From d7bb5dc1c7b40cef1f1737d39f8498b8dc14d590 Mon Sep 17 00:00:00 2001
From: Bob Mottram <bob@libreserver.org>
Date: Mon, 11 Apr 2022 13:13:04 +0100
Subject: [PATCH] Fixing edit diffs

---
 content.py | 74 +++++++++++++++++++++++-------------------------------
 daemon.py  |  2 +-
 tests.py   | 72 +++++++++++++++++++++++++++++++++++++++++++++++++---
 3 files changed, 100 insertions(+), 48 deletions(-)

diff --git a/content.py b/content.py
index 81c85631f..be1a07503 100644
--- a/content.py
+++ b/content.py
@@ -1426,51 +1426,39 @@ def content_diff(content: str, prev_content: str) -> str:
     """
     d = difflib.Differ()
     text1_lines = content.splitlines()
-    text2_lines = prev_content.splitlines()
-    diff = d.compare(text1_lines, text2_lines)
+    text1_sentences = []
+    for line in text1_lines:
+        sentences = line.split('.')
+        for sentence in sentences:
+            text1_sentences.append(sentence.strip())
+
+    text2_lines = prev_content.splitlines()
+    text2_sentences = []
+    for line in text2_lines:
+        sentences = line.split('.')
+        for sentence in sentences:
+            text2_sentences.append(sentence.strip())
+
+    diff = d.compare(text1_sentences, text2_sentences)
 
-    diff_content = content
     diff_text = ''
-    reference_text = ''
-    change_positions = ''
     for line in diff:
         if line.startswith('- '):
-            reference_text = line[2:]
-        elif line.startswith('? '):
-            change_positions = line[2:]
-            state = 0
-            ctr = 0
-            for changed_char in change_positions:
-                if state == 0:
-                    if changed_char == '-':
-                        diff_text += \
-                            '<label class="diff_add">' + reference_text[ctr]
-                        ctr += 1
-                        state = 1
-                        continue
-                    elif changed_char == '+':
-                        diff_text += \
-                            '<label class="diff_remove">' + reference_text[ctr]
-                        ctr += 1
-                        state = 1
-                        continue
-                elif state == 1:
-                    if changed_char != '-' and changed_char != '+':
-                        diff_text += '</label>' + reference_text[ctr]
-                        ctr += 1
-                        state = 0
-                        continue
-                diff_text += reference_text[ctr]
-                ctr += 1
-
-            if state == 1:
-                diff_text += '</label>'
-
-            while ctr < len(reference_text):
-                diff_text += reference_text[ctr]
-                ctr += 1
-            diff_content = diff_content.replace(reference_text, diff_text)
-    return diff_content
+            if not diff_text:
+                diff_text = '<p>'
+            else:
+                diff_text += '<br>'
+            diff_text += '<label class="diff_add">+ ' + line[2:] + '</label>'
+        elif line.startswith('+ '):
+            if not diff_text:
+                diff_text = '<p>'
+            else:
+                diff_text += '<br>'
+            diff_text += \
+                '<label class="diff_remove">- ' + line[2:] + '</label>'
+    if diff_text:
+        diff_text += '</p>'
+    return diff_text
 
 
 def create_edits_html(edits_json: {}, post_json_object: {},
@@ -1507,11 +1495,11 @@ def create_edits_html(edits_json: {}, post_json_object: {},
         datetime_object = \
             convert_published_to_local_timezone(datetime_object, timezone)
         modified_str = datetime_object.strftime("%a %b %d, %H:%M")
-        diff = '<p><b>' + modified_str + '</b></p><p>' + diff + '</p>'
+        diff = '<p><b>' + modified_str + '</b></p>' + diff
         edits_str += diff
         content = prev_content
     if not edits_str:
         return ''
     return '<br><details><summary class="cw">' + \
         translate['SHOW EDITS'] + '</summary>' + \
-        edits_str + '</details>\n'
+        edits_str + '</details>'
diff --git a/daemon.py b/daemon.py
index 3246da238..c9a4aa6f8 100644
--- a/daemon.py
+++ b/daemon.py
@@ -20792,7 +20792,7 @@ def run_daemon(check_actor_timeout: int,
         check_actor_timeout = 2
     httpd.check_actor_timeout = check_actor_timeout
 
-    # how many hours after a post was publushed can a reply be made
+    # how many hours after a post was published can a reply be made
     default_reply_interval_hrs = 9999999
     httpd.default_reply_interval_hrs = default_reply_interval_hrs
 
diff --git a/tests.py b/tests.py
index 2bc945046..6eacaa61f 100644
--- a/tests.py
+++ b/tests.py
@@ -129,6 +129,7 @@ from inbox import json_post_allows_comments
 from inbox import valid_inbox
 from inbox import valid_inbox_filenames
 from categories import guess_hashtag_category
+from content import create_edits_html
 from content import content_diff
 from content import bold_reading_string
 from content import safe_web_text
@@ -6921,15 +6922,78 @@ def _test_bold_reading() -> None:
 def _test_diff_content() -> None:
     print('diff_content')
     prev_content = \
-        'Some text before.\nThis is some content.\nThis is another line.'
+        'Some text before.\n' + \
+        'Starting sentence. This is some content.\nThis is another line.'
     content = \
         'Some text before.\nThis is some more content.\nThis is another line.'
     result = content_diff(content, prev_content)
-    expected = 'Some text before.\n' + \
-        'This is some <label class="diff_add">' + \
-        'more </label>content.\nThis is another line.'
+    expected = \
+        '<p><label class="diff_remove">' + \
+        '- Starting sentence</label><br><label class="diff_add">' + \
+        '+ This is some more content</label><br>' + \
+        '<label class="diff_remove">- This is some content</label></p>'
     assert result == expected
 
+    content = \
+        'Some text before.\nThis is content.\nThis line.'
+    result = content_diff(content, prev_content)
+    expected = \
+        '<p><label class="diff_remove">- Starting sentence</label><br>' + \
+        '<label class="diff_add">+ This is content</label><br>' + \
+        '<label class="diff_remove">- This is some content</label><br>' + \
+        '<label class="diff_add">+ This line</label><br>' + \
+        '<label class="diff_remove">- This is another line</label></p>'
+    assert result == expected
+
+    translate = {
+        "SHOW EDITS": "SHOW EDITS"
+    }
+    timezone = 'Europe/Berlin'
+    content1 = \
+        "<p>This is some content.</p>" + \
+        "<p>Some other content.</p>"
+    content2 = \
+        "<p>This is some previous content.</p>" + \
+        "<p>Some other previous content.</p>"
+    content3 = \
+        "<p>This is some more previous content.</p>" + \
+        "<p>Some other previous content.</p>"
+    post_json_object = {
+        "object": {
+            "content": content1,
+            "published": "2020-12-14T00:08:06Z"
+        }
+    }
+    edits_json = {
+        "2020-12-14T00:05:19Z": {
+            "object": {
+                "content": content3,
+                "published": "2020-12-14T00:05:19Z"
+            }
+        },
+        "2020-12-14T00:07:34Z": {
+            "object": {
+                "content": content2,
+                "published": "2020-12-14T00:07:34Z"
+            }
+        }
+    }
+    html_str = \
+        create_edits_html(edits_json, post_json_object, translate, timezone)
+    assert html_str
+    expected = \
+        '<br><details><summary class="cw">SHOW EDITS</summary>' + \
+        '<p><b>Mon Dec 14, 01:07</b></p><p><label class="diff_add">' + \
+        '+ This is some content</label><br><label class="diff_remove">' + \
+        '- This is some previous content</label><br>' + \
+        '<label class="diff_add">+ Some other content</label><br>' + \
+        '<label class="diff_remove">- Some other previous content' + \
+        '</label></p><p><b>Mon Dec 14, 01:05</b></p><p>' + \
+        '<label class="diff_add">+ This is some previous content' + \
+        '</label><br><label class="diff_remove">' + \
+        '- This is some more previous content</label></p></details>'
+    assert html_str == expected
+
 
 def run_all_tests():
     base_dir = os.getcwd()