ZDDC/classifier/js/copy.js

/**
 * ZDDC Classifier — copy-out (Classify & Copy mode).
 *
 * Copies the fully-classified source files into a SEPARATE output directory
 * under their canonical ZDDC names and folder layout
 *   <party>/{received,issued}/<DATE_TN (STATUS) - TITLE>/<TRACKING_REV (STATUS) - TITLE.ext>
 * The source is never modified — every operation is a read (getFile) on the
 * source and a write into the chosen output handle.
 *
 * Duplicate detection:
 *   - two sources → the same output path  = mapping conflict (skipped + reported)
 *   - target already exists, identical bytes (sha256) = skipped
 *   - target exists, different bytes       = left untouched + reported (no clobber)
 *
 * Built on the generic FS-Access shape (getDirectoryHandle/getFileHandle/
 * createWritable), so it works against a real handle today and a server-backed
 * output handle later without changing this logic.
 */
(function () {
    'use strict';

    var outputHandle = null;     // remembered for the session

    function C() { return window.app.modules.classify; }

    function collectFiles() {
        var out = [];
        (function walk(nodes) {
            (nodes || []).forEach(function (n) {
                (n.files || []).forEach(function (f) { out.push(f); });
                walk(n.children);
            });
        })(window.app.folderTree || []);
        return out;
    }

    // Files that are ready to copy: complete target, not excluded.
    function plan() {
        var c = C(), items = [];
        collectFiles().forEach(function (f) {
            var d = c.deriveTarget(f);
            if (d.excluded || !d.complete) return;
            items.push({ file: f, d: d, outRel: d.outPath + '/' + d.filename });
        });
        return items;
    }

    // Group by output path; >1 source for a path = a mapping conflict.
    function conflictsIn(items) {
        var by = {}, conflicts = [];
        items.forEach(function (p) { (by[p.outRel] = by[p.outRel] || []).push(p); });
        Object.keys(by).forEach(function (k) { if (by[k].length > 1) conflicts.push(k); });
        return { by: by, conflicts: conflicts };
    }

    function toast(msg, level) {
        if (window.zddc && window.zddc.toast) window.zddc.toast(msg, level);
    }
    function setStatus(text) {
        var el = document.getElementById('scanStatus');
        if (!el) return;
        el.textContent = text;
        el.classList.toggle('scanning', !!text);
    }

    async function chooseOutput() {
        if (!window.showDirectoryPicker) {
            toast('Copying to an output directory needs the File System Access API (use Chromium, or run via zddc-server).', 'error');
            return null;
        }
        try {
            var h = await window.showDirectoryPicker({ mode: 'readwrite', id: 'zddc-classifier-output' });
            outputHandle = h;
            C().setOutputName(h.name);
            return h;
        } catch (e) {
            if (e.name !== 'AbortError') toast('Could not open the output directory — ' + (e.message || e), 'error');
            return null;
        }
    }

    async function ensureDir(root, relPath) {
        var parts = relPath.split('/').filter(Boolean);
        var cur = root;
        for (var i = 0; i < parts.length; i++) {
            cur = await cur.getDirectoryHandle(parts[i], { create: true });
        }
        return cur;
    }

    // Resolve a target subdirectory WITHOUT creating it (null if any segment is
    // missing). Lets us check a file's existence cheaply on resume before paying
    // to create the folder chain.
    async function resolveDir(root, relPath, create) {
        var parts = relPath.split('/').filter(Boolean);
        var cur = root;
        for (var i = 0; i < parts.length; i++) {
            try { cur = await cur.getDirectoryHandle(parts[i], create ? { create: true } : undefined); }
            catch (e) { if (!create) return null; throw e; }
        }
        return cur;
    }

    // Resolve a source file's live handle. Fresh-scan files already carry one;
    // snapshot-loaded files resolve lazily from the workspace root by path.
    async function srcHandle(fileObj) {
        if (fileObj.handle) return fileObj.handle;
        if (!window.app.rootHandle) throw new Error('source directory not connected');
        return window.app.modules.scanner.resolveFileHandle(window.app.rootHandle, fileObj);
    }

    // Read a source file's bytes (a File or Blob). A zip member is extracted
    // from its archive (lazily reloaded from the root); a plain file is read
    // through its resolved handle. The source is never written either way.
    async function readSource(fileObj) {
        if (fileObj.isVirtual) {
            return window.app.modules.scanner.extractZipMember(window.app.rootHandle, fileObj);
        }
        return (await srcHandle(fileObj)).getFile();
    }

    // Copy one file. Returns 'copied' | 'skipped' (already present → resumable).
    // The existence check is a cheap stat/HEAD; a present target is left as-is so
    // re-running after an interruption skips the work already done — no source
    // read, no hashing. (Canonical ZDDC names ⇒ same name = same document, and
    // the server archive is WORM, so we never overwrite.)
    // SHA-256 of a source file's bytes, cached on the file object (reused by the
    // duplicate-conflict audit AND the post-copy verify).
    async function sourceSha(fileObj) {
        if (fileObj.sha256) return fileObj.sha256;
        var blob = await readSource(fileObj);
        var h = await window.zddc.crypto.sha256File(blob);
        fileObj.sha256 = h;
        return h;
    }
    async function writeTarget(out, p) {
        var dir = await ensureDir(out, p.d.outPath);
        var srcFile = await readSource(p.file);   // READ source (never write it)
        var fh = await dir.getFileHandle(p.d.filename, { create: true });
        var w = await fh.createWritable();
        await w.write(srcFile);
        await w.close();
    }
    async function copyOne(out, p) {
        // Cheap existence probe: resolve the dir WITHOUT creating it (the HTTP
        // handle doesn't verify here, but getFileHandle below does a HEAD).
        var probe = await resolveDir(out, p.d.outPath, false);
        if (probe) {
            try { await probe.getFileHandle(p.d.filename); return 'skipped'; }
            catch (e) { /* NotFound → write it below */ }
        }
        await writeTarget(out, p);
        return 'copied';
    }
    // Read the written target back and compare its SHA-256 to the source.
    async function verifyOne(out, p) {
        var dir = await resolveDir(out, p.d.outPath, false);
        if (!dir) return false;
        var fh; try { fh = await dir.getFileHandle(p.d.filename); } catch (e) { return false; }
        var th = await window.zddc.crypto.sha256File(await fh.getFile());
        return th === (await sourceSha(p.file));
    }
    async function removeTarget(out, p) {
        var dir = await resolveDir(out, p.d.outPath, false);
        if (dir && dir.removeEntry) { try { await dir.removeEntry(p.d.filename); } catch (e) { /* best effort */ } }
    }

    // Snapshot-loaded files have no live handle — re-grant read on the source
    // (one click) before we read any bytes (hashing or copying). Returns false
    // if the source can't be read.
    async function ensureSourceReadable(items) {
        if (!items.some(function (p) { return !p.file.handle; })) return true;
        if (!window.app.rootHandle) {
            toast('The source directory isn’t connected. Re-open the workspace to reconnect it.', 'error');
            return false;
        }
        var ok = await window.app.modules.persist.verifyPermission(window.app.rootHandle, false);
        if (!ok) { toast('Permission to read the source directory was denied.', 'error'); return false; }
        return true;
    }

    // Group fully-classified files by their canonical output name. Files with the
    // SAME tracking number + revision MUST have the same content: identical bytes
    // collapse to a single copy; differing bytes are a CONFLICT the user must fix.
    async function resolvePlan(items) {
        var by = {};
        items.forEach(function (p) { (by[p.outRel] = by[p.outRel] || []).push(p); });
        var todo = [], conflicts = [], conflictKeys = {}, dupeCount = 0, keys = Object.keys(by);
        for (var i = 0; i < keys.length; i++) {
            var group = by[keys[i]];
            if (group.length === 1) { todo.push(group[0]); continue; }
            var hashes = [], bad = false;
            for (var j = 0; j < group.length; j++) {
                try { hashes.push(await sourceSha(group[j].file)); } catch (e) { bad = true; hashes.push('ERR' + j); }
            }
            var distinct = {}; hashes.forEach(function (h) { distinct[h] = true; });
            if (!bad && Object.keys(distinct).length === 1) {
                todo.push(group[0]); dupeCount += group.length - 1;     // identical → one copy
            } else {
                conflicts.push(keys[i]);
                group.forEach(function (g) { conflictKeys[g.d.key] = true; });
            }
        }
        return { todo: todo, conflicts: conflicts, conflictKeys: conflictKeys, dupeCount: dupeCount };
    }

    // Pre-flight shared by Copy and the standalone "Check" button: hash colliding
    // names, flag conflicts in the UI, return the deduped todo (or null to abort).
    async function preflight(verb) {
        var items = plan();
        if (!items.length) {
            toast('Nothing ' + verb + ' yet — no files are fully classified (need a tracking leaf AND a transmittal).', 'warning');
            return null;
        }
        if (!(await ensureSourceReadable(items))) return null;
        setStatus('Checking for same-name/different-content conflicts…');
        var r = await resolvePlan(items);
        setStatus('');
        C().setHashConflicts(r.conflictKeys);
        if (r.conflicts.length) {
            toast(r.conflicts.length + ' same-name/different-content conflict(s) flagged (≠ in red): same tracking+revision, different bytes. Fix these before copying.', 'error');
        }
        if (r.dupeCount) toast(r.dupeCount + ' exact duplicate(s) collapse to one copy.', 'info');
        return r;
    }

    // Standalone audit (the "Check" button) — flag conflicts without copying.
    async function audit() {
        var r = await preflight('to check');
        if (r && !r.conflicts.length) {
            toast('No conflicts — ' + r.todo.length + ' file' + (r.todo.length === 1 ? '' : 's') + ' ready to copy.', 'success');
        }
        return r;
    }

    async function run() {
        if (!C().isEnabled()) return;
        var r = await preflight('to copy');
        if (!r) return;
        var todo = r.todo;
        if (!todo.length) { if (r.conflicts.length) toast('Resolve the flagged conflicts, then copy.', 'warning'); return; }

        // Where to file the canonical copies: the server archive (HTTP) or a local
        // folder. Both read the source, never write it, both resumable + verified.
        var dest = await chooseDestination(todo.length);
        if (!dest) return;
        return dest === 'server' ? copyToServer(todo) : copyToLocal(todo);
    }

    function summary(s, where) {
        var msg = 'Copy to ' + where + ' — ' + s.copied + ' copied & verified, ' + s.skipped + ' already there'
            + (s.verifyFailed ? (', ' + s.verifyFailed + ' FAILED verification (bad copy removed — re-run)') : '')
            + (s.errors ? (', ' + s.errors + ' errored (retry to resume)') : '') + '.';
        toast(msg, (s.errors || s.verifyFailed) ? 'warning' : 'success');
    }

    async function copyToLocal(todo) {
        var out = outputHandle || await chooseOutput();
        if (!out) return;
        if (!confirm('Copy ' + todo.length + ' file(s) into "' + out.name + '"?\n\n'
            + 'Written under <party>/<received|issued>/<transmittal>/ — pick your archive/ folder to file them directly. '
            + 'Re-running resumes (already-copied files are skipped). The source is not modified.')) return;
        var s = await copyTo(out, todo);
        summary(s, '"' + out.name + '"');
        return s;
    }

    // Copy straight into a project's archive on the server over HTTP (PUT per
    // file, mkdir as needed). Uses the zddc-source HTTP handle, so the SAME copy
    // engine writes <party>/<received|issued>/<transmittal>/<name> under
    // <project>/archive/. The user picks any project they can access.
    async function copyToServer(todo) {
        var src = window.zddc && window.zddc.source;
        if (!src || location.protocol === 'file:') {
            toast('Server copy needs the classifier to be served by a zddc-server (open it over http).', 'error');
            return;
        }
        var projects = await fetchAccessProjects();
        if (projects == null) { toast('Could not load your projects from the server.', 'error'); return; }
        if (!projects.length) { toast('No projects you can access on this server.', 'warning'); return; }
        var proj = await chooseProject(projects);
        if (!proj) return;
        var archive;
        try {
            var rel = proj.url || ('/' + proj.name + '/');
            if (rel.charAt(rel.length - 1) !== '/') rel += '/';
            archive = new URL(rel + 'archive/', location.origin).href;
        } catch (e) { toast('Bad project URL — ' + (e.message || e), 'error'); return; }
        var out = new src.HttpDirectoryHandle(archive, 'archive');
        var s = await copyTo(out, todo);
        summary(s, (proj.title || proj.name) + ' / archive');
        return s;
    }
    // The caller's accessible projects (read view from /.profile/access). Write
    // permission is enforced server-side on PUT, so a 403 surfaces per file.
    async function fetchAccessProjects() {
        try {
            var resp = await fetch('/.profile/access', { headers: { 'Accept': 'application/json' }, credentials: 'same-origin', cache: 'no-cache' });
            if (!resp.ok) return null;
            if ((resp.headers.get('Content-Type') || '').toLowerCase().indexOf('json') === -1) return null;
            var data = await resp.json();
            return Array.isArray(data.projects) ? data.projects : [];
        } catch (e) { return null; }
    }
    function chooseProject(projects) {
        return new Promise(function (resolve) {
            var done = false;
            function finish(v) { if (done) return; done = true; document.removeEventListener('keydown', onKey); back.remove(); resolve(v); }
            function onKey(e) { if (e.key === 'Escape') finish(null); }
            var back = document.createElement('div'); back.className = 'copy-choice__backdrop';
            var box = document.createElement('div'); box.className = 'copy-choice';
            var h = document.createElement('h3'); h.textContent = 'Copy to a project archive';
            var p = document.createElement('p');
            p.innerHTML = 'Files go to <code>&lt;project&gt;/archive/&lt;party&gt;/&lt;received|issued&gt;/&lt;transmittal&gt;/</code>. Pick a project you can access.';
            var sel = document.createElement('select'); sel.className = 'copy-choice__select';
            projects.forEach(function (pr, i) {
                var o = document.createElement('option'); o.value = String(i);
                o.textContent = pr.name + (pr.title ? ' — ' + pr.title : '');
                sel.appendChild(o);
            });
            var row = document.createElement('div'); row.className = 'copy-choice__btns';
            var go = document.createElement('button'); go.className = 'btn btn-primary'; go.textContent = 'Copy here';
            go.addEventListener('click', function () { finish(projects[Number(sel.value)] || null); });
            var cancel = document.createElement('button'); cancel.className = 'btn btn-secondary'; cancel.textContent = 'Cancel';
            cancel.addEventListener('click', function () { finish(null); });
            row.appendChild(go); row.appendChild(cancel);
            box.appendChild(h); box.appendChild(p); box.appendChild(sel); box.appendChild(row);
            back.appendChild(box);
            var pressedBackdrop = false;
            back.addEventListener('mousedown', function (e) { pressedBackdrop = (e.target === back); });
            back.addEventListener('click', function (e) { if (e.target === back && pressedBackdrop) finish(null); });
            document.addEventListener('keydown', onKey);
            document.body.appendChild(back);
        });
    }

    // Tiny modal: choose server archive vs local folder. Resolves 'server' |
    // 'local' | null. The server option is offered only over http(s).
    function chooseDestination(n) {
        return new Promise(function (resolve) {
            var done = false;
            function finish(v) { if (done) return; done = true; document.removeEventListener('keydown', onKey); back.remove(); resolve(v); }
            function onKey(e) { if (e.key === 'Escape') finish(null); }
            var onServer = location.protocol === 'http:' || location.protocol === 'https:';
            var back = document.createElement('div'); back.className = 'copy-choice__backdrop';
            var box = document.createElement('div'); box.className = 'copy-choice';
            var h = document.createElement('h3');
            h.textContent = 'Copy ' + n + ' classified file' + (n === 1 ? '' : 's');
            var p = document.createElement('p');
            p.innerHTML = 'Filed under <code>&lt;party&gt;/&lt;received|issued&gt;/&lt;transmittal&gt;/&lt;name&gt;</code>. '
                + 'Re-running resumes — files already present at the destination are skipped.';
            var row = document.createElement('div'); row.className = 'copy-choice__btns';
            function btn(label, cls, val, disabled) {
                var b = document.createElement('button'); b.className = 'btn ' + cls; b.textContent = label;
                if (disabled) { b.disabled = true; b.title = 'Open the classifier over a zddc-server to enable this'; }
                else b.addEventListener('click', function () { finish(val); });
                return b;
            }
            row.appendChild(btn('☁ Copy to server archive', 'btn-primary', 'server', !onServer));
            row.appendChild(btn('📁 Copy to a local folder…', onServer ? 'btn-secondary' : 'btn-primary', 'local'));
            row.appendChild(btn('Cancel', 'btn-secondary', null));
            box.appendChild(h); box.appendChild(p); box.appendChild(row);
            back.appendChild(box);
            var pressedBackdrop = false;
            back.addEventListener('mousedown', function (e) { pressedBackdrop = (e.target === back); });
            back.addEventListener('click', function (e) { if (e.target === back && pressedBackdrop) finish(null); });
            document.addEventListener('keydown', onKey);
            document.body.appendChild(back);
        });
    }

    // Run the copy loop over a ready list against an output handle. No picker,
    // no confirm — that's run()'s job; this is the engine (and the test seam).
    // Resumable: copyOne skips targets that already exist, so a re-run after an
    // interruption only does the remaining work.
    async function copyTo(out, todo) {
        var s = { copied: 0, skipped: 0, errors: 0, verifyFailed: 0 };
        var copied = [];
        for (var i = 0; i < todo.length; i++) {
            setStatus('Copying… ' + (i + 1) + '/' + todo.length + ' — ' + todo[i].d.filename);
            try {
                var r = await copyOne(out, todo[i]);
                s[r]++;
                if (r === 'copied') copied.push(todo[i]);
            } catch (e) {
                s.errors++;
                if (window.zddc && window.zddc.toast) {
                    window.zddc.toast('Failed to copy ' + todo[i].outRel + ' — ' + (e.message || e), 'error');
                }
            }
        }
        // Verification pass over JUST the files copied this run: read each target
        // back, compare SHA-256 to the source. One re-copy attempt on mismatch;
        // if it still fails, remove the bad target so a re-run re-copies it — so
        // resume converges on a fully-correct archive.
        for (var k = 0; k < copied.length; k++) {
            setStatus('Verifying… ' + (k + 1) + '/' + copied.length + ' — ' + copied[k].d.filename);
            try {
                if (await verifyOne(out, copied[k])) continue;
                await writeTarget(out, copied[k]);
                if (await verifyOne(out, copied[k])) continue;
                s.verifyFailed++;
                await removeTarget(out, copied[k]);
                if (window.zddc && window.zddc.toast) {
                    window.zddc.toast('Verification failed for ' + copied[k].outRel + ' — removed the bad copy; re-run to retry.', 'error');
                }
            } catch (e) {
                s.verifyFailed++;
                if (window.zddc && window.zddc.toast) {
                    window.zddc.toast('Verify error for ' + copied[k].outRel + ' — ' + (e.message || e), 'error');
                }
            }
        }
        setStatus('');
        return s;
    }

    function readyCount() { return plan().length; }

    window.app.modules.copy = {
        run: run,
        audit: audit,
        readyCount: readyCount,
        chooseOutput: chooseOutput,
        // shared with the MDL flow
        fetchAccessProjects: fetchAccessProjects,
        chooseProject: chooseProject,
        // test/advanced seams
        plan: plan,
        conflictsIn: conflictsIn,
        resolvePlan: resolvePlan,
        copyTo: copyTo,
    };
})();