spoi-importer-original/geonames_v2.sh

#!/bin/bash

# GeoNames.org
# ============

#All
# -------------
sources=("KM" "EH" "BF" "BJ" "BW" "CD" "CI" "CM" "EG" "GN" "GW" "KE" "LR" "LY" "MA" "SL" "DJ" "ET" "MW" "NA" "NG" "RW" "SO" "ZA" "TZ" "TG" "UG" "ZM" "ZW" "AO" "BI" "CF" "CG" "DZ" "ER" "GA" "GH" "SD" "TD" "MZ" "LS" "CV" "GM" "GQ" "MG" "ML" "MR" "MU" "NE" "SC" "SN" "SS" "ST" "SZ" "TN" "EC" "AR" "BO" "BR" "CL" "CO" "PY" "PE" "UY" "GY" "SR" "VE" "US" "CA" "MX" "BZ" "CU" "GT" "NI" "KN" "AG" "AI" "AW" "BB" "BM" "BS" "CR" "DM" "DO" "GD" "HN" "HT" "JM" "LC" "PA" "SV" "TT" "VC" "NZ" "FJ" "AU" "AS" "CK" "FM" "KI" "MH" "NR" "PG" "PW" "SB" "TL" "TO" "TV" "VU" "WS" "AZ" "BD" "CN" "ID" "IN" "IR" "IQ" "IL" "JO" "JP" "KZ" "LB" "MN" "MV" "MY" "NP" "KP" "PK" "PH" "RU" "KR" "LK" "SG" "SY" "TH" "TW" "TJ" "TM" "UZ" "VN" "YE" "SA" "QA" "OM" "MM" "LA" "KW" "KH" "KG" "BT" "BN" "BH" "AF" "AE" "AD" "AM" "AT" "BE" "BG" "HR" "CY" "CZ" "DE" "DK" "FI" "FR" "GE" "HU" "IS" "KS" "LI" "NO" "PL" "PT" "RO" "RS" "SK" "SI" "ES" "SE" "CH" "GB" "BY" "BA" "EE" "LV" "LT" "LU" "MK" "MT" "MD" "MC" "ME" "NL" "TR" "UA" "VA" "SM" "GR" "AL" "IE" "IT" "AQ" "AX" "BL" "BQ" "BV" "CC" "CW" "CX" "FK" "FO" "GF" "GG" "GI" "GL" "GP" "GC" "GU" "HK" "HM" "IM" "IO" "KY" "MF" "MO" "MP" "MQ" "MS" "NC" "NF" "NU" "PF" "PM" "PN" "PR" "PS" "RE" "SH" "SJ" "SX" "TC" "TF" "TK" "UM" "VG" "VI" "WF")




# Africa
# ------
#sources=("KM" "EH" "BF" "BJ" "BW" "CD" "CI" "CM" "EG" "GN" "GW" "KE" "LR" "LY" "MA" "SL")
#sources=("DJ" "ET" "MW" "NA" "NG" "RW" "SO" "ZA" "TZ" "TG" "UG" "ZM" "ZW" "AO" "BI" "CF" "CG" "DZ" "ER" "GA" "GH" "SD" "TD" "MZ" "LS" "CV" "GM" "GQ" "MG" "ML" "MR" "MU" "NE" "SC" "SN" "SS" "ST" "SZ" "TN") 

# South America
# -------------
#sources=("EC" "AR" "BO" "BR" "CL" "CO" "PY" "PE" "UY" "GY" "SR" "VE")

# North America
# -------------
#sources=("US" "CA" "MX")

# Central America
# ---------------
#sources=("BZ" "CU" "GT" "NI" "KN" "AG" "AI" "AW" "BB" "BM" "BS" "CR" "DM" "DO" "GD" "HN" "HT" "JM" "LC" "PA" "SV" "TT" "VC")

# Australia & Oceania
# --------------------
#sources=("NZ" "FJ" "AU" "AS" "CK" "FM" "KI" "MH" "NR" "PG" "PW" "SB" "TL" "TO" "TV" "VU" "WS")

# Asia
# ----
#sources=("AZ" "BD" "CN" "ID" "IN" "IR" "IQ" "IL" "JO" "JP" "KZ" "LB" "MN" "MV" "MY" "NP" "KP" "PK" "PH" "RU" "KR" "LK" "SG" "SY" "TH" "TW" "TJ" "TM" "UZ" "VN" "YE" "SA" "QA" "OM" "MM" "LA" "KW" "KH" "KG" "BT" "BN" "BH" "AF" "AE")

# Europe
# ------
#sources=("AD" "AM" "AT" "BE" "BG" "HR" "CY" "CZ" "DE" "DK" "FI" "FR" "GE" "HU" "IS" "KS" "LI" "NO" "PL" "PT" "RO" "RS" "SK" "SI" "ES" "SE" "CH" "GB" "BY" "BA" "EE" "LV" "LT" "LU" "MK" "MT" "MD" "MC" "ME" "NL" "TR" "UA" "VA" "SM" "GR" "AL" "IE" "IT")

# Other areas
# -----------
#sources=("AQ" "AX" "BL" "BQ" "BV" "CC" "CW" "CX" "FK" "FO" "GF" "GG" "GI" "GL" "GP" "GC" "GU" "HK" "HM" "IM" "IO" "KY" "MF" "MO" "MP" "MQ" "MS" "NC" "NF" "NU" "PF" "PM" "PN" "PR" "PS" "RE" "SH" "SJ" "SX" "TC" "TF" "TK" "UM" "VG" "VI" "WF")

sources=("CN")

# Not processed "JE" "YT"


for i in ${sources[*]}
do
echo $i
wget -P "/media/cerba/Data/Data/SPOI/Temp" "http://download.geonames.org/export/dump/${i}.zip"
cd /media/cerba/Data/Data/SPOI/Temp
unzip -u ${i}.zip
awk 'BEGIN { FS="\t"; OFS=";" } {$1=$1; print}' ${i}.txt > ${i}1.tmp
awk 'BEGIN { FS="&"; OFS="and" } {$1=$1; print}' ${i}1.tmp > ${i}.tmp
file_out="${i}.xml"

echo '<?xml version="1.0"?>' > $file_out
echo '<root>' >> $file_out

while IFS=$';' read -r -a arry
do
echo '<Record>' >> $file_out
echo '<A>'${arry[0]}'</A>' >> $file_out
echo '<B>'${arry[1]}'</B>' >> $file_out
echo '<C>'${arry[2]}'</C>' >> $file_out
#echo '<D>'${arry[3]}'</D>' >> $file_out
echo '<E>'${arry[4]}'</E>' >> $file_out
echo '<F>'${arry[5]}'</F>' >> $file_out
echo '<G>'${arry[6]}'</G>' >> $file_out
echo '<H>'${arry[7]}'</H>' >> $file_out
echo '<I>'${arry[8]}'</I>' >> $file_out
echo '</Record>' >> $file_out
done < ${i}.tmp


echo '</root>' >> $file_out
cd /home/cerba/Projekty/SPOI/Scripts
echo "xslt start"
java -Xms2560m -Xmx14560m -jar saxon9he.jar -s:/media/cerba/Data/Data/SPOI/Temp/${i}.xml -xsl:../XSLT/transform_geonames_download.xsl -o:/media/cerba/Data/Data/SPOI/RDF_output/${i}.rdf

rm -rf /media/cerba/Data/Data/SPOI/Temp
done