misc: add support for thumbnailing PDF files (#26632)

debian/control

@@ -25,7 +25,8 @@ Recommends: python-dns,
     python-qrcode,
     libjs-leaflet,
     python-magic,
-    python-docutils
+    python-docutils,
+    graphicsmagick
 Suggests: python-libxml2,
     python-lasso,
     python-psycopg2

wcs/fields.py

@@ -33,7 +33,7 @@ from django.utils.formats import date_format as django_date_format
 from qommon import _
 from qommon import evalutils
 from qommon.form import *
-from qommon.misc import localstrftime, strftime, date_format, ellipsize
+from qommon.misc import localstrftime, strftime, date_format, ellipsize, can_thumbnail
 from qommon.template import Template, TemplateError
 from qommon import get_cfg, get_logger
 
@@ -907,7 +907,7 @@ class FileField(WidgetField):
         t = TemplateIO(html=True)
         t += htmltext('<div class="file-field">')
         t += htmltext('<a download="%s" href="[download]?f=%s">') % (value.base_filename, self.id)
-        if value.content_type and value.content_type.startswith('image/'):
+        if can_thumbnail(value.content_type):
             t += htmltext('<img alt="" src="[download]?f=%s&thumbnail=1"/>') % self.id
         t += htmltext('<span>%s</span>') % value
         t += htmltext('</a></div>')

wcs/forms/common.py

@@ -81,17 +81,16 @@ class FileDirectory(Directory):
                 response.set_header(
                         'content-disposition', 'attachment; filename="%s"' % file.base_filename)
 
-        fp = file.get_file_pointer()
-        if self.thumbnails and file.content_type.startswith('image/'):
+        if self.thumbnails and misc.can_thumbnail(file.content_type):
             try:
-                thumbnail = misc.get_thumbnail(fp)
-            except misc.ThumbnailError:
-                pass
-            else:
+                thumbnail = misc.get_thumbnail(file.get_filename(),
+                        content_type=file.content_type)
                 response.set_content_type('image/png')
                 return thumbnail
+            except misc.ThumbnailError:
+                pass
 
-        return fp.read()
+        return file.get_file_pointer().read()
 
 
 class FilesDirectory(Directory):

wcs/forms/root.py

@@ -1123,16 +1123,16 @@ class FormPage(Directory, FormTemplateMixin):
         if tempfile['charset']:
             response.set_charset(tempfile['charset'])
 
-        file_pointer = get_session().get_tempfile_content(t).get_file_pointer()
         if get_request().form.get('thumbnail') == '1':
             try:
-                thumbnail = misc.get_thumbnail(file_pointer)
+                thumbnail = misc.get_thumbnail(get_session().get_tempfile_path(t),
+                        content_type=tempfile['content_type'])
             except misc.ThumbnailError:
                 pass
             else:
                 response.set_content_type('image/png')
                 return thumbnail
-        return file_pointer.read()
+        return get_session().get_tempfile_content(t).get_file_pointer().read()
 
     def validating(self, data):
         self.html_top(self.formdef.name)

wcs/qommon/form.py

@@ -71,7 +71,7 @@ from wcs.conditions import Condition, ValidationError
 from qommon import _, ngettext
 import misc
 from .humantime import humanduration2seconds, seconds2humanduration, timewords
-from .misc import strftime, C_
+from .misc import strftime, C_, HAS_GM
 from publisher import get_cfg
 from .template_utils import render_block_to_string
 
@@ -698,8 +698,11 @@ class FileWithPreviewWidget(CompositeWidget):
         if not temp:
             return False
 
-        filetype = mimetypes.guess_type(temp.get('orig_filename', ''))
-        if not (filetype and filetype[0] and filetype[0].startswith('image')):
+        filetype = (mimetypes.guess_type(temp.get('orig_filename', '')) or [''])[0]
+        if filetype == 'application/pdf':
+            return HAS_GM
+
+        if not filetype.startswith('image/'):
             return False
 
         if Image:
@@ -821,6 +824,12 @@ class PicklableUpload(Upload):
         # quack like UploadedFile
         return self.get_file_pointer()
 
+    def get_filename(self):
+        if not hasattr(self, 'qfilename'):
+            raise AttributeError('filename')
+        basedir = os.path.join(get_publisher().app_dir, 'uploads')
+        return os.path.join(basedir, self.qfilename)
+
     def get_content(self):
         if hasattr(self, 'qfilename'):
             filename = os.path.join(get_publisher().app_dir, 'uploads', self.qfilename)

wcs/qommon/misc.py

@@ -51,6 +51,12 @@ from urllib import urlencode, quote
 from urllib2 import urlparse
 from cStringIO import StringIO
 
+try:
+    subprocess.check_call(['which', 'gm'], stdout=open('/dev/null', 'w'))
+    HAS_GM = True
+except subprocess.CalledProcessError:
+    HAS_GM = False
+
 
 class ThumbnailError(Exception):
     pass
@@ -553,19 +559,32 @@ def file_digest(content, chunk_size=100000):
         digest.update(chunk)
     return digest.hexdigest()
 
-def get_thumbnail(fp):
-    if Image is None:
+
+def can_thumbnail(content_type):
+    if content_type == 'application/pdf' and not (HAS_GM and Image):
+        return False
+    if content_type and content_type.startswith('image/'):
+        return bool(Image is not None)
+    return False
+
+
+def get_thumbnail(filepath, content_type=None):
+    if not can_thumbnail(content_type or ''):
         raise ThumbnailError()
 
+    if content_type == 'application/pdf':
+        fp = StringIO(subprocess.check_output(
+                ['gm', 'convert', '-geometry', '500x', filepath, 'png:-']))
+    else:
+        fp = open(filepath)
+
     try:
         image = Image.open(fp)
         image.thumbnail((500, 300))
         image_thumb_fp = StringIO()
         image.save(image_thumb_fp, "PNG")
     except IOError:
-        # failed to create thumbnail; restore file pointer state and raise
-        # exception.
-        fp.seek(0)
+        # failed to create thumbnail.
         raise ThumbnailError()
     return image_thumb_fp.getvalue()
 

wcs/qommon/sessions.py

@@ -303,6 +303,14 @@ class Session(QommonSession, CaptchaSession, StorableObject):
             return None
         return misc.json_loads(open(filename).read())
 
+    def get_tempfile_path(self, token):
+        temp = self.get_tempfile(token)
+        if not temp:
+            return None
+        dirname = os.path.join(get_publisher().app_dir, 'tempfiles')
+        filename = os.path.join(dirname, temp['unsigned_token'])
+        return filename
+
     def get_tempfile_content(self, token):
         temp = self.get_tempfile(token)
         if not temp: