From 313a211ddef8db5cd155b7d20ed86adc2d591a14 Mon Sep 17 00:00:00 2001
From: Bob Mottram <bob@libreserver.org>
Date: Wed, 16 Jul 2025 20:26:48 +0100
Subject: [PATCH] Fix another case where long words were getting through and
 trashing the UX

---
 content.py |  9 ++++----
 tests.py   | 61 ++++++++++++++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 66 insertions(+), 4 deletions(-)

diff --git a/content.py b/content.py
index 71ba30a05..277fd6a8b 100644
--- a/content.py
+++ b/content.py
@@ -1113,6 +1113,7 @@ def remove_long_words(content: str, max_word_length: int,
                     content = content[:max_word_length]
                 return content
     content = content.replace('<p></p>', '<p> </p>')
+    content = content.replace('<br />', '<br>')
     words = content.split(' ')
     if not long_words_list:
         long_words_list: list[str] = []
@@ -1169,10 +1170,10 @@ def remove_long_words(content: str, max_word_length: int,
             end_of_line_char = '\n'
             if '<br>' in original_word_str:
                 end_of_line_char = ''
-            content = content.replace(word_str,
-                                      word_str[:max_word_length] +
-                                      end_of_line_char +
-                                      word_str[max_word_length:])
+            new_word_str = \
+                word_str[:max_word_length] + end_of_line_char + \
+                word_str[max_word_length:]
+            content = content.replace(word_str, new_word_str)
         else:
             content = content.replace(word_str,
                                       word_str[:max_word_length])
diff --git a/tests.py b/tests.py
index 543707d2a..0e88caba8 100644
--- a/tests.py
+++ b/tests.py
@@ -3990,6 +3990,67 @@ def _test_web_links():
     result_text = remove_long_words(example_text, 40, [])
     assert result_text == '<p>1. HAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHA</p>'
 
+    example_text = \
+        '<p>printing it as</p><p>2037140369160192492322925579391203607' + \
+        '9180457601678813485162666958812204044214622117979705473511687' + \
+        '4592660294244917367332676764786341343508013819160968359130462' + \
+        '2042992012189361768911860908296446835744287698513143677144449' + \
+        '7938552838757164161177222742794307083740349375668764576182918' + \
+        '1191081223077464252347315511376155447223453567674374804815128' + \
+        '6735404590608216087337598386531730585596091790505424832645417' + \
+        '4535501077237920206100984089409165176752643554839846882599844' + \
+        '5923991409873771880931926782681047189948065387778396896132986' + \
+        '1506909307742354583410543953848817545579436209308661274911375' + \
+        '0279438885690899872309557867300649792692083221832567745997106' + \
+        '4361272558974579895067960526717371469973690968110845204652565' + \
+        '7587689165458042953094138457672314834155438608636666321168512' + \
+        '4252782578218268564249508506018479212668540667664012944414293' + \
+        '2330851074677344997595343747355492702171737174875079240941536' + \
+        '5577708112608640981357760962144998520758441440673068086203309' + \
+        '6319770638261320199660717718939289964587472883312324782316607' + \
+        '1967237252667926931453847363561981705896614769576786119378801' + \
+        '2882830466910136185864274111171019637988349519287519085606121' + \
+        '2522339449147264219612788230902092720858942438882884398008862' + \
+        '3971347873853102233621429454276071388173023720330558429380460' + \
+        '24463<br />-----<br />135631432782117671533556239831160434454' + \
+        '7335460488150228839142323597554809416586757525360501260323585' + \
+        '5427725242735144512118460721691926719820293873391318272145617' + \
+        '2554791642059210748685514914488339456875132466046368022599988' + \
+        '5273244295738339238990530543987019381897315877208424689309203' + \
+        '4437647713420965928758468739219412456408063053612145484613397' + \
+        '7772016123436107774252651067596615212806590262272555509568180' + \
+        '4017509174747230962281596577708797815066474134136273694278897' + \
+        '3360027388222852071012491607526037246576762645075722993280572' + \
+        '0175203303419048479968580491030484921302651482301409754854835' + \
+        '8571983490854257241872545704643626052023136119167363679870698' + \
+        '6658596724473374548909147593122088363315605434015319945902182' + \
+        '4528805007798699801368901189663857156868661234311411008127692' + \
+        '0776737318215252920723091949275668214683758105365336904270751' + \
+        '4269780883475016823742824846496720523040081390146960035213499' + \
+        '4242397562282441575512596963230050874858503383101416745601160' + \
+        '7883569793778503255464751436889383035257241831991489377524576' + \
+        '6486902236709319426301135526407141692658938140338022458562435' + \
+        '2058492483177210403138206901075955949723649750887887380112918' + \
+        '6077468349611419064296645075152353982955366217878670283274959' + \
+        '6553942067490556190686268506650055832177364114438608728336525' + \
+        '0621382626679227962349526822499666150757963545641335569627420' + \
+        '8921258670239960403498282550000556886617270910694586165105339' + \
+        '7011099950229146620093059412209127775142890468769710327385614' + \
+        '0122556856706927520947510344871870199933577438968706707773791' + \
+        '5845161008648178701657616293931863560534070691425550081565559' + \
+        '3409676811957993996940903554947847842104284053634021057864830' + \
+        '6688290446849712231828336090313297361727030502882056705996134' + \
+        '4733039520306167881855566393478866394412600612994195061492943' + \
+        '1274728867456552977222295601679486173993905936303907550874137' + \
+        '0019675194211557555048425375304</p><p>would have been way more' + \
+        ' useful. at least the line count would give a clue about ' + \
+        'magnitudes</p>"'
+    result_text = remove_long_words(example_text, 40, [])
+    assert len(result_text) < len(example_text)
+    assert result_text == \
+        '<p>printing it as have been way more useful. at least the ' + \
+        'line count would give a clue about magnitudes</p>"</p>'
+
     example_text = \
         '<p>Tox address is 88AB9DED6F9FBEF43E105FB72060A2D89F9B93C74' + \
         '4E8C45AB3C5E42C361C837155AFCFD9D448 </p>'