From 1b75e73adebe4449f4f8501481f8c4a7d9beebc7 Mon Sep 17 00:00:00 2001
From: curben <2809763-curben@users.noreply.gitlab.com>
Date: Sat, 8 Feb 2020 03:32:36 +0000
Subject: [PATCH] build: safer shell script

- https://sipb.mit.edu/doc/safe-shell/
---
 script.sh       | 100 ++++++++++++++++++++++++------------------------
 utils/commit.sh |   2 +-
 2 files changed, 51 insertions(+), 51 deletions(-)

diff --git a/script.sh b/script.sh
index 885b9773..7f9f656e 100644
--- a/script.sh
+++ b/script.sh
@@ -1,88 +1,88 @@
 #!/bin/sh
 
-set -e -x
+set -efux -o pipefail
 
 ## Create a temporary working folder
-mkdir -p tmp/ && cd tmp/
+mkdir -p "tmp/" && cd "tmp/"
 
 
 ## Prepare datasets
-wget https://urlhaus.abuse.ch/downloads/csv/ -O ../src/URLhaus.csv
-wget https://s3-us-west-1.amazonaws.com/umbrella-static/top-1m.csv.zip -O top-1m.csv.zip
+wget "https://urlhaus.abuse.ch/downloads/csv/" -O "../src/URLhaus.csv"
+wget "https://s3-us-west-1.amazonaws.com/umbrella-static/top-1m.csv.zip" -O "top-1m.csv.zip"
 
-cp ../src/exclude.txt .
+cp "../src/exclude.txt" "."
 
 ## Clean up URLhaus.csv
-cat ../src/URLhaus.csv | \
+cat "../src/URLhaus.csv" | \
 # Convert DOS to Unix line ending
 dos2unix | \
 # Remove comment
-sed '/^#/d' | \
+sed "/^#/d" | \
 # Parse URLs
 cut -f 6 -d '"' | \
-cut -f 3- -d '/' | \
+cut -f 3- -d "/" | \
 # Domain must have at least a 'dot'
-grep -F '.' | \
+grep -F "." | \
 # Remove www.
-sed 's/^www\.//g' | \
-sort -u > urlhaus.txt
+sed "s/^www\.//g" | \
+sort -u > "urlhaus.txt"
 
 ## Parse domain and IP address only
-cat urlhaus.txt | \
-cut -f 1 -d '/' | \
-cut -f 1 -d ':' | \
-sort -u > urlhaus-domains.txt
+cat "urlhaus.txt" | \
+cut -f 1 -d "/" | \
+cut -f 1 -d ":" | \
+sort -u > "urlhaus-domains.txt"
 
-cat ../src/URLhaus.csv | \
+cat "../src/URLhaus.csv" | \
 dos2unix | \
-sed '/^#/d' | \
+sed "/^#/d" | \
 # Parse online URLs only
 grep '"online"' | \
 cut -f 6 -d '"' | \
-cut -f 3- -d '/' | \
-sed 's/^www\.//g' | \
-sort -u > urlhaus-online.txt
+cut -f 3- -d "/" | \
+sed "s/^www\.//g" | \
+sort -u > "urlhaus-online.txt"
 
-cat urlhaus-online.txt | \
-cut -f 1 -d '/' | \
-cut -f 1 -d ':' | \
-sort -u > urlhaus-domains-online.txt
+cat "urlhaus-online.txt" | \
+cut -f 1 -d "/" | \
+cut -f 1 -d ":" | \
+sort -u > "urlhaus-domains-online.txt"
 
 
 ## Parse the Cisco Umbrella 1 Million
-unzip -p top-1m.csv.zip | \
+unzip -p "top-1m.csv.zip" | \
 dos2unix | \
 # Parse domains only
-cut -f 2 -d ',' | \
-grep -F '.' | \
+cut -f 2 -d "," | \
+grep -F "." | \
 # Remove www.
-sed 's/^www\.//g' | \
-sort -u > top-1m.txt
+sed "s/^www\.//g" | \
+sort -u > "top-1m.txt"
 
 # Merge Umbrella and self-maintained top domains
-cat top-1m.txt exclude.txt | \
-sort -u > top-1m-well-known.txt
+cat "top-1m.txt" "exclude.txt" | \
+sort -u > "top-1m-well-known.txt"
 
 
 ## Parse popular domains from URLhaus
-cat urlhaus-domains.txt | \
+cat "urlhaus-domains.txt" | \
 # grep match whole line
-grep -Fx -f top-1m-well-known.txt > urlhaus-top-domains.txt
+grep -Fx -f "top-1m-well-known.txt" > "urlhaus-top-domains.txt"
 
 
 ## Parse domains from URLhaus excluding popular domains
-cat urlhaus-domains.txt | \
-grep -F -vf urlhaus-top-domains.txt > malware-domains.txt
+cat "urlhaus-domains.txt" | \
+grep -F -vf "urlhaus-top-domains.txt" > "malware-domains.txt"
 
-cat urlhaus-domains-online.txt | \
-grep -F -vf urlhaus-top-domains.txt > malware-domains-online.txt
+cat "urlhaus-domains-online.txt" | \
+grep -F -vf "urlhaus-top-domains.txt" > "malware-domains-online.txt"
 
 ## Parse malware URLs from popular domains
-cat urlhaus.txt | \
-grep -F -f urlhaus-top-domains.txt > malware-url-top-domains.txt
+cat "urlhaus.txt" | \
+grep -F -f "urlhaus-top-domains.txt" > "malware-url-top-domains.txt"
 
-cat urlhaus-online.txt | \
-grep -F -f urlhaus-top-domains.txt > malware-url-top-domains-online.txt
+cat "urlhaus-online.txt" | \
+grep -F -f "urlhaus-top-domains.txt" > "malware-url-top-domains-online.txt"
 
 
 ## Merge malware domains and URLs
@@ -95,14 +95,14 @@ FIFTH_LINE="! License: https://creativecommons.org/publicdomain/zero/1.0/"
 SIXTH_LINE="! Source: https://urlhaus.abuse.ch/api/"
 COMMENT="$FIRST_LINE\n$SECOND_LINE\n$THIRD_LINE\n$FOURTH_LINE\n$FIFTH_LINE\n$SIXTH_LINE"
 
-cat malware-domains.txt malware-url-top-domains.txt | \
+cat "malware-domains.txt" "malware-url-top-domains.txt" | \
 sort | \
-sed '1 i\'"$COMMENT"'' > ../urlhaus-filter.txt
+sed '1 i\'"$COMMENT"'' > "../urlhaus-filter.txt"
 
-cat malware-domains-online.txt malware-url-top-domains-online.txt | \
+cat "malware-domains-online.txt" "malware-url-top-domains-online.txt" | \
 sort | \
 sed '1 i\'"$COMMENT"'' | \
-sed '1s/Malicious/Online Malicious/' > ../urlhaus-filter-online.txt
+sed "1s/Malicious/Online Malicious/" > "../urlhaus-filter-online.txt"
 
 
 ## Host-only blocklist
@@ -113,13 +113,13 @@ FOURTH_LINE="# License: https://creativecommons.org/publicdomain/zero/1.0/"
 FIFTH_LINE="# Source: https://urlhaus.abuse.ch/api/"
 COMMENT="$FIRST_LINE\n$SECOND_LINE\n$THIRD_LINE\n$FOURTH_LINE\n$FIFTH_LINE"
 
-cat malware-domains.txt | \
+cat "malware-domains.txt" | \
 sort | \
-sed '1 i\'"$COMMENT"'' > ../urlhaus-filter-hosts.txt
+sed '1 i\'"$COMMENT"'' > "../urlhaus-filter-hosts.txt"
 
-cat malware-domains-online.txt | \
+cat "malware-domains-online.txt" | \
 sort | \
 sed '1 i\'"$COMMENT"'' | \
-sed '1s/Malicious/Online Malicious/' > ../urlhaus-filter-hosts-online.txt
+sed "1s/Malicious/Online Malicious/" > "../urlhaus-filter-hosts-online.txt"
 
-cd ../ && rm -r tmp/
+cd ../ && rm -r "tmp/"
diff --git a/utils/commit.sh b/utils/commit.sh
index 15faddd2..7b3d5fc3 100644
--- a/utils/commit.sh
+++ b/utils/commit.sh
@@ -1,6 +1,6 @@
 #!/bin/sh
 
-set -e -x
+set -efux -o pipefail
 
 ## Commit the filter update