debug

provsql/src/provenance.py

@@ -74,14 +74,10 @@ def preprocess_query(query, cur, conn):
 #fs: [formula1, formula2, ...]
 
 def process_formula(fs, tables):
-    elements = []
+    base_formula = ut.baseFormula(fs[0])
     for i, formula in enumerate(fs):
-        nf, idx = ut.remove1sAndAddTablename(formula, tables[i])
-        elements.append(((nf, idx)))
-    elements.sort(key=operator.itemgetter(1))
-    formula_tmp = [x[0] for x in elements]
-    output = " ".join(formula_tmp)
-    return output
+        base_formula = ut.remove1sAndAddTablename(formula, tables[i], base_formula)
+    return base_formula
 
 def get_header(query):
     parsed = sp.parse(query)[0]

provsql/src/provenanceUtils.py

@@ -2,52 +2,73 @@ def commitAndExecute(conn, cur, command):
     cur.execute(command)
     conn.commit()
 
-#find the formula and return it with the index it should be in the real formula
-def findFormula(formula):
-    mylist = ["⊕", "⊗", "⊖", "𝟙"]
+# Return formula composed with 𝟙.
+def baseFormula(formula):
+    charset = ["⊕", "⊗", "⊖", "𝟙", " ", "(", ")"]
     output = ""
-    state = "before first character"
-    idx = 0
-    for i, char in enumerate(formula):
-        if (char in mylist or char in ["(", ")", " "]) and state == "before first character":
-            if char == "𝟙":
-                idx += 1
-            continue
+    state = 0
+    for c in formula:
+        if c in charset:
+            if state == 1:
+                output = output + "𝟙" + c
+                state = 0
+            else:
+                output = output + c
         else:
-            if state == "after first character":
-                if char == "𝟙":
-                    break
-                else:
-                    output += char
+            if state == 1:
+                continue
             else:
-            #reach formula
-                for j in range(i, -1, -1):
-                    if formula[j] == " ":
-                        break
-                    else:
-                        output = formula[j] + output
-                state = "after first character"
-    print("find formula", output)
-    return (output.strip(), idx)
+                state = 1
+    return output
+            
+def updateBaseFormula(tname, count, base):
+    charset = ["⊕", "⊗", "⊖", " ", "(", ")"]
+    output = ""
+    tmp = 0
+    state = 0
+    for i, c in enumerate(base):
+        if c in charset:
+            if state == 1:
+                tmp += 1
+                state = 0
+            output += c
+            continue
+        elif c == "𝟙":
+            if tmp == count:
+                output = output + tname
+                output = output + base[i+1:]
+                break
+            else:
+                output += c
+                tmp += 1
+        else:
+            if state == 0:
+                state = 1
+            output += c
+    return output
             
 
-def remove1sAndAddTablename(formula, table):
 
-    nf, idx = findFormula(formula)
-    elements = nf.split(" ")
-    for i, e in enumerate(elements):
-        if e in ["⊕", "⊗", "⊖"]:
-            continue
+def remove1sAndAddTablename(formula, table, base):
+    charset = ["⊕", "⊗", "⊖", " ", "(", ")"]
+    count = 0
+    state = 0
+    tname = ""
+    for c in formula:
+        if c in charset:
+           if state == 1:
+                state = 0
+                base = updateBaseFormula(table + "." + tname, count, base)
+                tname = ""
+                count += 1
+           continue
+        elif c == "𝟙":
+            count += 1
         else:
-            if "(" in e:
-                eidx = e.index("(") + 1
-                ne = e[:eidx] + table + "." + e[eidx:]
-                elements[i] = ne
-            else:
-                ne = table + "." + e
-                elements[i] = ne
-    output = " ".join(elements)
-    return (output.strip(), idx)
+            if state == 0:
+                state = 1
+            tname = tname + c
+    return base
     
 def compareTwoRows(row1, row2):
     for i in range(1, len(row1)-1):