Add support from the HTTP range header

Author: satchamo

django/views/static.py

@@ -61,11 +61,32 @@ def serve(request, path, document_root=None, show_indexes=False):
         return HttpResponseNotModified()
     content_type, encoding = mimetypes.guess_type(fullpath)
     content_type = content_type or 'application/octet-stream'
-    response = StreamingHttpResponse(open(fullpath, 'rb'),
+    ranged_file = RangedFileReader(open(fullpath, 'rb'))
+    response = StreamingHttpResponse(ranged_file,
                                      content_type=content_type)
     response["Last-Modified"] = http_date(statobj.st_mtime)
     if stat.S_ISREG(statobj.st_mode):
-        response["Content-Length"] = statobj.st_size
+        size = statobj.st_size
+        response["Content-Length"] = size
+        response["Accept-Ranges"] = "bytes"
+        # Respect the Range header.
+        if "HTTP_RANGE" in request.META:
+            try:
+                ranges = parse_range_header(request.META['HTTP_RANGE'], size)
+            except ValueError:
+                ranges = None
+            # only handle syntactically valid headers, that are simple (no
+            # multipart byteranges)
+            if ranges is not None and len(ranges) == 1:
+                start, stop = ranges[0]
+                if stop > size:
+                    # requested range not satisfiable
+                    return HttpResponse(status=416)
+                ranged_file.start = start
+                ranged_file.stop = stop
+                response["Content-Range"] = "bytes %d-%d/%d" % (start, stop - 1, size)
+                response["Content-Length"] = stop - start
+                response.status_code = 206
     if encoding:
         response["Content-Encoding"] = encoding
     return response
@@ -144,3 +165,73 @@ def was_modified_since(header=None, mtime=0, size=0):
     except (AttributeError, ValueError, OverflowError):
         return True
     return False
+
+
+def parse_range_header(header, resource_size):
+    """
+    Parses a range header into a list of two-tuples (start, stop) where `start`
+    is the starting byte of the range (inclusive) and `stop` is the ending byte
+    position of the range (exclusive).
+
+    Returns None if the value of the header is not syntatically valid.
+    """
+    if not header or '=' not in header:
+        return None
+
+    ranges = []
+    units, range_ = header.split('=', 1)
+    units = units.strip().lower()
+
+    if units != "bytes":
+        return None
+
+    for val in range_.split(","):
+        val = val.strip()
+        if '-' not in val:
+            return None
+
+        if val.startswith("-"):
+            # suffix-byte-range-spec: this form specifies the last N bytes of an
+            # entity-body
+            start = resource_size + int(val)
+            if start < 0:
+                start = 0
+            stop = resource_size
+        else:
+            # byte-range-spec: first-byte-pos "-" [last-byte-pos]
+            start, stop = val.split("-", 1)
+            start = int(start)
+            # the +1 is here since we want the stopping point to be exclusive, whereas in
+            # the HTTP spec, the last-byte-pos is inclusive
+            stop = int(stop)+1 if stop else resource_size
+            if start >= stop:
+                return None
+
+        ranges.append((start, stop))
+
+    return ranges
+
+
+class RangedFileReader:
+    """
+    Wraps a file like object with an iterator that runs over part (or all) of
+    the file defined by start and stop. Blocks of block_size will be returned
+    from the starting position, up to, but not including the stop point.
+    """
+    block_size = 8192
+    def __init__(self, file_like, start=0, stop=float("inf"), block_size=None):
+        self.f = file_like
+        self.block_size = block_size or RangedFileReader.block_size
+        self.start = start
+        self.stop = stop
+
+    def __iter__(self):
+        self.f.seek(self.start)
+        position = self.start
+        while position < self.stop:
+            data = self.f.read(min(self.block_size, self.stop - position))
+            if not data:
+                break
+
+            yield data
+            position += self.block_size

tests/view_tests/tests/test_static.py

@@ -8,7 +8,7 @@
 from django.http import HttpResponseNotModified
 from django.test import SimpleTestCase, override_settings
 from django.utils.http import http_date
-from django.views.static import was_modified_since
+from django.views.static import was_modified_since, RangedFileReader
 
 from .. import urls
 from ..urls import media_dir
@@ -95,6 +95,51 @@ def test_404(self):
         response = self.client.get('/%s/non_existing_resource' % self.prefix)
         self.assertEqual(404, response.status_code)
 
+    def test_accept_ranges(self):
+        response = self.client.get('/%s/%s' % (self.prefix, "file.txt"))
+        self.assertEqual(response['Accept-Ranges'], "bytes")
+
+    def test_syntactically_invalid_ranges(self):
+        """
+        Test that a syntactically invalid byte range header is ignored and the
+        response gives back the whole resource as per RFC 2616, section 14.35.1
+        """
+        content = open(path.join(media_dir, "file.txt")).read()
+        invalid = ["megabytes=1-2", "bytes=", "bytes=3-2", "bytes=--5", "units", "bytes=-,"]
+        for range_ in invalid:
+            response = self.client.get('/%s/%s' % (self.prefix, "file.txt"), HTTP_RANGE=range_)
+            self.assertEqual(content, b''.join(response))
+
+    def test_unsatisfiable_range(self):
+        """Test that an unsatisfiable range results in a 416 HTTP status code"""
+        content = open(path.join(media_dir, "file.txt")).read()
+        # since byte ranges are *inclusive*, 0 to len(content) would be unsatisfiable
+        response = self.client.get('/%s/%s' % (self.prefix, "file.txt"), HTTP_RANGE="bytes=0-%d" % len(content))
+        self.assertEqual(response.status_code, 416)
+
+    def test_ranges(self):
+        # set the block size to something small so we do multiple iterations in
+        # the RangedFileReader class
+        original_block_size = RangedFileReader.block_size
+        RangedFileReader.block_size = 3
+
+        content = open(path.join(media_dir, "file.txt")).read()
+        # specify the range header, the expected response content, and the
+        # values of the content-range header byte positions
+        ranges = {
+            "bytes=0-10": (content[0:11], (0, 10)),
+            "bytes=9-9": (content[9:10], (9, 9)),
+            "bytes=-5": (content[len(content)-5:], (len(content)-5, len(content)-1)),
+            "bytes=3-": (content[3:], (3, len(content)-1)),
+            "bytes=-%d" % (len(content) + 1): (content, (0, len(content)-1)),
+        }
+        for range_, (expected_result, byte_positions) in ranges.items():
+            response = self.client.get('/%s/%s' % (self.prefix, "file.txt"), HTTP_RANGE=range_)
+            self.assertEqual(expected_result, b''.join(response))
+            self.assertEqual(int(response['Content-Length']), len(expected_result))
+            self.assertEqual(response['Content-Range'], "bytes %d-%d/%d" % (byte_positions + (len(content),)))
+
+        RangedFileReader.block_size = original_block_size
 
 class StaticHelperTest(StaticTests):
     """