From 6a79f61cb65d20e3da3b43abdc8326048320b13d Mon Sep 17 00:00:00 2001
From: JohnE <git@malloc.io>
Date: Thu, 17 Dec 2015 13:47:21 -0800
Subject: [PATCH] FIN: second index population complete, fingerprint count for
 each md5 complete (fp_count), collision detection complete, primary keys
 complete

---
 docs/devnotes             |  6 ++++-
 libs/fingerprint_index.py | 48 +++++++++++++++++++++++++++------------
 2 files changed, 39 insertions(+), 15 deletions(-)

diff --git a/docs/devnotes b/docs/devnotes
index e7b17c2..5ab623e 100644
--- a/docs/devnotes
+++ b/docs/devnotes
@@ -14,11 +14,12 @@
 "Creating... FINISHED."
 "168 fingerprints in the new index"
 
-
+[ Android Data ]
 -Processing of android data
   -removed verbose messsage for each fingerprint
   -fix error message to look good
 
+[ Module Cleanup ]
 -remove print statements (from modules)
 -add logging statements
 -add exception throwing
@@ -26,6 +27,9 @@
   -test md5 sum create of tables
   -test md5 sum create of entire db (tables concatenated)
 
+[ Incorporate with another Product ]
+-add module to other python project
+-add unit test to other python project
 
 
 [[ Code Snippets ]]
diff --git a/libs/fingerprint_index.py b/libs/fingerprint_index.py
index 4d96b01..d0c4bd6 100644
--- a/libs/fingerprint_index.py
+++ b/libs/fingerprint_index.py
@@ -57,7 +57,7 @@ class FingerprintIndex:
         ''')
       self.db_conn.execute('''
           CREATE TABLE md5_tables (
-          md5_table TEXT,
+          md5_table TEXT PRIMARY KEY,
           fp_list TEXT,
           fp_count INTEGER);
         ''')
@@ -85,7 +85,7 @@ class FingerprintIndex:
           fq_file = fp_dir + os.path.sep + file
           db.importJson(fq_file)
           self.__insertMod_md5_all(db.db_hash, db.table_hashes.values(), file)
-          #self.__insertMod_md5_tables(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()
@@ -108,13 +108,14 @@ class FingerprintIndex:
         ''', [md5_db, ','.join(md5_list), filename, 1])
     except sql.IntegrityError:
       try:
-        fp_list = self.__selectFileList(md5_db)
+        (fp_list, fp_count) = self.__selectFileList(md5_db)
         fp_list += ","+filename
+        fp_count += 1
         # logging.info("fp_list=={}".format(fp_list))
         self.db_conn.execute(
           '''
-          UPDATE md5_all SET fp_list=? WHERE md5_db=?
-          ''', [fp_list, md5_db])
+          UPDATE md5_all SET fp_list=?, fp_count=? WHERE md5_db=?
+          ''', [fp_list, fp_count, md5_db])
       except Exception as e:
         raise FingerprintIndexWrite("Error updating a row\n{}".format(e))
     except Exception as e:
@@ -127,24 +128,43 @@ class FingerprintIndex:
       try:
         self.db_conn.execute(
           '''
-          INSERT INTO md5_tables VALUES(?, ?)
-          ''', [md5_table, filename])
-      except:
+          INSERT INTO md5_tables VALUES(?, ?, ?)
+          ''', [md5_table, filename, 1])
+      except sql.IntegrityError:
         try:
-          # modify row, add filename
-          pass
-        except:
-          pass
+          (fp_list, fp_count) = self.__selectFileList222(md5_table)
+          fp_list += ","+filename
+          fp_count += 1          
+          self.db_conn.execute(
+            '''
+            UPDATE md5_tables SET fp_list=?, fp_count=? WHERE md5_table=?
+            ''', [fp_list, fp_count, md5_table])
+        except Exception as e:
+          raise FingerprintIndexWrite("Error updating a row\n{}".format(e))
+      except Exception as e:
+        raise FingerprintIndexWrite("Error inserting a row\n{}".format(e))
 
   #
   def __selectFileList(self, md5_db):
     try:
       rows = self.cur.execute(
         '''
-        SELECT fp_list from md5_all WHERE md5_db=?
+        SELECT fp_list, fp_count FROM md5_all WHERE md5_db=?
         ''', [md5_db])
       for row in rows:
-        return row[0]
+        return (row[0], row[1])
+    except:
+      raise FingerprintIndexWrite("Error selecting fp_list\n{}".format(e))
+
+  #
+  def __selectFileList222(self, md5_table):
+    try:
+      rows = self.cur.execute(
+        '''
+        SELECT fp_list, fp_count FROM md5_tables WHERE md5_table=?
+        ''', [md5_table])
+      for row in rows:
+        return (row[0], row[1])
     except:
       raise FingerprintIndexWrite("Error selecting fp_list\n{}".format(e))