Add support to set a lexer with a given filename. Closes issue #63.

This allows third-party API utils to set the lexer automatically based on the open filename.

docs/api.rst

@@ -62,6 +62,32 @@ values. Default: ``2592000``. In the default configuration valid values are:
 * 604800
 * 2592000
 
+``filename`` (optional)
+~~~~~~~~~~~~~~~~~~~~~~~
+
+Can also be set via GET. A filename which we use to determine a lexer, if
+``lexer`` is not set. In case we can't determine a file, the lexer will fallback
+to ``plain`` code (no highlighting). A given ``lexer`` will overwrite any
+filename! Example::
+
+    {
+        "url": "https://dpaste.de/xsWd",
+        "lexer": "",
+        "filename": "python",
+        "conent": "The text body of the snippet."
+    }
+
+This will create a ``python`` highlighted snippet. However in this example::
+
+    {
+        "url": "https://dpaste.de/xsWd",
+        "lexer": "php",
+        "filename": "python",
+        "conent": "The text body of the snippet."
+    }
+
+Since the lexer is set too, we will create a ``php`` highlighted snippet.
+
 .. hint:: You need to adjust the setting ``DPASTE_BASE_URL`` which is used
     to generate the full qualified URL in the API response. See :doc:`settings`.
 

dpaste/tests/test_api.py

@@ -3,9 +3,9 @@
 from django.core.urlresolvers import reverse
 from django.test.client import Client
 from django.test import TestCase
-from django.test.utils import override_settings
 
 from ..models import Snippet
+from ..highlight import PLAIN_CODE
 
 class SnippetAPITestCase(TestCase):
 
@@ -188,3 +188,53 @@ class SnippetAPITestCase(TestCase):
         self.assertEqual(response.status_code, 200)
         self.assertEqual(Snippet.objects.count(), 1)
         self.assertTrue(Snippet.objects.all()[0].expires)
+
+    def test_filename_not_given(self):
+        """
+        No lexer and no filename given returns a BadRequest.
+        """
+        response = self.client.post(self.api_url, {
+            'content': u"Hello Wörld.\n\tGood Bye",
+            'lexer': '',
+            'filename': ''
+        })
+        self.assertEqual(response.status_code, 400)
+
+    def test_filename_given(self):
+        """
+        No lexer and a Python filename will set a 'python' lexer.
+        """
+        response = self.client.post(self.api_url, {
+            'content': u"Hello Wörld.\n\tGood Bye",
+            'lexer': '',
+            'filename': 'helloworld.py'
+        })
+        self.assertEqual(response.status_code, 200)
+        self.assertEqual(Snippet.objects.count(), 1)
+        self.assertEqual(Snippet.objects.all()[0].lexer, 'python')
+
+    def test_awkward_filename_given(self):
+        """
+        A unknown filename will create a 'plain' code snippet.
+        """
+        response = self.client.post(self.api_url, {
+            'content': u"Hello Wörld.\n\tGood Bye",
+            'lexer': '',
+            'filename': 'helloworld.helloworld'
+        })
+        self.assertEqual(response.status_code, 200)
+        self.assertEqual(Snippet.objects.count(), 1)
+        self.assertEqual(Snippet.objects.all()[0].lexer, PLAIN_CODE)
+
+    def test_filename_and_lexer_given(self):
+        """
+        A given lexer will overwrite whats the filename guessing.
+        """
+        response = self.client.post(self.api_url, {
+            'content': u"Hello Wörld.\n\tGood Bye",
+            'lexer': 'php',
+            'filename': 'helloworld.py'
+        })
+        self.assertEqual(response.status_code, 200)
+        self.assertEqual(Snippet.objects.count(), 1)
+        self.assertEqual(Snippet.objects.all()[0].lexer, 'php')

dpaste/views.py

@@ -16,10 +16,13 @@ from django.views.defaults import (page_not_found as django_page_not_found,
     server_error as django_server_error)
 from django.views.decorators.csrf import csrf_exempt
 
+from pygments.lexers import get_lexer_for_filename
+from pygments.util import ClassNotFound
+
 from dpaste.forms import SnippetForm, get_expire_values, EXPIRE_CHOICES
 from dpaste.models import Snippet, ONETIME_LIMIT
-from dpaste.highlight import LEXER_WORDWRAP, LEXER_LIST
-from dpaste.highlight import LEXER_DEFAULT, LEXER_KEYS
+from dpaste.highlight import (LEXER_DEFAULT, LEXER_KEYS, LEXER_WORDWRAP,
+    LEXER_LIST, PLAIN_CODE)
 
 # -----------------------------------------------------------------------------
 # Snippet Handling
@@ -285,15 +288,32 @@ FORMAT_MAPPING = {
 def snippet_api(request):
     content = request.POST.get('content', '').strip()
     lexer = request.REQUEST.get('lexer', LEXER_DEFAULT).strip()
+    filename = request.REQUEST.get('filename', '').strip()
     expires = request.REQUEST.get('expires', '').strip()
     format = request.REQUEST.get('format', 'default').strip()
 
     if not content:
         return HttpResponseBadRequest('No content given')
 
-    if not lexer in LEXER_KEYS:
-        return HttpResponseBadRequest('Invalid lexer given. Valid lexers are: %s' %
-            ', '.join(LEXER_KEYS))
+    # We need at least a lexer or a filename
+    if not lexer and not filename:
+        return HttpResponseBadRequest('No lexer or filename given. Unable to '
+            'determine a highlight. Valid lexers are: %s' % ', '.join(LEXER_KEYS))
+
+    # A lexer is given, check if its valid at all
+    if lexer and lexer not in LEXER_KEYS:
+        return HttpResponseBadRequest('Invalid lexer "%s" given. Valid lexers are: %s' % (
+            lexer, ', '.join(LEXER_KEYS)))
+
+    # No lexer is given, but we have a filename, try to get the lexer out of it.
+    # In case Pygments cannot determine the lexer of the filename, we fallback
+    # to 'plain' code.
+    if not lexer and filename:
+        try:
+            lexer_cls = get_lexer_for_filename(filename)
+            lexer = lexer_cls.aliases[0]
+        except (ClassNotFound, IndexError):
+            lexer = PLAIN_CODE
 
     if expires:
         expire_options = [str(i) for i in dict(EXPIRE_CHOICES).keys()]