87/MassQueryLanguage

Author: mwang87

massql/msql.ebnf

@@ -103,12 +103,14 @@ function: functionscannum
     | functionscanrangesum
     | functionscanmz
     | functionscaninfo
+    | functionscanmaxint
 
 functionscannum: "scannum"
 functionscansum: "scansum" 
 functionscanrangesum: "scanrangesum"
 functionscanmz: "scanmz"
 functionscaninfo: "scaninfo"
+functionscanmaxint: "scanmaxint"
 
 booleanandconjunction: "AND"
     | "and"

massql/msql_engine.py

@@ -688,4 +688,28 @@ def _executecollate_query(parsed_dict, ms1_df, ms2_df):
 
                 return ms2_df
 
+        if parsed_dict["querytype"]["function"] == "functionscanmaxint":
+            # This is the biggest peak in the spectrum
+
+            if parsed_dict["querytype"]["datatype"] == "datams1data":
+                if len(ms1_df) == 0:
+                    return ms1_df
+
+                ms1sum_df = ms1_df.groupby("scan").max().reset_index()
+
+                ms1_df = ms1_df.groupby("scan").first().reset_index()
+                ms1_df["i"] = ms1sum_df["i"]
+
+                return ms1_df
+            if parsed_dict["querytype"]["datatype"] == "datams2data":
+                if len(ms2_df) == 0:
+                    return ms2_df
+
+                ms2sum_df = ms2_df.groupby("scan").max().reset_index()
+                
+                ms2_df = ms2_df.groupby("scan").first().reset_index()
+                ms2_df["i"] = ms2sum_df["i"]
+
+                return ms2_df
+
         print("APPLYING FUNCTION")    

tests/test_query.py

@@ -648,6 +648,15 @@ def test_waters_query():
 
     results_df = msql_engine.process_query(query, "tests/data/MMSRG_027.mzML")
 
+def test_scanmaxint_query():
+    query = """
+        QUERY scanmaxint(MS2DATA)
+    """
+
+    results_df = msql_engine.process_query(query, "tests/data/GNPS00002_A3_p.mzML")
+
+    print(results_df)
+
 def main():
     #msql_engine.init_ray()
 
@@ -718,7 +727,8 @@ def main():
     #test_ms2_mobility_variable2()
     #test_massdefect_ANY_query()
     #test_advanced_filters()
-    test_advanced_filters2()
+    #test_advanced_filters2()
+    test_scanmaxint_query()
 
 
 if __name__ == "__main__":