IO4IT 4.0.22__tar.gz → 4.0.23__tar.gz

io4it-4.0.23/PKG-INFO

@@ -0,0 +1,28 @@
+Metadata-Version: 2.4
+Name: io4it
+Version: 4.0.23
+Home-page: 
+Author: 
+Author-email: 
+Keywords: orange3 add-on
+Requires-Dist: torchvision
+Requires-Dist: torch
+Requires-Dist: pylatexenc
+Requires-Dist: docopt
+Requires-Dist: boto3
+Requires-Dist: opencv-python-headless
+Requires-Dist: docling
+Requires-Dist: docling-core
+Requires-Dist: pypandoc
+Requires-Dist: pypandoc-binary
+Requires-Dist: scikit-learn
+Requires-Dist: openai
+Requires-Dist: pip-system-certs
+Requires-Dist: docx2pdf
+Requires-Dist: msal
+Requires-Dist: exchangelib
+Requires-Dist: html2text
+Requires-Dist: ddgs
+Requires-Dist: CATEGORIT
+Dynamic: keywords
+Dynamic: requires-dist

io4it-4.0.23/io4it.egg-info/PKG-INFO

@@ -0,0 +1,28 @@
+Metadata-Version: 2.4
+Name: io4it
+Version: 4.0.23
+Home-page: 
+Author: 
+Author-email: 
+Keywords: orange3 add-on
+Requires-Dist: torchvision
+Requires-Dist: torch
+Requires-Dist: pylatexenc
+Requires-Dist: docopt
+Requires-Dist: boto3
+Requires-Dist: opencv-python-headless
+Requires-Dist: docling
+Requires-Dist: docling-core
+Requires-Dist: pypandoc
+Requires-Dist: pypandoc-binary
+Requires-Dist: scikit-learn
+Requires-Dist: openai
+Requires-Dist: pip-system-certs
+Requires-Dist: docx2pdf
+Requires-Dist: msal
+Requires-Dist: exchangelib
+Requires-Dist: html2text
+Requires-Dist: ddgs
+Requires-Dist: CATEGORIT
+Dynamic: keywords
+Dynamic: requires-dist

{io4it-4.0.22 → io4it-4.0.23}/io4it.egg-info/SOURCES.txt

@@ -11,11 +11,12 @@ orangecontrib/IO4IT/__init__.py
 orangecontrib/IO4IT/ocr_function/__init__.py
 orangecontrib/IO4IT/ocr_function/word_converter.py
 orangecontrib/IO4IT/utils/__init__.py
+orangecontrib/IO4IT/utils/custom_api.py
 orangecontrib/IO4IT/utils/environment_collector.py
+orangecontrib/IO4IT/utils/file2images.py
 orangecontrib/IO4IT/utils/keys_manager.py
 orangecontrib/IO4IT/utils/mail.py
 orangecontrib/IO4IT/utils/pool_exec_utils.py
-orangecontrib/IO4IT/utils/pptx2jpg.py
 orangecontrib/IO4IT/utils/secret_manager.py
 orangecontrib/IO4IT/utils/utils_md.py
 orangecontrib/IO4IT/utils/pst_extractor/TigerODMPST.dll
@@ -24,6 +25,7 @@ orangecontrib/IO4IT/utils/pst_extractor/export_pst.py
 orangecontrib/IO4IT/utils/pst_extractor/libpff.dll
 orangecontrib/IO4IT/utils/pst_extractor/z.dll
 orangecontrib/IO4IT/widgets/OWChatGpt.py
+orangecontrib/IO4IT/widgets/OWCustomLLMApi.py
 orangecontrib/IO4IT/widgets/OWDeep_Search.py
 orangecontrib/IO4IT/widgets/OWDoclingToMarkdown.py
 orangecontrib/IO4IT/widgets/OWExportMarkdown.py
@@ -58,9 +60,11 @@ orangecontrib/IO4IT/widgets/designer/ow_extract_tables_docx_to_xlsx.ui
 orangecontrib/IO4IT/widgets/designer/ow_file_ext_selector.ui
 orangecontrib/IO4IT/widgets/designer/ow_json_to_data_table.ui
 orangecontrib/IO4IT/widgets/designer/owchatgpt.ui
+orangecontrib/IO4IT/widgets/designer/owcustomllmapi.ui
 orangecontrib/IO4IT/widgets/designer/owdeepsearch.ui
 orangecontrib/IO4IT/widgets/designer/owdoclingtomarkdown.ui
 orangecontrib/IO4IT/widgets/designer/owexportmarkdown.ui
+orangecontrib/IO4IT/widgets/designer/owfile2images.ui
 orangecontrib/IO4IT/widgets/designer/owinboxmailmonitoring.ui
 orangecontrib/IO4IT/widgets/designer/owinsertdatatoexcel.ui
 orangecontrib/IO4IT/widgets/designer/owmailloader.ui
@@ -72,7 +76,6 @@ orangecontrib/IO4IT/widgets/designer/owofficenormalizer.ui
 orangecontrib/IO4IT/widgets/designer/owparserhtml.ui
 orangecontrib/IO4IT/widgets/designer/owpdftype.ui
 orangecontrib/IO4IT/widgets/designer/owpipauditor.ui
-orangecontrib/IO4IT/widgets/designer/owpptx2jpg.ui
 orangecontrib/IO4IT/widgets/designer/owprocesspoolexecutor.ui
 orangecontrib/IO4IT/widgets/designer/owspeechtotext.ui
 orangecontrib/IO4IT/widgets/designer/owvisualizationer.ui
@@ -86,6 +89,7 @@ orangecontrib/IO4IT/widgets/icons/dep_md_old.png
 orangecontrib/IO4IT/widgets/icons/download.png
 orangecontrib/IO4IT/widgets/icons/export_md.png
 orangecontrib/IO4IT/widgets/icons/extract_table.png
+orangecontrib/IO4IT/widgets/icons/file2images.png
 orangecontrib/IO4IT/widgets/icons/file_extensor.png
 orangecontrib/IO4IT/widgets/icons/html.png
 orangecontrib/IO4IT/widgets/icons/json-file.png
@@ -96,9 +100,9 @@ orangecontrib/IO4IT/widgets/icons/mail_writer.png
 orangecontrib/IO4IT/widgets/icons/md.png
 orangecontrib/IO4IT/widgets/icons/monitor-email.svg
 orangecontrib/IO4IT/widgets/icons/office_normalizer.png
+orangecontrib/IO4IT/widgets/icons/owcustomllmapi.svg
 orangecontrib/IO4IT/widgets/icons/owmd2html.svg
 orangecontrib/IO4IT/widgets/icons/pip_auditor.png
-orangecontrib/IO4IT/widgets/icons/pptx2jpg.png
 orangecontrib/IO4IT/widgets/icons/process_pool_executor.png
 orangecontrib/IO4IT/widgets/icons/speech_to_text.png
 orangecontrib/IO4IT/widgets/icons/upload.png

io4it-4.0.23/orangecontrib/IO4IT/utils/custom_api.py

@@ -0,0 +1,132 @@
+import base64
+import mimetypes
+import os
+
+import requests
+
+import Orange.data
+from Orange.data import StringVariable
+
+
+def build_headers(header_name, auth_scheme, api_key):
+    header_name = (header_name or "Authorization").strip()
+    auth_scheme = (auth_scheme or "").strip()
+    value = f"{auth_scheme} {api_key}".strip() if auth_scheme else str(api_key)
+    return {"Content-Type": "application/json", header_name: value}
+
+
+def image_to_data_uri(file_path):
+    """Encode a local image file as a base64 data URI."""
+    mime_type, _ = mimetypes.guess_type(file_path)
+    mime_type = mime_type or "application/octet-stream"
+    with open(file_path, "rb") as f:
+        encoded = base64.b64encode(f.read()).decode("utf-8")
+    return f"data:{mime_type};base64,{encoded}"
+
+
+def build_user_content(prompt, image_paths):
+    """Build the 'content' field of a user message, attaching images when provided."""
+    if not image_paths:
+        return str(prompt)
+
+    content = []
+    for image_path in image_paths:
+        image_path = image_path.strip().strip("'").strip('"')
+        if not image_path:
+            continue
+        if not os.path.exists(image_path):
+            content.append({"type": "text", "text": f"[ERROR] Image not found: {image_path}"})
+            continue
+        try:
+            content.append({"type": "image_url", "image_url": {"url": image_to_data_uri(image_path)}})
+        except Exception as e:
+            content.append({"type": "text", "text": f"[ERROR] Unable to read image {image_path}: {e}"})
+    content.append({"type": "text", "text": str(prompt)})
+    return content
+
+
+def call_chat_completion(prompt, system_prompt, config, image_paths=None):
+    """
+    Call a /chat/completions endpoint using the widely adopted chat-completions
+    message format. Vision-capable models can be reached by passing image_paths,
+    which are embedded in the user message as base64 data URIs.
+
+    config keys: base_url, route, header_name, auth_scheme, api_key,
+                 model, max_tokens, temperature, cert_path, cert_is_ca,
+                 proxy_url
+    """
+    url = config["base_url"].rstrip("/") + config["route"]
+    headers = build_headers(config["header_name"], config["auth_scheme"], config["api_key"])
+
+    messages = []
+    if system_prompt:
+        messages.append({"role": "system", "content": str(system_prompt)})
+    messages.append({"role": "user", "content": build_user_content(prompt, image_paths)})
+
+    payload = {
+        "model": config["model"],
+        "messages": messages,
+        "temperature": config["temperature"],
+        "max_tokens": config["max_tokens"],
+    }
+
+    kwargs = {"timeout": 600}
+    cert_path = config.get("cert_path")
+    if cert_path:
+        if config.get("cert_is_ca"):
+            kwargs["verify"] = cert_path
+        else:
+            kwargs["cert"] = cert_path
+
+    if config.get("use_proxy"):
+        # Explicitly set, even when empty: an empty value forces a direct
+        # connection and bypasses any http_proxy/https_proxy environment
+        # variable (e.g. the one Orange itself sets from its preferences),
+        # instead of silently inheriting it.
+        proxy_url = config.get("proxy_url") or ""
+        kwargs["proxies"] = {"http": proxy_url, "https": proxy_url}
+
+    response = requests.post(url, headers=headers, json=payload, **kwargs)
+    response.raise_for_status()
+    data = response.json()
+    return data["choices"][0]["message"]["content"]
+
+
+def generate_answers(data, config, progress_callback=None):
+    """
+    Runs call_chat_completion for every row of `data` (needs a 'prompt' column,
+    optionally a 'system_prompt' column and an 'image paths' column for
+    vision-capable models) and returns a new Table with an added 'Answer'
+    string column.
+
+    The 'image paths' column, when present, holds one or more image file
+    paths separated by ';' for that row.
+    """
+    domain = data.domain
+    has_system_prompt = "system_prompt" in domain
+    has_images = "image paths" in domain
+
+    prompts = data.get_column("prompt")
+    system_prompts = data.get_column("system_prompt") if has_system_prompt else None
+    image_paths_column = data.get_column("image paths") if has_images else None
+
+    answers = []
+    n_rows = len(data)
+    for i in range(n_rows):
+        prompt = prompts[i]
+        system_prompt = system_prompts[i] if system_prompts is not None else None
+        image_paths = image_paths_column[i].split(";") if image_paths_column is not None else None
+        try:
+            answer = call_chat_completion(prompt, system_prompt, config, image_paths)
+        except requests.HTTPError as e:
+            body = e.response.text if e.response is not None else str(e)
+            answer = f"[ERROR] HTTP {e.response.status_code if e.response is not None else '?'}: {body}"
+        except Exception as e:
+            answer = f"[ERROR] {e}"
+
+        answers.append(answer)
+        if progress_callback is not None:
+            progress_callback((int(100 * (i + 1) / max(n_rows, 1)), f"Row {i + 1}/{n_rows} done\n"))
+
+    answer_var = StringVariable("Answer")
+    return data.add_column(answer_var, answers)

io4it-4.0.23/orangecontrib/IO4IT/utils/file2images.py

@@ -0,0 +1,211 @@
+import sys
+import time
+import subprocess
+import shutil
+from pathlib import Path
+
+
+def get_conversion_method():
+    """Determines the best available conversion method."""
+    # 1. Check for MS Office
+    if sys.platform.startswith("win"):
+        try:
+            import win32com.client
+            # Try to see if the COM object is registered
+            win32com.client.GetActiveObject("PowerPoint.Application")
+            return "office"
+        except:
+            try:
+                # If not running, try to see if it's dispatchable
+                win32com.client.Dispatch("PowerPoint.Application")
+                return "office"
+            except:
+                pass
+    elif sys.platform == "darwin":
+        # Check if PowerPoint app exists on Mac
+        if Path("/Applications/Microsoft PowerPoint.app").exists():
+            return "office"
+
+    # 2. Check for LibreOffice (soffice)
+    if shutil.which("soffice"):
+        return "libreoffice"
+
+    # 3. Fallback to Aspose
+    return "aspose"
+
+
+def convert_via_office(src: Path, pdf_path: Path):
+    if sys.platform.startswith("win"):
+        import win32com.client
+        powerpoint = win32com.client.Dispatch("PowerPoint.Application")
+        try:
+            presentation = powerpoint.Presentations.Open(str(src), WithWindow=False)
+            presentation.SaveAs(str(pdf_path), 32)  # 32 = ppSaveAsPDF
+            presentation.Close()
+        finally:
+            # Note: We don't quit() powerpoint here to avoid killing other open docs
+            pass
+    elif sys.platform == "darwin":
+        script = f'tell application "Microsoft PowerPoint" to save (open POSIX file "{src}") in POSIX file "{pdf_path}" as PDF'
+        subprocess.run(["osascript", "-e", script], check=True)
+
+
+def convert_via_libreoffice(src: Path, out_dir: Path):
+    subprocess.run([
+        "soffice", "--headless", "--convert-to", "pdf",
+        "--outdir", str(out_dir), str(src)
+    ], check=True, capture_output=True)
+    return out_dir / f"{src.stem}.pdf"
+
+
+def parse_pages(spec, total):
+    """Convertit une spécification de pages en liste d'indices 1-based, triés et dédupliqués.
+
+    Exemples (total = nombre de pages du document) :
+        None / "" / "all" / "*"  -> toutes les pages
+        "1"                      -> [1]
+        "1,3,5"                  -> [1, 3, 5]
+        "1-3"                    -> [1, 2, 3]
+        "2-4,6"                  -> [2, 3, 4, 6]
+    Les pages hors intervalle [1, total] sont ignorées silencieusement.
+    """
+    if spec is None:
+        return list(range(1, total + 1))
+
+    spec = str(spec).strip().lower()
+    if spec in ("", "all", "toutes", "tout", "*"):
+        return list(range(1, total + 1))
+
+    pages = set()
+    for part in spec.split(","):
+        part = part.strip()
+        if not part:
+            continue
+        try:
+            if "-" in part:
+                a, b = part.split("-", 1)
+                a, b = int(a), int(b)
+                for p in range(min(a, b), max(a, b) + 1):
+                    pages.add(p)
+            else:
+                pages.add(int(part))
+        except ValueError:
+            # Fragment invalide (ex. "abc") -> on l'ignore
+            continue
+
+    return sorted(p for p in pages if 1 <= p <= total)
+
+
+# Formats gérés nativement par Pixmap.save() de PyMuPDF.
+# Les autres (tif/tiff, webp, bmp...) passent par Pillow via pix.pil_save().
+_NATIVE_FORMATS = {"jpg", "jpeg", "png", "pnm", "pgm", "ppm", "pbm", "pam", "tga", "ps", "psd"}
+
+
+def normalize_ext(out_ext):
+    """Nettoie l'extension demandée : minuscule, sans point initial, jpg par défaut."""
+    ext = str(out_ext or "jpg").strip().lower().lstrip(".")
+    if ext == "jpeg":
+        ext = "jpg"
+    return ext or "jpg"
+
+
+def normalize_dpi(dpi, default=150):
+    """Convertit le DPI en entier valide (> 0), sinon retombe sur `default`."""
+    try:
+        dpi = int(dpi)
+    except (TypeError, ValueError):
+        return default
+    return dpi if dpi > 0 else default
+
+
+def _save_pixmap(pix, img_path: Path, ext: str):
+    """Sauvegarde le pixmap au bon format selon l'extension."""
+    if ext in _NATIVE_FORMATS:
+        pix.save(str(img_path))
+    else:
+        # tif/tiff et autres formats -> nécessite Pillow
+        pix.pil_save(str(img_path))
+
+
+def process_one_file(file_path_str: str, out_dir_str: str = None,
+                     pages: str = "all", out_ext: str = "jpg",
+                     dpi: int = 150, unique_folder: bool = True):
+    """Convertit un PPTX (ou PDF) en images.
+
+    Args:
+        file_path_str : chemin du fichier source (.pptx ou .pdf).
+        out_dir_str   : dossier de sortie. Si None/vide, on retombe sur le
+                        comportement historique (images_<nom> ou images/).
+        pages         : pages/slides à extraire ("all", "1", "1,3", "1-3"...).
+        out_ext       : format d'image de sortie ("jpg", "png", "tif"...).
+        dpi           : résolution de rendu en points par pouce (défaut 150).
+        unique_folder : utilisé seulement quand out_dir_str n'est pas fourni.
+
+    Retour : [src, "img1|img2|...", "ok"/"nok", durée, message]
+    """
+    try:
+        import fitz  # PyMuPDF
+    except ImportError:
+        return [file_path_str, "", "nok", "0.0", "Erreur: PyMuPDF non installé"]
+
+    t0 = time.time()
+    src = Path(file_path_str).resolve()
+    ext_out = normalize_ext(out_ext)
+    dpi_out = normalize_dpi(dpi)
+
+    # --- Dossier de sortie : colonne prioritaire, sinon fallback historique ---
+    if out_dir_str and str(out_dir_str).strip():
+        out_dir = Path(str(out_dir_str).strip())
+    else:
+        out_dir = src.parent / (f"images_{src.stem}" if unique_folder else "images")
+    out_dir.mkdir(parents=True, exist_ok=True)
+
+    img_list = []
+    ext = src.suffix.lower()
+    method = "pdf"          # méthode par défaut si l'entrée est déjà un PDF
+    temp_pdf = False        # True si on a généré un PDF intermédiaire à supprimer
+
+    try:
+        # --- PHASE 1 : obtenir un PDF ---
+        if ext == ".pdf":
+            pdf_path = src
+        else:
+            method = get_conversion_method()
+            if method == "aspose":
+                # import aspose.slides as slides  # (fallback désactivé)
+                return [str(src), "", "nok", f"{time.time()-t0:.2f}",
+                        "Conversion impossible sans LibreOffice ou MS Office"]
+
+            pdf_path = out_dir / f"{src.stem}.pdf"
+            temp_pdf = True
+            if method == "office":
+                convert_via_office(src, pdf_path)
+            else:  # libreoffice
+                convert_via_libreoffice(src, out_dir)
+
+        # --- PHASE 2 : rasteriser uniquement les pages demandées ---
+        doc = fitz.open(str(pdf_path))
+        wanted = parse_pages(pages, doc.page_count)
+
+        for i in wanted:
+            page = doc.load_page(i - 1)  # fitz est 0-based
+            pix = page.get_pixmap(dpi=dpi_out)
+            img_path = out_dir / f"{src.stem}_slide_{i:03d}.{ext_out}"
+            _save_pixmap(pix, img_path, ext_out)
+            img_list.append(str(img_path))
+        doc.close()
+
+        if temp_pdf and pdf_path.exists():
+            pdf_path.unlink()
+
+        duration = time.time() - t0
+        if not img_list:
+            return [str(src), "", "nok", f"{duration:.2f}",
+                    f"Aucune page valide pour la spec '{pages}' ({method})"]
+
+        return [str(src), "|".join(img_list), "ok", f"{duration:.2f}",
+                f"{len(img_list)} image(s) .{ext_out} @ {dpi_out} DPI ({method})"]
+
+    except Exception as e:
+        return [str(src), "", "nok", f"{time.time()-t0:.2f}",
+                f"Method {method} failed: {str(e)}"]