fix fetching mentions when no webmention file exists

- Avoid relative paths, so I can run them from any working directory.
- Make a dedicated curl-wrapping shell script instead of re-defining the
  same alias everywhere.
- Support extended offline periods: allow get-webmentions.sh to fall
  back to the cached copy of my webmentions for up to a day, and don't
  accidentally destroy it; keep changes atomic.
- Verify that the fetched webmentions are legit before replacing the
  cached ones.
- Make shellcheck happy about quoting in vnu.sh by passing the list of
  files with xargs instead of a shell variable.

Makefile.online

@@ -24,7 +24,7 @@ LIGHTHOUSE_ARGS += --budget-path linter-configs/budget.json --output html --outp
 
 hint-online:
 	@echo "Running webhint"
-	@hint --config linter-configs/hintrc $(URLS)
+	hint --config linter-configs/hintrc $(URLS)
 lighthouse:
 	mkdir -p $(CHROME_PROFILE)
 	CHROME_PATH=$(CHROME_PATH) CHROME_PROFILE=$(CHROME_PROFILE) JS_FLAGS='' lighthouse $(URLS) $(LIGHTHOUSE_ARGS)
@@ -40,8 +40,9 @@ axe-ff:
 	@echo "Running axe with Firefox"
 	@scripts/bin/axe-ff $(OUTPUT_DIR) $(URLS)
 
+
 .PHONY: axe axe-ff
 
 .PHONY: all-extra
-all-extra: axe-ff validate-json equal-access htmlproofer lint-css validate-feeds
+all-extra: axe-ff equal-access lint-local
 

linter-configs/browserslistrc

@@ -7,6 +7,7 @@
 
 > 0.1%
 last 5 versions
-Firefox ESR
+Firefox >= 57
+iOS >= 9.1
 not dead
-IE 9-11
+IE 11

linter-configs/hintrc

@@ -3,7 +3,8 @@
 		"name": "puppeteer",
 		"options": {
 			"browser": "Chromium",
-			"headless": true
+			"headless": true,
+			"puppeteerOptions": {"args": ["--no-sandbox"]}
 		}
 	},
 	"extends": [
@@ -43,8 +44,6 @@
 			"error",
 			{
 				"ignore": [
-					"time",
-					"picture",
 					"a[download]",
 					"a[referrerpolicy]",
 					"a[translate]",
@@ -52,7 +51,6 @@
 					"samp[translate]",
 					"span[translate]",
 					"img[decoding]",
-					"meta[name=color-scheme]",
 					"meta[name=theme-color]"
 				]
 			}
@@ -64,12 +62,20 @@
 				"maxAgeTarget": 10800
 			}
 		],
-		"http-compression": "off",
+		"http-compression": ["error", {
+			"resource": {
+				"zopfli": false
+			},
+			"html": {
+				"zopfli": false
+			}
+		}],
 		"https-only": "error",
 		"sri": "error",
-		"detect-css-reflows": "off",
-		"detect-css-reflows/layout": "off",
-		"detect-css-reflows/paint": "off",
+		"no-http-redirects": "error",
+		"detect-css-reflows/layout": "hint",
+		"detect-css-reflows/paint": "hint",
+		"detect-css-reflows/composite": "hint",
 		"manifest-exists": "error",
 		"manifest-file-extension": "error",
 		"manifest-is-valid": "error",
@@ -83,10 +89,7 @@
 		"no-vulnerable-javascript-libraries": "off",
 		"html-checker": "off",
 		"image-optimization-cloudinary": "off",
-		"no-disallowed-headers": [ "error",
-			{
-				"ignore": ["x-frame-options"]
-			}],
+		"no-disallowed-headers": "error",
 		"meta-charset-utf-8": "error",
 		"disown-opener": "error",
 		"content-type": ["error", {

scripts/bin/axe-ff

@@ -1,7 +1,9 @@
 #!/bin/sh
+
+set -e -u
 # Runs axe on every page of my sitemap
 # first arg is output directory, after that comes urls from a sitemap. We replace the urls with equivalent local file paths.
-PATH="scripts/bin:$PATH"
+PATH="$(dirname "$0"):$PATH"
 output_dir="$1"
 shift
 urls="$(echo "$*" | tr ' ' '\n' | sd '^https?://[^/]*' "file://$PWD/$output_dir" | sd '/$' '/index.html' | tr '\n' ' ')"

scripts/bin/validate-feed

@@ -39,7 +39,6 @@ done
 
 url="$1"
 
-
 find_rel_mention_count() {
 	curl -s "$url" | grep -cF 'rel="mentioned"'
 }

scripts/connectivity-check.sh

@@ -1,14 +1,14 @@
 #!/bin/sh
 set -e -u
 
-alias curl_cmd='curl --proto "=https" --proto-default https --http2 -siSL --tlsv1.3 --cert-status'
-
 ipv6_success=1
 ipv4_success=1
 
-curl_cmd -6 'seirdy.one/ip' || ipv6_success=0
+curl_wrapper="$(dirname "$0")/curl-wrapper.sh"
+
+"$curl_wrapper" -6 'https://seirdy.one/ip' || ipv6_success=0
 echo
-curl_cmd -4 'seirdy.one/ip' || ipv4_success=0
+"$curl_wrapper" -4 'https://seirdy.one/ip' || ipv4_success=0
 echo
 
 if [ "$ipv6_success" = 0 ] && [ "$ipv4_success" = 0 ]; then

scripts/curl-wrapper.sh

@@ -0,0 +1,3 @@
+#!/bin/sh
+# a curl wrapper for seirdy.one
+curl --proto "=https" --tlsv1.3 --cert-status -sS -m10 "$@"

scripts/get-webmentions.sh

@@ -2,18 +2,50 @@
 #
 # Script to fetch all approved webmentions from webmentiond as a big json response.
 # Uses POSIX and cURL in CI, also uses any pass/pash-compatible pwmngr otherwise
-# The response is cached for 90 minutes.
+# The response is cached for 90 minutes. If fetching fresh webmentions fails, fall back to the cached version for up to one day (1440 minutes). If cached webmentions are older than a day, error out.
 
 set -e -u
 
+dirname="$(dirname "$0")"
+curl_wrapper="$dirname/curl-wrapper.sh"
 auth_url='https://seirdy.one/webmentions/authenticate/access-key'
 webmentions_url='https://seirdy.one/webmentions/manage/mentions?limit=9999&status=approved'
-webmentions_file="$(realpath data/webmentions.json)"
+webmentions_file="$dirname/../data/webmentions.json"
 
-# just a little curl wrapper I use on seirdy.one
-alias ccurl='curl --proto "=https" --proto-default https --tlsv1.3 --cert-status'
+stale_after_minutes=90
+fallback_stale_after_minutes=1440
 
-# use a long-lived key (password) to fetch a short-lived bearer token.
+skip_check=0
+times_run=0
+
+check_cached_webmentions() {
+	times_run="$((times_run + 1))"
+	if [ "$skip_check" = 0 ]; then
+		expires_in="$fallback_stale_after_minutes"
+		if [ $# -gt 0 ]; then
+			expires_in="$1"
+		fi
+		exit_status=1
+		if [ -f "$webmentions_file" ]; then
+			old_webmention_file="$(find "$webmentions_file" -mmin +"$expires_in")"
+			if [ "$old_webmention_file" = "" ]; then
+				echo 'Using cached webmentions'
+				skip_check=1
+				exit_status=0
+				return 0
+			fi
+			return 1
+		fi
+		if [ "$exit_status" = 1 ] && [ "$times_run" = 2 ]; then
+			echo "Webmentions are outdated. failed to fetch for over a day."
+			exit "$exit_status"
+		fi
+	fi
+}
+
+trap check_cached_webmentions EXIT
+
+# Grab my long-lived key (password). We will use this to authenticate.
 key() {
 	set +u
 	if [ -n "$BUILD_SUBMITTER" ]; then
@@ -24,23 +56,36 @@ key() {
 	set -u
 }
 
+# Fetch a short-lived access token from my webmention receiver.
 token() {
 	key_response="$(key)"
-	ccurl -sX POST "$auth_url" -d "key=$key_response"
+	"$curl_wrapper" -sX POST "$auth_url" -d "key=$key_response"
 }
 
-# use that token to fetch all webmentions
+# Verify that the webmentions file has downloaded succesfully by ensuring that it starts and ends with a given substring.
+# The total number of webmentions should be at least a 3-digit number.
+verify_webmentions() {
+	grep -E '^\{"items":\[\{"id":".*,"total":[0-9]{3}([0-9]*)?\}$' "$webmentions_file.tmp" >/dev/null
+}
+
+# use the token to fetch all webmentions.
 fetch_webmentions() {
+	echo 'Fetching webmentions'
 	token_response="$(token)"
-	ccurl --compressed -H "Authorization: Bearer $token_response" "$webmentions_url" -o "$webmentions_file"
+	"$curl_wrapper" --compressed -H "Authorization: Bearer $token_response" "$webmentions_url" -o "$webmentions_file.tmp" || return 1
+	if verify_webmentions; then
+		mv "$webmentions_file.tmp" "$webmentions_file"
+	else
+		echo 'Error: webmentions failed to verify'
+		exit 1
+	fi
 }
 
 # fetch webmentions if we don't have a fresh copy already.
-
-if [ -f "$webmentions_file" ] \
-	&& [ "$(find "$webmentions_file" -mmin +90)" = "" ]; then
-	echo 'Using cached webmentions'
+if [ -f "$webmentions_file" ]; then
+	if ! check_cached_webmentions "$stale_after_minutes"; then
+		fetch_webmentions
+	fi
 else
-	echo 'Fetching webmentions'
 	fetch_webmentions
 fi

scripts/populate-webrings.sh

@@ -8,7 +8,6 @@
 
 set -e -u
 
-
 # the name of this program
 progname="$(basename "$0")"
 dirname="$(dirname "$0")"
@@ -58,15 +57,15 @@ done
 endless_orbit() {
 	printf 'Endless Orbit,'
 	curl -sSL --compressed https://linkyblog.neocities.org/onionring/onionring-variables.js \
-	| grep -C 1 https://seirdy.one/ \
-	| tr -d "'\n" | sed 's|https://seirdy.one/|https://linkyblog.neocities.org/webring.html|'
+		| grep -C 1 https://seirdy.one/ \
+		| tr -d "'\n" | sed 's|https://seirdy.one/|https://linkyblog.neocities.org/webring.html|'
 	echo 'null'
 }
 
 if [ "$dry_run" = '1' ]; then
 	endless_orbit
 elif [ -f "$webrings_dest" ]; then
-		echo "webrings file already generated"
+	echo "webrings file already generated"
 else
 	endless_orbit | cat "$webrings_src" - >"$webrings_dest"
 fi

scripts/vnu.sh

@@ -1,6 +1,6 @@
 #!/bin/sh
 set -e -u
-pwd="$(dirname "$0")"
+dirname="$(dirname "$0")"
 output_dir="$1"
 
 find_files_to_analyze() {
@@ -8,12 +8,9 @@ find_files_to_analyze() {
 		| grep -Ev '(bimi\.svg|search/index\.x?html)$'
 }
 
-files_to_analyze="$(find_files_to_analyze)"
+# files_to_analyze="$(find_files_to_analyze)"
 
 # we skip the BIMI icon (VNU can't handle SVG 1.2) and the search page (it has raw templates).
-vnu \
-	--stdout \
-	--format json \
-	--also-check-svg \
-	$files_to_analyze \
-	| sh "$pwd/filter-vnu.sh"
+find_files_to_analyze \
+	| xargs vnu --stdout --format json --also-check-svg \
+	| sh "$dirname/filter-vnu.sh"

scripts/xhtmlize-single-file.sh

@@ -23,7 +23,7 @@ cleanup() {
 }
 trap cleanup EXIT
 
-run_tidy () {
+run_tidy() {
 	tidy -asxhtml -config linter-configs/tidy.conf 2>/dev/null || true
 }
 
@@ -36,13 +36,13 @@ sed 7d "$html_file" | xmllint --format --encode UTF-8 --noent - | tail -n +2 >"$
 	# shellcheck disable=SC2016 # these are regex statements, not shell expressions
 	#shellcheck source=/home/rkumar/Executables/ghq/git.sr.ht/~seirdy/seirdy.one/scripts/xhtmlize.sh
 	sed \
-			-e '1,7d' \
-			-e 's|</span>(&nbsp;)?.span itemprop="familyName|</span>&#160;<span itemprop="familyName"|' \
-			-e 's|class="u-photo photo"[^<]*<|class="u-photo photo"/> <|' \
-			-E \
-			-e 's|([a-z])<data|\1 <data|' \
-			-e 's#</span>(<a[^>]*rel="(nofollow ugc|ugc nofollow)"([^>]*)?>liked</a>)#</span> \1#' \
-			-e 's#^[\t\s]*<(code|/pre)#<\1#' \
-			"$tmp_file" \
+		-e '1,7d' \
+		-e 's|</span>(&nbsp;)?.span itemprop="familyName|</span>&#160;<span itemprop="familyName"|' \
+		-e 's|class="u-photo photo"[^<]*<|class="u-photo photo"/> <|' \
+		-E \
+		-e 's|([a-z])<data|\1 <data|' \
+		-e 's#</span>(<a[^>]*rel="(nofollow ugc|ugc nofollow)"([^>]*)?>liked</a>)#</span> \1#' \
+		-e 's#^[\t\s]*<(code|/pre)#<\1#' \
+		"$tmp_file" \
 		| awk '/^<\/code>/{printf "%s",$0;next}7'
 } >"$html_file"