hash_funcs for st.cache_data and st.cache_resource

lib/streamlit/runtime/caching/cache_data_api.py

@@ -44,6 +44,7 @@
     MsgData,
     MultiCacheResults,
 )
+from streamlit.runtime.caching.hashing import HashFuncsDict
 from streamlit.runtime.caching.storage import (
     CacheStorage,
     CacheStorageContext,
@@ -80,11 +81,13 @@ def __init__(
         max_entries: int | None,
         ttl: float | timedelta | str | None,
         allow_widgets: bool,
+        hash_funcs: HashFuncsDict | None = None,
     ):
         super().__init__(
             func,
             show_spinner=show_spinner,
             allow_widgets=allow_widgets,
+            hash_funcs=hash_funcs,
         )
         self.persist = persist
         self.max_entries = max_entries
@@ -365,6 +368,7 @@ def __call__(
         show_spinner: bool | str = True,
         persist: CachePersistType | bool = None,
         experimental_allow_widgets: bool = False,
+        hash_funcs: HashFuncsDict | None = None,
     ):
         return self._decorator(
             func,
@@ -373,6 +377,7 @@ def __call__(
             persist=persist,
             show_spinner=show_spinner,
             experimental_allow_widgets=experimental_allow_widgets,
+            hash_funcs=hash_funcs,
         )
 
     def _decorator(
@@ -384,6 +389,7 @@ def _decorator(
         show_spinner: bool | str,
         persist: CachePersistType | bool,
         experimental_allow_widgets: bool,
+        hash_funcs: HashFuncsDict | None = None,
     ):
         """Decorator to cache functions that return data (e.g. dataframe transforms, database queries, ML inference).
 
@@ -529,6 +535,7 @@ def wrapper(f):
                     max_entries=max_entries,
                     ttl=ttl,
                     allow_widgets=experimental_allow_widgets,
+                    hash_funcs=hash_funcs,
                 )
             )
 
@@ -543,6 +550,7 @@ def wrapper(f):
                 max_entries=max_entries,
                 ttl=ttl,
                 allow_widgets=experimental_allow_widgets,
+                hash_funcs=hash_funcs,
             )
         )
 

lib/streamlit/runtime/caching/cache_resource_api.py

@@ -45,6 +45,7 @@
     MsgData,
     MultiCacheResults,
 )
+from streamlit.runtime.caching.hashing import HashFuncsDict
 from streamlit.runtime.metrics_util import gather_metrics
 from streamlit.runtime.scriptrunner.script_run_context import get_script_run_ctx
 from streamlit.runtime.stats import CacheStat, CacheStatsProvider
@@ -153,11 +154,13 @@ def __init__(
         ttl: float | timedelta | str | None,
         validate: ValidateFunc | None,
         allow_widgets: bool,
+        hash_funcs: HashFuncsDict | None = None,
     ):
         super().__init__(
             func,
             show_spinner=show_spinner,
             allow_widgets=allow_widgets,
+            hash_funcs=hash_funcs,
         )
         self.max_entries = max_entries
         self.ttl = ttl
@@ -245,6 +248,7 @@ def __call__(
         show_spinner: bool | str = True,
         validate: ValidateFunc | None = None,
         experimental_allow_widgets: bool = False,
+        hash_funcs: HashFuncsDict | None = None,
     ):
         return self._decorator(
             func,
@@ -253,6 +257,7 @@ def __call__(
             show_spinner=show_spinner,
             validate=validate,
             experimental_allow_widgets=experimental_allow_widgets,
+            hash_funcs=hash_funcs,
         )
 
     def _decorator(
@@ -264,6 +269,7 @@ def _decorator(
         show_spinner: bool | str,
         validate: ValidateFunc | None,
         experimental_allow_widgets: bool,
+        hash_funcs: HashFuncsDict | None = None,
     ):
         """Decorator to cache functions that return global resources (e.g. database connections, ML models).
 
@@ -389,6 +395,7 @@ def _decorator(
                     ttl=ttl,
                     validate=validate,
                     allow_widgets=experimental_allow_widgets,
+                    hash_funcs=hash_funcs,
                 )
             )
 
@@ -400,6 +407,7 @@ def _decorator(
                 ttl=ttl,
                 validate=validate,
                 allow_widgets=experimental_allow_widgets,
+                hash_funcs=hash_funcs,
             )
         )
 

lib/streamlit/runtime/caching/cache_utils.py

@@ -50,7 +50,7 @@
     MsgData,
     replay_cached_messages,
 )
-from streamlit.runtime.caching.hashing import update_hash
+from streamlit.runtime.caching.hashing import HashFuncsDict, update_hash
 
 _LOGGER = get_logger(__name__)
 
@@ -169,10 +169,12 @@ def __init__(
         func: types.FunctionType,
         show_spinner: bool | str,
         allow_widgets: bool,
+        hash_funcs: HashFuncsDict | None,
     ):
         self.func = func
         self.show_spinner = show_spinner
         self.allow_widgets = allow_widgets
+        self.hash_funcs = hash_funcs
 
     @property
     def cache_type(self) -> CacheType:
@@ -254,6 +256,7 @@ def _get_or_create_cached_value(
             func=self._info.func,
             func_args=func_args,
             func_kwargs=func_kwargs,
+            hash_funcs=self._info.hash_funcs,
         )
 
         try:
@@ -351,6 +354,7 @@ def _make_value_key(
     func: types.FunctionType,
     func_args: tuple[Any, ...],
     func_kwargs: dict[str, Any],
+    hash_funcs: HashFuncsDict | None,
 ) -> str:
     """Create the key for a value within a cache.
 
@@ -388,9 +392,20 @@ def _make_value_key(
 
         try:
             update_hash(
-                (arg_name, arg_value),
+                arg_name,
                 hasher=args_hasher,
                 cache_type=cache_type,
+                hash_source=func,
+            )
+            # we call update_hash twice here, first time for `arg_name`
+            # without `hash_funcs`, and second time for `arg_value` with hash_funcs
+            # to evaluate user defined `hash_funcs` only for computing `arg_value` hash.
+            update_hash(
+                arg_value,
+                hasher=args_hasher,
+                cache_type=cache_type,
+                hash_funcs=hash_funcs,
+                hash_source=func,
             )
         except UnhashableTypeError as exc:
             raise UnhashableParamError(cache_type, func, arg_name, arg_value, exc)
@@ -417,6 +432,7 @@ def _make_function_key(cache_type: CacheType, func: types.FunctionType) -> str:
         (func.__module__, func.__qualname__),
         hasher=func_hasher,
         cache_type=cache_type,
+        hash_source=func,
     )
 
     # Include the function's source code in its hash. If the source code can't
@@ -432,9 +448,7 @@ def _make_function_key(cache_type: CacheType, func: types.FunctionType) -> str:
         source_code = func.__code__.co_code
 
     update_hash(
-        source_code,
-        hasher=func_hasher,
-        cache_type=cache_type,
+        source_code, hasher=func_hasher, cache_type=cache_type, hash_source=func
     )
 
     cache_key = func_hasher.hexdigest()