IO4IT 4.0.26__tar.gz → 4.0.27__tar.gz

{io4it-4.0.26 → io4it-4.0.27}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: io4it
-Version: 4.0.26
+Version: 4.0.27
 Home-page: 
 Author: 
 Author-email: 

{io4it-4.0.26 → io4it-4.0.27}/io4it.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: io4it
-Version: 4.0.26
+Version: 4.0.27
 Home-page: 
 Author: 
 Author-email: 

{io4it-4.0.26 → io4it-4.0.27}/orangecontrib/IO4IT/utils/custom_api.py

@@ -32,11 +32,14 @@ def search_for_streamed_token(decoded_line):
     dict_regex = r"\{[\s\S]*\}"
     match = re.search(dict_regex, decoded_line)
     if not match:
-        return "Token not found"
-    
+        return ""
+
     json_data = json.loads(match.group())
     delta = json_data["choices"][0]["delta"]
-    token = next((delta[key] for key in known_keys if key in delta), "Token not found")
+    # .get(key) (not "key in delta") also skips a key whose value is
+    # JSON null, e.g. {"delta": {"role": "assistant", "content": null}},
+    # which some OpenAI-compatible servers (llama-server included) send.
+    token = next((delta[key] for key in known_keys if delta.get(key)), "")
     return token
 
 
@@ -58,6 +61,7 @@ def call_chat_completion(messages, config, stream=False, workflow_id="Request_RA
         "messages": messages,
         "temperature": config["temperature"],
         "max_tokens": config["max_tokens"],
+        "stream": stream,
     }
 
     kwargs = {"timeout": 600}

{io4it-4.0.26 → io4it-4.0.27}/orangecontrib/IO4IT/widgets/OWCustomLLMApi.py

@@ -12,12 +12,14 @@ from AnyQt.QtWidgets import (
 
 if "site-packages/Orange/widgets" in os.path.dirname(os.path.abspath(__file__)).replace("\\", "/"):
     from Orange.widgets.orangecontrib.IO4IT.utils import custom_api
+    from Orange.widgets.orangecontrib.IO4IT.utils import keys_manager
     from Orange.widgets.orangecontrib.AAIT.utils import thread_management
     from Orange.widgets.orangecontrib.AAIT.llm.answers_llama import identify_table_type
     from Orange.widgets.orangecontrib.AAIT.utils.import_uic import uic
     from Orange.widgets.orangecontrib.AAIT.utils.initialize_from_ini import apply_modification_from_python_file
 else:
     from orangecontrib.IO4IT.utils import custom_api
+    from orangecontrib.IO4IT.utils import keys_manager
     from orangecontrib.AAIT.utils import thread_management
     from orangecontrib.AAIT.llm.answers_llama import identify_table_type
     from orangecontrib.AAIT.utils.import_uic import uic
@@ -51,6 +53,7 @@ class OWCustomLLMApi(widget.OWWidget):
     header_name = Setting("Authorization")
     auth_scheme = Setting("Bearer")
     api_key = Setting("")
+    key_service = Setting("")
     cert_path = Setting("")
     cert_is_ca = Setting(False)
     use_proxy = Setting(False)
@@ -79,7 +82,7 @@ class OWCustomLLMApi(widget.OWWidget):
     def __init__(self):
         super().__init__()
         self.setFixedWidth(700)
-        self.setFixedHeight(695)
+        self.setFixedHeight(733)
         uic.loadUi(self.gui, self)
 
         self.label_description = self.findChild(QLabel, 'Description')
@@ -104,6 +107,10 @@ class OWCustomLLMApi(widget.OWWidget):
         self.line_api_key.setText(self.api_key)
         self.line_api_key.editingFinished.connect(self.update_parameters)
 
+        self.line_key_service = self.findChild(QLineEdit, 'lineKeyService')
+        self.line_key_service.setText(self.key_service)
+        self.line_key_service.editingFinished.connect(self.update_parameters)
+
         self.line_cert_path = self.findChild(QLineEdit, 'lineCertPath')
         self.line_cert_path.setText(self.cert_path)
         self.line_cert_path.editingFinished.connect(self.update_parameters)
@@ -144,6 +151,7 @@ class OWCustomLLMApi(widget.OWWidget):
         self.thread = None
         self.can_run = True
         self.result = None
+        self._resolved_api_key = None
 
         self.post_initialized()
 
@@ -153,6 +161,7 @@ class OWCustomLLMApi(widget.OWWidget):
         self.header_name = self.line_header_name.text().strip() or "Authorization"
         self.auth_scheme = self.line_auth_scheme.text().strip()
         self.api_key = self.line_api_key.text()
+        self.key_service = self.line_key_service.text().strip()
         self.cert_path = self.line_cert_path.text().strip()
         self.cert_is_ca = self.check_cert_is_ca.isChecked()
         self.use_proxy = self.check_use_proxy.isChecked()
@@ -167,13 +176,36 @@ class OWCustomLLMApi(widget.OWWidget):
             self.line_cert_path.setText(path)
             self.update_parameters()
 
+    def resolve_api_key(self):
+        """
+        Resolve the API key to actually use. A "Key manager service" name
+        takes priority: the real key is fetched from the secure key store
+        (Key Manager tool) and never persisted in the workflow file. Falls
+        back to the plain-text "API key" field (stored in the .ows file)
+        for legacy workflows that don't use the key store.
+        """
+        if self.key_service:
+            cfg = keys_manager.lire_config_api(self.key_service)
+            if not cfg or not cfg.get("api_key"):
+                self.error(
+                    f"No API key found for service '{self.key_service}' in the Key Manager. "
+                    f"Register it there (Create / overwrite api key), or clear this field to "
+                    f"use the 'API key' field instead."
+                )
+                return None
+            return cfg["api_key"]
+        if not self.api_key:
+            self.error("API key is required (either via 'Key manager service' or the 'API key' field).")
+            return None
+        return self.api_key
+
     def build_config(self):
         return {
             "base_url": self.base_url,
             "route": self.route,
             "header_name": self.header_name,
             "auth_scheme": self.auth_scheme,
-            "api_key": self.api_key,
+            "api_key": self._resolved_api_key,
             "cert_path": self.cert_path,
             "cert_is_ca": self.cert_is_ca,
             "use_proxy": self.use_proxy,
@@ -210,8 +242,8 @@ class OWCustomLLMApi(widget.OWWidget):
             self.Outputs.data.send(None)
             return
 
-        if not self.api_key:
-            self.error("API key is required.")
+        self._resolved_api_key = self.resolve_api_key()
+        if self._resolved_api_key is None:
             self.Outputs.data.send(None)
             return
 

{io4it-4.0.26 → io4it-4.0.27}/orangecontrib/IO4IT/widgets/OWDoclingToMarkdown.py

@@ -14,6 +14,7 @@ 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
+from docling.datamodel.pipeline_options import PdfPipelineOptions, RapidOcrOptions
 from docling.document_converter import (
     DocumentConverter,
     PdfFormatOption,
@@ -31,6 +32,45 @@ else:
     from orangecontrib.AAIT.utils.import_uic import uic
 
 
+# --------- helpers Docling ----------
+def _build_converter(force_ocr: bool = False):
+    """Construit le DocumentConverter.
+    force_ocr=False : configuration d'origine (options PDF par défaut).
+    force_ocr=True  : OCR RapidOCR forcé sur toute la page (PDF image).
+    """
+    pdf_format = PdfFormatOption(
+        pipeline_cls=StandardPdfPipeline,
+        backend=PyPdfiumDocumentBackend
+    )
+    if force_ocr:
+        pdf_options = PdfPipelineOptions()
+        pdf_options.do_ocr = True
+        pdf_options.ocr_options = RapidOcrOptions(force_full_page_ocr=True)
+        pdf_format = PdfFormatOption(
+            pipeline_cls=StandardPdfPipeline,
+            backend=PyPdfiumDocumentBackend,
+            pipeline_options=pdf_options,
+        )
+
+    # Docling minimal config (inspiré du snippet)
+    return DocumentConverter(
+        allowed_formats=[InputFormat.PDF, InputFormat.DOCX, InputFormat.PPTX],
+        format_options={
+            InputFormat.PDF: pdf_format,
+            InputFormat.DOCX: WordFormatOption(
+                pipeline_cls=SimplePipeline
+            ),
+            # PPTX: pas d'option spécifique; géré par défaut
+        },
+    )
+
+
+def _has_text(md: str, min_chars: int = 20) -> bool:
+    """True si le Markdown contient du vrai texte (hors commentaires <!-- image -->)."""
+    cleaned = re.sub(r"<!--.*?-->", "", md, flags=re.DOTALL)
+    return sum(c.isalnum() for c in cleaned) >= min_chars
+
+
 # --------- worker stateless : convertit 1 fichier avec Docling ----------
 def _convert_one_file(file_path_str: str):
     """Convertit un fichier (PDF/DOCX/PPTX) en Markdown via Docling.
@@ -67,7 +107,18 @@ def _convert_one_file(file_path_str: str):
             },
         )
         doc = doc_converter.convert(str(src)).document
+        md = doc.export_to_markdown(page_break_placeholder="<-!-- page-break --!->")
+        # 1er passage : comportement d'origine
+        doc = _build_converter().convert(str(src)).document
         md = doc.export_to_markdown()
+        message = ""
+
+        # Repli : PDF sans texte exploitable (PDF image) -> OCR pleine page
+        if src.suffix.lower() == ".pdf" and not _has_text(md):
+            doc = _build_converter(force_ocr=True).convert(str(src)).document
+            md = doc.export_to_markdown()
+            message = "ocr pleine page (pdf sans texte natif)"
+
         out_md.write_text(md, encoding="utf-8")
         status, message = "ok", ""
     except Exception as e:

{io4it-4.0.26 → io4it-4.0.27}/orangecontrib/IO4IT/widgets/designer/owcustomllmapi.ui

@@ -7,7 +7,7 @@
     <x>0</x>
     <y>0</y>
     <width>700</width>
-    <height>697</height>
+    <height>733</height>
    </rect>
   </property>
   <property name="windowTitle">
@@ -35,7 +35,7 @@
      <x>20</x>
      <y>50</y>
      <width>661</width>
-     <height>281</height>
+     <height>317</height>
     </rect>
    </property>
    <property name="title">
@@ -155,11 +155,14 @@
       <height>29</height>
      </rect>
     </property>
+    <property name="toolTip">
+     <string>Only used when "Key manager service" below is empty. Stored in plain text in the workflow file (.ows) — prefer the Key Manager service field when possible.</string>
+    </property>
     <property name="echoMode">
      <enum>QLineEdit::Password</enum>
     </property>
    </widget>
-   <widget class="QLabel" name="labelCertPath">
+   <widget class="QLabel" name="labelKeyService">
     <property name="geometry">
      <rect>
       <x>10</x>
@@ -168,6 +171,32 @@
       <height>21</height>
      </rect>
     </property>
+    <property name="text">
+     <string>Key manager service :</string>
+    </property>
+   </widget>
+   <widget class="QLineEdit" name="lineKeyService">
+    <property name="geometry">
+     <rect>
+      <x>150</x>
+      <y>170</y>
+      <width>491</width>
+      <height>29</height>
+     </rect>
+    </property>
+    <property name="toolTip">
+     <string>Name of an API key registered via the standalone Key Manager tool ("Create / overwrite api key"). Takes priority over the API key field above and is never stored in the workflow file.</string>
+    </property>
+   </widget>
+   <widget class="QLabel" name="labelCertPath">
+    <property name="geometry">
+     <rect>
+      <x>10</x>
+      <y>210</y>
+      <width>131</width>
+      <height>21</height>
+     </rect>
+    </property>
     <property name="text">
      <string>.pem file :</string>
     </property>
@@ -176,7 +205,7 @@
     <property name="geometry">
      <rect>
       <x>150</x>
-      <y>170</y>
+      <y>206</y>
       <width>410</width>
       <height>29</height>
      </rect>
@@ -186,7 +215,7 @@
     <property name="geometry">
      <rect>
       <x>566</x>
-      <y>170</y>
+      <y>206</y>
       <width>75</width>
       <height>29</height>
      </rect>
@@ -199,7 +228,7 @@
     <property name="geometry">
      <rect>
       <x>150</x>
-      <y>204</y>
+      <y>240</y>
       <width>491</width>
       <height>21</height>
      </rect>
@@ -212,7 +241,7 @@
     <property name="geometry">
      <rect>
       <x>10</x>
-      <y>240</y>
+      <y>276</y>
       <width>131</width>
       <height>21</height>
      </rect>
@@ -228,7 +257,7 @@
     <property name="geometry">
      <rect>
       <x>150</x>
-      <y>236</y>
+      <y>272</y>
       <width>491</width>
       <height>29</height>
      </rect>
@@ -242,7 +271,7 @@
    <property name="geometry">
     <rect>
      <x>20</x>
-     <y>340</y>
+     <y>376</y>
      <width>661</width>
      <height>121</height>
     </rect>
@@ -339,7 +368,7 @@
    <property name="geometry">
     <rect>
      <x>20</x>
-     <y>472</y>
+     <y>508</y>
      <width>661</width>
      <height>31</height>
     </rect>
@@ -352,7 +381,7 @@
    <property name="geometry">
     <rect>
      <x>20</x>
-     <y>512</y>
+     <y>548</y>
      <width>661</width>
      <height>151</height>
     </rect>

{io4it-4.0.26 → io4it-4.0.27}/orangecontrib/IO4IT/widgets/designer/owfile2images.ui

@@ -7,19 +7,19 @@
     <x>0</x>
     <y>0</y>
     <width>470</width>
-    <height>520</height>
+    <height>550</height>
    </rect>
   </property>
   <property name="minimumSize">
    <size>
     <width>470</width>
-    <height>520</height>
+    <height>550</height>
    </size>
   </property>
   <property name="maximumSize">
    <size>
     <width>470</width>
-    <height>520</height>
+    <height>550</height>
    </size>
   </property>
   <property name="windowTitle">
@@ -197,7 +197,7 @@ Input must contain a column with the absolute path to the .pptx / .pdf files.</s
      <x>160</x>
      <y>475</y>
      <width>281</width>
-     <height>36</height>
+     <height>32</height>
     </rect>
    </property>
    <property name="styleSheet">

{io4it-4.0.26 → io4it-4.0.27}/setup.py

@@ -2,7 +2,7 @@ from setuptools import setup, find_packages
 
 # Configuration
 NAME = "io4it"
-VERSION = "4.0.26"
+VERSION = "4.0.27"
 
 INSTALL_REQUIRES = [
     "torchvision",