Fix #708: .filename is private, and don't combine over ourselves

CHANGES.rst

@@ -29,10 +29,17 @@ Unreleased
 - Combining data stored in SQLite now goes about twice as fast, fixing `issue
   761`_.  Thanks, Stephan Richter.
 
+- The ``filename`` attribute on `CoverageData` object has been made private.
+  You can use the ``filename`` method to get the actual file name being used
+  to store data, and the ``base_filename`` method to get the original filename
+  before parallelizing suffixes were added.  This is part of fixing `issue
+  708`_.
+
 - Line numbers in the HTML report now align properly with source lines, even
   when Chrome's minimum font size is set, fixing `issue 748`_.  Thanks Wen Ye.
 
 .. _issue 702: https://github.com/nedbat/coveragepy/issues/702
+.. _issue 708: https://github.com/nedbat/coveragepy/issues/708
 .. _issue 746: https://github.com/nedbat/coveragepy/issues/746
 .. _issue 748: https://github.com/nedbat/coveragepy/issues/748
 .. _issue 761: https://github.com/nedbat/coveragepy/issues/761

coverage/cmdline.py

@@ -667,7 +667,7 @@ def do_debug(self, args):
                 self.coverage.load()
                 data = self.coverage.get_data()
                 print(info_header("data"))
-                print("path: %s" % self.coverage.get_data().filename)
+                print("path: %s" % self.coverage.get_data().filename())
                 if data:
                     print("has_arcs: %r" % data.has_arcs())
                     summary = line_counts(data, fullpath=True)

coverage/data.py

@@ -710,7 +710,7 @@ def combine_parallel_data(data, aliases=None, data_paths=None, strict=False):
     """
     # Because of the os.path.abspath in the constructor, data_dir will
     # never be an empty string.
-    data_dir, local = os.path.split(data.filename)
+    data_dir, local = os.path.split(data.base_filename())
     localdot = local + '.*'
 
     data_paths = data_paths or [data_dir]
@@ -729,6 +729,12 @@ def combine_parallel_data(data, aliases=None, data_paths=None, strict=False):
 
     files_combined = 0
     for f in files_to_combine:
+        if f == data.filename():
+            # Sometimes we are combining into a file which is one of the
+            # parallel files.  Skip that file.
+            if data._debug.should('dataio'):
+                data._debug.write("Skipping combining ourself: %r" % (f,))
+            continue
         if data._debug.should('dataio'):
             data._debug.write("Combining data file %r" % (f,))
         try:

coverage/sqldata.py

@@ -97,10 +97,10 @@ def __init__(self, basename=None, suffix=None, warn=None, debug=None):
         self._current_context_id = None
 
     def _choose_filename(self):
-        self.filename = self._basename
+        self._filename = self._basename
         suffix = filename_suffix(self._suffix)
         if suffix:
-            self.filename += "." + suffix
+            self._filename += "." + suffix
 
     def _reset(self):
         if self._dbs:
@@ -113,8 +113,8 @@ def _reset(self):
 
     def _create_db(self):
         if self._debug.should('dataio'):
-            self._debug.write("Creating data file {!r}".format(self.filename))
-        self._dbs[get_thread_id()] = Sqlite(self.filename, self._debug)
+            self._debug.write("Creating data file {!r}".format(self._filename))
+        self._dbs[get_thread_id()] = Sqlite(self._filename, self._debug)
         with self._dbs[get_thread_id()] as db:
             for stmt in SCHEMA.split(';'):
                 stmt = " ".join(stmt.strip().split())
@@ -128,22 +128,22 @@ def _create_db(self):
 
     def _open_db(self):
         if self._debug.should('dataio'):
-            self._debug.write("Opening data file {!r}".format(self.filename))
-        self._dbs[get_thread_id()] = Sqlite(self.filename, self._debug)
+            self._debug.write("Opening data file {!r}".format(self._filename))
+        self._dbs[get_thread_id()] = Sqlite(self._filename, self._debug)
         with self._dbs[get_thread_id()] as db:
             try:
                 schema_version, = db.execute("select version from coverage_schema").fetchone()
             except Exception as exc:
                 raise CoverageException(
                     "Data file {!r} doesn't seem to be a coverage data file: {}".format(
-                        self.filename, exc
+                        self._filename, exc
                     )
                 )
             else:
                 if schema_version != SCHEMA_VERSION:
                     raise CoverageException(
                         "Couldn't use data file {!r}: wrong schema: {} instead of {}".format(
-                            self.filename, schema_version, SCHEMA_VERSION
+                            self._filename, schema_version, SCHEMA_VERSION
                         )
                     )
 
@@ -155,14 +155,14 @@ def _open_db(self):
 
     def _connect(self):
         if get_thread_id() not in self._dbs:
-            if os.path.exists(self.filename):
+            if os.path.exists(self._filename):
                 self._open_db()
             else:
                 self._create_db()
         return self._dbs[get_thread_id()]
 
     def __nonzero__(self):
-        if (get_thread_id() not in self._dbs and not os.path.exists(self.filename)):
+        if (get_thread_id() not in self._dbs and not os.path.exists(self._filename)):
             return False
         try:
             with self._connect() as con:
@@ -221,6 +221,14 @@ def _set_context_id(self):
                 cur = con.execute("insert into context (context) values (?)", (context,))
                 self._current_context_id = cur.lastrowid
 
+    def base_filename(self):
+        """The base filename for storing data."""
+        return self._basename
+
+    def filename(self):
+        """Where is the data stored?"""
+        return self._filename
+
     def add_lines(self, line_data):
         """Add measured line data.
 
@@ -477,10 +485,10 @@ def erase(self, parallel=False):
         """
         self._reset()
         if self._debug.should('dataio'):
-            self._debug.write("Erasing data file {!r}".format(self.filename))
-        file_be_gone(self.filename)
+            self._debug.write("Erasing data file {!r}".format(self._filename))
+        file_be_gone(self._filename)
         if parallel:
-            data_dir, local = os.path.split(self.filename)
+            data_dir, local = os.path.split(self._filename)
             localdot = local + '.*'
             pattern = os.path.join(os.path.abspath(data_dir), localdot)
             for filename in glob.glob(pattern):

tests/test_cmdline.py

@@ -691,7 +691,7 @@ def test_debug_data(self):
             2 files:
             file1.py: 17 lines [a_plugin]
             file2.py: 23 lines
-            """).replace("FILENAME", data.filename))
+            """).replace("FILENAME", data.filename()))
 
     def test_debug_data_with_no_data(self):
         data = CoverageData()
@@ -700,7 +700,7 @@ def test_debug_data_with_no_data(self):
             -- data ------------------------------------------------------
             path: FILENAME
             No data collected
-            """).replace("FILENAME", data.filename))
+            """).replace("FILENAME", data.filename()))
 
 
 class CmdLineStdoutTest(BaseCmdLineTest):