Update title normalization test

cinemas/riversidestudios.co.uk/transform.js

@@ -9,9 +9,26 @@ const {
 } = require("../../common/utils");
 const attributes = require("./attributes");
 
+function formatTable($, $el) {
+  return $el
+    .find("tr")
+    .map((_, row) => {
+      const cells = $(row).find("td");
+      if (cells.length < 2) return null;
+      const key = getText(cells.eq(0)).replace(/:$/, "");
+      const value = getText(cells.eq(1));
+      return key && value ? `${key}: ${value}` : null;
+    })
+    .get()
+    .filter(Boolean)
+    .join("\n");
+}
+
 function getOverviewData(pageData) {
   const $ = cheerio.load(pageData);
   const data = {};
+
+  // First, try to get data from .event-details .event-detail-section
   $(".event-details .event-detail-section").each(function () {
     const key = basicNormalize(getText($(this).find("div").eq(0)));
     const value = getText($(this).find("div").eq(1));
@@ -30,13 +47,62 @@ function getOverviewData(pageData) {
       }
     }
   });
+
+  // Also try .te figure table (newer format)
+  $(".te figure table tr").each(function () {
+    const cells = $(this).find("td");
+    if (cells.length < 2) return;
+    const key = basicNormalize(getText(cells.eq(0)));
+    const value = getText(cells.eq(1));
+    switch (key) {
+      case "year of release:": {
+        data.year = value;
+        break;
+      }
+      case "director:": {
+        data.directors = value;
+        break;
+      }
+      case "cast:": {
+        data.actors = value;
+        break;
+      }
+      case "age rating:": {
+        data.classification = value;
+        break;
+      }
+    }
+  });
+
   return data;
 }
 
+/**
+ * Formats the .te content for use as overview text in matchingHints.
+ * Ensures good spacing between paragraphs and formats tables nicely.
+ */
+function formatOverviewText(pageData) {
+  const $ = cheerio.load(pageData);
+  const $te = $(".te");
+  if (!$te.length) return "";
+  return $te
+    .children()
+    .map((_, el) => {
+      const $el = $(el);
+      const $table = $el.find("table");
+      return $table.length ? formatTable($, $table) : getText($el);
+    })
+    .get()
+    .filter(Boolean)
+    .join("\n\n");
+}
+
 async function transform({ movieListPage, moviePages }, sourcedEvents) {
   const movies = movieListPage.map((movieData) => {
     const { title, url: urlRaw } = movieData;
-    const { year, directors, actors } = getOverviewData(moviePages[urlRaw]);
+    const { year, directors, actors, classification } = getOverviewData(
+      moviePages[urlRaw],
+    );
     const url = encodeURI(urlRaw);
     const showingId = generateShowingId(attributes, movieData.id);
 
@@ -45,7 +111,7 @@ async function transform({ movieListPage, moviePages }, sourcedEvents) {
       year,
       directors,
       actors,
-      classification: movieData.age_rating_class,
+      classification: movieData.age_rating_class || classification,
     });
 
     const performances = Object.keys(movieData.performances).flatMap(
@@ -102,7 +168,7 @@ async function transform({ movieListPage, moviePages }, sourcedEvents) {
       overview,
       performances: uniquePerformances,
       matchingHints: {
-        overview: getText(cheerio.load(moviePages[urlRaw])(".te")),
+        overview: formatOverviewText(moviePages[urlRaw]),
       },
     };
   });

common/known-removable-phrases.js

@@ -100,6 +100,7 @@ const knownRemovablePhrases = [
   " with score by Pet Shop Boys",
   " with live score by Maxim Melton",
   " with live music from millie turner",
+  " with live original score",
   " with original score by Body Orchestra",
   " Music by Pet Shop Boys",
   " with Ranjit S. Ruprai",
@@ -643,6 +644,7 @@ const knownRemovablePhrases = [
   "SAFAR ",
   "Scared To Dance:",
   "Schools Screening IWD26:",
+  "Doc Screening & Discussion",
   "Screening & Social",
   "Screening & Q&A:",
   "Screening + Q&A:",
@@ -715,6 +717,7 @@ const knownRemovablePhrases = [
   "UFF 2025:",
   "UKJFF 2025:",
   "UKJFF:",
+  "Anti-Valentine's Day:",
   "Valentine's Day:",
   "Galentine's Day:",
   "Galentine’s Day:",
@@ -724,6 +727,7 @@ const knownRemovablePhrases = [
   "Visions of Ukraine:",
   "Visions of an Otherworld:",
   "Wallace & Gromit in ",
+  "Wallace & Gromit :",
   "William Shakespeare’s",
   "Windrush Caribbean Film Festival 2025:",
   "Women of Almodóvar:",
@@ -830,6 +834,7 @@ const knownRemovablePhrases = [
   "Preview Screenings",
   "preview screening",
   "Previews",
+  "[Preview]",
   "preview",
   "live action",
   "Sapphic",

common/tests/test-titles.json

@@ -50774,5 +50774,181 @@
   {
     "input": "White Men Can't Jump",
     "output": "white men cant jump"
+  },
+  {
+    "input": "THE CONSPIRACISTS + LIVE Q&A",
+    "output": "conspiracists"
+  },
+  {
+    "input": "Parent & Baby Screening: The Voice of Hind Rajab",
+    "output": "voice of hind rajab"
+  },
+  {
+    "input": "EcoDocs at Castlehaven Presents: The Grab Doc Screening & Discussion",
+    "output": "grab"
+  },
+  {
+    "input": "[Preview] A Private Life [Vie privée]",
+    "output": "a private life"
+  },
+  {
+    "input": "Mardaani 3 (Hindi)",
+    "output": "mardaani 3"
+  },
+  {
+    "input": "Euphoria (Telugu)",
+    "output": "euphoria"
+  },
+  {
+    "input": "Gandhi Talks (Hindi)",
+    "output": "gandhi talks"
+  },
+  {
+    "input": "Lockdown (Tamil)",
+    "output": "lockdown"
+  },
+  {
+    "input": "Paatki (Gujarati)",
+    "output": "paatki"
+  },
+  {
+    "input": "Prakambanam (Malayalam)",
+    "output": "prakambanam"
+  },
+  {
+    "input": "Sri Chidambaram (Telugu)",
+    "output": "sri chidambaram"
+  },
+  {
+    "input": "Valathu Vashathe Kallan (Malayalam)",
+    "output": "valathu vashathe kallan"
+  },
+  {
+    "input": "Film Jam Sunday Screening and Social (Feb)",
+    "output": "film jam sunday screening social"
+  },
+  {
+    "input": "The Testament of Ann Lee",
+    "output": "testament of ann lee"
+  },
+  {
+    "input": "Toddler Club: Lilo & Stitch (2002)",
+    "output": "lilo stitch (2002)"
+  },
+  {
+    "input": "Largo - cinema event",
+    "output": "largo"
+  },
+  {
+    "input": "3rd Emerging Filmmakers Showcase",
+    "output": "3rd emerging filmmakers showcase"
+  },
+  {
+    "input": "The Glass Essays",
+    "output": "glass essays"
+  },
+  {
+    "input": "Mardaani 3",
+    "output": "mardaani 3"
+  },
+  {
+    "input": "Prakambanam",
+    "output": "prakambanam"
+  },
+  {
+    "input": "Sinners (IMAX)",
+    "output": "sinners"
+  },
+  {
+    "input": "Valathu Vashathe Kallan",
+    "output": "valathu vashathe kallan"
+  },
+  {
+    "input": "Secret Screening (02-02-26)",
+    "output": "mystery movie"
+  },
+  {
+    "input": "Arco + Q&A with Natalie Portman and Ugo Bienvenu",
+    "output": "arco"
+  },
+  {
+    "input": "Crime 101 + Bart Layton Director Q&A",
+    "output": "crime 101"
+  },
+  {
+    "input": "Molly Vs the Machines + Recorded Q&A",
+    "output": "molly vs the machines"
+  },
+  {
+    "input": "Drink & Dine: One Battle After Another",
+    "output": "one battle after another"
+  },
+  {
+    "input": "Crystal Palace Friends of Palestine presents All That's Left of You + Q&A",
+    "output": "all thats left of you"
+  },
+  {
+    "input": "Dead Man's Wire",
+    "output": "dead mans wire"
+  },
+  {
+    "input": "Wallace & Gromit : A Matter of Loaf and Death",
+    "output": "a matter of loaf death"
+  },
+  {
+    "input": "Category H: OBJECT LOVERS! Christine & Killer Sofa",
+    "output": "object lovers christine killer sofa"
+  },
+  {
+    "input": "Molly Vs THE MACHINES + Discussion",
+    "output": "molly vs the machines"
+  },
+  {
+    "input": "Exploding Cinema",
+    "output": "exploding cinema"
+  },
+  {
+    "input": "Paratha Club",
+    "output": "paratha club"
+  },
+  {
+    "input": "Peckham Filmmakers - Winter Showcase",
+    "output": "peckham filmmakers"
+  },
+  {
+    "input": "Anti-Valentine's Day: Switchblade Romance",
+    "output": "switchblade romance"
+  },
+  {
+    "input": "Valentine's Day: 100 Nights of Hero",
+    "output": "100 nights of hero"
+  },
+  {
+    "input": "Valentine's Day: Pillion",
+    "output": "pillion"
+  },
+  {
+    "input": "Valentine's Day: The History of Sound",
+    "output": "history of sound"
+  },
+  {
+    "input": "'L'Inferno' (1911) with live original score",
+    "output": "linferno (1911)"
+  },
+  {
+    "input": "Guest Event - Anything that Bleeds: Book Launch",
+    "output": "anything that bleeds"
+  },
+  {
+    "input": "A Room With a View + Q&A with Keith Lodwick",
+    "output": "a room with a view"
+  },
+  {
+    "input": "The London International Animation Festival presents Wonderful Animated Shorts for 3 -12 year-olds",
+    "output": "wonderful animated shorts for 3"
+  },
+  {
+    "input": "LSFF x Nunhead Community Cinema: London Lives",
+    "output": "lsff x nunhead community cinema london lives"
   }
 ]