feat(corpus): add save-to-corpus + private corpus search scope

- POST /api/save-to-corpus.php — saves tool output text to user's default CaveauAI corpus via ClientRagPipeline - api/case/upload.php — dual-writes uploaded PDFs to CaveauAI client_documents (best-effort) - assets/js/corpus-save.js — shared <dialog> handler for .js-save-corpus buttons on all tool pages - includes/layout_footer.php — injects corpus-save.js + shared save dialog markup - korrespond/deep-research/barnevernet/discrepancy JS — save-to-corpus buttons on output sections - api/search.php + LegalTools::search() — corpus_scope param ('shared'|'private'|'both'), merges personal CaveauAI corpus with shared legal library when 'both' - includes/tool_form.php + assets/js/tools.js — corpus scope radio toggle shown on search tab - api/user-docs.php — add POST upload method for non-SSO authenticated users Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
2026-05-22 17:50:32 +02:00
parent ed329f9d05
commit b014638f39
13 changed files with 465 additions and 33 deletions
@@ -0,0 +1,95 @@
+<?php
+/**
+ * POST /api/save-to-corpus.php
+ *
+ * Save tool output text into the user's CaveauAI corpus.
+ * Uses dbnToolsBootCaveau() to call ClientRagPipeline directly via filesystem include.
+ *
+ * Request body (JSON, max 500 KB):
+ *   title       string (required)
+ *   content     string (required, min 30 chars)
+ *   source_tool string (optional, slug)
+ *   tags        string (optional, comma-separated)
+ */
+
+declare(strict_types=1);
+
+require_once dirname(__DIR__) . '/includes/bootstrap.php';
+
+dbnToolsRequireMethod('POST');
+dbnToolsRequireAuth();
+
+$user     = dbnToolsAuthenticatedUser();
+$clientId = (int)($user['client_id'] ?? 0);
+if ($clientId <= 0) {
+    dbnToolsError('No linked CaveauAI workspace. Log in via the CaveauAI portal first.', 403, 'no_workspace');
+}
+
+$input      = dbnToolsJsonInput(500_000);
+$title      = trim($input['title'] ?? '');
+$content    = trim($input['content'] ?? '');
+$sourceTool = substr(preg_replace('/[^a-z0-9\-_]/', '', strtolower($input['source_tool'] ?? '')), 0, 64) ?: null;
+$rawTags    = trim($input['tags'] ?? '');
+$tags       = json_encode(
+    array_values(array_filter(array_map('trim', explode(',', $rawTags)))),
+    JSON_UNESCAPED_UNICODE
+);
+
+if ($title === '') {
+    dbnToolsError('title is required.', 400, 'bad_request');
+}
+if (strlen($content) < 30) {
+    dbnToolsError('content too short (min 30 chars).', 400, 'bad_request');
+}
+if (strlen($content) > 2_000_000) {
+    dbnToolsError('content exceeds 2 MB limit.', 400, 'too_large');
+}
+
+// Load CaveauAI platform (getDb, ClientRagPipeline, etc.)
+dbnToolsBootCaveau();
+
+try {
+    $db = getDb();
+} catch (Throwable $e) {
+    dbnToolsError('CaveauAI database unavailable: ' . $e->getMessage(), 503, 'db_unavailable');
+}
+
+// Resolve default corpus for this client
+$stmt = $db->prepare('SELECT id FROM client_corpora WHERE client_id = ? AND is_default = 1 LIMIT 1');
+$stmt->execute([$clientId]);
+$corpusId = (int)($stmt->fetchColumn() ?: 0);
+if ($corpusId === 0) {
+    dbnToolsError(
+        'No default corpus found for your account. Set one up in the CaveauAI portal.',
+        409,
+        'no_corpus'
+    );
+}
+
+$wordCount = str_word_count($content);
+
+$ins = $db->prepare("
+    INSERT INTO client_documents
+      (client_id, corpus_id, title, source_type, content, category,
+       tags, import_method, source_tool, word_count, status)
+    VALUES (?, ?, ?, 'text', ?, 'tool-output', ?, 'tool_output', ?, ?, 'pending')
+");
+$ins->execute([$clientId, $corpusId, $title, $content, $tags, $sourceTool, $wordCount]);
+$docId = (int)$db->lastInsertId();
+
+try {
+    $rag    = new ClientRagPipeline($clientId);
+    $chunks = $rag->ingestDocument($docId);
+} catch (Throwable $e) {
+    // Document is saved but not indexed — mark error and return partial success
+    $db->prepare("UPDATE client_documents SET status='error', error_message=? WHERE id=?")
+       ->execute([$e->getMessage(), $docId]);
+    dbnToolsError(
+        'Saved to corpus but indexing failed: ' . $e->getMessage(),
+        500,
+        'index_failed',
+        ['document_id' => $docId]
+    );
+}
+
+dbnToolsRespond(['ok' => true, 'document_id' => $docId, 'chunks' => $chunks], 201);