add wordpress migration

dnscms/news/models.py

@@ -5,6 +5,7 @@ from wagtail.admin.panels import FieldPanel
 from wagtail.fields import RichTextField
 from wagtail.models import Page
 from wagtail.search import index
+from wagtail_wordpress_import.models import WPImportedPageMixin
 
 from dnscms.fields import CommonStreamField
 
@@ -27,7 +28,7 @@ class NewsIndex(Page):
     search_fields = []
 
 
-class NewsPage(Page):
+class NewsPage(WPImportedPageMixin, Page):
     subpage_types = []
     parent_page_types = ["news.NewsIndex"]
     show_in_menus = False
@@ -74,3 +75,54 @@ class NewsPage(Page):
         index.SearchField("lead", boost=1),
         index.SearchField("body"),
     ]
+
+    settings_panels = Page.settings_panels + WPImportedPageMixin.wordpress_panels
+
+    def import_wordpress_data(self, data):
+        import html
+
+        from bs4 import BeautifulSoup
+
+        def generate_excerpt(html_content):
+            soup = BeautifulSoup(html_content, features="lxml")
+            VALID_TAGS = ["div", "p"]
+
+            for tag in soup.findAll("p"):
+                if tag.name not in VALID_TAGS:
+                    tag.remove()
+
+            text = soup.get_text().strip()
+            words = text.split(" ")
+            if len(words) < 26:
+                return text
+            return " ".join(words[:25]) + " [...]"
+
+        # Wagtail page model fields
+        self.title = html.unescape(data["title"])
+        self.slug = data["slug"]
+        self.first_published_at = data["first_published_at"]
+        self.last_published_at = data["last_published_at"]
+        self.latest_revision_created_at = data["latest_revision_created_at"]
+        self.search_description = data["search_description"]
+
+        # debug fields
+        self.wp_post_id = data["wp_post_id"]
+        self.wp_post_type = data["wp_post_type"]
+        self.wp_link = data["wp_link"]
+        self.wp_raw_content = data["wp_raw_content"]
+        self.wp_block_json = data["wp_block_json"]
+        self.wp_processed_content = data["wp_processed_content"]
+        self.wp_normalized_styles = data["wp_normalized_styles"]
+        self.wp_post_meta = data["wp_post_meta"]
+
+        # own model fields
+        self.body = data["body"] or ""
+
+        meta = data["wp_post_meta"]
+
+        written_excerpt = meta.get("excerpt_encoded")
+        generated_excerpt = ""
+        if not written_excerpt:
+            generated_excerpt = generate_excerpt(self.wp_processed_content)
+
+        self.excerpt = written_excerpt or generated_excerpt or "[...]"