From 9ec9cc5f87b65c3ca964da14781ba44a759c7070 Mon Sep 17 00:00:00 2001
From: deeplow <deeplower@protonmail.com>
Date: Mon, 7 Aug 2023 13:04:37 +0100
Subject: [PATCH] Replace armor guards that indicate isolated output

---
 dangerzone/isolation_provider/base.py      | 10 ++++++----
 dangerzone/isolation_provider/container.py | 18 +++++++++++++-----
 tests/test_docs_large                      |  2 +-
 3 files changed, 20 insertions(+), 10 deletions(-)

diff --git a/dangerzone/isolation_provider/base.py b/dangerzone/isolation_provider/base.py
index f054372..00552c9 100644
--- a/dangerzone/isolation_provider/base.py
+++ b/dangerzone/isolation_provider/base.py
@@ -11,8 +11,10 @@ from ..util import replace_control_chars
 log = logging.getLogger(__name__)
 
 MAX_CONVERSION_LOG_CHARS = 150 * 50  # up to ~150 lines of 50 characters
-CONVERSION_LOG_START = "-----CONVERSION LOG START-----"
-CONVERSION_LOG_END = "-----CONVERSION LOG END-----"
+DOC_TO_PIXELS_LOG_START = "----- DOC TO PIXELS LOG START -----"
+DOC_TO_PIXELS_LOG_END = "----- DOC TO PIXELS LOG END -----"
+PIXELS_TO_PDF_LOG_START = "----- PIXELS TO PDF LOG START -----"
+PIXELS_TO_PDF_LOG_END = "----- PIXELS TO PDF LOG END -----"
 
 
 class IsolationProvider(ABC):
@@ -90,8 +92,8 @@ class IsolationProvider(ABC):
         conversion_string = replace_control_chars(untrusted_conversion_str)
 
         # Add armor (gpg-style)
-        armor_start = f"{CONVERSION_LOG_START}\n"
-        armor_end = CONVERSION_LOG_END
+        armor_start = f"{DOC_TO_PIXELS_LOG_START}\n"
+        armor_end = DOC_TO_PIXELS_LOG_END
         return armor_start + conversion_string + armor_end
 
 
diff --git a/dangerzone/isolation_provider/container.py b/dangerzone/isolation_provider/container.py
index 02ff7b7..9d00985 100644
--- a/dangerzone/isolation_provider/container.py
+++ b/dangerzone/isolation_provider/container.py
@@ -18,7 +18,12 @@ from ..util import (
     get_tmp_dir,
     replace_control_chars,
 )
-from .base import MAX_CONVERSION_LOG_CHARS, IsolationProvider
+from .base import (
+    MAX_CONVERSION_LOG_CHARS,
+    PIXELS_TO_PDF_LOG_END,
+    PIXELS_TO_PDF_LOG_START,
+    IsolationProvider,
+)
 
 # Define startupinfo for subprocesses
 if platform.system() == "Windows":
@@ -339,10 +344,13 @@ class Container(IsolationProvider):
 
         if getattr(sys, "dangerzone_dev", False):
             log_path = safe_dir / "captured_output.txt"
-            with open(log_path, "r", encoding="ascii", errors="replace") as f:
-                untrusted_log = f.read(MAX_CONVERSION_LOG_CHARS)
-            text = f"Container output: (pixels to PDF)\n{self.sanitize_conversion_str(untrusted_log)}"
-            log.info(text)
+            if log_path.exists():  # If first stage failed this may not exist
+                with open(log_path, "r", encoding="ascii", errors="replace") as f:
+                    text = (
+                        f"Container output: (pixels to PDF)\n"
+                        f"{PIXELS_TO_PDF_LOG_START}{f.read()}{PIXELS_TO_PDF_LOG_END}"
+                    )
+                    log.info(text)
 
         return success
 
diff --git a/tests/test_docs_large b/tests/test_docs_large
index 51419aa..0068ffc 160000
--- a/tests/test_docs_large
+++ b/tests/test_docs_large
@@ -1 +1 @@
-Subproject commit 51419aa36bd4701927d1610fb544e18a4692f1db
+Subproject commit 0068ffcb67f45fe9e3a082649493b7c8db5d1473