Project

General

Profile

« Previous | Next » 

Revision 11442

Added by Paul Sarando about 11 years ago

Update load-geoscrub-input.sh to download from URL.

Removed logic to dump input data directly from the vegbien database and
to download the input from a URL provided by AMK instead.
Also updated this script to download the file into an input data
directory, rather than just into the current working directory.

View differences:

derived/biengeo/load-geoscrub-input.sh
1 
# Bash script to dump geoscrub_input table (created by AMK) from the
1 
#!/bin/bash
2 

  		




  
  3
  
    
# Bash script to download geoscrub_input table dump (created by AMK) from the


  		




  2
  4
  
    
# vegbien database, and load it into the geoscrub database (i.e., the


  		




  3
  5
  
    
# postgis database prepped with geonames.org data, GADM2 data, and


  		




  4
  6
  
    
# associated mapping tables).


  		




  ......




  14
  16
  
    
# NCEAS


  		




  15
  17
  
    
# Created Nov 2012


  		




  16
  18
  
    

  		




  17
  
  
    
# dump distinct records from vegbien


  		




  18
  
  
    
psql --host vegbiendev -U bien vegbien \


  		




  19
  
  
    
    -c 'COPY "public.2012-11-04-07-34-10.r5984".geoscrub_input


  		




  20
  
  
    
    TO STDOUT


  		




  21
  
  
    
    WITH CSV' > geoscrub-corpus.csv


  		




  
  19
  
    
# Note, to force data to download from DATA_URL, ensure the DATAFILE is deleted


  		




  
  20
  
    
# before running this script.


  		




  
  21
  
    
DATA_URL="http://fs.vegpath.org/exports/geoscrub_input.no_header.cols=country,stateProvince,county,decimalLatitude,decimalLongitude.csv"


  		




  
  22
  
    
DATADIR="$(dirname $0)/input"


  		




  
  23
  
    
DATAFILE="${DATADIR}/geoscrub-corpus.csv"


  		




  22
  24
  
    

  		




  
  25
  
    
if [[ ! -d "$DATADIR" ]]; then


  		




  
  26
  
    
    echo "making directory ${DATADIR}"


  		




  
  27
  
    
    mkdir -p "$DATADIR"


  		




  
  28
  
    

  		




  
  29
  
    
    if [[ $? != 0 ]]; then


  		




  
  30
  
    
        echo "Could not create directory ${DATADIR}"


  		




  
  31
  
    
        exit 1


  		




  
  32
  
    
    fi


  		




  
  33
  
    
fi


  		




  
  34
  
    

  		




  
  35
  
    
if [[ ! -r "$DATAFILE" ]]; then


  		




  
  36
  
    
    # download distinct records from vegbien


  		




  
  37
  
    
    wget -O "$DATAFILE" "$DATA_URL"


  		




  
  38
  
    

  		




  
  39
  
    
    if [[ $? != 0 ]]; then


  		




  
  40
  
    
        echo "Could not download input to ${DATAFILE}"


  		




  
  41
  
    
        exit 1


  		




  
  42
  
    
    fi


  		




  
  43
  
    
fi


  		




  
  44
  
    

  		




  23
  45
  
    
# generate table


  		




  24
  46
  
    
psql -c \


  		




  25
  47
  
    
   'CREATE TABLE vegbien_geoscrub (


  		




  ......




  31
  53
  
    
    )' geoscrub


  		




  32
  54
  
    

  		




  33
  55
  
    
# load


  		




  34
  
  
    
psql -c "COPY vegbien_geoscrub FROM '$DATADIR/geoscrub-corpus.csv' WITH CSV" geoscrub


  		




  
  56
  
    
psql -c "COPY vegbien_geoscrub FROM '${DATAFILE}' WITH CSV" geoscrub


  		




  35
  57
  
    

  		




  36
  
  
    

  		




  37
  58
  
    

  		












Also available in:  Unified diff