[glom/import_csv_refactored] Import: Fixed tests to fit new parser API

[Michael Hasselmann <mikhas src gnome org>]

commit b3eca39111a4960eb14f42f302d459284b439d6b
Author: Michael Hasselmann <michaelh openismus com>
Date:   Fri Sep 25 13:18:19 2009 +0200

    Import: Fixed tests to fit new parser API
    
    * glom/import_csv/csv_parser.[h|cc]: Added a finished_parsing signal since the
    max_row count logic is flawed
    (https://bugzilla.gnome.org/show_bug.cgi?id=588233#c16). Also moved common
    parts of on_stream_read(.) (now on_buffer_read(.)) into
    copy_buffer_and_continue_reading(.), to avoid copy'n'paste bugs.
    
    * tests/import/utils.[h|cc]: Added mainloop functionality to trigger the idle
    parse sequence of the CsvParser for the testcases.
    
    * tests/import/test_[parsing,signals].cc: Fixed the tests to fit new parser
    API.

 glom/import_csv/csv_parser.cc |  169 ++++++++++++++++-------------------------
 glom/import_csv/csv_parser.h  |   25 ++++---
 tests/import/test_parsing.cc  |  147 +++++++++++++++++------------------
 tests/import/test_signals.cc  |   42 +++++------
 tests/import/utils.cc         |   65 +++++++++++++++-
 tests/import/utils.h          |   11 +++
 6 files changed, 244 insertions(+), 215 deletions(-)
---
diff --git a/glom/import_csv/csv_parser.cc b/glom/import_csv/csv_parser.cc
index 19889f9..ccb1f5d 100644
--- a/glom/import_csv/csv_parser.cc
+++ b/glom/import_csv/csv_parser.cc
@@ -140,6 +140,11 @@ CsvParser::type_signal_encoding_error CsvParser::signal_encoding_error() const
   return m_signal_encoding_error;
 }
 
+CsvParser::type_signal_finished_parsing CsvParser::signal_finished_parsing() const
+{
+  return m_finished_parsing;
+}
+
 CsvParser::type_signal_line_scanned CsvParser::signal_line_scanned() const
 {
   return m_signal_line_scanned;
@@ -236,8 +241,8 @@ void CsvParser::clear()
   // Set to current encoding I guess ...
   //m_conv("UTF-8", encoding),
   m_input_position= 0;
-  // Disconnect signal handlers, too? Nah, I don't think so ...
-  //m_idle_connection.disconnect();
+  // Disconnect signal handlers, too.
+  m_idle_connection.disconnect();
   m_line_number = 0;
   set_state(STATE_NONE);
 }
@@ -388,7 +393,9 @@ bool CsvParser::on_idle_parse()
     }
 
     // We have parsed the whole file. We have finished.
+    // TODO: To only emit signal_finished_parsing here is *not* enough.
     set_state(STATE_PARSED);
+    signal_finished_parsing().emit();
   }
 
   // Continue if there are more bytes to process
@@ -424,132 +431,98 @@ void CsvParser::do_line_scanned(const Glib::ustring& line, guint line_number)
     row.push_back(field);
   }
 
-  signal_line_scanned().emit(line, line_number);
+  signal_line_scanned().emit(row, line_number);
 }
 
 void CsvParser::on_file_read(const Glib::RefPtr<Gio::AsyncResult>& result)
 {
+  // TODO: Introduce CsvParser::is_idle_handler_connected() instead?
+  if(!m_idle_connection.connected())
+  {
+    m_idle_connection = Glib::signal_idle().connect(sigc::mem_fun(*this, &CsvParser::on_idle_parse));
+  }
+
 #ifdef GLIBMM_EXCEPTIONS_ENABLED
   try
   {
     m_stream = m_file->read_finish(result);
 
     m_buffer.reset(new Buffer);
-    m_stream->read_async(m_buffer->buf, sizeof(m_buffer->buf), sigc::mem_fun(*this, &CsvParser::on_stream_read));
+    m_stream->read_async(m_buffer->buf, sizeof(m_buffer->buf), sigc::mem_fun(*this, &CsvParser::on_buffer_read));
   }
-  catch(const Glib::Exception& error)
+  catch(const Glib::Exception& ex)
   {
-    signal_file_read_error().emit( error.what() );
+    signal_file_read_error().emit(ex.what());
     clear();
-    // TODO: Response?
   }
 #else
-    std::auto_ptr<Glib::Error> error;
-    m_stream = m_file->read_finish(result, error);
-    if (!error.get())
-    {
-      m_buffer.reset(new Buffer);
-      m_stream->read_async(m_buffer->buf, sizeof(m_buffer->buf), sigc::mem_fun(*this, &CsvParser::on_stream_read));
-    }
-    else
-    {
-      signal_file_read_error().emit( error->what() );
-      clear();
-    }
-#endif    
+  std::auto_ptr<Glib::Error> error;
+  m_stream = m_file->read_finish(result, error);
+  if (!error.get())
+  {
+    m_buffer.reset(new Buffer);
+    m_stream->read_async(m_buffer->buf, sizeof(m_buffer->buf), sigc::mem_fun(*this, &CsvParser::on_buffer_read));
+  }
+  else
+  {
+    signal_file_read_error().emit(error->what());
+    clear();
+  }
+#endif
 }
 
+void CsvParser::copy_buffer_and_continue_reading(gssize size)
+{
+  if(size > 0)
+  {
+    m_raw.insert(m_raw.end(), m_buffer->buf, m_buffer->buf + size);
 
-void CsvParser::on_stream_read(const Glib::RefPtr<Gio::AsyncResult>& result)
+    m_buffer.reset(new Buffer);
+    m_stream->read_async(m_buffer->buf, sizeof(m_buffer->buf), sigc::mem_fun(*this, &CsvParser::on_buffer_read));
+  }
+  else // When size == 0 we finished reading.
+  {
+    //TODO: put in proper data reset method?
+    m_buffer.reset(0);
+    m_stream.reset();
+    m_file.reset();
+  }
+}
+
+void CsvParser::on_buffer_read(const Glib::RefPtr<Gio::AsyncResult>& result)
 {
 #ifdef GLIBMM_EXCEPTIONS_ENABLED
   try
   {
     const gssize size = m_stream->read_finish(result);
-    m_raw.insert(m_raw.end(), m_buffer->buf, m_buffer->buf + size);
-
-    // If the parser already exists, but it is currently not parsing because it waits
-    // for new input, then continue parsing.
-    // TODO: Introduce CsvParser::is_idle_handler_connected() instead?
-    if(!m_idle_connection.connected())
-    {
-      m_idle_connection = Glib::signal_idle().connect(sigc::mem_fun(*this, &CsvParser::on_idle_parse));
-    }
-    // If the parser does not exist yet, then create a new parser, except when the
-    // current encoding does not work for the file, in which case the user must first
-    // choose another encoding.
-    else if(m_state != CsvParser::STATE_ENCODING_ERROR)
-    {
-      begin_parse();
-    }
-
-    if(size > 0)
-    {
-      // Read the next few bytes
-      m_stream->read_async(m_buffer->buf, sizeof(m_buffer->buf), sigc::mem_fun(*this, &CsvParser::on_stream_read));
-    }
-    else
-    {
-      // Finished reading
-      m_buffer.reset(0);
-      m_stream.reset();
-      m_file.reset();
-    }
+    copy_buffer_and_continue_reading(size);
   }
-  catch(const Glib::Exception& error)
+  catch(const Glib::Exception& ex)
   {
-    signal_file_read_error().emit( error.what() );
+     signal_file_read_error().emit(ex.what());
     clear();
-    // TODO: Response?
   }
 #else
-    std::auto_ptr<Glib::Error> error;
-    const gssize size = m_stream->read_finish(result, error);
-    if (!error.get())
-    {
-      m_raw.insert(m_raw.end(), m_buffer->buf, m_buffer->buf + size);
-
-      // If the parser already exists, but it is currently not parsing because it waits
-      // for new input, then continue parsing.
-      if(!m_idle_connection.connected())
-      {
-        m_idle_connection = Glib::signal_idle().connect(sigc::mem_fun(*m_parser.get(), &CsvParser::on_idle_parse));
-      }
-      // If the parser does not exist yet, then create a new parser, except when the
-      // current encoding does not work for the file ,in which case the user must first
-      // choose another encoding.
-      else if(m_state != CsvParser::ENCODING_ERROR)
-      {
-        begin_parse();
-      }
-
-      if(size > 0)
-      {
-        // Read the next few bytes
-        m_stream->read_async(m_buffer->buf, sizeof(m_buffer->buf), sigc::mem_fun(*this, &CsvParser::on_stream_read));
-      }
-      else
-      {
-        // Finished reading
-        m_buffer.reset(0);
-        m_stream.reset();
-        m_file.reset();
-      }
-    }
-    if (error.get())
-    {
-      signal_file_read_error().emit( error->what() );
-      clear();
-    }
+  std::auto_ptr<Glib::Error> error;
+  const gssize size = m_stream->read_finish(result, error);
+  if (!error.get())
+  {
+    copy_buffer_and_continue_reading(size)
+  }
+  else
+  {
+    signal_file_read_error().emit(error->what());
+    clear();
+  }
 #endif
 }
 
-
 void CsvParser::on_file_query_info(const Glib::RefPtr<Gio::AsyncResult>& result)
 {
 #ifdef GLIBMM_EXCEPTIONS_ENABLED
   try
   {
+    // Why is m_file null? Did we clear the parser before reading the file info?
     Glib::RefPtr<Gio::FileInfo> info = m_file->query_info_finish(result);
     if(info)
       signal_have_display_name().emit(info->get_display_name());
@@ -568,17 +541,7 @@ void CsvParser::on_file_query_info(const Glib::RefPtr<Gio::AsyncResult>& result)
   }
   else
     std::cerr << "Failed to fetch display name of uri " << m_file->get_uri() << ": " << error->what() << std::endl;
-#endif    
-}
-
-
-//TODO This seems to be superfluous - we already connect the idle handler elsewhere.
-void CsvParser::begin_parse()
-{
-  clear();
-
-  set_state(STATE_PARSING);
-  m_idle_connection = Glib::signal_idle().connect(sigc::mem_fun(*this, &CsvParser::on_idle_parse));
+#endif
 }
 
 void CsvParser::set_state(State state)
diff --git a/glom/import_csv/csv_parser.h b/glom/import_csv/csv_parser.h
index 41f77e3..93554aa 100644
--- a/glom/import_csv/csv_parser.h
+++ b/glom/import_csv/csv_parser.h
@@ -48,6 +48,9 @@ class CsvParser
 {
 public:
 
+  typedef std::vector<Glib::ustring> type_row_strings;
+  typedef std::vector<type_row_strings> type_rows;
+
   //TODO: Avoid having to specify an initial encoding.
   explicit CsvParser(const std::string& encoding_charset);
 
@@ -99,13 +102,20 @@ public:
   type_signal_encoding_error signal_encoding_error() const;
 
 
-  typedef sigc::signal<void, std::string, unsigned int> type_signal_line_scanned;
+  typedef sigc::signal<void, type_row_strings, unsigned int> type_signal_line_scanned;
 
   /** This signal will be emitted each time the parser has scanned a line. TODO: Do we mean row instead of line? - A row contain a newline.
    */
   type_signal_line_scanned signal_line_scanned() const;
 
 
+  typedef sigc::signal<void> type_signal_finished_parsing;
+
+  /** This signal will be emitted when the parser successfully finished to parse a file.
+   */
+  type_signal_finished_parsing signal_finished_parsing() const;
+
+
   typedef sigc::signal<void> type_signal_state_changed;
 
   /** This signal will be emitted when the state changes.
@@ -126,17 +136,10 @@ public:
   void set_file_and_start_parsing(const std::string& uri);
 
 private:
-
-  typedef std::vector<Glib::ustring> type_row_strings;
-  typedef std::vector<type_row_strings> type_rows;
-
-
   // In order to not make the UI feel sluggish during larger imports we parse
   // on chunk at a time in the idle handler.
-public:  // public because it is needed for testing => no main loop
   bool on_idle_parse();
 
-private:
   void begin_parse();
 
   static const gunichar DELIMITER = ',';
@@ -147,12 +150,11 @@ private:
   void do_line_scanned(const Glib::ustring& current_line, guint line_number);
 
   //TODO: Document this:
-public: // public because it is needed for testing => no main loop
   static Glib::ustring::const_iterator advance_field(const Glib::ustring::const_iterator& iter, const Glib::ustring::const_iterator& end, Glib::ustring& field);
 
-private:
   void on_file_read(const Glib::RefPtr<Gio::AsyncResult>& result);
-  void on_stream_read(const Glib::RefPtr<Gio::AsyncResult>& result);
+  void copy_buffer_and_continue_reading(gssize size);
+  void on_buffer_read(const Glib::RefPtr<Gio::AsyncResult>& result);
   void on_file_query_info(const Glib::RefPtr<Gio::AsyncResult>& result);
 
   void set_state(State state);
@@ -180,6 +182,7 @@ private:
   type_signal_have_display_name m_signal_have_display_name;
   type_signal_encoding_error m_signal_encoding_error;
   type_signal_line_scanned m_signal_line_scanned;
+  type_signal_finished_parsing m_finished_parsing;
   type_signal_state_changed m_signal_state_changed;
 
   Glib::RefPtr<Gio::File> m_file;
diff --git a/tests/import/test_parsing.cc b/tests/import/test_parsing.cc
index 733acbd..b538232 100644
--- a/tests/import/test_parsing.cc
+++ b/tests/import/test_parsing.cc
@@ -8,22 +8,32 @@
 namespace
 {
 
+void print_tokens();
+
 typedef std::vector<std::string> type_tokens;
 
 type_tokens& get_tokens_instance()
 {
-  static type_tokens tokens;
-  return tokens;
+  static type_tokens type_tokens;
+  return type_tokens;
 }
 
 
-void on_line_scanned(const Glib::ustring& line, guint line_number);
+void on_line_scanned(const std::vector<Glib::ustring>& row, guint /*line_number*/)
+{
+  for(std::vector<Glib::ustring>::const_iterator iter = row.begin();
+      iter != row.end();
+      ++iter)
+  {
+    get_tokens_instance().push_back(*iter);
+  }
+}
 
 void print_tokens()
 {
   for(type_tokens::const_iterator iter = get_tokens_instance().begin();
-       iter != get_tokens_instance().end();
-       ++iter)
+      iter != get_tokens_instance().end();
+      ++iter)
   {
     std::cout << " [" << *iter << "] ";
   }
@@ -34,7 +44,7 @@ void print_tokens()
 bool check_tokens(const std::string& regex)
 {
   Glib::RefPtr<Glib::Regex> check;
-  
+
   #ifdef GLIBMM_EXCEPTIONS_ENABLED
   try
   {
@@ -44,7 +54,7 @@ bool check_tokens(const std::string& regex)
   {
     std::cerr << "Glib::Regex::create() failed: " << ex.what() << std::endl;
     return false;
-  } 
+  }
   #else
   std::auto_ptr<Glib::Error> ex;
   check = Glib::Regex::create(regex, static_cast<Glib::RegexCompileFlags>(0), static_cast<Glib::RegexMatchFlags>(0), ex);
@@ -54,10 +64,10 @@ bool check_tokens(const std::string& regex)
     return false;
   }
   #endif
- 
-  if(!check)
+
+  if(!check && 0 == get_tokens_instance().size())
     return false;
-     
+
   for(type_tokens::const_iterator iter = get_tokens_instance().begin();
        iter != get_tokens_instance().end();
        ++iter)
@@ -69,23 +79,9 @@ bool check_tokens(const std::string& regex)
   return true;
 }
 
-void on_line_scanned(const Glib::ustring& line, guint /*line_number*/)
+void connect_signals(Glom::CsvParser& parser)
 {
-  Glib::ustring field;
-  Glib::ustring::const_iterator line_iter(line.begin());
-
-  while(line_iter != line.end())
-  {
-    line_iter = Glom::CsvParser::advance_field(line_iter, line.end(), field);
-    get_tokens_instance().push_back(field);
-
-    // Manually have to skip separators.
-    if(',' == *line_iter)
-    {
-      get_tokens_instance().push_back(",");
-      ++line_iter;
-    }
-  }
+  parser.signal_line_scanned().connect(sigc::ptr_fun(&on_line_scanned));
 }
 
 } // namespace
@@ -93,110 +89,109 @@ void on_line_scanned(const Glib::ustring& line, guint /*line_number*/)
 // Testcases
 int main(int argc, char* argv[])
 {
+  Glib::thread_init();
   Gtk::Main gtk(argc, argv);
 
-  Glom::CsvParser parser("UTF-8");
-  parser.signal_line_scanned().connect(sigc::ptr_fun(&on_line_scanned));
-
   bool result = true;
   std::stringstream report;
 
   // test_dquoted_string
   {
-    const char raw_line[] = "\"a \"\"quoted\"\" token\",\"sans quotes\"\n";
-    ImportTests::set_parser_contents(parser, raw_line, sizeof(raw_line));
+    const char raw[] = "\"a \"\"quoted\"\" token\",\"sans quotes\"\n";
+    bool finished_parsing = ImportTests::run_parser_from_buffer(&connect_signals, raw, sizeof(raw));
 
-    parser.set_file_and_start_parsing("./tests/import/data/dquoted_string.csv");
-
-    while(parser.on_idle_parse())
-    {}
+    bool passed = (finished_parsing &&
+                   check_tokens("^(a \"quoted\" token|sans quotes)$") &&
+                   2 == get_tokens_instance().size());
+    get_tokens_instance().clear();
 
-    bool passed = check_tokens("^(a \"quoted\" token|,|sans quotes)$");
     if(!ImportTests::check("test_dquoted_string", passed, report))
       result = false;
-
-    get_tokens_instance().clear();
-    parser.clear();
   }
 
   // test_skip_on_no_ending_newline
   {
-    const char raw_line[] = "\"this\",\"line\",\"will\",\"be\",\"skipped\"";
-    ImportTests::set_parser_contents(parser, raw_line, sizeof(raw_line));
+    const char raw[] = "\"token in first line\"\n\"2nd token\", \"but\", \"this\",\"line\",\"will\",\"be\",\"skipped\"";
+    bool finished_parsing = ImportTests::run_parser_from_buffer(&connect_signals, raw, sizeof(raw));
 
-    while(parser.on_idle_parse())
-    {}
+    bool passed = (finished_parsing &&
+                   check_tokens("token in first line") &&
+                   1 == get_tokens_instance().size());
+    get_tokens_instance().clear();
 
-    bool passed = (get_tokens_instance().size() == 0);
     if(!ImportTests::check("test_skip_on_no_ending_newline", passed, report))
       result = false;
-
-    get_tokens_instance().clear();
-    parser.clear();
   }
 
   // test_skip_on_no_quotes_around_token
   {
-    const char raw_line[] = "this,line,contains,only,empty,tokens\n";
-    ImportTests::set_parser_contents(parser, raw_line, sizeof(raw_line));
+    const char raw[] = "this,line,contains,only,empty,tokens\n";
+    bool finished_parsing = ImportTests::run_parser_from_buffer(&connect_signals, raw, sizeof(raw));
 
-    while(parser.on_idle_parse())
-    {}
+    bool passed = (finished_parsing &&
+                   check_tokens("^$") &&
+                   6 == get_tokens_instance().size());
+    get_tokens_instance().clear();
 
-    bool passed = check_tokens("^(|,)$");
     if(!ImportTests::check("test_skip_on_no_quotes_around_token", passed, report))
       result = false;
-
-    get_tokens_instance().clear();
-    parser.clear();
   }
 
   // test_skip_spaces_around_separators
   {
-    const char raw_line[] = "\"spaces\" , \"around\", \"separators\"\n";
-    ImportTests::set_parser_contents(parser, raw_line, sizeof(raw_line));
+    const char raw[] = "\"spaces\" , \"around\", \"separators\"\n";
+    bool finished_parsing = ImportTests::run_parser_from_buffer(&connect_signals, raw, sizeof(raw));
 
-    while(parser.on_idle_parse())
-    {}
+    bool passed = (finished_parsing &&
+                   check_tokens("^(spaces|around|separators)$") &&
+                   3 == get_tokens_instance().size());
+    get_tokens_instance().clear();
 
-    bool passed = (get_tokens_instance().size() == 5);
     if(!ImportTests::check("test_skip_spaces_around_separators", passed, report))
       result = false;
 
-    get_tokens_instance().clear();
-    parser.clear();
   }
 
   // test_fail_on_non_comma_separators
   {
-    const char raw_line[] = "\"cannot\"\t\"tokenize\"\t\"this\"\n";
-    ImportTests::set_parser_contents(parser, raw_line, sizeof(raw_line));
+    const char raw[] = "\"cannot\"\t\"tokenize\"\t\"this\"\n";
+    bool finished_parsing = ImportTests::run_parser_from_buffer(&connect_signals, raw, sizeof(raw));
 
-    while(parser.on_idle_parse())
-    {}
+    bool passed = (finished_parsing &&
+                   check_tokens("^cannottokenizethis$") &&
+                   1 == get_tokens_instance().size());
+    get_tokens_instance().clear();
 
-    bool passed = check_tokens("^cannottokenizethis$");
     if(!ImportTests::check("test_fail_on_non_comma_separators", passed, report))
       result = false;
-
-    get_tokens_instance().clear();
-    parser.clear();
   }
 
   // test_parse_newline_inside_quotes
   {
-    const char raw_line[] = "\"cell with\nnewline\"\n\"token on next line\"";
-    ImportTests::set_parser_contents(parser, raw_line, sizeof(raw_line));
+    const char raw[] = "\"cell with\nnewline\"\n\"token on next line\"\n";
+    bool finished_parsing = ImportTests::run_parser_from_buffer(&connect_signals, raw, sizeof(raw));
 
-    while(parser.on_idle_parse())
-    {}
+    bool passed = (finished_parsing &&
+                   check_tokens("^(cell with\nnewline|token on next line)$") &&
+                   2 == get_tokens_instance().size());
+    get_tokens_instance().clear();
 
-    bool passed = check_tokens("^(cell with\nnewline|token on next line)$");
     if(!ImportTests::check("test_parse_newline_inside_quotes", passed, report))
       result = false;
+  }
 
+  // test_fail_on_non_matching_quotes
+  {
+    const char raw[] = "\"token\"\nthis quote has no partner\",\"token\"\n";
+    bool finished_parsing = ImportTests::run_parser_from_buffer(&connect_signals, raw, sizeof(raw));
+
+    bool passed = (finished_parsing &&
+                   check_tokens("token") &&
+                   1 == get_tokens_instance().size());
     get_tokens_instance().clear();
-    parser.clear();
+
+    if(!ImportTests::check("test_fail_on_non_matching_quotes", passed, report))
+      result = false;
   }
 
   if(!result)
diff --git a/tests/import/test_signals.cc b/tests/import/test_signals.cc
index 25ddbe7..b269201 100644
--- a/tests/import/test_signals.cc
+++ b/tests/import/test_signals.cc
@@ -44,17 +44,20 @@ void print_signal_counts()
   std::cout << "encoding errors: " << get_encoding_error_count_instance() << std::endl;
 }
 
+void connect_signals(Glom::CsvParser& parser)
+{
+  parser.signal_line_scanned().connect(sigc::hide(sigc::hide(&on_line_scanned)));
+  parser.signal_encoding_error().connect(sigc::ptr_fun(&on_encoding_error));
+}
+
 } // namespace
 
 // Testcases
 int main(int argc, char* argv[])
 {
+  Glib::thread_init();
   Gtk::Main gtk(argc, argv);
 
-  Glom::CsvParser parser("UTF-8");
-  parser.signal_line_scanned().connect(sigc::hide(sigc::hide(&on_line_scanned)));
-  parser.signal_encoding_error().connect(sigc::ptr_fun(&on_encoding_error));
-
   bool result = true;
   std::stringstream report;
 
@@ -62,41 +65,36 @@ int main(int argc, char* argv[])
   {
     // 2 CSV lines, first one contains newlines inside quotes
     const char raw[] = "\"some\n quoted\r\n newlines\n\", \"token2\"\n\"token3\"\n";
-    ImportTests::set_parser_contents(parser, raw, sizeof(raw));
+    bool finished_parsing = ImportTests::run_parser_from_buffer(&connect_signals, raw, sizeof(raw));
 
-    while(parser.on_idle_parse())
-    {}
-
-    bool passed = (2 == get_line_scanned_count_instance() &&
+    bool passed = (finished_parsing &&
+                   2 == get_line_scanned_count_instance() &&
                    0 == get_encoding_error_count_instance());
 
     if(!ImportTests::check("test_ignore_quoted_newlines", passed, report))
       result = false;
 
     reset_signal_counts();
-    parser.clear();
   }
 
   // test_ignore_empty_lines
   {
     // 5 CSV lines, but only 2 contain data
     const char raw[] = "token1\n\n\n\ntoken2, token3\n";
-    ImportTests::set_parser_contents(parser, raw, sizeof(raw));
-
-    while(parser.on_idle_parse())
-    {}
+    bool finished_parsing = ImportTests::run_parser_from_buffer(&connect_signals, raw, sizeof(raw));
 
-    const bool passed = (2 == get_line_scanned_count_instance() &&
+    bool passed = (finished_parsing &&
+                   2 == get_line_scanned_count_instance() &&
                    0 == get_encoding_error_count_instance());
 
     if(!ImportTests::check("test_ignore_empty_lines", passed, report))
       result = false;
 
     reset_signal_counts();
-    parser.clear();
   }
 
-  // test_wrong_encoding
+// TODO: Cannot currently run this test in a sane fashion, fix me!
+/*
   {
     const char* const encoding_arr[] = {"UTF-8", "UCS-2"};
     type_encodings encodings(encoding_arr, encoding_arr + G_N_ELEMENTS(encoding_arr));
@@ -141,24 +139,22 @@ int main(int argc, char* argv[])
     reset_signal_counts();
     parser.clear();
   }
+*/
 
   // test_incomplete_chars
   {
     // An incomplete Unicode sequence.
     const char raw[] = "\0xc0\n";
-    ImportTests::set_parser_contents(parser, raw, sizeof(raw));
+    bool finished_parsing = ImportTests::run_parser_from_buffer(&connect_signals, raw, sizeof(raw));
 
-    while(parser.on_idle_parse())
-    {}
-
-    const bool passed = (1 == get_encoding_error_count_instance() &&
+    bool passed = (finished_parsing &&
+                   1 == get_encoding_error_count_instance() &&
                    0 == get_line_scanned_count_instance());
 
     if(!ImportTests::check("test_incomplete_chars", passed, report))
       result = false;
 
     reset_signal_counts();
-    parser.clear();
   }
 
   if(!result)
diff --git a/tests/import/utils.cc b/tests/import/utils.cc
index 16de7af..2f3abd4 100644
--- a/tests/import/utils.cc
+++ b/tests/import/utils.cc
@@ -13,8 +13,69 @@ bool check(const std::string& name, bool test, std::stringstream& report)
 
 void set_parser_contents(Glom::CsvParser& /*parser*/, const char* /*input*/, guint /*size*/)
 {
-  // Do not read terminating null byte.
-  //parser.m_raw = std::vector<char>(input, input + size -1);
+}
+
+std::string create_file_from_buffer(const char* input, guint size)
+{
+  // use Glib's file utilities to get an unique tmp filename
+  std::string tmp_filename;
+  int tmp_file_handle = Glib::file_open_tmp(tmp_filename, "glom_testdata");
+
+  if (0 < tmp_file_handle)
+  {
+    ssize_t result = write(tmp_file_handle, input, size);
+    g_assert(-1 != result); // g_return_with_val would still be wrong here, I think?
+
+    close(tmp_file_handle);
+  }
+
+  return Glib::filename_to_uri(tmp_filename);
+}
+
+Glib::RefPtr<Glib::MainLoop>& get_mainloop_instance()
+{
+  static Glib::RefPtr<Glib::MainLoop> mainloop(0);
+  return mainloop;
+}
+
+bool& get_result_instance()
+{
+  static bool result = true;
+  return result;
+}
+
+void on_mainloop_killed_by_watchdog()
+{
+  get_result_instance() = false; // Comment out if you want to run tests and get useful results even with a non-working finished_parsing signal.
+  get_mainloop_instance()->quit();
+}
+
+bool run_parser_from_buffer(void (*connect_parser_signals)(Glom::CsvParser& parser), const char* input, guint size)
+{
+    get_result_instance() = true;
+
+    get_mainloop_instance().reset();
+    get_mainloop_instance() = Glib::MainLoop::create();
+
+    Glom::CsvParser parser("UTF-8");
+
+    parser.signal_finished_parsing().connect(sigc::mem_fun(*get_mainloop_instance().operator->(), &Glib::MainLoop::quit));
+    // Install a watchdog for the mainloop, no test should need longer than 3
+    // seconds. Also, we need to guard against being stuck in the mainloop.
+    // Infinitely running tests are useless.
+    get_mainloop_instance()->get_context()->signal_timeout().connect_seconds_once(sigc::ptr_fun(&on_mainloop_killed_by_watchdog), 3);
+
+    (*connect_parser_signals)(parser);
+
+    const std::string file_name = create_file_from_buffer(input, size);
+    parser.set_file_and_start_parsing(file_name);
+
+    get_mainloop_instance()->run();
+
+    int result = unlink(Glib::filename_from_uri(file_name).c_str());
+    g_assert(-1 != result);
+
+    return get_result_instance();
 }
 
 } //namespace ImportTests
diff --git a/tests/import/utils.h b/tests/import/utils.h
index 778dc1d..76cc14a 100644
--- a/tests/import/utils.h
+++ b/tests/import/utils.h
@@ -3,6 +3,8 @@
 
 #include <glom/import_csv/csv_parser.h>
 #include <iostream>
+#include <unistd.h>
+#include <errno.h>
 
 namespace ImportTests
 {
@@ -12,5 +14,14 @@ bool check(const std::string& name, bool test, std::stringstream& report);
 /// This takes a @a size argument so we can test parsing of null bytes.
 void set_parser_contents(Glom::CsvParser& parser, const char* input, guint size);
 
+// Returns the file name of the temporary created file, which will contain the buffer's contents.
+std::string create_file_from_buffer(const char* input, guint size);
+
+bool run_parser_from_buffer(void (*connect_parser_signals)(Glom::CsvParser& parser), const char* input, guint size);
+
+void on_mainloop_killed_by_watchdog();
+Glib::RefPtr<Glib::MainLoop>& get_mainloop_instance();
+bool& get_result_instance();
+
 } //namespace ImportTests
 #endif