From 502b472c0796fcb5bd5a5f57673bf27314497a76 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Ondr=CC=8Cej=20Ny=CC=81vlt?= <ondrej@nyv.lt>
Date: Fri, 25 Aug 2023 22:41:25 +0200
Subject: [PATCH] Customize diff to html rendering

---
 processor/app.py | 26 ++++++++++++++++++++++++--
 1 file changed, 24 insertions(+), 2 deletions(-)

diff --git a/processor/app.py b/processor/app.py
index 292b110..fae924c 100644
--- a/processor/app.py
+++ b/processor/app.py
@@ -11,6 +11,28 @@ import hashlib
 from diff_match_patch import diff_match_patch
 
 
+class DiffThing(diff_match_patch):
+    def diff_html(self, diffs):
+        """Like diff_prettyHtml, but without inline style attributes
+        (makes it easier to style it with stylesheets).
+        """
+        html = []
+        for op, data in diffs:
+            text = (
+                data.replace("&", "&amp;")
+                .replace("<", "&lt;")
+                .replace(">", "&gt;")
+                .replace("\n", "&para;<br>")
+            )
+            if op == self.DIFF_INSERT:
+                html.append("<ins>%s</ins>" % text)
+            elif op == self.DIFF_DELETE:
+                html.append('<del">%s</del>' % text)
+            elif op == self.DIFF_EQUAL:
+                html.append("<span>%s</span>" % text)
+        return "".join(html)
+
+
 #
 # Idea block:
 #
@@ -25,7 +47,7 @@ REDIS_ARTICLE_EXPIRE_SEC = 604800
 config = confuse.Configuration("headline", __name__)
 config.set_file(CONFIG_FILE)
 
-dmp = diff_match_patch()
+dmp = DiffThing()
 
 rc = redis.Redis(host="redis", port=6379, db=0)
 
@@ -88,7 +110,7 @@ def write_article(article, rc):
 def process_diff(old, new, rss_id):
     diff = dmp.diff_main(old["title"], new["title"])
     dmp.diff_cleanupSemantic(diff)
-    html_diff = dmp.diff_prettyHtml(diff)
+    html_diff = dmp.diff_html(diff)
     # print(old['link'])
     # print(diff)