dwani 0.1.20__tar.gz → 0.1.22__tar.gz

{dwani-0.1.20 → dwani-0.1.22}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: dwani
-Version: 0.1.20
+Version: 0.1.22
 Summary: Multimodal API for Indian + European languages (Chat, Vision, TTS, ASR, Translate, Docs)
 Author-email: sachin <python@dwani.ai>
 License: MIT License

{dwani-0.1.20 → dwani-0.1.22}/dwani/__init__.py

@@ -72,10 +72,16 @@ class document:
     @staticmethod
     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 query_page(file_path, page_number=1,prompt="list the key points", query_lang="eng_Latn", tgt_lang="kan_Knda", model="gemma3"):
+        return _get_client().query_page(file_path, page_number, prompt, query_lang, 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)
+    def query_all(file_path, prompt="list the key points", query_lang="eng_Latn", tgt_lang="kan_Knda", model="gemma3"):
+        return _get_client().query_all(file_path, prompt, query_lang, tgt_lang, model)
     
     @staticmethod
     def run_doc_query_kannada(file_path, page_number=1, prompt="list key points", src_lang="kan_Latn", tgt_lang="kan_Knda", model="gemma3"):

{dwani-0.1.20 → dwani-0.1.22}/dwani/client.py

@@ -59,9 +59,14 @@ class DwaniClient:
         from .docs import extract
         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
-        return doc_query(self, file_path, page_number=page_number, prompt=prompt, src_lang=src_lang, tgt_lang=tgt_lang, model=model)
+    def query_page(self, file_path, page_number=1, prompt="list the key points", query_lang="eng_Latn", tgt_lang="kan_Knda", model="gemma3"):
+        from .docs import query_page
+        return query_page(self, file_path, page_number=page_number, prompt=prompt, query_lang=query_lang, tgt_lang=tgt_lang, model=model)
+
+    def query_all(self, file_path, prompt="list the key points", query_lang="eng_Latn", tgt_lang="kan_Knda", model="gemma3"):
+        from .docs import query_all
+        return query_all(self, file_path, prompt=prompt, query_lang=query_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="kan_Knda", model="gemma3"):
         from .docs import doc_query_kannada

{dwani-0.1.20 → dwani-0.1.22}/dwani/docs.py

@@ -10,7 +10,14 @@ logger = logging.getLogger(__name__)
 language_options = [
     ("English", "eng_Latn"),
     ("Kannada", "kan_Knda"),
-    ("Hindi", "hin_Deva"),
+    ("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"),
     ("German", "deu_Latn") 
@@ -53,6 +60,8 @@ def document_ocr_all(client, file_path, model="gemma3"):
         try:
             resp = requests.post(
                 f"{client.api_base}/v1/extract-text-all",
+#TODO - test -chunk
+#                f"{client.api_base}/v1/extract-text-all-chunk",
                 headers=client._headers(),
                 files=files,
                 data=data,
@@ -211,13 +220,13 @@ def extract(client, file_path, page_number=1, tgt_lang="kan_Knda", model="gemma3
 
     return resp.json()
 
-def doc_query(
+def query_page(
     client,
     file_path,
     page_number=1,
     prompt="list the key points",
     tgt_lang="kan_Knda",
-    src_lang="eng_Latn",
+    query_lang="eng_Latn",
     model="gemma3"
 ):
     """Query a document with a custom prompt and language options."""
@@ -233,7 +242,8 @@ def doc_query(
     
     tgt_lang_code = normalize_language(tgt_lang)
     
-    src_lang_code = normalize_language(src_lang)
+    query_lang_code = normalize_language(query_lang)
+
 
     url = f"{client.api_base}/v1/indic-custom-prompt-pdf"
     headers = client._headers()
@@ -243,7 +253,7 @@ def doc_query(
             "page_number": str(page_number),
             "prompt": prompt,
             "tgt_lang": tgt_lang_code,
-            "src_lang": src_lang_code,
+            "query_lang": query_lang_code,
             "model": model
         }
 
@@ -253,6 +263,7 @@ def doc_query(
                 headers=headers,
                 files=files,
                 data=data,
+                #params=params,
                 timeout=90
             )
             resp.raise_for_status()
@@ -264,6 +275,56 @@ def doc_query(
 
     return resp.json()
 
+def query_all(
+    client,
+    file_path,
+    prompt="list the key points",
+    tgt_lang="kan_Knda",
+    query_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}, prompt={prompt}, tgt_lang={tgt_lang}, model={model}")
+    validate_model(model)
+    
+    if not file_path.lower().endswith('.pdf'):
+        raise ValueError("File must be a PDF")
+    if not prompt.strip():
+        raise ValueError("Prompt cannot be empty")
+    
+    tgt_lang_code = normalize_language(tgt_lang)
+    
+    query_lang_code = normalize_language(query_lang)
+
+    url = f"{client.api_base}/v1/indic-custom-prompt-pdf-all"
+    headers = client._headers()
+    with open(file_path, "rb") as f:
+        files = {"file": (file_path, f, "application/pdf")}
+        data = {
+            "prompt": prompt,
+            "tgt_lang": tgt_lang_code,
+            "query_lang": query_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:
+            logger.error(f"Doc query request failed: {str(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(
     client,
     file_path,
@@ -320,43 +381,50 @@ def doc_query_kannada(
 
 class Documents:
     @staticmethod
-    def run_ocr_number(file_path, page_number=2,model="gemma3"):
+    def run_ocr_number(file_path, page_number=1,model="gemma3"):
         from .client import DwaniClient
         client = DwaniClient()
-        return document_ocr_page(client, file_path, page_number, model)
+        return document_ocr_page(client, file_path=file_path, page_number=page_number, model=model)
     @staticmethod
     def run_ocr_all(file_path, model="gemma3"):
         from .client import DwaniClient
         client = DwaniClient()
-        return document_ocr_all(client, file_path, model)
+        return document_ocr_all(client, file_path=file_path, model=model)
     
     @staticmethod
     def summarize_page(file_path, page_number=1, tgt_lang="kan_Knda", model="gemma3"):
         from .client import DwaniClient
         client = DwaniClient()
-        return document_summarize_page(client, file_path, page_number, tgt_lang, model)
+        return document_summarize_page(client, file_path=file_path, page_number=page_number, tgt_lang=tgt_lang, model=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)
+        return document_summarize_all(client, file_path=file_path, tgt_lang=tgt_lang, model=model)
 
     @staticmethod
     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, tgt_lang, model)
+        return extract(client, file_path=file_path, page_number=page_number, tgt_lang=tgt_lang, model=model)
+    
+    @staticmethod
+    def query_page(file_path, page_number=1, prompt="list the key points", query_lang="eng_Latn", tgt_lang="kan_Knda", model="gemma3"):
+        from .client import DwaniClient
+        client = DwaniClient()
+        return query_page(client, file_path=file_path, page_number=page_number, prompt=prompt, query_lang=query_lang, tgt_lang=tgt_lang, model=model)
     
     @staticmethod
-    def run_doc_query(file_path, page_number=1, prompt="list the key points", tgt_lang="kan_Knda", model="gemma3"):
+    def query_all(file_path, prompt="list the key points", query_lang="eng_Latn", tgt_lang="kan_Knda", model="gemma3"):
         from .client import DwaniClient
         client = DwaniClient()
-        return doc_query(client, file_path, page_number, prompt, tgt_lang, model)
+        return query_all(client, file_path=file_path, prompt=prompt, query_lang=query_lang, tgt_lang=tgt_lang, model=model)
     
+
     @staticmethod
     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, tgt_lang, model)
+        return doc_query_kannada(client, file_path=file_path, page_number=page_number, prompt=prompt, tgt_lang=tgt_lang, model=model)

{dwani-0.1.20 → dwani-0.1.22}/pyproject.toml

@@ -5,7 +5,7 @@ build-backend = "setuptools.build_meta"
 [project]
 name = "dwani"
 
-version = "0.1.20"
+version = "0.1.22"
 description = "Multimodal API for Indian + European languages (Chat, Vision, TTS, ASR, Translate, Docs)"
 authors = [
     { name="sachin", email="python@dwani.ai" }