From 30b2aab7b645309eed6f328e00aed6eaa8e551ca Mon Sep 17 00:00:00 2001
From: "Mr. Senko" <atodorov@mrsenko.com>
Date: Sun, 11 Jun 2017 02:28:32 +0300
Subject: [PATCH] Add recursion stop

---
 pelican/contents.py                 | 25 ++++++++++++++++++++++---
 pelican/tests/content/include5.html |  1 +
 pelican/tests/content/include6.html |  1 +
 pelican/tests/test_contents.py      |  9 +++++++++
 4 files changed, 33 insertions(+), 3 deletions(-)
 create mode 100644 pelican/tests/content/include5.html
 create mode 100644 pelican/tests/content/include6.html

diff --git a/pelican/contents.py b/pelican/contents.py
index f6e08147..96713014 100644
--- a/pelican/contents.py
+++ b/pelican/contents.py
@@ -321,10 +321,19 @@ class Content(object):
         """
         regex = r"""[{|]include[|}](?P<path>[\w./]+)"""
         hrefs = re.compile(regex, re.X)
+        processed_paths = []
+        # In Python 3.x we can use the `nonlocal` declaration, in `replacer()`,
+        # to tell Python we mean to assign to the `source_path` variable from
+        # `_update_includes()`.
+        # In Python 2.x we simply can't assign to `source_path` in `replacer()`.
+        # However, we work around this by not assigning to the variable itself,
+        # but using a mutable container to keep track about the current working
+        # directory while doing the recursion.
+        source_dir = [source_path]
 
         def replacer(m):
             path = m.group('path')
-            path = self._path_replacer(path, source_path)
+            path = self._path_replacer(path, source_dir[0])
             path = posixize_path(
                     os.path.abspath(
                         os.path.join(self.settings['PATH'], path)
@@ -343,15 +352,25 @@ class Content(object):
                 logger.warning("Unable to read `%s`, skipping include.", path)
                 return ''.join(('{include}', m.group('path')))
 
+            # recursion stop
+            if path in processed_paths:
+                raise RuntimeError("Circular inclusion detected for '%s'" % path)
+            processed_paths.append(path)
+
             reader = self.readers.reader_classes[ext](self.settings)
             text, meta = reader.read(path)
 
             # if we recurse into another file to perform more includes
             # self._path_replacer needs to know in which directory
             # it operates otherwise it produces wrong paths
-            source_dir = posixize_path(os.path.dirname(path))
+            source_dir[0] = posixize_path(os.path.dirname(path))
+            current_source_dir = source_dir[0]
 
-            text = self._update_includes(text, source_dir)
+            # recursively replace other includes
+            text = hrefs.sub(replacer, text)
+
+            # restore source dir
+            source_dir[0] = current_source_dir
             return text
 
         return hrefs.sub(replacer, content)
diff --git a/pelican/tests/content/include5.html b/pelican/tests/content/include5.html
new file mode 100644
index 00000000..b5018e36
--- /dev/null
+++ b/pelican/tests/content/include5.html
@@ -0,0 +1 @@
+{include}include6.html
diff --git a/pelican/tests/content/include6.html b/pelican/tests/content/include6.html
new file mode 100644
index 00000000..e70cf4e6
--- /dev/null
+++ b/pelican/tests/content/include6.html
@@ -0,0 +1 @@
+{include}include5.html
diff --git a/pelican/tests/test_contents.py b/pelican/tests/test_contents.py
index 9546ab8a..660d33f9 100644
--- a/pelican/tests/test_contents.py
+++ b/pelican/tests/test_contents.py
@@ -558,6 +558,15 @@ class TestPage(LoggedTestCase):
             ' Included content is above'
         )
 
+        # recursion loop, include5.html includes include6.html
+        # and the other way around
+        args['content'] = (
+            'There is a simple include here '
+            '{include}include5.html'
+        )
+        with self.assertRaisesRegex(RuntimeError, 'Circular inclusion detected'):
+            Page(**args).get_content('http://notmyidea.org')
+
     def test_multiple_authors(self):
         """Test article with multiple authors."""
         args = self.page_kwargs.copy()