示例#1
0
def test_get_model(bento_neo4j):
  (b,h)=bento_neo4j
  the_mdb = MDB(uri=b)
  assert the_mdb
  ObjectMap.clear_cache()
  m = Model(handle='ICDC',mdb=the_mdb)
  m.dget()

  with m.drv.session() as session:
    result = session.run('match (n:node) where n.model="ICDC" return count(n)')
    assert len(m.nodes) == result.single().value()
    result = session.run('match (n:relationship) where n.model="ICDC" return count(n)')
    assert len(m.edges) == result.single().value()
    result = session.run('match (p:property)<--(n:node) where p.model="ICDC" and  n.model="ICDC" return count(p)')
    assert len(m.props) == result.single().value()
    result = session.run(
      'match (s:node)<-[:has_src]-(e:relationship)-[:has_dst]->(d:node) where e.model="ICDC" return s,e,d')
    for rec in result:
      (s,e,d) = (rec['s'],rec['e'],rec['d'])
      triplet = (e['handle'], s['handle'], d['handle'])
      assert m.edges[triplet].handle == e['handle']
      assert m.edges[triplet].src.handle == s['handle']
      assert m.edges[triplet].dst.handle == d['handle']

    result = session.run(
      'match (n:node)-[:has_property]->(p:property) where (n.model="ICDC") return n, collect(p) as pp')
    for rec in result:
      for p in rec['pp']:
        key = (rec['n']['handle'], p['handle'])
        assert m.props[key]
        assert m.props[key].neoid == p.id
        assert m.nodes[rec['n']['handle']].props[p['handle']].neoid == p.id

    result = session.run(
      'match (t:term)<-[:has_term]-(v:value_set)<-[:has_value_set]-(p:property) where p.model="ICDC" return p, v, collect(t) as tt')
    for rec in result:
      (p, v, tt) = (rec['p'],rec['v'],rec['tt'])
      [op] = [ x for x in m.props.values() if x.handle == p['handle'] ]
      vs = op.value_set
      assert op
      assert set( op.values ) == { t['value'] for t in tt }
示例#2
0
def test_create_model():
    model = Model('test')
    case = Node({"handle": "case"})
    #  set_trace()
    case.props['days_to_enrollment'] = Property(
        {"handle": 'days_to_enrollment'})
    model.add_node(case)
    assert isinstance(model.nodes['case'], Node)
    assert model.props[('case', 'days_to_enrollment')]
    model.add_node({"handle": "sample"})
    assert model.nodes["sample"]
    assert isinstance(model.nodes["sample"], Node)
    assert model.nodes["sample"].model == 'test'
    case_id = Property({"handle": "case_id", "value_domain": "string"})
    model.add_prop(case, case_id)
    assert model.props[("case", "case_id")]
    assert model.props[("case", "case_id")].value_domain == 'string'
    assert 'case_id' in model.nodes['case'].props
    sample = model.nodes["sample"]
    of_case = Edge({"handle": "of_case", "src": sample, "dst": case})
    of_case.props['operator'] = Property({
        "handle": "operator",
        "value_domain": "boolean"
    })
    model.add_edge(of_case)
    assert model.edges[('of_case', 'sample', 'case')]
    assert model.contains(of_case.props['operator'])
    assert of_case.props['operator'].model == 'test'
    assert model.props[('of_case', 'sample', 'case', 'operator')]
    assert model.props[('of_case', 'sample', 'case',
                        'operator')].value_domain == 'boolean'
    dx = Property({"handle": "diagnosis", "value_domain": "value_set"})
    tm = Term({"value": "CRS"})
    model.add_prop(case, dx)
    model.add_terms(dx, tm, 'rockin_pneumonia', 'fungusamongus')
    assert {x.value
            for x in dx.terms.values()
            } == {'CRS', 'rockin_pneumonia', 'fungusamongus'}
示例#3
0
def test_put_model(bento_neo4j):
  (b,h)=bento_neo4j
  the_mdb = MDB(uri=b)
  assert the_mdb
  ObjectMap.clear_cache()
  m = Model(handle='ICDC',mdb=the_mdb)
  m.dget()
  prop = m.props[('sample','sample_type')]
  sample = m.nodes['sample']
  edge = m.edges[('on_visit','sample', 'visit')]
  term = Term({"value":"electric_boogaloo"})
  m.add_terms(prop, term)
  node = m.nodes['lab_exam']
  m.dput()
  with m.drv.session() as session:
    result = session.run('match (v:value_set)-->(t:term {value:"electric_boogaloo"}) return v,t')
    rec = result.single()
    assert rec['v'].id == prop.value_set.neoid
    assert rec['t'].id == term.neoid
    assert rec['t']['value'] == term.value
    result = session.run('match (n:node {handle:"lab_exam"}) return n')
    rec = result.single()
    assert rec['n'].id == node.neoid

  term = m.props[('demographic','sex')].terms['M']
  assert term.concept
  assert term.concept._id == "337c0e4f-506a-4f4e-95f6-07c3462b81ff"

  concept = term.concept
  assert term in concept.belongs.values()
  term.concept=None
  assert not term in concept.belongs.values()
  assert ('concept',concept) in term.removed_entities
  m.dput()
  with m.drv.session() as session:
    result = session.run('match (t:term) where id(t)=$id return t',{"id":term.neoid})
    assert result.single() # term there
    result = session.run('match (c:concept) where id(c)=$id return c',{"id":concept.neoid})
    assert result.single() # concept there
    result = session.run('match (t:term)-->(c:concept) where id(t)=$id return t',{"id":term.neoid})
    assert not result.single() # but link is gone
  
    concept._id="heydude"
    term.concept = concept
    prop.model = None
    assert not prop.model

  m.dput()

  with m.drv.session() as session:
    result = session.run('match (t:term)--(c:concept) where id(t)=$id return c',{"id":term.neoid})
    s = result.single()
    assert s
    assert s['c'].id == concept.neoid
    assert s['c']['id'] == "heydude"
    result = session.run('match (p:property) where id(p)=$id return p',{"id":prop.neoid})
    s = result.single()
    assert s
    assert s['p'].id == prop.neoid
    assert not 'model' in s['p']

  prop.model = 'ICDC'
  at_enrollment = m.edges[('at_enrollment','prior_surgery','enrollment')]
  prior_surgery = m.nodes['prior_surgery']
  with m.drv.session() as session:
    result = session.run('match (n:node)<-[:has_src]-(r:relationship {handle:"at_enrollment"})-[:has_dst]->(:node {handle:"enrollment"}) where id(n)=$id return r',{"id":prior_surgery.neoid})
    s = result.single()
    assert s

  m.rm_edge(at_enrollment)
  assert not at_enrollment.src
  assert not at_enrollment.dst
  assert not at_enrollment in m.edges_out(prior_surgery)
  m.dput()
  with m.drv.session() as session:
    result = session.run('match (n:node)<-[:has_src]-(r:relationship {handle:"at_enrollment"})-[:has_dst]->(:node {handle:"enrollment"}) where id(n)=$id return r',{"id":prior_surgery.neoid})
    s = result.single()
    assert not s
    result = session.run('match (e:relationship) where id(e)=$id return e',{"id":at_enrollment.neoid})
    s = result.single()
    assert s
示例#4
0
def test_init_model():
    with pytest.raises(ArgError, match=".*requires arg 'handle'"):
        Model()
    m = Model('test')
    assert m
    assert m.handle == 'test'
示例#5
0
 def create_model(self):
     """Create :class:`Model` instance from loaded YAML
     Note: This is brittle, since the syntax of MDF is hard-coded into this method."""
     if not self.schema.keys():
         raise ValueError("attribute 'schema' not set - are yamls loaded?")
     self._model = Model(handle=self.handle)
     ynodes = self.schema["Nodes"]
     yedges = self.schema["Relationships"]
     ypropdefs = self.schema["PropDefinitions"]
     yunps = self.schema.get("UniversalNodeProperties")
     yurps = self.schema.get("UniversalRelationshipProperties")
     # create nodes
     for n in ynodes:
         yn = ynodes[n]
         init = {"handle": n, "model": self.handle}
         for a in ["category", "desc"]:
             if yn.get(a):
                 init[a] = yn[a]
         node = self._model.add_node(init)
         if yn.get("Tags"):
             tags = CollValue({}, owner=node, owner_key="tags")
             for t in yn["Tags"]:
                 tags[t] = Tag({"value": t})
             node["tags"] = tags
     # create edges (relationships)
     for e in yedges:
         ye = yedges[e]
         for ends in ye["Ends"]:
             init = {
                 "handle": e,
                 "model": self.handle,
                 "src": self._model.nodes[ends["Src"]],
                 "dst": self._model.nodes[ends["Dst"]],
                 "multiplicity": ends.get("Mul") or ye.get("Mul")
                 or MDF.default_mult,
                 "desc": ends.get("Desc") or ye.get("Desc"),
             }
             edge = self._model.add_edge(init)
             Tags = ye.get("Tags") or ends.get("Tags")
             if Tags:
                 tags = CollValue({}, owner=edge, owner_key="tags")
                 for t in Tags:
                     tags[t] = Tag({"value": t})
                     edge["tags"] = tags
     # create properties
     for ent in ChainMap(self._model.nodes, self._model.edges).values():
         if isinstance(ent, Node):
             pnames = ynodes[ent.handle]["Props"]
             if yunps:
                 pnames.extend(
                     yunps["mayHave"] if yunps.get("mayHave") else [])
                 pnames.extend(
                     yunps["mustHave"] if yunps.get("mustHave") else [])
         elif isinstance(ent, Edge):
             # props elts appearing Ends hash take
             # precedence over Props elt in the
             # handle's hash
             (hdl, src, dst) = ent.triplet
             [end] = [
                 e for e in yedges[hdl]["Ends"]
                 if e["Src"] == src and e["Dst"] == dst
             ]
             pnames = end.get("Props") or yedges[hdl].get("Props")
             if yurps:
                 pnames.extend(
                     yurps["mayHave"] if yurps.get("mayHave") else [])
                 pnames.extend(
                     yurps["mustHave"] if yurps.get("mustHave") else [])
         else:
             raise AttributeError(
                 "unhandled entity type {type} for properties".format(
                     type=type(ent).__name__))
         if pnames:
             for pname in pnames:
                 ypdef = ypropdefs.get(pname)
                 if not ypdef:
                     warn(
                         "property '{pname}' does not have a corresponding propdef"
                         .format(pname=pname))
                     break
                 init = {"handle": pname, "model": self.handle}
                 if ypdef.get("Type"):
                     init.update(self.calc_value_domain(ypdef["Type"]))
                 else:
                     init["value_domain"] = MDF.default_type
                 prop = self._model.add_prop(ent, init)
                 ent.props[prop.handle] = prop
                 if ypdef.get("Tags"):
                     tags = CollValue({}, owner=node, owner_key="tags")
                     for t in ypdef["Tags"]:
                         tags[t] = Tag({"value": t})
                     prop["tags"] = tags
     return self._model
示例#6
0
class MDF(object):
    default_mult = "one_to_one"
    default_type = "TBD"

    def __init__(self, *yaml_files, handle=None):
        """Create a :class:`Model` from MDF YAML files.
        :param str|file|url *yaml_files: MDF filenames or file objects, in desired merge order
        :param str handle: Handle (name) for the resulting Model
        :attribute model: the :class:`bento_meta.model.Model` created"""
        if not handle or not isinstance(handle, str):
            raise ArgError("arg handle= must be a str - name for model")
        self.handle = handle
        self.files = yaml_files
        self.schema = om.MergedOptions()
        self._model = None
        if self.files:
            self.load_yaml()
            self.create_model()
        else:
            warn("No MDF files provided to constructor")
        pass

    @property
    def model(self):
        """The :class:`bento_meta.model.Model` object created from the MDF input"""
        return self._model

    def load_yaml(self):
        """Load YAML files or open file handles specified in constructor"""
        yloader = yaml.loader.Loader
        for f in self.files:
            if isinstance(f, str):
                if re.match("(?:file|https?)://", f):
                    response = requests.get(f)
                    if not response.ok:
                        raise ArgError(
                            "Fetching url {} returned code {}".format(
                                response.url, response.status_code))
                    response.encoding = "utf8"
                    f = response.text
                else:
                    f = open(f, "r")
            try:
                yml = yaml.load(f, Loader=yloader)
                self.schema.update(yml)
            except ConstructorError as ce:
                print("YAML constructor failed in '{fn}':\n{e}".format(
                    fn=f.name, e=ce))
                raise ce
            except ParserError as pe:
                print("YAML parser failed in '{fn}':\n{e}".format(fn=f.name,
                                                                  e=pe))
                raise pe
            except Exception:
                raise

    def create_model(self):
        """Create :class:`Model` instance from loaded YAML
        Note: This is brittle, since the syntax of MDF is hard-coded into this method."""
        if not self.schema.keys():
            raise ValueError("attribute 'schema' not set - are yamls loaded?")
        self._model = Model(handle=self.handle)
        ynodes = self.schema["Nodes"]
        yedges = self.schema["Relationships"]
        ypropdefs = self.schema["PropDefinitions"]
        yunps = self.schema.get("UniversalNodeProperties")
        yurps = self.schema.get("UniversalRelationshipProperties")
        # create nodes
        for n in ynodes:
            yn = ynodes[n]
            init = {"handle": n, "model": self.handle}
            for a in ["category", "desc"]:
                if yn.get(a):
                    init[a] = yn[a]
            node = self._model.add_node(init)
            if yn.get("Tags"):
                tags = CollValue({}, owner=node, owner_key="tags")
                for t in yn["Tags"]:
                    tags[t] = Tag({"value": t})
                node["tags"] = tags
        # create edges (relationships)
        for e in yedges:
            ye = yedges[e]
            for ends in ye["Ends"]:
                init = {
                    "handle": e,
                    "model": self.handle,
                    "src": self._model.nodes[ends["Src"]],
                    "dst": self._model.nodes[ends["Dst"]],
                    "multiplicity": ends.get("Mul") or ye.get("Mul")
                    or MDF.default_mult,
                    "desc": ends.get("Desc") or ye.get("Desc"),
                }
                edge = self._model.add_edge(init)
                Tags = ye.get("Tags") or ends.get("Tags")
                if Tags:
                    tags = CollValue({}, owner=edge, owner_key="tags")
                    for t in Tags:
                        tags[t] = Tag({"value": t})
                        edge["tags"] = tags
        # create properties
        for ent in ChainMap(self._model.nodes, self._model.edges).values():
            if isinstance(ent, Node):
                pnames = ynodes[ent.handle]["Props"]
                if yunps:
                    pnames.extend(
                        yunps["mayHave"] if yunps.get("mayHave") else [])
                    pnames.extend(
                        yunps["mustHave"] if yunps.get("mustHave") else [])
            elif isinstance(ent, Edge):
                # props elts appearing Ends hash take
                # precedence over Props elt in the
                # handle's hash
                (hdl, src, dst) = ent.triplet
                [end] = [
                    e for e in yedges[hdl]["Ends"]
                    if e["Src"] == src and e["Dst"] == dst
                ]
                pnames = end.get("Props") or yedges[hdl].get("Props")
                if yurps:
                    pnames.extend(
                        yurps["mayHave"] if yurps.get("mayHave") else [])
                    pnames.extend(
                        yurps["mustHave"] if yurps.get("mustHave") else [])
            else:
                raise AttributeError(
                    "unhandled entity type {type} for properties".format(
                        type=type(ent).__name__))
            if pnames:
                for pname in pnames:
                    ypdef = ypropdefs.get(pname)
                    if not ypdef:
                        warn(
                            "property '{pname}' does not have a corresponding propdef"
                            .format(pname=pname))
                        break
                    init = {"handle": pname, "model": self.handle}
                    if ypdef.get("Type"):
                        init.update(self.calc_value_domain(ypdef["Type"]))
                    else:
                        init["value_domain"] = MDF.default_type
                    prop = self._model.add_prop(ent, init)
                    ent.props[prop.handle] = prop
                    if ypdef.get("Tags"):
                        tags = CollValue({}, owner=node, owner_key="tags")
                        for t in ypdef["Tags"]:
                            tags[t] = Tag({"value": t})
                        prop["tags"] = tags
        return self._model

    def calc_value_domain(self, typedef):
        if isinstance(typedef, dict):
            if typedef.get("pattern"):
                return {
                    "value_domain": "regexp",
                    "pattern": typedef["pattern"]
                }
            elif typedef.get("units"):
                return {
                    "value_domain": typedef.get("value_type"),
                    "units": ";".join(typedef.get("units")),
                }
            else:
                # punt
                warn("MDF type descriptor unrecognized: json looks like {j}".
                     format(j=json.dumps(typedef)))
                return {"value_domain": json.dumps(typedef)}
        elif isinstance(typedef,
                        list):  # a valueset: create value set and term objs
            vs = ValueSet({"_id": str(uuid4())})
            vs.handle = self.handle + vs._id[0:8]
            if re.match("^(?:https?|bolt)://", typedef[0]):  # looks like url
                vs.url = typedef[0]
            else:  # an enum
                for t in typedef:
                    vs.terms[t] = Term({"value": t})
            return {"value_domain": "value_set", "value_set": vs}
        else:
            return {"value_domain": MDF.default_type}