ChatterBot 1.2.9__tar.gz → 1.2.11__tar.gz

{chatterbot-1.2.9 → chatterbot-1.2.11}/ChatterBot.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: ChatterBot
-Version: 1.2.9
+Version: 1.2.11
 Summary: ChatterBot is a machine learning, conversational dialog engine
 Author: Gunther Cox
 License-Expression: BSD-3-Clause
@@ -153,16 +153,10 @@ section of the documentation.
 
 See release notes for changes https://github.com/gunthercox/ChatterBot/releases
 
-# Development pattern for contributors
+# Contributing
 
-1. [Create a fork](https://help.github.com/articles/fork-a-repo/) of
-   the [main ChatterBot repository](https://github.com/gunthercox/ChatterBot) on GitHub.
-2. Make your changes in a branch named something different from `master`, e.g. create
-   a new branch `my-pull-request`.
-3. [Create a pull request](https://help.github.com/articles/creating-a-pull-request/).
-4. Please follow the [Python style guide for PEP-8](https://www.python.org/dev/peps/pep-0008/).
-5. Use the projects [built-in automated testing](https://docs.chatterbot.us/testing/).
-   to help make sure that your contribution is free from errors.
+Contributions are welcomed, to help ensure a smooth process please start with the contributing guidelines in our documentation:
+https://docs.chatterbot.us/contributing/
 
 # Sponsors
 

{chatterbot-1.2.9 → chatterbot-1.2.11}/ChatterBot.egg-info/SOURCES.txt

@@ -77,6 +77,7 @@ tests/test_benchmarks.py
 tests/test_chatbot.py
 tests/test_cli.py
 tests/test_comparisons.py
+tests/test_connection_pool.py
 tests/test_conversations.py
 tests/test_corpus.py
 tests/test_examples.py
@@ -84,6 +85,7 @@ tests/test_filters.py
 tests/test_initialization.py
 tests/test_languages.py
 tests/test_parsing.py
+tests/test_poc_vulnerability.py
 tests/test_preprocessors.py
 tests/test_response_selection.py
 tests/test_search.py

{chatterbot-1.2.9 → chatterbot-1.2.11}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: ChatterBot
-Version: 1.2.9
+Version: 1.2.11
 Summary: ChatterBot is a machine learning, conversational dialog engine
 Author: Gunther Cox
 License-Expression: BSD-3-Clause
@@ -153,16 +153,10 @@ section of the documentation.
 
 See release notes for changes https://github.com/gunthercox/ChatterBot/releases
 
-# Development pattern for contributors
+# Contributing
 
-1. [Create a fork](https://help.github.com/articles/fork-a-repo/) of
-   the [main ChatterBot repository](https://github.com/gunthercox/ChatterBot) on GitHub.
-2. Make your changes in a branch named something different from `master`, e.g. create
-   a new branch `my-pull-request`.
-3. [Create a pull request](https://help.github.com/articles/creating-a-pull-request/).
-4. Please follow the [Python style guide for PEP-8](https://www.python.org/dev/peps/pep-0008/).
-5. Use the projects [built-in automated testing](https://docs.chatterbot.us/testing/).
-   to help make sure that your contribution is free from errors.
+Contributions are welcomed, to help ensure a smooth process please start with the contributing guidelines in our documentation:
+https://docs.chatterbot.us/contributing/
 
 # Sponsors
 

{chatterbot-1.2.9 → chatterbot-1.2.11}/README.md

@@ -92,16 +92,10 @@ section of the documentation.
 
 See release notes for changes https://github.com/gunthercox/ChatterBot/releases
 
-# Development pattern for contributors
-
-1. [Create a fork](https://help.github.com/articles/fork-a-repo/) of
-   the [main ChatterBot repository](https://github.com/gunthercox/ChatterBot) on GitHub.
-2. Make your changes in a branch named something different from `master`, e.g. create
-   a new branch `my-pull-request`.
-3. [Create a pull request](https://help.github.com/articles/creating-a-pull-request/).
-4. Please follow the [Python style guide for PEP-8](https://www.python.org/dev/peps/pep-0008/).
-5. Use the projects [built-in automated testing](https://docs.chatterbot.us/testing/).
-   to help make sure that your contribution is free from errors.
+# Contributing
+
+Contributions are welcomed, to help ensure a smooth process please start with the contributing guidelines in our documentation:
+https://docs.chatterbot.us/contributing/
 
 # Sponsors
 

{chatterbot-1.2.9 → chatterbot-1.2.11}/chatterbot/__init__.py

@@ -4,7 +4,7 @@ ChatterBot is a machine learning, conversational dialog engine.
 from .chatterbot import ChatBot
 
 
-__version__ = '1.2.9'
+__version__ = '1.2.11'
 
 __all__ = (
     'ChatBot',

{chatterbot-1.2.9 → chatterbot-1.2.11}/chatterbot/chatterbot.py

@@ -2,7 +2,7 @@ import logging
 from typing import Union
 from chatterbot.storage import StorageAdapter
 from chatterbot.logic import LogicAdapter
-from chatterbot.search import TextSearch, IndexedTextSearch
+from chatterbot.search import TextSearch, IndexedTextSearch, SemanticVectorSearch
 from chatterbot.tagging import PosLemmaTagger
 from chatterbot.conversation import Statement
 from chatterbot import languages
@@ -74,41 +74,60 @@ class ChatBot(object):
 
         tagger_language = kwargs.get('tagger_language', languages.ENG)
 
-        try:
-            Tagger = kwargs.get('tagger', PosLemmaTagger)
-
-            # Allow instances to be provided for performance optimization
-            # (Example: a pre-loaded model in a tagger when unit testing)
-            if not isinstance(Tagger, type):
-                self.tagger = Tagger
-            else:
-                self.tagger = Tagger(language=tagger_language)
-        except IOError as io_error:
-            # Return a more helpful error message if possible
-            if "Can't find model" in str(io_error):
-                model_name = utils.get_model_for_language(tagger_language)
-                if hasattr(tagger_language, 'ENGLISH_NAME'):
-                    language_name = tagger_language.ENGLISH_NAME
+        # Check if storage adapter has a preferred tagger
+        PreferredTagger = self.storage.get_preferred_tagger()
+
+        if PreferredTagger is not None:
+            # Storage adapter specifies its own tagger
+            self.tagger = PreferredTagger(language=tagger_language)
+        else:
+            # Use default or user-specified tagger
+            try:
+                Tagger = kwargs.get('tagger', PosLemmaTagger)
+
+                # Allow instances to be provided for performance optimization
+                # (Example: a pre-loaded model in a tagger when unit testing)
+                if not isinstance(Tagger, type):
+                    self.tagger = Tagger
                 else:
-                    language_name = tagger_language
-                raise self.ChatBotException(
-                    'Setup error:\n'
-                    f'The Spacy model for "{language_name}" language is missing.\n'
-                    'Please install the model using the command:\n\n'
-                    f'python -m spacy download {model_name}\n\n'
-                    'See https://spacy.io/usage/models for more information about available models.'
-                ) from io_error
-            else:
-                raise io_error
+                    self.tagger = Tagger(language=tagger_language)
+            except IOError as io_error:
+                # Return a more helpful error message if possible
+                if "Can't find model" in str(io_error):
+                    model_name = utils.get_model_for_language(tagger_language)
+                    if hasattr(tagger_language, 'ENGLISH_NAME'):
+                        language_name = tagger_language.ENGLISH_NAME
+                    else:
+                        language_name = tagger_language
+                    raise self.ChatBotException(
+                        'Setup error:\n'
+                        f'The Spacy model for "{language_name}" language is missing.\n'
+                        'Please install the model using the command:\n\n'
+                        f'python -m spacy download {model_name}\n\n'
+                        'See https://spacy.io/usage/models for more information about available models.'
+                    ) from io_error
+                else:
+                    raise io_error
 
+        # Initialize search algorithms
         primary_search_algorithm = IndexedTextSearch(self, **kwargs)
         text_search_algorithm = TextSearch(self, **kwargs)
+        semantic_vector_search_algorithm = SemanticVectorSearch(self, **kwargs)
 
         self.search_algorithms = {
             primary_search_algorithm.name: primary_search_algorithm,
-            text_search_algorithm.name: text_search_algorithm
+            text_search_algorithm.name: text_search_algorithm,
+            semantic_vector_search_algorithm.name: semantic_vector_search_algorithm
         }
 
+        # Check if storage adapter has a preferred search algorithm
+        preferred_search_algorithm = self.storage.get_preferred_search_algorithm()
+        if preferred_search_algorithm and preferred_search_algorithm in self.search_algorithms:
+            # Set as default for logic adapters that don't specify their own search algorithm
+            # This ensures BestMatch and other adapters use the optimal search method
+            self.logger.info(f'Storage adapter prefers search algorithm: {preferred_search_algorithm}')
+            kwargs.setdefault('search_algorithm_name', preferred_search_algorithm)
+
         for adapter in logic_adapters:
             utils.validate_adapter_class(adapter, LogicAdapter)
             logic_adapter = utils.initialize_class(adapter, self, **kwargs)
@@ -191,15 +210,22 @@ class ChatBot(object):
                 input_statement.in_response_to = previous_statement.text
 
         # Make sure the input statement has its search text saved
-
-        if not input_statement.search_text:
-            _search_text = self.tagger.get_text_index_string(input_statement.text)
-            input_statement.search_text = _search_text
-
-        if not input_statement.search_in_response_to and input_statement.in_response_to:
-            input_statement.search_in_response_to = self.tagger.get_text_index_string(
-                input_statement.in_response_to
-            )
+        if not self.tagger.needs_text_indexing():
+            # Tagger doesn't transform text, use it directly
+            if not input_statement.search_text:
+                input_statement.search_text = input_statement.text
+            if not input_statement.search_in_response_to and input_statement.in_response_to:
+                input_statement.search_in_response_to = input_statement.in_response_to
+        else:
+            # Use tagger for text indexing or transformations
+            if not input_statement.search_text:
+                _search_text = self.tagger.get_text_index_string(input_statement.text)
+                input_statement.search_text = _search_text
+
+            if not input_statement.search_in_response_to and input_statement.in_response_to:
+                input_statement.search_in_response_to = self.tagger.get_text_index_string(
+                    input_statement.in_response_to
+                )
 
         response = self.generate_response(
             input_statement,

{chatterbot-1.2.9 → chatterbot-1.2.11}/chatterbot/search.py

@@ -157,3 +157,73 @@ class TextSearch:
                 if confidence >= 1.0:
                     self.chatbot.logger.info('Exact match found, stopping search')
                     break
+
+
+class SemanticVectorSearch:
+    """
+    Semantic vector search for storage adapters that use vector embeddings.
+    Does not require a tagger or comparison function - relies on the storage
+    adapter's native vector similarity search capabilities.
+
+    :param search_page_size:
+        The maximum number of records to load into memory at a time when searching.
+        Defaults to 1000
+    """
+
+    name = 'semantic_vector_search'
+
+    def __init__(self, chatbot, **kwargs):
+        self.chatbot = chatbot
+
+        self.search_page_size = kwargs.get(
+            'search_page_size', 1000
+        )
+
+    def search(self, input_statement, **additional_parameters):
+        """
+        Search for semantically similar statements using vector similarity.
+        Confidence scores are calculated by the storage adapter based on
+        vector distances and returned in the results.
+
+        :param input_statement: A statement.
+        :type input_statement: chatterbot.conversation.Statement
+
+        :param **additional_parameters: Additional parameters to be passed
+            to the ``filter`` method of the storage adapter when searching.
+
+        :rtype: Generator yielding one closest matching statement at a time.
+        """
+        self.chatbot.logger.info('Beginning semantic vector search')
+
+        search_parameters = {
+            'search_in_response_to_contains': input_statement.text,
+            'persona_not_startswith': 'bot:',
+            'page_size': self.search_page_size
+        }
+
+        if additional_parameters:
+            search_parameters.update(additional_parameters)
+
+        statement_list = self.chatbot.storage.filter(**search_parameters)
+
+        best_confidence_so_far = 0
+
+        self.chatbot.logger.info('Processing search results')
+
+        # Yield statements with confidence scores from vector similarity
+        for statement in statement_list:
+            # Confidence should already be set by the storage adapter
+            confidence = getattr(statement, 'confidence', 0.0)
+
+            if confidence > best_confidence_so_far:
+                best_confidence_so_far = confidence
+
+                self.chatbot.logger.info('Similar statement found: {} {}'.format(
+                    statement.in_response_to, confidence
+                ))
+
+                yield statement
+
+                if confidence >= 1.0:
+                    self.chatbot.logger.info('Exact match found, stopping search')
+                    break

{chatterbot-1.2.9 → chatterbot-1.2.11}/chatterbot/storage/redis.py

@@ -30,13 +30,19 @@ class RedisVectorStorageAdapter(StorageAdapter):
         in the future and its behavior has not yet been finalized.
 
     The RedisVectorStorageAdapter allows ChatterBot to store conversation
-    data in a redis instance.
+    data in a redis instance using vector embeddings for semantic similarity search.
 
     All parameters are optional, by default a redis instance on localhost is assumed.
 
     :keyword database_uri: eg: redis://localhost:6379/0',
         The database_uri can be specified to choose a redis instance.
     :type database_uri: str
+
+    NOTES:
+    * Unlike other database based storage adapters, the RedisVectorStorageAdapter
+      does not leverage `search_text` and `search_in_response_to` fields for indexing.
+      Instead, it uses vector embeddings to find similar statements based on
+      semantic similarity. This allows for more flexible and context-aware matching.
     """
 
     class RedisMetaDataType:
@@ -100,6 +106,21 @@ class RedisVectorStorageAdapter(StorageAdapter):
 
         self.vector_store = RedisVectorStore(embeddings, config=config)
 
+    def get_preferred_tagger(self):
+        """
+        Redis uses vector embeddings and doesn't need POS-lemma indexing.
+        Returns NoOpTagger to avoid unnecessary spaCy processing.
+        """
+        from chatterbot.tagging import NoOpTagger
+        return NoOpTagger
+
+    def get_preferred_search_algorithm(self):
+        """
+        Redis uses semantic vector search instead of text-based matching.
+        Returns the name of the SemanticVectorSearch algorithm.
+        """
+        return 'semantic_vector_search'
+
     def get_statement_model(self):
         """
         Return the statement model.
@@ -127,6 +148,16 @@ class RedisVectorStorageAdapter(StorageAdapter):
 
         values.update(document.metadata)
 
+        # Convert Unix timestamp back to datetime for StatementObject
+        # Redis may return this as int, float, or string representation
+        if 'created_at' in values:
+            created_at_value = values['created_at']
+            if isinstance(created_at_value, str):
+                # Convert string to float first
+                created_at_value = float(created_at_value)
+            if isinstance(created_at_value, (int, float)):
+                values['created_at'] = datetime.fromtimestamp(created_at_value)
+
         tags = values['tags']
         values['tags'] = list(set(tags.split('|') if tags else []))
 
@@ -177,6 +208,7 @@ class RedisVectorStorageAdapter(StorageAdapter):
             - exclude_text
             - exclude_text_words
             - persona_not_startswith
+            - search_text_contains
             - search_in_response_to_contains
             - order_by
         """
@@ -245,27 +277,26 @@ class RedisVectorStorageAdapter(StorageAdapter):
             else:
                 filter_condition = query
 
-        # Handle search_text parameter (used by BestMatch logic adapter)
-        # BestMatch uses search_text to find statements with matching indexed text.
-        # Since Redis doesn't store search_text as a field, we approximate this by:
-        # 1. Using the search_text value as a semantic query against in_response_to
-        # 2. This finds statements that are responses to similar inputs
-        # The effect is similar to BestMatch's Phase 2: finding alternate responses
-        if 'search_text' in kwargs:
-            _search_text = kwargs.get('search_text', '')
-
-            # Get embedding for the search text
-            # Note: search_text may be indexed (e.g., "NOUN:cat VERB:run") so this
-            # approximates finding responses to semantically similar queries
-            embedding = self.vector_store.embeddings.embed_query(_search_text)
+        if 'search_text_contains' in kwargs:
+            # Find statements whose text (responses) are similar.
+            #
+            # Use semantic similarity on the search query itself. This finds responses
+            # that would be semantically appropriate, even if they don't share exact words.
+            #
+            # Our vectors are of 'in_response_to' (what was said TO the bot),
+            # not 'text' (what the bot said). So we use the query as if it were an input,
+            # and find statements that would respond to similar inputs. The result is
+            # statements whose context (in_response_to) is similar, which tends to yield
+            # similar responses.
+            _search_query = kwargs['search_text_contains']
+
+            # Use vector similarity to find statements responding to similar contexts
+            embedding = self.vector_store.embeddings.embed_query(_search_query)
 
-            # Build return fields from metadata schema
             return_fields = [
                 'text', 'in_response_to', 'conversation', 'persona', 'tags', 'created_at'
             ]
 
-            # Use direct index query via RedisVL
-            # Search on the vectorized content (in_response_to) to find similar response patterns
             query = VectorQuery(
                 vector=embedding,
                 vector_field_name='embedding',
@@ -274,20 +305,35 @@ class RedisVectorStorageAdapter(StorageAdapter):
                 filter_expression=filter_condition
             )
 
-            # Execute query
             results = self.vector_store.index.query(query)
 
-            # Convert results to Document objects
             Document = self.get_statement_model()
             documents = []
-            for result in results:
-                # Extract metadata and content
+
+            # Calculate confidence from vector distances
+            # Results are ordered by similarity (best match first)
+            for idx, result in enumerate(results):
                 in_response_to = result.get('in_response_to', '')
 
-                # Convert created_at from integer (YYMMDD) to datetime
-                created_at_int = int(result.get('created_at', 0))
-                if created_at_int:
-                    created_at = datetime.strptime(str(created_at_int), '%y%m%d')
+                # Redis vector_score is cosine distance (lower is better)
+                # Convert to confidence: confidence = 1 - distance
+                # If vector_score not available, use result order
+                vector_score = result.get('vector_score')
+                if vector_score is not None:
+                    # Cosine distance ranges from 0 (identical) to 2 (opposite)
+                    # Normalize to confidence: 1.0 (identical) to 0.0 (opposite)
+                    confidence = max(0.0, 1.0 - (float(vector_score) / 2.0))
+                else:
+                    # Fallback: use result order (first result = highest confidence)
+                    # Start at 0.95 for first result, decay by 0.05 per position
+                    confidence = max(0.0, 0.95 - (idx * 0.05))
+
+                # Parse timestamp
+                created_at_value = result.get('created_at', 0)
+                if isinstance(created_at_value, str):
+                    created_at = datetime.fromtimestamp(float(created_at_value))
+                elif created_at_value:
+                    created_at = datetime.fromtimestamp(float(created_at_value))
                 else:
                     created_at = datetime.now()
 
@@ -297,6 +343,7 @@ class RedisVectorStorageAdapter(StorageAdapter):
                     'persona': result.get('persona', ''),
                     'tags': result.get('tags', ''),
                     'created_at': created_at,
+                    'confidence': confidence,
                 }
                 doc = Document(
                     page_content=in_response_to,
@@ -307,6 +354,23 @@ class RedisVectorStorageAdapter(StorageAdapter):
 
             return [self.model_to_object(document) for document in documents]
 
+        # Redis uses vector similarity: we search for statements whose actual
+        # text field is semantically similar to the text that produced this search_text.
+        # This is stored in the closest_match.text field, but BestMatch only passes
+        # search_text. Since we can't reverse POS tags to original text (for now),
+        # we treat this parameter as a signal to do text-based similarity search.
+        #
+        # Note: The caller should ideally pass the actual text, but for compatibility
+        # we'll work with what we receive. In practice, search_text_contains is the
+        # better parameter for this use case.
+        if 'search_text' in kwargs:
+            # For now, we'll treat search_text as a filter-only parameter
+            # and fall through to the regular query_search below.
+            # This prevents the broken behavior of embedding POS tags.
+            # The proper fix requires BestMatch to pass additional context
+            # or use search_text_contains instead.
+            pass
+
         ordering = kwargs.get('order_by', None)
 
         if ordering:
@@ -341,14 +405,31 @@ class RedisVectorStorageAdapter(StorageAdapter):
             # Convert results to Document objects
             Document = self.get_statement_model()
             documents = []
-            for result in results:
+
+            # Calculate confidence from vector distances
+            # Results are ordered by similarity (best match first)
+            for idx, result in enumerate(results):
                 # Extract metadata and content
                 in_response_to = result.get('in_response_to', '')
 
-                # Convert created_at from integer (YYMMDD) to datetime
-                created_at_int = int(result.get('created_at', 0))
-                if created_at_int:
-                    created_at = datetime.strptime(str(created_at_int), '%y%m%d')
+                # Redis vector_score is cosine distance (lower is better)
+                # Convert to confidence: confidence = 1 - distance
+                # If vector_score not available, use result order
+                vector_score = result.get('vector_score')
+                if vector_score is not None:
+                    # Cosine distance ranges from 0 (identical) to 2 (opposite)
+                    # Normalize to confidence: 1.0 (identical) to 0.0 (opposite)
+                    confidence = max(0.0, 1.0 - (float(vector_score) / 2.0))
+                else:
+                    # Fallback: use result order (first result = highest confidence)
+                    # Start at 0.95 for first result, decay by 0.05 per position
+                    confidence = max(0.0, 0.95 - (idx * 0.05))
+
+                # Convert Unix timestamp back to datetime
+                # Redis returns numeric fields as strings
+                created_at_timestamp = result.get('created_at', '0')
+                if created_at_timestamp and created_at_timestamp != '0':
+                    created_at = datetime.fromtimestamp(float(created_at_timestamp))
                 else:
                     created_at = datetime.now()
 
@@ -358,6 +439,7 @@ class RedisVectorStorageAdapter(StorageAdapter):
                     'persona': result.get('persona', ''),
                     'tags': result.get('tags', ''),
                     'created_at': created_at,
+                    'confidence': confidence,
                 }
                 doc = Document(
                     page_content=in_response_to,
@@ -395,9 +477,9 @@ class RedisVectorStorageAdapter(StorageAdapter):
         metadata = {
             'text': text,
             'category': kwargs.get('category', ''),
-            # NOTE: `created_at` must have a valid numeric value or results will
-            # not be returned for similarity_search for some reason
-            'created_at': kwargs.get('created_at') or int(_default_date.strftime('%y%m%d')),
+            # Store created_at as Unix timestamp with microseconds (float)
+            # This provides full datetime precision while maintaining Redis NUMERIC field compatibility
+            'created_at': kwargs.get('created_at') or _default_date.timestamp(),
             'tags': '|'.join(unique_tags) if unique_tags else '',
             'conversation': kwargs.get('conversation', ''),
             'persona': kwargs.get('persona', ''),
@@ -427,7 +509,7 @@ class RedisVectorStorageAdapter(StorageAdapter):
                 metadata={
                     'text': statement.text,
                     'conversation': statement.conversation or '',
-                    'created_at': int(statement.created_at.strftime('%y%m%d')),
+                    'created_at': statement.created_at.timestamp(),
                     'persona': statement.persona or '',
                     # Prevent duplicate tag entries in the database
                     'tags': '|'.join(
@@ -452,7 +534,7 @@ class RedisVectorStorageAdapter(StorageAdapter):
         metadata = {
             'text': statement.text,
             'conversation': statement.conversation or '',
-            'created_at': int(statement.created_at.strftime('%y%m%d')),
+            'created_at': statement.created_at.timestamp(),
             'persona': statement.persona or '',
             'tags': '|'.join(unique_tags) if unique_tags else '',
         }
@@ -508,11 +590,9 @@ class RedisVectorStorageAdapter(StorageAdapter):
                 # Parse the metadata
                 metadata = json.loads(data[b'_metadata_json'].decode())
 
-                # Convert created_at from integer (YYMMDD) back to datetime
-                if 'created_at' in metadata and isinstance(metadata['created_at'], int):
-                    created_at_str = str(metadata['created_at'])
-                    # Parse YYMMDD format
-                    metadata['created_at'] = datetime.strptime(created_at_str, '%y%m%d')
+                # Convert created_at from Unix timestamp back to datetime
+                if 'created_at' in metadata and isinstance(metadata['created_at'], (int, float)):
+                    metadata['created_at'] = datetime.fromtimestamp(metadata['created_at'])
 
                 # Get the in_response_to from the hash
                 in_response_to = data.get(b'in_response_to', b'').decode()