WIP: write database schema to file JSON

libs/fingerprint.py

@@ -59,15 +59,20 @@ class DBSchema:
     return
 
 
-  def writeFingerprint(self):
+  def debugFingerprint(self):
     keys = self.tables.keys()
     for key in keys:
       print "[[ TABLE: <" + key + "> ]]"
       tableDef = self.tables[key]
-      # hehehe = tableDef.SQLstr()
       print str(tableDef.SQLstr())
       tableDef.toJSON()
-    return
+
+
+  def writeFingerprint(self, filehandle):
+    keys = self.tables.keys()
+    for key in keys:
+      tableDef = self.tables[key]
+      tableDef.toFile(filehandle);
 
 
   def getErrorString(self, errorCode):
@@ -94,8 +99,7 @@ class TableDefinition:
     self.tableName = ""
     self.sqlStr    = ""
     self.fields    = {}
-    # self.pkeys     = []
+
-    return
 
   def loadTable(self, tableName, sqlStr):
     self.tableName = tableName
@@ -104,16 +108,12 @@ class TableDefinition:
     if results:
       colstr = results.group(1)
       print "[[ TABLE: <" + tableName + "> ]]"
-      # print "FIELDS: " + colstr
       columns = colstr.split(',')
       for col in columns:
         newField = self.__parseCreateStr(col.strip())
         if newField:
           self.fields[newField['name']] = newField
 
-    return
-
-
   # Table Definition
   #
   #  CREATE TABLE contacts (_id INTEGER PRIMARY KEY AUTOINCREMENT,name_raw_contact_id INTEGER REFERENCES raw_contacts(_id),
@@ -131,7 +131,7 @@ class TableDefinition:
       newField = {}
       #                     photo_id INTEGER REFERENCES data(_id)
       #                     name_raw_contact_id INTEGER REFERENCES raw_contacts(_id)
-      results = re.match(r'(\w+)\s+(\w+)\s+REFERENCESS\s+(\W+)', sqltext)
+      results = re.match(r'(?:[`|\"|\'])*(\w+)(?:[`|\"|\'])*\s+(\w+)\s+REFERENCESS\s+(\W+)', sqltext)
       if results:
         newField['name'] = results.group(1)
         newField['datatype'] = results.group(2)
@@ -140,7 +140,16 @@ class TableDefinition:
         return newField
       #                     pinned INTEGER NOT NULL DEFAULT 2147483647
       #                     send_to_voicemail INTEGER NOT NULL DEFAULT 0
-      results = re.match(r'(\w+)\s+(\w+)\s+NOT NULL\s+DEFAULT\s+(\w+)', sqltext)
+      results = re.match(r'(?:[`|\"|\'])*(\w+)(?:[`|\"|\'])*\s+(\w+)\s+NOT NULL\s+DEFAULT\s+(\w+)', sqltext)
+      if results:
+        newField['name'] = results.group(1)
+        newField['datatype'] = results.group(2)
+        newField['notnull'] = True
+        newField['default'] = results.group(3)
+        return newField
+      #                     pinned INTEGER DEFAULT 2147483647
+      #                     send_to_voicemail INTEGER DEFAULT 0
+      results = re.match(r'(?:[`|\"|\'])*(\w+)(?:[`|\"|\'])*\s+(\w+)\s+DEFAULT\s+(\w+)', sqltext)
       if results:
         newField['name'] = results.group(1)
         newField['datatype'] = results.group(2)
@@ -148,7 +157,15 @@ class TableDefinition:
         newField['default'] = results.group(3)
         return newField
       #                     _id INTEGER PRIMARY KEY AUTOINCREMENT
-      results = re.match(r'(\w+)\s+(\w+)\s+PRIMARY KEY\s+AUTOINCREMENT', sqltext)
+      results = re.match(r'(?:[`|\"|\'])*(\w+)(?:[`|\"|\'])*\s+(\w+)\s+PRIMARY KEY\s+AUTOINCREMENT', sqltext)
+      if results:
+        newField['name'] = results.group(1)
+        newField['datatype'] = results.group(2)
+        newField['primarykey'] = True
+        newField['autoincrement'] = True
+        return newField
+      #                     _id INTEGER PRIMARY KEY
+      results = re.match(r'(?:[`|\"|\'])*(\w+)(?:[`|\"|\'])*\s+(\w+)\s+PRIMARY KEY', sqltext)
       if results:
         newField['name'] = results.group(1)
         newField['datatype'] = results.group(2)
@@ -156,26 +173,26 @@ class TableDefinition:
         newField['autoincrement'] = True
         return newField
       #                     FileID INTEGER NOT NULL
-      results = re.match(r'(\w+)\s+(\w+)\s+NOT NULL', sqltext)
+      results = re.match(r'(?:[`|\"|\'])*(\w+)(?:[`|\"|\'])*\s+(\w+)\s+NOT NULL', sqltext)
       if results:
         newField['name'] = results.group(1)
         newField['datatype'] = results.group(2)
         newField['notnull'] = True
         return newField
       #                     PRIMARY KEY (field_name, 
-      results = re.match(r'PRIMARY KEY \((\w+)\,?', sqltext)
+      results = re.match(r'PRIMARY KEY \((?:[`|\"|\'])*(\w+)(?:[`|\"|\'])*\,?', sqltext)
       if results:
         field = self.fields[results.group(1)]
         field['primarykey'] = True
         return False
       #                     custom_ringtone TEXT
-      results = re.match(r'(\w+)\s+(\w+)', sqltext)
+      results = re.match(r'(?:[`|\"|\'])*(\w+)(?:[`|\"|\'])*\s+(\w+)', sqltext)
       if results:
         newField['name'] = results.group(1)
         newField['datatype'] = results.group(2)
         return newField
       #                     field_name)
-      results = re.match(r'\,?(\w+)\)', sqltext)
+      results = re.match(r'(?:[`|\"|\'])*(\w+)(?:[`|\"|\'])*\)', sqltext)
       if results:
         field = self.fields[results.group(1)]
         field['primarykey'] = True
@@ -198,6 +215,10 @@ class TableDefinition:
     print json.dumps(self.fields)
 
 
+  def toFile(self, filehandle):
+    json.dump(self.fields, filehandle, sort_keys=True, indent=4)
+
+
   def __str__(self):
     global delimeter
     retstr = ""

main.py

@@ -8,15 +8,17 @@ from libs import fingerprint
 
 
 def main():
-  (filein, verbose) = parseArgs()
+  (filein, fileout, verbose) = parseArgs()
   #retVal = fingerprint.scanDB(filein)
 
   db = fingerprint.DBSchema()
   retVal = db.scanDBFile(filein)
 
   if (retVal > 0):
-    print "\n\n"
+    fh = open(fileout, "w")
-    db.writeFingerprint()
+    db.debugFingerprint()
+    db.writeFingerprint(fh)
+    fh.close()
   else:
     print db.getErrorString(retVal)
 
@@ -31,7 +33,8 @@ def parseArgs():
   args = parser.parse_args()
   if (args.verbose):
     verbose = args.verbose
-  return (args.file, verbose)
+  fileout = args.file + "_" + timestr
+  return (args.file, fileout, verbose)
 
 
 if __name__ == "__main__":