From 7077862382ea148ab7158c622fec126dcc178a2b Mon Sep 17 00:00:00 2001
From: James Ketrenos <james.p.ketrenos@intel.com>
Date: Fri, 28 Sep 2018 12:26:57 -0700
Subject: [PATCH] Prevent duplicate hash entries in hash transaction

---
 server/scanner.js | 101 +++++++++++++++++++++++++++-------------------
 1 file changed, 60 insertions(+), 41 deletions(-)

diff --git a/server/scanner.js b/server/scanner.js
index c6e26a6..3bcb34d 100755
--- a/server/scanner.js
+++ b/server/scanner.js
@@ -212,55 +212,74 @@ function processBlock(items) {
         lastMessage = moment();
       }
     });
-}, {
-    concurrency: 5
+  }, {
+    concurrency: 10
   }).then(function() {
     let toProcess = processing.length, lastMessage = moment();
-    /* Needs to be one at a time in case there are multiple HASH collisions */
-    return photoDB.sequelize.transaction(function(transaction) {
-      return Promise.mapSeries(processing, function(asset) {
-        return photoDB.sequelize.query("SELECT photohashes.*,photos.filename,albums.path FROM photohashes " +
-          "LEFT JOIN photos ON (photos.id=photohashes.photoId) " +
-          "LEFT JOIN albums ON (albums.id=photos.albumId) " +
-          "WHERE hash=:hash OR photoId=:id", {
-          replacements: asset,
-          type: photoDB.sequelize.QueryTypes.SELECT
-        }).then(function(results) {
-          let query;
-          if (results.length == 0) {
-            query = "INSERT INTO photohashes (hash,photoId) VALUES(:hash,:id)";
-          } else if (results[0].hash != asset.hash) {
-            query = "UPDATE photohashes SET hash=:hash WHERE photoId=:id)";
-          } else if (results[0].photoId != asset.id) {
-            console.log("Duplicate asset: " + 
-              "'" + asset.album.path + asset.filename + "' is a copy of " +
-              "'" + results[0].path + results[0].filename + "'");
-            duplicates.push(asset);
-            return;
-          }
-
-          /* Even if the hash doesn't need to be updated, the entry needs to be scanned */
-          needsProcessing.push(asset);
-
-          if (!query) {
-            return;
-          }
-
-          return photoDB.sequelize.query(query, {
+    /* Needs to be one at a time in case there are multiple HASH collisions. To speed 
+     * up commits to the DB we will batch these into 100 record transactions where HASH
+     * collions are done via DB query *AND* in-memory table lookup in the current batch */
+    let batchSize = 100, batches = [];
+    while (processing.length) {
+      batches.push(processing.splice(0, batchSize));
+    }
+    return Promise.mapSeries(batches, function(batch) {
+      return photoDB.sequelize.transaction(function(transaction) {
+        return Promise.mapSeries(batch, function(asset, index) {
+          return photoDB.sequelize.query("SELECT photohashes.*,photos.filename,albums.path FROM photohashes " +
+            "LEFT JOIN photos ON (photos.id=photohashes.photoId) " +
+            "LEFT JOIN albums ON (albums.id=photos.albumId) " +
+            "WHERE hash=:hash OR photoId=:id", {
             replacements: asset,
-            transaction: transaction
+            type: photoDB.sequelize.QueryTypes.SELECT
+          }).then(function(results) {
+            let query;
+
+            /* If this asset exists in this transaction block, push it into the results */
+            for (let i = 0; i < index; i++) {
+              if (batch[i].hash == asset.hash) {
+                results.push(batch[i]);
+              }
+            }
+
+            if (results.length == 0) {
+              query = "INSERT INTO photohashes (hash,photoId) VALUES(:hash,:id)";
+            } else if (results[0].hash != asset.hash) {
+              query = "UPDATE photohashes SET hash=:hash WHERE photoId=:id)";
+            } else if (results[0].photoId != asset.id) {
+              console.log("Duplicate asset: " + 
+                "'" + asset.album.path + asset.filename + "' is a copy of " +
+                "'" + results[0].path + results[0].filename + "'");
+              duplicates.push(asset);
+              return;
+            }
+
+            /* Even if the hash doesn't need to be updated, the entry needs to be scanned */
+            needsProcessing.push(asset);
+
+            if (!query) {
+              return;
+            }
+
+            return photoDB.sequelize.query(query, {
+              replacements: asset,
+              transaction: transaction
+            });
+          }).then(function() {
+            toProcess--;
+            if (moment().add(-5, 'seconds') > lastMessage) {
+              console.log("Hash items to be checked: "  + toProcess);
+              lastMessage = moment();
+            }
           });
-        }).then(function() {
-          toProcess--;
-          if (moment().add(-5, 'seconds') > lastMessage) {
-            console.log("Hash items to be checked: "  + toProcess);
-            lastMessage = moment();
-          }
         });
       });
+    }).catch(function(error) {
+      console.log("Error commiting HASH transactions");
+      throw error;
     });
   }).then(function() {
-    let toProcess = processing.length, lastMessage = moment();
+    let toProcess = needsProcessing.length, lastMessage = moment();
     console.log(needsProcessing.length + " assets need to have metadata extracted");
     return Promise.map(needsProcessing, function(asset) {
       var path = asset.album.path,