dwani 0.1.9__py3-none-any.whl → 0.1.11__py3-none-any.whl

dwani/__init__.py

@@ -4,10 +4,10 @@ from .audio import Audio
 from .vision import Vision
 from .asr import ASR
 from .translate import Translate
-from .exceptions import DhwaniAPIError
+from .exceptions import DwaniAPIError
 from .docs import Documents
 
-__all__ = ["DwaniClient", "Chat", "Audio", "Vision", "ASR", "DhwaniAPIError", "Translate", "Documents"]
+__all__ = ["DwaniClient", "Chat", "Audio", "Vision", "ASR", "DwaniAPIError", "Translate", "Documents"]
 
 # Optionally, instantiate a default client for convenience
 api_key = None
@@ -24,6 +24,9 @@ class chat:
     @staticmethod
     def create(prompt, src_lang, tgt_lang, model="gemma3"):
         return _get_client().chat(prompt, src_lang, tgt_lang, model)
+    @staticmethod
+    def direct(prompt, model="gemma3", system_prompt =""):
+        return _get_client().chat_direct(prompt, model, system_prompt)
 
 class audio:
     @staticmethod
@@ -34,6 +37,9 @@ class vision:
     @staticmethod
     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)
+    @staticmethod
+    def caption_direct(file_path, query="describe the image", model="gemma3", system_prompt=""):
+        return _get_client().caption_direct(file_path, query, model, system_prompt)
 
 class asr:
     @staticmethod

dwani/asr.py

@@ -1,4 +1,4 @@
-from .exceptions import DhwaniAPIError
+from .exceptions import DwaniAPIError
 import requests
 
 # Allowed languages (case-sensitive for display, but we'll handle case-insensitively)
@@ -13,7 +13,9 @@ ALLOWED_LANGUAGES = [
     "Odia",
     "Punjabi",
     "Tamil",
-    "Telugu"
+    "Telugu",
+    "English",
+    "German"
 ]
 
 def validate_language(language):
@@ -43,7 +45,7 @@ def asr_transcribe(client, file_path, language):
             files=files
         )
     if resp.status_code != 200:
-        raise DhwaniAPIError(resp)
+        raise DwaniAPIError(resp)
     return resp.json()
 
 class ASR:

dwani/audio.py

@@ -1,4 +1,4 @@
-from .exceptions import DhwaniAPIError
+from .exceptions import DwaniAPIError
 import requests
 
 def audio_speech(client, input, response_format="mp3", output_file=None):
@@ -14,7 +14,7 @@ def audio_speech(client, input, response_format="mp3", output_file=None):
         stream=True
     )
     if resp.status_code != 200:
-        raise DhwaniAPIError(resp)
+        raise DwaniAPIError(resp)
     if output_file:
         with open(output_file, "wb") as f:
             for chunk in resp.iter_content(chunk_size=8192):

dwani/chat.py

@@ -1,4 +1,4 @@
-from .exceptions import DhwaniAPIError
+from .exceptions import DwaniAPIError
 import requests
 
 # Language options mapping
@@ -7,14 +7,15 @@ 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"),
+    ("German","deu_Latn"),
 ]
 
 # Create a dictionary for language name to code mapping
@@ -35,9 +36,25 @@ 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_direct(client, prompt, model="gemma3", system_prompt=""):
+    url = f"{client.api_base}/v1/chat_direct"
+    payload = {
+        "prompt": prompt,
+        "model": model,
+        "system_prompt":system_prompt
+    }
+    resp = requests.post(
+        url,
+        headers={**client._headers(), "Content-Type": "application/json"},
+        json=payload
+    )
+    if resp.status_code != 200:
+        raise DwaniAPIError(resp)
+    return resp.json()
+
 def chat_create(client, prompt, src_lang, tgt_lang, model="gemma3"):
     # Validate model
-    valid_models = ["gemma3", "qwen3", "deepseek-r1"]
+    valid_models = ["gemma3", "qwen3", "deepseek-r1", "sarvam-m"]
     if model not in valid_models:
         raise ValueError(f"Unsupported model: {model}. Supported models: {valid_models}")
     
@@ -58,11 +75,15 @@ def chat_create(client, prompt, src_lang, tgt_lang, model="gemma3"):
         json=payload
     )
     if resp.status_code != 200:
-        raise DhwaniAPIError(resp)
+        raise DwaniAPIError(resp)
     return resp.json()
 
 class Chat:
     @staticmethod
     def create(prompt, src_lang, tgt_lang, model="gemma3"):
         from . import _get_client
-        return _get_client().chat(prompt, src_lang, tgt_lang, model)
+        return _get_client().chat(prompt, src_lang, tgt_lang, model)
+    @staticmethod
+    def direct(prompt, model="gemma3", system_prompt=""):
+        from . import _get_client
+        return _get_client().chat_direct(prompt, model, system_prompt)

dwani/client.py

@@ -1,6 +1,6 @@
 import os
 import requests
-from .exceptions import DhwaniAPIError
+from .exceptions import DwaniAPIError
 
 class DwaniClient:
     def __init__(self, api_key=None, api_base=None):
@@ -22,7 +22,12 @@ class DwaniClient:
     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, model=model)
-    
+
+
+    def chat_direct(self, prompt, model="gemma3", system_prompt=""):
+        from .chat import chat_direct
+        return chat_direct(self, prompt=prompt, model=model, system_prompt=system_prompt)
+
     def speech(self, input, response_format="mp3"):
         from .audio import audio_speech
         return audio_speech(self, input=input, response_format=response_format)
@@ -31,6 +36,10 @@ class DwaniClient:
         from .vision import vision_caption
         return vision_caption(self, file_path=file_path, query=query, src_lang=src_lang, tgt_lang=tgt_lang, model=model)
 
+    def caption_direct(self, file_path, query="describe the image", model="gemma3", system_prompt=""):
+        from .vision import vision_direct
+        return vision_direct(self, file_path=file_path, query=query, model=model, system_prompt=system_prompt)
+
     def transcribe(self, file_path, language=None):
         from .asr import asr_transcribe
         return asr_transcribe(self, file_path=file_path, language=language)

dwani/docs.py

@@ -1,5 +1,6 @@
 import requests
-from .exceptions import DhwaniAPIError
+
+from .exceptions import DwaniAPIError
 import logging
 
 # Set up logging
@@ -11,7 +12,8 @@ language_options = [
     ("Kannada", "kan_Knda"),
     ("Hindi", "hin_Deva"),
     ("Tamil", "tam_Taml"),
-    ("Telugu", "tel_Telu")
+    ("Telugu", "tel_Telu"),
+    ("German", "deu_Latn") 
 ]
 
 # Create dictionaries for language name to code and code to code mapping
@@ -61,7 +63,7 @@ def document_ocr(client, file_path, language=None, model="gemma3"):
             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)
+            raise DwaniAPIError(resp) if 'resp' in locals() else DwaniAPIError.from_exception(e)
     
     logger.debug(f"OCR response: {resp.status_code}")
     return resp.json()
@@ -89,6 +91,7 @@ def document_summarize(client, file_path, page_number=1, src_lang="eng_Latn", tg
             "tgt_lang": tgt_lang_code,
             "model": model
         }
+
         try:
             resp = requests.post(
                 url,
@@ -100,9 +103,10 @@ def document_summarize(client, file_path, page_number=1, src_lang="eng_Latn", tg
             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)
+            raise DwaniAPIError(resp) if 'resp' in locals() else DwaniAPIError.from_exception(e)
     
     logger.debug(f"Summarize response: {resp.status_code}")
+
     return resp.json()
 
 def extract(client, file_path, page_number=1, src_lang="eng_Latn", tgt_lang="kan_Knda", model="gemma3"):
@@ -122,6 +126,7 @@ def extract(client, file_path, page_number=1, src_lang="eng_Latn", tgt_lang="kan
     headers = client._headers()
     with open(file_path, "rb") as f:
         files = {"file": (file_path, f, "application/pdf")}
+
         data = {
             "page_number": str(page_number),
             "src_lang": src_lang_code,
@@ -139,9 +144,10 @@ def extract(client, file_path, page_number=1, src_lang="eng_Latn", tgt_lang="kan
             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)
+            raise DwaniAPIError(resp) if 'resp' in locals() else DwaniAPIError.from_exception(e)
     
     logger.debug(f"Extract response: {resp.status_code}")
+
     return resp.json()
 
 def doc_query(
@@ -178,6 +184,7 @@ def doc_query(
             "tgt_lang": tgt_lang_code,
             "model": model
         }
+
         try:
             resp = requests.post(
                 url,
@@ -189,9 +196,10 @@ def doc_query(
             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)
+            raise DwaniAPIError(resp) if 'resp' in locals() else DwaniAPIError.from_exception(e)
     
     logger.debug(f"Doc query response: {resp.status_code}")
+
     return resp.json()
 
 def doc_query_kannada(
@@ -221,6 +229,7 @@ def doc_query_kannada(
     headers = client._headers()
     with open(file_path, "rb") as f:
         files = {"file": (file_path, f, "application/pdf")}
+
         data = {
             "page_number": str(page_number),
             "prompt": prompt,
@@ -239,9 +248,10 @@ def doc_query_kannada(
             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)
+            raise DwaniAPIError(resp) if 'resp' in locals() else DwaniAPIError.from_exception(e)
     
     logger.debug(f"Doc query Kannada response: {resp.status_code}")
+
     return resp.json()
 
 class Documents:

dwani/exceptions.py

@@ -1,4 +1,4 @@
-class DhwaniAPIError(Exception):
+class DwaniAPIError(Exception):
     def __init__(self, response):
         super().__init__(f"API Error {response.status_code}: {response.text}")
         self.status_code = response.status_code

dwani/translate.py

@@ -1,4 +1,4 @@
-from .exceptions import DhwaniAPIError
+from .exceptions import DwaniAPIError
 import requests
 
 # Language options mapping
@@ -14,7 +14,8 @@ language_options = [
     ("Odia", "ory_Orya"),
     ("Punjabi", "pan_Guru"),
     ("Tamil", "tam_Taml"),
-    ("Telugu", "tel_Telu") 
+    ("Telugu", "tel_Telu"),
+    ("German", "deu_Latn") 
 ]
 
 # Create dictionaries for language name to code and code to code mapping
@@ -62,7 +63,7 @@ def run_translate(client, sentences, src_lang, tgt_lang):
         json=payload
     )
     if resp.status_code != 200:
-        raise DhwaniAPIError(resp)
+        raise DwaniAPIError(resp)
     return resp.json()
 
 class Translate:

dwani/vision.py

@@ -1,4 +1,4 @@
-from .exceptions import DhwaniAPIError
+from .exceptions import DwaniAPIError
 import requests
 
 # Language options mapping
@@ -14,7 +14,8 @@ language_options = [
     ("Odia", "ory_Orya"),
     ("Punjabi", "pan_Guru"),
     ("Tamil", "tam_Taml"),
-    ("Telugu", "tel_Telu") 
+    ("Telugu", "tel_Telu"),
+    ("German","deu_Latn") 
 ]
 
 # Create dictionaries for language name to code and code to code mapping
@@ -35,9 +36,32 @@ 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_direct(client, file_path, query="describe this image", model="gemma3", system_prompt=""):
+    url = (
+        f"{client.api_base}/v1/visual_query_direct"
+        f"?model={model}"
+    )
+    headers = {
+        **client._headers(),
+        "accept": "application/json"
+    }
+    with open(file_path, "rb") as f:
+        files = {"file": (file_path, f, "image/png")}
+        data = {"query": query, "system_prompt": system_prompt}
+        resp = requests.post(
+            url,
+            headers=headers,
+            files=files,
+            data=data
+        )
+    if resp.status_code != 200:
+        raise DwaniAPIError(resp)
+    return resp.json()
+
+
 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"]
+    valid_models = ["gemma3", "qwen2.5vl", "moondream", "smolvla"]
     if model not in valid_models:
         raise ValueError(f"Unsupported model: {model}. Supported models: {valid_models}")
     
@@ -64,11 +88,15 @@ def vision_caption(client, file_path, query="describe the image", src_lang="eng_
             data=data
         )
     if resp.status_code != 200:
-        raise DhwaniAPIError(resp)
+        raise DwaniAPIError(resp)
     return resp.json()
 
 class Vision:
     @staticmethod
     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(file_path, query, src_lang, tgt_lang, model)
+        return _get_client().caption(file_path, query, src_lang, tgt_lang, model)
+    @staticmethod
+    def caption_direct(file_path, query="describe the image", model="gemma3", system_prompt=""):
+        from . import _get_client
+        return _get_client().caption_direct(file_path, query, model, system_prompt)

{dwani-0.1.9.dist-info → dwani-0.1.11.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: dwani
-Version: 0.1.9
+Version: 0.1.11
 Summary: Multimodal API for Indian languages (Chat, Vision, TTS, ASR, Translate, Docs)
 Author-email: sachin <python@dwani.ai>
 License: MIT License
@@ -59,10 +59,10 @@ dwani.api_base = os.getenv("DWANI_API_BASE_URL")
 ### Text Query 
 ---
 - With model selection
-  - Supported models : gemma3 (default), qwen3
+  - gemma3 (default), qwen3, sarvam-m
 
 ---
-- gemma3
+- gemma3 - with translation
 ```python
 resp = dwani.Chat.create(prompt="Hello!", src_lang="english", tgt_lang="kannada", model="gemma3")
 print(resp)
@@ -70,13 +70,23 @@ print(resp)
 ```json
 {'response': 'ನಮಸ್ತೆ! ಭಾರತ ಮತ್ತು ಕರ್ನಾಟಕವನ್ನು ಗಮನದಲ್ಲಿಟ್ಟುಕೊಂಡು ಇಂದು ನಿಮ್ಮ ಪ್ರಶ್ನೆಗಳಿಗೆ ನಾನು ನಿಮಗೆ ಹೇಗೆ ಸಹಾಯ ಮಾಡಲಿ?'}
 ```
+
+- gemma3 - without translation
+```python
+resp = dwani.Chat.direct(prompt="Hello!", model="gemma3")
+print(resp)
+```
+```json
+{'response': 'Hello! I am Dwani, ready to assist you with information pertaining to India, specifically Karnataka. '}
+```
+
 ---
 ### Vision Query
 ---
 - With model selection
-  - Supported models : gemma3 (default), moondream
-- gemma3
+  - gemma3 (default), moondream, smolvla
 
+- gemma3 - with translation
 ```python
 result = dwani.Vision.caption(
     file_path="image.png",
@@ -90,6 +100,19 @@ 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.'}
 ```
+- gemma3 - without translation
+```python
+result = dwani.Vision.caption_direct(
+    file_path="image.png",
+    query="Describe this logo",
+    model="gemma3"
+)
+print(result)
+```
+```json
+ {'answer': 'The logo displays a publishing overview stating that changes are under review, with a production rollout initiated at version sixty-four point one point one, expanding to sixteen countries/regions including Australia and Bangladesh.'}
+```
+
 ---
 ### Speech to Text -  Automatic Speech Recognition (ASR)
 ---
@@ -108,13 +131,13 @@ resp = dwani.Translate.run_translate(sentences="hi, i am gaganyatri", src_lang="
 print(resp)
 ```
 ```json
-{'translations': ['ಹಾಯ್']}
+{'translations': ['ಹಾಯ್, ನಾನು ಗಗನಯಾತ್ರಿ']}
 ```
 ---
 ### Text to Speech -  Speech Synthesis
 ---
 ```python
-response = dwani.Audio.speech(input="ಕರ್ನಾಟಕ ದ ರಾಜಧಾನಿ ಯಾವುದು", response_format="wav")
+response = dwani.Audio.speech(input="ಕರ್ನಾಟಕದ ರಾಜಧಾನಿ ಯಾವುದು", response_format="wav")
 with open("output.wav", "wb") as f:
     f.write(response)
 ```

dwani-0.1.11.dist-info/RECORD

@@ -0,0 +1,14 @@
+dwani/__init__.py,sha256=8Q1qdF0g6xWEy1_PK6qxG-EeZUp_QOwYBqNv_y6EyHI,3010
+dwani/asr.py,sha256=BAdqivQd57NJZX1dSY-J6EFi8TDdyuhf_AyCPcQ0M7w,1719
+dwani/audio.py,sha256=MWsIZazL91c2wa5AE1YY78l9RKaJwNFFHIajuwl43Jg,886
+dwani/chat.py,sha256=Tui52XBhUyDyN2rOFoLme4oB0Q8fkD9_0tFDAnRzoaU,2979
+dwani/client.py,sha256=VG7MFCF4yLAWyD037YcI3QTUMxRfJZrWTDK-JAurTnY,3356
+dwani/docs.py,sha256=Cp0Gtudug79GH25toB-Npl35ZFA0TM32oZF2xH1VmNY,10598
+dwani/exceptions.py,sha256=n06dPmR20rS4T3sJBWHQhGxzg4SJKzird9Hx0YTwwo0,226
+dwani/translate.py,sha256=-6UHV5hu1oBxuDlGlGYp13bFDayKWwo1rBkJhE-LRMs,2568
+dwani/vision.py,sha256=sjkudW2Jb_PEbRcoZy_S1Wno6K5icZz42pHcpD1FMGs,3607
+dwani-0.1.11.dist-info/licenses/LICENSE,sha256=IAD8tbwWZbPWHXgYjabHoMv0aaUzZUYzYiEbfhTCisY,1070
+dwani-0.1.11.dist-info/METADATA,sha256=6dhzLxq9g_unE9uqNFYwMGBXAB3WvjAMyDnvhYjkDqo,5791
+dwani-0.1.11.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+dwani-0.1.11.dist-info/top_level.txt,sha256=AM5EhkyuO_EXQFR9JIxEV6tAYMCCyc-a1dLifpCGBUk,6
+dwani-0.1.11.dist-info/RECORD,,

dwani-0.1.9.dist-info/RECORD

@@ -1,14 +0,0 @@
-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,,