fixes to image tables

Forums/DB_Connection/db_connection.py

@@ -118,9 +118,9 @@ def verifyImage(cur, base64Image):
 
     try:
 
-        cur.execute("lock table market_images IN ACCESS EXCLUSIVE MODE")
+        cur.execute("lock table forum_images IN ACCESS EXCLUSIVE MODE")
 
-        cur.execute("select image_id from market_images where hash_image = %(hashImage)s limit 1",
+        cur.execute("select image_id from forum_images where hash_image = %(hashImage)s limit 1",
                     {'hashImage': generate_image_hash(base64Image)})
 
         recset = cur.fetchall()
@@ -273,7 +273,7 @@ def getLastImage(cur):
 
     try:
 
-        cur.execute("select image_id from market_images order by image_id desc limit 1")
+        cur.execute("select image_id from forum_images order by image_id desc limit 1")
 
         recset = cur.fetchall()
 
@@ -406,7 +406,7 @@ def create_author(cur, row, forumId):
              "signature_user, image_user, dateinserted_user) Values (%s, %s, %s, %s, %s, %s, %s, %s, %s)"
 
        recset = [userId, forumId,
-                 row[2], 'Nan', 'Nan', 'Nan', 'Nan', 'Nan', #telling the create_user function to not track changes here
+                 row[2], 'Nan', 'Nan', 'Nan', 'Nan', None, #telling the create_user function to not track changes here
                  row[8]]
 
        cur.execute(sql, recset)
@@ -451,7 +451,7 @@ def create_user(cur, row, forumId, index):
         recset = cur.fetchall()
 
         imageId = recset[0]['image_user']
-        if not is_same_image(row[9][index], recset[0]['image_user'])):
+        if not is_same_image(cur, row[9][index], recset[0]['image_user']):
             imageId = create_image(cur, row[9][index], row[8])
 
         if (str(recset[0]['status_user']) != str(row[11][index] if row[11][index] != '-1' else None) or
@@ -464,7 +464,7 @@ def create_user(cur, row, forumId, index):
                 str(recset[0]['reputation_user']) != 'Nan' or
                 str(recset[0]['interest_user']) != 'Nan' or
                 str(recset[0]['signature_user']) != 'Nan' or
-                str(recset[0]['image_user']) != 'Nan'):
+                str(recset[0]['image_user']) != str(None)):
 
                 userVersionId = int(getLastUserVersion(cur, userId) + 1)
 
@@ -537,7 +537,7 @@ def create_posts(cur, row, forumId, topicId):
                 recset = cur.fetchall()
 
                 imageId = recset[0]['image_post']
-                if not is_same_image(row[18][i], recset[0]['image_post'])):
+                if not is_same_image(cur, row[18][i], recset[0]['image_post']):
                     imageId = create_image(cur, row[18][i], row[8])
 
                 if (str(recset[0]['content_post']) != str(row[15][i]) or
@@ -566,23 +566,23 @@ def create_posts(cur, row, forumId, topicId):
                                       'dateinserted_post': row[8],
                                       'postId': postId})
 
-def is_same_image(newBase64Image, oldImageId):
+def is_same_image(cur, newBase64Image, oldImageId):
 
-    if newBase64Image == "-1" and oldImageId == "-1":
+    if newBase64Image == "-1" and str(oldImageId) is str(None):
         return True
 
-    if newBase64Image == "-1": # and oldImageId != "-1"
+    if newBase64Image == "-1": # and str(oldImageId) != str(None)
         return False
 
-    if oldImageId == "-1": # and newBase64Image != "-1"
+    if str(oldImageId) == str(None): # and newBase64Image != "-1"
         return False
 
-    sql = "select * from market_images where image_id = %(imageId)s"
+    sql = "select * from forum_images where image_id = %(imageId)s"
     cur.execute(sql, {'imageId': oldImageId})
 
     recset = cur.fetchall()
 
-    decImage = decode_decrypt_image_in_base64(base64Image)
+    decImage = decode_decrypt_image_in_base64(newBase64Image)
 
     hash1 = generate_image_hash(decImage)
     hash2 = recset[0]['hash_image']
@@ -590,19 +590,21 @@ def is_same_image(newBase64Image, oldImageId):
     if hash1 == hash2:
         return True
     
-    features1 = json.loads(extract_hidden_layer_output(dec_string))
+    features1 = json.loads(extract_hidden_layer_output(decImage))
     features2 = json.loads(recset[0]['resnet50_image'])
 
     similarity = 1 - distance.cosine(features1, features2)
 
-    return similarity < 0.8
+    return similarity >= 0.8
 
 def create_image(cur, base64Image, date):
 
     if base64Image == "-1":
         return None
 
-    imageId = verifyImage(base64Image)
+    decImage = decode_decrypt_image_in_base64(base64Image)
+
+    imageId = verifyImage(cur, decImage)
 
     if not imageId:
        imageId = int(getLastImage(cur) + 1)
@@ -611,15 +613,13 @@ def create_image(cur, base64Image, date):
        newImage = False
 
     if newImage:
-        
-        decImage = decode_decrypt_image_in_base64(base64Image)
 
         hashImage = generate_image_hash(decImage)
         hloFeatures = extract_hidden_layer_output(decImage)
         keypoints, descriptors = extract_keypoints(decImage)
 
-        sql = "Insert into market_images(image_id, hash_image, base64_image, " \
-              "resnet50_image, siftkey_image, siftdesc_image, dateinserted_image) " \ 
+        sql = "Insert into forum_images(image_id, hash_image, base64_image, " \
+              "resnet50_image, siftkey_image, siftdesc_image, dateinserted_image) " \
               "Values (%s, %s, %s, %s, %s, %s, %s)"
 
         recset = [imageId,
@@ -662,11 +662,11 @@ def create_database(cur, con):
 
         sql = "create table forum_images(image_id integer not null, hash_image character varying(64) not null, base64_image character varying(10000000) not null, " \
               "resnet50_image character varying(1000000) null, siftkey_image character varying(1000000) null, siftdesc_image character varying(1000000) null, " \
-              "dateinserted_item timestamp(6) with time zone not null, " \
-              "constraint items_pk primary key (image_id))"
+              "dateinserted_image timestamp(6) with time zone not null, " \
+              "constraint forum_images_pk primary key (image_id))"
         cur.execute(sql)
 
-        sql = "create unique index unique_image ON forum_images USING btree (hash_image ASC NULLS LAST)"
+        sql = "create unique index unique_forum_image ON forum_images USING btree (hash_image ASC NULLS LAST)"
         cur.execute(sql)
 
         sql = "create table forums (forum_id integer NOT NULL, name_forum character varying(255) NOT NULL, url_forum " \

Forums/Utilities/utilities.py

@@ -89,6 +89,9 @@ def extract_keypoints(image_string):
 
     keypoints, descriptors = sift.detectAndCompute(img, None)
 
+    if len(keypoints) == 0:
+        return None, None
+
     return json.dumps(wrap_keypoints(keypoints)), json.dumps(descriptors.tolist())
 
 

MarketPlaces/DB_Connection/db_connection.py

@@ -273,7 +273,7 @@ def create_vendor(cur, row, marketId):
         recset = cur.fetchall()
 
         imageId = recset[0]['image_vendor']
-        if not is_same_image(row[21], recset[0]['image_vendor'])):
+        if not is_same_image(cur, row[21], recset[0]['image_vendor']):
             imageId = create_image(cur, row[21], row[23])
 
         if (str(recset[0]['rating_vendor']) != str(row[2] if row[2] != '-1' else None) or # there was a change in the vendor information
@@ -362,7 +362,7 @@ def create_items(cur, row, marketId, vendorId):
             recset = cur.fetchall()
 
             imageId = recset[0]['image_item']
-            if not is_same_image(row[20], recset[0]['image_item'])):
+            if not is_same_image(cur, row[20], recset[0]['image_item']):
                 imageId = create_image(cur, row[20], row[23])
 
             if (str(recset[0]['vendor_id']) != str(vendorId) or
@@ -389,7 +389,7 @@ def create_items(cur, row, marketId, vendorId):
 
                 sql = "Insert into items_history (item_id, version_item, market_id, vendor_id, name_item, description_item, cve_item, ms_item, category_item, " \
                       "views_item, reviews_item, rating_item, dateadded_item, btc_item, usd_item, euro_item, quantitysold_item, " \
-                      "quantityleft_item, shippedfrom_item, shippedto_item, lastseen_item, image_item, href_item, dateinserted_item, " \
+                      "quantityleft_item, shippedfrom_item, shippedto_item, image_item, href_item, lastseen_item, dateinserted_item, " \
                       "classification_item) Values (%s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s, " \
                       "%s, %s, %s, %s, %s, %s)"
 
@@ -425,7 +425,7 @@ def create_items(cur, row, marketId, vendorId):
                       "rating_item = %(rating_item)s, dateadded_item = %(dateadded_item)s, btc_item = %(btc_item)s, " \
                       "usd_item = %(usd_item)s, euro_item = %(euro_item)s, quantitysold_item = %(quantitysold_item)s, " \
                       "quantityleft_item = %(quantityleft_item)s, shippedfrom_item = %(shippedfrom_item)s, shippedto_item = %(shippedto_item)s, " \
-                      "image_item = %(image_item)s, lastseen_item = %(lastseen_item)s, dateinserted_item = %(dateinserted_item)s." \
+                      "image_item = %(image_item)s, lastseen_item = %(lastseen_item)s, dateinserted_item = %(dateinserted_item)s, " \
                       "classification_item = %(classification_item)s where item_id = %(itemId)s"
 
                 cur.execute(sql, {'vendor_id': vendorId,
@@ -460,15 +460,15 @@ def create_items(cur, row, marketId, vendorId):
 
     return itemId
 
-def is_same_image(newBase64Image, oldImageId):
+def is_same_image(cur, newBase64Image, oldImageId):
 
-    if newBase64Image == "-1" and oldImageId == "-1":
+    if newBase64Image == "-1" and str(oldImageId) == str(None):
         return True
 
-    if newBase64Image == "-1": # and oldImageId != "-1"
+    if newBase64Image == "-1": # and str(oldImageId) != str(None)
         return False
 
-    if oldImageId == "-1": # and newBase64Image != "-1"
+    if str(oldImageId) == str(None): # and newBase64Image != "-1"
         return False
 
     sql = "select * from market_images where image_id = %(imageId)s"
@@ -476,7 +476,7 @@ def is_same_image(newBase64Image, oldImageId):
 
     recset = cur.fetchall()
 
-    decImage = decode_decrypt_image_in_base64(base64Image)
+    decImage = decode_decrypt_image_in_base64(newBase64Image)
 
     hash1 = generate_image_hash(decImage)
     hash2 = recset[0]['hash_image']
@@ -484,19 +484,21 @@ def is_same_image(newBase64Image, oldImageId):
     if hash1 == hash2:
         return True
     
-    features1 = json.loads(extract_hidden_layer_output(dec_string))
+    features1 = json.loads(extract_hidden_layer_output(decImage))
     features2 = json.loads(recset[0]['resnet50_image'])
 
     similarity = 1 - distance.cosine(features1, features2)
 
-    return similarity < 0.8
+    return similarity >= 0.8
 
 def create_image(cur, base64Image, date):
 
     if base64Image == "-1":
         return None
 
-    imageId = verifyImage(base64Image)
+    decImage = decode_decrypt_image_in_base64(base64Image)
+
+    imageId = verifyImage(cur, decImage)
 
     if not imageId:
        imageId = int(getLastImage(cur) + 1)
@@ -505,15 +507,13 @@ def create_image(cur, base64Image, date):
        newImage = False
 
     if newImage:
-        
-        decImage = decode_decrypt_image_in_base64(base64Image)
 
         hashImage = generate_image_hash(decImage)
         hloFeatures = extract_hidden_layer_output(decImage)
         keypoints, descriptors = extract_keypoints(decImage)
 
         sql = "Insert into market_images(image_id, hash_image, base64_image, " \
-              "resnet50_image, siftkey_image, siftdesc_image, dateinserted_image) " \ 
+              "resnet50_image, siftkey_image, siftdesc_image, dateinserted_image) " \
               "Values (%s, %s, %s, %s, %s, %s, %s)"
 
         recset = [imageId,
@@ -556,11 +556,11 @@ def create_database(cur, con):
 
         sql = "create table market_images(image_id integer not null, hash_image character varying(64) not null, base64_image character varying(10000000) not null, " \
               "resnet50_image character varying(1000000) null, siftkey_image character varying(1000000) null, siftdesc_image character varying(1000000) null, " \
-              "dateinserted_item timestamp(6) with time zone not null, " \
-              "constraint items_pk primary key (image_id))"
+              "dateinserted_image timestamp(6) with time zone not null, " \
+              "constraint market_images_pk primary key (image_id))"
         cur.execute(sql)
 
-        sql = "create unique index unique_image ON market_images USING btree (hash_image ASC NULLS LAST)"
+        sql = "create unique index unique_market_image ON market_images USING btree (hash_image ASC NULLS LAST)"
         cur.execute(sql)
 
         sql = "create table marketplaces(market_id integer not null, name_market character varying(255) not null, " \

MarketPlaces/Utilities/utilities.py

@@ -89,6 +89,9 @@ def extract_keypoints(image_string):
 
     keypoints, descriptors = sift.detectAndCompute(img, None)
 
+    if len(keypoints) == 0:
+        return None, None
+
     return json.dumps(wrap_keypoints(keypoints)), json.dumps(descriptors.tolist())