update: CSS per-page split and lightningcss unusedSymbols report

docs: record CSS template inventory + decide unusedSymbols-before-split ordering

feat(css): add content-source collector + dynamic-class safelist for unusedSymbols report

- scripts/css-content-sources.mjs: buildCorpus() gathers templates/public/src
  PHP + first-party js/app JS (vendor excluded), returns {corpus, sources,
  totalBytes, safelist, prefixes}
- Mined 22 exact runtime classes + 5 DB/state-derived suffix prefixes from
  status-badge.php, SystemController statusClass/logLineClass, and class=<?=?>
  ternaries
- docs/css-split-analysis.md notes content-corpus section
- td: #11 collect-content-sources done; feeds #12 report script

feat(css): unusedSymbols report script + just css-report recipe

- scripts/css-unused-report.mjs: per-bundle class/id extraction vs buildCorpus()
  corpus + safelists; measures reclaimable bytes via lightningcss transform
  unusedSymbols (report-only, no stripping to disk)
- just css-report: rebuild CSS then run the report
- css-content-sources.mjs: add VENDOR_CLASS_PREFIXES (filepond--*, htmx-*)
- RESULT: 216,383B total, ~6.2KB (2.9%) reclaimable; FilePond/HTMX exclusion
  corrected inflated 26% (56KB) false-positive down to honest 2.9%
- docs/css-split-analysis.md findings table + TODO 12/13 done

docs(css): record go/no-go decision — split NO-GO, pruning conditional-go

- Decision analysis in docs/css-split-analysis.md
- ~6.2KB (2.9%) reclaimable of 216KB; base.min.css only 484B (2.3%)
- SPLIT NO-GO: base.css already well-used; parked u/w/x/y(/z) as deferred
- PRUNING conditional-go on hand-verifiable dead selers from source, never dist;
  re-run just css-report after each edit; keep needs-review + vendor-prefix cls
- td: task 14 done; split stream 41 tasks -> 2 pending / 32 done / 7 deferred

todo: defer CSS pruning stream (10/u/w/x/y/z), context updated

TODO.md

@@ -1,16 +1,13 @@
 # TODO
 
-> Last updated: 2026-08-26
-> Context: Recherche insensible aux accents sur toutes les barres de recherche (publique + backoffice)
+> Last updated: 2026-08-28
+> Context: CSS waste strategy — full stream DEFERRED (tasks 10/u/w/x/y/z): unusedSymbols report showed only ~6.2KB/2.9% of 216KB reclaimable; pruning + per-page split parked (docs/css-split-analysis.md). #t (Interdit admin route) confirmed already implemented via /admin/media.php + media-viewer.php + recapitulatif link. No pending live tasks.
 
 ## In Progress
 
 ## Pending
 
 ## Completed
-- [x] #normalize-synopsis-whitespace & Passages à la ligne inopinés dans les synopsis : jugés INDÉTECTABLES/non-corrigeables automatiquement (une coupure de ligne interne peut être une vraie frontière de mot → espace, ou une coupure du mot → à réjoindre ; indistinguables sans dictionnaire, toute tentative corrompt l'entrée p.ex. "poursuivantsur"). Migration 045 SUPPRIMÉE ; dites à l'admin qu'il n'y a pas de nettoyage fiable des données existantes. Conservation de la NORMALISATION SÛRE sur les nouvelles saisies : fin de ligne \r\n/\r → \n, pertes de lignes vides multiples → \n\n, espaces/tabulations/NBSP → espace unique, trim — retour à la ligne simple intra-paragraphe laissé intact.
-- [x] #recherche-insensible-aux-accents Recherche insensible aux accents sur toutes les barres de recherche : "Théophile" et "Theophile" donnent maintenant les mêmes résultats. Implémentation via une fonction SQLite UDF `accfold()` (app/src/AccentFolding.php, UDF enregistré dans Database::registerSqliteFunctions) qui plie les caractères accentués latin (é→e, ç→c, …) vers leur base ASCII ; chaque condition de recherche compare la colonne repliée au terme replié (`accfold(column) LIKE accfold(:term)`). Appliqué à la recherche publique (searchTheses/countSearchResults sur titre/sous-titre/synopsis/auteurs/promoteurs/tags), au popover étudiant (getThesesByAuthorName/getThesesForAuthors) et à la recherche backoffice par auteur (getThesesList/getThesesListCount). NULL-safe, enregistré sur les connexions prod et de test (TestDatabase). Pas d'intl/iconv dispo → map de translitération manuelle (Western Latin-1 + Latin Extended-A). Évite aussi la dépréciation PHP 8.5 de PDO::sqliteCreateFunction() : création de la connexion via \Pdo\Sqlite (createFunction()) quand dispo, repli sur PDO (sqliteCreateFunction()) sur anciens runtimes ; tests passent de `@dataProvider` docblock à l'attribut #[DataProvider] (fin de la dépréciation PHPUnit). 294 tests verts sans dépréciation ; phpstan OK.
-- [x] #untrack-build-caches Détrache .phpunit.result.cache et .php-cs-fixer.cache (déjà gitignorés) pour qu'ils ne réapparaissent plus comme modifications dans le working copy.
 - [x] #audit-all-docs-and [!high] Audit all docs/ and classify accurate vs stale
 - [x] #rewrite-development-md-to-match [!high] Rewrite development.md to match current just dev / app/ layout / PHPUnit
 - [x] #rewrite-deployment-md-to-match [!high] Rewrite deployment.md to match just deploy / /var/www/xamxam/ / backup
@@ -38,8 +35,19 @@
 - [x] #default-exemplaire-baiu Default Exemplaire BAIU + ERG to true on student (partage) submission in ThesisCreateController, replicating admin add-form default
 - [x] #m-style-links-on Style links on licences, charte, a propos pages (underline + accent hover)
 - [x] #recherche-backoffice-par-identifiant Recherche backoffice par identifiant : ajouter t.identifier a la WHERE de getThesesList/getThesesListCount
-- [x] #admin-open-interdit-files Open 'Interdit' thesis files from backoffice: dedicated /admin/media.php route (AdminAuth session-gated) that bypasses the Interdit visibility gate in MediaController, keeping path jail + MIME checks; recap page links Interdit files to it. Also: file links open via /admin/media-viewer.php HTML wrapper so the tab shows the original file name instead of media.php; media.php sets a meaningful Content-Disposition filename too. Hardened: admin media route restricted to thesis-file prefixes only (defense-in-depth over the MIME/jail checks)
+- [x] #add-admin-only-route-to [!low] Add admin-only route to open Interdit thesis files from backoffice --context repo
+- [x] #identify-which-templates-load [!high] Identify which templates load base.min.css and the page types — DONE inventory: docs/css-split-analysis.md. Key finding: head.php loads base.min.css on EVERY full page (App::render => templates/head.php unconditional).  is the per-page-type mechanism (public/tfe/repertoire/content-page/not-found/form/admin/system). Centralize inventory in docs/css-split-analysis.md. Gates both split (w) and unusedSymbols corpus (11).
+- [x] #collect-content-sources-for [!high] Collect content sources for the unusedSymbols pass — Built scripts/css-content-sources.mjs: buildCorpus() returns {corpus, sources, totalBytes, safelist, prefixes}. Gathers templates/public/src PHP + first-party js/app JS (vendor excluded). Mined 22 exact dynamic classes (status-*, log-*, input-error, active, disabled, btn--*, fhb-*, admin-icon-btn--*, status-published/pending/badge) + 5 suffix prefixes (status-access--*, toc-level-*, admin-import-log__item--*, admin-body, student-body) from status-badge.php, SystemController statusClass/logLineClass, class=<?=...?> ternaries. Documented docs/css-split-analysis.md. Feeds task 12 report.
+- [x] #write-report-script-to [!medium] Write report script to compute unused selectors + reclaimable bytes — Built scripts/css-unused-report.mjs + just css-report recipe. Extracts class/id symbols per dist bundle, matches against buildCorpus() corpus + safelists, measures reclaimable bytes via lightningcss transform unusedSymbols (no stripping to disk). FULL RESULT: 216,383B total, ~6.2KB (2.9%) reclaimable. See docs/css-split-analysis.md table. Feeds task 14 go/no-go.
+- [x] #flag-or-safelist-dynamic [!medium] Flag or safelist dynamic classes before trusting the report — Vendor + dynamic safelists shipped. Dynamic: 22 exact classes (status-*, log-*, input-error, active, disabled, btn--*, fhb-*, admin-icon-btn--*, status-badge/published/pending) + 5 DB/state suffix prefixes (status-access--*, toc-level-*, admin-import-log__item--*, admin-body, student-body). VENDOR: filepond--* and htmx-* prefixes excluded (runtime vendor JS assembly) — without these the report inflated to 26% (56KB); with them honest 2.9%. Report surfaces needs-review (kept, not flagged removable).
+- [x] #summarize-findings-and-decide [!low] Summarize findings and decide whether to proceed to pruning — Decision recorded in docs/css-split-analysis.md. Data: 216,383B total, ~6.2KB (2.9%) reclaimable; base.min.css only 484B. SPLIT: NO-GO (base.css already well-used; split adds complexity for ~2% of one bundle) -> defer u-w-x-y. PRUNING: CONDITIONAL-GO on hand-verifiable dead set only, from SOURCE css files, re-run just css-report after each edit, never touch needs-review/vendor-prefix classes.
 
 ## Deferred / Blocked
 - [ ] #just-setup-backs-a [!medium] just setup backs a stale setup-dev.sh (clones php-live-reload, legacy admin/data/ dirs) — needs rewrite or removal
 - [ ] #csv-export-writes-cc2r-exemplaire [!low] CSV export writes CC2r/Exemplaire columns that import's $cell mapping ignores — document or wire up
+- [ ] #split-base-min-css-into-per-page [!medium] Split base.min.css into per-page bundles to reduce cross-page CSS waste — NO-GO (task 14): per-page split parked. base.min.css already well-used (only 484B/2.3% reclaimable); split adds build/template complexity for ~2% of one bundle. See docs/css-split-analysis.md decision.
+- [ ] #define-per-page-type-css-bundle [!medium] Define per-page-type CSS bundle boundaries and entry files — Using the template->page-type map, decide which source files (components/*, utilities.css, reset, typography, etc.) belong in each page-type bundle. Document the mapping so shared-vs-page-specific splits are explicit and reviewable.
+- [ ] #implement-per-page-bundles-in [!medium] Implement per-page bundles in scripts/build-css.mjs — Extend build-css.mjs with new bundle targets following the existing bundleCss/concatBundle pattern. Emit per-page-type .min.css outputs into app/public/assets/dist/. Keep base.min.css for pages not yet split.
+- [ ] #wire-templates-to-load [!medium] Wire templates to load their page-type bundle instead of base.min.css — Update each template's <link> tags to point at the appropriate per-page bundle. Ensure no page loads CSS it doesn't need and that shared pages still get full styling.
+- [ ] #measure-before-after-css-bytes [!low] Measure before/after CSS bytes shipped per page type — Moot: part of split stream which is NO-GO (task 14). Skip unless split is revived.
+- [ ] #add-lightningcss-unusedsymbols-report-only [!high] Add lightningcss unusedSymbols report-only pass for base.min.css — SEQUENCING: do this (unusedSymbols report-only) BEFORE the per-page split (u). Reasoning: high priority, zero-risk diagnostic-only, produces reclaimable-byte go/no-go (14) against ~216KB baseline; if pruning is not worthwhile the split may not be either. Reuses inventory from v (docs/css-split-analysis.md). Start: 11 collect-content-sources.

app/public/assets/css/admin.css

@@ -1453,16 +1453,11 @@ th.admin-ap-col {
   gap: var(--space-3xs);
   border-top: 1px solid var(--border-primary);
   padding: var(--space-xs) 0;
-	display: grid;
-	grid-template-columns: 1fr;
-	gap: var(--space-3xs);
-	border-top: 1px solid var(--border-primary);
 }
 
 .param-form > div:first-of-type {
   border-top: none;
   padding-top: 0;
-	border-top: none;
 }
 
 .param-form input[type="password"] {

app/public/assets/css/tfe.css

@@ -4,17 +4,14 @@
    ============================================================ */
 
 .tfe-main {
+  display: flex;
   flex: 1;
   min-height: 0;
   overflow: hidden;
-	display: flex;
-	flex: 1;
-	min-height: 0;
-	overflow: hidden;
-	/* Left padding keeps content aligned with the rest of the page;
-	   no right padding so the right column (and its scrollbar) sits
-	   flush against the viewport edge. */
-	padding-left: var(--space-m);
+  /* Left padding keeps content aligned with the rest of the page;
+     no right padding so the right column (and its scrollbar) sits
+     flush against the viewport edge. */
+  padding-left: var(--space-m);
 }
 
 /* Stacked article layout — fills main height */
@@ -60,15 +57,6 @@
   padding: var(--space-xl) var(--space-m) var(--space-xl) var(--space-m);
 }
 
-/* Files container inside right column — flex kept for PDF expand-to-fill behavior */
-	display: flex;
-	flex-direction: column;
-	overflow-y: auto;
-	min-width: 0;
-	padding-top: var(--space-l);
-	padding-bottom: var(--space-xl);
-}
-
 /* Left column: stacks author, title, meta, synopsis — scrolls independently */
 .tfe-left-column {
 	gap: var(--space-m);

docs/css-split-analysis.md

@@ -0,0 +1,162 @@
+# CSS split / unusedSymbols — Template → Page-type inventory
+
+Status: analysis (gates `define-per-page-type-css-bundle` and the unusedSymbols content corpus)
+
+## How CSS is loaded today (head.php)
+
+`App::render()` (app/src/App.php) always `include`s `app/templates/head.php` for a
+full page render. head.php unconditionally emits:
+
+```html
+<link rel="stylesheet" href="/assets/dist/base.min.css">
+```
+
+`base.min.css` = the whole `style.css` `@import` chain:
+reset → colors → typography → base → all `components/*` → utilities. So **every
+full page ships the entire component+utility set**, even pages that barely use it
+(the cross-page waste the split targets).
+
+On top of it, head.php renders `$extraCss` array entries (admin prepends
+`admin.min.css` via `$extraCssAdmin`). Page-specific `.min.css` bundles load as
+`$extraCss`.
+
+Fragments / HTMX partials (`app/templates/partials/*`, `app/public/**/fragments`)
+are included **directly**, not via `App::render()`, so they do NOT load head.php.
+
+## Page-type map (entry → templates → extra css → bodyClass)
+
+### Public (non-admin)
+
+| Page type            | Template(s)                     | extraCss (on top of base)      | bodyClass    | Notes |
+|----------------------|---------------------------------|--------------------------------|--------------|-------|
+| home                 | `public/home.php`               | `public.min.css`               | `home-body`  | HomeController |
+| tfe                  | `public/tfe.php`                | `tfe.min.css`                  | `tfe-body`   | TfeController |
+| repertoire/search    | `public/repertoire.php`, `public/search.php` | `repertoire.min.css` | `search-body` | SearchController |
+| content-page (about / licence / charte) | `public/about.php`, `public/licence.php`, `public/charte.php` | `content-page.min.css` | `apropos-body` | AboutController, LicenceController, CharteController |
+| not-found            | `public/not-found.php`          | `not-found.min.css`            | `page-not-found` | Dispatcher |
+| partage (student form / recap) | `partage/form-page.php` (templates/), `partage/recapitulatif.php`, `partage/retry-email.php`, `partage/index.php` | `form.min.css` (+ `filepond`) | — | FormBootstrap injects `extraCss = [form.min.css]`; partage/index sets `$filepondBase` |
+
+### Admin
+
+head.php prepends `admin.min.css` (`$extraCssAdmin`) whenever `$isAdmin` is set.
+
+| Page type       | Entry (app/public/admin/*.php)     | extra extraCss / extraCssAdmin | bodyClass |
+|-----------------|------------------------------------|--------------------------------|-----------|
+| login / reset   | `login.php`, `password-reset.php`, `request-reset.php` | — | `admin-body` |
+| index           | `index.php`                        | `filepond*.css` (extraCssAdmin) | `admin-body` |
+| recapitulatif   | `recapitulatif.php`                | — | `admin-body student-body` (student mode) |
+| add / edit      | `add.php`, `edit.php`, `contenus-edit.php` | form deps | `admin-body` |
+| contenu mgmt    | `contenus.php`, `tags.php`, `acces*.php`, `cleanup.php`, `account.php`, etc. | — | `admin-body` |
+| parametres      | `parametres.php`                   | `system.min.css` (extraCssAdmin) | `admin-body` |
+
+## Dist bundles present (app/public/assets/dist/)
+
+From `scripts/build-css.mjs`: `base.min.css`, `admin.min.css`, `form.min.css`,
+`public.min.css`, `tfe.min.css`, `repertoire.min.css`, `content-page.min.css`,
+`not-found.min.css`, `system.min.css`, `file-access.min.css`, `common.min.css`.
+
+## Shared core (potential overlap between bundles)
+
+`base.min.css` already contains everything: reset, colors, typography, base,
+components/{links,focus,forms,tables,dialog,details,media,buttons,badges,toast,
+pagination,header,search,toc}, utilities.
+
+The split must decide which of those move OUT of the global base into page-type
+bundles; the components used on every page (reset, colors, typography, base,
+header, footer, search? — see component usage below) stay in a slim base.
+
+## Next step
+
+Component-by-component usage audit: which templates actually reference each
+`components/*` class / utility. That drives `define-per-page-type-css-bundle`
+and feeds the unusedSymbols content corpus.
+
+## Content corpus (for unusedSymbols report)
+
+`scripts/css-content-sources.mjs` builds the content corpus scanned against the
+CSS symbols:
+
+- **Sources**: app/templates/*.php, app/public/*.php, app/src/*.php
+  (controllers + icon.php helper), app/public/assets/js/app/*.js (first-party).
+- **Vendor JS excluded** (htmx/filepond/pdf): their class names are internal to
+  their own bundled CSS, and including them would only *under*-report project
+  usage — the safe direction.
+- **Dynamic-class handling**: `buildCorpus()` returns a safelist of 22 exact
+  runtime class names (status-ok/warn/err/unknown, log-*, input-error, active,
+  disabled, btn--*, fhb-*, admin-icon-btn--*, status-published/pending/badge)
+  plus 5 prefix patterns (status-access--*, toc-level-*, admin-import-log__item--*,
+  admin-body, student-body) whose suffix is DB/state-derived.
+
+Run: `node scripts/css-content-sources.mjs` (prints inventory).
+The report script (scripts/css-unused-report.mjs, task 12) imports `buildCorpus()`.
+
+## unusedSymbols report — findings (diagnostic, task 12/14 go-no-go)
+
+`just css-report` rebuilds CSS then runs `scripts/css-unused-report.mjs`:
+for each dist/*.min.css it extracts class/id symbols, checks them against the
+content corpus + dynamic/vendor safelists, and measures bytes lightningcss would
+reclaim (no stripping applied).
+
+**Result (216,383 B total): ~6.2 KB (2.9%) reclaimable.** Per bundle:
+
+| bundle | orig B | reclaim B | candidate-unused |
+|---|---|---|---|
+| admin.min.css | 55016 | 3060 | admin-import-results*, n-grid, n-section, param-*, admin-toggle*, admin-dialog--sheet, admin-maintenance-* |
+| base.min.css | 21375 | 484 | btn--success, btn--blue, btn--yellow |
+| form.min.css | 41855 | 731 | mode-toggle, licence-generalites, file-preview-list |
+| partage-form.min.css | 37229 | 731 | (same) |
+| form-base.min.css | 19110 | 751 | (same) |
+| public.min.css | 4047 | 228 | card__media--placeholder |
+| system.min.css | 7408 | 185 | sys-status-section |
+| content-page.min.css | 3683 | 32 | heading-permalink |
+
+**Vendor classes excluded** (would over-report + reclaim): filepond--*, htmx-*.
+They are assembled by vendor JS at runtime, so they never appear as literals.
+
+See TODO task 14 for the go/no-go evaluation of this data.
+
+## Decision (task 14): per-page split and CSS pruning
+
+Data: ~6.2 KB (2.9%) reclaimable across 216 KB total. base.min.css itself only
+484 B (2.3%) reclaimable. Waste is concentrated in admin/form/partage-form
+(shared FilePond + form CSS).
+
+**Diagnosis:** base.css is already well-used; the per-page SPLIT would shave
+minimal real weight (base.min.css is ~all-referenced). The unused-symbol PRUNING
+opportunity is small and concentrated in bundles that are already page-type-
+specific (form/admin).
+
+### Split — NO-GO
+
+Splitting base.min.css into per-page bundles was premised on cross-page waste of
+component+utility CSS. The report shows that premise is weak: nearly all of
+base.min.css's classes are referenced somewhere, and shaving a redundant handful
+is a pruning task, not a split. A split adds build/template complexity for ~2%
+of one bundle. → Do not split. Park tasks u-w-x-y as deferred.
+
+### Pruning — CONDITIONAL-GO (narrow, safe subset only)
+
+The only pruning worth doing is hand-verifiable dead selectors, removed from
+SOURCE css files (never from dist) so the next build drops them:
+
+- base.min.css: btn--success, btn--blue, btn--yellow (3, 484 B)
+- public.min.css: card__media--placeholder (228 B)
+- system.min.css: sys-status-section (185 B)
+- content-page.min.css: heading-permalink (32 B)
+- form*.min.css: mode-toggle, mode-toggle--back, licence-generalites,
+  file-preview-list (731 B, duplicated across form-base/form/partage-form)
+- admin.min.css: admin-import-results*, n-grid, n-section, param-*,
+  admin-toggle*, admin-dialog--sheet, admin-maintenance-* (3060 B)
+
+Safety rules:
+1. Delete from app/public/assets/css/** sources, never dist.
+2. Rebuild + re-run `just css-report` after each edit; the class must vanish
+   from the candidate list (not just move to needs-review).
+3. Never remove a `needs-review` (dynamic emitter) or vendor-prefix class
+   (filepond--*, htmx-*).
+4. btn--success/blue/yellow are safe (no live $-built variants); confirm they are
+   not emitted by a helper before deleting.
+5. Green path: `just build-css` + smoke admin/login/form/public pages after.
+
+**Verdict:** park the split (NO-GO); optionally prune the ~6.2 KB dead set as a
+follow-up, lowest-risk slice first (base/public/system/content-page).

justfile

@@ -70,6 +70,13 @@ build-css:
 build-js:
     @node scripts/build-js.mjs
 
+# Diagnostic: report unused CSS symbols per bundle (rebuilds then reports).
+# Does NOT strip anything. Run after changing templates/JS to refresh the corpus.
+[group('build')]
+css-report:
+    @node scripts/build-css.mjs
+    @node scripts/css-unused-report.mjs
+
 [group('build')]
 build-install:
     @npm ci

scripts/css-content-sources.mjs

@@ -0,0 +1,210 @@
+#!/usr/bin/env node
+
+/**
+ * Content corpus for the unusedSymbols report (task 11).
+ *
+ * Reliable unused-selector detection needs a "content corpus": all the text that
+ * carries CSS class/id names that could appear in the HTML at runtime. We feed
+ * lightningcss's `unusedSymbols` option the set of symbols that do NOT appear in
+ * this corpus. Anything absent from the corpus is a candidate for pruning.
+ *
+ * Sources gathered (in logical order):
+ *   - app/templates/*.php   recursively  (full-page templates + partials/fragments)
+ *   - app/public/*.php      recursively  (admin\/partage\/root dispatchers + fragments)
+ *   - app/src/*.php         recursively  (controllers, icon.php, helpers)
+ *   - app/public/assets/js/app/*.js      (first-party JS toggling/inserting classes)
+ *
+ * Vendor JS is EXCLUDED: minified third-party bundles (htmx, filepond, pdf)
+ * reference class names internal to their own CSS, and those names would only
+ * *under*-report project usage — the safe direction. Their CSS is already
+ * independently bundled by build-css.mjs.
+ *
+ * Dynamic-class emitters (ternaries, string concat with `$var` inside
+ * class="...", icon() helper) are handled in two ways:
+ *   1. The literal class names in the PHP source are still captured by the raw
+ *      scan (the static part of `class="btn btn--<?= ... ?>"` yields "btn").
+ *   2. Names produced only at runtime (the `$var` part) are NOT literals — we
+ *      list them in DYNAMIC_CLASS_SAFELIST (task 13) so the report flags them as
+ *      "needs review" instead of silently marking them removable.
+ *
+ * Export:
+ *   buildCorpus()       -> { corpus, sources, totalBytes }
+ *   DYNAMIC_CLASS_SAFELIST -> Set<string>       exact symbols always kept
+ *   DYNAMIC_CLASS_PREFIXES -> string[]          prefix patterns (x--*) kept
+ */
+
+import { readFileSync } from "node:fs";
+import { readdirSync, statSync } from "node:fs";
+import { dirname, extname, join, relative, resolve } from "node:path";
+import { fileURLToPath } from "node:url";
+
+const __dirname = dirname(fileURLToPath(import.meta.url));
+const root = resolve(__dirname, "..");
+const appDir = resolve(root, "app");
+
+/**
+ * Exact class names produced ONLY at runtime (PHP `$var`, ternaries, helper
+ * return values) — the raw corpus text always contains them because they appear
+ * as the literal value in the source (e.g. `return 'status-ok'`, `'active'`).
+ * Listing them here makes the link between dynamic emitters and the CSS explicit
+ * and reviewable. If a reviewer later confirms one is truly unused, it can be
+ * dropped from this set.
+ *
+ * Mined from: status-badge.php ($cssClass), SystemController::statusClass /
+ * logLineClass, and the dynamic class="<?= ... ?>" ternaries in templates/public.
+ */
+const DYNAMIC_CLASS_SAFELIST = new Set([
+  // status-badge.php ($cssClass) — publication + access states
+  "status-published",
+  "status-pending",
+  "status-badge",
+  // SystemController::statusClass($status)
+  "status-ok",
+  "status-warn",
+  "status-err",
+  "status-unknown",
+  // SystemController::logLineClass($line)
+  "log-crit",
+  "log-error",
+  "log-warn",
+  "log-notice",
+  // form field validation state
+  "input-error",
+  // conditional tab / button / pagination state
+  "active",
+  "disabled",
+  "btn--primary",
+  "btn--secondary",
+  "pagination-btn",
+  // admin form-help-inline toggles
+  "fhb-dot--on",
+  "fhb-dot--off",
+  "fhb-inline--disabled",
+  // admin icon button publish states
+  "admin-icon-btn--publish",
+  "admin-icon-btn--unpublish",
+]);
+
+/**
+ * Prefix patterns whose SUFFIX is derived from runtime/DB data (so the exact
+ * symbol is never a literal in the corpus). The report must keep ANY symbol that
+ * starts with one of these prefixes. E.g. `status-access--{slug}` (DB slug),
+ * `toc-level-{n}` (computed), `admin-import-log__item--{type}` (DB type).
+ */
+const DYNAMIC_CLASS_PREFIXES = [
+  "status-access", // status-access--{accessSlug}
+  "toc-level", // toc-level-{n}
+  "admin-import-log__item", // admin-import-log__item--{type}
+  "admin-body", // from $bodyClass var (also student-body)
+  "student-body",
+];
+
+/**
+ * Vendor-generated class families. These are assembled by third-party JS at
+ * runtime (string concat), so no literal appears in the first-party corpus — a
+ * naive scan would (mis-)report them as unused, inflating reclaimable bytes.
+ *
+ * - `filepond--*`  : FilePond builds its DOM/sceleton from class name parts
+ *                    (filepond--item, --panel-root, --action-*, ...) at runtime
+ *                    (see vendor/filepond.min.js). Its CSS is bundled into
+ *                    admin/form/partage-form. MUST stay.
+ * - `htmx-*`       : HTMX toggles htmx-settling/htmx-swapping during swaps.
+ * - add more as the report surfaces vendor families.
+ */
+const VENDOR_CLASS_PREFIXES = [
+  "filepond", // filepond--* (also filepond--image-preview-*, --action-*)
+  "htmx", // htmx-settling, htmx-swapping, htmx-request, htmx-adding-class
+];
+const SCAN_GLOBS = [
+  join(appDir, "templates"), // app/templates/**/*.php
+  join(appDir, "public"), // app/public/**/*.php (fragments + dispatch)
+  join(appDir, "src"), // app/src/**/*.php (controllers, icon.php, helpers)
+  join(appDir, "public/assets/js/app"), // first-party JS
+];
+
+/**
+ * Try <index>.php-style values; returns paths to read.
+ * @param {string} dir
+ */
+function collectPhp(dir) {
+  const out = [];
+  const walk = (d) => {
+    let entries;
+    try {
+      entries = readdirSync(d, { withFileTypes: true });
+    } catch {
+      return;
+    }
+    for (const e of entries) {
+      const p = join(d, e.name);
+      if (e.isDirectory()) {
+        // Skip heavy/generated subtrees we never want in the corpus.
+        if (["assets", "vendor", "node_modules", "dist"].includes(e.name)) continue;
+        // For app/public/assets we only want js/app, handled as its own root.
+        if (e.name === "css" || e.name === "img" || e.name === "icons" || e.name === "fonts") continue;
+        walk(p);
+      } else if (e.isFile() && extname(e.name) === ".php") {
+        out.push(p);
+      }
+    }
+  };
+  walk(dir);
+  return out.sort();
+}
+
+function collectJs(appJsDir) {
+  const out = [];
+  let entries;
+  try {
+    entries = readdirSync(appJsDir, { withFileTypes: true });
+  } catch {
+    return out;
+  }
+  for (const e of entries) {
+    if (e.isFile() && extname(e.name) === ".js") {
+      out.push(join(appJsDir, e.name));
+    }
+  }
+  return out.sort();
+}
+
+/**
+ * Build the full content corpus.
+ * @returns {{ corpus: string, sources: Array<{path,bytes}>, totalBytes: number, safelist: Set<string> }}
+ */
+export function buildCorpus() {
+  const files = [];
+  for (const dir of SCAN_GLOBS) {
+    if (dir.endsWith("js/app")) {
+      files.push(...collectJs(dir).map((f) => ({ path: f, bytes: statSync(f).size })));
+    } else {
+      files.push(...collectPhp(dir).map((f) => ({ path: f, bytes: statSync(f).size })));
+    }
+  }
+  files.sort((a, b) => a.path.localeCompare(b.path));
+
+  const parts = [];
+  for (const f of files) {
+    parts.push(readFileSync(f.path, "utf8"));
+  }
+
+  return {
+    corpus: parts.join("\n"),
+    sources: files,
+    totalBytes: files.reduce((s, f) => s + f.bytes, 0),
+    safelist: DYNAMIC_CLASS_SAFELIST,
+    prefixes: DYNAMIC_CLASS_PREFIXES,
+    vendorPrefixes: VENDOR_CLASS_PREFIXES,
+  };
+}
+
+// CLI: print an inventory summary.
+if (process.argv[1] === fileURLToPath(import.meta.url)) {
+  const c = buildCorpus();
+  console.log(`Content corpus: ${c.corpus.length.toLocaleString()} chars across ${c.sources.length} files`);
+  for (const s of c.sources) {
+    console.log(`  ${relative(root, s.path).padEnd(70)} ${s.bytes.toLocaleString()} B`);
+  }
+  console.log(`Total source bytes: ${c.totalBytes.toLocaleString()} B`);
+  console.log(`Dynamic safelist: ${c.safelist.size} exact + ${c.prefixes.length} prefixes`);
+}

scripts/css-unused-report.mjs

@@ -0,0 +1,168 @@
+#!/usr/bin/env node
+
+/**
+ * CSS unusedSymbols report — DIAGNOSTIC ONLY. Does not modify any bundle.
+ *
+ * Computes which class/id selectors in each built `.min.css` bundle are never
+ * referenced by the runtime content corpus (templates, first-party JS, helpers),
+ * then measures how many bytes lightningcss would reclaim if those unused
+ * symbols were stripped. No stripping is applied to disk.
+ *
+ * Method:
+ *   1. Extract every class/id symbol from the minified CSS text.
+ *   2. Build the content corpus via buildCorpus() (scripts/css-content-sources.mjs).
+ *   3. A symbol is "candidate-unused" if its name does NOT appear (word-bounded)
+ *      anywhere in the corpus and is not covered by the dynamic safelist/prefixes.
+ *   4. Symbols in the dynamic safelist/prefixes but absent from corpus are listed
+ *      as "needs review" (kept, never flagged removable).
+ *   5. Reclaimable bytes = size(minified) - size(minified with `unusedSymbols`
+ *      set to the candidate-unused set).
+ *
+ * Run: node scripts/css-unused-report.mjs [bundle ...]
+ *      (no args => all dist/*.min.css)
+ */
+
+import { readFileSync, readdirSync } from "node:fs";
+import { dirname, resolve } from "node:path";
+import { fileURLToPath } from "node:url";
+import { transform } from "lightningcss";
+import { buildCorpus } from "./css-content-sources.mjs";
+
+const __dirname = dirname(fileURLToPath(import.meta.url));
+const root = resolve(__dirname, "..");
+const distDir = resolve(root, "app/public/assets/dist");
+
+const TARGETS = { chrome: 115 << 16, firefox: 115 << 16, safari: 16 << 16 };
+
+/** Extract distinct `.class` and `#id` symbols from CSS text. */
+function extractSymbols(css) {
+  const names = new Set();
+  const re = /[^\\]([.#])(-?[_a-zA-Z][_a-zA-Z0-9-]*)/g;
+  let m = re.exec(css);
+  while (m) {
+    const prefix = m[1]; // '.' or '#'
+    const name = m[2];
+    // Skip CSS escapes/urls: names immediately followed by nothing else are fine.
+    // url("...png") file extensions (e.g. .otf, .svg) are not real symbols; drop
+    // tokens that are part of a url() or @import by checking the preceding char.
+    const before = css[m.index - 1];
+    // Exclude when preceded by `.` inside a url (e.g. `styles.otf` → ends `.otf`)
+    // by requiring the char before the prefix to be whitespace, `{`, `,`, `(`, `[`, `>`, `~`, `+`, `:`, or start.
+    if (/[\w-]/.test(before)) {
+      m = re.exec(css); // advance
+      continue;
+    }
+    names.add(prefix + name);
+    m = re.exec(css); // advance
+  }
+  return names;
+}
+
+/** True if `name` (a CSS identifier) appears as a whole token in corpus. */
+function symbolInCorpus(name, corpus) {
+  // Escape regex special chars (names are already [A-Za-z0-9-], keep simple).
+  const re = new RegExp(`\\b${name}\\b`, "g");
+  return re.test(corpus);
+}
+
+/** Compress whitespace for a canonical multi-term search on dynamic prefixes. */
+function analyze(bundleName) {
+  const file = resolve(distDir, bundleName);
+  let css;
+  try {
+    css = readFileSync(file, "utf8");
+  } catch {
+    return { bundle: bundleName, error: "not found" };
+  }
+  const { corpus, safelist, prefixes, vendorPrefixes } = buildCorpus();
+
+  const symbols = extractSymbols(css);
+  const unused = [];
+  const needsReview = [];
+  const used = [];
+
+  for (const sym of symbols) {
+    const name = sym.slice(1); // drop '.' or '#'
+    const idOrClass = sym[0];
+    // Id symbols are handled separately; unusedSymbols supports ids, but to keep
+    // the report conservative we only strip class symbols here. (ids are usually
+    // JS/app entry points).
+    const inCorpus = idOrClass === "." ? symbolInCorpus(name, corpus) : true;
+    const safelisted = safelist.has(name);
+    const prefixSafe = prefixes.some((p) => name.startsWith(p));
+    const vendorSafe = vendorPrefixes.some((p) => name.startsWith(p));
+
+    if (inCorpus || prefixSafe || vendorSafe) {
+      used.push(name);
+    } else if (safelisted) {
+      needsReview.push(name); // dynamic emitter exists, but not as literal — keep
+    } else {
+      unused.push(name);
+    }
+  }
+
+  // Reclaimable bytes: original vs stripped.
+  const originalSize = Buffer.byteLength(css, "utf8");
+
+  // NOTE: unusedSymbols in lightningcss expects bare names (no '.'/'#' prefix).
+  const stripped = transform({
+    filename: file,
+    code: Buffer.from(css, "utf8"),
+    minify: true,
+    targets: TARGETS,
+    unusedSymbols: unused,
+  });
+  const strippedSize = Buffer.byteLength(stripped.code, "utf8");
+  const reclaimable = originalSize - strippedSize;
+
+  return {
+    bundle: bundleName,
+    symbols: symbols.size,
+    used,
+    needsReview,
+    unused,
+    originalSize,
+    strippedSize,
+    reclaimable,
+  };
+}
+
+function printReport(results) {
+  let totalOriginal = 0;
+  let totalReclaim = 0;
+  console.log("\nCSS unusedSymbols report (diagnostic only — no stripping applied)\n");
+  for (const r of results) {
+    if (r.error) {
+      console.log(`  ✗ ${r.bundle}: ${r.error}`);
+      continue;
+    }
+    totalOriginal += r.originalSize;
+    totalReclaim += r.reclaimable;
+    console.log(`\n══ ${r.bundle} ══`);
+    console.log(`  symbols: ${r.symbols}   original: ${r.originalSize} B   reclaimable: ${r.reclaimable} B (${((r.reclaimable / r.originalSize) * 100).toFixed(1)}%)`);
+    if (r.unused.length) {
+      console.log(`  ── ${r.unused.length} candidate-unused (would be stripped):`);
+      console.log(`     ${r.unused.join(" ")}`);
+    }
+    if (r.needsReview.length) {
+      console.log(`  ── ${r.needsReview.length} needs-review (dynamic emitter, kept):`);
+      console.log(`     ${r.needsReview.join(" ")}`);
+    }
+  }
+  console.log(`\n────────────────────────────────────`);
+  console.log(`TOTAL original: ${totalOriginal} B   reclaimable: ${totalReclaim} B (${((totalReclaim / totalOriginal) * 100).toFixed(1)}%)`);
+}
+
+function main() {
+  const args = process.argv.slice(2);
+  let bundles;
+  if (args.length) {
+    bundles = args;
+  } else {
+    bundles = readdirSync(distDir).filter((f) => f.endsWith(".min.css"));
+  }
+  const results = bundles.map(analyze);
+  printReport(results);
+}
+
+main();