From 2312d54984814bd841054376259701e7ae4df8ae Mon Sep 17 00:00:00 2001
From: JohnE <git@malloc.io>
Date: Mon, 14 Dec 2015 20:56:10 -0800
Subject: [PATCH] FIN: updated the insert into md5_all table with collision
 detection, updates the row and adds another fingerprint to the fp_list column

---
 docs/devnotes             |  5 ++--
 libs/fingerprint_index.py | 51 +++++++++++++++++++++++++++------------
 2 files changed, 38 insertions(+), 18 deletions(-)

diff --git a/docs/devnotes b/docs/devnotes
index bd86d78..e7b17c2 100644
--- a/docs/devnotes
+++ b/docs/devnotes
@@ -2,9 +2,8 @@
 [[ 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 column, fp_count NUMBER (number of fingerprints for this md5 schema)
+  -complete populating the second index
 
 -add "-idxf" "--index_force" to force a re-index
   -first give info about current index
diff --git a/libs/fingerprint_index.py b/libs/fingerprint_index.py
index 0cc14d8..5b0d03e 100644
--- a/libs/fingerprint_index.py
+++ b/libs/fingerprint_index.py
@@ -23,6 +23,7 @@ class FingerprintIndex:
   #
   def __init__(self):
     self.db_conn = None
+    self.cur = None
     return
 
   #
@@ -51,18 +52,18 @@ class FingerprintIndex:
           CREATE TABLE md5_all (
           md5_db TEXT PRIMARY KEY,
           md5_list TEXT,
-          fp_name TEXT);
+          fp_list TEXT);
         ''')
       self.db_conn.execute('''
           CREATE TABLE md5_tables (
-          md5_list TEXT,
-          fp_name TEXT);
+          md5_table TEXT,
+          fp_list TEXT);
         ''')
       logging.info("Successfully created index table")      
       self.__populateIndex(fp_dir) 
       logging.info("Successfully populated the index")
-    except:
-      raise FingerprintIndexWrite("Error creating an index file")
+    except Exception as e:
+      raise FingerprintIndexWrite("Error creating an index file\n{}".format(e))
     finally:
       if self.db_conn:
         self.db_conn.close()
@@ -74,6 +75,7 @@ class FingerprintIndex:
     failCount = 0
     finCount = 0
     try:
+      self.cur  = self.db_conn.cursor()
       db = FingerprintDB()
       files = os.listdir(fp_dir)
       for file in files:
@@ -85,7 +87,8 @@ class FingerprintIndex:
           finCount = finCount+1
           if ((finCount % 5) == 0):
             self.db_conn.commit()
-        except:
+        except Exception as e:
+          logging.error(e)
           failCount = failCount+1
     except:
       pass
@@ -94,24 +97,31 @@ class FingerprintIndex:
     logging.info("Completed populating the index. Completed: {}  Failed: {} ".format(str(finCount), str(failCount)))
 
   #
-  def __insertMod_md5_all(self, md5_all, md5_list, filename):
+  def __insertMod_md5_all(self, md5_db, 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_all VALUES(?, ?, ?)
-        ''', [md5_all, ','.join(md5_list), filename])
-    except sql.IntegrityError as e:
-      logging.error("*** Got here bitches!!!")      
+        ''', [md5_db, ','.join(md5_list), filename])
+    except sql.IntegrityError:
+      try:
+        fp_list = self.__selectFileList(md5_db)
+        fp_list += ","+filename
+        # logging.info("fp_list=={}".format(fp_list))
+        self.db_conn.execute(
+          '''
+          UPDATE md5_all SET fp_list=? WHERE md5_db=?
+          ''', [fp_list, md5_db])
+      except Exception as e:
+        raise FingerprintIndexWrite("Error updating a row\n{}".format(e))
     except Exception as 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")
+      raise FingerprintIndexWrite("Error inserting a row\n{}".format(e))
 
   #
-  def __insertMod_md5_tables(self, md5_list, filename):
+  def __insertMod_md5_tables(self, md5_db, filename):
     # insert the md5 of the table schemas
-    for md5_table in md5_list:
+    for md5_table in md5_db:
       try:
         self.db_conn.execute(
           '''
@@ -124,6 +134,17 @@ class FingerprintIndex:
         except:
           pass
 
+  #
+  def __selectFileList(self, md5_db):
+    try:
+      rows = self.cur.execute(
+        '''
+        SELECT fp_list from md5_all WHERE md5_db=?
+        ''', [md5_db])
+      for row in rows:
+        return row[0]
+    except:
+      raise FingerprintIndexWrite("Error selecting fp_list\n{}".format(e))
 
   #
   def __checkIntegrity(self):