[tor-commits] [metrics-tasks/master] Merge adjacent ranges in city databases (#6471).

[karsten@xxxxxxxxxxxxxx]

commit 22bd9dce19c603a978c0847cf8647dabcc60bae8
Author: Karsten Loesing <karsten.loesing@xxxxxxx>
Date:   Mon Nov 5 22:50:15 2012 -0500

    Merge adjacent ranges in city databases (#6471).
---
 .../torproject/task6471/DatabaseImporterImpl.java  |   21 ++++++++++++++++---
 1 files changed, 17 insertions(+), 4 deletions(-)

diff --git a/task-6471/java/src/org/torproject/task6471/DatabaseImporterImpl.java b/task-6471/java/src/org/torproject/task6471/DatabaseImporterImpl.java
index 0f7df77..407d8ca 100644
--- a/task-6471/java/src/org/torproject/task6471/DatabaseImporterImpl.java
+++ b/task-6471/java/src/org/torproject/task6471/DatabaseImporterImpl.java
@@ -194,6 +194,8 @@ public class DatabaseImporterImpl extends DatabaseImpl
       }
       br.close();
       /* Parse blocks file and add ranges to the database. */
+      long lastStartAddress = 0L, lastEndAddress = -2L;
+      String lastCode = null;
       br = new BufferedReader(new FileReader(blocksFile));
       while ((line = br.readLine()) != null) {
         if (!line.startsWith("\"")) {
@@ -211,7 +213,21 @@ public class DatabaseImporterImpl extends DatabaseImpl
           break;
         }
         String code = locations.get(location);
-        this.addRange(databaseDate, startAddress, endAddress, code);
+        if (lastStartAddress == 0L) {
+          lastStartAddress = startAddress;
+          lastCode = code;
+        } else if (lastEndAddress + 1L != startAddress ||
+            !code.equals(lastCode)) {
+          this.addRange(databaseDate, lastStartAddress, lastEndAddress,
+              lastCode);
+          lastStartAddress = startAddress;
+          lastCode = code;
+        }
+        lastEndAddress = endAddress;
+      }
+      if (lastStartAddress != 0L) {
+        this.addRange(databaseDate, lastStartAddress, lastEndAddress,
+            lastCode);
       }
       br.close();
     } catch (IOException e) {
@@ -493,9 +509,6 @@ public class DatabaseImporterImpl extends DatabaseImpl
    * interface, because the caller needs to make sure that repairTree()
    * is called prior to any lookupAddress() calls.  No further checks are
    * performed that the tree is repaired before look up an address.
-   *
-   * TODO While repairing the tree, we might also optimize it by merging
-   * adjacent address ranges with the same database date ranges.
    */
   void repairTree() {
     if (this.addedDatabaseDate < 0) {



_______________________________________________
tor-commits mailing list
tor-commits@xxxxxxxxxxxxxxxxxxxx
https://lists.torproject.org/cgi-bin/mailman/listinfo/tor-commits