v3.2.0: LLM-извлечение + сравнение допников — кнопка Сравнить

site/app.py

@@ -13,6 +13,7 @@ from api import api_bp
 from routes.main import index
 from routes.parse import parse
 from routes.misc import health, logs
+from routes.llm import process as llm_process
 
 load_dotenv()
 
@@ -43,6 +44,9 @@ class ContractsApp:
         # Парсинг: SSE-поток
         self.app.add_url_rule("/parse/<cid>", "parse", parse)
 
+        # LLM: извлечение + сравнение
+        self.app.add_url_rule("/llm/process/<cid>", "llm_process", llm_process)
+
         # Служебные
         self.app.add_url_rule("/health", "health", health)
         self.app.add_url_rule("/logs", "logs", logs)

site/routes/llm.py

@@ -0,0 +1,130 @@
+"""routes/llm.py — LLM-извлечение + сравнение: один SSE-поток."""
+
+import json as _json
+import time as _time
+
+from flask import Response
+
+import db
+import extractor
+import differ
+
+
+def process(cid):
+    """
+    GET /llm/process/<cid> — SSE-поток полного цикла:
+
+      1. Для каждого файла договора: parsed_text → LLM → spec_rows
+      2. Сравнение допников через differ.diff()
+      3. Результат → SSE-сообщения в браузер
+    """
+    def generate():
+        start_time = _time.time()
+
+        # ── Получить все файлы договора ─────────────────────
+        supps, _ = db.query(
+            "SELECT s.id as supp_id, s.type, d.id as doc_id, d.filename, d.parsed_text "
+            "FROM supplements s JOIN documents d ON s.document_id = d.id "
+            "WHERE s.contract_id = %s AND d.parsed_text IS NOT NULL "
+            "ORDER BY s.created_at",
+            (cid,),
+        )
+        if not supps:
+            yield f"data: {_json.dumps({'type': 'error', 'message': 'Нет распарсенных файлов. Сначала нажмите Парсинг.'})}\n\n"
+            return
+
+        supp_rows = [dict(zip(supps["columns"], r)) for r in supps["rows"]]
+        extracted = {}  # supp_id → [rows]
+
+        # ── Шаг 1: LLM-извлечение для каждого файла ─────────
+        for s in supp_rows:
+            # Пропустить уже извлечённые
+            existing, _ = db.query(
+                "SELECT 1 FROM spec_rows WHERE supplement_id = %s LIMIT 1",
+                (s["supp_id"],),
+            )
+            if existing and existing["rows"]:
+                # Загрузить существующие
+                rows_res, _ = db.query(
+                    "SELECT row_num, name, price, qty, sum, date_start "
+                    "FROM spec_rows WHERE supplement_id = %s ORDER BY row_num",
+                    (s["supp_id"],),
+                )
+                rows = [dict(zip(rows_res["columns"], r)) for r in rows_res["rows"]] if rows_res else []
+                extracted[s["supp_id"]] = rows
+                yield f"data: {_json.dumps({'type': 'extract_skip', 'name': s['filename'], 'reason': 'уже извлечено', 'count': len(rows)})}\n\n"
+                continue
+
+            yield f"data: {_json.dumps({'type': 'extract_start', 'name': s['filename']})}\n\n"
+
+            t0 = _time.time()
+            result = extractor.extract(s["parsed_text"])
+
+            if "error" in result:
+                yield f"data: {_json.dumps({'type': 'extract_error', 'name': s['filename'], 'error': result['error']})}\n\n"
+                continue
+
+            rows = result.get("rows", [])
+            unresolved = result.get("unresolved", [])
+            elapsed = round(_time.time() - t0, 2)
+
+            # Сохранить в БД
+            db.execute("DELETE FROM spec_rows WHERE supplement_id = %s", (s["supp_id"],))
+            for row in rows:
+                db.execute(
+                    "INSERT INTO spec_rows (supplement_id, row_num, name, price, qty, sum, date_start) "
+                    "VALUES (%s, %s, %s, %s, %s, %s, %s)",
+                    (
+                        s["supp_id"],
+                        row.get("row_num"),
+                        row.get("name"),
+                        row.get("price"),
+                        row.get("qty"),
+                        row.get("sum"),
+                        row.get("date_start"),
+                    ),
+                )
+
+            extracted[s["supp_id"]] = rows
+            yield f"data: {_json.dumps({'type': 'extract_done', 'name': s['filename'], 'time_s': elapsed, 'count': len(rows), 'unresolved': unresolved})}\n\n"
+
+        # ── Шаг 2: Сравнение допников ────────────────────────
+        yield f"data: {_json.dumps({'type': 'diff_start', 'files': len(supp_rows)})}\n\n"
+
+        # Группируем: initial + amendments
+        initial_rows = None
+        amendments = []
+
+        for s in supp_rows:
+            if s["type"] == "initial":
+                initial_rows = extracted.get(s["supp_id"], [])
+            else:
+                amendments.append({
+                    "supp_id": s["supp_id"],
+                    "filename": s["filename"],
+                    "type": s["type"],
+                    "rows": extracted.get(s["supp_id"], []),
+                })
+
+        all_changes = []
+
+        if initial_rows is not None:
+            for am in amendments:
+                if not am["rows"]:
+                    continue
+                d = differ.diff(initial_rows, am["rows"])
+                changes = d.get("changes", [])
+                summary = d.get("summary", {})
+                all_changes.append({
+                    "filename": am["filename"],
+                    "type": am["type"],
+                    "changes": changes,
+                    "summary": summary,
+                })
+                yield f"data: {_json.dumps({'type': 'diff_file', 'name': am['filename'], 'changes': len(changes), 'summary': summary})}\n\n"
+
+        # ── Итог ────────────────────────────────────────────
+        total_time = round(_time.time() - start_time, 2)
+        yield f"data: {_json.dumps({'type': 'done', 'total_time_s': total_time, 'all_changes': all_changes})}\n\n"
+
+    return Response(generate(), mimetype="text/event-stream")

site/templates/upload.html

@@ -54,12 +54,19 @@
     .modal-body .kv .k { color: var(--muted); width: 110px; flex-shrink: 0; }
     .modal-body .kv .v { word-break: break-all; }
     .modal-body pre { background: var(--brand-grey-light); padding: 10px; border-radius: 6px; font-size: 12px; max-height: 200px; overflow-y: auto; white-space: pre-wrap; margin-top: 4px; }
+    .diff-added { background: rgba(34,197,94,.1); }
+    .diff-deleted { background: rgba(239,68,68,.1); text-decoration: line-through; }
+    .diff-changed { background: rgba(234,179,8,.1); }
+    .diff-added td { color: var(--green); }
+    .diff-deleted td { color: var(--destructive); }
+    .diff-field-old { color: var(--destructive); text-decoration: line-through; font-size: 11px; }
+    .diff-field-new { color: var(--green); font-weight: 600; }
   </style>
 </head>
 <body>
   <div class="topbar">
     <img src="{{ url_for('static', filename='nubes-logo.svg') }}" alt="Nubes">
-    <span class="title">Сверка договоров <span style="font-weight:400;color:var(--muted);font-size:12px;">v3.1.1</span></span>
+    <span class="title">Сверка договоров <span style="font-weight:400;color:var(--muted);font-size:12px;">v3.2.0</span></span>
   </div>
 
   <div class="content">
@@ -83,8 +90,21 @@
         <button class="btn btn-primary" id="parseBtn" style="width:100%;justify-content:center;margin-top:12px;" disabled>
           <i data-lucide="play" style="width:16px;height:16px;"></i> Парсинг
         </button>
+
+        <button class="btn btn-primary" id="llmBtn" style="width:100%;justify-content:center;margin-top:8px;display:none;">
+          <i data-lucide="scale" style="width:16px;height:16px;"></i> Сравнить
+        </button>
       </div>
     </div>
+
+    <!-- Результаты сравнения -->
+    <div class="card" id="diffCard" style="display:none;">
+      <div class="card-header">
+        <i data-lucide="file-diff" style="width:18px;height:18px;"></i>
+        Результаты сравнения <span id="diffStatus" style="font-weight:400;font-size:12px;margin-left:8px;"></span>
+      </div>
+      <div class="card-body" id="diffBody"></div>
+    </div>
   </div>
 
   <div class="modal-overlay" id="modalOverlay" onclick="closeModal(event)">
@@ -318,6 +338,12 @@
         es.close();
         parseBtn.disabled = true;
         parseBtn.innerHTML = '<i data-lucide="check" style="width:16px;height:16px;"></i> ✓ Готово';
+
+        // Показать кнопку «Сравнить»
+        var llmBtn = document.getElementById('llmBtn');
+        llmBtn.style.display = 'block';
+        llmBtn.disabled = false;
+        llmBtn.innerHTML = '<i data-lucide="scale" style="width:16px;height:16px;"></i> Сравнить';
         lucide.createIcons();
 
         var summaryRow = document.createElement('tr');
@@ -410,6 +436,134 @@
   window.showInfo = showInfo;
   window.closeModal = closeModal;
 
+  // ── LLM: Сравнить ────────────────────────────────────────────
+
+  document.getElementById('llmBtn').addEventListener('click', function() {
+    if (!contractId) return;
+
+    var llmBtn = document.getElementById('llmBtn');
+    llmBtn.disabled = true;
+    llmBtn.innerHTML = '<span style="display:inline-block;width:16px;height:16px;border:2px solid rgba(255,255,255,.3);border-top-color:#fff;border-radius:50%;animation:spin .6s linear infinite;"></span> Анализ...';
+
+    var diffCard = document.getElementById('diffCard');
+    var diffBody = document.getElementById('diffBody');
+    var diffStatus = document.getElementById('diffStatus');
+    diffCard.style.display = 'block';
+    diffBody.innerHTML = '<span style="color:var(--muted);">⏳ LLM-извлечение...</span>';
+
+    var es = new EventSource('/llm/process/' + contractId);
+    var filesDone = 0;
+
+    es.onmessage = function(e) {
+      var d = JSON.parse(e.data);
+
+      if (d.type === 'extract_start') {
+        diffBody.innerHTML += '<div style="font-size:12px;color:var(--muted);">🔍 ' + d.name + ' — извлечение...</div>';
+      }
+
+      else if (d.type === 'extract_done') {
+        filesDone++;
+        diffStatus.textContent = '✓ ' + filesDone + ' файлов, ' + d.count + ' строк';
+        diffBody.innerHTML += '<div style="font-size:12px;color:var(--green);">✓ ' + d.name + ' — ' + d.count + ' строк (' + d.time_s + 'с)</div>';
+      }
+
+      else if (d.type === 'extract_skip') {
+        filesDone++;
+        diffBody.innerHTML += '<div style="font-size:12px;color:var(--muted);">⏭ ' + d.name + ' — уже извлечено (' + d.count + ' строк)</div>';
+      }
+
+      else if (d.type === 'extract_error') {
+        diffBody.innerHTML += '<div style="font-size:12px;color:var(--destructive);">✗ ' + d.name + ': ' + d.error + '</div>';
+      }
+
+      else if (d.type === 'diff_start') {
+        diffBody.innerHTML += '<div style="margin-top:8px;font-weight:600;">📊 Сравнение допников:</div>';
+      }
+
+      else if (d.type === 'diff_file') {
+        var added = d.summary ? (d.summary.added || 0) : 0;
+        var changed = d.summary ? (d.summary.changed || 0) : 0;
+        var deleted = d.summary ? (d.summary.deleted || 0) : 0;
+        diffBody.innerHTML += '<div style="font-size:12px;">  ' + d.name + ': <span style="color:var(--green);">+' + added + '</span> <span style="color:#eab308;">~' + changed + '</span> <span style="color:var(--destructive);">-' + deleted + '</span></div>';
+      }
+
+      else if (d.type === 'done') {
+        es.close();
+        llmBtn.innerHTML = '<i data-lucide="check" style="width:16px;height:16px;"></i> ✓ Готово';
+        lucide.createIcons();
+        diffStatus.textContent = '✓ ' + d.total_time_s + 'с';
+
+        // Построить таблицу изменений
+        var all = d.all_changes || [];
+        if (all.length === 0) {
+          diffBody.innerHTML += '<div style="color:var(--muted);margin-top:8px;">Нет допников для сравнения.</div>';
+          return;
+        }
+
+        all.forEach(function(ch) {
+          diffBody.innerHTML += '<div style="font-weight:600;margin-top:12px;">📄 ' + ch.filename + ' (' + ch.type + ')</div>';
+
+          if (!ch.changes || ch.changes.length === 0) {
+            diffBody.innerHTML += '<div style="color:var(--muted);font-size:12px;">Нет изменений</div>';
+            return;
+          }
+
+          var tbl = '<div class="table-wrap" style="margin-top:4px;"><table><thead><tr>' +
+            '<th>№</th><th>Услуга</th><th>Цена</th><th>Объём</th><th>Сумма</th><th>Начало</th></tr></thead><tbody>';
+
+          ch.changes.forEach(function(c) {
+            var cls = '';
+            if (c.change_type === 'added') cls = 'diff-added';
+            else if (c.change_type === 'deleted') cls = 'diff-deleted';
+            else if (c.change_type === 'changed') cls = 'diff-changed';
+
+            var vals = c.new_values || c.old_values || {};
+
+            var price = _diffCell(c, 'price');
+            var qty = _diffCell(c, 'qty');
+            var sum = _diffCell(c, 'sum');
+
+            tbl += '<tr class="' + cls + '">' +
+              '<td>' + (c.row_num || '') + '</td>' +
+              '<td>' + (vals.name || '') + '</td>' +
+              '<td class="num-cell">' + price + '</td>' +
+              '<td class="num-cell">' + qty + '</td>' +
+              '<td class="num-cell">' + sum + '</td>' +
+              '<td>' + (vals.date_start || '') + '</td>' +
+              '</tr>';
+          });
+
+          tbl += '</tbody></table></div>';
+          diffBody.innerHTML += tbl;
+        });
+      }
+
+      else if (d.type === 'error') {
+        diffBody.innerHTML += '<div style="color:var(--destructive);margin-top:8px;">✗ ' + d.message + '</div>';
+      }
+    };
+
+    es.addEventListener('error', function() {
+      es.close();
+      llmBtn.innerHTML = '<i data-lucide="scale" style="width:16px;height:16px;"></i> Сравнить';
+      llmBtn.disabled = false;
+      lucide.createIcons();
+    });
+  });
+
+  function _diffCell(c, field) {
+    if (c.change_type === 'changed' && c.old_values && c.new_values) {
+      var ov = c.old_values[field];
+      var nv = c.new_values[field];
+      if (ov !== nv && ov !== undefined) {
+        return '<span class="diff-field-old">' + (ov !== null ? ov : '—') + '</span> → <span class="diff-field-new">' + (nv !== null ? nv : '—') + '</span>';
+      }
+    }
+    var vals = c.new_values || c.old_values || {};
+    var v = vals[field];
+    return v !== null && v !== undefined ? v : '—';
+  }
+
   var style = document.createElement('style');
   style.textContent = '@keyframes spin { to { transform: rotate(360deg); } }';
   document.head.appendChild(style);