perf: avoid needless sql operations. #1538

If the set of arcs is empty, skip the SQL operations.  We also need to
allow setting a file tracer for an unmeasured file, to avoid the Cython
problem whose fix caused the performance issue in the first place.

TBH, I don't know why we had to prevent file tracers on unmeasured
files.  Perhaps pytest-cov has changed to avoid the behavior that caused
problems.

3 files changed, 23 insertions, 20 deletions

diff --git a/CHANGES.rst b/CHANGES.rst
index 185f20bf..745564c5 100644
--- a/CHANGES.rst
+++ b/CHANGES.rst
@@ -20,13 +20,20 @@ development at the same time, such as 4.5.x and 5.0.
 Unreleased
 ----------
 
+- Performance: fixed a slow-down with dynamic contexts that appeared in 6.4.3,
+  closing `issue 1538`_.  Hopefully this doesn't break the `Cython change`_
+  that fixed `issue 972`_.  Thanks to Mathieu Kniewallner for the deep
+  investigative work and comprehensive issue report.
+
 - Added: the debug output file can now be specified with ``[run] debug_file``
   in the configuration file.  Closes `issue 1319`_.
 
 - Typing: all product and test code has type annotations.
 
+.. _Cython change: https://github.com/nedbat/coveragepy/pull/1347
+.. _issue 972: https://github.com/nedbat/coveragepy/issues/972
 .. _issue 1319: https://github.com/nedbat/coveragepy/issues/1319
-
+.. _issue 1538: https://github.com/nedbat/coveragepy/issues/1538
 
 .. scriv-start-here
 
diff --git a/coverage/sqldata.py b/coverage/sqldata.py
index da66ad09..1cb8abe4 100644
--- a/coverage/sqldata.py
+++ b/coverage/sqldata.py
@@ -528,6 +528,8 @@ class CoverageData(AutoReprMixin):
         with self._connect() as con:
             self._set_context_id()
             for filename, arcs in arc_data.items():
+                if not arcs:
+                    continue
                 file_id = self._file_id(filename, add=True)
                 data = [(file_id, self._current_context_id, fromno, tono) for fromno, tono in arcs]
                 con.executemany_void(
@@ -571,12 +573,7 @@ class CoverageData(AutoReprMixin):
         self._start_using()
         with self._connect() as con:
             for filename, plugin_name in file_tracers.items():
-                file_id = self._file_id(filename)
-                if file_id is None:
-                    raise DataError(
-                        f"Can't add file tracer data for unmeasured file '{filename}'"
-                    )
-
+                file_id = self._file_id(filename, add=True)
                 existing_plugin = self.file_tracer(filename)
                 if existing_plugin:
                     if existing_plugin != plugin_name:
@@ -1213,10 +1210,9 @@ class SqliteDb(AutoReprMixin):
         else:
             raise AssertionError(f"SQL {sql!r} shouldn't return {len(rows)} rows")
 
-    def _executemany(self, sql: str, data: Iterable[Any]) -> sqlite3.Cursor:
+    def _executemany(self, sql: str, data: List[Any]) -> sqlite3.Cursor:
         """Same as :meth:`python:sqlite3.Connection.executemany`."""
         if self.debug.should("sql"):
-            data = list(data)
             final = ":" if self.debug.should("sqldata") else ""
             self.debug.write(f"Executing many {sql!r} with {len(data)} rows{final}")
             if self.debug.should("sqldata"):
@@ -1233,7 +1229,9 @@ class SqliteDb(AutoReprMixin):
 
     def executemany_void(self, sql: str, data: Iterable[Any]) -> None:
         """Same as :meth:`python:sqlite3.Connection.executemany` when you don't need the cursor."""
-        self._executemany(sql, data).close()
+        data = list(data)
+        if data:
+            self._executemany(sql, data).close()
 
     def executescript(self, script: str) -> None:
         """Same as :meth:`python:sqlite3.Connection.executescript`."""
diff --git a/tests/test_data.py b/tests/test_data.py
index ab4e03f7..5953ba36 100644
--- a/tests/test_data.py
+++ b/tests/test_data.py
@@ -187,6 +187,14 @@ class CoverageDataTest(CoverageTest):
         assert_line_counts(covdata, SUMMARY_3_4)
         assert_measured_files(covdata, MEASURED_FILES_3_4)
 
+    def test_ok_to_add_empty_arcs(self) -> None:
+        covdata = DebugCoverageData()
+        covdata.add_arcs(ARCS_3)
+        covdata.add_arcs(ARCS_4)
+        covdata.add_arcs(dict.fromkeys(ARCS_3, set()))
+        assert_line_counts(covdata, SUMMARY_3_4)
+        assert_measured_files(covdata, MEASURED_FILES_3_4)
+
     @pytest.mark.parametrize("klass", [CoverageData, DebugCoverageData])
     def test_cant_add_arcs_with_lines(self, klass: TCoverageData) -> None:
         covdata = klass()
@@ -350,16 +358,6 @@ class CoverageDataTest(CoverageTest):
         assert covdata.file_tracer("p1.foo") == "p1.plugin"
         assert covdata.file_tracer("main.py") == ""
 
-    def test_cant_file_tracer_unmeasured_files(self) -> None:
-        covdata = DebugCoverageData()
-        msg = "Can't add file tracer data for unmeasured file 'p1.foo'"
-        with pytest.raises(DataError, match=msg):
-            covdata.add_file_tracers({"p1.foo": "p1.plugin"})
-
-        covdata.add_lines({"p2.html": [10, 11, 12]})
-        with pytest.raises(DataError, match=msg):
-            covdata.add_file_tracers({"p1.foo": "p1.plugin"})
-
     def test_cant_change_file_tracer_name(self) -> None:
         covdata = DebugCoverageData()
         covdata.add_lines({"p1.foo": [1, 2, 3]})