Build: check for _build/html directory and fail if exists (#10126)

* Build: check for `_build/html` directory and fail if exists

Continuation of #10038
Closes #10036

* Update exception message

Co-authored-by: Anthony <[email protected]>

---------

Co-authored-by: Anthony <[email protected]>

Author: humitos

readthedocs/doc_builder/director.py

@@ -332,6 +332,30 @@ def run_build_job(self, job):
         for command in commands:
             environment.run(command, escape_command=False, cwd=cwd)
 
+    def check_old_output_directory(self):
+        """
+        Check if there the directory '_build/html' exists and fail the build if so.
+
+        Read the Docs used to build artifacts into '_build/html' and there are
+        some projects with this path hardcoded in their files. Those builds are
+        having unexpected behavior since we are not using that path anymore.
+
+        In case we detect they are keep using that path, we fail the build
+        explaining this.
+        """
+        command = self.build_environment.run(
+            "test",
+            "-x",
+            "_build/html",
+            cwd=self.data.project.checkout_path(self.data.version.slug),
+            record=False,
+        )
+        if command.exit_code == 0:
+            log.warning(
+                "Directory '_build/html' exists. This may lead to unexpected behavior."
+            )
+            raise BuildUserError(BuildUserError.BUILD_OUTPUT_OLD_DIRECTORY_USED)
+
     def run_build_commands(self):
         reshim_commands = (
             {"pip", "install"},

readthedocs/doc_builder/exceptions.py

@@ -52,6 +52,11 @@ class BuildUserError(BuildBaseException):
         "and it is not currently supported. "
         'Please, remove all the files but the "{artifact_type}" your want to upload.'
     )
+    BUILD_OUTPUT_OLD_DIRECTORY_USED = gettext_noop(
+        "Some files were detected in an unsupported output path, '_build/html'. "
+        "Ensure your project is configured to use the output path "
+        "'$READTHEDOCS_OUTPUT/html' instead."
+    )
     PDF_COMMAND_FAILED = gettext_noop(
         "PDF generation failed. "
         "The build log below contains information on what errors caused the failure."

readthedocs/projects/tasks/builds.py

@@ -754,6 +754,7 @@ def execute(self):
                     self.update_build(state=BUILD_STATE_BUILDING)
                     self.data.build_director.build()
             finally:
+                self.data.build_director.check_old_output_directory()
                 self.data.build_data = self.collect_build_data()
 
         # At this point, the user's build already succeeded.

readthedocs/projects/tests/test_build_tasks.py

@@ -782,6 +782,13 @@ def test_build_commands_executed(
                     cwd=mock.ANY,
                     record=False,
                 ),
+                mock.call(
+                    "test",
+                    "-x",
+                    "_build/html",
+                    record=False,
+                    cwd=mock.ANY,
+                ),
                 # FIXME: I think we are hitting this issue here:
                 # https://github.com/pytest-dev/pytest-mock/issues/234
                 mock.call("lsb_release", "--description", record=False, demux=True),
@@ -806,14 +813,6 @@ def test_build_commands_executed(
                     record=False,
                     demux=True,
                 ),
-                mock.call(
-                    "test",
-                    "-x",
-                    "_build/html",
-                    record=False,
-                    demux=True,
-                    cwd=mock.ANY,
-                ),
             ]
         )
 

readthedocs/telemetry/collectors.py

@@ -93,27 +93,8 @@ def collect(self):
         }
         data["doctool"] = self._get_doctool()
 
-        # NOTE: Check if there are files at `_build/html` (the old output directory)
-        # and log these projects so we can communicate with their maintainers
-        #
-        # This temporal and should be removed in the future once we have
-        # decided what to do with this collected data.
-        self._check_using_old_output_directory()
-
         return data
 
-    def _check_using_old_output_directory(self):
-        code, _, _ = self.run(
-            "test",
-            "-x",
-            "_build/html",
-            cwd=self.project.checkout_path(self.version.slug),
-        )
-        if code == 0:
-            log.warning(
-                "Directory '_build/html' exists. This may lead to unexpected behavior."
-            )
-
     def _get_doctool_name(self):
         if self.version.is_sphinx_type:
             return "sphinx"