projects.mako.cc - redirect-tools/blob - example/01-extract_redirects_wrapper.sh

   1 #!/bin/bash
   2
   3 RUN=$(expr $1 + 1)
   4 INDEX=$(printf "%03d" "$RUN")
   5
   6 CUR_DIR="/nfs/home/B/bhill/condor_jobs/extract_redirects-20140412"
   7 DATA_DIR="/nfs/home/B/bhill/shared_space/barnstar"
   8 INPUT_FILE=$(find ${DATA_DIR}/wp-enwiki-xml -name '*7z' | sed -n ${RUN}p)
   9 OUTPUT_FILE="${DATA_DIR}/wp-enwiki-redir/wp_edits_redir_${INDEX}.tsv.bz2"
  10
  11 # print material out
  12 7za x -so "${INPUT_FILE}" | /usr/local/bin/python2.7 ${CUR_DIR}/extract_redirects.py | bzip2 -c - > ${OUTPUT_FILE}