Bash HTTP(S) download script (pure bash HTTP download for systems without curl/wget/fetch)

bash_wget.sh

#!/bin/bash
#
# bash_wget.sh
#
# Tyler Whittaker <[email protected]>
# 
# BASH HTTP(S) DOWNLOAD SCRIPT
# - downloads a file from HTTP server in almost pure bash (no curl/wget required)
# - requires `openssl` binary for HTTPS download (alternatively, just use curl/wget)
#
# usage: `bash_wget.sh http://www.website.farts/coolfile.zip`
#
# (intended as a fallback for systems without wget/curl/fetch)

# exit immediately on failure
set -e
# enable this to see each command as it executes (i.e. debug mode)
#set -x

# check if URL argument is provided
if [ $# -ne 1 ]; then
    echo "error: please provide exactly one URL as an argument"
    echo "usage: $0 <url>"
    exit 1
fi

url="$1"

# validate URL format (basic check)
if [[ ! $url =~ ^https?:// ]]; then
    echo "error: Invalid URL format. URL must start with http:// or https://"
    exit 1
fi

# extract filename from URL
filename=$(basename "$url")

# if filename is empty or contains only special characters, use a default name
if [ -z "$filename" ] || [[ "$filename" =~ ^[[:punct:]]+$ ]]; then
    filename="downloaded_file"
fi

echo "> downloading $url to $filename..."

# parse URL into components
proto=${url%%://*}
temp=${url#*://}
host=${temp%%/*}
uri=${temp#*/}

# set default port based on protocol
if [ "$proto" = "https" ]; then
    port=443
else
    port=80
fi

# extract port if specified in URL
if [[ $host =~ : ]]; then
    port=${host#*:}
    host=${host%:*}
fi

# function to properly format HTTP request with correct line endings
format_request() {
    printf "GET /%s HTTP/1.1\r\n" "$uri"
    printf "Host: %s\r\n" "$host"
    printf "User-Agent: Bash-Download-Script\r\n"
    printf "Accept: */*\r\n"
    printf "Connection: close\r\n"
    printf "\r\n"
}

# function to extract content after headers
extract_content() {
    local input_file="$1"
    local output_file="$2"
    local start_line
    
    # find the empty line that separates headers from content
    start_line=$(awk '/^\r$/ {print NR; exit}' "$input_file")
    
    if [ -n "$start_line" ]; then
        # add 1 to skip the empty line itself
        start_line=$((start_line + 1))
        # extract everything after the headers
        tail -n +"$start_line" "$input_file" > "$output_file"
        return 0
    else
        return 1
    fi
}

# function to handle HTTPS connections
handle_https() {
    local temp_file=$(mktemp)
    local temp_content=$(mktemp)
    
    # send request and receive response using openssl with proper SSL options
    (format_request; sleep 2) | \
    openssl s_client -connect "${host}:${port}" \
                    -quiet \
                    -no_ssl3 \
                    -no_tls1 \
                    -no_tls1_1 \
                    -ign_eof 2>/dev/null > "$temp_file"

    # check if we received any data
    if [ ! -s "$temp_file" ]; then
        rm -f "$temp_file" "$temp_content"
        return 1
    fi

    # check status code
    local status_code
    status_code=$(head -n 1 "$temp_file" | grep -o '[0-9]\{3\}')
    
    if [ "$status_code" != "200" ]; then
        echo "error: server returned status code $status_code"
        rm -f "$temp_file" "$temp_content"
        return 1
    fi

    # extract content after headers
    if ! extract_content "$temp_file" "$temp_content"; then
        echo "error: failed to extract content from response"
        rm -f "$temp_file" "$temp_content"
        return 1
    fi

    # move content to final destination
    mv "$temp_content" "$filename"
    rm -f "$temp_file"
    return 0
}

# function to handle HTTP connections
handle_http() {
    local temp_file=$(mktemp)
    local temp_content=$(mktemp)
    
    # create TCP connection and send request
    exec 3<>/dev/tcp/"$host"/"$port"
    format_request >&3
    cat <&3 > "$temp_file"
    exec 3>&-
    exec 3<&-
    
    # check if we received any data
    if [ ! -s "$temp_file" ]; then
        rm -f "$temp_file" "$temp_content"
        return 1
    fi

    # check status code
    local status_code
    status_code=$(head -n 1 "$temp_file" | grep -o '[0-9]\{3\}')
    
    if [ "$status_code" != "200" ]; then
        echo "error: server returned status code $status_code"
        cat "$temp_file" >&2
        rm -f "$temp_file" "$temp_content"
        return 1
    fi

    # extract content after headers
    if ! extract_content "$temp_file" "$temp_content"; then
        echo "error: failed to extract content from response"
        rm -f "$temp_file" "$temp_content"
        return 1
    fi

    # move content to final destination
    mv "$temp_content" "$filename"
    rm -f "$temp_file"
    return 0
}

# download file based on protocol
if [ "$proto" = "https" ]; then
    if ! command -v openssl >/dev/null; then
        echo "error: OpenSSL is required for HTTPS downloads"
        exit 1
    fi
    
    if ! handle_https; then
        echo "error: HTTPS download failed"
        exit 1
    fi
else
    if ! handle_http; then
        echo "error: HTTP download failed"
        exit 1
    fi
fi

# verify download was successful
if [ -s "$filename" ]; then
    echo "download completed successfully"
    exit 0
else
    echo "error: download produced empty file"
    rm -f "$filename"
    exit 1
fi