tiger_import

#!/bin/bash

TMP="/tmp/tiger-import.$$"
SHPS="edges"
DBFS="featnames addr"
BASE=$(dirname $0)
PATH=$PATH:$BASE
SQL="$BASE/sql"
HELPER_LIB="$BASE/../lib/geocoder/us/sqlite3.so"
SHP2SQLITE=src/shp2sqlite/shp2sqlite 
DATABASE=$1
SOURCE=$2
shift
shift

mkdir -p $TMP || exit 1

# Initialize the database if it doesn't exist.
# Added places back in after adding the "drop if exists" directive to the SQL file. theduckylittle, 2011/12/15
[ ! -r $DATABASE ] && cat ${SQL}/{create,place}.sql | sqlite3 $DATABASE
#[ ! -r $DATABASE ] && cat ${SQL}/create.sql | sqlite3 $DATABASE
 
# Marshal the county directories to import.
#
# If no directory was given on the command-line, read a list of county IDs from STDIN.
if [ x"$1" != x"" ]; then
    cat
else
    # Otherwise, find all of the IDs from the contents of the directory structure.
    ls $SOURCE/tl_*_edges.zip | while read file; do
        file=$(basename $file)
        code=${file##tl_????_}
        echo ${code%%_edges.zip}
    done
fi | sort | while read code; do
    echo "--- $code"
    # Unpack the county files into the temp directory.
    for file in $SHPS $DBFS; do
	ZIP=$(ls $SOURCE/*_${code}_${file}.zip 2>/dev/null)
	SHP=$(ls $SOURCE/*_${code}_${file}.* 2>/dev/null)
	if [ x"$ZIP" != x"" ]; then
	    unzip -q $ZIP -d $TMP
	elif [ x"$SHP" != x"" ]; then
	    ln -s $SHP $TMP
	fi
    done
    # Generate an SQL stream to feed into the sqlite3 binary.
    # Start by loading the helper libs and initializing the temporary tables
    #   that will hold the TIGER data before ETL.
    (echo ".load $HELPER_LIB" && \
     cat ${SQL}/setup.sql && \
     for file in $SHPS; do
       # Convert each Shapefile into SQL statements.
	${SHP2SQLITE} -aS ${TMP}/*_${file}.shp tiger_${file}
     done && \
     for file in $DBFS; do
       # Convert each DBF into SQL statements likewise.
       shp2sqlite -an ${TMP}/*_${file}.dbf tiger_${file}
     done && \
     cat ${SQL}/convert.sql) | sqlite3 $DATABASE
    # Finally, do the transform/load phase (convert.sql)
    # and clean up the temporary files.
    rm -f $TMP/*
done 2>&1 | tee import-$$.log
rm -rf $TMP