fix: skip .doc binary files unchanged + bump v0.4/v1.0.187

2026-06-29 16:07:56 +04:00
parent 2f81fb3645
commit 58bc73385d
4 changed files with 26 additions and 10 deletions
@@ -208,7 +208,10 @@ class TwoPassObfuscator:
                all_texts[fname] = text
                if doc is not None:
                    all_docx[fname] = doc
-                self._scan_regex(text)
+                elif text == "[DOC binary — not parsed]":
+                    pass  # .doc без изменений
+                else:
+                    self._scan_regex(text)

            if self._llm_client:
                self._scan_llm_ner(all_texts)
@@ -217,7 +220,10 @@ class TwoPassObfuscator:
            results = []
            for fname, content, ctype in files:
                obf_content = content
-                if fname in all_docx:
+                if fname.endswith('.doc'):
+                    # .doc — бинарный, оставляем как есть
+                    pass
+                elif fname in all_docx:
                    obf_content = self._replace_in_docx(all_docx[fname])
                elif fname.endswith('.pdf'):
                    txt = all_texts.get(fname, '')
@@ -310,8 +316,10 @@ class TwoPassObfuscator:
                            text += "\n" + " | ".join(str(c) if c else "" for c in row)

        elif ext == '.doc':
-            # .doc: читаем как текст (конвертация через libreoffice не используется)
-            text = content.decode('utf-8', errors='replace')
+            # .doc — бинарный формат, без libreoffice не парсим
+            # Пропускаем без изменений (не обфусцируем)
+            text = "[DOC binary — not parsed]"
+            return text, None

        else:
            text = content.decode('utf-8', errors='replace')
@@ -64,7 +64,7 @@
  <a href="/">Сверка договоров</a>
  <span class="sep">|</span>
  <strong>DrHider</strong>
-  <span style="font-size:11px;color:var(--muted);">v0.3</span>
+  <span style="font-size:11px;color:var(--muted);">v0.4</span>
 </header>

 <main>
@@ -73,7 +73,7 @@
 <body>
 <div class="topbar">
  <img src="/static/logo.svg" alt="Nubes">
-  <span class="title">Сверка договоров — LLM AI-driven Event Sourcing <span style="font-weight:400;color:var(--muted);font-size:12px;">v1.0.186-flask</span></span>
+  <span class="title">Сверка договоров — LLM AI-driven Event Sourcing <span style="font-weight:400;color:var(--muted);font-size:12px;">v1.0.187-flask</span></span>
  <div id="pipelineStepper" style="display:flex;gap:8px;font-size:11px;align-items:center;color:var(--muted);">
    <span id="stepUpload">○ Загрузка</span><span>→</span>
    <span id="stepClassify">○ Классификация</span><span>→</span>