support custom Azure model names

setup.py

@@ -47,7 +47,7 @@
 # https://packaging.python.org/en/latest/guides/distributing-packages-using-setuptools/
 setup(
     name=package,
-    version="0.2.15",
+    version="0.2.16",
     python_requires=">=3.8, <3.13",
     description=f"UniqueBible App is a cross-platform & offline bible application, integrated with high-quality resources and unique features. Developers: Eliran Wong and Oliver Tseng",
     long_description=long_description,

uniquebible/__init__.py

@@ -135,7 +135,7 @@ def isServerAlive(ip, port):
 # check latest version of azure api at https://learn.microsoft.com/en-us/azure/ai-services/openai/reference
 config.azure_api_version = "2024-10-21"
 
-def is_CJK(self, text):
+def is_CJK(text):
     for char in text:
         if 'CJK' in unicodedata.name(char):
             return True
@@ -214,6 +214,27 @@ def getMistralApi_key() -> str:
     else:
         return ""
 
+def getOpenAIClient():
+    # priority in order: azure > github > openai
+    if config.azureApi_key:
+        return AzureOpenAI(azure_endpoint=re.sub("/models[/]*$", "", config.azureBaseUrl),api_version=config.azure_api_version,api_key=config.azureApi_key)
+    if config.githubApi_key:
+        return OpenAI(api_key=getGithubApi_key(),base_url="https://models.inference.ai.azure.com")
+    return OpenAI()
+
+def extract_text(filepath):
+    try:
+        from markitdown import MarkItDown
+        filepath = filepath.rstrip()
+        if os.path.isfile(filepath):
+            if re.search("(\.jpg|\.jpeg|\.png)$", filepath.lower()):
+                md = MarkItDown(llm_client=getOpenAIClient(), llm_model="gpt-4o")
+            else:
+                md = MarkItDown()
+            return md.convert(filepath)
+    except:
+        return "Install markitdown first!"
+
 def getChatResponse(backend, chatMessages) -> Optional[str]:
     if not isLLMReady(backend) or not backend in config.llm_backends:
         return None

uniquebible/gui/WorkSpace.py

@@ -1,4 +1,4 @@
-from uniquebible import config
+from uniquebible import config, extract_text
 import re, os, base64, glob, webbrowser, markdown
 from datetime import datetime
 import uniquebible.shortcut as sc
@@ -197,11 +197,10 @@ def extractTextFromDocument(self, editable=False):
                 html = config.mainWindow.htmlWrapper(html, True, html=False if fileName.lower().endswith(".md") else True)
             else:
                 try:
-                    import textract
-                    html = textract.process(fileName).decode()
+                    html = extract_text(fileName)
                     html = config.mainWindow.htmlWrapper(html, True, html=False)
                 except:
-                    self.parent.displayMessage("Optional package 'textract' is not installed!")
+                    self.parent.displayMessage("Optional package 'markitdown' is not installed!")
             self.addHtmlContent(html, editable, os.path.basename(fileName))
 
     def fixNoteFont(self, note):

uniquebible/latest_changes.txt

@@ -1,5 +1,11 @@
 PIP package:
 
+0.2.16
+
+* replace textextract package with markitdown
+
+* support custom model names deployed via Azure service
+
 0.2.7-0.2.11
 
 * added support of using FREE Github API key

uniquebible/plugins/menu/Bible Chat.py

@@ -78,12 +78,18 @@ def __init__(self, parent=None):
         self.apiModelBox = QComboBox()
         initialIndex = 0
         index = 0
-        if config.llm_backend in ("openai", "github", "azure"):
+        if config.llm_backend in ("openai", "github"):
             for key in ("gpt-4o", "gpt-4o-mini"):
                 self.apiModelBox.addItem(key)
                 if key == config.openaiApi_chat_model:
                     initialIndex = index
                 index += 1
+        elif config.llm_backend == "azure":
+            for key in config.azureOpenAIModels: # users can manually change config.azureOpenAIModels to match custom deployed model names
+                self.apiModelBox.addItem(key)
+                if key == config.openaiApi_chat_model:
+                    initialIndex = index
+                index += 1
         elif config.llm_backend == "google":
             for key in ("gemini-2.0-flash-exp", "gemini-1.5-flash", "gemini-1.5-flash-8b", "gemini-1.5-pro"):
                 self.apiModelBox.addItem(key)

uniquebible/util/ConfigUtil.py

@@ -332,6 +332,9 @@ def updateModules(module, isInstalled):
         setConfig("azureApi_key", """
         # Azure API Key""",
         "")
+        setConfig("azureOpenAIModels", """
+        # users can manually change config.azureOpenAIModels to match custom model names deployed via Azure service""",
+        ["gpt-4o", "gpt-4o-mini"])
         setConfig("azureBaseUrl", """
         # Github API inference endpoint""",
         "")

uniquebible/util/LocalCliHandler.py

@@ -1,6 +1,6 @@
 # flake8: noqa
 import re, pprint, os, requests, platform, pydoc, markdown, sys, subprocess, json, shutil, webbrowser, traceback, textwrap, wcwidth, unicodedata
-from uniquebible import config
+from uniquebible import config, extract_text
 import threading, time
 #from duckduckgo_search import ddg
 from functools import partial
@@ -2228,20 +2228,19 @@ def downloadyoutube(self):
         return ""
 
     def opentext(self, editMode=False):
-        if ("Textract" in config.enabled):
+        if ("Markitdown" in config.enabled):
             self.print(self.divider)
             userInput = self.getPath.getFilePath(check_isfile=True, empty_to_cancel=True)
             if userInput:
-                import textract
-                content = textract.process(userInput).decode()
+                content = extract_text(userInput)
                 if editMode:
                     self.multilineEditor(content)
                     return ""
                 else:
                     return content
             else:
                 return self.cancelAction()
-        self.printToolNotFound("textract")
+        self.printToolNotFound("markitdown")
         return ""
 
     def printToolNotFound(self, tool):

uniquebible/util/checkup.py

@@ -238,9 +238,9 @@ def isTranslateInstalled():
     except:
         return False
 
-def isTextractInstalled():
+def isMarkitdownInstalled():
     try:
-        import textract
+        from markitdown import MarkItDown
         return True
     except:
         return False
@@ -743,7 +743,7 @@ def runTerminalMode():
     ("word-forms", "Generate English Word Forms", isWordformsInstalled),
     ("lemmagen3", "Lemmatizer", isLemmagen3Installed),
     ("chinese-english-lookup", "Chinese-to-English word definition", isChineseEnglishLookupInstalled),
-    #("textract", "Extract text from document", isTextractInstalled),
+    ("markitdown", "Extract text from document", isMarkitdownInstalled),
     ("tabulate", "Pretty-print tabular data", isTabulateInstalled),
     #("apsw", "Another Python SQLite Wrapper", isApswInstalled),
     ("pyluach", "Hebrew (Jewish) calendar dates", isPyluachInstalled),
@@ -799,7 +799,7 @@ def runTerminalMode():
     ("word-forms", "Generate English Word Forms", isWordformsInstalled),
     ("lemmagen3", "Lemmatizer", isLemmagen3Installed),
     ("chinese-english-lookup", "Chinese-to-English word definition", isChineseEnglishLookupInstalled),
-    #("textract", "Extract text from document", isTextractInstalled),
+    ("markitdown", "Extract text from document", isMarkitdownInstalled),
     ("tabulate", "Pretty-print tabular data", isTabulateInstalled),
     #("apsw", "Another Python SQLite Wrapper", isApswInstalled),
     ("pyluach", "Hebrew (Jewish) calendar dates", isPyluachInstalled),

uniquebible/util/terminal_text_editor.py

@@ -1,4 +1,4 @@
-from uniquebible import config
+from uniquebible import config, extract_text
 import re, os
 from uniquebible.util.TextUtil import TextUtil
 from prompt_toolkit.formatted_text import HTML
@@ -685,8 +685,7 @@ def extractFile(self, filepath="", getTextOnly=False):
 
     def extractFileText(self, filepath):
         if os.path.isfile(filepath):
-            import textract
-            text = text = textract.process(filepath).decode()
+            text = extract_text(filepath)
             # to prevent corrupting original file, users need to specify a file path for saving
             #self.filepath = filepath
             self.savedText = text

uniquebible/util/text_editor_checkup.py

@@ -72,9 +72,9 @@ def isTranslateInstalled():
         return False
 
 # optional
-def isTextractInstalled():
+def isMarkitdownInstalled():
     try:
-        import textract
+        from markitdown import MarkItDown
         return True
     except:
         return False
@@ -234,7 +234,7 @@ def updateModules(module, isInstalled):
     ("pyperclip", "Cross-platform clipboard utilities", isPyperclipInstalled),
     ("Pygments", "Syntax highlighting package", isPygmentsInstalled),
     ("translate", "Google Translate", isTranslateInstalled),
-    #("textract", "Extract text from document", isTextractInstalled),
+    ("markitdown", "Extract text from document", isMarkitdownInstalled),
 ]
 for module, feature, isInstalled in optional:
     checkModule = re.sub("-|_", "", module)