From fe2eb86fafd4fd655847b95a2ae12868b70926de Mon Sep 17 00:00:00 2001
From: Stuart Axon <stuaxo2@yahoo.com>
Date: Thu, 12 Oct 2017 00:45:59 +0100
Subject: [PATCH] pelican-import, fix urls of attached files

This patch wp-attach modifies ```<img``` and  ```<a href``` urls to point at the attachment path - so  ```http://example.com/wp-content/uploads/blah.png```   becomes    ```wp-content/uploads/blah.png```

This needs wp-content in pelicanconf STATIC_FILES

I thought I could use {filename} but since the importer is generates source html this doesn't work.
It would be nice if there was a way to customise the output path, but not sure how that would work.
---
 pelican/tests/test_importer.py  |  6 +++++-
 pelican/tools/pelican_import.py | 30 +++++++++++++++++++++++-------
 2 files changed, 28 insertions(+), 8 deletions(-)

diff --git a/pelican/tests/test_importer.py b/pelican/tests/test_importer.py
index 6af59212..63d17539 100644
--- a/pelican/tests/test_importer.py
+++ b/pelican/tests/test_importer.py
@@ -234,7 +234,11 @@ class TestWordpressXmlImporter(unittest.TestCase):
             with open(WORDPRESS_DECODED_CONTENT_SAMPLE, 'r') as decoded_file:
                 decoded_content = decoded_file.read()
                 self.assertEqual(
-                    decode_wp_content(encoded_content, br=False),
+                    decode_wp_content(
+                      encoded_content,
+                      attached_files=None,
+                      br=False
+                    ),
                     decoded_content)
 
     def test_preserve_verbatim_formatting(self):
diff --git a/pelican/tools/pelican_import.py b/pelican/tools/pelican_import.py
index 25fc45e5..26df1a4d 100755
--- a/pelican/tools/pelican_import.py
+++ b/pelican/tools/pelican_import.py
@@ -3,6 +3,7 @@
 from __future__ import print_function, unicode_literals
 
 import argparse
+import collections
 import logging
 import os
 import re
@@ -29,11 +30,22 @@ except ImportError:
 logger = logging.getLogger(__name__)
 
 
-def decode_wp_content(content, br=True):
+def decode_wp_content(content, attached_files=None, br=True):
     pre_tags = {}
     if content.strip() == "":
         return ""
 
+    if attached_files:
+        for path, urls in attached_files.items():
+            for url in urls:
+                content = re.sub(r'(<a\s+[^>]*href=")%s(")' % url,
+                                 r'\1/%s\2' % path,
+                                 content)
+
+                content = re.sub(r'(<img\s+[^>]*src=")%s(")' % url,
+                                 r'\1/%s\2' % path,
+                                 content)
+
     content += "\n"
     if "<pre" in content:
         pre_parts = content.split("</pre>")
@@ -651,10 +663,13 @@ def get_attachments(xml):
 
 
 def download_attachments(output_path, urls):
-    """Downloads WordPress attachments and returns a list of paths to
-    attachments that can be associated with a post (relative path to output
-    directory). Files that fail to download, will not be added to posts"""
-    locations = []
+    """Downloads WordPress attachments and returns a returns a dict {url:path} of
+        attachments that can be associated with a post (relative path to output
+        directory). Files that fail to download, will not be added to posts
+
+    {relpath: {set of urls}}
+    """
+    locations = collections.defaultdict(set)
     for url in urls:
         path = urlparse(url).path
         # teardown path and rebuild to negate any errors with
@@ -671,7 +686,8 @@ def download_attachments(output_path, urls):
         print('downloading {}'.format(filename))
         try:
             urlretrieve(url, os.path.join(full_path, filename))
-            locations.append(os.path.join(localpath, filename))
+            relpath = os.path.join(localpath, filename)
+            locations[relpath].add(url)
         except (URLError, IOError) as e:
             # Python 2.7 throws an IOError rather Than URLError
             logger.warning("No file could be downloaded from %s\n%s", url, e)
@@ -720,7 +736,7 @@ def fields2pelican(
                 # Replace newlines with paragraphs wrapped with <p> so
                 # HTML is valid before conversion
                 if in_markup == 'wp-html':
-                    new_content = decode_wp_content(content)
+                    new_content = decode_wp_content(content, attached_files)
                 else:
                     paragraphs = content.splitlines()
                     paragraphs = ['<p>{0}</p>'.format(p) for p in paragraphs]