From 6f8aa6ee09accbb46db6927dac0e61ca64b7d143 Mon Sep 17 00:00:00 2001
From: Jake Howard <git@theorangeone.net>
Date: Thu, 25 May 2017 20:34:41 +0100
Subject: [PATCH] Calculate wordcount better and make togglable

---
 md_pdf/assets/templates/footer-template.html |  4 +++-
 md_pdf/build/context.py                      |  4 +++-
 md_pdf/config/validate.py                    | 10 +++++++++-
 md_pdf/utils.py                              | 12 ++++++++++++
 test-files/mdp.yml                           |  1 +
 5 files changed, 28 insertions(+), 3 deletions(-)

diff --git a/md_pdf/assets/templates/footer-template.html b/md_pdf/assets/templates/footer-template.html
index ceabfa5..2ea9fde 100644
--- a/md_pdf/assets/templates/footer-template.html
+++ b/md_pdf/assets/templates/footer-template.html
@@ -11,7 +11,9 @@
           Page <span class="page"></span> of <span class="topage"></span>
         </td>
         <td>
-          Total Words: {{ word_count }}
+          {% if word_count %}
+            Total Words: {{ word_count }}
+          {% endif %}
         </td>
       </tr>
     </table>
diff --git a/md_pdf/build/context.py b/md_pdf/build/context.py
index e380d42..0a6d127 100644
--- a/md_pdf/build/context.py
+++ b/md_pdf/build/context.py
@@ -1,5 +1,6 @@
 from md_pdf.consts import TEMPLATES_DIR, STATIC_DIR
 from word_count import word_count
+from md_pdf.utils import get_plain_text
 
 
 EXTRA_CONTEXT = {
@@ -12,6 +13,7 @@ def get_context(config, content):
     context = config['context'].copy()
     context['title'] = config['title']
     context = dict(context, **EXTRA_CONTEXT, **{
-        'word_count': word_count(content)
     })
+    if config.get('show_word_count'):
+        context['word_count'] = word_count(get_plain_text(content))
     return context
diff --git a/md_pdf/config/validate.py b/md_pdf/config/validate.py
index d7e0199..c8dc41e 100644
--- a/md_pdf/config/validate.py
+++ b/md_pdf/config/validate.py
@@ -76,6 +76,13 @@ def validate_toc(config):
         raise ConfigValidationException("Table of contents key should be either true or false")
 
 
+def validate_wordcount(config):
+    if 'show_word_count' not in config:
+        return
+    if type(config['show_word_count']) != bool:
+        raise ConfigValidationException("Show word count key should be either true or false")
+
+
 def validate_config(config):
     logger.debug("Validating Config...")
     for validator in [
@@ -84,7 +91,8 @@ def validate_config(config):
         test_output,
         validate_bibliography,
         validate_context,
-        validate_toc
+        validate_toc,
+        validate_wordcount
     ]:
         validator(config)
     logger.debug("Config Ok!")
diff --git a/md_pdf/utils.py b/md_pdf/utils.py
index 69a0e0b..101e639 100644
--- a/md_pdf/utils.py
+++ b/md_pdf/utils.py
@@ -1,6 +1,7 @@
 import shutil
 import os
 import logging
+from bs4 import BeautifulSoup
 
 logger = logging.getLogger(__file__)
 
@@ -19,3 +20,14 @@ def safe_list_get(l, idx, default):
         return l[idx]
     except IndexError:
         return default
+
+
+def get_plain_text(content):
+    soup = BeautifulSoup(content, 'html.parser')
+    body = soup.find('body')
+    try:
+        body.find('h1', class_='references-title').extract()
+        body.find('div', class_='references').extract()
+    except AttributeError:
+        pass
+    return body.text
diff --git a/test-files/mdp.yml b/test-files/mdp.yml
index 661ce22..538845b 100644
--- a/test-files/mdp.yml
+++ b/test-files/mdp.yml
@@ -13,3 +13,4 @@ context:
   turnitin_number: 789123
 title: test title
 toc: true
+show_word_count: true