fix: lowercase shop page folders (theme menu 404s), absolute home links; QA suite v3 with live crawl + canonical check

https://projects.knownelement.com/issues/928

pages/01.home/home.md

@@ -9,5 +9,5 @@ Austin, Texas — fabrication, printing, and maker services.
 
 Bookings are open — schedule a shop consultation from the link above.
 
-- [Shop Manual](shop) — how the shop runs: areas, tools, equipment, supplies, safety.
+- [Shop Manual](/shop) — how the shop runs: areas, tools, equipment, supplies, safety.
 - [Book a visit](https://booking.startinglineproductions.com) — schedule a shop consultation.

qa/site-qa.sh

@@ -1,88 +1,93 @@
 #!/bin/bash
 # site-qa.sh - QA for the Grav Website repo and (with -l BASE) the live site.
+#
 #   ./site-qa.sh              repo checks: structure, frontmatter, link targets
 #   ./site-qa.sh -l BASE      live checks: every known route returns 200, all
-#                             internal links/assets resolve, pages have content
+#                             internal links/assets resolve, pages have content,
-# Exit non-zero on any failure. Used by CI and the post-deploy sync job.
+#                             canonical URLs are HTTPS
+#
+# Exit non-zero on any failure. Used by Gitea Actions, the local pre-push
+# hook, and the post-deploy verification in the site's sync job.
 set -uo pipefail
 cd "$(dirname "$0")/.."
 FAIL=0
 fail() { echo "FAIL: $*"; FAIL=$((FAIL+1)); }
-route_of() { # pages-relative dir -> URL route (lowercase, strip NN. prefixes)
+
-  echo "$1" | tr 'A-Z' 'a-z' | sed -E 's|/([0-9]+\.)|/|g; s|^([0-9]+\.)||'
+# pages-relative dir -> URL route: lowercase, strip NN. numeric prefixes
+route_of() {
+  local p=${1#/}
+  p=$(echo "$p" | tr 'A-Z' 'a-z' | sed -E 's@(^|/)[0-9]+[.]@\1@g')
+  printf '%s' "${p:-/}"
 }
 
-repo_checks() {
+# collect route -> file map from the pages tree
-  declare -A ROUTES
+declare -A ROUTES PFILE
-  while IFS= read -r f; do
+while IFS= read -r f; do
   grep -q '^title:' "$f" || fail "no title frontmatter: $f"
-    d=$(dirname "$f"); r=$(route_of "${d#pages/}")
+  d=${f#pages/}            # e.g. 02.shop/01.Welcome/page.md
-    ROUTES["${r:-/}"]=1
+  d=${d%/*}                # 02.shop/01.Welcome
-  done < <(find pages -name '*.md')
+  d=${d%/page}             # (page.md pages keep their folder route)
-  while IFS= read -r f; do
+  r="/$(route_of "$d")"
-    d=$(dirname "$f")
+  # Grav's homepage alias: the configured home folder is served at /
+  [ "$r" = "/home" ] && r="/"
+  ROUTES["$r"]=$f
+  PFILE["$r"]=$f
+done < <(find pages -name '*.md')
+ROUTES["/"]=${PFILE["home"]:-}
+
+repo_checks() {
+  local f r link t
+  for r in "${!ROUTES[@]}"; do
+    f=${ROUTES[$r]}
+    [ -n "$f" ] || continue
     for link in $(grep -oE '\]\([^)#]+' "$f" | sed -E 's/^\]\(//'); do
       case "$link" in http*|mailto:*) continue ;; esac
-      t=$(realpath -m --relative-to=. "$(dirname "$f")/$link" 2>/dev/null) || { fail "unresolvable link $link in $f"; continue; }
+      t="$r/$link"
-      t=${t#pages}; t=$(route_of "$t")
+      t=$(awk -v p="$t" 'BEGIN{n=split(p,seg,"/");out="";for(i=1;i<=n;i++){s=tolower(seg[i]);sub(/^[0-9]+\./,"",s);if(s==""||s==".")continue;if(s==".."){sub(/\/[^\/]*$/,"",out)}else{out=out"/"s}};print (out==""?"/":out)}')
-      [ -n "${ROUTES[${t:-/}]:-}" ] || fail "broken link '$link' in $f (no route '${t:-/}')"
+      [ -n "${ROUTES[$t]:-}" ] || fail "broken link '$link' in ${PFILE[$r]} (no route '$t')"
+    done
   done
-  done < <(find pages -name '*.md')
   echo "repo: ${#ROUTES[@]} routes checked"
 }
 
 live_checks() {
   local BASE=${1%/}
-  local ROUTES=("/")
+  local ROUTELIST=("/") SITEMAP=/tmp/qa-sitemap.txt CHECKED=/tmp/qa-checked.txt
-  while IFS= read -r f; do
+  local r
-    d=$(dirname "$f"); r=$(route_of "${d#pages/}")
+  for r in "${!ROUTES[@]}"; do [ "$r" = "/" ] || ROUTELIST+=("/$r"); done
-    ROUTES+=("/${r#/}")
-  done < <(find pages -name '*.md' | grep -v 'pages/01.home/home.md')
-  local SITEMAP=/tmp/qa-sitemap.txt CHECKED=/tmp/qa-checked.txt
   : > $SITEMAP; : > $CHECKED
-  for r in $(printf '%s\n' "${ROUTES[@]}" | sort -u); do
+  for r in $(printf '%s\n' "${ROUTELIST[@]}" | sort -u); do
     code=$(curl -s -o /dev/null -w '%{http_code}' -m 20 --retry 1 "$BASE$r")
     if [ "$code" = "200" ]; then echo "$BASE$r" >> $SITEMAP; else fail "route $r -> $code"; fi
     sleep 0.2
   done
+  can=$(curl -s -m 15 "$BASE/" | grep -oE '<link rel="canonical" href="[^"]*"' | head -1)
+  case "$can" in
+    *https://*) : ;;
+    "") fail "no canonical link on home" ;;
+    *) fail "canonical not https: $can" ;;
+  esac
+  local url body upath link code
   for url in $(cat $SITEMAP); do
     body=$(curl -s -m 20 "$url"); sleep 0.2
     [ ${#body} -lt 500 ] && fail "page too small ($url, ${#body} bytes)"
-    upath=${url#$BASE}; [ -z "$upath" ] && upath=/
+    upath=${url#$BASE}; upath=${upath:-/}
     for link in $(echo "$body" | grep -oE 'href="[^"]*"' | sed 's/href="//; s/"$//' | sort -u); do
       case "$link" in
         http*|mailto:*|\#*) [[ "$link" == "$BASE"* ]] || continue ;;
         /*) link="$BASE$link" ;;
-        *)  link="$BASE$(realpath -m "$upath/../$link" | sed 's|//\+|/|g')" ;;
+        *)  link="$BASE$(realpath -m "${upath%/}/..${upath%/}/$link" 2>/dev/null | sed 's|//\+|/|g')" ;;
       esac
       grep -qxF "$link" $CHECKED && continue
       echo "$link" >> $CHECKED
       case "$link" in
-        *.css|*.js|*.png|*.jpg|*.jpeg|*.svg|*.ico|*.woff*|*.map|*.webp|*.txt|*.xml) ;;
+        *.css|*.js|*.png|*.jpg|*.jpeg|*.svg|*.ico|*.woff*|*.map|*.webp|*.xml) continue ;;
-        *) continue ;;
       esac
       code=$(curl -s -o /dev/null -w '%{http_code}' -m 20 "$link"); sleep 0.1
-      [ "$code" != "200" ] && fail "asset $link -> $code"
+      [ "$code" != "200" ] && fail "link $link -> $code"
-    done
-    for link in $(echo "$body" | grep -oE 'href="[^"]*"' | sed 's/href="//; s/"$//' | sort -u); do
-      case "$link" in http*|mailto:*|\#*) continue ;; esac
-      case "$link" in
-        /*) L="$BASE$link" ;;
-        *)  L="$BASE$(realpath -m "$upath/../$link" | sed 's|//\+|/|g')" ;;
-      esac
-      grep -qxF "$L" $CHECKED && continue
-      echo "$L" >> $CHECKED
-      case "$L" in
-        *.css|*.js|*.png|*.jpg|*.jpeg|*.svg|*.ico|*.woff*|*.map|*.webp|*.txt|*.xml) continue ;;
-      esac
-      code=$(curl -s -o /dev/null -w '%{http_code}' -m 20 "$L"); sleep 0.1
-      [ "$code" != "200" ] && fail "link $L -> $code"
     done
   done
-  # canonical must be https (reverse-proxy scheme detection)
+  echo "checked ${#ROUTES[@]} routes, $(wc -l < $CHECKED) links/assets"
-  can=$(curl -s -m 15 "$BASE/" | grep -oE '<link rel="canonical" href="[^"]*"' | head -1)
-  case "$can" in *https://*) : ;; "") fail "no canonical link on home" ;; *) fail "canonical not https: $can" ;; esac
-  echo "checked $(wc -l < $SITEMAP) routes, $(wc -l < $CHECKED) links/assets"
 }
 
 case "${1:-}" in