analyze_html: display the last url of each shortcode

This allows for finding examples of where they are used in
the existing site.

Author: ahayzen-kdab

wagtail_wordpress_import/analysis.py

@@ -20,6 +20,7 @@ def __init__(self):
         self.styles_unique_pages = Counter()
         self.classes_unique_pages = Counter()
         self.shortcodes_unique_pages = Counter()
+        self.shortcodes_page_url = {}
 
     @classmethod
     def find_all_tags(cls, dom):
@@ -91,7 +92,7 @@ def find_all_shortcodes(cls, dom):
 
         return shortcodes
 
-    def analyze(self, html):
+    def analyze(self, html, page_url):
         self.total += 1
 
         try:
@@ -117,3 +118,6 @@ def analyze(self, html):
         self.styles_unique_pages.update(styles.keys())
         self.classes_unique_pages.update(classes.keys())
         self.shortcodes_unique_pages.update(shortcodes.keys())
+
+        for shortcode in shortcodes.keys():
+            self.shortcodes_page_url[shortcode] = page_url

wagtail_wordpress_import/importers/wordpress.py

@@ -263,9 +263,9 @@ def analyze_html(self, html_analyzer, *, page_types, page_statuses):
                     item.get("wp:post_type") in page_types
                     and item.get("wp:status") in page_statuses
                 ):
-
                     html_analyzer.analyze(
-                        filter_linebreaks_wp(item.get("content:encoded"))
+                        filter_linebreaks_wp(item.get("content:encoded")),
+                        item.get("link"),
                     )
 
     def connect_richtext_page_links(self, imported_pages):

wagtail_wordpress_import/management/commands/analyze_html_content.py

@@ -112,10 +112,11 @@ def handle(self, **options):
             "Shortcode",
             "Pages used on",
             "Total occurrences",
+            "Last URL",
         ]
         for shortcode, total_pages in analyzer.shortcodes_unique_pages.most_common():
             shortcodes_table.add_row(
-                [shortcode, total_pages, analyzer.shortcodes_total[shortcode]]
+                [shortcode, total_pages, analyzer.shortcodes_total[shortcode], analyzer.shortcodes_page_url[shortcode]]
             )
 
         self.stdout.write("Most commonly used shortcodes")