WIP: second index is created and working on populating it

docs/devnotes

@@ -1,4 +1,31 @@
 
+[[ TODO ]]
+
+-MD5 Index
+  -do I create an index of MD5 table hashes?
+    (how do I lookup each table to compare)
+  -add index comparison feature
+
+-add "-idxf" "--index_force" to force a re-index
+  -first give info about current index
+  -next, create a new index
+  -give info about new index
+"152 fingerprints in the index"
+"Creating a new index. Moved previous to backup"
+"Creating... FINISHED."
+"168 fingerprints in the new index"
+
+
+-Processing of android data
+  -removed verbose messsage for each fingerprint
+  -fix error message to look good
+
+-remove print statements (from modules)
+-add logging statements
+-add exception throwing
+-add unit testing
+  -test md5 sum create of tables
+  -test md5 sum create of entire db (tables concatenated)
 
 
 

libs/fingerprint.py

@@ -35,7 +35,7 @@ class FingerprintDB:
 
   sqlmaster = "SELECT name, sql FROM sqlite_master WHERE type='table'"
   # version of the scanner used to create the fingerprint
-  scanner_ver = "0.90"
+  scanner_ver = "0.95"
   # version of the json file format, this # is inserted in the json fingerprint file and can be used to determine what is supported at the time of that version
   format_ver  = "0.91"
 

libs/fingerprint_index.py

@@ -48,10 +48,15 @@ class FingerprintIndex:
     try:
       self.db_conn = sql.connect(fq_fpidx)
       self.db_conn.execute('''
-          CREATE TABLE md5_index (
-          md5_all TEXT PRIMARY KEY,
+          CREATE TABLE md5_all (
+          md5_db TEXT PRIMARY KEY,
           md5_list TEXT,
-          file_name TEXT);
+          fp_name TEXT);
+        ''')
+      self.db_conn.execute('''
+          CREATE TABLE md5_tables (
+          md5_list TEXT,
+          fp_name TEXT);
         ''')
       logging.info("Successfully created index table")      
       self.__populateIndex(fp_dir) 
@@ -68,22 +73,18 @@ class FingerprintIndex:
     """ read each file, pull md5, add row to database """
     failCount = 0
     finCount = 0
-    # self.db_conn.execute("INSERT INTO md5_index VALUES(?, ?, ?)", ["AAA", "BBB", "CCC"])
-    # self.db_conn.execute("INSERT INTO md5_index VALUES('DDD', 'EEE', 'FFF')")
     try:
       db = FingerprintDB()
       files = os.listdir(fp_dir)
-      # print ("Populating DB, files=={}".format(files))
       for file in files:
         try:
           fq_file = fp_dir + os.path.sep + file
-          # print ("importJson file=={}".format(fq_file))
           db.importJson(fq_file)
-          self.__insertRecord(db.db_hash, db.table_hashes.values(), file)
-          # print("db_hash=={}".format(db.db_hash))
-          # print("table_hashes={}".format(db.table_hashes))
-          #md5_all = __createMD5Index(dbht)
+          self.__insertMod_md5_all(db.db_hash, db.table_hashes.values(), file)
+          #self.__insertMod_md5_tables(db.table_hashes.values(), file)
           finCount = finCount+1
+          if ((finCount % 5) == 0):
+            self.db_conn.commit()
         except:
           failCount = failCount+1
     except:
@@ -93,17 +94,37 @@ class FingerprintIndex:
     logging.info("Completed populating the index. Completed: {}  Failed: {} ".format(str(finCount), str(failCount)))
 
   #
-  def __insertRecord(self, md5_all, md5_list, filename):
+  def __insertMod_md5_all(self, md5_all, md5_list, filename):
     try:
       # logging.info("INSERT INTO md5_index VALUES(?, ?, ?): {}; {}; {}".format(md5_all, str(md5_list), filename))
       self.db_conn.execute(
         '''
-        INSERT INTO md5_index VALUES(?, ?, ?)
+        INSERT INTO md5_all VALUES(?, ?, ?)
         ''', [md5_all, ','.join(md5_list), filename])
+    except sql.IntegrityError as e:
+      logging.error("*** Got here bitches!!!")      
     except Exception as e:
-      logging.error("Error inserting a row\n{}".format(e))
+      logging.error("Error inserting a row\n{}\n{}".format(e, e.__class__.__name__))
+      logging.error("md5_all:  {}\nmd5_list:  {}\nfilename:  {}".format(md5_all, ','.join(md5_list), filename))
       raise FingerprintIndexWrite("Error inserting a row")
 
+  #
+  def __insertMod_md5_tables(self, md5_list, filename):
+    # insert the md5 of the table schemas
+    for md5_table in md5_list:
+      try:
+        self.db_conn.execute(
+          '''
+          INSERT INTO md5_tables VALUES(?, ?)
+          ''', [md5_table, filename])
+      except:
+        try:
+          # modify row, add filename
+          pass
+        except:
+          pass
+
+
   #
   def __checkIntegrity(self):
     """ Sanity check the number of files against the index rows """