dwani 0.1.10__py3-none-any.whl → 0.1.12__py3-none-any.whl

dwani/__init__.py

@@ -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/chat.py

@@ -36,9 +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}")
     
@@ -66,4 +82,8 @@ 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

@@ -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/translate.py

@@ -25,21 +25,27 @@ lang_code_to_code = {code: code for _, code in language_options}
 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 split_into_sentences(text):
+    """Split a string into sentences based on full stops."""
+    if not text.strip():
+        return []
+    # Split on full stops, preserving non-empty sentences
+    sentences = [s.strip() for s in text.split('.') if s.strip()]
+    return sentences
+
 def run_translate(client, sentences, src_lang, tgt_lang):
-    # Convert single string to list if necessary
+    """Translate sentences in a single API call."""
+    # Convert single string to list of sentences if necessary
     if isinstance(sentences, str):
-        sentences = [sentences]
+        sentences = split_into_sentences(sentences)
     elif not isinstance(sentences, list):
         raise ValueError("sentences must be a string or a list of strings")
     
@@ -70,4 +76,4 @@ class Translate:
     @staticmethod
     def run_translate(sentences, src_lang, tgt_lang):
         from . import _get_client
-        return _get_client().translate(sentences, src_lang, tgt_lang)
+        return run_translate(_get_client(), sentences, src_lang, tgt_lang)

dwani/vision.py

@@ -36,9 +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}")
     
@@ -72,4 +95,8 @@ 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-0.1.10.dist-info → dwani-0.1.12.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: dwani
-Version: 0.1.10
+Version: 0.1.12
 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
@@ -59,10 +59,10 @@ dwani.api_base = os.getenv("DWANI_API_BASE_URL")
 ### Text Query 
 ---
 - With model selection
-  - Supported models : gemma3 (default), qwen3, deepseek-r1-8b, sarvam-m
+  - 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-0.1.12.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=c03N8-tN49IBcTA6GMOkrJ3MaVzZ12RnYdLQwRbEeoQ,2794
+dwani/vision.py,sha256=sjkudW2Jb_PEbRcoZy_S1Wno6K5icZz42pHcpD1FMGs,3607
+dwani-0.1.12.dist-info/licenses/LICENSE,sha256=IAD8tbwWZbPWHXgYjabHoMv0aaUzZUYzYiEbfhTCisY,1070
+dwani-0.1.12.dist-info/METADATA,sha256=aj_vKOlWgfznm_c_RqT82VYKl8FBia0CDQLbf40ZoJU,5791
+dwani-0.1.12.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+dwani-0.1.12.dist-info/top_level.txt,sha256=AM5EhkyuO_EXQFR9JIxEV6tAYMCCyc-a1dLifpCGBUk,6
+dwani-0.1.12.dist-info/RECORD,,

dwani-0.1.10.dist-info/RECORD

@@ -1,14 +0,0 @@
-dwani/__init__.py,sha256=k1fWBnAp5zHQaYnOpUwzPIngqzVO4wIQr1wp5kPyzfE,2663
-dwani/asr.py,sha256=BAdqivQd57NJZX1dSY-J6EFi8TDdyuhf_AyCPcQ0M7w,1719
-dwani/audio.py,sha256=MWsIZazL91c2wa5AE1YY78l9RKaJwNFFHIajuwl43Jg,886
-dwani/chat.py,sha256=K3OJHQcRhU0aVmWBqajZqbfZg_Q5Dfm6Es3YMSpkxGY,2332
-dwani/client.py,sha256=UsRLoYZgj25F-qCGlATvElG6r3EWxqndeMv696cBk1w,2904
-dwani/docs.py,sha256=Cp0Gtudug79GH25toB-Npl35ZFA0TM32oZF2xH1VmNY,10598
-dwani/exceptions.py,sha256=n06dPmR20rS4T3sJBWHQhGxzg4SJKzird9Hx0YTwwo0,226
-dwani/translate.py,sha256=-6UHV5hu1oBxuDlGlGYp13bFDayKWwo1rBkJhE-LRMs,2568
-dwani/vision.py,sha256=9tRPhEXFQ3n-80XxVCs1qrEKqvzsoxGQTOKs2fTwQTI,2699
-dwani-0.1.10.dist-info/licenses/LICENSE,sha256=IAD8tbwWZbPWHXgYjabHoMv0aaUzZUYzYiEbfhTCisY,1070
-dwani-0.1.10.dist-info/METADATA,sha256=YUBayRe_IiFtbeJhr3Wu9trPBRGykVDngdzZGD8_2pk,5062
-dwani-0.1.10.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-dwani-0.1.10.dist-info/top_level.txt,sha256=AM5EhkyuO_EXQFR9JIxEV6tAYMCCyc-a1dLifpCGBUk,6
-dwani-0.1.10.dist-info/RECORD,,