Balance synthetic movie and special path bases

tools/schema_v2_synthetic_augment/src/main.rs

@@ -963,7 +963,7 @@ fn path_seed_compatible_with_base(kind: PathSeedKind, base_kind: BaseMediaKind)
 }
 
 fn simple_path_leaf_records() -> Vec<Record> {
-    vec![
+    let mut records = vec![
         char_record_from_spans(
             "03.mkv",
             &[(0, 2, Entity::Episode)],
@@ -988,23 +988,67 @@ fn simple_path_leaf_records() -> Vec<Record> {
             "path_leaf_sxe",
             "path_aug",
         ),
-        char_record_from_spans(
-            "Movie.mkv",
-            &[(0, 5, Entity::Tag)],
-            "schema_v2_synthetic_path_leaf",
-            "path_leaf_movie",
-            "path_leaf_movie",
-            "path_aug",
-        ),
-        char_record_from_spans(
-            "NCOP2.mkv",
-            &[(0, 5, Entity::Special)],
-            "schema_v2_synthetic_path_leaf",
-            "path_leaf_special",
-            "path_leaf_special",
-            "path_aug",
-        ),
-    ]
+    ];
+
+    const MOVIE_TAGS: [&str; 6] = [
+        "Movie",
+        "Gekijouban",
+        "The Movie",
+        "Film",
+        "BDMovie",
+        "劇場版",
+    ];
+    const MOVIE_SUFFIXES: [&str; 12] = [
+        "",
+        " 01",
+        " 02",
+        " 2004",
+        " 1080p",
+        " BDRip",
+        " WEB-DL",
+        " Remux",
+        " Complete",
+        " Director Cut",
+        " Part A",
+        " Part B",
+    ];
+    for (tag_idx, tag) in MOVIE_TAGS.iter().enumerate() {
+        for (suffix_idx, suffix) in MOVIE_SUFFIXES.iter().enumerate() {
+            let filename = format!("{tag}{suffix}.mkv");
+            let template_id = format!("path_leaf_movie_{tag_idx}_{suffix_idx}");
+            records.push(char_record_from_spans(
+                &filename,
+                &[(0, tag.len(), Entity::Tag)],
+                "schema_v2_synthetic_path_leaf",
+                &template_id,
+                "path_leaf_movie",
+                "path_aug",
+            ));
+        }
+    }
+
+    const SPECIAL_TAGS: [&str; 12] = [
+        "NCOP", "NCOP2", "NCED", "NCED2", "OP", "OP2", "ED", "ED1", "PV", "CM", "Menu", "Trailer",
+    ];
+    const SPECIAL_SUFFIXES: [&str; 8] =
+        ["", " 01", " 02", " A", " B", " Clean", " Creditless", " v2"];
+    for (tag_idx, tag) in SPECIAL_TAGS.iter().enumerate() {
+        for (suffix_idx, suffix) in SPECIAL_SUFFIXES.iter().enumerate() {
+            let special_text = format!("{tag}{suffix}");
+            let filename = format!("{special_text}.mkv");
+            let template_id = format!("path_leaf_special_{tag_idx}_{suffix_idx}");
+            records.push(char_record_from_spans(
+                &filename,
+                &[(0, special_text.len(), Entity::Special)],
+                "schema_v2_synthetic_path_leaf",
+                &template_id,
+                "path_leaf_special",
+                "path_aug",
+            ));
+        }
+    }
+
+    records
 }
 
 fn char_record_from_spans(