dwani 0.1.18__tar.gz → 0.1.20__tar.gz

{dwani-0.1.18 → dwani-0.1.20}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: dwani
-Version: 0.1.18
+Version: 0.1.20
 Summary: Multimodal API for Indian + European languages (Chat, Vision, TTS, ASR, Translate, Docs)
 Author-email: sachin <python@dwani.ai>
 License: MIT License
@@ -78,7 +78,7 @@ dwani.api_base = os.getenv("DWANI_API_BASE_URL")
 
 #### Document - OCR
 ```python
-result = dwani.Documents.run_ocr_number(file_path="dwani-workshop.pdf", page_number=1, model="gemma3")
+result = dwani.Documents.run_ocr_page(file_path="dwani-workshop.pdf", page_number=1, model="gemma3")
 print(result)
 ```
 ```json
@@ -86,6 +86,17 @@ print(result)
 ```
 
 
+#### Document - Summary
+
+```python
+result = dwani.Documents.summarize_all(
+            file_path="dwani-workshop.pdf", model="gemma3" , tgt_lang="english"  
+    )
+
+print("Document Query Response: gemma3- ", result["summary"])
+```
+
+
 ### Text Query 
 ---
 - gemma3 (default)
@@ -168,4 +179,16 @@ python -m build
 
 python -m twine upload dist/*
 
+-->
+
+<!--
+Without Batch  
+2025-07-14 13:39:50,330 - dwani_api - INFO - Request to /indic-summarize-pdf-all took 245.381 seconds
+INFO:dwani_api:Request to /indic-summarize-pdf-all took 245.381 seconds
+
+With Batch
+
+vllm serve google/gemma-3-4b-it --served-model-name gemma3 --host 0.0.0.0 --port 9000 --gpu-memory-utilization 0.8 --tensor-parallel-size 1 --max-model-len 65536     --dtype bfloat16 
+
+
 -->

{dwani-0.1.18 → dwani-0.1.20}/README.md

@@ -41,7 +41,7 @@ dwani.api_base = os.getenv("DWANI_API_BASE_URL")
 
 #### Document - OCR
 ```python
-result = dwani.Documents.run_ocr_number(file_path="dwani-workshop.pdf", page_number=1, model="gemma3")
+result = dwani.Documents.run_ocr_page(file_path="dwani-workshop.pdf", page_number=1, model="gemma3")
 print(result)
 ```
 ```json
@@ -49,6 +49,17 @@ print(result)
 ```
 
 
+#### Document - Summary
+
+```python
+result = dwani.Documents.summarize_all(
+            file_path="dwani-workshop.pdf", model="gemma3" , tgt_lang="english"  
+    )
+
+print("Document Query Response: gemma3- ", result["summary"])
+```
+
+
 ### Text Query 
 ---
 - gemma3 (default)
@@ -131,4 +142,16 @@ python -m build
 
 python -m twine upload dist/*
 
+-->
+
+<!--
+Without Batch  
+2025-07-14 13:39:50,330 - dwani_api - INFO - Request to /indic-summarize-pdf-all took 245.381 seconds
+INFO:dwani_api:Request to /indic-summarize-pdf-all took 245.381 seconds
+
+With Batch
+
+vllm serve google/gemma-3-4b-it --served-model-name gemma3 --host 0.0.0.0 --port 9000 --gpu-memory-utilization 0.8 --tensor-parallel-size 1 --max-model-len 65536     --dtype bfloat16 
+
+
 -->

{dwani-0.1.18 → dwani-0.1.20}/dwani/__init__.py

@@ -53,25 +53,30 @@ class translate:
 
 class document:
     @staticmethod
-    def run_ocr_number(file_path, page_number=1, model="gemma3"):
-        return _get_client().document_ocr_number(file_path, page_number, model)
+    def run_ocr_page(file_path, page_number=1, model="gemma3"):
+        return _get_client().document_ocr_page(file_path, page_number, model)
     
     @staticmethod
     def run_ocr_all(file_path, model="gemma3"):
         return _get_client().document_ocr_all(file_path, model)
     
     @staticmethod
-    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)
+    def run_summarize_page(file_path, page_number=1, tgt_lang="kan_Knda", model="gemma3"):
+        return _get_client().document_summarize_page(file_path, page_number, tgt_lang, model)
+    
+
+    @staticmethod
+    def run_summarize_all(file_path,  tgt_lang="kan_Knda", model="gemma3"):
+        return _get_client().document_summarize_all(file_path, tgt_lang, model)
     
     @staticmethod
-    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)
+    def run_extract(file_path, page_number=1, tgt_lang="kan_Knda", model="gemma3"):
+        return _get_client().extract(file_path, page_number, tgt_lang, model)
     
     @staticmethod
     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(file_path, page_number=1, prompt="list key points", src_lang="eng_Latn", tgt_lang="kan_Knda", model="gemma3"):
+    def run_doc_query_kannada(file_path, page_number=1, prompt="list key points", src_lang="kan_Latn", tgt_lang="kan_Knda", model="gemma3"):
         return _get_client().doc_query_kannada(file_path, page_number, prompt, src_lang, tgt_lang, model)

{dwani-0.1.18 → dwani-0.1.20}/dwani/asr.py

@@ -42,7 +42,8 @@ def asr_transcribe(client, file_path, language):
         resp = requests.post(
             f"{client.api_base}/v1/transcribe/?language={api_language}",
             headers=client._headers(),
-            files=files
+            files=files,
+            timeout=90
         )
     if resp.status_code != 200:
         raise DwaniAPIError(resp)

{dwani-0.1.18 → dwani-0.1.20}/dwani/audio.py

@@ -12,7 +12,8 @@ def audio_speech(client, input, response_format="mp3", output_file=None, languag
         headers={**client._headers(), "accept": "application/json"},
         params=params,
         data='',  # Empty body, as in the curl example
-        stream=True
+        stream=True,
+        timeout=90
     )
     if resp.status_code != 200:
         raise DwaniAPIError(resp)

{dwani-0.1.18 → dwani-0.1.20}/dwani/chat.py

@@ -46,7 +46,8 @@ def chat_direct(client, prompt, model="gemma3", system_prompt=""):
     resp = requests.post(
         url,
         headers={**client._headers(), "Content-Type": "application/json"},
-        json=payload
+        json=payload,
+        timeout=90
     )
     if resp.status_code != 200:
         raise DwaniAPIError(resp)
@@ -72,7 +73,8 @@ def chat_create(client, prompt, src_lang, tgt_lang, model="gemma3"):
     resp = requests.post(
         url,
         headers={**client._headers(), "Content-Type": "application/json"},
-        json=payload
+        json=payload,
+        timeout=90
     )
     if resp.status_code != 200:
         raise DwaniAPIError(resp)

{dwani-0.1.18 → dwani-0.1.20}/dwani/client.py

@@ -51,13 +51,13 @@ class DwaniClient:
         from .docs import document_ocr_all
         return document_ocr_all(self, file_path=file_path, model=model)
 
-    def document_summarize(self, file_path, page_number=1, src_lang="eng_Latn", tgt_lang="kan_Knda", model="gemma3"):
+    def document_summarize(self, file_path, page_number=1, tgt_lang="kan_Knda", model="gemma3"):
         from .docs import document_summarize
-        return document_summarize(self, file_path, page_number, src_lang, tgt_lang, model)
+        return document_summarize(self, file_path, page_number, tgt_lang, model)
 
-    def extract(self, file_path, page_number=1, src_lang="eng_Latn", tgt_lang="kan_Knda", model="gemma3"):
+    def extract(self, file_path, page_number=1, 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, model=model)
+        return extract(self, file_path=file_path, page_number=page_number, 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", model="gemma3"):
         from .docs import doc_query

{dwani-0.1.18 → dwani-0.1.20}/dwani/docs.py

@@ -56,7 +56,7 @@ def document_ocr_all(client, file_path, model="gemma3"):
                 headers=client._headers(),
                 files=files,
                 data=data,
-                timeout=60
+                timeout=90
             )
             resp.raise_for_status()
         except requests.RequestException as e:
@@ -67,7 +67,7 @@ def document_ocr_all(client, file_path, model="gemma3"):
     return resp.json()
 
 
-def document_ocr_number(client, file_path, page_number, model="gemma3"):
+def document_ocr_page(client, file_path, page_number, model="gemma3"):
     """OCR a document (image/PDF) and return extracted text."""
     logger.debug(f"Calling document_ocr: file_path={file_path}, model={model}")
     validate_model(model)
@@ -85,7 +85,7 @@ def document_ocr_number(client, file_path, page_number, model="gemma3"):
                 headers=client._headers(),
                 files=files,
                 params=params,
-                timeout=60
+                timeout=90
             )
             resp.raise_for_status()
         except requests.RequestException as e:
@@ -94,9 +94,10 @@ def document_ocr_number(client, file_path, page_number, model="gemma3"):
     
     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", model="gemma3"):
+
+def document_summarize_page(client, file_path, page_number=1, tgt_lang="kan_Knda", model="gemma3"):
     """Summarize a PDF document with language and page number options."""
-    logger.debug(f"Calling document_summarize: file_path={file_path}, page_number={page_number}, src_lang={src_lang}, tgt_lang={tgt_lang}, model={model}")
+    logger.debug(f"Calling document_summarize: file_path={file_path}, page_number={page_number}, tgt_lang={tgt_lang}, model={model}")
     validate_model(model)
     
     if not file_path.lower().endswith('.pdf'):
@@ -104,7 +105,6 @@ def document_summarize(client, file_path, page_number=1, src_lang="eng_Latn", tg
     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)
     
     url = f"{client.api_base}/v1/indic-summarize-pdf"
@@ -113,7 +113,6 @@ def document_summarize(client, file_path, page_number=1, src_lang="eng_Latn", tg
         files = {"file": (file_path, f, "application/pdf")}
         data = {
             "page_number": str(page_number),
-            "src_lang": src_lang_code,
             "tgt_lang": tgt_lang_code,
             "model": model
         }
@@ -124,7 +123,44 @@ def document_summarize(client, file_path, page_number=1, src_lang="eng_Latn", tg
                 headers=headers,
                 files=files,
                 data=data,
-                timeout=60
+                timeout=90
+            )
+            resp.raise_for_status()
+        except requests.RequestException as e:
+            logger.error(f"Summarize request failed: {str(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 document_summarize_all(client, file_path, tgt_lang="kan_Knda", model="gemma3"):
+    """Summarize a PDF document with language """
+    logger.debug(f"Calling document_summarize: file_path={file_path}, tgt_lang={tgt_lang}, model={model}")
+    validate_model(model)
+    
+    if not file_path.lower().endswith('.pdf'):
+        raise ValueError("File must be a PDF")
+    
+    tgt_lang_code = normalize_language(tgt_lang)
+    
+    url = f"{client.api_base}/v1/indic-summarize-pdf-all"
+    headers = client._headers()
+    with open(file_path, "rb") as f:
+        files = {"file": (file_path, f, "application/pdf")}
+        data = {
+            "tgt_lang": tgt_lang_code,
+            "model": model
+        }
+
+        try:
+            resp = requests.post(
+                url,
+                headers=headers,
+                files=files,
+                data=data,
+                timeout=90
             )
             resp.raise_for_status()
         except requests.RequestException as e:
@@ -135,9 +171,10 @@ def document_summarize(client, file_path, page_number=1, src_lang="eng_Latn", tg
 
     return resp.json()
 
-def extract(client, file_path, page_number=1, src_lang="eng_Latn", tgt_lang="kan_Knda", model="gemma3"):
+
+def extract(client, file_path, page_number=1, 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}")
+    logger.debug(f"Calling extract: file_path={file_path}, page_number={page_number}, tgt_lang={tgt_lang}, model={model}")
     validate_model(model)
     
     if not file_path.lower().endswith('.pdf'):
@@ -145,7 +182,6 @@ def extract(client, file_path, page_number=1, src_lang="eng_Latn", tgt_lang="kan
     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)
     
     url = f"{client.api_base}/v1/indic-extract-text/"
@@ -155,7 +191,6 @@ def extract(client, file_path, page_number=1, src_lang="eng_Latn", tgt_lang="kan
 
         data = {
             "page_number": str(page_number),
-            "src_lang": src_lang_code,
             "tgt_lang": tgt_lang_code,
             "model": model
         }
@@ -165,7 +200,7 @@ def extract(client, file_path, page_number=1, src_lang="eng_Latn", tgt_lang="kan
                 headers=headers,
                 files=files,
                 data=data,
-                timeout=60
+                timeout=90
             )
             resp.raise_for_status()
         except requests.RequestException as e:
@@ -181,12 +216,12 @@ def doc_query(
     file_path,
     page_number=1,
     prompt="list the key points",
-    src_lang="eng_Latn",
     tgt_lang="kan_Knda",
+    src_lang="eng_Latn",
     model="gemma3"
 ):
     """Query a document with a custom prompt and language options."""
-    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}")
+    logger.debug(f"Calling doc_query: file_path={file_path}, page_number={page_number}, prompt={prompt}, tgt_lang={tgt_lang}, model={model}")
     validate_model(model)
     
     if not file_path.lower().endswith('.pdf'):
@@ -196,9 +231,10 @@ def doc_query(
     if not prompt.strip():
         raise ValueError("Prompt cannot be empty")
     
-    src_lang_code = normalize_language(src_lang)
     tgt_lang_code = normalize_language(tgt_lang)
     
+    src_lang_code = normalize_language(src_lang)
+
     url = f"{client.api_base}/v1/indic-custom-prompt-pdf"
     headers = client._headers()
     with open(file_path, "rb") as f:
@@ -206,8 +242,8 @@ def doc_query(
         data = {
             "page_number": str(page_number),
             "prompt": prompt,
-            "src_lang": src_lang_code,
             "tgt_lang": tgt_lang_code,
+            "src_lang": src_lang_code,
             "model": model
         }
 
@@ -217,7 +253,7 @@ def doc_query(
                 headers=headers,
                 files=files,
                 data=data,
-                timeout=60
+                timeout=90
             )
             resp.raise_for_status()
         except requests.RequestException as e:
@@ -233,12 +269,12 @@ def doc_query_kannada(
     file_path,
     page_number=1,
     prompt="list key points",
-    src_lang="eng_Latn",
     tgt_lang="kan_Knda",
+    src_lang="kan_Knda",
     model="gemma3"
 ):
     """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}")
+    logger.debug(f"Calling doc_query_kannada: file_path={file_path}, page_number={page_number}, prompt={prompt}, tgt_lang={tgt_lang}, model={model}")
     validate_model(model)
     
     if not file_path.lower().endswith('.pdf'):
@@ -248,8 +284,10 @@ def doc_query_kannada(
     if not prompt.strip():
         raise ValueError("Prompt cannot be empty")
     
-    src_lang_code = normalize_language(src_lang)
     tgt_lang_code = normalize_language(tgt_lang) if tgt_lang else "kan_Knda"
+
+    src_lang_code = normalize_language(src_lang) 
+    
     
     url = f"{client.api_base}/v1/indic-custom-prompt-pdf"
     headers = client._headers()
@@ -259,8 +297,8 @@ def doc_query_kannada(
         data = {
             "page_number": str(page_number),
             "prompt": prompt,
-            "src_lang": src_lang_code,
             "tgt_lang": tgt_lang_code,
+            "src_lang": src_lang_code,
             "model": model
         }
         try:
@@ -269,7 +307,7 @@ def doc_query_kannada(
                 headers=headers,
                 files=files,
                 data=data,
-                timeout=60
+                timeout=90
             )
             resp.raise_for_status()
         except requests.RequestException as e:
@@ -285,7 +323,7 @@ class Documents:
     def run_ocr_number(file_path, page_number=2,model="gemma3"):
         from .client import DwaniClient
         client = DwaniClient()
-        return document_ocr_number(client, file_path, page_number, model)
+        return document_ocr_page(client, file_path, page_number, model)
     @staticmethod
     def run_ocr_all(file_path, model="gemma3"):
         from .client import DwaniClient
@@ -293,25 +331,32 @@ class Documents:
         return document_ocr_all(client, file_path, model)
     
     @staticmethod
-    def summarize(file_path, page_number=1, src_lang="eng_Latn", tgt_lang="kan_Knda", model="gemma3"):
+    def summarize_page(file_path, page_number=1, 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)
-    
+        return document_summarize_page(client, file_path, page_number, tgt_lang, model)
+
+
+    @staticmethod
+    def summarize_all(file_path, tgt_lang="kan_Knda", model="gemma3"):
+        from .client import DwaniClient
+        client = DwaniClient()
+        return document_summarize_all(client, file_path, tgt_lang, model)
+
     @staticmethod
-    def run_extract(file_path, page_number=1, src_lang="eng_Latn", tgt_lang="kan_Knda", model="gemma3"):
+    def run_extract(file_path, page_number=1, tgt_lang="kan_Knda", model="gemma3"):
         from .client import DwaniClient
         client = DwaniClient()
-        return extract(client, file_path, page_number, src_lang, tgt_lang, model)
+        return extract(client, file_path, page_number, tgt_lang, model)
     
     @staticmethod
-    def run_doc_query(file_path, page_number=1, prompt="list the key points", src_lang="eng_Latn", tgt_lang="kan_Knda", model="gemma3"):
+    def run_doc_query(file_path, page_number=1, prompt="list the key points", 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)
+        return doc_query(client, file_path, page_number, prompt, tgt_lang, model)
     
     @staticmethod
-    def run_doc_query_kannada(file_path, page_number=1, prompt="list key points", src_lang="eng_Latn", tgt_lang="kan_Knda", model="gemma3"):
+    def run_doc_query_kannada(file_path, page_number=1, prompt="list key points", 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)
+        return doc_query_kannada(client, file_path, page_number, prompt, tgt_lang, model)

{dwani-0.1.18 → dwani-0.1.20}/dwani/translate.py

@@ -66,7 +66,8 @@ def run_translate(client, sentences, src_lang, tgt_lang):
     resp = requests.post(
         url,
         headers={**client._headers(), "Content-Type": "application/json", "accept": "application/json"},
-        json=payload
+        json=payload,
+        timeout=90
     )
     if resp.status_code != 200:
         raise DwaniAPIError(resp)

{dwani-0.1.18 → dwani-0.1.20}/dwani/vision.py

@@ -52,7 +52,8 @@ def vision_direct(client, file_path, query="describe this image", model="gemma3"
             url,
             headers=headers,
             files=files,
-            data=data
+            data=data,
+            timeout=90
         )
     if resp.status_code != 200:
         raise DwaniAPIError(resp)
@@ -84,7 +85,8 @@ def vision_caption(client, file_path, query="describe the image", src_lang="eng_
             url,
             headers=headers,
             files=files,
-            data=data
+            data=data,
+            timeout=90
         )
     if resp.status_code != 200:
         raise DwaniAPIError(resp)

{dwani-0.1.18 → dwani-0.1.20}/pyproject.toml

@@ -5,7 +5,7 @@ build-backend = "setuptools.build_meta"
 [project]
 name = "dwani"
 
-version = "0.1.18"
+version = "0.1.20"
 description = "Multimodal API for Indian + European languages (Chat, Vision, TTS, ASR, Translate, Docs)"
 authors = [
     { name="sachin", email="python@dwani.ai" }