Merge pull request #137 from alistairmagee/clean_summary

Add "Clean Summary" plugin

clean_summary/README.md

@@ -0,0 +1,34 @@
+#Clean Summary Plugin#
+
+Plugin to clean your summary of excess images. Images can take up much more
+space than text and lead to summaries being different sizes on archive and 
+index pages. With this plugin you can specify a maximum number of images that
+will appear in your summaries.
+
+There is also an option to include a minimum of one image.
+
+##Settings##
+
+This plugin has two settings. `CLEAN_SUMMARY_MAXIMUM` which takes an int, and 
+`CLEAN_SUMMARY_MINIMUM_ONE` which takes a boolean. They default to 0 and False.
+
+`CLEAN_SUMMARY_MAXIMUM` sets the maximum number of images that will appear in 
+your summary.
+
+if `CLEAN_SUMMARY_MINIMUM_ONE` is set to `True` and your summary doesn't already
+contain an image, the plugin will add the first image in your article (if one 
+exists) to the beginning of the summary.
+
+##Requirements##
+
+Requires Beautiful Soup:
+
+    pip install BeautifulSoup4
+
+
+##Usage with Summary Plugin##
+
+If using the summary plugin, make sure summary appears in your plugins before
+clean summary. Eg.
+
+    PLUGINS = ['summary', 'clean_summary', ... ]

clean_summary/__init__.py

@@ -0,0 +1 @@
+from .clean_summary import *

clean_summary/clean_summary.py

@@ -0,0 +1,38 @@
+"""
+Clean Summary
+-------------
+
+adds option to specify maximum number of images to appear in article summary
+also adds option to include an image by default if one exists in your article
+"""
+
+from pelican import signals
+from pelican.contents import Content, Article
+from bs4 import BeautifulSoup
+from six import text_type
+
+def clean_summary(instance):
+    if "CLEAN_SUMMARY_MAXIMUM" in instance.settings:
+        maximum_images = instance.settings["CLEAN_SUMMARY_MAXIMUM"]
+    else:
+        maximum_images = 0
+    if "CLEAN_SUMMARY_MINIMUM_ONE" in instance.settings:
+        minimum_one = instance.settings['CLEAN_SUMMARY_MINIMUM_ONE']
+    else:
+        minimum_one = False
+    if type(instance) == Article:
+        summary = instance.summary
+        summary = BeautifulSoup(instance.summary, 'html.parser')
+        images = summary.findAll('img')
+        if (len(images) > maximum_images):
+            for image in images[maximum_images:]:
+                image.extract()
+        if len(images) < 1 and minimum_one: #try to find one
+            content = BeautifulSoup(instance.content, 'html.parser')
+            first_image = content.find('img')
+            if first_image:
+                summary.insert(0, first_image)
+        instance._summary = text_type(summary)
+
+def register():
+    signals.content_object_init.connect(clean_summary)