langchain-timbr 2.1.2__tar.gz → 2.1.4__tar.gz

{langchain_timbr-2.1.2 → langchain_timbr-2.1.4}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: langchain-timbr
-Version: 2.1.2
+Version: 2.1.4
 Summary: LangChain & LangGraph extensions that parse LLM prompts into Timbr semantic SQL and execute them.
 Project-URL: Homepage, https://github.com/WPSemantix/langchain-timbr
 Project-URL: Documentation, https://docs.timbr.ai/doc/docs/integration/langchain-sdk/

{langchain_timbr-2.1.2 → langchain_timbr-2.1.4}/src/langchain_timbr/_version.py

@@ -28,7 +28,7 @@ version_tuple: VERSION_TUPLE
 commit_id: COMMIT_ID
 __commit_id__: COMMIT_ID
 
-__version__ = version = '2.1.2'
-__version_tuple__ = version_tuple = (2, 1, 2)
+__version__ = version = '2.1.4'
+__version_tuple__ = version_tuple = (2, 1, 4)
 
 __commit_id__ = commit_id = None

{langchain_timbr-2.1.2 → langchain_timbr-2.1.4}/src/langchain_timbr/langchain/execute_timbr_query_chain.py

@@ -169,6 +169,7 @@ class ExecuteTimbrQueryChain(Chain):
             "verify_ssl": self._verify_ssl,
             "is_jwt": self._is_jwt,
             "jwt_tenant_id": self._jwt_tenant_id,
+            "additional_headers": {"results-limit": str(self._max_limit)},
             **self._conn_params,
         }
 
@@ -265,7 +266,8 @@ class ExecuteTimbrQueryChain(Chain):
                 rows = run_query(
                     sql,
                     self._get_conn_params(),
-                    llm_prompt=prompt
+                    llm_prompt=prompt,
+                    use_query_limit=True,
                 ) if is_sql_valid and is_sql_not_tried else []
                 
                 if iteration < self._no_results_max_retries:

{langchain_timbr-2.1.2 → langchain_timbr-2.1.4}/src/langchain_timbr/langchain/generate_answer_chain.py

@@ -23,6 +23,7 @@ class GenerateAnswerChain(Chain):
         is_jwt: Optional[bool] = False,
         jwt_tenant_id: Optional[str] = None,
         conn_params: Optional[dict] = None,
+        note: Optional[str] = '',
         debug: Optional[bool] = False,
         **kwargs,
     ):
@@ -33,6 +34,7 @@ class GenerateAnswerChain(Chain):
         :param verify_ssl: Whether to verify SSL certificates (default is True).
         :param is_jwt: Whether to use JWT authentication (default is False).
         :param jwt_tenant_id: JWT tenant ID for multi-tenant environments (required when is_jwt=True).
+        :param note: Optional additional note to extend our llm prompt
         :param conn_params: Extra Timbr connection parameters sent with every request (e.g., 'x-api-impersonate-user').
         
         ## Example
@@ -77,6 +79,7 @@ class GenerateAnswerChain(Chain):
         self._jwt_tenant_id = jwt_tenant_id
         self._debug = to_boolean(debug)
         self._conn_params = conn_params or {}
+        self._note = note
 
 
     @property
@@ -116,6 +119,7 @@ class GenerateAnswerChain(Chain):
             conn_params=self._get_conn_params(),
             results=rows,
             sql=sql,
+            note=self._note,
             debug=self._debug,
         )
 

{langchain_timbr-2.1.2 → langchain_timbr-2.1.4}/src/langchain_timbr/langchain/generate_timbr_sql_chain.py

@@ -160,6 +160,7 @@ class GenerateTimbrSqlChain(Chain):
             "verify_ssl": self._verify_ssl,
             "is_jwt": self._is_jwt,
             "jwt_tenant_id": self._jwt_tenant_id,
+            "additional_headers": {"results-limit": str(self._max_limit)},
             **self._conn_params,
         }
 

{langchain_timbr-2.1.2 → langchain_timbr-2.1.4}/src/langchain_timbr/langchain/timbr_sql_agent.py

@@ -126,6 +126,7 @@ class TimbrSqlAgent(BaseSingleActionAgent):
             is_jwt=to_boolean(is_jwt),
             jwt_tenant_id=jwt_tenant_id,
             conn_params=conn_params,
+            note=note,
             debug=to_boolean(debug),
         ) if self._generate_answer else None
 

{langchain_timbr-2.1.2 → langchain_timbr-2.1.4}/src/langchain_timbr/langchain/validate_timbr_sql_chain.py

@@ -161,6 +161,7 @@ class ValidateTimbrSqlChain(Chain):
             "verify_ssl": self._verify_ssl,
             "is_jwt": self._is_jwt,
             "jwt_tenant_id": self._jwt_tenant_id,
+            "additional_headers": {"results-limit": str(self._max_limit)},
             **self._conn_params,
         }
 

{langchain_timbr-2.1.2 → langchain_timbr-2.1.4}/src/langchain_timbr/llm_wrapper/llm_wrapper.py

@@ -203,7 +203,7 @@ class LlmWrapper(LLM):
       if azure_tenant_id and azure_client_id:
         from azure.identity import ClientSecretCredential, get_bearer_token_provider
         azure_client_secret = pop_param_value(params, ['azure_client_secret', 'llm_client_secret'], default=api_key)
-        scope = pop_param_value(params, ['azure_scope', 'llm_scope'], default=config.llm_scope)
+        scope = pop_param_value(params, ['azure_scope', 'llm_scope', 'scope'], default=config.llm_scope)
         credential = ClientSecretCredential(
           tenant_id=azure_tenant_id,
           client_id=azure_client_id,

{langchain_timbr-2.1.2 → langchain_timbr-2.1.4}/src/langchain_timbr/timbr_llm_connector.py

@@ -115,6 +115,7 @@ class TimbrLlmConnector:
             "verify_ssl": self.verify_ssl,
             "is_jwt": self.is_jwt,
             "jwt_tenant_id": self.jwt_tenant_id,
+            "additional_headers": {"results-limit": str(self.max_limit)},
             **self.conn_params,
         }
 

{langchain_timbr-2.1.2 → langchain_timbr-2.1.4}/src/langchain_timbr/utils/general.py

@@ -1,6 +1,7 @@
 import os
 from typing import Any, Optional, Union
 import json
+import re
 
 ### A global helper functions to use across the project
 
@@ -34,6 +35,7 @@ def to_integer(value) -> int:
 def parse_additional_params(value) -> dict:
     """
     Parse additional parameters from string format 'a=1,b=2' or return dict as-is.
+    Handles JSON values correctly, including nested structures with commas.
     
     Args:
         value: String in format 'key=value,key2=value2', JSON string, or dict
@@ -49,25 +51,156 @@ def parse_additional_params(value) -> dict:
             stripped_value = value.strip()
             if stripped_value.startswith('{') and stripped_value.endswith('}'):
                 try:
-                    return json.loads(stripped_value)
+                    return _try_parse_json_value(stripped_value)
                 except json.JSONDecodeError:
                     pass
             
-            # Fall back to key=value parsing
+            # Check if complex parsing is needed (presence of nested structures)
+            needs_complex_parsing = any(char in value for char in ['{', '}', '[', ']', '(', ')'])
+            
+            if not needs_complex_parsing:
+                # Fast path: simple key=value pairs
+                params = {}
+                for pair in value.split(','):
+                    if '=' in pair:
+                        key, val = pair.split('=', 1)
+                        params[key.strip().lower()] = _try_parse_json_value(val.strip())
+                return params
+            
+            # Complex parsing that handles JSON values with commas
             params = {}
-            for pair in (value.split('&') if '&' in value else value.split(',')):
-                if '=' in pair:
-                    key, val = pair.split('=', 1)
-                    params[key.strip().lower()] = val.strip()
-                elif ':' in pair:
-                    key, val = pair.split(':', 1)
-                    params[key.strip().lower()] = val.strip()
+            i = 0
+            while i < len(value):
+                # Find the next key=value pair
+                equals_pos = value.find('=', i)
+                if equals_pos == -1:
+                    break
+                
+                # Extract the key
+                key = value[i:equals_pos].strip()
+                
+                # Find where the value starts
+                value_start = equals_pos + 1
+                
+                # Determine where the value ends (considering nested structures)
+                value_end = _find_value_end(value, value_start)
+                
+                # Extract and parse the value
+                val = value[value_start:value_end].strip()
+                
+                # Try to parse the value as JSON if it looks like JSON
+                parsed_val = _try_parse_json_value(val)
+                
+                params[key.lower()] = parsed_val
+                
+                # Move to the next parameter (skip comma if present)
+                i = value_end
+                if i < len(value) and value[i] == ',':
+                    i += 1
+            
             return params
         return {}
     except Exception as e:
         raise ValueError(f"Failed to parse additional parameters: {e}")
 
 
+def _find_value_end(text: str, start: int) -> int:
+    """
+    Find the end position of a parameter value, considering nested structures.
+    
+    Args:
+        text: The full text being parsed
+        start: Starting position of the value
+        
+    Returns:
+        End position of the value
+    """
+    depth = {'brace': 0, 'bracket': 0, 'paren': 0}
+    in_quotes = False
+    quote_char = None
+    i = start
+    
+    while i < len(text):
+        char = text[i]
+        
+        # Handle quotes
+        if char in ('"', "'"):
+            if not in_quotes:
+                in_quotes = True
+                quote_char = char
+            elif char == quote_char:
+                # Check if it's escaped
+                if i > 0 and text[i-1] != '\\':
+                    in_quotes = False
+                    quote_char = None
+        
+        # Only process structural characters if not in quotes
+        elif not in_quotes:
+            if char == '{':
+                depth['brace'] += 1
+            elif char == '}':
+                depth['brace'] -= 1
+            elif char == '[':
+                depth['bracket'] += 1
+            elif char == ']':
+                depth['bracket'] -= 1
+            elif char == '(':
+                depth['paren'] += 1
+            elif char == ')':
+                depth['paren'] -= 1
+            elif char == ',' and all(d == 0 for d in depth.values()):
+                # Found a comma at the top level - this is the end of the value
+                return i
+        
+        i += 1
+    
+    return i
+
+
+def _try_parse_json_value(val: str) -> Any:
+    """
+    Try to parse a string value as JSON. If it fails, return the original string.
+    Supports both single and double quotes for JSON-like structures.
+    
+    Args:
+        val: String value to parse
+        
+    Returns:
+        Parsed JSON value or original string
+    """
+    val = val.strip()
+    
+    # Check if it looks like JSON
+    if (val.startswith('{') and val.endswith('}')) or \
+       (val.startswith('[') and val.endswith(']')) or \
+       val in ('true', 'false', 'null') or \
+       (val.startswith('"') and val.endswith('"')):
+        try:
+            return json.loads(val)
+        except json.JSONDecodeError:
+            # Try converting single quotes to double quotes for Python-style dicts
+            if val.startswith('{') or val.startswith('['):
+                try:
+                    # Replace single quotes with double quotes, handling escaped quotes
+                    normalized = val.replace("\\'", "<<<ESCAPED_SINGLE>>>")
+                    normalized = normalized.replace("'", '"')
+                    normalized = normalized.replace("<<<ESCAPED_SINGLE>>>", "'")
+                    return json.loads(normalized)
+                except json.JSONDecodeError:
+                    pass
+    # Try to parse as number - validate format first
+    if re.match(r'^-?\d+(\.\d+)?$', val):
+        try:
+            if '.' in val:
+                return float(val)
+            return int(val)
+        except ValueError:
+            pass
+        pass
+    
+    return val
+
+
 def is_llm_type(llm_type, enum_value):
     """Check if llm_type equals the enum value or its name, case-insensitive."""
     if llm_type == enum_value:

{langchain_timbr-2.1.2 → langchain_timbr-2.1.4}/src/langchain_timbr/utils/timbr_llm_utils.py

@@ -545,8 +545,9 @@ def answer_question(
     conn_params: dict,
     results: str,
     sql: Optional[str] = None,
-    debug: Optional[bool] = False,
     timeout: Optional[int] = None,
+    note: Optional[str] = '',
+    debug: Optional[bool] = False,
 ) -> dict[str, Any]:
     # Use config default timeout if none provided
     if timeout is None:
@@ -558,6 +559,7 @@ def answer_question(
         question=question,
         formatted_rows=results,
         additional_context=f"SQL QUERY:\n{sql}\n\n" if sql else "",
+        note=note,
     )
     
     apx_token_count = _calculate_token_count(llm, prompt)

{langchain_timbr-2.1.2 → langchain_timbr-2.1.4}/src/langchain_timbr/utils/timbr_utils.py

@@ -70,7 +70,7 @@ def cache_with_version_check(func):
     return wrapper
 
 
-def run_query(sql: str, conn_params: dict, llm_prompt: Optional[str] = None) -> list[list]:
+def run_query(sql: str, conn_params: dict, llm_prompt: Optional[str] = None, use_query_limit = False) -> list[list]:
     if not conn_params:
         raise("Please provide connection params.")
 
@@ -79,9 +79,22 @@ def run_query(sql: str, conn_params: dict, llm_prompt: Optional[str] = None) ->
         clean_prompt = llm_prompt.replace('\r\n', ' ').replace('\n', ' ').replace('?', '')
         query = f"-- LLM: {clean_prompt}\n{sql}"
 
+    query_conn_params = conn_params
+    if not use_query_limit:
+        # Remove results-limit
+        if 'additional_headers' in conn_params and 'results-limit' in conn_params['additional_headers']:
+            query_upper = query.strip().upper()
+            if query_upper.startswith('SHOW') or query_upper.startswith('DESC'):
+                query_conn_params = conn_params.copy()
+                query_conn_params['additional_headers'] = conn_params['additional_headers'].copy()
+                del query_conn_params['additional_headers']['results-limit']
+                # If no other additional_headers remain, delete the key entirely
+                if not query_conn_params['additional_headers']:
+                    del query_conn_params['additional_headers']
+
     results = timbr_http_connector.run_query(
         query=query,
-        **conn_params,
+        **query_conn_params,
     )
 
     return results