Bug 793467 - GnuCash crashes when trying to open a binary file instead of a CSV

The cause was an uncaught exception from boost::tokenizer.
Fix and add test case.

gnucash/import-export/csv-imp/assistant-csv-price-import.cpp

@@ -717,7 +717,7 @@ CsvImpPriceAssist::file_confirm_cb ()
     catch (std::range_error &e)
     {
         /* Parsing failed ... */
-        gnc_error_dialog (GTK_WINDOW(csv_imp_asst), "%s", e.what());
+        gnc_error_dialog (GTK_WINDOW(csv_imp_asst), "%s", _(e.what()));
         return;
     }
     /* Get settings store and populate */

gnucash/import-export/csv-imp/assistant-csv-trans-import.cpp

@@ -723,7 +723,7 @@ CsvImpTransAssist::file_confirm_cb ()
     catch (std::range_error &e)
     {
         /* Parsing failed ... */
-        gnc_error_dialog (GTK_WINDOW (csv_imp_asst), "%s", e.what());
+        gnc_error_dialog (GTK_WINDOW (csv_imp_asst), "%s", _(e.what()));
         return;
     }
 

gnucash/import-export/csv-imp/gnc-csv-tokenizer.cpp

@@ -11,6 +11,10 @@
 #include <boost/locale.hpp>
 #include <boost/algorithm/string.hpp>
 
+extern "C" {
+    #include <glib/gi18n.h>
+}
+
 void
 GncCsvTokenizer::set_separators(const std::string& separators)
 {
@@ -34,33 +38,40 @@ int GncCsvTokenizer::tokenize()
     m_tokenized_contents.clear();
     std::istringstream in_stream(m_utf8_contents);
 
-    while (std::getline (in_stream, buffer))
+    try
     {
-        // --- deal with line breaks in quoted strings
+        while (std::getline (in_stream, buffer))
-        buffer = boost::trim_copy (buffer); // Removes trailing newline and spaces
-        last_quote = buffer.find_first_of('"');
-        while (last_quote != std::string::npos)
         {
-            if (last_quote == 0) // Test separately because last_quote - 1 would be out of range
+            // --- deal with line breaks in quoted strings
-                inside_quotes = !inside_quotes;
+            buffer = boost::trim_copy (buffer); // Removes trailing newline and spaces
-            else if (buffer[ last_quote - 1 ] != '\\')
+            last_quote = buffer.find_first_of('"');
-                inside_quotes = !inside_quotes;
+            while (last_quote != std::string::npos)
+            {
+                if (last_quote == 0) // Test separately because last_quote - 1 would be out of range
+                    inside_quotes = !inside_quotes;
+                else if (buffer[ last_quote - 1 ] != '\\')
+                    inside_quotes = !inside_quotes;
 
-            last_quote = buffer.find_first_of('"',last_quote+1);
+                last_quote = buffer.find_first_of('"',last_quote+1);
+            }
+
+            line.append(buffer);
+            if (inside_quotes)
+            {
+                line.append(" ");
+                continue;
+            }
+            // ---
+
+            Tokenizer tok(line, sep);
+            vec.assign(tok.begin(),tok.end());
+            m_tokenized_contents.push_back(vec);
+            line.clear();
         }
-
+    }
-        line.append(buffer);
+    catch (boost::escaped_list_error &e)
-        if (inside_quotes)
+    {
-        {
+        throw (std::range_error N_("There was an error parsing the file."));
-            line.append(" ");
-            continue;
-        }
-        // ---
-
-        Tokenizer tok(line, sep);
-        vec.assign(tok.begin(),tok.end());
-        m_tokenized_contents.push_back(vec);
-        line.clear();
     }
 
     return 0;

gnucash/import-export/csv-imp/gnc-tx-import.cpp

@@ -404,7 +404,7 @@ void GncTxImport::tokenize (bool guessColTypes)
     /* If it failed, generate an error. */
     if (m_parsed_lines.size() == 0)
     {
-        throw (std::range_error ("Tokenizing failed."));
+        throw (std::range_error (N_("There was an error parsing the file.")));
         return;
     }
 

gnucash/import-export/csv-imp/test/test-tokenizer.cpp

@@ -138,6 +138,17 @@ TEST_F (GncTokenizerTest, tokenize_from_csv_file)
  * independently.
  */
 
+/* First test whether we're properly catching boost::tokenizer throws
+ * This happens when the input data has invalid escape sequences */
+TEST_F (GncTokenizerTest, tokenize_binary_data)
+{
+    GncCsvTokenizer *csvtok = dynamic_cast<GncCsvTokenizer*>(csv_tok.get());
+    csvtok->set_separators (",");
+
+    set_utf8_contents (csv_tok, R"(\764Test,Something)");
+    EXPECT_THROW (csv_tok->tokenize(), std::range_error);
+}
+
 /* This helper function will run the parse step on the given data
  * with the parser as configured by the calling test function.
  * This allows the same code to be used with different csv test strings