io4it 2.1.0.8__tar.gz → 2.1.1__tar.gz

{io4it-2.1.0.8 → io4it-2.1.1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: io4it
-Version: 2.1.0.8
+Version: 2.1.1
 Home-page: 
 Author: 
 Author-email: 

{io4it-2.1.0.8 → io4it-2.1.1}/io4it.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: io4it
-Version: 2.1.0.8
+Version: 2.1.1
 Home-page: 
 Author: 
 Author-email: 

{io4it-2.1.0.8 → io4it-2.1.1}/orangecontrib/IO4IT/widgets/OWDoclingMarkdownizerSimple.py

@@ -8,6 +8,9 @@ from Orange.widgets import widget
 from Orange.widgets.utils.signals import Input, Output
 from Orange.data import Domain, StringVariable, Table, DiscreteVariable
 
+# --- Ajouts pour l'écriture Excel ---
+from openpyxl import Workbook
+
 # --- Docling (unique lib utilisée pour la conversion) ---
 from docling.backend.pypdfium2_backend import PyPdfiumDocumentBackend
 from docling.datamodel.base_models import InputFormat
@@ -208,9 +211,30 @@ class OWDoclingMarkdownizerSimple(widget.OWWidget):
         self.Outputs.status_data.send(status_table)
 
     def _run_conversion(self, files, progress_callback):
-        """Main function to run the conversion, supports sequential and parallel modes."""
         results = []
 
+        # Gère le chemin du fichier Excel
+        base_name = "conversion_results"  # Nom de base pour le fichier Excel
+        # Définir le chemin du dossier de sortie de Docling
+        if files:
+            first_file_path = Path(files[0])
+            out_dir = first_file_path.parent / "conversion_markdown"
+        else:
+            out_dir = Path.cwd() / "conversion_markdown"
+
+        excel_path = out_dir / f"{base_name}.xlsx"
+        counter = 1
+        while excel_path.exists():
+            excel_path = out_dir / f"{base_name}_{counter}.xlsx"
+            counter += 1
+
+        # Initialise le classeur et la feuille Excel
+        wb = Workbook()
+        ws = wb.active
+        ws.title = "Conversion Results"
+        headers = ["input_path", "output_md", "status", "duration_sec", "message"]
+        ws.append(headers)
+
         if self.external_executor is None:
             # --- Mode simple séquentiel ---
             for i, p in enumerate(files):
@@ -220,8 +244,8 @@ class OWDoclingMarkdownizerSimple(widget.OWWidget):
 
                 row = _convert_one_file(path_str)
                 results.append(row)
-
-                # Mise à jour du tableau avec le résultat et envoi immédiat
+                ws.append(row)
+                wb.save(excel_path)
                 self.status_update_signal.emit([row[0], row[2], row[4]])
 
                 progress_callback((i + 1) / len(files) * 100)
@@ -237,13 +261,16 @@ class OWDoclingMarkdownizerSimple(widget.OWWidget):
                 try:
                     row = fut.result()
                     results.append(row)
-                    # Mise à jour du résultat de la future et envoi
+                    ws.append(row)
+                    wb.save(excel_path)
                     self.status_update_signal.emit([row[0], row[2], row[4]])
                 except Exception as e:
                     # Gestion des erreurs de la future et envoi
                     row = [file_path_str, str((Path(file_path_str).parent / 'a_md' / f"{Path(file_path_str).stem}.md")),
                            "nok", "0.00", f"FutureError: {e}"]
                     results.append(row)
+                    ws.append(row)
+                    wb.save(excel_path)
                     self.status_update_signal.emit([row[0], "nok", f"FutureError: {e}"])
 
                 progress_callback(i / len(files) * 100)

{io4it-2.1.0.8 → io4it-2.1.1}/orangecontrib/IO4IT/widgets/OWOfficeNormalizer.py

@@ -9,6 +9,9 @@ from Orange.widgets import widget
 from Orange.widgets.utils.signals import Input, Output
 from Orange.data import Domain, StringVariable, Table, DiscreteVariable
 
+# --- Ajout pour l'écriture Excel ---
+from openpyxl import Workbook
+
 if "site-packages/Orange/widgets" in os.path.dirname(os.path.abspath(__file__)).replace("\\", "/"):
     from Orange.widgets.orangecontrib.IO4IT.utils import utils_md
     from Orange.widgets.orangecontrib.AAIT.utils.import_uic import uic
@@ -20,7 +23,7 @@ else:
 class OWOfficeNormalizer(widget.OWWidget):
     name = "Office Normalizer"
     description = "Convertit .doc→.docx et .ppt→.pptx via COM (Windows + Office)"
-    category = "AAIT - LLM INTEGRATION"
+    category = "IO4IT"
     icon = "icons/office_normalizer.png"
     if "site-packages/Orange/widgets" in os.path.dirname(os.path.abspath(__file__)).replace("\\", "/"):
         icon = "icons_dev/office_normalizer.png"
@@ -44,7 +47,7 @@ class OWOfficeNormalizer(widget.OWWidget):
         self.data = None
         self.autorun = True
         self.result = None
-        self.processed_statuses = []  # To store incremental status data
+        self.processed_statuses = []
         self.post_initialized()
 
     @Inputs.data
@@ -67,7 +70,7 @@ class OWOfficeNormalizer(widget.OWWidget):
             return
 
         self.progressBarInit()
-        self.processed_statuses = []  # Reset status list for a new run
+        self.processed_statuses = []
         self.Outputs.status_data.send(None)
 
         # Process files directly without a separate thread
@@ -91,6 +94,21 @@ class OWOfficeNormalizer(widget.OWWidget):
         output_base_dir = common_path / "office_normalisation"
         output_base_dir.mkdir(parents=True, exist_ok=True)
 
+        # Gère le nom du fichier Excel avec incrémentation
+        base_name = "normalization_results"
+        excel_path = output_base_dir / f"{base_name}.xlsx"
+        counter = 1
+        while excel_path.exists():
+            excel_path = output_base_dir / f"{base_name}_{counter}.xlsx"
+            counter += 1
+
+        # Initialise le classeur Excel
+        wb = Workbook()
+        ws = wb.active
+        ws.title = "Normalization Results"
+        headers = ["src_path", "dst_path", "status", "details"]
+        ws.append(headers)
+
         for i, path_str in enumerate(file_paths):
             self.progressBarSet(i / total_files * 100)
 
@@ -136,7 +154,12 @@ class OWOfficeNormalizer(widget.OWWidget):
                     details = f"error: {error_msg}"
                     status_text = f"ko: {details}"
 
-            # Append to the final results list
+            # Ajoute la ligne de résultat à la table Excel et la sauvegarde
+            result_row = [path_str, dst_path, status_short, details]
+            ws.append(result_row)
+            wb.save(excel_path)
+
+            # Append to the final results list for Orange table
             rows.append([path_str, dst_path, status_text])
 
             # Append to the status update list and send the incremental table

{io4it-2.1.0.8 → io4it-2.1.1}/orangecontrib/IO4IT/widgets/OWPdfType.py

@@ -9,6 +9,9 @@ from Orange.data import Domain, StringVariable, Table, DiscreteVariable
 from Orange.widgets import widget
 from Orange.widgets.utils.signals import Input, Output
 
+# --- Ajout pour l'écriture Excel ---
+from openpyxl import Workbook
+
 # Les imports sont adaptés pour correspondre au style de l'autre script
 if "site-packages/Orange/widgets" in os.path.dirname(os.path.abspath(__file__)).replace("\\", "/"):
     from Orange.widgets.orangecontrib.IO4IT.utils import utils_md
@@ -23,7 +26,7 @@ else:
 class OWPdfType(widget.OWWidget):
     name = "PDF Type"
     description = "Checks if a PDF is text-based or image-based"
-    category = "AAIT - LLM INTEGRATION"
+    category = "IO4IT"
     icon = "icons/check_pdf.png"
     if "site-packages/Orange/widgets" in os.path.dirname(os.path.abspath(__file__)).replace("\\", "/"):
         icon = "icons_dev/check_pdf.png"
@@ -99,7 +102,32 @@ class OWPdfType(widget.OWWidget):
 
     def _process_pdfs(self, in_data: Table, progress_callback: callable, status_callback: callable) -> tuple[
         Table | None, Table | None]:
+
+        # Extraction des chemins de fichiers avant de commencer le traitement
         paths = [str(x) for x in in_data.get_column("file_path")]
+
+        # --- Gérer le nom du fichier Excel avec incrémentation ---
+        excel_output_dir = Path.cwd() / "pdf_check_results"
+        if paths:
+            first_file_path = Path(paths[0])
+            excel_output_dir = first_file_path.parent / "pdf_check_results"
+
+        excel_output_dir.mkdir(parents=True, exist_ok=True)
+
+        base_name = "pdf_check_results"
+        excel_path = excel_output_dir / f"{base_name}.xlsx"
+        counter = 1
+        while excel_path.exists():
+            excel_path = excel_output_dir / f"{base_name}_{counter}.xlsx"
+            counter += 1
+
+        # --- Initialiser le classeur Excel ---
+        wb = Workbook()
+        ws = wb.active
+        ws.title = "PDF Check Results"
+        headers = ["file_path", "status", "details"]
+        ws.append(headers)
+
         text_indices = []
         image_indices = []
 
@@ -108,21 +136,36 @@ class OWPdfType(widget.OWWidget):
             progress_callback(i / total_files * 100)
 
             fp = Path(p)
+            result_row = [p, "", ""]  # Initialisation de la ligne de résultat
 
             if not fp.exists() or fp.suffix.lower() != ".pdf":
-                status_callback([p, "ko", "Invalid file or not a PDF"])
+                result_row[1] = "ko"
+                result_row[2] = "Invalid file or not a PDF"
+                status_callback(result_row)
+                ws.append(result_row)
+                wb.save(excel_path)
                 continue
 
             try:
                 is_text = utils_md.is_pdf_text_based(fp)
                 if is_text:
                     text_indices.append(i)
-                    status_callback([p, "ok", "Text-based PDF"])
+                    result_row[1] = "ok"
+                    result_row[2] = "Text-based PDF"
                 else:
                     image_indices.append(i)
-                    status_callback([p, "ok", "Image-based PDF"])
+                    result_row[1] = "ok"
+                    result_row[2] = "Image-based PDF"
+
+                status_callback(result_row)
+                ws.append(result_row)
+                wb.save(excel_path)
             except Exception as e:
-                status_callback([p, "ko", f"Error: {str(e)}"])
+                result_row[1] = "ko"
+                result_row[2] = f"Error: {str(e)}"
+                status_callback(result_row)
+                ws.append(result_row)
+                wb.save(excel_path)
 
         progress_callback(100)
 

{io4it-2.1.0.8 → io4it-2.1.1}/setup.py

@@ -2,7 +2,7 @@ from setuptools import setup, find_packages
 
 # Configuration
 NAME = "io4it"
-VERSION = "2.1.0.8"
+VERSION = "2.1.1"
 
 INSTALL_REQUIRES = [
     "pylatexenc",