File: Allow selecting a file with an arbitrary extension

Author: janezd

Orange/widgets/data/owfile.py

@@ -147,8 +147,11 @@ class Warning(widget.OWWidget.Warning):
     class Error(widget.OWWidget.Error):
         file_not_found = Msg("File not found.")
         missing_reader = Msg("Missing reader.")
+        select_file_type = Msg("Select file type.")
         sheet_error = Msg("Error listing available sheets.")
         unknown = Msg("Read error:\n{}")
+        unknown_select = Msg(
+            "Read error, possibly due to incorrect choice of file type:\n{}")
 
     UserAdviceMessages = [
         widget.Message(
@@ -264,7 +267,7 @@ def package(w):
         self.reader_combo = QComboBox(self)
         self.reader_combo.setSizePolicy(Policy.Expanding, Policy.Fixed)
         self.reader_combo.setMinimumSize(QSize(100, 1))
-        self.reader_combo.activated[int].connect(self.select_reader)
+        self.reader_combo.activated[int].connect(self.on_reader_change)
 
         box.layout().addWidget(self.reader_combo)
         layout.addWidget(box, 0, 1)
@@ -327,6 +330,10 @@ def select_sheet(self):
         self.recent_paths[0].sheet = self.sheet_combo.currentText()
         self.load_data()
 
+    def on_reader_change(self, n):
+        self.select_reader(n)
+        self.load_data()
+
     def select_reader(self, n):
         if self.source != self.LOCAL_FILE:
             return  # ignore for URL's
@@ -335,14 +342,11 @@ def select_reader(self, n):
             path = self.recent_paths[0]
             if n == 0:  # default
                 path.file_format = None
-                self.load_data()
             elif n <= len(self.available_readers):
                 reader = self.available_readers[n - 1]
                 path.file_format = reader.qualified_name()
-                self.load_data()
             else:  # the rest include just qualified names
                 path.file_format = self.reader_combo.itemText(n)
-                self.load_data()
 
     def _url_set(self):
         index = self.url_combo.currentIndex()
@@ -373,7 +377,9 @@ def browse_file(self, in_demos=False):
         else:
             start_file = self.last_path() or os.path.expanduser("~/")
 
-        filename, reader, _ = open_filename_dialog(start_file, None, self.available_readers)
+        filename, reader, _ = open_filename_dialog(
+            start_file, None, self.available_readers,
+            add_all="*")
         if not filename:
             return
         self.add_path(filename)
@@ -415,20 +421,20 @@ def _try_load(self):
             if not url:
                 return self.Information.no_file_selected
 
-        def mark_problematic_reader():
-            self.reader_combo.setItemData(self.reader_combo.currentIndex(),
-                                          QBrush(Qt.red), Qt.ForegroundRole)
-
         try:
             self.reader = self._get_reader()  # also sets current reader index
             assert self.reader is not None
         except MissingReaderException:
-            mark_problematic_reader()
-            return self.Error.missing_reader
+            if self.reader_combo.currentIndex() > 0:
+                return self.Error.missing_reader
+            else:
+                return self.Error.select_file_type
         except Exception as ex:
-            mark_problematic_reader()
             log.exception(ex)
-            return lambda x=ex: self.Error.unknown(str(x))
+            if self.reader_combo.currentIndex() > 0:
+                return lambda x=ex: self.Error.unknown(str(x))
+            else:
+                return lambda x=ex: self.Error.unknown_select(str(x))
 
         try:
             self._update_sheet_combo()
@@ -439,7 +445,6 @@ def mark_problematic_reader():
             try:
                 data = self.reader.read()
             except Exception as ex:
-                mark_problematic_reader()
                 log.exception(ex)
                 return lambda x=ex: self.Error.unknown(str(x))
             if warnings:
@@ -455,9 +460,25 @@ def mark_problematic_reader():
         return None
 
     def _get_reader(self) -> FileFormat:
+        """
+        Get the reader for the current file.
+
+        For local files, this also observes the stored settings and the reader
+        combo, as follows:
+
+        1. If the file format is known (from stored settings), use it and set
+           the reader combo to the corresponding index (as in settings)
+        2. Otherwise, detect it from the extension and set the combo to
+           Auto detect, overriding any previous user-set choice
+        3. Otherwise, use the current combo state.
+
+        Returns:
+            FileFormat: reader instance
+        """
         if self.source == self.LOCAL_FILE:
             path = self.last_path()
             self.reader_combo.setEnabled(True)
+
             if self.recent_paths and self.recent_paths[0].file_format:
                 qname = self.recent_paths[0].file_format
                 qname_index = {r.qualified_name(): i for i, r in enumerate(self.available_readers)}
@@ -471,11 +492,22 @@ def _get_reader(self) -> FileFormat:
                 try:
                     reader_class = class_from_qualified_name(qname)
                 except Exception as ex:
-                    raise MissingReaderException(f'Can not find reader "{qname}"') from ex
+                    raise MissingReaderException(f'Can not fdind reader "{qname}"') from ex
                 reader = reader_class(path)
+
             else:
-                self.reader_combo.setCurrentIndex(0)
-                reader = FileFormat.get_reader(path)
+                old_idx = self.reader_combo.currentIndex()
+                try:
+                    self.reader_combo.setCurrentIndex(0)
+                    reader = FileFormat.get_reader(path)
+                except MissingReaderException:
+                    if old_idx == 0:
+                        raise
+                    # Set the path for the current file format,
+                    # and repeat the call to return the corresponding reader
+                    self.select_reader(old_idx)
+                    return self._get_reader()
+
             if self.recent_paths and self.recent_paths[0].sheet:
                 reader.select_sheet(self.recent_paths[0].sheet)
             return reader
@@ -504,12 +536,21 @@ def _select_active_sheet(self):
             self.sheet_combo.setCurrentIndex(0)
 
     def _initialize_reader_combo(self):
-        self.reader_combo.clear()
-        filters = [format_filter(f) for f in self.available_readers]
-        self.reader_combo.addItems([DEFAULT_READER_TEXT] + filters)
-        self.reader_combo.setCurrentIndex(0)
-        self.reader_combo.setDisabled(True)
-        # additional readers may be added in self._get_reader()
+        # Reset to initial state without losing the current index or
+        # emitting any signals.
+        combo = self.reader_combo
+        if not combo.count():
+            filters = [format_filter(f) for f in self.available_readers]
+            combo.addItems([DEFAULT_READER_TEXT] + filters)
+            combo.setCurrentIndex(0)
+        else:
+            # additional readers may be added in self._get_reader()
+            n = len(self.available_readers) + 1
+            if combo.currentIndex() >= n:
+                combo.setCurrentIndex(0)
+            while combo.count() > n:
+                combo.removeItem(combo.count() - 1)
+        combo.setDisabled(True)
 
     @staticmethod
     def _describe(table):

Orange/widgets/data/tests/test_owfile.py

@@ -361,13 +361,13 @@ def test_reader_custom_tab(self):
         outdata = self.get_output(self.widget.Outputs.data)
         self.assertEqual(len(outdata), 150)  # loaded iris
 
-    def test_no_reader_extension(self):
+    def test_unknown_extension(self):
         with named_file("", suffix=".xyz_unknown") as fn:
             no_reader = RecentPath(fn, None, None)
             self.widget = self.create_widget(OWFile,
                                              stored_settings={"recent_paths": [no_reader]})
             self.widget.load_data()
-        self.assertTrue(self.widget.Error.missing_reader.is_shown())
+        self.assertTrue(self.widget.Error.select_file_type.is_shown())
 
     def test_fail_sheets(self):
         with named_file("", suffix=".failed_sheet") as fn: