msstreamdownloader/downloadPOSIX.sh

#!/bin/sh
# Don't link mrsh to sh, since it does not allow line continuations in here-documents.
# Don't use tcsh, it doesn't allow multiple lines in braced or parenthesized blocks.
# Use dash, yash, bash, posh, ksh, mksh, zsh, pbosh

# Run entire script in subshell to prevent variable pollution
(
# https://unix.stackexchange.com/a/188365
# If and only if bash is in POSIX-mode, which can be forced by setting the
# POSIXLY_CORRECT variable (to any value), then the special built-ins, see
# https://www.gnu.org/software/bash/manual/html_node/Special-Builtins.html,
# are found BEFORE functions during lookup. `unset' is one of these special
# built-ins. However, when using `shopt -s expand_aliases', alias expansion in
# non-interactive shells is supported, so aliasing unset is a problem. 
# This can be circumvented by escaping any of the letters of the command, usually the first.
#
# The COMMANDS_ variable contains the commands, space delimited (therefore the IFS),
# which are supposed to be "made safe".
IFS="
	 "
POSIXLY_CORRECT='1'
COMMANDS_='builtin unalias unset read printf command exit type . tr mkdir wc sed grep xargs ffmpeg jq wget'
# shellcheck disable=SC2086
\unset -f -- ${COMMANDS_-} 2>/dev/null
# shellcheck disable=SC2086
\unalias -- ${COMMANDS_-} 2>/dev/null || true

command -v -- wget >/dev/null 2>&1 || {
    printf '\033[31m%s\033[m\n' "Please install wget"
    exit 1
}

command -v -- jq >/dev/null 2>&1 || {
    printf '\033[31m%s\033[m\n' "Please install jq"
    exit 1
}

command -v -- ffmpeg >/dev/null 2>&1 || {
    printf '\033[31m%s\033[m\n' "Please install ffmpeg"
    exit 1
}

# .env contains a line like
# BASE64_BEARER_TOKEN='<literal base64 encoded string>'
# Retrieve that base64 token from a request from the browser
# network tab network tabfrom a logged in MS Stream (Classic) tab
. './.env'

OUTDIR='./videos'

read -r API_URL <<- EOM
https://euwe-1.api.microsoftstream.com/api/videos?\
\$top=4\
&\
\$skip=0\
&\
\$orderby=metrics/trendingScore asc\
&\
\$filter=\
		published \
		and \
		(\
				(\
						state eq 'Completed' and contentSource ne 'livestream'\
				) \
				or \
				(\
						contentSource eq 'livestream' \
						and not \
						(\
								state eq 'Completed' \
								and not \
								liveEvent/LiveEventOptions/OnDemandOptions/EnablePlayback\
						)\
				)\
		)\
&\
api-version=1.4-private
EOM

API_URL="$(
    printf '%s' "${API_URL-}" |
    tr -d -- '\t'
)"

mkdir -p -- "${OUTDIR-}" || {
    printf '\033[31m%s\033[m\n' 'Aborting'
    exit 2
}

# '>' without quotes is delimiter, since it is one of the few characters disallowed in a URL. Title must not include this character
# https://www.ietf.org/rfc/rfc2396.txt
# 2.4.3. Excluded US-ASCII Characters
m3u8_manifest_urls_and_metadata="$(
    wget --no-verbose                                               \
        --quiet                                                     \
        --output-document -                                         \
        --header="authorization: Bearer ${BASE64_BEARER_TOKEN-}"     \
        -- "${API_URL-}"                                             |
        jq --raw-output                                             \
        '.value[] | .name + ">" + .media.duration[2:] + ">" + (.playbackUrls | map(select(.mimeType == "application/vnd.apple.mpegurl")) | .[].playbackUrl)'
)"

[ -z "${m3u8_manifest_urls_and_metadata-}" ] && {
    printf '\033[31m%s\033[m\n' 'Error GETting manifest urls (response empty) or Bearer token invalid/expired. Exiting.'
    exit 3
}

idx='1'
total="$(printf '%s\n' "${m3u8_manifest_urls_and_metadata-}" | wc -l)"

IFS='
'
for m3u8_manifest_url_and_metadata in ${m3u8_manifest_urls_and_metadata-}
do
    IFS='>' read -r title length m3u8_manifest_url <<- EOM
				${m3u8_manifest_url_and_metadata-}
		EOM

    length="$(printf '%s' "${length-}" | sed -e 's/\..*S/S/g' -e 's/M/:/g' -e 's/H/:/g')"

    filepath="${OUTDIR-}/${title-}.mp4"

    printf '\033[32m%s\033[m\n' "Download '${filepath-}' (${idx-}/${total-})"

    ffmpeg -hide_banner                                                \
        -loglevel error                                                \
        -headers "authorization: Bearer ${BASE64_BEARER_TOKEN-}"        \
        -i "${m3u8_manifest_url-}"                                      \
        -progress -                                                    \
        -c copy                                                        \
        "${filepath-}"                                                  |
            grep --line-buffered 'out_time='                           |
            sed -u -e 's/out_time=//g' -e 's/\..*//g' -e 's/^-//g'     |
            xargs -I{} printf '%s\r' "{} of ${length-}       "

    printf '\n\033[33m%s\033[m\n' "Done"

    idx="$(( idx + 1 ))"
done

unset -v POSIXLY_CORRECT        \
idx                             \
total                           \
API_URL                         \
filepath                        \
m3u8_manifest_url               \
m3u8_highest_res_url            \
m3u8_manifest_url_and_metadata  \
m3u8_manifest_urls_and_metadata \
OUTDIR                          \
BASE64_BEARER_TOKEN

IFS="
	 "
)