✨ add needs_import_cache_size configuration

docs/configuration.rst

@@ -1645,7 +1645,18 @@ keys:
             The related CSS class definition must be done by the user, e.g. by :ref:`own_css`.
             (*optional*) (*default*: ``external_link``)
 
+.. _needs_import_cache_size:
 
+needs_import_cache_size
+~~~~~~~~~~~~~~~~~~~~~~~
+
+.. versionadded:: 3.1.0
+
+Sets the maximum number of needs cached by the :ref:`needimport` directive,
+which is used to avoid multiple reads of the same file.
+Note, setting this value too high may lead to high memory usage during the sphinx build.
+
+Default: :need_config_default:`import_cache_size`
 
 .. _needs_needextend_strict:
 

docs/directives/needimport.rst

@@ -30,6 +30,11 @@ The directive also supports URL as argument to download ``needs.json`` from remo
 
    .. needimport:: https://my_company.com/docs/remote-needs.json
 
+.. seealso::
+
+    :ref:`needs_import_cache_size`,
+    to control the cache size for imported needs.
+
 Options
 -------
 

sphinx_needs/config.py

@@ -440,6 +440,10 @@ def __setattr__(self, name: str, value: Any) -> None:
         default_factory=list, metadata={"rebuild": "html", "types": (list,)}
     )
     """List of external sources to load needs from."""
+    import_cache_size: int = field(
+        default=100, metadata={"rebuild": "", "types": (int,)}
+    )
+    """Maximum number of imported needs to cache."""
     builder_filter: str = field(
         default="is_external==False", metadata={"rebuild": "html", "types": (str,)}
     )

sphinx_needs/directives/needimport.py

@@ -3,7 +3,9 @@
 import json
 import os
 import re
-from typing import Sequence
+import threading
+from copy import deepcopy
+from typing import Any, OrderedDict, Sequence
 from urllib.parse import urlparse
 
 import requests
@@ -52,7 +54,8 @@ class NeedimportDirective(SphinxDirective):
 
     @measure_time("needimport")
     def run(self) -> Sequence[nodes.Node]:
-        # needs_list = {}
+        needs_config = NeedsSphinxConfig(self.config)
+
         version = self.options.get("version")
         filter_string = self.options.get("filter")
         id_prefix = self.options.get("id_prefix", "")
@@ -111,21 +114,34 @@ def run(self) -> Sequence[nodes.Node]:
                 raise ReferenceError(
                     f"Could not load needs import file {correct_need_import_path}"
                 )
+            mtime = os.path.getmtime(correct_need_import_path)
 
-            try:
-                with open(correct_need_import_path) as needs_file:
-                    needs_import_list = json.load(needs_file)
-            except (OSError, json.JSONDecodeError) as e:
-                # TODO: Add exception handling
-                raise SphinxNeedsFileException(correct_need_import_path) from e
-
-            errors = check_needs_data(needs_import_list)
-            if errors.schema:
-                logger.info(
-                    f"Schema validation errors detected in file {correct_need_import_path}:"
-                )
-                for error in errors.schema:
-                    logger.info(f'  {error.message} -> {".".join(error.path)}')
+            if (
+                needs_import_list := _FileCache.get(correct_need_import_path, mtime)
+            ) is None:
+                try:
+                    with open(correct_need_import_path) as needs_file:
+                        needs_import_list = json.load(needs_file)
+                except (OSError, json.JSONDecodeError) as e:
+                    # TODO: Add exception handling
+                    raise SphinxNeedsFileException(correct_need_import_path) from e
+
+                errors = check_needs_data(needs_import_list)
+                if errors.schema:
+                    logger.info(
+                        f"Schema validation errors detected in file {correct_need_import_path}:"
+                    )
+                    for error in errors.schema:
+                        logger.info(f'  {error.message} -> {".".join(error.path)}')
+                else:
+                    _FileCache.set(
+                        correct_need_import_path,
+                        mtime,
+                        needs_import_list,
+                        needs_config.import_cache_size,
+                    )
+
+            self.env.note_dependency(correct_need_import_path)
 
         if version is None:
             try:
@@ -141,17 +157,17 @@ def run(self) -> Sequence[nodes.Node]:
                 f"Version {version} not found in needs import file {correct_need_import_path}"
             )
 
-        needs_config = NeedsSphinxConfig(self.config)
         data = needs_import_list["versions"][version]
 
+        # TODO this is not exactly NeedsInfoType, because the export removes/adds some keys
+        needs_list: dict[str, NeedsInfoType] = data["needs"]
+
         if ids := self.options.get("ids"):
             id_list = [i.strip() for i in ids.split(",") if i.strip()]
-            data["needs"] = {
+            needs_list = {
                 key: data["needs"][key] for key in id_list if key in data["needs"]
             }
 
-        # TODO this is not exactly NeedsInfoType, because the export removes/adds some keys
-        needs_list: dict[str, NeedsInfoType] = data["needs"]
         if schema := data.get("needs_schema"):
             # Set defaults from schema
             defaults = {
@@ -160,7 +176,8 @@ def run(self) -> Sequence[nodes.Node]:
                 if "default" in value
             }
             needs_list = {
-                key: {**defaults, **value} for key, value in needs_list.items()
+                key: {**defaults, **value}  # type: ignore[typeddict-item]
+                for key, value in needs_list.items()
             }
 
         # Filter imported needs
@@ -169,7 +186,8 @@ def run(self) -> Sequence[nodes.Node]:
             if filter_string is None:
                 needs_list_filtered[key] = need
             else:
-                filter_context = need.copy()
+                # we deepcopy here, to ensure that the original data is not modified
+                filter_context = deepcopy(need)
 
                 # Support both ways of addressing the description, as "description" is used in json file, but
                 # "content" is the sphinx internal name for this kind of information
@@ -185,7 +203,9 @@ def run(self) -> Sequence[nodes.Node]:
                         location=(self.env.docname, self.lineno),
                     )
 
-        needs_list = needs_list_filtered
+        # note we need to deepcopy here, as we are going to modify the data,
+        # but we want to ensure data referenced from the cache is not modified
+        needs_list = deepcopy(needs_list_filtered)
 
         # tags update
         if tags := [
@@ -265,6 +285,41 @@ def docname(self) -> str:
         return self.env.docname
 
 
+class _ImportCache:
+    """A simple cache for imported needs,
+    mapping a (path, mtime) to a dictionary of needs.
+    that is thread safe,
+    and has a maximum size when adding new items.
+    """
+
+    def __init__(self) -> None:
+        self._cache: OrderedDict[tuple[str, float], dict[str, Any]] = OrderedDict()
+        self._need_count = 0
+        self._lock = threading.Lock()
+
+    def set(
+        self, path: str, mtime: float, value: dict[str, Any], max_size: int
+    ) -> None:
+        with self._lock:
+            self._cache[(path, mtime)] = value
+            self._need_count += len(value)
+            max_size = max(max_size, 0)
+            while self._need_count > max_size:
+                _, value = self._cache.popitem(last=False)
+                self._need_count -= len(value)
+
+    def get(self, path: str, mtime: float) -> dict[str, Any] | None:
+        with self._lock:
+            return self._cache.get((path, mtime), None)
+
+    def __repr__(self) -> str:
+        with self._lock:
+            return f"{self.__class__.__name__}({list(self._cache)})"
+
+
+_FileCache = _ImportCache()
+
+
 class VersionNotFound(BaseException):
     pass