Add precheck streams capability

When requested, QPDFWriter will do more aggress prechecking of streams
to make sure it can actually succeed in decoding them before
attempting to do so. This will allow preservation of raw data even
when the raw data is corrupted relative to the specified filters.

5 files changed, 57 insertions, 18 deletions

diff --git a/libqpdf/QPDF.cc b/libqpdf/QPDF.cc
index 32c8cdf9..b5c1212c 100644
--- a/libqpdf/QPDF.cc
+++ b/libqpdf/QPDF.cc
@@ -2134,12 +2134,14 @@ QPDF::getCompressibleObjGens()
     return result;
 }
 
-void
+bool
 QPDF::pipeStreamData(int objid, int generation,
 		     qpdf_offset_t offset, size_t length,
 		     QPDFObjectHandle stream_dict,
-		     Pipeline* pipeline)
+		     Pipeline* pipeline,
+                     bool suppress_warnings)
 {
+    bool success = false;
     std::vector<PointerHolder<Pipeline> > to_delete;
     if (this->encrypted)
     {
@@ -2165,21 +2167,29 @@ QPDF::pipeStreamData(int objid, int generation,
 	    length -= len;
 	    pipeline->write(QUtil::unsigned_char_pointer(buf), len);
 	}
+        success = true;
     }
     catch (QPDFExc& e)
     {
-	warn(e);
+        if (! suppress_warnings)
+        {
+            warn(e);
+        }
     }
     catch (std::runtime_error& e)
     {
-	QTC::TC("qpdf", "QPDF decoding error warning");
-	warn(QPDFExc(qpdf_e_damaged_pdf, this->file->getName(),
-		     "", this->file->getLastOffset(),
-		     "error decoding stream data for object " +
-		     QUtil::int_to_string(objid) + " " +
-		     QUtil::int_to_string(generation) + ": " + e.what()));
+        if (! suppress_warnings)
+        {
+            QTC::TC("qpdf", "QPDF decoding error warning");
+            warn(QPDFExc(qpdf_e_damaged_pdf, this->file->getName(),
+                         "", this->file->getLastOffset(),
+                         "error decoding stream data for object " +
+                         QUtil::int_to_string(objid) + " " +
+                         QUtil::int_to_string(generation) + ": " + e.what()));
+        }
     }
     pipeline->finish();
+    return success;
 }
 
 void
diff --git a/libqpdf/QPDFObjectHandle.cc b/libqpdf/QPDFObjectHandle.cc
index 7618cdf3..bac233df 100644
--- a/libqpdf/QPDFObjectHandle.cc
+++ b/libqpdf/QPDFObjectHandle.cc
@@ -496,11 +496,12 @@ QPDFObjectHandle::getRawStreamData()
 
 bool
 QPDFObjectHandle::pipeStreamData(Pipeline* p, bool filter,
-				 bool normalize, bool compress)
+				 bool normalize, bool compress,
+                                 bool suppress_warnings)
 {
     assertStream();
     return dynamic_cast<QPDF_Stream*>(obj.getPointer())->pipeStreamData(
-	p, filter, normalize, compress);
+	p, filter, normalize, compress, suppress_warnings);
 }
 
 void
diff --git a/libqpdf/QPDFWriter.cc b/libqpdf/QPDFWriter.cc
index 01748fc7..59e306fc 100644
--- a/libqpdf/QPDFWriter.cc
+++ b/libqpdf/QPDFWriter.cc
@@ -57,6 +57,7 @@ QPDFWriter::init()
     stream_data_mode_set = false;
     stream_data_mode = qpdf_s_compress;
     qdf_mode = false;
+    precheck_streams = false;
     static_id = false;
     suppress_original_object_ids = false;
     direct_stream_lengths = true;
@@ -177,6 +178,12 @@ QPDFWriter::setQDFMode(bool val)
 }
 
 void
+QPDFWriter::setPrecheckStreams(bool val)
+{
+    this->precheck_streams = val;
+}
+
+void
 QPDFWriter::setMinimumPDFVersion(std::string const& version)
 {
     setMinimumPDFVersion(version, 0);
@@ -1522,6 +1529,21 @@ QPDFWriter::unparseObject(QPDFObjectHandle object, int level,
 
 	flags |= f_stream;
 
+        if (filter && this->precheck_streams)
+        {
+            try
+            {
+                QTC::TC("qpdf", "QPDFWriter precheck stream");
+                Pl_Discard discard;
+                filter = object.pipeStreamData(
+                    &discard, true, false, false, true);
+            }
+            catch (std::exception)
+            {
+                filter = false;
+            }
+        }
+
 	pushPipeline(new Pl_Buffer("stream data"));
 	activatePipelineStack();
 	bool filtered =
diff --git a/libqpdf/QPDF_Stream.cc b/libqpdf/QPDF_Stream.cc
index b4d14441..31d583b8 100644
--- a/libqpdf/QPDF_Stream.cc
+++ b/libqpdf/QPDF_Stream.cc
@@ -85,7 +85,7 @@ PointerHolder<Buffer>
 QPDF_Stream::getStreamData()
 {
     Pl_Buffer buf("stream data buffer");
-    if (! pipeStreamData(&buf, true, false, false))
+    if (! pipeStreamData(&buf, true, false, false, false))
     {
 	throw std::logic_error("getStreamData called on unfilterable stream");
     }
@@ -97,7 +97,7 @@ PointerHolder<Buffer>
 QPDF_Stream::getRawStreamData()
 {
     Pl_Buffer buf("stream data buffer");
-    pipeStreamData(&buf, false, false, false);
+    pipeStreamData(&buf, false, false, false, false);
     QTC::TC("qpdf", "QPDF_Stream getRawStreamData");
     return buf.getBuffer();
 }
@@ -351,7 +351,8 @@ QPDF_Stream::filterable(std::vector<std::string>& filters,
 
 bool
 QPDF_Stream::pipeStreamData(Pipeline* pipeline, bool filter,
-			    bool normalize, bool compress)
+			    bool normalize, bool compress,
+                            bool suppress_warnings)
 {
     std::vector<std::string> filters;
     int predictor = 1;
@@ -487,9 +488,13 @@ QPDF_Stream::pipeStreamData(Pipeline* pipeline, bool filter,
     else
     {
 	QTC::TC("qpdf", "QPDF_Stream pipe original stream data");
-	QPDF::Pipe::pipeStreamData(this->qpdf, this->objid, this->generation,
-				   this->offset, this->length,
-				   this->stream_dict, pipeline);
+	if (! QPDF::Pipe::pipeStreamData(this->qpdf, this->objid, this->generation,
+                                         this->offset, this->length,
+                                         this->stream_dict, pipeline,
+                                         suppress_warnings))
+        {
+            filter = false;
+        }
     }
 
     return filter;
diff --git a/libqpdf/qpdf/QPDF_Stream.hh b/libqpdf/qpdf/QPDF_Stream.hh
index fa405d70..d053fd0f 100644
--- a/libqpdf/qpdf/QPDF_Stream.hh
+++ b/libqpdf/qpdf/QPDF_Stream.hh
@@ -23,7 +23,8 @@ class QPDF_Stream: public QPDFObject
 
     // See comments in QPDFObjectHandle.hh for these methods.
     bool pipeStreamData(Pipeline*, bool filter,
-			bool normalize, bool compress);
+			bool normalize, bool compress,
+                        bool suppress_warnings);
     PointerHolder<Buffer> getStreamData();
     PointerHolder<Buffer> getRawStreamData();
     void replaceStreamData(PointerHolder<Buffer> data,