pysfi 0.1.12__py3-none-any.whl → 0.1.13__py3-none-any.whl

sfi/docscan/docscan_gui.py

@@ -129,7 +129,7 @@ class ConfigManager:
         """
         if config_file is None:
             # Use user home directory for config
-            config_dir = Path.home() / ".sfi"
+            config_dir = Path.home() / ".pysfi"
             config_dir.mkdir(exist_ok=True)
             config_file = config_dir / "docscan_gui.json"
         self.config_file = config_file
@@ -286,7 +286,9 @@ class SettingsDialog(QDialog):
         layout = QVBoxLayout()
 
         # Language Settings Group
-        self.language_group = QGroupBox(t("language_settings", default="Language Settings"))
+        self.language_group = QGroupBox(
+            t("language_settings", default="Language Settings")
+        )
         language_layout = QHBoxLayout()
         language_layout.setSpacing(10)
 
@@ -302,13 +304,18 @@ class SettingsDialog(QDialog):
         layout.addWidget(self.language_group)
 
         # Processing Options Group
-        processing_group = QGroupBox(t("processing_options", default="Processing Options"))
+        processing_group = QGroupBox(
+            t("processing_options", default="Processing Options")
+        )
         processing_layout = QVBoxLayout()
         processing_layout.setSpacing(10)
 
         self.ocr_checkbox = QCheckBox(t("use_pdf_ocr"))
         self.ocr_checkbox.setToolTip(
-            t("ocr_tooltip", default="Enable OCR for scanned PDF files to extract text from images")
+            t(
+                "ocr_tooltip",
+                default="Enable OCR for scanned PDF files to extract text from images",
+            )
         )
 
         self.process_pool_checkbox = QCheckBox(t("use_process_pool"))
@@ -325,7 +332,9 @@ class SettingsDialog(QDialog):
         layout.addWidget(processing_group)
 
         # Performance Settings Group
-        performance_group = QGroupBox(t("performance_settings", default="Performance Settings"))
+        performance_group = QGroupBox(
+            t("performance_settings", default="Performance Settings")
+        )
         performance_layout = QFormLayout()
         performance_layout.setSpacing(12)
 
@@ -335,7 +344,10 @@ class SettingsDialog(QDialog):
         self.thread_spin.setMaximum(16)
         self.thread_spin.setValue(4)
         self.thread_spin.setToolTip(
-            t("threads_tooltip", default="Number of worker threads (higher values may improve speed but use more CPU)")
+            t(
+                "threads_tooltip",
+                default="Number of worker threads (higher values may improve speed but use more CPU)",
+            )
         )
         performance_layout.addRow(t("threads"), self.thread_spin)
 
@@ -367,7 +379,9 @@ class SettingsDialog(QDialog):
 
         # OK and Cancel buttons
         button_box = QDialogButtonBox()
-        button_box.setStandardButtons(QDialogButtonBox.StandardButton.Ok | QDialogButtonBox.StandardButton.Cancel)  # type: ignore
+        button_box.setStandardButtons(
+            QDialogButtonBox.StandardButton.Ok | QDialogButtonBox.StandardButton.Cancel
+        )  # type: ignore
         button_box.accepted.connect(self.accept)  # type: ignore
         button_box.rejected.connect(self.reject)  # type: ignore
         button_layout.addWidget(button_box)
@@ -388,7 +402,9 @@ class SettingsDialog(QDialog):
 
         # Update dialog title and UI elements immediately
         self.setWindowTitle(t("scan_options_tab"))
-        self.language_group.setTitle(t("language_settings", default="Language Settings"))
+        self.language_group.setTitle(
+            t("language_settings", default="Language Settings")
+        )
         self.lang_label.setText(t("language_label", default="Language:"))
 
         # Update other group boxes
@@ -397,11 +413,17 @@ class SettingsDialog(QDialog):
             if item.widget() and isinstance(item.widget(), QGroupBox):
                 group_box = item.widget()
                 if "Language" in group_box.title():  # type: ignore
-                    group_box.setTitle(t("language_settings", default="Language Settings"))  # type: ignore
+                    group_box.setTitle(
+                        t("language_settings", default="Language Settings")
+                    )  # type: ignore
                 elif "Processing" in group_box.title():  # type: ignore
-                    group_box.setTitle(t("processing_options", default="Processing Options"))  # type: ignore
+                    group_box.setTitle(
+                        t("processing_options", default="Processing Options")
+                    )  # type: ignore
                 elif "Performance" in group_box.title():  # type: ignore
-                    group_box.setTitle(t("performance_settings", default="Performance Settings"))  # type: ignore
+                    group_box.setTitle(
+                        t("performance_settings", default="Performance Settings")
+                    )  # type: ignore
 
         # Show temporary message in dialog
         logger.info(f"Language preview: {'中文' if lang == 'zh_CN' else 'English'}")
@@ -410,7 +432,9 @@ class SettingsDialog(QDialog):
         """Load current settings into dialog."""
         if self.config_manager:
             self.ocr_checkbox.setChecked(self.config_manager.get("use_pdf_ocr", False))
-            self.process_pool_checkbox.setChecked(self.config_manager.get("use_process_pool", False))
+            self.process_pool_checkbox.setChecked(
+                self.config_manager.get("use_process_pool", False)
+            )
             self.thread_spin.setValue(self.config_manager.get("threads", 4))
             self.batch_spin.setValue(self.config_manager.get("batch_size", 50))
 
@@ -428,7 +452,9 @@ class SettingsDialog(QDialog):
         # Save settings to config
         if self.config_manager:
             self.config_manager.set("use_pdf_ocr", self.ocr_checkbox.isChecked())
-            self.config_manager.set("use_process_pool", self.process_pool_checkbox.isChecked())
+            self.config_manager.set(
+                "use_process_pool", self.process_pool_checkbox.isChecked()
+            )
             self.config_manager.set("threads", self.thread_spin.value())
             self.config_manager.set("batch_size", self.batch_spin.value())
 
@@ -446,7 +472,9 @@ class SettingsDialog(QDialog):
         """Save settings when OK is clicked and apply language immediately."""
         if self.config_manager:
             self.config_manager.set("use_pdf_ocr", self.ocr_checkbox.isChecked())
-            self.config_manager.set("use_process_pool", self.process_pool_checkbox.isChecked())
+            self.config_manager.set(
+                "use_process_pool", self.process_pool_checkbox.isChecked()
+            )
             self.config_manager.set("threads", self.thread_spin.value())
             self.config_manager.set("batch_size", self.batch_spin.value())
 
@@ -483,7 +511,9 @@ class SettingsDialog(QDialog):
         # Update menu bar
         self.main_window.file_menu.setTitle(t("file_menu", default="&File"))
         self.main_window.save_action.setText(t("save_results", default="&Save Results"))
-        self.main_window.clear_action.setText(t("clear_results", default="&Clear Results"))
+        self.main_window.clear_action.setText(
+            t("clear_results", default="&Clear Results")
+        )
 
         # Update input section
         for i in range(self.main_window.centralWidget().layout().count()):
@@ -655,16 +685,25 @@ class DocScanGUI(QMainWindow):
         types_layout = QHBoxLayout()
         types_label = QLabel(t("file_types"))
         self.types_edit = QLineEdit(t("default_file_types"))
-        self.types_edit.setToolTip(t("file_types_tooltip", default="File types to scan (comma separated)"))
+        self.types_edit.setToolTip(
+            t("file_types_tooltip", default="File types to scan (comma separated)")
+        )
         types_layout.addWidget(types_label)
         types_layout.addWidget(self.types_edit)
         input_layout.addLayout(types_layout)
 
         # Checkbox for including image formats
-        self.include_images_checkbox = QCheckBox(t("include_image_formats", default="Include Image Formats"))
-        self.include_images_checkbox.setChecked(self.config_manager.get("include_image_formats", False))
+        self.include_images_checkbox = QCheckBox(
+            t("include_image_formats", default="Include Image Formats")
+        )
+        self.include_images_checkbox.setChecked(
+            self.config_manager.get("include_image_formats", False)
+        )
         self.include_images_checkbox.setToolTip(
-            t("include_image_formats_tooltip", default="Include image formats (jpg, jpeg, png, gif, bmp, tiff) in scan")
+            t(
+                "include_image_formats_tooltip",
+                default="Include image formats (jpg, jpeg, png, gif, bmp, tiff) in scan",
+            )
         )
         self.include_images_checkbox.stateChanged.connect(self._toggle_image_formats)  # type: ignore
         input_layout.addWidget(self.include_images_checkbox)
@@ -680,14 +719,20 @@ class DocScanGUI(QMainWindow):
         if state == 2:  # Checked
             # Add image formats if not already present
             current_types = self.types_edit.text()
-            if not any(img_type in current_types for img_type in image_types.split(",")):
+            if not any(
+                img_type in current_types for img_type in image_types.split(",")
+            ):
                 all_types = f"{base_types},{image_types}"
                 self.types_edit.setText(all_types)
         else:  # Unchecked
             # Remove image formats
             current_types = self.types_edit.text()
             types_list = [t.strip() for t in current_types.split(",")]
-            filtered_types = [t for t in types_list if t not in ["jpg", "jpeg", "png", "gif", "bmp", "tiff"]]
+            filtered_types = [
+                t
+                for t in types_list
+                if t not in ["jpg", "jpeg", "png", "gif", "bmp", "tiff"]
+            ]
             self.types_edit.setText(",".join(filtered_types))
 
     def _create_actions_section(self, parent_layout: QVBoxLayout) -> None:
@@ -753,7 +798,12 @@ class DocScanGUI(QMainWindow):
         # Results table
         self.results_table = QTableWidget()
         self.results_table.setColumnCount(4)
-        self.results_table.setHorizontalHeaderLabels([t("file"), t("type"), t("matches"), t("time")])
+        self.results_table.setHorizontalHeaderLabels([
+            t("file"),
+            t("type"),
+            t("matches"),
+            t("time"),
+        ])
         self.results_table.horizontalHeader().setStretchLastSection(True)
         results_layout.addWidget(QLabel(t("match_details")))
         results_layout.addWidget(self.results_table)
@@ -789,7 +839,9 @@ class DocScanGUI(QMainWindow):
         recent_dirs = self.config_manager.get("recent_directories", [])
         start_dir = recent_dirs[0] if recent_dirs else str(Path.cwd())
 
-        dir_path = QFileDialog.getExistingDirectory(self, t("select_input_directory"), start_dir)
+        dir_path = QFileDialog.getExistingDirectory(
+            self, t("select_input_directory"), start_dir
+        )
         if dir_path:
             self.dir_edit.setText(str(Path(dir_path)))
 
@@ -803,11 +855,15 @@ class DocScanGUI(QMainWindow):
             input_dir = Path(dir_text)
             if input_dir.exists() and input_dir.is_dir():
                 # Search for rules.json or rules*.json files
-                rule_files = list(input_dir.glob("rules.json")) + list(input_dir.glob("rules*.json"))
+                rule_files = list(input_dir.glob("rules.json")) + list(
+                    input_dir.glob("rules*.json")
+                )
 
                 if rule_files:
                     # Use the first matching file, prefer exact "rules.json"
-                    exact_match = next((f for f in rule_files if f.name == "rules.json"), None)
+                    exact_match = next(
+                        (f for f in rule_files if f.name == "rules.json"), None
+                    )
                     rules_file = exact_match if exact_match else rule_files[0]
                     self.rules_edit.setText(str(rules_file.resolve()))
             # Validate rules file after directory change
@@ -820,9 +876,13 @@ class DocScanGUI(QMainWindow):
         """Open file browser dialog for rules file."""
         # Get recent rules files for initial path
         recent_files = self.config_manager.get("recent_rules_files", [])
-        start_dir = str(Path(recent_files[0]).parent) if recent_files else str(Path.cwd())
+        start_dir = (
+            str(Path(recent_files[0]).parent) if recent_files else str(Path.cwd())
+        )
 
-        file_path, _ = QFileDialog.getOpenFileName(self, t("select_rules_file"), start_dir, t("json_files"))
+        file_path, _ = QFileDialog.getOpenFileName(
+            self, t("select_rules_file"), start_dir, t("json_files")
+        )
         if file_path:
             self.rules_edit.setText(str(Path(file_path)))
             self._validate_rules_file()
@@ -856,7 +916,9 @@ class DocScanGUI(QMainWindow):
                 """)
                 if self.rules_validation_label:
                     self.rules_validation_label.setText("✓")
-                    self.rules_validation_label.setStyleSheet("color: #4CAF50; font-weight: bold;")
+                    self.rules_validation_label.setStyleSheet(
+                        "color: #4CAF50; font-weight: bold;"
+                    )
             else:
                 # File doesn't exist - show red border and X
                 self.rules_edit.setStyleSheet("""
@@ -868,7 +930,9 @@ class DocScanGUI(QMainWindow):
                 """)
                 if self.rules_validation_label:
                     self.rules_validation_label.setText("✗")
-                    self.rules_validation_label.setStyleSheet("color: #F44336; font-weight: bold;")
+                    self.rules_validation_label.setStyleSheet(
+                        "color: #F44336; font-weight: bold;"
+                    )
         except Exception:
             # Invalid path - show red border and X
             self.rules_edit.setStyleSheet("""
@@ -880,7 +944,9 @@ class DocScanGUI(QMainWindow):
             """)
             if self.rules_validation_label:
                 self.rules_validation_label.setText("✗")
-                self.rules_validation_label.setStyleSheet("color: #F44336; font-weight: bold;")
+                self.rules_validation_label.setStyleSheet(
+                    "color: #F44336; font-weight: bold;"
+                )
 
     def _load_rules(self) -> list[Rule]:
         """Load rules from JSON file.
@@ -981,8 +1047,14 @@ class DocScanGUI(QMainWindow):
         # Update summary
         scan_info = results.get("scan_info", {})
         processed = scan_info.get("files_processed", scan_info.get("total_files", 0))
-        self.files_label.setText(t("files_scanned").replace("0", f"{processed}/{scan_info.get('total_files', 0)}"))
-        self.matches_label.setText(f"{t('files_with_matches')} {scan_info.get('files_with_matches', 0)}")
+        self.files_label.setText(
+            t("files_scanned").replace(
+                "0", f"{processed}/{scan_info.get('total_files', 0)}"
+            )
+        )
+        self.matches_label.setText(
+            f"{t('files_with_matches')} {scan_info.get('files_with_matches', 0)}"
+        )
 
         # Update progress bar to 100%
         self.progress_bar.setValue(100)
@@ -1005,9 +1077,15 @@ class DocScanGUI(QMainWindow):
         # Determine status: completed takes precedence if scan finished normally, otherwise stopped
         # A scan is considered completed if it processed files or was intentionally stopped after starting
         scan_info = results.get("scan_info", {})
-        files_processed = scan_info.get("files_processed", scan_info.get("total_files", 0))
+        files_processed = scan_info.get(
+            "files_processed", scan_info.get("total_files", 0)
+        )
         was_stopped = results.get("stopped", False)
-        status = t("scan_completed") if files_processed > 0 or not was_stopped else t("scan_stopped")
+        status = (
+            t("scan_completed")
+            if files_processed > 0 or not was_stopped
+            else t("scan_stopped")
+        )
         self._log_message(status)
         self._log_message(t("found_matches_files", count=len(matches)))
 
@@ -1100,7 +1178,9 @@ class DocScanGUI(QMainWindow):
         for match in match_data.get("matches", []):
             details.append(f"Rule: {match.get('rule_name', '')}")
             details.append(f"Description: {match.get('rule_description', '')}")
-            details.append(f"Line {match.get('line_number', 0)}: {match.get('match', '')}")
+            details.append(
+                f"Line {match.get('line_number', 0)}: {match.get('match', '')}"
+            )
             details.append("\nContext:")
             for ctx_line in match.get("context", []):
                 details.append(f"  {ctx_line}")
@@ -1115,16 +1195,22 @@ class DocScanGUI(QMainWindow):
             return
 
         default_name = f"scan_results_{datetime.now().strftime('%Y%m%d_%H%M%S')}.json"
-        file_path, _ = QFileDialog.getSaveFileName(self, "Save Results", default_name, "JSON Files (*.json)")
+        file_path, _ = QFileDialog.getSaveFileName(
+            self, "Save Results", default_name, "JSON Files (*.json)"
+        )
 
         if file_path:
             try:
                 with open(file_path, "w", encoding="utf-8") as f:
                     json.dump(self.scan_results, f, indent=2, ensure_ascii=False)
                 self._log_message(t("results_saved_to", path=file_path))
-                QMessageBox.information(self, t("success"), t("results_saved_to", path=file_path))
+                QMessageBox.information(
+                    self, t("success"), t("results_saved_to", path=file_path)
+                )
             except Exception as e:
-                QMessageBox.critical(self, t("error"), t("failed_to_save_results", error=e))
+                QMessageBox.critical(
+                    self, t("error"), t("failed_to_save_results", error=e)
+                )
 
     def _open_results(self) -> None:
         """Open and load previously saved scan results from JSON file."""
@@ -1152,9 +1238,17 @@ class DocScanGUI(QMainWindow):
 
             # Update summary
             scan_info = results.get("scan_info", {})
-            processed = scan_info.get("files_processed", scan_info.get("total_files", 0))
-            self.files_label.setText(t("files_scanned").replace("0", f"{processed}/{scan_info.get('total_files', 0)}"))
-            self.matches_label.setText(f"{t('files_with_matches')} {scan_info.get('files_with_matches', 0)}")
+            processed = scan_info.get(
+                "files_processed", scan_info.get("total_files", 0)
+            )
+            self.files_label.setText(
+                t("files_scanned").replace(
+                    "0", f"{processed}/{scan_info.get('total_files', 0)}"
+                )
+            )
+            self.matches_label.setText(
+                f"{t('files_with_matches')} {scan_info.get('files_with_matches', 0)}"
+            )
 
             # Update progress bar to 100% since this is completed work
             self.progress_bar.setValue(100)
@@ -1167,9 +1261,13 @@ class DocScanGUI(QMainWindow):
                 file_path = match_data.get("file_path", "")
                 file_type = match_data.get("file_type", "")
                 match_count = len(match_data.get("matches", []))
-                proc_time = match_data.get("metadata", {}).get("processing_time_seconds", 0)
+                proc_time = match_data.get("metadata", {}).get(
+                    "processing_time_seconds", 0
+                )
 
-                self.results_table.setItem(row, 0, QTableWidgetItem(Path(file_path).name))
+                self.results_table.setItem(
+                    row, 0, QTableWidgetItem(Path(file_path).name)
+                )
                 self.results_table.setItem(row, 1, QTableWidgetItem(file_type))
                 self.results_table.setItem(row, 2, QTableWidgetItem(str(match_count)))
                 self.results_table.setItem(row, 3, QTableWidgetItem(f"{proc_time:.3f}"))
@@ -1180,10 +1278,14 @@ class DocScanGUI(QMainWindow):
 
             # Log the action
             self._log_message(t("loaded_results_from", path=file_path))
-            QMessageBox.information(self, t("success"), t("results_loaded_successfully", path=file_path))
+            QMessageBox.information(
+                self, t("success"), t("results_loaded_successfully", path=file_path)
+            )
 
         except Exception as e:
-            QMessageBox.critical(self, t("error"), t("failed_to_load_results", error=str(e)))
+            QMessageBox.critical(
+                self, t("error"), t("failed_to_load_results", error=str(e))
+            )
 
     def _clear_results(self) -> None:
         """Clear all results and logs."""
@@ -1227,7 +1329,9 @@ class DocScanGUI(QMainWindow):
         self.rules_edit.setText(rules_file)
 
         # Restore file types
-        file_types = self.config_manager.get("file_types", "pdf,docx,xlsx,pptx,txt,odt,rtf,epub,csv,xml,html,md")
+        file_types = self.config_manager.get(
+            "file_types", "pdf,docx,xlsx,pptx,txt,odt,rtf,epub,csv,xml,html,md"
+        )
         self.types_edit.setText(file_types)
 
         # Validate rules file after loading config