From 36055c51726264c43110c2225ba9bdc2fdcce5f2 Mon Sep 17 00:00:00 2001
From: Peter Wu <peter@lekensteyn.nl>
Date: Wed, 11 Jun 2014 13:36:41 +0200
Subject: More numbers for estimating duration

---
 src/database/BrandAnalyzerQueue.java |  4 ++++
 src/main/Analyzor.java               | 19 +++++++++++++++++++
 2 files changed, 23 insertions(+)

diff --git a/src/database/BrandAnalyzerQueue.java b/src/database/BrandAnalyzerQueue.java
index 21d0290..f6b0a40 100644
--- a/src/database/BrandAnalyzerQueue.java
+++ b/src/database/BrandAnalyzerQueue.java
@@ -49,6 +49,7 @@ public class BrandAnalyzerQueue implements Runnable {
     }
 
     private void fillQueue() throws SQLException, InterruptedException {
+        int count = 0;
         while (data.next()) {
             List<String> brands = checker.getBrands(data.getString("text"));
             // if there is no brand, add a dummy so we know it got checked
@@ -58,7 +59,10 @@ public class BrandAnalyzerQueue implements Runnable {
             long tweetid = data.getLong("tweetid");
             Result result = new Result(tweetid, brands);
             queue.put(result);
+            ++count;
         }
+        getLogger().log(Level.INFO, "Finished with queuing " + count
+                + " analyses of tweets");
     }
 
     public Result next() {
diff --git a/src/main/Analyzor.java b/src/main/Analyzor.java
index 9e26c97..2589e66 100644
--- a/src/main/Analyzor.java
+++ b/src/main/Analyzor.java
@@ -254,6 +254,20 @@ public class Analyzor {
         writer.close();
     }
 
+    /**
+     * @return The number of rows that would be returned by the query.
+     */
+    private int getQuerySize(String query) throws SQLException {
+        PreparedStatement statement;
+        String countQuery = "WITH q AS (" + query + ") SELECT COUNT(*) FROM q";
+        statement = connection.prepareStatement(countQuery);
+        ResultSet result = statement.executeQuery();
+        result.next();
+        int count = result.getInt("count");
+        result.close();
+        return count;
+    }
+
     /**
      * Obtain the brands of select tweet texts.
      *
@@ -275,6 +289,11 @@ public class Analyzor {
         if (queryText.isEmpty()) {
             queryText = "SELECT tweetid, text FROM tweet";
         }
+
+        // print expected count, this will increase the initial startup time
+        // (5s for 10M results), but will give an accurate ETA estimator
+        System.out.println("Expected query size: " + getQuerySize(queryText));
+
         // allow the use of cursors for less memory usage
         connection.setAutoCommit(false);
         statement = connection.prepareStatement(queryText);
-- 
cgit v1.2.1