FIN: database schema compare

libs/fingerprint.py

@@ -48,17 +48,17 @@ class DBSchema:
     return 1
 
 
-  def importJson(self, filein):
-    self.tables = __importDBSchema(filein)
+  def importJson(self, filejson):
+    self.tablesJson = self.__importDBSchema(filejson)
 
 
-  def compareDBSchema(self, filein):
-    self.tablesJson = __importDBSchema(filein)
-    result = __DBSchemaCompare()
+  def compareDB(self, filejson):
+    self.tablesJson = self.__importDBSchema(filejson)
+    result = self.__DBSchemaCompare()
     return result
 
 
-  def __importDBSchema(self, filein)
+  def __importDBSchema(self, filein):
     tables = {}
     try:
       fh = open(filein, "r")
@@ -67,16 +67,73 @@ class DBSchema:
       tb = jsonData['tables']
       keys = tb.keys()
       for key in keys:
+        print "[[ Table <" + key + "> imported ]]"
         newTable = TableDefinition()
         newTable.importTable(key, dbmt[key], tb[key])
         tables[key] = newTable
 
     except Exception, e:
+      print "ERROR: problem loading json file: " + filein
       print e
 
     return tables
 
 
+  def __DBSchemaCompare(self):
+    # the json database schema definition is what our tools is expecting...
+    # ...so we use it as the baseline
+    # look for table, if exists, compare each 
+    # if exists, compare each field
+    # else, add to unknown tables...or do a fuzzy compare (look at number of fields, field names)
+    for tableName in self.tablesJson.keys():
+      table = self.tables[tableName]
+      print "[[ Comparing Table: " + tableName + " ]]"
+      if (table):
+        self.__CompareTable(self.tablesJson[tableName], table)
+      else:
+        self.__FuzzyTable()
+
+    return
+
+
+  #
+  # Compare the Table Definitions. 
+  # Compare Table 1 (Json table) to Table 2
+  # 
+  def __CompareTable(self, tb1, tb2):
+    fieldsTotalCount = 0
+    fieldsErrorCount = 0
+    propTotalCount = 0
+    propErrorCount = 0
+
+    fields1 = tb1.fields
+    fields2 = tb2.fields
+    for field in fields1.keys():
+      field1 = fields1[field]
+      fieldsTotalCount += 1
+      if (fields2.has_key(field)):
+        field2 = fields1[field]
+        for properties in field1.keys():
+          propTotalCount += 1
+          if not field2.has_key(properties):
+            propErrorCount += 1
+      else:
+        fieldsErrorCount += 1
+
+    if (propErrorCount == 0 and fieldsErrorCount == 0):
+      print "100% compatible"
+    else:
+      totals = propTotalCount + fieldsTotalCount
+      errors = propErrorCount + fieldsErrorCount
+      print "Table difference found: " + str(errors)
+      #print str((errors/totals) * 100) + '% compatible   total == ' + str(totals) + "   errors == " + str(errors)
+
+
+  # look at un-identified tables and try to match fields by their properties
+  def __FuzzyTable():
+    return
+
+
   def __openDB(self, filein):
     conn = sqlite3.connect(filein)
     cur = conn.cursor()
@@ -93,10 +150,17 @@ class DBSchema:
 
 
   def debugFingerprint(self):
-    keys = self.tables.keys()
+    if self.tables:
+      myDict = self.tables
+    elif self.tablesJson:
+      myDict = self.tablesJson
+    else:
+      return
+
+    keys = myDict.keys()
     for key in keys:
       print "[[ TABLE: <" + key + "> ]]"
-      tableDef = self.tables[key]
+      tableDef = myDict[key]
       #print str(tableDef.SQLstr())
       tableDef.toJSON()
 

main.py

@@ -11,7 +11,11 @@ def main():
   (filein, fileout2, filejson, verbose) = parseArgs()
 
   db = fingerprint.DBSchema()
-  if (filein):
+
+  if (filein and filejson):
+    db.scanDBFile(filein)
+    db.compareDB(filejson)
+  elif (filein):
     retVal = db.scanDBFile(filein)
     if (retVal > 0):
       fh = open(fileout2, "w")
@@ -21,7 +25,7 @@ def main():
     else:
       print db.getErrorString(retVal)
   elif (filejson):
-    db.importDBSchema(filejson)
+    db.importJson(filejson)
     db.debugFingerprint()