feat: add additional MF query cases

.gitignore

@@ -0,0 +1,4 @@
+__pycache__/
+*.py[cod]
+.venv/
+.env

codegen.py

@@ -204,8 +204,6 @@ def _emit_filter(spec: PhiSpec) -> str:
             or_clauses.append("" + " and ".join(and_group) + "")
 
     condition = " or ".join(or_clauses)
-    if len(condition.strip()) == 0:
-        condition = True
 
     lines.append(f"        if ({condition}):")
     lines.append("             filtered_mf_struct[_key] = entry")
@@ -234,4 +232,4 @@ def main() -> None:
 
 
 if __name__ == "__main__":
-    main()
+    main()

phi_input_case2.txt

@@ -1,18 +1,18 @@
-# Case 2: Count and Max quant per customer
 """
-SELECT
-    cust,
-    COUNT(*),
-    MAX(quant)
-FROM sales
-WHERE quant IS NOT NULL
-GROUP BY cust;
+select state, month, a_sum_quant, b_avg_quant
+from sales
+where year = 2018
+group by state, month; a, b
+such that a_state = state and a_month = month
+    and b_state = state and b_month = month
+having a_sum_quant >= 3 * b_avg_quant
 """
-S: cust, 1_count_*, 1_max_quant
-n: 1
-V: cust
-F: 1_count_*, 1_max_quant
 
+S: state, month, 1_sum_quant, 2_avg_quant
+n: 2
+V: state, month
+F: 1_sum_quant, 2_avg_quant
 sigma:
-  1: (cust == g_cust) and (quant is not None)
-G:
+  1: state == g_state and month == g_month and year == 2018
+  2: state == g_state and month == g_month and year == 2018
+G: 1_sum_quant >= 3 * 2_avg_quant

phi_input_case3.txt

@@ -1,30 +1,11 @@
 """
-WITH a AS (
-    SELECT
-        state,
-        month,
-        SUM(quant) AS a_sum_quant
-    FROM sales
-    WHERE state = 'CA'
-    GROUP BY state, month
-),
-b AS (
-    SELECT
-        state,
-        MAX(quant) AS b_max_quant
-    FROM sales
-    GROUP BY state
-)
-SELECT
-    a.state,
-    a.month,
-    a.a_sum_quant,
-    b.b_max_quant
-FROM a
-JOIN b
-  ON a.state = b.state
-WHERE a.a_sum_quant >= b.b_max_quant * 2;
-
+select state, month, a_sum_quant, b_max_quant
+from sales
+where state = 'CA'
+group by state,month; a,b 
+such that a_state = state and a_month = month
+  and b_state = state
+having a_sum_quant >= b_max_quant * 2
 """
 S: state, month, 1_sum_quant, 2_max_quant
 n: 2

phi_parser.py

@@ -118,7 +118,6 @@ def parse_phi_file(path: str) -> PhiSpec:
     grouping_attrs = _split_list(V)
 
     aggs: List[AggSpec] = []
-    having: List[List[str]] = []
     if F:
         for item in _split_list(F):
             m = _AGG_RE.match(item)
@@ -152,6 +151,8 @@ def parse_phi_file(path: str) -> PhiSpec:
         # Split OR (case-insensitive)
         or_blocks = re.split(r'\s+OR\s+', G, flags=re.IGNORECASE)
 
+        having: List[List[str]] = []
+
         for block in or_blocks:
             and_parts = re.split(r'\s+AND\s+', block, flags=re.IGNORECASE)
             rewrite = []
@@ -197,5 +198,4 @@ def _split_list(s: str) -> List[str]:
 
 
 def _first_token(s: str) -> str:
-    return s.strip().split()[0]
-
+    return s.strip().split()[0]

qpe_case1.py

@@ -0,0 +1,97 @@
+from __future__ import annotations
+import os
+import psycopg2
+import psycopg2.extras
+from dotenv import load_dotenv
+
+# Auto-generated by codegen.py. Re-generate anytime.
+
+def _safe_eval_predicate(expr: str, env: dict) -> bool:
+    """
+    English comments:
+    Evaluate a predicate expression in a restricted environment.
+    - Variables are taken from the given env dict.
+    - Supports Python boolean logic (and/or/not).
+    """
+    if not expr:
+        return True
+    return bool(eval(expr, {"__builtins__": {}}, env))
+
+
+def run_query():
+    try:
+        load_dotenv()
+        user = os.getenv("USER")
+        password = os.getenv("PASSWORD")
+        dbname = os.getenv("DBNAME")
+        host = os.getenv("HOST", "localhost")
+        port = os.getenv("PORT", "5432")
+
+        conn = psycopg2.connect(
+            dbname=dbname,
+            user=user,
+            password=password,
+            host=host,
+            port=port,
+            cursor_factory=psycopg2.extras.DictCursor
+        )
+        cur = conn.cursor()
+
+    except:
+        raise RuntimeError("Incorrect USER/PASSWORD/DBNAME in .env")
+
+    # mf_struct maps grouping key tuple -> entry dict
+    mf_struct = {}
+
+    # SCAN 0: initialize mf_struct entries for distinct grouping keys
+    cur.execute('SELECT * FROM sales')
+    for row in cur:
+        key = (row['prod'], row['month'])
+        if key not in mf_struct:
+            entry = {}
+            entry['prod'] = row['prod']
+            entry['month'] = row['month']
+            entry['1_sum_quant'] = 0
+            entry['2_sum_quant'] = 0
+            mf_struct[key] = entry
+
+    # SCAN 1: compute aggregates for grouping variable 1
+    cur.execute('SELECT * FROM sales')
+    for row in cur:
+        for _key, entry in mf_struct.items():
+            env = dict(row)
+            env['g_prod'] = entry.get('prod')
+            env['g_month'] = entry.get('month')
+            if not _safe_eval_predicate('prod == g_prod and month == g_month', env):
+                continue
+            entry['1_sum_quant'] += (row['quant'] if row['quant'] is not None else 0)
+
+    # SCAN 2: compute aggregates for grouping variable 2
+    cur.execute('SELECT * FROM sales')
+    for row in cur:
+        for _key, entry in mf_struct.items():
+            env = dict(row)
+            env['g_prod'] = entry.get('prod')
+            env['g_month'] = entry.get('month')
+            if not _safe_eval_predicate('prod == g_prod', env):
+                continue
+            entry['2_sum_quant'] += (row['quant'] if row['quant'] is not None else 0)
+
+    filtered_mf_struct = {}
+    for _key, entry in mf_struct.items():
+        if (entry['1_sum_quant'] > 0.9*entry['2_sum_quant']/10 and (entry['1_sum_quant'] > 40000 or entry['2_sum_quant']<500000) and entry['month'] == 11):
+             filtered_mf_struct[_key] = entry
+
+    # Output
+    out_cols = ['prod', 'month', '1_sum_quant', '2_sum_quant']
+    print("\t".join(out_cols))
+    for _key, entry in filtered_mf_struct.items():
+        row_out = [str(entry.get(c, "")) for c in out_cols]
+        print("\t".join(row_out))
+
+    cur.close()
+    conn.close()
+
+
+if __name__ == "__main__":
+    run_query()

qpe_case2.py

@@ -0,0 +1,103 @@
+from __future__ import annotations
+import os
+import psycopg2
+import psycopg2.extras
+from dotenv import load_dotenv
+
+# Auto-generated by codegen.py. Re-generate anytime.
+
+def _safe_eval_predicate(expr: str, env: dict) -> bool:
+    """
+    English comments:
+    Evaluate a predicate expression in a restricted environment.
+    - Variables are taken from the given env dict.
+    - Supports Python boolean logic (and/or/not).
+    """
+    if not expr:
+        return True
+    return bool(eval(expr, {"__builtins__": {}}, env))
+
+
+def run_query():
+    try:
+        load_dotenv()
+        user = os.getenv("USER")
+        password = os.getenv("PASSWORD")
+        dbname = os.getenv("DBNAME")
+        host = os.getenv("HOST", "localhost")
+        port = os.getenv("PORT", "5432")
+
+        conn = psycopg2.connect(
+            dbname=dbname,
+            user=user,
+            password=password,
+            host=host,
+            port=port,
+            cursor_factory=psycopg2.extras.DictCursor
+        )
+        cur = conn.cursor()
+
+    except:
+        raise RuntimeError("Incorrect USER/PASSWORD/DBNAME in .env")
+
+    # mf_struct maps grouping key tuple -> entry dict
+    mf_struct = {}
+
+    # SCAN 0: initialize mf_struct entries for distinct grouping keys
+    cur.execute('SELECT * FROM sales')
+    for row in cur:
+        key = (row['state'], row['month'])
+        if key not in mf_struct:
+            entry = {}
+            entry['state'] = row['state']
+            entry['month'] = row['month']
+            entry['1_sum_quant'] = 0
+            entry['2_avg_quant__sum'] = 0
+            entry['2_avg_quant__count'] = 0
+            entry['2_avg_quant'] = 0
+            mf_struct[key] = entry
+
+    # SCAN 1: compute aggregates for grouping variable 1
+    cur.execute('SELECT * FROM sales')
+    for row in cur:
+        for _key, entry in mf_struct.items():
+            env = dict(row)
+            env['g_state'] = entry.get('state')
+            env['g_month'] = entry.get('month')
+            if not _safe_eval_predicate('state == g_state and month == g_month and year == 2018', env):
+                continue
+            entry['1_sum_quant'] += (row['quant'] if row['quant'] is not None else 0)
+
+    # SCAN 2: compute aggregates for grouping variable 2
+    cur.execute('SELECT * FROM sales')
+    for row in cur:
+        for _key, entry in mf_struct.items():
+            env = dict(row)
+            env['g_state'] = entry.get('state')
+            env['g_month'] = entry.get('month')
+            if not _safe_eval_predicate('state == g_state and month == g_month and year == 2018', env):
+                continue
+            val = row['quant']
+            if val is not None:
+                entry['2_avg_quant__sum'] += val
+                entry['2_avg_quant__count'] += 1
+                entry['2_avg_quant'] = entry['2_avg_quant__sum'] / entry['2_avg_quant__count']
+
+    filtered_mf_struct = {}
+    for _key, entry in mf_struct.items():
+        if (entry['1_sum_quant'] >= 3 * entry['2_avg_quant']):
+             filtered_mf_struct[_key] = entry
+
+    # Output
+    out_cols = ['state', 'month', '1_sum_quant', '2_avg_quant']
+    print("\t".join(out_cols))
+    for _key, entry in filtered_mf_struct.items():
+        row_out = [str(entry.get(c, "")) for c in out_cols]
+        print("\t".join(row_out))
+
+    cur.close()
+    conn.close()
+
+
+if __name__ == "__main__":
+    run_query()