remove_unused_sedml_elements.py

import libsedml
import os


def hasTime(plot):
    doc = plot.getSedDocument()
    for curve in plot.getListOfCurves():
        xdg = doc.getDataGenerator(curve.getXDataReference())
        var = xdg.getVariable(0)
        if var.isSetSymbol():
            if var.getSymbol() == "urn:sedml:symbol:time":
                return True
    return False


def checkRepeats(plot):
    doc = plot.getSedDocument()
    for curve in plot.getListOfCurves():
        ydg = doc.getDataGenerator(curve.getYDataReference())
        var = ydg.getVariable(0)
        if var.isSetTaskReference():
            task = doc.getTask(var.getTaskReference())
            if isinstance(task, libsedml.SedRepeatedTask):
                range = task.getRange(0)
                if isinstance(range, libsedml.SedUniformRange):
                    return (True, range.getNumberOfPoints())
                if isinstance(range, libsedml.SedVectorRange):
                    return (True, range.getNumValues())
                raise("Unimplemented Range type:\n" + range.toSed())
    return (False, 1)


def choose_between(plotlist):
    unlikely = []
    possible = []
    good = []
    for plot in plotlist:
        if not isinstance(plot, libsedml.SedPlot2D):
            raise("Unable to parse surface plots.")
        hastime = hasTime(plot)
        isrepeat, reps = checkRepeats(plot)
        if hastime:
            if isrepeat:
                if reps <= 30:
                    good.append(plot)
                else:
                    unlikely.append(plot)
            else:
                possible.append(plot)
        else:
            if isrepeat:
                good.append(plot)
            else:
                possible.append(plot)
        # print(plot.toSed())
    if len(good) > 0:
        # If there was a better option, don't use the worse one
        unlikely.extend(possible)
        return unlikely
    if len(possible) > 0:
        return unlikely
    return []


def remove_duplicate_plots(doc):
    """ Remove any duplicate plots

    Args:
        doc (:obj:`libsedml.SedMLDocument`): Document
    """

    changed = False

    # Collect plot names
    names = {}
    for plot in doc.getListOfOutputs():
        if isinstance(plot, libsedml.SedPlot):
            if plot.isSetName():
                name = plot.getName()
                if name not in names:
                    names[name] = []
                names[name].append(plot)

    for name in names:
        if len(names[name]) > 1:
            for plot in choose_between(names[name]):
                doc.removeOutput(plot.getId())
                changed = True

    # write corrected SED-ML
    return changed


def remove_unused_datagens(doc):
    """ Remove any duplicate data generators

    Args:
        doc (:obj:`libsedml.SedMLDocument`): Document
    """

    changed = False

    # Collect referenced data generators
    used_datagens = set()
    for output in doc.getListOfOutputs():
        if isinstance(output, libsedml.SedReport):
            for dataset in output.getListOfDataSets():
                used_datagens.add(dataset.getDataReference())
        elif isinstance(output, libsedml.SedPlot2D):
            for curve in output.getListOfCurves():
                used_datagens.add(curve.getXDataReference())
                used_datagens.add(curve.getYDataReference())
        elif isinstance(output, libsedml.SedSurface):
            for surface in output.getListOfSurfaces():
                used_datagens.add(surface.getXDataReference())
                used_datagens.add(surface.getYDataReference())
                used_datagens.add(surface.getZDataReference())

    unused_datagens = []
    dgs = doc.getListOfDataGenerators()
    for datagen in dgs:
        dgid = datagen.getId()
        if dgid not in used_datagens:
            unused_datagens.append(dgid)

    for unused_datagen in unused_datagens:
        dgs.remove(unused_datagen)
        changed = True

    # write corrected SED-ML
    return changed


def remove_unused_tasks(doc):
    changed = False

    # get tasks used by data generators
    used_tasks = set()
    for datagen in doc.getListOfDataGenerators():
        for variable in datagen.getListOfVariables():
            if variable.isSetTaskReference():
                used_tasks.add(variable.getTaskReference())

    # add used subtasks
    used_parent_tasks = list(used_tasks)
    while used_parent_tasks:
        parent_task = doc.getTask(used_parent_tasks.pop())
        if isinstance(parent_task, libsedml.SedRepeatedTask):
            for subtask in parent_task.getListOfSubTasks():
                used_parent_tasks.append(subtask.getTask())
                used_tasks.add(subtask.getTask())

    # determine unused tasks
    unused_tasks = set()
    for task in doc.getListOfTasks():
        task_id = task.getId()
        if task_id not in used_tasks:
            unused_tasks.add(task_id)

    # remove unused tasks
    for task_id in unused_tasks:
        doc.removeTask(task_id)
        changed = True

    # return whether at least one task was removed
    return changed


def remove_unused_sims_and_mods(doc):
    changed = False
    used_sims = set()
    used_mods = set()
    for task in doc.getListOfTasks():
        try:
            used_mods.add(task.getModelReference())
            used_sims.add(task.getSimulationReference())
        except Exception:
            pass
    unused_sims = set()
    unused_mods = set()
    for mod in doc.getListOfModels():
        mid = mod.getId()
        if mid not in used_mods:
            unused_mods.add(mid)
    for sim in doc.getListOfSimulations():
        sid = sim.getId()
        if sid not in used_sims:
            unused_sims.add(sid)
    for sid in unused_sims:
        doc.removeSimulation(sid)
        changed = True
    for mid in unused_mods:
        doc.removeModel(mid)
        changed = True
    return changed


def run(id, sedml_filenames):
    removed_sedml_files = []
    for sedml_filename in sedml_filenames:
        if "Parmar2017_Deficient_Rich_tracer" in sedml_filename or "Parmar2017_Adequate_tracer" in sedml_filename:
            continue
        doc = libsedml.readSedMLFromFile(sedml_filename)
        changed = False
        changed = remove_duplicate_plots(doc) or changed
        changed = remove_unused_datagens(doc) or changed
        changed = remove_unused_tasks(doc) or changed
        changed = remove_unused_sims_and_mods(doc) or changed
        if (changed):
            # print("Modified", id, os.path.basename(sedml_filename))
            if doc.getNumModels() == 0:
                # Remove the SED-ML file entirely
                os.remove(sedml_filename)
                removed_sedml_files.append(sedml_filename)
            else:
                libsedml.writeSedMLToFile(doc, sedml_filename)
    for sedml_filename in removed_sedml_files:
        sedml_filenames.remove(sedml_filename)