bash scripts for extraction commands

extract_json_expanded.sh

@@ -0,0 +1,16 @@
+#!/bin/bash
+
+WikiExtractor.py \
+/scratch/wikipedia-base/dumps/dump02212018/enwiki-latest-pages-articles.xml.bz2 \
+--processes $1 \
+--output /scratch/wikipedia-base/extracted/json-expanded \
+--bytes 10M \
+--compress \
+--json \
+--links \
+--sections \
+--lists \
+--templates templates.txt \
+--min_text_length 0 \
+--filter_disambig_pages \
+--keep_tables

extract_json_not_expanded.sh

@@ -0,0 +1,16 @@
+#!/bin/bash
+
+WikiExtractor.py \
+/scratch/wikipedia-base/dumps/dump02212018/enwiki-latest-pages-articles.xml.bz2 \
+--processes $1 \
+--output /scratch/wikipedia-base/extracted/json-not-expanded \
+--bytes 10M \
+--compress \
+--json \
+--links \
+--sections \
+--lists \
+--no-templates \
+--min_text_length 0 \
+--filter_disambig_pages \
+--keep_tables