1 files changed, 52 insertions, 0 deletions
diff --git a/app/blog/parse.py b/app/blog/parse.py
new file mode 100755
index 0000000..27f5165
--- /dev/null
+++ b/app/blog/parse.py
@@ -0,0 +1,52 @@
+#!/usr/bin/python
+import os, datetime
+from dateutil.parser import parse as dateparser
+from os.path import abspath,dirname
+from django.core.exceptions import ObjectDoesNotExist
+from blog.models import Entry
+
+from django.conf import settings
+
+def parse_file(filepath):
+    data = {}
+    contents = open(filepath).read()
+    raw = contents.splitlines()
+    for line in raw[1:]:
+        if line == '---':
+            break
+        else:
+            k,v = line.split(':', 1)
+            data[k.strip()] = v.strip()
+    body = "\n".join(line.strip() for line in raw[1:])
+    data["body_markdown"] = body.split('---')[1]
+    return data
+
+"""    
+    now I need a function to query the db for the title and date
+    if there's no entry then it's new and we add it and publish
+    What about edits though? Crap, edits. That means we need to check lastmod
+    and that's notoriously inaccurate. damn.
+from blog.parse import *
+crawl_dir()
+
+"""
+def crawl_dir():
+    file_root = settings.POSTS_DIR
+    file_list = os.listdir(file_root)
+    file_list = filter(lambda item: not (item.startswith('README') or item.startswith('updategithub.php') or item.startswith('.') or item.endswith('~')),file_list)
+    for f in file_list: 
+        fpath = file_root+"/"+f
+        last_mod = datetime.datetime.fromtimestamp(os.path.getmtime(fpath))
+        last_run = datetime.datetime.fromtimestamp(os.path.getmtime(abspath(dirname(__file__))+'/last_run'))
+        if last_mod > last_run: 
+            print "needs an update"
+            data = parse_file(fpath)
+            date = dateparser(data['pub_date'])
+            try:
+                row = Entry.objects.get(title=str(data['title']),pub_date=date) 
+                print row.title, date
+            except ObjectDoesNotExist:
+                print data['title'] + str(date) + " = not found"
+    last_mod_dump = open(abspath(dirname(__file__))+'/last_run','w')
+    print last_mod_dump
+    print >> last_mod_dump, str(datetime.datetime.now())