Update title normalization test

common/is-sport-showing.js

@@ -19,6 +19,7 @@ const sportShowings = [
   /EUROPA LEAGUE QUALIFIER/i,
   /EUROPA LEAGUE FINAL/i,
   /Enterprise National League/i,
+  /Fifa World Cup/i,
   /FIFA Club World/i,
   /FIFA CWC/i,
   /NON-LEAGUE FINALS?/i,

common/known-removable-phrases.js

@@ -41,6 +41,7 @@ const knownRemovablePhrases = [
   "UK Charity Screening",
   "London Gala Screening",
   "Feature Screening",
+  "Repeat Screening",
   "***LONDON***",
   "***VHS***",
   "*ADVANCE*",
@@ -308,6 +309,7 @@ const knownRemovablePhrases = [
   "with Tim Robey",
   "with TOMISIN ADEPEJU",
   "with tommy wiseau live on stage",
+  "with Rebelle Voices",
   "with Sammy Baloji",
   "with Selina Robertson & Isabel Moir",
   "with Special Guest Sviatoslav Vakarchuk",
@@ -330,6 +332,9 @@ const knownRemovablePhrases = [
   "with Morgan M. Page",
   "with Ronan Bennett",
   "with score by Joe Harvey-Whyte",
+  "WITH HORRORITUAL!",
+  "with Free Chai & Samosas!",
+  "with Odorama!",
   "w/ Reece Shearsmith intro",
   "w/ Ben Wheatley Q&A",
   "w/ Peter Strickland Q&A",
@@ -512,8 +517,10 @@ const knownRemovablePhrases = [
   "by Rebecca Salvadori",
   "by Louisa May Alcott",
   "by Paromita Vohra",
+  "by Renátó Olasz",
   "by Fassbinder",
   "by Joe Harvey-Whyte",
+  "C.O.F.F.I.N '",
   "COLD ISLANDERS:",
   "Cold War Visions:",
   "Mamoru Hosoda's",
@@ -546,6 +553,7 @@ const knownRemovablePhrases = [
   "CELLULOID JAM!",
   "CinemaLive:",
   "CINESOCIAL:",
+  "CINEMA MENTIRÉ:",
   "Cinema Film Screening Konrad Wolf Film Series:",
   "Cinema Film Screening Konrad Wolf:",
   "Cinema Film Screening & Talk:",
@@ -564,6 +572,7 @@ const knownRemovablePhrases = [
   "Clermont-Ferrand:",
   "Closing Ceremony: “Shaw Festival – London 2026”",
   "Create 2026:",
+  "Copy of Copy of ",
   "Copy of ",
   "CIFF2025:",
   "CYFF:",
@@ -659,6 +668,7 @@ const knownRemovablePhrases = [
   "Senior Community Cinema x The Old Ways:",
   "Senior Community Cinema:",
   "Sneak Peek & Intro",
+  "Sneak Peek 2.0:",
   "The Old Ways:",
   "SCS: The Old Ways",
   "FILM CLUB FRIDAYS:",
@@ -750,6 +760,7 @@ const knownRemovablePhrases = [
   "The Woman Behind Visconti's Masterpiece",
   "Cinebug turns 1:",
   "Hussein Kamal's",
+  "Ido Fluk:",
   "IFF:",
   "IFF 2024:",
   "IFF 2025:",
@@ -986,6 +997,7 @@ const knownRemovablePhrases = [
   ": Staff Picks!",
   "Stephanie Rothman:",
   "Movie Subtitled",
+  "Movie Bottomless Brunch Party",
   "Subtitled screen ",
   "SUBTITLED ",
   "SUBTITLED:",
@@ -1076,6 +1088,7 @@ const knownRemovablePhrases = [
   ", a queer romance",
   "Documentary Screenings",
   "Documentary Screening",
+  "Documentary Premiere",
   "1994 version",
   "3rd annual screening",
   "4th annual screening",
@@ -1167,7 +1180,9 @@ const knownRemovablePhrases = [
   "Short Film Screenings",
   "Short Film Screening",
   "Bloomsday Film Screening",
+  ": London film screening",
   "Film Screening",
+  "Film premiere and",
   "Film Premiere:",
   "Fundraising Screening",
   "Fundraiser Q&A:",

common/normalize-title.js

@@ -497,7 +497,10 @@ function normalizeTitle(title, options) {
     [/(classic )?secret scre(e|a)(n|m)ing( \d+)?/i, "mystery movie"],
     [/secret (classic )?bollywood cinema/i, "mystery movie"],
     [/scre(e|a)(n|m) unseen/i, "mystery movie"],
-    [/(Orange Box )?Secret Film Screenings?/i, "mystery movie"],
+    [
+      /(Orange Box )?Secret Film Screenings?(: Summer Series)?/i,
+      "mystery movie",
+    ],
     [/^.* \+ mystery movie/i, "mystery movie"],
     ["vhs film", "movie"],
     [/The Bill Reunion \d+/i, "The Bill Reunion"],
@@ -728,11 +731,19 @@ function normalizeTitle(title, options) {
     ["A Night of Latin Jazz - ", "A Night of Latin Jazz: "],
     ["Classic Night - ", "Classic Night: "],
     ["Weird Wednesday - ", "Weird Wednesday: "],
+    ["Bar Trash: 4TH BIRTHDAY - ", "Bar Trash: 4TH BIRTHDAY: "],
+    ["Elon Musk Unveiled -", "Elon Musk Unveiled: "],
+    ["Goethe-Kino - ", "Goethe-Kino: "],
     ["The Life + Legacy", "The Life and Legacy"],
     [" - Oggi", ": Oggi"],
     [" x metropolis", " metropolis"],
     ["Nick Drake: A Skin Too Few", "A Skin Too Few: The Days of Nick Drake"],
     ["(Screening) / ", "(Screening) & "],
+    [
+      "BEYOND ILLUSION - MAGIC DOCUMENTARY",
+      "Beyond Illusion: The Making of a Magician - ",
+    ],
+    [/The Band - The Show/i, "The Band The Show"],
     // Variant families collapsed from known-removable-phrases.js
     // Each pattern covers multiple near-identical string entries that shared a common structure
     [/dog[- ]?friendly(?:\s+screening)?[:\s]*/i, ""],
@@ -987,6 +998,7 @@ function normalizeTitle(title, options) {
     .replace(/^the (?=\S+\s+(?![[(]))/i, "")
     .replace(/([a-z])-([a-z])/gi, "$1$2")
     .replace(/\s+q&a$/i, "")
+    .replace("?s", "s")
     .replace(/\s3d$/i, "")
     .replace(/[+?]$/, "")
     .replace(/\(\d{4}-[^)]+\)$/, "") // Remove any date range suffixes

common/standardize-prefixing-for-theatre-performances.js

@@ -74,8 +74,11 @@ function standardizePrefixingForMetropolitanOperaPerformances(title, options) {
   title = title.replace(/\s+&\s+/, " and ").replace(/\s+-\s+/, ": ");
 
   // Update if "met opera" is a suffix
-  if (title.toLowerCase().includes(": met opera ")) {
-    title = `The Metropolitan Opera: ${title.replace(/: Met Opera /i, " ")}`;
+  if (
+    title.toLowerCase().includes(": met opera ") ||
+    title.toLowerCase().includes(": the met opera ")
+  ) {
+    title = `The Metropolitan Opera: ${title.replace(/: (The )?Met Opera /i, " ")}`;
   }
 
   let updatedPrefixTitle = metOperaPrefixes.reduce(
@@ -144,6 +147,8 @@ const rboPrefixes = [
   /Royal Opera House[:|\s]/i,
   /^The Royal Ballet[:|\s]/i,
   /^The Royal Opera[:|\s]/i,
+  /: The Royal Ballet[:|\s]/i,
+  /: The Royal Opera[:|\s]/i,
   /RB&O Live:/i,
   /RB&O:/i,
   /Live From Royal Ballet/i,
@@ -152,6 +157,7 @@ const rboPrefixes = [
 function standardizePrefixingForRoyalBalletOperaPerformances(title) {
   title = title
     .replace(/Captured Live /i, "")
+    .replace(/-\s?Live\s?-/i, "-")
     .replace(/Hoffman(\s|$)/i, "Hoffmann$1")
     .replace(/\s+&\s+/, " and ")
     .replace(/\s+-\s+/, ": ");
@@ -257,7 +263,8 @@ function standardizePrefixingForTheatrePerformances(
       lowercaseTitle.includes("the metropolitan opera:")) ||
     (lowercaseTitle.startsWith("rbo ") &&
       lowercaseTitle.includes("the met opera -")) ||
-    lowercaseTitle.includes(" - met opera")
+    lowercaseTitle.includes(" - met opera") ||
+    lowercaseTitle.includes(" - the met opera")
   ) {
     return standardizePrefixingForMetropolitanOperaPerformances(title, options);
   }
@@ -270,7 +277,9 @@ function standardizePrefixingForTheatrePerformances(
     lowercaseTitle.startsWith("royal opera") ||
     lowercaseTitle.startsWith("royal ballet") ||
     lowercaseTitle.startsWith("the royal opera") ||
+    lowercaseTitle.includes("- the royal opera") ||
     lowercaseTitle.startsWith("the royal ballet") ||
+    lowercaseTitle.includes("- the royal ballet") ||
     lowercaseTitle.startsWith("roh royal opera") ||
     lowercaseTitle.startsWith("roh ") ||
     lowercaseTitle.startsWith("roh: ") ||