Finalized dblp and tested the entire pipeline.

dblp/download_prepare.sh

@@ -18,9 +18,14 @@ gzip -d dblp-2017-11-01.xml.gz
 # Convert XML to bracket notation.
 ./dblp_to_bracket.py
 
+# Remove 'www' entries.
+awk '!/{www{key{homepages/' dblp.bracket > dblp_no_www.bracket
+
 # Sort the dataset.
-./../utilities/sort_dataset.sh dblp.bracket
+./../utilities/sort_dataset.sh dblp_no_www.bracket > dblp_no_www_sorted.bracket
 
 # Tidy up.
-# rm *xml*
-# rm *.dtd
+rm *xml*
+rm *.dtd
+rm dblp.bracket
+rm dblp_no_www.bracket

dblp/tidy-up.sh

-#!/bin/bash
-
-# Delete all downloaded files.
-rm *xml*
-rm *.dtd

utilities/sort_dataset.sh

@@ -32,4 +32,4 @@
 # NOTE: We substract the escaped brackets because they're part of node labels.
 # 
 cat $input | awk '{print gsub("{","{")-gsub("\\\\{","\\{"), $0}' | \
-sort -n --buffer-size=4G | cut -d' ' -f2-
+sort -n --buffer-size=10G | cut -d' ' -f2-