df

cmdlet/add_tag.py

@@ -47,6 +47,13 @@ def _normalize_title_for_extract(text: str) -> str:
     s = s.replace("\u2011", "-")  # non-breaking hyphen
     s = s.replace("\u2012", "-")  # figure dash
     s = s.replace("\u2015", "-")  # horizontal bar
+
+    # Collapse any whitespace runs (including newlines/tabs) to a single space.
+    # Some sources wrap the artist name or title across lines.
+    try:
+        s = re.sub(r"\s+", " ", s).strip()
+    except Exception:
+        s = " ".join(s.split())
     return s
 
 
@@ -70,7 +77,7 @@ def _literal_to_title_pattern_regex(literal: str) -> str:
         if ch.isspace():
             while i < len(literal) and literal[i].isspace():
                 i += 1
-            out.append(r"\\s*")
+            out.append(r"\s*")
             continue
         out.append(re.escape(ch))
         i += 1
@@ -95,7 +102,7 @@ def _compile_extract_template(template: str) -> tuple[re.Pattern[str], List[str]
         raise ValueError("extract template must contain at least one (field)")
 
     field_names: List[str] = []
-    parts: List[str] = [r"^\\s*"]
+    parts: List[str] = [r"^\s*"]
     last_end = 0
 
     for idx, m in enumerate(matches):
@@ -108,18 +115,24 @@ def _compile_extract_template(template: str) -> tuple[re.Pattern[str], List[str]
             raise ValueError(f"invalid field name '{raw_name}' (use A-Z, 0-9, underscore)")
         field_names.append(raw_name)
 
+        name_lower = raw_name.lower()
         is_last = idx == (len(matches) - 1)
         if is_last:
             parts.append(fr"(?P<{raw_name}>.+)")
         else:
-            parts.append(fr"(?P<{raw_name}>.+?)")
+            # Heuristic: common numeric fields should capture full digit runs.
+            # This avoids ambiguous splits like track='2', title='3 ...'.
+            if name_lower in {"disk", "disc", "cd", "track", "trk", "episode", "ep", "season", "year"}:
+                parts.append(fr"(?P<{raw_name}>\d+)")
+            else:
+                parts.append(fr"(?P<{raw_name}>.+?)")
 
         last_end = m.end()
 
     tail = tpl[last_end:]
     if tail:
         parts.append(_literal_to_title_pattern_regex(tail))
-    parts.append(r"\\s*$")
+    parts.append(r"\s*$")
 
     rx = "".join(parts)
     return re.compile(rx, flags=re.IGNORECASE), field_names