def __init__(self, doc, req_m, req_s, req_b, completeness=None):
self.doc = doc
self.flat_doc = doc.getFlattenedTree()
self.flat_doc_static = list(doc.getFlattenedTree())
self.docInfo = MusDocInfo(doc).get()
self.musicEl = doc.getElementsByName("music")[0]
self.measures = self.musicEl.getDescendantsByName("measure")
self.ema_exp = EmaExpression(self.docInfo, req_m, req_s, req_b,
completeness)
self.ema_measures = self.ema_exp.get()
self.compiled_exp = self.ema_exp.getCompiled()
def __init__(self, doc, req_m, req_s, req_b, completeness=None):
self.doc = doc
self.flat_doc = doc.getFlattenedTree()
self.flat_doc_static = list(doc.getFlattenedTree())
self.docInfo = MusDocInfo(doc).get()
self.musicEl = doc.getElementsByName("music")[0]
self.measures = self.musicEl.getDescendantsByName("measure")
self.ema_exp = EmaExpression(self.docInfo,
req_m,
req_s,
req_b,
completeness)
self.ema_measures = self.ema_exp.get()
self.compiled_exp = self.ema_exp.getCompiled()
class MeiSlicer(object):
"""Class for slicing an MEI doc given a split EMA expresion. """
def __init__(self, doc, req_m, req_s, req_b, completeness=None):
self.doc = doc
self.flat_doc = doc.getFlattenedTree()
self.flat_doc_static = list(doc.getFlattenedTree())
self.docInfo = MusDocInfo(doc).get()
self.musicEl = doc.getElementsByName("music")[0]
self.measures = self.musicEl.getDescendantsByName("measure")
self.ema_exp = EmaExpression(self.docInfo, req_m, req_s, req_b,
completeness)
self.ema_measures = self.ema_exp.get()
self.compiled_exp = self.ema_exp.getCompiled()
def slice(self):
""" Return a modified MEI doc containing the selected notation
provided a EMA expression of measures, staves, and beats."""
# parse general beats information
self.beatsInfo = self.docInfo["beats"]
self.timeChanges = self.beatsInfo.keys()
self.timeChanges.sort(key=int)
# Process measure ranges and store boundary measures
boundary_mm = []
for em in self.ema_measures:
boundary_mm.append(em.measures[0])
boundary_mm.append(em.measures[-1])
self.processContigRange(em)
# Recursively remove remaining data in between
# measure ranges before returning modified doc
# First remove measures in between
to_remove = []
middle_boundaries = boundary_mm[1:-1]
for bm in middle_boundaries[::2]:
i = middle_boundaries.index(bm)
try:
start_m = self.measures[bm.idx - 1]
start_m_pos = start_m.getPositionInDocument()
end_m = self.measures[middle_boundaries[i + 1].idx - 1]
end_m_pos = end_m.getPositionInDocument()
for el in self.flat_doc[start_m_pos + 1:end_m_pos]:
if el.hasAncestor("measure"):
if el.getAncestor(
"measure").getId() != start_m.getId():
to_remove.append(el)
else:
to_remove.append(el)
except IndexError:
pass
m_first = self.measures[boundary_mm[0].idx - 1]
m_final = self.measures[boundary_mm[-1].idx - 1]
# List of elements to keep
keep = ["meiHead"]
def _removeBefore(curEl):
parent = curEl.getParent()
if parent:
# Casting to list to avoid modfying the sequence
for el in list(curEl.getPeers()):
if el == curEl:
break
else:
if not el.getName() in keep:
parent.removeChild(el)
return _removeBefore(parent)
return curEl
def _removeAfter(curEl):
parent = curEl.getParent()
if parent:
removing = False
# Casting to list to avoid modfying the sequence
for el in list(curEl.getPeers()):
if removing:
if not el.getName() in keep:
parent.removeChild(el)
elif el == curEl:
removing = True
return _removeAfter(parent)
return curEl
def _findLowestCommonAncestor(el1, el2):
while not el1 == el2:
el1 = el1.getParent()
el2 = el2.getParent()
return el1
def _removeUnusedStaffDefs(scoredef, start_ema_m):
numbers = set()
for em in self.ema_measures:
if start_ema_m == em.measures[0]:
for m in em.measures:
for ema_s in m.staves:
numbers.add(ema_s.number)
break
for staffd in scoredef.getDescendantsByName("staffDef"):
if staffd.hasAttribute("n"):
val = staffd.getAttribute("n").getValue()
if int(val) not in numbers:
staffd.getParent().removeChild(staffd)
# TODO! Remove definitions of unselected staves WITHIN RANGE
# Compute closest score definition to start measure
preceding = self.flat_doc_static[:m_first.getPositionInDocument()]
first_scoreDef = None
for el in reversed(preceding):
if el.getName() == "scoreDef":
first_scoreDef = el
break
# Recursively remove elements before and after selected measures
_removeBefore(m_first)
_removeAfter(m_final)
# Compute closest score definition to start measure of each range
b_scoreDef = first_scoreDef
for i, bm in enumerate(
boundary_mm[::2]): # list comprehension get only start mm
b_measure = self.measures[bm.idx - 1]
preceding = self.flat_doc_static[:b_measure.getPositionInDocument(
)]
for el in reversed(preceding):
if el.getName() == "scoreDef":
try:
s_id = b_scoreDef.getId()
if not s_id == el.getId():
b_scoreDef = el
# Re-attach computed score definition
sd_copy = MeiElement(b_scoreDef)
_removeUnusedStaffDefs(sd_copy, bm)
b_measure.getParent().addChildBefore(
b_measure, sd_copy)
except AttributeError:
b_scoreDef = el
break
for el in to_remove:
el.getParent().removeChild(el)
if "raw" in self.ema_exp.completenessOptions:
lca = _findLowestCommonAncestor(m_first, m_final)
self.doc.setRootElement(lca)
# re-attach computed score definition if required by
# completeness = signature
if "signature" in self.ema_exp.completenessOptions:
m_first.getParent().addChildBefore(m_first, first_scoreDef)
return self.doc
def processContigRange(self, ema_range):
""" Process a contigous range of measures give an MEI doc and an
EmaExpression.EmaMeasureRange object """
# get all the spanners for total extension of meaure selection
# (including gap measures, if present)
# NB: Doing this for every range may be inefficient for larger files
spanners = self.getMultiMeasureSpanners(ema_range.measures[0].idx - 1,
ema_range.measures[-1].idx - 1)
# Let's start with measures
for i, ema_m in enumerate(ema_range.measures):
is_first_m = i == 0
is_last_m = i == len(ema_range.measures) - 1
# Get requested measure
measure = self.measures[ema_m.idx - 1]
events = measure.getChildren()
# determine current measure beat info
meter = None
for change in self.timeChanges:
if int(change) + 1 <= ema_m.idx:
meter = self.beatsInfo[change]
# Get list of staff numbers in current measure
sds = [
int(sd.getAttribute("n").getValue())
for sd in measure.getClosestStaffDefs()
]
# Set aside selected staff numbers
s_nos = []
# Proceed to locate requested staves
for ema_s in ema_m.staves:
if ema_s.number not in sds:
# CAREFUL: there may be issues with "silent" staves
# that may be defined by missing from current measure.
# TODO: Write test, fix.
raise BadApiRequest("Requested staff is not defined")
s_nos.append(ema_s.number)
for s_i, staff in enumerate(measure.getChildrenByName("staff")):
s_no = s_i
if staff.hasAttribute("n"):
s_no = int(staff.getAttribute("n").getValue())
if s_no in s_nos:
ema_s = ema_m.staves[s_nos.index(s_no)]
# Get other elements affecting the staff, e.g. slurs
around = []
for el in list(events):
if self._isInStaff(el, s_no):
around.append(el)
# Create sets of elements marked for selection, removal and
# cutting. Elements are not removed or cut immediately to make
# sure that beat calcualtions are accurate.
marked_as_selected = MeiElementSet()
marked_as_space = MeiElementSet()
marked_for_removal = MeiElementSet()
marked_for_cutting = MeiElementSet()
# Now locate the requested beat ranges within the staff
for b_i, ema_beat_range in enumerate(ema_s.beat_ranges):
is_first_b = i == 0
is_last_b = i == len(ema_s.beat_ranges) - 1
def _remove(el):
""" Determine whether a removed element needs to
be converted to a space or removed altogether"""
if is_last_b and is_last_m:
marked_for_removal.add(el)
else:
marked_as_space.add(el)
# shorten them names
tstamp_first = ema_beat_range.tstamp_first
tstamp_final = ema_beat_range.tstamp_final
co = self.ema_exp.completenessOptions
# check that the requested beats actually fit in the meter
if tstamp_first > int(meter["count"])+1 or \
tstamp_final > int(meter["count"])+1:
raise BadApiRequest(
"Request beat is out of measure bounds")
# Find all descendants with att.duration.musical (@dur)
for layer in staff.getDescendantsByName("layer"):
cur_beat = 1.0
is_first_match = True
for el in layer.getDescendants():
if el.hasAttribute(
"dur"
) and not el.hasAttribute("grace"):
dur = self._calculateDur(el, meter)
# TODO still problems with non-consecutive beat ranges
# e.g. @1@3
# exclude descendants at and in between tstamps
if cur_beat >= tstamp_first:
# We round to 4 decimal places to avoid issues caused by
# tuplet-related calculations, which are admittedly not
# well expressed in floating numbers.
if round(cur_beat, 4) <= round(
tstamp_final, 4):
marked_as_selected.add(el)
if is_first_match and "cut" in co:
marked_for_cutting.add(el)
is_first_match = False
# discard from removal set if it had
# been placed there from other beat
# range
marked_as_space.discard(el)
# Cut the duration of the last element
# if completeness = cut
needs_cut = cur_beat + dur > tstamp_final + 1
if needs_cut and "cut" in co:
marked_for_cutting.add(el)
is_first_match = False
elif not marked_as_selected.get(el):
_remove(el)
elif not marked_as_selected.get(el):
marked_as_space.add(el)
# continue
cur_beat += dur
# select elements affecting the staff occurring
# within beat range
for event in around:
if not marked_as_selected.get(event):
if event.hasAttribute("tstamp"):
ts = float(
event.getAttribute(
"tstamp").getValue())
if ts < 1 and "cut" not in self.ema_exp.completenessOptions:
ts = 1
ts2_att = None
if event.hasAttribute("tstamp2"):
ts2_att = event.getAttribute("tstamp2")
if ts > tstamp_final or (
not ts2_att and ts < tstamp_first):
marked_for_removal.add(event)
elif ts2_att:
ts2 = ts2_att.getValue()
if "+" not in ts2:
if ts2 < tstamp_first:
marked_for_removal.add(event)
elif ts2 == tstamp_final:
marked_as_selected.add(event)
marked_for_removal.discard(
event)
if ts < tstamp_first and ts2 >= tstamp_final:
marked_as_selected.add(event)
marked_for_removal.discard(
event)
else:
marked_for_removal.add(event)
else:
marked_as_selected.add(event)
else:
marked_as_selected.add(event)
elif event.hasAttribute("startid"):
startid = (event.getAttribute(
"startid").getValue().replace("#", ""))
target = self.doc.getElementById(startid)
if not target:
msg = """Unsupported Encoding: attribute
startid on element {0} does not point to any
element in the document.""".format(
event.getName())
raise UnsupportedEncoding(
re.sub(r'\s+', ' ', msg.strip()))
# Make sure the target event is in the same measure
event_m = event.getAncestor(
"measure").getId()
target_m = target.getAncestor(
"measure").getId()
if not event_m == target_m:
msg = """Unsupported Encoding: attribute
startid on element {0} does not point to an
element in the same measure.""".format(
event.getName())
raise UnsupportedEncoding(
re.sub(r'\s+', ' ', msg.strip()))
else:
if marked_as_selected.get(target):
marked_as_selected.add(event)
marked_for_removal.discard(event)
elif not event.hasAttribute("endid"):
marked_for_removal.add(event)
else:
# Skip if event starts after latest
# selected element with duration
pos = target.getPositionInDocument(
)
is_ahead = False
for i in reversed(
marked_as_selected.
getElements()):
if i.hasAttribute("dur"):
if pos > i.getPositionInDocument(
):
marked_for_removal.add(
event)
is_ahead = True
break
if not is_ahead:
# last chance to keep it:
# must start before and end after
# latest selected element with duration
endid = (
event.getAttribute("endid")
.getValue().replace(
"#", ""))
target2 = self.doc.getElementById(
endid)
if marked_as_selected.get(
target2):
marked_as_selected.add(
event)
marked_for_removal.discard(
event)
else:
pos2 = target2.getPositionInDocument(
)
for i in reversed(
marked_as_selected.
getElements()):
if i.hasAttribute(
"dur"):
if pos2 > i.getPositionInDocument(
):
marked_as_selected.add(
event)
marked_for_removal.discard(
event)
else:
marked_for_removal.add(
event)
break
# Remove elements marked for removal
for el in marked_for_removal:
el.getParent().removeChild(el)
# Replace elements marked as spaces with actual spaces,
# unless completion = nospace, then remove the elements.
for el in marked_as_space:
parent = el.getParent()
if "nospace" not in self.ema_exp.completenessOptions:
space = MeiElement("space")
space.setId(el.id)
space.addAttribute(el.getAttribute("dur"))
if el.getAttribute("dots"):
space.addAttribute(el.getAttribute("dots"))
elif el.getChildrenByName("dot"):
dots = str(len(el.getChildrenByName("dot")))
space.addAttribute(MeiAttribute("dots", dots))
parent.addChildBefore(el, space)
el.getParent().removeChild(el)
else:
# Remove this staff and its attached events
staff.getParent().removeChild(staff)
for el in list(events):
if self._isInStaff(el, s_no):
el.getParent().removeChild(el)
# At the first measure, also add relevant multi-measure spanners
# for each selected staff
if is_first_m:
for evs in spanners.values():
for event_id in evs:
ev = self.doc.getElementById(event_id)
# Spanners starting outside of beat ranges
# may be already gone
if ev:
# Determine staff of event for id changes
for ema_s in ema_m.staves:
staff_no = self._isInStaff(ev, ema_s.number)
if staff_no and staff_no in s_nos:
# If the event is attached to more than one staff, just
# consider it attached to the its first one
staff_no = staff_no[0]
staff = None
for staff_candidate in measure.getDescendantsByName(
"staff"):
if staff_candidate.hasAttribute("n"):
n = int(
staff_candidate.getAttribute(
"n").getValue())
if n == staff_no:
staff = staff_candidate
# Truncate event to start at the beginning of the beat range
if ev.hasAttribute("startid"):
# Set startid to the first event still on staff,
# at the first available layer
try:
layer = (
staff.getChildrenByName("layer"))
first_id = layer[0].getChildren(
)[0].getId()
ev.getAttribute("startid").setValue(
"#" + first_id)
except IndexError:
msg = """
Unsupported encoding. Omas attempted to adjust the
starting point of a selected multi-measure element
that starts before the selection, but the staff or
layer could not be located.
"""
msg = re.sub(r'\s+', ' ', msg.strip())
raise UnsupportedEncoding(msg)
if ev.hasAttribute("tstamp"):
# Set tstamp to first in beat selection
tstamp_first = 0
for e_s in ema_m.staves:
if e_s.number == staff_no:
tstamp_first = e_s.beat_ranges[
0].tstamp_first
ev.getAttribute("tstamp").setValue(
str(tstamp_first))
# Truncate to end of range if completeness = cut
# (actual beat cutting will happen when beat ranges are procesed)
if "cut" in self.ema_exp.completenessOptions:
if ev.hasAttribute("tstamp2"):
att = ev.getAttribute("tstamp2")
t2 = att.getValue()
p = re.compile(r"([1-9]+)(?=m\+)")
multimeasure = p.match(t2)
if multimeasure:
new_val = len(mm) - 1
att.setValue(
p.sub(str(new_val), t2))
# Otherwise adjust tspan2 value to correct distance.
# E.g. given 4 measures with a spanner originating
# in 1 and ending in 4 and a selection of measures 2 and 3,
# change @tspan2 from 3m+X to 2m+X
else:
if ev.hasAttribute("tstamp2"):
att = ev.getAttribute("tstamp2")
t2 = att.getValue()
p = re.compile(r"([1-9]+)(?=m\+)")
multimeasure = p.match(t2)
if multimeasure:
dis = evs[event_id]["distance"]
new_val = int(
multimeasure.group(1)) - dis
att.setValue(
p.sub(str(new_val), t2))
# move element to first measure and add it to selected
# events "around" the staff.
ev.moveTo(measure)
return self.doc
def getMultiMeasureSpanners(self, start, end=-1):
"""Return a dictionary of spanning elements encompassing,
landing or starting within selected measures"""
mm = self.musicEl.getDescendantsByName("measure")
try:
start_m = mm[start]
except IndexError:
raise BadApiRequest("Requested measure does not exist.")
table = {}
# Template of table dictionary (for reference):
# {
# "_targetMeasureID_" : {
# "_eventID_" : {
# "origin" : "_originMeasureID_",
# "distance" : 0,
# "startid" : "_startID_",
# "endid" : "_endID_",
# "tstamp" : "_beat_",
# "tstamp2" : "_Xm+beat_"
# }
# }
# }
def _calculateDistance(origin):
""" Calcualte distance of origin measure from
first measure in selection """
# Cast MeiElementList to python list
list_mm = list(mm)
return list_mm.index(start_m) - list_mm.index(origin)
# Exclude end measure index from request,
# unless last measure is requested (creates table for whole file).
if end == -1:
end = len(mm) + 1
# Look back through measures and build a table of events
# spanning multiple measures via tstamp/tstamp2 or startid/endid pairs.
for i, prev_m in enumerate(mm[:end]):
m_id = prev_m.getId()
for el in prev_m.getDescendants():
# Determine destination based on startid/endid pairs.
# If not found, look for tstamp/tstamp2 pairs before moving on.
if el.hasAttribute("endid"):
endid = el.getAttribute("endid").getValue()
endid = endid.replace("#", "")
target_el = self.doc.getElementById(endid)
if target_el:
if target_el.hasAncestor("measure"):
# This could be a comparison of objects,
# but comparing ids feels safer.
destination = target_el.getAncestor("measure")
if destination.getId() != prev_m.getId():
# Create table entry
dest_id = destination.getId()
distance = _calculateDistance(prev_m)
el_id = el.getId()
if dest_id not in table:
table[dest_id] = {}
table[dest_id][el_id] = {
"origin": m_id,
"distance": distance,
"endid": endid
}
if el.hasAttribute("startid"):
startid = (el.getAttribute(
"startid").getValue().replace("#", ""))
table[dest_id][el_id]["startid"] = startid
elif el.hasAttribute("tstamp2"):
t2 = el.getAttribute("tstamp2").getValue()
multiMesSpan = re.match(r"([1-9]+)m\+", t2)
if multiMesSpan:
destination = mm[i + int(multiMesSpan.group(1))]
# Create table entry
dest_id = destination.getId()
distance = _calculateDistance(prev_m)
el_id = el.getId()
if dest_id not in table:
table[dest_id] = {}
table[dest_id][el_id] = {
"origin": m_id,
"distance": distance,
"tstamp2": t2
}
if el.hasAttribute("tstamp"):
table[dest_id][el_id]["tstamp"] = (
el.getAttribute("tstamp").getValue())
return table
def _calculateDur(self, element, meter):
""" Determine the duration of an element given a meter """
# TODO: beware of @duration.default - though not very common
dur_val = element.getAttribute("dur").getValue()
if dur_val == "breve":
dur_val = "0.5"
elif dur_val == "long":
dur_val = "0.25"
duration = float(dur_val)
relative_dur = float(int(meter["unit"]) / duration)
dots = 0
if element.getAttribute("dots"):
dots = int(element.getAttribute("dots").getValue())
elif element.getChildrenByName("dot"):
dots = len(element.getChildrenByName("dot"))
dot_dur = duration
for d in range(1, int(dots) + 1):
dot_dur = dot_dur * 2
relative_dur += float(int(meter["unit"]) / dot_dur)
# Is this element contained in a tuplet element?
# (TODO account for tupletspan)
if element.hasAncestor("tuplet"):
tupl = element.getAncestor("tuplet")
numbase = tupl.getAttribute("numbase")
num = tupl.getAttribute("num")
if not num or not numbase:
raise UnsupportedEncoding(
"Cannot understand tuplet beat: both @num and @numbase must be present"
)
else:
tupl_ratio = float(numbase.getValue()) / float(num.getValue())
relative_dur = relative_dur * tupl_ratio
return relative_dur
def _cutDuration(self, element, meter):
""" Cut the duration of given element to the final beat """
element.getAttribute("dur").setValue(str(meter["unit"]))
# Remove dots if any
element.removeAttribute("dots")
element.removeChildrenByName("dot")
def _isInStaff(self, el, s_no):
""" Get all staff numbers of element if element is in given staff"""
values = self._getSelectedStaffNosFor(el)
if s_no in values:
return values
else:
values = []
return values
def _getSelectedStaffNosFor(self, el):
""" Get staff numbers of element if the staves in given measure
are selected"""
# TODO: CAREFUL - EDITORIAL MARKUP MAY OBFUSCATE THIS
values = []
if el.hasAttribute("staff"):
# Split value of @staff, as it may contain multiple values.
values = el.getAttribute("staff").getValue().split()
values = [int(x) for x in values]
return values
class MeiSlicer(object):
"""Class for slicing an MEI doc given a split EMA expresion. """
def __init__(self, doc, req_m, req_s, req_b, completeness=None):
self.doc = doc
self.flat_doc = doc.getFlattenedTree()
self.flat_doc_static = list(doc.getFlattenedTree())
self.docInfo = MusDocInfo(doc).get()
self.musicEl = doc.getElementsByName("music")[0]
self.measures = self.musicEl.getDescendantsByName("measure")
self.ema_exp = EmaExpression(self.docInfo,
req_m,
req_s,
req_b,
completeness)
self.ema_measures = self.ema_exp.get()
self.compiled_exp = self.ema_exp.getCompiled()
def slice(self):
""" Return a modified MEI doc containing the selected notation
provided a EMA expression of measures, staves, and beats."""
# parse general beats information
self.beatsInfo = self.docInfo["beats"]
self.timeChanges = self.beatsInfo.keys()
self.timeChanges.sort(key=int)
# Process measure ranges and store boundary measures
boundary_mm = []
for em in self.ema_measures:
boundary_mm.append(em.measures[0])
boundary_mm.append(em.measures[-1])
self.processContigRange(em)
# Recursively remove remaining data in between
# measure ranges before returning modified doc
# First remove measures in between
to_remove = []
middle_boundaries = boundary_mm[1:-1]
for bm in middle_boundaries[::2]:
i = middle_boundaries.index(bm)
try:
start_m = self.measures[bm.idx-1]
start_m_pos = start_m.getPositionInDocument()
end_m = self.measures[middle_boundaries[i+1].idx-1]
end_m_pos = end_m.getPositionInDocument()
for el in self.flat_doc[start_m_pos+1:end_m_pos]:
if el.hasAncestor("measure"):
if el.getAncestor("measure").getId() != start_m.getId():
to_remove.append(el)
else:
to_remove.append(el)
except IndexError:
pass
m_first = self.measures[boundary_mm[0].idx-1]
m_final = self.measures[boundary_mm[-1].idx-1]
# List of elements to keep
keep = ["meiHead"]
def _removeBefore(curEl):
parent = curEl.getParent()
if parent:
# Casting to list to avoid modfying the sequence
for el in list(curEl.getPeers()):
if el == curEl:
break
else:
if not el.getName() in keep:
parent.removeChild(el)
return _removeBefore(parent)
return curEl
def _removeAfter(curEl):
parent = curEl.getParent()
if parent:
removing = False
# Casting to list to avoid modfying the sequence
for el in list(curEl.getPeers()):
if removing:
if not el.getName() in keep:
parent.removeChild(el)
elif el == curEl:
removing = True
return _removeAfter(parent)
return curEl
def _findLowestCommonAncestor(el1, el2):
while not el1 == el2:
el1 = el1.getParent()
el2 = el2.getParent()
return el1
def _removeUnusedStaffDefs(scoredef, start_ema_m):
numbers = set()
for em in self.ema_measures:
if start_ema_m == em.measures[0]:
for m in em.measures:
for ema_s in m.staves:
numbers.add(ema_s.number)
break
for staffd in scoredef.getDescendantsByName("staffDef"):
if staffd.hasAttribute("n"):
val = staffd.getAttribute("n").getValue()
if int(val) not in numbers:
staffd.getParent().removeChild(staffd)
# TODO! Remove definitions of unselected staves WITHIN RANGE
# Compute closest score definition to start measure
preceding = self.flat_doc_static[:m_first.getPositionInDocument()]
first_scoreDef = None
for el in reversed(preceding):
if el.getName() == "scoreDef":
first_scoreDef = el
break
# Recursively remove elements before and after selected measures
_removeBefore(m_first)
_removeAfter(m_final)
# Compute closest score definition to start measure of each range
b_scoreDef = first_scoreDef
for i, bm in enumerate(boundary_mm[::2]): # list comprehension get only start mm
b_measure = self.measures[bm.idx-1]
preceding = self.flat_doc_static[:b_measure.getPositionInDocument()]
for el in reversed(preceding):
if el.getName() == "scoreDef":
try:
s_id = b_scoreDef.getId()
if not s_id == el.getId():
b_scoreDef = el
# Re-attach computed score definition
sd_copy = MeiElement(b_scoreDef)
_removeUnusedStaffDefs(sd_copy, bm)
b_measure.getParent().addChildBefore(b_measure, sd_copy)
except AttributeError:
b_scoreDef = el
break
for el in to_remove:
el.getParent().removeChild(el)
if "raw" in self.ema_exp.completenessOptions:
lca = _findLowestCommonAncestor(m_first, m_final)
self.doc.setRootElement(lca)
# re-attach computed score definition if required by
# completeness = signature
if "signature" in self.ema_exp.completenessOptions:
m_first.getParent().addChildBefore(m_first, first_scoreDef)
return self.doc
def processContigRange(self, ema_range):
""" Process a contigous range of measures give an MEI doc and an
EmaExpression.EmaMeasureRange object """
# get all the spanners for total extension of meaure selection
# (including gap measures, if present)
# NB: Doing this for every range may be inefficient for larger files
spanners = self.getMultiMeasureSpanners(ema_range.measures[0].idx-1,
ema_range.measures[-1].idx-1)
# Let's start with measures
for i, ema_m in enumerate(ema_range.measures):
is_first_m = i == 0
is_last_m = i == len(ema_range.measures)-1
# Get requested measure
measure = self.measures[ema_m.idx-1]
events = measure.getChildren()
# determine current measure beat info
meter = None
for change in self.timeChanges:
if int(change)+1 <= ema_m.idx:
meter = self.beatsInfo[change]
# Get list of staff numbers in current measure
sds = [int(sd.getAttribute("n").getValue())
for sd in measure.getClosestStaffDefs()]
# Set aside selected staff numbers
s_nos = []
# Proceed to locate requested staves
for ema_s in ema_m.staves:
if ema_s.number not in sds:
# CAREFUL: there may be issues with "silent" staves
# that may be defined by missing from current measure.
# TODO: Write test, fix.
raise BadApiRequest("Requested staff is not defined")
s_nos.append(ema_s.number)
for s_i, staff in enumerate(measure.getChildrenByName("staff")):
s_no = s_i
if staff.hasAttribute("n"):
s_no = int(staff.getAttribute("n").getValue())
if s_no in s_nos:
ema_s = ema_m.staves[s_nos.index(s_no)]
# Get other elements affecting the staff, e.g. slurs
around = []
for el in list(events):
if self._isInStaff(el, s_no):
around.append(el)
# Create sets of elements marked for selection, removal and
# cutting. Elements are not removed or cut immediately to make
# sure that beat calcualtions are accurate.
marked_as_selected = MeiElementSet()
marked_as_space = MeiElementSet()
marked_for_removal = MeiElementSet()
marked_for_cutting = MeiElementSet()
# Now locate the requested beat ranges within the staff
for b_i, ema_beat_range in enumerate(ema_s.beat_ranges):
is_first_b = i == 0
is_last_b = i == len(ema_s.beat_ranges)-1
def _remove(el):
""" Determine whether a removed element needs to
be converted to a space or removed altogether"""
if is_last_b and is_last_m:
marked_for_removal.add(el)
else:
marked_as_space.add(el)
# shorten them names
tstamp_first = ema_beat_range.tstamp_first
tstamp_final = ema_beat_range.tstamp_final
co = self.ema_exp.completenessOptions
# check that the requested beats actually fit in the meter
if tstamp_first > int(meter["count"])+1 or \
tstamp_final > int(meter["count"])+1:
raise BadApiRequest(
"Request beat is out of measure bounds")
# Find all descendants with att.duration.musical (@dur)
for layer in staff.getDescendantsByName("layer"):
cur_beat = 1.0
is_first_match = True
for el in layer.getDescendants():
if el.hasAttribute("dur") and not el.hasAttribute("grace"):
dur = self._calculateDur(el, meter)
# TODO still problems with non-consecutive beat ranges
# e.g. @1@3
# exclude descendants at and in between tstamps
if cur_beat >= tstamp_first:
# We round to 4 decimal places to avoid issues caused by
# tuplet-related calculations, which are admittedly not
# well expressed in floating numbers.
if round(cur_beat, 4) <= round(tstamp_final, 4):
marked_as_selected.add(el)
if is_first_match and "cut" in co:
marked_for_cutting.add(el)
is_first_match = False
# discard from removal set if it had
# been placed there from other beat
# range
marked_as_space.discard(el)
# Cut the duration of the last element
# if completeness = cut
needs_cut = cur_beat+dur > tstamp_final+1
if needs_cut and "cut" in co:
marked_for_cutting.add(el)
is_first_match = False
elif not marked_as_selected.get(el):
_remove(el)
elif not marked_as_selected.get(el):
marked_as_space.add(el)
# continue
cur_beat += dur
# select elements affecting the staff occurring
# within beat range
for event in around:
if not marked_as_selected.get(event):
if event.hasAttribute("tstamp"):
ts = float(event.getAttribute("tstamp").getValue())
if ts < 1 and "cut" not in self.ema_exp.completenessOptions:
ts = 1
ts2_att = None
if event.hasAttribute("tstamp2"):
ts2_att = event.getAttribute("tstamp2")
if ts > tstamp_final or (not ts2_att and ts < tstamp_first):
marked_for_removal.add(event)
elif ts2_att:
ts2 = ts2_att.getValue()
if "+" not in ts2:
if ts2 < tstamp_first:
marked_for_removal.add(event)
elif ts2 == tstamp_final:
marked_as_selected.add(event)
marked_for_removal.discard(event)
if ts < tstamp_first and ts2 >= tstamp_final:
marked_as_selected.add(event)
marked_for_removal.discard(event)
else:
marked_for_removal.add(event)
else:
marked_as_selected.add(event)
else:
marked_as_selected.add(event)
elif event.hasAttribute("startid"):
startid = (
event.getAttribute("startid")
.getValue()
.replace("#", "")
)
target = self.doc.getElementById(startid)
if not target:
msg = """Unsupported Encoding: attribute
startid on element {0} does not point to any
element in the document.""".format(
event.getName())
raise UnsupportedEncoding(
re.sub(r'\s+', ' ', msg.strip()))
# Make sure the target event is in the same measure
event_m = event.getAncestor("measure").getId()
target_m = target.getAncestor("measure").getId()
if not event_m == target_m:
msg = """Unsupported Encoding: attribute
startid on element {0} does not point to an
element in the same measure.""".format(
event.getName())
raise UnsupportedEncoding(
re.sub(r'\s+', ' ', msg.strip()))
else:
if marked_as_selected.get(target):
marked_as_selected.add(event)
marked_for_removal.discard(event)
elif not event.hasAttribute("endid"):
marked_for_removal.add(event)
else:
# Skip if event starts after latest
# selected element with duration
pos = target.getPositionInDocument()
is_ahead = False
for i in reversed(marked_as_selected.getElements()):
if i.hasAttribute("dur"):
if pos > i.getPositionInDocument():
marked_for_removal.add(event)
is_ahead = True
break
if not is_ahead:
# last chance to keep it:
# must start before and end after
# latest selected element with duration
endid = (
event.getAttribute("endid")
.getValue()
.replace("#", "")
)
target2 = self.doc.getElementById(endid)
if marked_as_selected.get(target2):
marked_as_selected.add(event)
marked_for_removal.discard(event)
else:
pos2 = target2.getPositionInDocument()
for i in reversed(marked_as_selected.getElements()):
if i.hasAttribute("dur"):
if pos2 > i.getPositionInDocument():
marked_as_selected.add(event)
marked_for_removal.discard(event)
else:
marked_for_removal.add(event)
break
# Remove elements marked for removal
for el in marked_for_removal:
el.getParent().removeChild(el)
# Replace elements marked as spaces with actual spaces,
# unless completion = nospace, then remove the elements.
for el in marked_as_space:
parent = el.getParent()
if "nospace" not in self.ema_exp.completenessOptions:
space = MeiElement("space")
space.setId(el.id)
space.addAttribute(el.getAttribute("dur"))
if el.getAttribute("dots"):
space.addAttribute(el.getAttribute("dots"))
elif el.getChildrenByName("dot"):
dots = str(len(el.getChildrenByName("dot")))
space.addAttribute(MeiAttribute("dots", dots))
parent.addChildBefore(el, space)
el.getParent().removeChild(el)
else:
# Remove this staff and its attached events
staff.getParent().removeChild(staff)
for el in list(events):
if self._isInStaff(el, s_no):
el.getParent().removeChild(el)
# At the first measure, also add relevant multi-measure spanners
# for each selected staff
if is_first_m:
for evs in spanners.values():
for event_id in evs:
ev = self.doc.getElementById(event_id)
# Spanners starting outside of beat ranges
# may be already gone
if ev:
# Determine staff of event for id changes
for ema_s in ema_m.staves:
staff_no = self._isInStaff(ev, ema_s.number)
if staff_no and staff_no in s_nos:
# If the event is attached to more than one staff, just
# consider it attached to the its first one
staff_no = staff_no[0]
staff = None
for staff_candidate in measure.getDescendantsByName("staff"):
if staff_candidate.hasAttribute("n"):
n = int(staff_candidate.getAttribute("n").getValue())
if n == staff_no:
staff = staff_candidate
# Truncate event to start at the beginning of the beat range
if ev.hasAttribute("startid"):
# Set startid to the first event still on staff,
# at the first available layer
try:
layer = (
staff.getChildrenByName("layer")
)
first_id = layer[0].getChildren()[0].getId()
ev.getAttribute("startid").setValue("#"+first_id)
except IndexError:
msg = """
Unsupported encoding. Omas attempted to adjust the
starting point of a selected multi-measure element
that starts before the selection, but the staff or
layer could not be located.
"""
msg = re.sub(r'\s+', ' ', msg.strip())
raise UnsupportedEncoding(msg)
if ev.hasAttribute("tstamp"):
# Set tstamp to first in beat selection
tstamp_first = 0
for e_s in ema_m.staves:
if e_s.number == staff_no:
tstamp_first = e_s.beat_ranges[0].tstamp_first
ev.getAttribute("tstamp").setValue(str(tstamp_first))
# Truncate to end of range if completeness = cut
# (actual beat cutting will happen when beat ranges are procesed)
if "cut" in self.ema_exp.completenessOptions:
if ev.hasAttribute("tstamp2"):
att = ev.getAttribute("tstamp2")
t2 = att.getValue()
p = re.compile(r"([1-9]+)(?=m\+)")
multimeasure = p.match(t2)
if multimeasure:
new_val = len(mm) - 1
att.setValue(p.sub(str(new_val), t2))
# Otherwise adjust tspan2 value to correct distance.
# E.g. given 4 measures with a spanner originating
# in 1 and ending in 4 and a selection of measures 2 and 3,
# change @tspan2 from 3m+X to 2m+X
else:
if ev.hasAttribute("tstamp2"):
att = ev.getAttribute("tstamp2")
t2 = att.getValue()
p = re.compile(r"([1-9]+)(?=m\+)")
multimeasure = p.match(t2)
if multimeasure:
dis = evs[event_id]["distance"]
new_val = int(multimeasure.group(1)) - dis
att.setValue(p.sub(str(new_val), t2))
# move element to first measure and add it to selected
# events "around" the staff.
ev.moveTo(measure)
return self.doc
def getMultiMeasureSpanners(self, start, end=-1):
"""Return a dictionary of spanning elements encompassing,
landing or starting within selected measures"""
mm = self.musicEl.getDescendantsByName("measure")
try:
start_m = mm[start]
except IndexError:
raise BadApiRequest("Requested measure does not exist.")
table = {}
# Template of table dictionary (for reference):
# {
# "_targetMeasureID_" : {
# "_eventID_" : {
# "origin" : "_originMeasureID_",
# "distance" : 0,
# "startid" : "_startID_",
# "endid" : "_endID_",
# "tstamp" : "_beat_",
# "tstamp2" : "_Xm+beat_"
# }
# }
# }
def _calculateDistance(origin):
""" Calcualte distance of origin measure from
first measure in selection """
# Cast MeiElementList to python list
list_mm = list(mm)
return list_mm.index(start_m) - list_mm.index(origin)
# Exclude end measure index from request,
# unless last measure is requested (creates table for whole file).
if end == -1:
end = len(mm) + 1
# Look back through measures and build a table of events
# spanning multiple measures via tstamp/tstamp2 or startid/endid pairs.
for i, prev_m in enumerate(mm[:end]):
m_id = prev_m.getId()
for el in prev_m.getDescendants():
# Determine destination based on startid/endid pairs.
# If not found, look for tstamp/tstamp2 pairs before moving on.
if el.hasAttribute("endid"):
endid = el.getAttribute("endid").getValue()
endid = endid.replace("#", "")
target_el = self.doc.getElementById(endid)
if target_el:
if target_el.hasAncestor("measure"):
# This could be a comparison of objects,
# but comparing ids feels safer.
destination = target_el.getAncestor("measure")
if destination.getId() != prev_m.getId():
# Create table entry
dest_id = destination.getId()
distance = _calculateDistance(prev_m)
el_id = el.getId()
if dest_id not in table:
table[dest_id] = {}
table[dest_id][el_id] = {
"origin": m_id,
"distance": distance,
"endid": endid
}
if el.hasAttribute("startid"):
startid = (
el
.getAttribute("startid")
.getValue()
.replace("#", "")
)
table[dest_id][el_id]["startid"] = startid
elif el.hasAttribute("tstamp2"):
t2 = el.getAttribute("tstamp2").getValue()
multiMesSpan = re.match(r"([1-9]+)m\+", t2)
if multiMesSpan:
destination = mm[i + int(multiMesSpan.group(1))]
# Create table entry
dest_id = destination.getId()
distance = _calculateDistance(prev_m)
el_id = el.getId()
if dest_id not in table:
table[dest_id] = {}
table[dest_id][el_id] = {
"origin": m_id,
"distance": distance,
"tstamp2": t2
}
if el.hasAttribute("tstamp"):
table[dest_id][el_id]["tstamp"] = (
el
.getAttribute("tstamp")
.getValue()
)
return table
def _calculateDur(self, element, meter):
""" Determine the duration of an element given a meter """
# TODO: beware of @duration.default - though not very common
dur_val = element.getAttribute("dur").getValue()
if dur_val == "breve":
dur_val = "0.5"
elif dur_val == "long":
dur_val = "0.25"
duration = float(dur_val)
relative_dur = float(int(meter["unit"]) / duration)
dots = 0
if element.getAttribute("dots"):
dots = int(element.getAttribute("dots").getValue())
elif element.getChildrenByName("dot"):
dots = len(element.getChildrenByName("dot"))
dot_dur = duration
for d in range(1, int(dots)+1):
dot_dur = dot_dur * 2
relative_dur += float(int(meter["unit"]) / dot_dur)
# Is this element contained in a tuplet element?
# (TODO account for tupletspan)
if element.hasAncestor("tuplet"):
tupl = element.getAncestor("tuplet")
numbase = tupl.getAttribute("numbase")
num = tupl.getAttribute("num")
if not num or not numbase:
raise UnsupportedEncoding(
"Cannot understand tuplet beat: both @num and @numbase must be present")
else:
tupl_ratio = float(numbase.getValue()) / float(num.getValue())
relative_dur = relative_dur * tupl_ratio
return relative_dur
def _cutDuration(self, element, meter):
""" Cut the duration of given element to the final beat """
element.getAttribute("dur").setValue(str(meter["unit"]))
# Remove dots if any
element.removeAttribute("dots")
element.removeChildrenByName("dot")
def _isInStaff(self, el, s_no):
""" Get all staff numbers of element if element is in given staff"""
values = self._getSelectedStaffNosFor(el)
if s_no in values:
return values
else:
values = []
return values
def _getSelectedStaffNosFor(self, el):
""" Get staff numbers of element if the staves in given measure
are selected"""
# TODO: CAREFUL - EDITORIAL MARKUP MAY OBFUSCATE THIS
values = []
if el.hasAttribute("staff"):
# Split value of @staff, as it may contain multiple values.
values = el.getAttribute("staff").getValue().split()
values = [int(x) for x in values]
return values
