layout as other widgets

orangecontrib/storynavigation/modules/settinganalysis.py

@@ -23,6 +23,7 @@ class SettingAnalyzer:
         language (str): ISO string of the language of the input text
         n_segments (int): Number of segments to split each text into
         text_tuples (list): binary tuple: text (str) and storyid
+        story_elements (list of lists): tokens with their Spacy analysis
         callback: function in widget to show the progress of this process
     """
 
@@ -34,9 +35,10 @@ class SettingAnalyzer:
     ENTITY_CACHE_FILE_NAME = "entity_cache.json"
 
 
-    def __init__(self, language, n_segments, text_tuples, story_elements, callback=None):
+    def __init__(self, language, n_segments, text_tuples, story_elements, user_defined_entities, callback=None):
         self.text_tuples = text_tuples
         self.n_segments = n_segments
+        self.user_defined_entities = user_defined_entities
         self.callback = callback
 
         self.__setup_required_nlp_resources(language)
@@ -91,16 +93,16 @@ def __setup_required_nlp_resources(self, language):
         """
         if language == constants.NL:
             self.model = constants.NL_SPACY_MODEL
-            self.entity_list = constants.NL_ENTITIES_FILE.read_text(encoding="utf-8").strip().split(os.linesep)
+            #self.entity_list = constants.NL_ENTITIES_FILE.read_text(encoding="utf-8").strip().split(os.linesep)
             self.time_words = constants.NL_TIME_WORDS_FILE.read_text(encoding="utf-8").strip().split(os.linesep)
         elif language == constants.EN:
             self.model = constants.EN_SPACY_MODEL
-            self.entity_list = constants.EN_ENTITIES_FILE.read_text(encoding="utf-8").strip().split(os.linesep)
+            #self.entity_list = constants.EN_ENTITIES_FILE.read_text(encoding="utf-8").strip().split(os.linesep)
             self.time_words = constants.EN_TIME_WORDS_FILE.read_text(encoding="utf-8").strip().split(os.linesep)
         else:
             raise ValueError(f"settingsanalysis.py: unknown language {language}")
 
-        self.entity_list = [line.split(",") for line in self.entity_list]
+        # self.entity_list = [line.split(",") for line in self.entity_list]
 
     def __sort_and_filter_results(self, results):
         results = [(x[0], x[1], int(x[2]), x[3], x[4]) for x in results]
@@ -120,11 +122,11 @@ def __process_texts(self, nlp, text_tuples, entities, callback=None):
         return self.__sort_and_filter_results(results)
 
 
-    def __analyze_text_with_list(self, text, nlp, entity_list):
+    def __analyze_text_with_list(self, text, nlp, user_defined_entities):
         matcher = Matcher(nlp.vocab)
         for entity_group in self.ENTITY_GROUPS:
             patterns = [[{"lower": entity_token} for entity_token in entity_text.lower().split()]
-                for entity_label, entity_text in entity_list
+                for entity_text, entity_label in list(user_defined_entities.items())
                     if entity_label in entity_group]
             matcher.add(entity_group[0], patterns)
         tokens = nlp(text)
@@ -170,7 +172,7 @@ def __filter_dates(self, combined_analysis):
 
 
     def __process_text(self, text_id, text, nlp, spacy_analysis):
-        list_analysis = self.__analyze_text_with_list(text, nlp, self.entity_list)
+        list_analysis = self.__analyze_text_with_list(text, nlp, self.user_defined_entities)
         combined_analysis = self.__combine_analyses(spacy_analysis, list_analysis)
         combined_analysis = self.__expand_locations(combined_analysis)
         combined_analysis = self.__filter_dates(combined_analysis)

orangecontrib/storynavigation/widgets/OWSNSettingAnalysis.py

@@ -14,6 +14,12 @@
 from storynavigation.modules.settinganalysis import SettingAnalyzer
 import storynavigation.modules.util as util
 
+from AnyQt.QtWidgets import QAbstractItemView, QHeaderView, QTableView
+from storynavigation.widgets.OWSNActorAnalysis import DocumentTableView, DocumentListModel, DocumentsFilterProxyModel
+from typing import Set
+from orangecontrib.text.widgets.utils import widgets
+from Orange.data.io import FileFormat
+
 
 class OWSNSettingAnalysis(OWWidget, ConcurrentWidgetMixin):
     name = 'Setting Analysis'
@@ -31,6 +37,12 @@ class OWSNSettingAnalysis(OWWidget, ConcurrentWidgetMixin):
                       "GPE": "lemonchiffon",
                       "LOC": "lightgreen",
                       "TIME": "thistle", }
+    dlgFormats = (
+        "All readable files ({});;".format(
+            '*' + ' *'.join(FileFormat.readers.keys())) +
+        ";;".join("{} (*{})".format(f.DESCRIPTION, ' *'.join(f.EXTENSIONS))
+                  for f in sorted(set(FileFormat.readers.values()),
+                                  key=list(FileFormat.readers.values()).index)))
 
     class Inputs:
         story_elements = Input("Story elements", Table)
@@ -50,12 +62,59 @@ def __init__(self):
 
         self.make_language_selection_menu()
 
+        self.recent_files = []
+        self.user_defined_entities = {}
+        # next 9 lines copied from Corpus widget
+        fbox = gui.widgetBox(self.controlArea, "Corpus file", orientation=0)
+        self.file_widget = widgets.FileWidget(
+            recent_files=self.recent_files, icon_size=(16, 16),
+            on_open=self.open_file, dialog_format=self.dlgFormats,
+            dialog_title='Open Orange Document Corpus',
+            reload_label='Reload', browse_label='Browse',
+            allow_empty=False, minimal_width=250,
+        )
+        fbox.layout().addWidget(self.file_widget)
+
+        # next 18 lines copied from CorpusViewer widget
         self.splitter = QSplitter(orientation=Qt.Horizontal, childrenCollapsible=False)
+        self.doc_list = DocumentTableView()
+        self.doc_list.setSelectionBehavior(QTableView.SelectRows)
+        self.doc_list.setSelectionMode(QTableView.ExtendedSelection)
+        self.doc_list.setEditTriggers(QAbstractItemView.NoEditTriggers)
+        self.doc_list.horizontalHeader().setSectionResizeMode(QHeaderView.Stretch)
+        self.doc_list.horizontalHeader().setVisible(False)
+        self.splitter.addWidget(self.doc_list)
+
+        self.doc_list_model = DocumentListModel()
+        proxy_model = DocumentsFilterProxyModel()
+        proxy_model.setSourceModel(self.doc_list_model)
+        self.doc_list.setModel(proxy_model)
+        self.doc_list.selectionModel().selectionChanged.connect(self.selection_changed)
         self.doc_webview = gui.WebviewWidget(self.splitter, debug=False)
         self.doc_webview.setHtml("")
         self.mainArea.layout().addWidget(self.splitter)
 
 
+    def open_file(self, user_defined_entities_file):
+        self.user_defined_entities = {}
+        with open(user_defined_entities_file) as file:
+            for line in file:
+                fields = line.strip().split(",")
+                self.user_defined_entities[fields[1]] = fields[0]
+        if self.story_elements:
+            self.set_story_elements(self.story_elements)
+
+
+    def get_selected_indexes(self) -> Set[int]:
+        m = self.doc_list.model().mapToSource
+        return [int(m(i).row()) for i in self.doc_list.selectionModel().selectedRows()]
+
+
+    def selection_changed(self) -> None:
+        self.stories_selected = self.get_selected_indexes()
+        self.__visualize_text_data()
+
+
     def make_language_selection_menu(self):
         self.select_language_combo = gui.comboBox(
             widget=self.controlArea,
@@ -81,6 +140,15 @@ def set_story_elements(self, story_elements=None):
             self.text_tuples = self.make_text_tuples(story_elements)
             self.__action_analyze_setting_wrapper()
 
+            self.doc_list_model.setup_data(self.make_document_names(self.text_tuples), None)
+
+
+    def make_document_names(self, text_tuples):
+        document_names = []
+        for text, text_id in text_tuples:
+            document_names.append("Document " + str(int(text_id) + 1))
+        return document_names
+
 
     def make_text_tuples(self, story_elements):
         story_elements_df = util.convert_orangetable_to_dataframe(story_elements)
@@ -124,12 +192,13 @@ def move_progress_bar(progress):
              n_segments=int(self.n_segments),
              text_tuples=self.text_tuples,
              story_elements=self.story_elements,
+             user_defined_entities=self.user_defined_entities,
              callback=move_progress_bar
         )
 
 
     def on_done(self, result) -> None:
-        self.__visualize_text_data(self.text_tuples, self.analyzer.settings_analysis)
+        self.__visualize_text_data()
         self.Outputs.dataset_level_data.send(table_from_frame(self.analyzer.settings_analysis))
 
 
@@ -149,9 +218,9 @@ def __insert_entity_color_in_story_text(self, story_text, start, end, label):
         return story_text
 
 
-    def __add_entity_colors_to_story_text(self, story_text, story_id, settings_analysis):
-        for index, row in settings_analysis.loc[
-                             settings_analysis["storyid"] == "ST" + str(story_id)].iloc[::-1].iterrows():
+    def __add_entity_colors_to_story_text(self, story_text, story_id):
+        for index, row in self.analyzer.settings_analysis.loc[
+                             self.analyzer.settings_analysis["storyid"] == "ST" + str(story_id)].iloc[::-1].iterrows():
            start = int(row["character id"])
            end = start + len(row["text"])
            story_text = self.__insert_entity_color_in_story_text(story_text,
@@ -165,12 +234,13 @@ def __add_paragraphs_to_story_text(self, story_text):
         return "<p>" + re.sub("\n\n", "<p>", story_text)
 
 
-    def __visualize_text_data(self, text_tuples, settings_analysis):
+    def __visualize_text_data(self):
         html_text = "<html><body>"
         html_text += self.__make_entity_bar_for_html()
-        for story_text, story_id in text_tuples:
-            story_text = self.__add_entity_colors_to_story_text(story_text, story_id, settings_analysis)
-            html_text += "<hr>" + self.__add_paragraphs_to_story_text(story_text)
+        for story_text, story_id in self.text_tuples:
+            if len(self.stories_selected) == 0 or int(story_id) in self.stories_selected:
+                story_text = self.__add_entity_colors_to_story_text(story_text, story_id)
+                html_text += "<hr>" + self.__add_paragraphs_to_story_text(story_text)
         html_text += "</body></html>"
         self.doc_webview.setHtml(html_text)