pythongh-130914: Make graphlib.TopologicalSorter.prepare() idempotent (python#131317)

Closes python#130914: Make graphlib.TopologicalSorter.prepare() idempotent

Relax the rules so that `.prepare()` can be called multiple times, provided that no work has been passed out by `.get_ready()` yet.

Author: lordmauve

Doc/library/graphlib.rst

@@ -106,6 +106,14 @@
       function, the graph cannot be modified, and therefore no more nodes can be
       added using :meth:`~TopologicalSorter.add`.
 
+      A :exc:`ValueError` will be raised if the sort has been started by
+      :meth:`~.static_order` or :meth:`~.get_ready`.
+
+      .. versionchanged:: next
+
+         ``prepare()`` can now be called more than once as long as the sort has
+         not started. Previously this raised :exc:`ValueError`.
+
    .. method:: is_active()
 
       Returns ``True`` if more progress can be made and ``False`` otherwise.

Doc/whatsnew/3.14.rst

@@ -607,6 +607,15 @@ getopt
 * Add support for returning intermixed options and non-option arguments in order.
   (Contributed by Serhiy Storchaka in :gh:`126390`.)
 
+
+graphlib
+--------
+
+* Allow :meth:`graphlib.TopologicalSorter.prepare` to be called more than once
+  as long as sorting has not started.
+  (Contributed by Daniel Pope in :gh:`130914`)
+
+
 http
 ----
 

Lib/graphlib.py

@@ -90,13 +90,17 @@ def prepare(self):
         still be used to obtain as many nodes as possible until cycles block more
         progress. After a call to this function, the graph cannot be modified and
         therefore no more nodes can be added using "add".
+
+        Raise ValueError if nodes have already been passed out of the sorter.
+
         """
-        if self._ready_nodes is not None:
-            raise ValueError("cannot prepare() more than once")
+        if self._npassedout > 0:
+            raise ValueError("cannot prepare() after starting sort")
 
-        self._ready_nodes = [
-            i.node for i in self._node2info.values() if i.npredecessors == 0
-        ]
+        if self._ready_nodes is None:
+            self._ready_nodes = [
+                i.node for i in self._node2info.values() if i.npredecessors == 0
+            ]
         # ready_nodes is set before we look for cycles on purpose:
         # if the user wants to catch the CycleError, that's fine,
         # they can continue using the instance to grab as many

Lib/test/test_graphlib.py

@@ -140,9 +140,21 @@ def test_calls_before_prepare(self):
     def test_prepare_multiple_times(self):
         ts = graphlib.TopologicalSorter()
         ts.prepare()
-        with self.assertRaisesRegex(ValueError, r"cannot prepare\(\) more than once"):
+        ts.prepare()
+
+    def test_prepare_after_pass_out(self):
+        ts = graphlib.TopologicalSorter({'a': 'bc'})
+        ts.prepare()
+        self.assertEqual(set(ts.get_ready()), {'b', 'c'})
+        with self.assertRaisesRegex(ValueError, r"cannot prepare\(\) after starting sort"):
             ts.prepare()
 
+    def test_prepare_cycleerror_each_time(self):
+        ts = graphlib.TopologicalSorter({'a': 'b', 'b': 'a'})
+        for attempt in range(1, 4):
+            with self.assertRaises(graphlib.CycleError, msg=f"{attempt=}"):
+                ts.prepare()
+
     def test_invalid_nodes_in_done(self):
         ts = graphlib.TopologicalSorter()
         ts.add(1, 2, 3, 4)

Misc/ACKS

@@ -1483,6 +1483,7 @@ Michael Pomraning
 Martin Pool
 Iustin Pop
 Claudiu Popa
+Daniel Pope
 Nick Pope
 John Popplewell
 Matheus Vieira Portela

Misc/NEWS.d/next/Library/2025-03-16-08-00-29.gh-issue-130914.6z883_.rst

@@ -0,0 +1,3 @@
+Allow :meth:`graphlib.TopologicalSorter.prepare` to be called more than once
+as long as sorting has not started.
+Patch by Daniel Pope.