From 1c0b2a4263c0193cf7205ec80033959ac699b6c7 Mon Sep 17 00:00:00 2001
From: Gonzalo Odiard <godiard@gmail.com>
Date: Mon, 05 Dec 2011 07:31:24 +0000
Subject: create_index.py use seek_bzip2 to create the indes needed to decombress blocks

---
diff --git a/create_index.py b/create_index.py
new file mode 100755
index 0000000..b46663f
--- /dev/null
+++ b/create_index.py
@@ -0,0 +1,79 @@
+#!/usr/bin/env python
+# -*- coding: utf-8 -*-
+# create index
+# use https://bitbucket.org/james_taylor/seek-bzip2
+
+import codecs
+import os
+from subprocess import call, Popen, PIPE, STDOUT
+
+input_xml_file_name = './eswiki-20111112-pages-articles.xml'
+
+def create_index():
+    output_file  = open("%s.processed.idx" % input_xml_file_name, mode='w')
+    num_block = 1
+    index_file  = open("%s.processed.bz2t" % input_xml_file_name, mode='r')
+    index_line = index_file.readline()
+    while index_line:
+        parts = index_line.split()
+        block_start = int(parts[0])
+        print "Block %d starts at %d" % (num_block, block_start) 
+        position = 0
+        # extract the block
+        bzip_file = open("%s.processed.bz2" % input_xml_file_name, mode='r')
+        cmd = ['./seek-bzip2/seek-bunzip', str(block_start)]
+        p = Popen(cmd, stdin=bzip_file, stdout=PIPE, stderr=STDOUT,
+                close_fds=True)
+        data_line = p.stdout.readline()
+        while data_line:
+            position += len(data_line)
+            #print data_line
+            if len(data_line) == 2:
+                if ord(data_line[0]) == 1:
+                    title = p.stdout.readline()
+                    position += len(title)
+                    # read article size
+                    # size
+                    size_line = p.stdout.readline()
+                    position += len(size_line)
+                    # \02
+                    data_line = p.stdout.readline()
+                    position += len(data_line)
+                    output_file.write("%s %d %d\n" % \
+                        (title[0:-1], num_block, position))
+                    print "Article %s block %d position %d" % \
+                        (title[0:-1], num_block, position)
+
+            data_line = p.stdout.readline()
+
+        num_block += 1
+        index_line = index_file.readline()
+
+    output_file.close()
+
+def create_bzip_table():
+    """
+    ../seek-bzip2/seek-bzip2/bzip-table <
+    eswiki-20110810-pages-articles.xml.processed.bz2 >
+    eswiki-20110810-pages-articles.xml.processed.bz2t 
+    """
+    cmd = ['./seek-bzip2/bzip-table']
+    bzip_file = open('%s.processed.bz2' % input_xml_file_name, mode='r')
+    table_file = open('%s.processed.bz2t' % input_xml_file_name, mode='w')
+    call(cmd, stdin=bzip_file, stdout=table_file, close_fds=True)
+
+
+print 'Compressing .processed file'
+if not os.path.exists('%s.processed.bz2' % input_xml_file_name):
+    cmd = ['bzip2', '-zk', '%s.processed' % input_xml_file_name]
+    p = call(cmd)
+else:
+    print '.bz2 already exists. Skipping'
+
+print 'Creating bzip2 table file'
+create_bzip_table()
+
+print 'Creating index file'
+create_index()
+
+
--
cgit v0.9.1