|
1 | 1 | #!/bin/bash
|
2 | 2 |
|
3 |
| -if [ ! -d wiki_indri_index ]; |
4 |
| -then |
5 |
| - wget http://web.eecs.umich.edu/~jahausw/download/wiki_indri_index.tar.gz |
| 3 | +if [ -d wiki_indri_index ]; then |
| 4 | + echo "Wikipedia index already downloaded, skipping" |
| 5 | + exit 1 |
| 6 | +else |
| 7 | + echo "Wikipedia index not found: OK" |
| 8 | +fi |
| 9 | + |
| 10 | +FREE=`df -k --output=avail "$PWD" | tail -n1` |
| 11 | +if [[ $FREE -lt 11523524 ]]; then |
| 12 | + read -p "Not enough space to download. Do you want to continue (Y/n)? " -n 1 -r |
| 13 | + if [[ ! $REPLY =~ ^[Yy]$ ]] |
| 14 | + then |
| 15 | + echo "" |
| 16 | + echo "Aborting." |
| 17 | + exit 1 |
| 18 | + fi |
| 19 | + echo "" |
| 20 | +else |
| 21 | + echo "Enough space to download.: OK" |
| 22 | +fi; |
| 23 | + |
| 24 | +echo "Starting download." |
| 25 | + |
| 26 | +wget -c http://web.eecs.umich.edu/~jahausw/download/wiki_indri_index.tar.gz |
| 27 | + |
| 28 | +if [ $? -ne 0 ]; then |
| 29 | + echo "wget return code.........: FAIL" |
| 30 | + echo "wget exited with nonzero return, please retry" |
| 31 | + exit 1; |
| 32 | +else |
| 33 | + echo "wget return code.........: OK" |
| 34 | +fi |
| 35 | + |
| 36 | +FILECHECKSUM="322ae59fd0473c6d055ea827ce791c758be1d632e67956ae385eed77aff2f4f3e84db2643e90845b35e1b6204822e9e6" |
| 37 | + |
| 38 | +echo "Download completed. Checking file." |
| 39 | +if [ "$(sha384sum wiki_indri_index.tar.gz | cut -d' ' -f1)" == $FILECHECKSUM ]; then |
| 40 | + echo "sha384sum check..........: OK" |
| 41 | + echo "Decompressing data" |
6 | 42 | tar xzvf wiki_indri_index.tar.gz
|
7 | 43 | else
|
8 |
| - echo "Wikipedia index already downloaded, skipping" |
| 44 | + echo "sha384sum check..........: FAIL" |
| 45 | + echo "Broken download:" |
| 46 | + echo "- Delete the file wiki_indri_index.tar.gz" |
| 47 | + echo "- Run this script again" |
9 | 48 | fi
|
10 | 49 |
|
11 | 50 | echo "******************************************************"
|
|
0 commit comments