Refine high-frequency DMHY labels

tools/rust_dmhy_template_apply/src/main.rs

@@ -203,6 +203,10 @@ static CJK_SEASON_EMBEDDED_RE: Lazy<Regex> = Lazy::new(|| {
 });
 static ASCII_SEASON_SUFFIX_RE: Lazy<Regex> =
     Lazy::new(|| Regex::new(r"(?i)^(.+?)[\s_.-]+(S\d{1,2})$").unwrap());
+static ORDINAL_SEASON_TOKEN_RE: Lazy<Regex> =
+    Lazy::new(|| Regex::new(r"(?i)^\d{1,2}(?:st|nd|rd|th)$").unwrap());
+static SEASON_WORD_RE: Lazy<Regex> =
+    Lazy::new(|| Regex::new(r"(?i)^(?:Season|Saison)$").unwrap());
 static CJK_TITLE_LANG_PREFIX_RE: Lazy<Regex> =
     Lazy::new(|| Regex::new(r"^(.+?)(国日双语|國日雙語|日语版|日語版|国语版|國語版|双语|雙語)(第?)$").unwrap());
 static SEASON_VALUE_RE: Lazy<Regex> = Lazy::new(|| Regex::new(r"(?i)^S(\d{1,2})$").unwrap());
@@ -219,7 +223,7 @@ static DATE_RANGE_MIXED_RE: Lazy<Regex> = Lazy::new(|| {
 static CJK_DATE_RE: Lazy<Regex> =
     Lazy::new(|| Regex::new(r"^(?:19|20)\d{2}年\d{1,2}月\d{1,2}日$").unwrap());
 static LANG_RE: Lazy<Regex> = Lazy::new(|| {
-    Regex::new(r"(?i)^(?:CHS|CHT|ZHS|ZHT|GB|BIG5|JPN?|JP|JA|JAP|ENG|EN|SC|TC|简[体體]?|繁[体體]?|简日|繁日|字幕|内封|外挂|Sub|Subs|MSubs?)$").unwrap()
+    Regex::new(r"(?i)^(?:CHS|CHT|ZHS|ZHT|GB|BIG5|JPN?|JP|JA|JAP|JPTC|JPSC|ENG|EN|SC|TC|简[体體]?|繁[体體]?|简日|繁日|字幕|内封|外挂|Sub|Subs|MSubs?)$").unwrap()
 });
 static MEDIA_RE: Lazy<Regex> = Lazy::new(|| {
     Regex::new(r"(?i)^(?:WEB|WEB[-_. ]?DL|WEB[-_. ]?Rip|BDRip|BluRay|BDMV|BD|DVDRip|DVD|HD|UHD|HDTV|TVRip|REMUX|xvid|x26[45]|h\.?26[45]|HEVC|AVC|AV1|YUV\d+P?\d*|AAC\s*\d*(?:\.\d+)?|DDP\s*\d*(?:\.\d+)?|FLAC|MP3|DTS|HDMA|DTS-HDMA|E?AC3x?\d*(?:\.\d+)?|Opus|WMV\d*|\d(?:\.\d)?ch|10[-_. ]?bit|8[-_. ]?bit|Hi10p?|Ma10p|ASSx?\d*|SRTx?\d*|SUP|R\d[A-Z]*|NoSub|MKV|MP4|AVI|RAW|Raws?)$").unwrap()
@@ -1210,7 +1214,10 @@ fn process_filename(
             }
         }
     };
-    if recipe.count.unwrap_or(0) <= args.audit_max_count && has_blocking_low_frequency_warning(&record)
+    let warnings = audit_warnings(&record);
+    if warnings.iter().any(|warning| warning == "no_title")
+        || (recipe.count.unwrap_or(0) <= args.audit_max_count
+            && has_blocking_warnings(&warnings))
     {
         return Processed::Skipped {
             reason: "low_frequency_audit_warning",
@@ -1231,8 +1238,8 @@ fn process_filename(
     }
 }
 
-fn has_blocking_low_frequency_warning(record: &Record) -> bool {
-    audit_warnings(record).iter().any(|warning| {
+fn has_blocking_warnings(warnings: &[String]) -> bool {
+    warnings.iter().any(|warning| {
         matches!(
             warning.as_str(),
             "ambiguous_no_episode_title"
@@ -3092,6 +3099,13 @@ fn is_standalone_separator(token: &str) -> bool {
             .is_some_and(|ch| ch.is_whitespace() || !ch.is_alphanumeric())
 }
 
+fn is_unicode_roman_season(token: &str) -> bool {
+    matches!(
+        token,
+        "Ⅰ" | "Ⅱ" | "Ⅲ" | "Ⅳ" | "Ⅴ" | "Ⅵ" | "Ⅶ" | "Ⅷ" | "Ⅸ" | "Ⅹ" | "Ⅺ" | "Ⅻ"
+    )
+}
+
 fn split_cjk_title_lang_prefix(token: &str) -> Option<(Vec<String>, Vec<String>)> {
     let caps = CJK_TITLE_LANG_PREFIX_RE.captures(token)?;
     let title = caps.get(1)?.as_str();
@@ -3253,6 +3267,70 @@ fn smooth_title_spans(tokens: &[String], labels: &[String]) -> Vec<String> {
                 continue;
             }
         }
+        if label == "B-TITLE" && matches!(token.to_ascii_lowercase().as_str(), "ep" | "episode") {
+            let next_episode = (index + 1..tokens.len()).find(|&cursor| {
+                !joiners.contains(&tokens[cursor].as_str()) || labels[cursor] != "O"
+            });
+            if next_episode.is_some_and(|cursor| labels[cursor] == "B-EPISODE") {
+                output[index] = "O".to_string();
+                continue;
+            }
+        }
+        if label == "B-TITLE" && is_unicode_roman_season(token) {
+            let previous_title_word = (0..index).rev().find(|&cursor| {
+                output[cursor] == "B-TITLE"
+                    && tokens[cursor]
+                        .chars()
+                        .any(|ch| ch.is_alphanumeric() || ('\u{4e00}'..='\u{9fff}').contains(&ch))
+            });
+            let previous_word = previous_title_word.map(|cursor| tokens[cursor].to_ascii_lowercase());
+            if previous_title_word.is_some()
+                && !matches!(previous_word.as_deref(), Some("lupin"))
+            {
+                output[index] = "B-SEASON".to_string();
+                continue;
+            }
+        }
+        if label == "B-TITLE" && ORDINAL_SEASON_TOKEN_RE.is_match(token) {
+            let next_word = (index + 1..tokens.len()).find(|&cursor| {
+                !joiners.contains(&tokens[cursor].as_str())
+                    && tokens[cursor].chars().any(|ch| ch.is_alphabetic())
+            });
+            if next_word.is_some_and(|cursor| {
+                labels[cursor] == "B-TITLE" && SEASON_WORD_RE.is_match(&tokens[cursor])
+            }) {
+                output[index] = "B-SEASON".to_string();
+                if let Some(cursor) = next_word {
+                    for joiner_index in index + 1..cursor {
+                        if joiners.contains(&tokens[joiner_index].as_str()) {
+                            output[joiner_index] = "B-SEASON".to_string();
+                        }
+                    }
+                    output[cursor] = "B-SEASON".to_string();
+                }
+                continue;
+            }
+        }
+        if label == "O"
+            && token.chars().all(|ch| ch.is_ascii_digit())
+            && token.len() <= 3
+        {
+            let previous_non_space = (0..index)
+                .rev()
+                .find(|&cursor| !tokens[cursor].chars().all(char::is_whitespace));
+            let next_non_space = (index + 1..tokens.len())
+                .find(|&cursor| !tokens[cursor].chars().all(char::is_whitespace));
+            if previous_non_space.is_some_and(|cursor| tokens[cursor] == "[")
+                && next_non_space.is_some_and(|cursor| tokens[cursor] == "]")
+                && output[..index].iter().any(|label| label == "B-TITLE")
+                && output[index + 1..]
+                    .iter()
+                    .any(|label| matches!(label.as_str(), "B-SOURCE" | "B-RESOLUTION"))
+            {
+                output[index] = "B-EPISODE".to_string();
+                continue;
+            }
+        }
         if label == "B-EPISODE" && token.chars().all(|ch| ch.is_ascii_digit()) {
             let previous_non_space = (0..index)
                 .rev()
@@ -3275,13 +3353,19 @@ fn smooth_title_spans(tokens: &[String], labels: &[String]) -> Vec<String> {
                     continue;
                 }
             }
-            let previous_word = (0..index)
-                .rev()
-                .find(|&cursor| {
-                    !joiners.contains(&tokens[cursor].as_str()) && labels[cursor] != "O"
-                        || tokens[cursor].chars().any(|ch| ch.is_alphabetic())
-                })
-                .map(|cursor| tokens[cursor].to_ascii_lowercase());
+            let mut previous_word = None;
+            for cursor in (0..index).rev() {
+                if matches!(tokens[cursor].as_str(), "]" | "】" | ")" | "）") {
+                    break;
+                }
+                if joiners.contains(&tokens[cursor].as_str()) {
+                    continue;
+                }
+                if tokens[cursor].chars().any(|ch| ch.is_alphabetic()) {
+                    previous_word = Some(tokens[cursor].to_ascii_lowercase());
+                }
+                break;
+            }
             if matches!(previous_word.as_deref(), Some("lesson" | "part" | "no")) {
                 output[index] = "O".to_string();
                 continue;
@@ -3560,14 +3644,16 @@ mod tests {
         let bang_season =
             labels_for("[LoliHouse] Bang Dream! 2nd Season - 01 [BDRip 1080p HEVC-10bit FLAC]");
         assert!(bang_season.contains(&("Bang".to_string(), "B-TITLE".to_string())));
-        assert!(bang_season.contains(&("Season".to_string(), "B-TITLE".to_string())));
+        assert!(bang_season.contains(&("2nd".to_string(), "B-SEASON".to_string())));
+        assert!(bang_season.contains(&("Season".to_string(), "B-SEASON".to_string())));
         assert!(bang_season.contains(&("01".to_string(), "B-EPISODE".to_string())));
         assert!(!bang_season.contains(&("01".to_string(), "B-SEASON".to_string())));
 
         let basket =
             labels_for("[Nekomoe kissaten&VCB-Studio] Fruits Basket 1st Season [24][1080p][x264_aac][sc]");
         assert!(basket.contains(&("Fruits".to_string(), "B-TITLE".to_string())));
-        assert!(basket.contains(&("Season".to_string(), "B-TITLE".to_string())));
+        assert!(basket.contains(&("1st".to_string(), "B-SEASON".to_string())));
+        assert!(basket.contains(&("Season".to_string(), "B-SEASON".to_string())));
         assert!(basket.contains(&("24".to_string(), "B-EPISODE".to_string())));
         assert!(!basket.contains(&("24".to_string(), "B-SEASON".to_string())));
 
@@ -4012,5 +4098,52 @@ mod tests {
         let bilingual = labels_for("辉夜大小姐想让我告白~天才们的恋爱头脑战~.S2-01.中日双语.云光字幕组.[1080p]");
         assert!(bilingual.contains(&("中日".to_string(), "B-SOURCE".to_string())));
         assert!(!bilingual.contains(&("中日".to_string(), "B-TITLE".to_string())));
+
+        let nekomoe_lang = labels_for("[Nekomoe kissaten][UniteUp!][05][720p][JPTC]");
+        assert!(nekomoe_lang.contains(&("JPTC".to_string(), "B-SOURCE".to_string())));
+        assert!(!nekomoe_lang.contains(&("JPTC".to_string(), "B-TITLE".to_string())));
+
+        let hayate =
+            labels_for("[漏勺rip][Hayate_the_combat_butler_2nd_Season][23][BDrip X264 AAC 720P]");
+        assert!(hayate.contains(&("Hayate".to_string(), "B-TITLE".to_string())));
+        assert!(hayate.contains(&("2nd".to_string(), "B-SEASON".to_string())));
+        assert!(hayate.contains(&("Season".to_string(), "B-SEASON".to_string())));
+        assert!(hayate.contains(&("23".to_string(), "B-EPISODE".to_string())));
+
+        let jade = labels_for("[GM-Team][国漫][诛仙 第2季][Jade Dynasty ��][2024][12][AVC][GB][1080P]");
+        assert!(jade.contains(&("Jade".to_string(), "B-TITLE".to_string())));
+        assert!(jade.contains(&("Dynasty".to_string(), "B-TITLE".to_string())));
+        assert!(jade.contains(&("Ⅱ".to_string(), "B-SEASON".to_string())));
+        assert!(jade.contains(&("12".to_string(), "B-EPISODE".to_string())));
+
+        let yu_no = labels_for(
+            "[JYFanSub][Kono_Yo_no_Hate_de_Koi_wo_Utau_Shoujo_YU-NO][23][BIG5][720P][AVC]",
+        );
+        assert!(yu_no.contains(&("NO".to_string(), "B-TITLE".to_string())));
+        assert!(yu_no.contains(&("23".to_string(), "B-EPISODE".to_string())));
+
+        let fox = labels_for(
+            "[GM-Team][国漫][狐妖小红娘 尾生篇][Fox Spirit Matchmaker Ⅷ][2019][05][AVC][GB][1080P]",
+        );
+        assert!(fox.contains(&("Fox".to_string(), "B-TITLE".to_string())));
+        assert!(fox.contains(&("Ⅷ".to_string(), "B-SEASON".to_string())));
+
+        let kage = labels_for("[LKSUB][Kage no Jitsuryokusha ni Naritakute! 2nd Season][03][GB][720P]");
+        assert!(kage.contains(&("2nd".to_string(), "B-SEASON".to_string())));
+        assert!(kage.contains(&(" ".to_string(), "B-SEASON".to_string())));
+        assert!(kage.contains(&("Season".to_string(), "B-SEASON".to_string())));
+
+        let tiger = labels_for("[虎面人W][Tiger Mask W][01][简日][720p]");
+        assert!(tiger.contains(&("Tiger".to_string(), "B-TITLE".to_string())));
+        assert!(tiger.contains(&("W".to_string(), "B-TITLE".to_string())));
+        assert!(tiger.contains(&("01".to_string(), "B-EPISODE".to_string())));
+
+        let date_live_special =
+            labels_for("[ANK-Raws] デート・ア・ライブⅡ CM01 (BDrip 1920x1080 HEVC-YUV420P10 FLAC)");
+        assert!(date_live_special.contains(&("Ⅱ".to_string(), "B-SEASON".to_string())));
+        assert!(date_live_special.contains(&("CM01".to_string(), "B-SPECIAL".to_string())));
+
+        let ep_only = dmhy_record("Ep.25", "tpl_test", &suggested_roles("TEXT SEP EPISODE")).unwrap();
+        assert!(audit_warnings(&ep_only).contains(&"no_title".to_string()));
     }
 }