dwani 0.1.7__py3-none-any.whl → 0.1.9__py3-none-any.whl

dwani/__init__.py

@@ -1,4 +1,4 @@
-from .client import DhwaniClient
+from .client import DwaniClient
 from .chat import Chat
 from .audio import Audio
 from .vision import Vision
@@ -7,59 +7,61 @@ from .translate import Translate
 from .exceptions import DhwaniAPIError
 from .docs import Documents
 
-__all__ = ["DhwaniClient", "Chat", "Audio", "Vision", "ASR", "DhwaniAPIError", "Translate", "Documents"]
+__all__ = ["DwaniClient", "Chat", "Audio", "Vision", "ASR", "DhwaniAPIError", "Translate", "Documents"]
 
 # Optionally, instantiate a default client for convenience
 api_key = None
-api_base = "http://localhost:7860"
+api_base = "http://0.0.0.0:8000"
 
 def _get_client():
     global _client
     if "_client" not in globals() or _client is None:
-        from .client import DhwaniClient
-        globals()["_client"] = DhwaniClient(api_key=api_key, api_base=api_base)
-    return globals()["_client"]
+        from .client import DwaniClient
+        globals()["_client"] = DwaniClient(api_key=api_key, api_base=api_base)
+    return _client
 
 class chat:
     @staticmethod
-    def create(prompt, **kwargs):
-        return _get_client().chat(prompt, **kwargs)
+    def create(prompt, src_lang, tgt_lang, model="gemma3"):
+        return _get_client().chat(prompt, src_lang, tgt_lang, model)
 
 class audio:
     @staticmethod
-    def speech(*args, **kwargs):
-        return _get_client().speech(*args, **kwargs)
+    def speech(input, response_format="wav"):
+        return _get_client().speech(input, response_format)
 
 class vision:
     @staticmethod
-    def caption(*args, **kwargs):
-        return _get_client().caption(*args, **kwargs)
+    def caption(file_path, query="describe the image", src_lang="eng_Latn", tgt_lang="kan_Knda", model="gemma3"):
+        return _get_client().caption(file_path, query, src_lang, tgt_lang, model)
 
 class asr:
     @staticmethod
-    def transcribe(*args, **kwargs):
-        return _get_client().transcribe(*args, **kwargs)
-
+    def transcribe(file_path, language="kannada"):
+        return _get_client().transcribe(file_path, language)
 
 class translate:
     @staticmethod
-    def run_translate(*args, **kwargs):
-        return _get_client().translate(*args, **kwargs)
-    
+    def run_translate(sentences, src_lang="kan_Knda", tgt_lang="eng_Latn"):
+        return _get_client().translate(sentences, src_lang, tgt_lang)
 
 class document:
     @staticmethod
-    def run_ocr(*args, **kwargs):
-        return _get_client().ocr(*args, **kwargs)
+    def run_ocr(file_path, language="eng_Latn", model="gemma3"):
+        return _get_client().document_ocr(file_path, language, model)
+    
     @staticmethod
-    def run_summarize(*args, **kwargs):
-        return _get_client().summarize(*args, **kwargs)
+    def run_summarize(file_path, page_number=1, src_lang="eng_Latn", tgt_lang="kan_Knda", model="gemma3"):
+        return _get_client().document_summarize(file_path, page_number, src_lang, tgt_lang, model)
+    
     @staticmethod
-    def run_extract(*args, **kwargs):
-        return _get_client().extract(*args, **kwargs)
+    def run_extract(file_path, page_number=1, src_lang="eng_Latn", tgt_lang="kan_Knda", model="gemma3"):
+        return _get_client().extract(file_path, page_number, src_lang, tgt_lang, model)
+    
     @staticmethod
-    def run_doc_query(*args, **kwargs):
-        return _get_client().doc_query(*args, **kwargs)
+    def run_doc_query(file_path, page_number=1, prompt="list the key points", src_lang="eng_Latn", tgt_lang="kan_Knda", model="gemma3"):
+        return _get_client().doc_query(file_path, page_number, prompt, src_lang, tgt_lang, model)
+    
     @staticmethod
-    def run_doc_query_kannada(*args, **kwargs):
-        return _get_client().doc_query_kannada(*args, **kwargs)
+    def run_doc_query_kannada(file_path, page_number=1, prompt="list key points", src_lang="eng_Latn", tgt_lang="kan_Knda", model="gemma3"):
+        return _get_client().doc_query_kannada(file_path, page_number, prompt, src_lang, tgt_lang, model)

dwani/chat.py

@@ -7,14 +7,14 @@ language_options = [
     ("Kannada", "kan_Knda"),
     ("Hindi", "hin_Deva"), 
     ("Assamese", "asm_Beng"),
-    ("Bengali","ben_Beng"),
-    ("Gujarati","guj_Gujr"),
-    ("Malayalam","mal_Mlym"),
-    ("Marathi","mar_Deva"),
-    ("Odia","ory_Orya"),
-    ("Punjabi","pan_Guru"),
-    ("Tamil","tam_Taml"),
-    ("Telugu","tel_Telu") 
+    ("Bengali", "ben_Beng"),
+    ("Gujarati", "guj_Gujr"),
+    ("Malayalam", "mal_Mlym"),
+    ("Marathi", "mar_Deva"),
+    ("Odia", "ory_Orya"),
+    ("Punjabi", "pan_Guru"),
+    ("Tamil", "tam_Taml"),
+    ("Telugu", "tel_Telu") 
 ]
 
 # Create a dictionary for language name to code mapping
@@ -35,7 +35,12 @@ def normalize_language(lang):
     supported_langs = list(lang_name_to_code.keys()) + list(lang_code_to_code.keys())
     raise ValueError(f"Unsupported language: {lang}. Supported languages: {supported_langs}")
 
-def chat_create(client, prompt, src_lang, tgt_lang, **kwargs):
+def chat_create(client, prompt, src_lang, tgt_lang, model="gemma3"):
+    # Validate model
+    valid_models = ["gemma3", "qwen3", "deepseek-r1"]
+    if model not in valid_models:
+        raise ValueError(f"Unsupported model: {model}. Supported models: {valid_models}")
+    
     # Normalize source and target languages
     src_lang_code = normalize_language(src_lang)
     tgt_lang_code = normalize_language(tgt_lang)
@@ -44,9 +49,9 @@ def chat_create(client, prompt, src_lang, tgt_lang, **kwargs):
     payload = {
         "prompt": prompt,
         "src_lang": src_lang_code,
-        "tgt_lang": tgt_lang_code
+        "tgt_lang": tgt_lang_code,
+        "model": model
     }
-    payload.update(kwargs)
     resp = requests.post(
         url,
         headers={**client._headers(), "Content-Type": "application/json"},
@@ -58,6 +63,6 @@ def chat_create(client, prompt, src_lang, tgt_lang, **kwargs):
 
 class Chat:
     @staticmethod
-    def create(prompt, src_lang, tgt_lang, **kwargs):
+    def create(prompt, src_lang, tgt_lang, model="gemma3"):
         from . import _get_client
-        return _get_client().chat(prompt, src_lang, tgt_lang, **kwargs)
+        return _get_client().chat(prompt, src_lang, tgt_lang, model)

dwani/client.py

@@ -2,53 +2,55 @@ import os
 import requests
 from .exceptions import DhwaniAPIError
 
-class DhwaniClient:
+class DwaniClient:
     def __init__(self, api_key=None, api_base=None):
         self.api_key = api_key or os.getenv("DWANI_API_KEY")
-        self.api_base = api_base or os.getenv("DWANI_API_BASE_URL", "http://localhost:8000")
+        self.api_base = api_base or os.getenv("DWANI_API_BASE_URL", "http://0.0.0.0:8000")
         if not self.api_key:
-            raise ValueError("DHWANI_API_KEY not set")
+            raise ValueError("DWANI_API_KEY not set")
 
     def _headers(self):
-        return {"X-API-Key": self.api_key}
+        return {
+            "X-API-Key": self.api_key,
+            "Accept": "application/json"
+        }
 
-    def translate(self, sentences, src_lang, tgt_lang, **kwargs):
+    def translate(self, sentences, src_lang, tgt_lang):
         from .translate import run_translate
-        return run_translate(self, sentences=sentences, src_lang=src_lang, tgt_lang=tgt_lang, **kwargs)
+        return run_translate(self, sentences=sentences, src_lang=src_lang, tgt_lang=tgt_lang)
 
-    def chat(self, prompt, src_lang, tgt_lang, **kwargs):
+    def chat(self, prompt, src_lang, tgt_lang, model="gemma3"):
         from .chat import chat_create
-        return chat_create(self, prompt=prompt, src_lang=src_lang, tgt_lang=tgt_lang, **kwargs)
+        return chat_create(self, prompt=prompt, src_lang=src_lang, tgt_lang=tgt_lang, model=model)
     
-    def speech(self, input, response_format="mp3", **kwargs):
+    def speech(self, input, response_format="mp3"):
         from .audio import audio_speech
-        return audio_speech(self, input=input, response_format=response_format, **kwargs)
+        return audio_speech(self, input=input, response_format=response_format)
 
-    def caption(self, file_path, query="describe the image", src_lang="eng_Latn", tgt_lang="kan_Knda", **kwargs):
+    def caption(self, file_path, query="describe the image", src_lang="eng_Latn", tgt_lang="kan_Knda", model="gemma3"):
         from .vision import vision_caption
-        return vision_caption(self, file_path=file_path, query=query, src_lang=src_lang, tgt_lang=tgt_lang, **kwargs)
+        return vision_caption(self, file_path=file_path, query=query, src_lang=src_lang, tgt_lang=tgt_lang, model=model)
 
-    def transcribe(self, file_path, language=None, **kwargs):
+    def transcribe(self, file_path, language=None):
         from .asr import asr_transcribe
-        return asr_transcribe(self, file_path=file_path, language=language, **kwargs)
+        return asr_transcribe(self, file_path=file_path, language=language)
     
-    def document_ocr(self, file_path, language=None, **kwargs):
+    def document_ocr(self, file_path, language=None, model="gemma3"):
         from .docs import document_ocr
-        return document_ocr(self, file_path=file_path, language=language, **kwargs)
+        return document_ocr(self, file_path=file_path, language=language, model=model)
 
-    def document_summarize(self, file_path, page_number=1, src_lang="eng_Latn", tgt_lang="kan_Knda", **kwargs):
+    def document_summarize(self, file_path, page_number=1, src_lang="eng_Latn", tgt_lang="kan_Knda", model="gemma3"):
         from .docs import document_summarize
-        return document_summarize(self, file_path, page_number, src_lang, tgt_lang, **kwargs)
+        return document_summarize(self, file_path, page_number, src_lang, tgt_lang, model)
 
-    def extract(self, file_path, page_number=1, src_lang="eng_Latn", tgt_lang="kan_Knda", **kwargs):
+    def extract(self, file_path, page_number=1, src_lang="eng_Latn", tgt_lang="kan_Knda", model="gemma3"):
         from .docs import extract
-        return extract(self, file_path=file_path, page_number=page_number, src_lang=src_lang,tgt_lang=tgt_lang, **kwargs)
+        return extract(self, file_path=file_path, page_number=page_number, src_lang=src_lang, tgt_lang=tgt_lang, model=model)
 
-
-    def doc_query( self, file_path, page_number=1, prompt="list the key points", src_lang="eng_Latn", tgt_lang="kan_Knda" , **kwargs ):
+    def doc_query(self, file_path, page_number=1, prompt="list the key points", src_lang="eng_Latn", tgt_lang="kan_Knda", model="gemma3"):
         from .docs import doc_query
-        return doc_query( self, file_path, page_number=page_number, prompt=prompt, src_lang=src_lang, tgt_lang=tgt_lang , **kwargs )
+        return doc_query(self, file_path, page_number=page_number, prompt=prompt, src_lang=src_lang, tgt_lang=tgt_lang, model=model)
 
-    def doc_query_kannada(self, file_path, page_number=1, prompt="list key points", src_lang="eng_Latn", language=None, **kwargs):
+    def doc_query_kannada(self, file_path, page_number=1, prompt="list key points", src_lang="eng_Latn", language="kan_Knda", model="gemma3"):
         from .docs import doc_query_kannada
-        return doc_query_kannada(self, file_path=file_path, page_number=page_number, prompt=prompt, src_lang=src_lang, language=language, **kwargs)
+        return doc_query_kannada(self, file_path=file_path, page_number=page_number, prompt=prompt, src_lang=src_lang, language=language, model=model)

dwani/docs.py

@@ -1,62 +1,81 @@
 import requests
 from .exceptions import DhwaniAPIError
+import logging
 
-# Language options mapping
+# Set up logging
+logger = logging.getLogger(__name__)
+
+# Language options mapping (aligned with server’s SUPPORTED_LANGUAGES)
 language_options = [
     ("English", "eng_Latn"),
     ("Kannada", "kan_Knda"),
-    ("Hindi", "hin_Deva"), 
-    ("Assamese", "asm_Beng"),
-    ("Bengali", "ben_Beng"),
-    ("Gujarati", "guj_Gujr"),
-    ("Malayalam", "mal_Mlym"),
-    ("Marathi", "mar_Deva"),
-    ("Odia", "ory_Orya"),
-    ("Punjabi", "pan_Guru"),
+    ("Hindi", "hin_Deva"),
     ("Tamil", "tam_Taml"),
-    ("Telugu", "tel_Telu") 
+    ("Telugu", "tel_Telu")
 ]
 
 # Create dictionaries for language name to code and code to code mapping
 lang_name_to_code = {name.lower(): code for name, code in language_options}
 lang_code_to_code = {code: code for _, code in language_options}
 
+# Supported models (aligned with server)
+VALID_MODELS = ["gemma3", "moondream", "qwen2.5vl", "qwen3", "sarvam-m", "deepseek-r1"]
+
 def normalize_language(lang):
     """Convert language input (name or code) to language code."""
     lang = lang.strip()
-    # Check if input is a language name (case-insensitive)
     lang_lower = lang.lower()
     if lang_lower in lang_name_to_code:
         return lang_name_to_code[lang_lower]
-    # Check if input is a language code
     if lang in lang_code_to_code:
         return lang_code_to_code[lang]
-    # Raise error if language is not supported
     supported_langs = list(lang_name_to_code.keys()) + list(lang_code_to_code.keys())
     raise ValueError(f"Unsupported language: {lang}. Supported languages: {supported_langs}")
 
-def document_ocr(client, file_path, language=None):
+def validate_model(model):
+    """Validate the model against supported models."""
+    if model not in VALID_MODELS:
+        raise ValueError(f"Unsupported model: {model}. Supported models: {VALID_MODELS}")
+    return model
+
+def document_ocr(client, file_path, language=None, model="gemma3"):
     """OCR a document (image/PDF) and return extracted text."""
-    data = {}
+    logger.debug(f"Calling document_ocr: file_path={file_path}, language={language}, model={model}")
+    validate_model(model)
+    
+    data = {"model": model}
     if language:
-        # Normalize the language input
         data["language"] = normalize_language(language)
     
     with open(file_path, "rb") as f:
-        files = {"file": f}
-        resp = requests.post(
-            f"{client.api_base}/v1/document/ocr",
-            headers=client._headers(),
-            files=files,
-            data=data
-        )
-    if resp.status_code != 200:
-        raise DhwaniAPIError(resp)
+        mime_type = "application/pdf" if file_path.lower().endswith('.pdf') else "image/png"
+        files = {"file": (file_path, f, mime_type)}
+        try:
+            resp = requests.post(
+                f"{client.api_base}/v1/document/ocr",
+                headers=client._headers(),
+                files=files,
+                data=data,
+                timeout=60
+            )
+            resp.raise_for_status()
+        except requests.RequestException as e:
+            logger.error(f"OCR request failed: {str(e)}")
+            raise DhwaniAPIError(resp) if 'resp' in locals() else DhwaniAPIError.from_exception(e)
+    
+    logger.debug(f"OCR response: {resp.status_code}")
     return resp.json()
 
-def document_summarize(client, file_path, page_number=1, src_lang="eng_Latn", tgt_lang="kan_Knda"):
+def document_summarize(client, file_path, page_number=1, src_lang="eng_Latn", tgt_lang="kan_Knda", model="gemma3"):
     """Summarize a PDF document with language and page number options."""
-    # Normalize source and target languages
+    logger.debug(f"Calling document_summarize: file_path={file_path}, page_number={page_number}, src_lang={src_lang}, tgt_lang={tgt_lang}, model={model}")
+    validate_model(model)
+    
+    if not file_path.lower().endswith('.pdf'):
+        raise ValueError("File must be a PDF")
+    if page_number < 1:
+        raise ValueError("Page number must be at least 1")
+    
     src_lang_code = normalize_language(src_lang)
     tgt_lang_code = normalize_language(tgt_lang)
     
@@ -67,41 +86,62 @@ def document_summarize(client, file_path, page_number=1, src_lang="eng_Latn", tg
         data = {
             "page_number": str(page_number),
             "src_lang": src_lang_code,
-            "tgt_lang": tgt_lang_code
+            "tgt_lang": tgt_lang_code,
+            "model": model
         }
-        resp = requests.post(
-            url,
-            headers=headers,
-            files=files,
-            data=data
-        )
-    if resp.status_code != 200:
-        raise DhwaniAPIError(resp)
+        try:
+            resp = requests.post(
+                url,
+                headers=headers,
+                files=files,
+                data=data,
+                timeout=60
+            )
+            resp.raise_for_status()
+        except requests.RequestException as e:
+            logger.error(f"Summarize request failed: {str(e)}")
+            raise DhwaniAPIError(resp) if 'resp' in locals() else DhwaniAPIError.from_exception(e)
+    
+    logger.debug(f"Summarize response: {resp.status_code}")
     return resp.json()
 
-def extract(client, file_path, page_number, src_lang, tgt_lang):
-    """
-    Extract and translate text from a document (image/PDF) using query parameters.
-    """
-    # Normalize source and target languages
+def extract(client, file_path, page_number=1, src_lang="eng_Latn", tgt_lang="kan_Knda", model="gemma3"):
+    """Extract and translate text from a PDF document using form data."""
+    logger.debug(f"Calling extract: file_path={file_path}, page_number={page_number}, src_lang={src_lang}, tgt_lang={tgt_lang}, model={model}")
+    validate_model(model)
+    
+    if not file_path.lower().endswith('.pdf'):
+        raise ValueError("File must be a PDF")
+    if page_number < 1:
+        raise ValueError("Page number must be at least 1")
+    
     src_lang_code = normalize_language(src_lang)
     tgt_lang_code = normalize_language(tgt_lang)
     
-    # Build the URL with query parameters
-    url = (
-        f"{client.api_base}/v1/indic-extract-text/"
-        f"?page_number={page_number}&src_lang={src_lang_code}&tgt_lang={tgt_lang_code}"
-    )
+    url = f"{client.api_base}/v1/indic-extract-text/"
     headers = client._headers()
     with open(file_path, "rb") as f:
         files = {"file": (file_path, f, "application/pdf")}
-        resp = requests.post(
-            url,
-            headers=headers,
-            files=files
-        )
-    if resp.status_code != 200:
-        raise DhwaniAPIError(resp)
+        data = {
+            "page_number": str(page_number),
+            "src_lang": src_lang_code,
+            "tgt_lang": tgt_lang_code,
+            "model": model
+        }
+        try:
+            resp = requests.post(
+                url,
+                headers=headers,
+                files=files,
+                data=data,
+                timeout=60
+            )
+            resp.raise_for_status()
+        except requests.RequestException as e:
+            logger.error(f"Extract request failed: {str(e)}")
+            raise DhwaniAPIError(resp) if 'resp' in locals() else DhwaniAPIError.from_exception(e)
+    
+    logger.debug(f"Extract response: {resp.status_code}")
     return resp.json()
 
 def doc_query(
@@ -110,10 +150,20 @@ def doc_query(
     page_number=1,
     prompt="list the key points",
     src_lang="eng_Latn",
-    tgt_lang="kan_Knda"
+    tgt_lang="kan_Knda",
+    model="gemma3"
 ):
     """Query a document with a custom prompt and language options."""
-    # Normalize source and target languages
+    logger.debug(f"Calling doc_query: file_path={file_path}, page_number={page_number}, prompt={prompt}, src_lang={src_lang}, tgt_lang={tgt_lang}, model={model}")
+    validate_model(model)
+    
+    if not file_path.lower().endswith('.pdf'):
+        raise ValueError("File must be a PDF")
+    if page_number < 1:
+        raise ValueError("Page number must be at least 1")
+    if not prompt.strip():
+        raise ValueError("Prompt cannot be empty")
+    
     src_lang_code = normalize_language(src_lang)
     tgt_lang_code = normalize_language(tgt_lang)
     
@@ -124,74 +174,103 @@ def doc_query(
         data = {
             "page_number": str(page_number),
             "prompt": prompt,
-            "source_language": src_lang_code,
-            "target_language": tgt_lang_code
+            "src_lang": src_lang_code,
+            "tgt_lang": tgt_lang_code,
+            "model": model
         }
-        resp = requests.post(
-            url,
-            headers=headers,
-            files=files,
-            data=data
-        )
-    if resp.status_code != 200:
-        raise DhwaniAPIError(resp)
+        try:
+            resp = requests.post(
+                url,
+                headers=headers,
+                files=files,
+                data=data,
+                timeout=60
+            )
+            resp.raise_for_status()
+        except requests.RequestException as e:
+            logger.error(f"Doc query request failed: {str(e)}")
+            raise DhwaniAPIError(resp) if 'resp' in locals() else DhwaniAPIError.from_exception(e)
+    
+    logger.debug(f"Doc query response: {resp.status_code}")
     return resp.json()
 
 def doc_query_kannada(
-    client, 
-    file_path, 
-    page_number=1, 
-    prompt="list key points", 
+    client,
+    file_path,
+    page_number=1,
+    prompt="list key points",
     src_lang="eng_Latn",
-    language=None
+    tgt_lang="kan_Knda",
+    model="gemma3"
 ):
-    """Summarize a document (image/PDF/text) with custom prompt and language."""
-    # Normalize source language and optional language parameter
+    """Query a document with a custom prompt, outputting in Kannada."""
+    logger.debug(f"Calling doc_query_kannada: file_path={file_path}, page_number={page_number}, prompt={prompt}, src_lang={src_lang}, tgt_lang={tgt_lang}, model={model}")
+    validate_model(model)
+    
+    if not file_path.lower().endswith('.pdf'):
+        raise ValueError("File must be a PDF")
+    if page_number < 1:
+        raise ValueError("Page number must be at least 1")
+    if not prompt.strip():
+        raise ValueError("Prompt cannot be empty")
+    
     src_lang_code = normalize_language(src_lang)
-    data = {
-        "page_number": str(page_number),
-        "prompt": prompt,
-        "src_lang": src_lang_code,
-    }
-    if language:
-        data["language"] = normalize_language(language)
+    tgt_lang_code = normalize_language(tgt_lang) if tgt_lang else "kan_Knda"
     
-    url = f"{client.api_base}/v1/indic-custom-prompt-kannada-pdf"
+    url = f"{client.api_base}/v1/indic-custom-prompt-pdf"
     headers = client._headers()
     with open(file_path, "rb") as f:
         files = {"file": (file_path, f, "application/pdf")}
-        resp = requests.post(
-            url,
-            headers=headers,
-            files=files,
-            data=data
-        )
-    if resp.status_code != 200:
-        raise DhwaniAPIError(resp)
+        data = {
+            "page_number": str(page_number),
+            "prompt": prompt,
+            "src_lang": src_lang_code,
+            "tgt_lang": tgt_lang_code,
+            "model": model
+        }
+        try:
+            resp = requests.post(
+                url,
+                headers=headers,
+                files=files,
+                data=data,
+                timeout=60
+            )
+            resp.raise_for_status()
+        except requests.RequestException as e:
+            logger.error(f"Doc query Kannada request failed: {str(e)}")
+            raise DhwaniAPIError(resp) if 'resp' in locals() else DhwaniAPIError.from_exception(e)
+    
+    logger.debug(f"Doc query Kannada response: {resp.status_code}")
     return resp.json()
 
 class Documents:
     @staticmethod
-    def ocr(file_path, language=None):
-        from . import _get_client
-        return _get_client().document_ocr(file_path, language)
-
+    def ocr(file_path, language=None, model="gemma3"):
+        from .client import DwaniClient
+        client = DwaniClient()
+        return document_ocr(client, file_path, language, model)
+    
     @staticmethod
-    def summarize(*args, **kwargs):
-        from . import _get_client
-        return _get_client().document_summarize(*args, **kwargs)
+    def summarize(file_path, page_number=1, src_lang="eng_Latn", tgt_lang="kan_Knda", model="gemma3"):
+        from .client import DwaniClient
+        client = DwaniClient()
+        return document_summarize(client, file_path, page_number, src_lang, tgt_lang, model)
     
     @staticmethod
-    def run_extract(*args, **kwargs):
-        from . import _get_client
-        return _get_client().extract(*args, **kwargs)
+    def run_extract(file_path, page_number=1, src_lang="eng_Latn", tgt_lang="kan_Knda", model="gemma3"):
+        from .client import DwaniClient
+        client = DwaniClient()
+        return extract(client, file_path, page_number, src_lang, tgt_lang, model)
     
     @staticmethod
-    def run_doc_query(*args, **kwargs):
-        from . import _get_client
-        return _get_client().doc_query(*args, **kwargs)
+    def run_doc_query(file_path, page_number=1, prompt="list the key points", src_lang="eng_Latn", tgt_lang="kan_Knda", model="gemma3"):
+        from .client import DwaniClient
+        client = DwaniClient()
+        return doc_query(client, file_path, page_number, prompt, src_lang, tgt_lang, model)
     
     @staticmethod
-    def run_doc_query_kannada(*args, **kwargs):
-        from . import _get_client
-        return _get_client().doc_query_kannada(*args, **kwargs)
+    def run_doc_query_kannada(file_path, page_number=1, prompt="list key points", src_lang="eng_Latn", tgt_lang="kan_Knda", model="gemma3"):
+        from .client import DwaniClient
+        client = DwaniClient()
+        return doc_query_kannada(client, file_path, page_number, prompt, src_lang, tgt_lang, model)

dwani/translate.py

@@ -35,7 +35,17 @@ def normalize_language(lang):
     supported_langs = list(lang_name_to_code.keys()) + list(lang_code_to_code.keys())
     raise ValueError(f"Unsupported language: {lang}. Supported languages: {supported_langs}")
 
-def run_translate(client, sentences, src_lang, tgt_lang, **kwargs):
+def run_translate(client, sentences, src_lang, tgt_lang):
+    # Convert single string to list if necessary
+    if isinstance(sentences, str):
+        sentences = [sentences]
+    elif not isinstance(sentences, list):
+        raise ValueError("sentences must be a string or a list of strings")
+    
+    # Validate that all elements in the list are strings
+    if not all(isinstance(s, str) for s in sentences):
+        raise ValueError("All sentences must be strings")
+    
     # Normalize source and target languages
     src_lang_code = normalize_language(src_lang)
     tgt_lang_code = normalize_language(tgt_lang)
@@ -46,7 +56,6 @@ def run_translate(client, sentences, src_lang, tgt_lang, **kwargs):
         "src_lang": src_lang_code,
         "tgt_lang": tgt_lang_code
     }
-    payload.update(kwargs)
     resp = requests.post(
         url,
         headers={**client._headers(), "Content-Type": "application/json", "accept": "application/json"},
@@ -58,6 +67,6 @@ def run_translate(client, sentences, src_lang, tgt_lang, **kwargs):
 
 class Translate:
     @staticmethod
-    def run_translate(sentences, src_lang, tgt_lang, **kwargs):
+    def run_translate(sentences, src_lang, tgt_lang):
         from . import _get_client
-        return _get_client().translate(sentences, src_lang, tgt_lang, **kwargs)
+        return _get_client().translate(sentences, src_lang, tgt_lang)

dwani/vision.py

@@ -35,7 +35,12 @@ def normalize_language(lang):
     supported_langs = list(lang_name_to_code.keys()) + list(lang_code_to_code.keys())
     raise ValueError(f"Unsupported language: {lang}. Supported languages: {supported_langs}")
 
-def vision_caption(client, file_path, query="describe the image", src_lang="eng_Latn", tgt_lang="kan_Knda"):
+def vision_caption(client, file_path, query="describe the image", src_lang="eng_Latn", tgt_lang="kan_Knda", model="gemma3"):
+    # Validate model
+    valid_models = ["gemma3", "qwen2.5vl", "moondream"]
+    if model not in valid_models:
+        raise ValueError(f"Unsupported model: {model}. Supported models: {valid_models}")
+    
     # Normalize source and target languages
     src_lang_code = normalize_language(src_lang)
     tgt_lang_code = normalize_language(tgt_lang)
@@ -43,7 +48,7 @@ def vision_caption(client, file_path, query="describe the image", src_lang="eng_
     # Build the endpoint using the client's api_base
     url = (
         f"{client.api_base}/v1/indic_visual_query"
-        f"?src_lang={src_lang_code}&tgt_lang={tgt_lang_code}"
+        f"?src_lang={src_lang_code}&tgt_lang={tgt_lang_code}&model={model}"
     )
     headers = {
         **client._headers(),
@@ -64,6 +69,6 @@ def vision_caption(client, file_path, query="describe the image", src_lang="eng_
 
 class Vision:
     @staticmethod
-    def caption(*args, **kwargs):
+    def caption(file_path, query="describe the image", src_lang="eng_Latn", tgt_lang="kan_Knda", model="gemma3"):
         from . import _get_client
-        return _get_client().caption(*args, **kwargs)
+        return _get_client().caption(file_path, query, src_lang, tgt_lang, model)

{dwani-0.1.7.dist-info → dwani-0.1.9.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: dwani
-Version: 0.1.7
+Version: 0.1.9
 Summary: Multimodal API for Indian languages (Chat, Vision, TTS, ASR, Translate, Docs)
 Author-email: sachin <python@dwani.ai>
 License: MIT License
@@ -39,7 +39,7 @@ Dynamic: license-file
 
 ### Install the library
 ```bash
-pip install dwani
+pip install --upgrade dwani
 ```
 
 ### Languages supported
@@ -55,33 +55,44 @@ dwani.api_key = os.getenv("DWANI_API_KEY")
 dwani.api_base = os.getenv("DWANI_API_BASE_URL")
 ```
 
-### Examples
 
-#### Text Query 
+### Text Query 
+---
+- With model selection
+  - Supported models : gemma3 (default), qwen3
+
+---
+- gemma3
 ```python
-resp = dwani.Chat.create(prompt="Hello!", src_lang="eng_Latn", tgt_lang="kan_Knda")
+resp = dwani.Chat.create(prompt="Hello!", src_lang="english", tgt_lang="kannada", model="gemma3")
 print(resp)
 ```
 ```json
 {'response': 'ನಮಸ್ತೆ! ಭಾರತ ಮತ್ತು ಕರ್ನಾಟಕವನ್ನು ಗಮನದಲ್ಲಿಟ್ಟುಕೊಂಡು ಇಂದು ನಿಮ್ಮ ಪ್ರಶ್ನೆಗಳಿಗೆ ನಾನು ನಿಮಗೆ ಹೇಗೆ ಸಹಾಯ ಮಾಡಲಿ?'}
 ```
+---
+### Vision Query
+---
+- With model selection
+  - Supported models : gemma3 (default), moondream
+- gemma3
 
-
-#### Vision Query
 ```python
 result = dwani.Vision.caption(
     file_path="image.png",
     query="Describe this logo",
-    src_lang="eng_Latn",
-    tgt_lang="kan_Knda"
+    src_lang="english",
+    tgt_lang="kannada",
+    model="gemma3"
 )
 print(result)
 ```
 ```json
 {'answer': 'ಒಂದು ವಾಕ್ಯದಲ್ಲಿ ಚಿತ್ರದ ಸಾರಾಂಶವನ್ನು ಇಲ್ಲಿ ನೀಡಲಾಗಿದೆಃ ಪ್ರಕಟಣೆಯ ಅವಲೋಕನವು ಪ್ರಸ್ತುತ ಅರವತ್ತನಾಲ್ಕು ದೇಶಗಳು/ಪ್ರದೇಶಗಳನ್ನು ಸೇರಿಸಲಾಗಿದೆ ಮತ್ತು ಇನ್ನೂ ಹದಿನಾರು ಪ್ರದೇಶಗಳನ್ನು ಸೇರಿಸಬೇಕಾಗಿದೆ. ಒದಗಿಸಲಾದ ಚಿತ್ರದಲ್ಲಿ ಲಾಂಛನವು ಕಾಣಿಸುವುದಿಲ್ಲ.'}
 ```
-
-#### Speech to Text -  Automatic Speech Recognition (ASR)
+---
+### Speech to Text -  Automatic Speech Recognition (ASR)
+---
 ```python
 result = dwani.ASR.transcribe(file_path="kannada_sample.wav", language="kannada")
 print(result)
@@ -89,26 +100,28 @@ print(result)
 ```json
 {'text': 'ಕರ್ನಾಟಕ ದ ರಾಜಧಾನಿ ಯಾವುದು'}
 ```
-
+---
 ### Translate
+---
 ```python
-resp = dwani.Translate.run_translate(sentences=["hi"], src_lang="eng_Latn", tgt_lang="kan_Knda")
+resp = dwani.Translate.run_translate(sentences="hi, i am gaganyatri", src_lang="english", tgt_lang="kannada")
 print(resp)
 ```
 ```json
 {'translations': ['ಹಾಯ್']}
 ```
-#### Text to Speech -  Speech Synthesis
-
+---
+### Text to Speech -  Speech Synthesis
+---
 ```python
-response = dwani.Audio.speech(input="ಕರ್ನಾಟಕ ದ ರಾಜಧಾನಿ ಯಾವುದು", response_format="mp3")
-with open("output.mp3", "wb") as f:
+response = dwani.Audio.speech(input="ಕರ್ನಾಟಕ ದ ರಾಜಧಾನಿ ಯಾವುದು", response_format="wav")
+with open("output.wav", "wb") as f:
     f.write(response)
 ```
 
 #### Document - Extract Text
 ```python
-result = dwani.Documents.run_extract(file_path = "dwani-workshop.pdf", page_number=1, src_lang="eng_Latn",tgt_lang="kan_Knda" )
+result = dwani.Documents.run_extract(file_path = "dwani-workshop.pdf", page_number=1, src_lang="english",tgt_lang="kannada" )
 print(result)
 ```
 ```json

dwani-0.1.9.dist-info/RECORD

@@ -0,0 +1,14 @@
+dwani/__init__.py,sha256=JcbP7N6J-is-r5g5aDM8OluuCD1V5HxT3TgMtLwcH8s,2665
+dwani/asr.py,sha256=3LYrLOaMhc5eXKFSoi63C8KAvwZI2NcuO25pwTfSVe0,1692
+dwani/audio.py,sha256=Q9vw4uBxGy1vQzmiZjZGrY8hkAEQNkGhjz5OcnpFEQQ,888
+dwani/chat.py,sha256=a6Bd0Skx9Fi4UVCj_-FfUR0wt3y8ep1AV7Q7kEqvpzA,2315
+dwani/client.py,sha256=sDSA1F1Ixh08uaSf4tuzsOm72oEAUi9w3dUiP3fyvUk,2905
+dwani/docs.py,sha256=PBCUHyulcV1AYX7WcX_uKLkYjUQ48zAZ9PK9Rrvhy6s,10571
+dwani/exceptions.py,sha256=qEN5ukqlnN7v-kHNEnISWFMpPMt6uTft9mPsTXJ4LVA,227
+dwani/translate.py,sha256=nYqKX7TDz6hds2Ih-CWXWkS8Bd_4KXVY_NG7erhtS_8,2542
+dwani/vision.py,sha256=rfmcLFPdZC1MLdYAG3aRdCW22-gkXfjqm6WYZJ1Ac2k,2674
+dwani-0.1.9.dist-info/licenses/LICENSE,sha256=IAD8tbwWZbPWHXgYjabHoMv0aaUzZUYzYiEbfhTCisY,1070
+dwani-0.1.9.dist-info/METADATA,sha256=hjS9WvvbnDJ3IZQkkg7PV4sRnbyBJmrxD0kz5Q4TFuc,5045
+dwani-0.1.9.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+dwani-0.1.9.dist-info/top_level.txt,sha256=AM5EhkyuO_EXQFR9JIxEV6tAYMCCyc-a1dLifpCGBUk,6
+dwani-0.1.9.dist-info/RECORD,,

{dwani-0.1.7.dist-info → dwani-0.1.9.dist-info}/WHEEL

@@ -1,5 +1,5 @@
 Wheel-Version: 1.0
-Generator: setuptools (80.7.1)
+Generator: setuptools (80.9.0)
 Root-Is-Purelib: true
 Tag: py3-none-any
 

dwani-0.1.7.dist-info/RECORD

@@ -1,14 +0,0 @@
-dwani/__init__.py,sha256=ldO5OND7DvJlbxaQ0R57Cc73jJTnCSslDDt4I4r-Op8,1895
-dwani/asr.py,sha256=3LYrLOaMhc5eXKFSoi63C8KAvwZI2NcuO25pwTfSVe0,1692
-dwani/audio.py,sha256=Q9vw4uBxGy1vQzmiZjZGrY8hkAEQNkGhjz5OcnpFEQQ,888
-dwani/chat.py,sha256=dQCl8lLQczwnAsvYlTZowd471ktRVZcW3w8gZ5Wpzms,2097
-dwani/client.py,sha256=OrnwqxBQMfEZ1iQEleFigNujiZve3ox53yv5aSmB3iQ,2849
-dwani/docs.py,sha256=EO41opJwfDFsNmH6nQl-HOsyWravCnj1f5ZDgxSZECI,6323
-dwani/exceptions.py,sha256=qEN5ukqlnN7v-kHNEnISWFMpPMt6uTft9mPsTXJ4LVA,227
-dwani/translate.py,sha256=IJiKrYIfwdJKc_PjlZKVRAwzpQDst_2MF_B_huxid_E,2185
-dwani/vision.py,sha256=wN7WkMRVmLrZnBJxnam7vihTXWGlWJ4JqXgyrp-tbrg,2330
-dwani-0.1.7.dist-info/licenses/LICENSE,sha256=IAD8tbwWZbPWHXgYjabHoMv0aaUzZUYzYiEbfhTCisY,1070
-dwani-0.1.7.dist-info/METADATA,sha256=ggOY4wss1nwN6PzhB0BZG5332emrP6C76qwjaK2mzSs,4810
-dwani-0.1.7.dist-info/WHEEL,sha256=Nw36Djuh_5VDukK0H78QzOX-_FQEo6V37m3nkm96gtU,91
-dwani-0.1.7.dist-info/top_level.txt,sha256=AM5EhkyuO_EXQFR9JIxEV6tAYMCCyc-a1dLifpCGBUk,6
-dwani-0.1.7.dist-info/RECORD,,