From 068072658221329c680730dab197c73d75539d89 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Carlos=20Cancino-Chac=C3=B3n?=
 <carloscancinochacon@gmail.com>
Date: Sat, 18 Jan 2025 15:07:45 +0100
Subject: [PATCH] fix issue parsing incorrectly formatted fingering in MusicXML

---
 partitura/io/importmusicxml.py | 27 ++++++++++++++++++++++++---
 partitura/score.py             | 33 ++++++++++++++++++++++++++++++++-
 partitura/utils/misc.py        | 23 ++++++++++++++++++++++-
 3 files changed, 78 insertions(+), 5 deletions(-)

diff --git a/partitura/io/importmusicxml.py b/partitura/io/importmusicxml.py
index 69b7a04a..2632432c 100644
--- a/partitura/io/importmusicxml.py
+++ b/partitura/io/importmusicxml.py
@@ -21,7 +21,7 @@
 import partitura.score as score
 from partitura.score import assign_note_ids
 from partitura.utils import ensure_notearray
-from partitura.utils.misc import deprecated_alias, deprecated_parameter, PathLike
+from partitura.utils.misc import deprecated_alias, deprecated_parameter, PathLike, parse_ints
 
 __all__ = ["load_musicxml", "musicxml_to_notearray"]
 
@@ -1677,8 +1677,29 @@ def get_technical_notations(e: etree._Element) -> List[score.NoteTechnicalNotati
 
 def parse_fingering(e: etree._Element) -> score.Fingering:
 
-    fingering = score.Fingering(fingering=int(e.text))
-
+    try:
+        # There seems to be a few cases with fingerings encoded like 4_1.
+        # This is not standard in MusicXML according to the documentation, 
+        # but since it appears in files from the web, and can be displayed 
+        # with MuseScore, the solution for now is just to take the fist value.
+        finger_info = parse_ints(e.text)
+    except Exception as e:
+        # Do not raise an error if fingering info cannot be parsed, insted
+        # just set it as None.
+        warnings.warn(f"Cannot parse fingering info for {e.text}!")
+        finger_info = [None]
+
+    is_alternate = e.attrib.get("alternate", False)
+    is_substitution = e.attrib.get("substitution", False)
+    placement = e.attrib.get("placement", None)
+
+    # If there is more than one finger, only take the first one
+    fingering = score.Fingering(
+        fingering=finger_info[0],
+        is_substitution=is_alternate,
+        is_alternate=is_alternate,
+        placement=placement,
+    )
     return fingering
 
 
diff --git a/partitura/score.py b/partitura/score.py
index 49477288..454c59cd 100644
--- a/partitura/score.py
+++ b/partitura/score.py
@@ -3396,12 +3396,43 @@ def __init__(self, type: str, info: Optional[Any] = None) -> None:
 
 
 class Fingering(NoteTechnicalNotation):
-    def __init__(self, fingering: int) -> None:
+    """
+    This object represents fingering. For now, it supports attributes
+    present in MusicXML:
+
+    https://www.w3.org/2021/06/musicxml40/musicxml-reference/elements/fingering/
+
+    Parameters
+    ----------
+    fingering : Optional[int]
+        Fingering information. Can be None (usually the result of incorrect parsing of fingering).
+
+    is_substitution: bool
+        Whether this fingering is a substitution in the middle of a note. Default is False
+    
+    is_alternate: bool
+        Whether this fingering is an alternative fingering. Default is False
+
+    placement: str
+        Placement of the fingering (above or below a note)
+    """
+
+    def __init__(
+        self,
+        fingering: Optional[int],
+        is_substitution: bool = False,
+        placement: Optional[str] = None,
+        is_alternate: bool = False,
+    ) -> None:
         super().__init__(
             type="fingering",
             info=fingering,
         )
         self.fingering = fingering
+        self.is_alternate = is_alternate
+        self.alternative_fingering = []
+        self.is_substitution = is_substitution
+        self.placement = placement
 
 
 class PartGroup(object):
diff --git a/partitura/utils/misc.py b/partitura/utils/misc.py
index 3695e189..c38c13e8 100644
--- a/partitura/utils/misc.py
+++ b/partitura/utils/misc.py
@@ -8,8 +8,9 @@
 import warnings
 from urllib.request import urlopen
 from shutil import copyfileobj
+import re
 
-from typing import Union, Callable, Dict, Any, Iterable, Optional
+from typing import Union, Callable, Dict, Any, Iterable, Optional, List
 
 import numpy as np
 
@@ -280,3 +281,23 @@ def download_file(
     """
     with urlopen(url) as in_stream, open(out, "wb") as out_file:
         copyfileobj(in_stream, out_file)
+
+
+def parse_ints(input_string: str) -> List[int]:
+    """
+    Parse all numbers from a given string where numbers are separated by spaces or tabs.
+
+    Parameters
+    ----------
+    input_string : str
+        The input string containing numbers separated by spaces or tabs.
+
+    Returns
+    -------
+    List[int]
+        A list of integers extracted from the input string.
+    """
+    # Regular expression to match numbers
+    pattern = r'\d+'
+    # Find all matches and convert them to integers
+    return list(map(int, re.findall(pattern, input_string)))
\ No newline at end of file