facebook2wordpress.py

@@ -0,0 +1,85 @@
+import json
+import arrow
+import os
+import logging
+from shlex import quote
+
+from pprint import pprint
+
+base = '/web/blog.pakuauk.com/exports/facebook-export'
+with open('facebook-export/posts/posts.json', 'rt') as f:
+    data = json.loads(f.read())
+data = data['status_updates']
+
+
+for c, post in enumerate(data):
+    #print('--- doing %d ---' % (c))
+    if 'title' not in post:
+        #logging.error('#%s no title', c)
+        continue
+    if 'new photo' not in post['title']:
+        #logging.error('#%s not a photo', c)
+        continue
+    if 'attachments' not in post:
+        #logging.error('#%s no attachments', c)
+        continue
+    if 'data' not in post['attachments'][0]:
+        #logging.error('#%s no data n attachments[0]', c)
+        continue
+    if 'data' not in post:
+        #logging.error('#%s no data in post', c)
+        continue
+    if not len (post['attachments'][0]['data']):
+        #logging.error('#%s no attachments', c)
+        continue
+
+    out = []
+    dt = arrow.get(post['timestamp'])
+    content = ''
+    if 'post' in post['data'][0]:
+        content = post['data'][0]['post']
+
+    if len(post['attachments'][0]['data']) == 1:
+        t = 'image'
+    else:
+        t = 'gallery'
+
+
+    cmd = "post_id=$(wp post create --porcelain --post_category='Facebook' --post_date='%s' --post_content=%s --post_title='%s')" % (
+        str(dt),
+        quote(content),
+        dt.format('YYYY-MM-DD HH:mm')
+    )
+    out.append(cmd)
+
+    for cntr, media in enumerate(post['attachments'][0]['data']):
+        media = media['media']
+
+        if 'uri' not in media:
+            logging.error('missing uri')
+            continue
+        fpath = "%s/%s" % (base,media['uri'])
+        fdt = arrow.get(media['creation_timestamp'])
+        if 'description' in media:
+            fcontent = media['description']
+        elif 'post' in post['data'][0]:
+            fcontent = post['data'][0]['post']
+
+        if cntr == 0:
+            f = '--featured_image'
+        else:
+            f = ''
+        out.append("wp-cli media import %s --post_id=${post_id} --title='%s' --caption=%s %s" % (
+            fpath,
+            os.path.basename(media['uri']),
+            quote(fcontent),
+            f
+        ))
+
+    if len(out) == 1:
+        continue
+    else:
+        for l in out:
+            print(l)
+        print('')
+        print('')