Python lists.DictList类代码示例

        def post(sql):
            # FIND OUT THE default DOMAIN SIZES
            result = self.db.column_query(sql)
            num_edges = len(edges)
            for e, edge in enumerate(edges):
                domain = edge.domain
                if domain.type == "default":
                    domain.type = "set"
                    parts = set(result[e])
                    domain.partitions = [{"index": i, "value": p} for i, p in enumerate(parts)]
                    domain.map = {p: i for i, p in enumerate(parts)}
                else:
                    Log.error("Do not know what to do here, yet")

            # FILL THE DATA CUBE
            maps = [(unwrap(e.domain.map), result[i]) for i, e in enumerate(edges)]
            cubes = DictList()
            for c, s in enumerate(select):
                data = Matrix(*[len(e.domain.partitions) + (1 if e.allow_nulls else 0) for e in edges])
                for rownum, value in enumerate(result[c + num_edges]):
                    coord = [m[r[rownum]] for m, r in maps]
                    data[coord] = value
                cubes.append(data)

            if isinstance(query.select, list):
                return cubes
            else:
                return cubes[0]

class _Stats(WindowFunction):
    """
    TRACK STATS, BUT IGNORE OUTLIERS
    """

    def __init__(self, middle=None, *args, **kwargs):
        object.__init__(self)
        self.middle = middle
        self.samples = DictList()

    def add(self, value):
        if value == None:
            return
        self.samples.append(value)

    def sub(self, value):
        if value == None:
            return
        self.samples.remove(value)

    def merge(self, agg):
        Log.error("Do not know how to handle")

    def end(self):
        ignore = Math.ceiling(len(self.samples) * (1 - self.middle) / 2)
        if ignore * 2 >= len(self.samples):
            return stats.Stats()
        output = stats.Stats(samples=sorted(self.samples)[ignore:len(self.samples) - ignore:])
        output.samples = list(self.samples)
        return output

def _select(template, data, fields, depth):
    output = DictList()
    deep_path = []
    deep_fields = UniqueIndex(["name"])
    for d in data:
        if isinstance(d, Dict):
            Log.error("programmer error, _select can not handle Dict")

        record = template.copy()
        children = None
        for f in fields:
            index, c = _select_deep(d, f, depth, record)
            children = c if children is None else children
            if index:
                path = f.value[0:index:]
                if not deep_fields[f]:
                    deep_fields.add(f)  # KEEP TRACK OF WHICH FIELDS NEED DEEPER SELECT
                short = MIN(len(deep_path), len(path))
                if path[:short:] != deep_path[:short:]:
                    Log.error("Dangerous to select into more than one branch at time")
                if len(deep_path) < len(path):
                    deep_path = path
        if not children:
            output.append(record)
        else:
            output.extend(_select(record, children, deep_fields, depth + 1))

    return output

    def select(self, fields):
        if isinstance(fields, Mapping):
            fields=fields.value

        if isinstance(fields, basestring):
            # RETURN LIST OF VALUES
            if len(split_field(fields)) == 1:
                if self.path[0] == fields:
                    return [d[1] for d in self.data]
                else:
                    return [d[0][fields] for d in self.data]
            else:
                keys = split_field(fields)
                depth = coalesce(MIN([i for i, (k, p) in enumerate(zip(keys, self.path)) if k != p]), len(self.path))  # LENGTH OF COMMON PREFIX
                short_key = keys[depth:]

                output = DictList()
                _select1((wrap(d[depth]) for d in self.data), short_key, 0, output)
                return output

        if isinstance(fields, list):
            output = DictList()

            meta = []
            for f in fields:
                if hasattr(f.value, "__call__"):
                    meta.append((f.name, f.value))
                else:
                    meta.append((f.name, functools.partial(lambda v, d: d[v], f.value)))

            for row in self._values():
                agg = Dict()
                for name, f in meta:
                    agg[name] = f(row)

                output.append(agg)

            return output

            # meta = []
            # for f in fields:
            #     keys = split_field(f.value)
            #     depth = coalesce(MIN([i for i, (k, p) in enumerate(zip(keys, self.path)) if k != p]), len(self.path))  # LENGTH OF COMMON PREFIX
            #     short_key = join_field(keys[depth:])
            #
            #     meta.append((f.name, depth, short_key))
            #
            # for row in self._data:
            #     agg = Dict()
            #     for name, depth, short_key in meta:
            #         if short_key:
            #             agg[name] = row[depth][short_key]
            #         else:
            #             agg[name] = row[depth]
            #     output.append(agg)
            # return output

        Log.error("multiselect over FlatList not supported")

 def more():
     output = DictList()
     for i in range(size):
         try:
             output.append(iterator.next())
         except StopIteration:
             done.append(True)
             break
     return output

    def update(self, command):
        """
        EXPECTING command == {"set":term, "where":where}
        THE set CLAUSE IS A DICT MAPPING NAMES TO VALUES
        THE where CLAUSE IS AN ES FILTER
        """
        command = wrap(command)
        schema = self._es.get_schema()

        # GET IDS OF DOCUMENTS
        results = self._es.search(
            {
                "fields": listwrap(schema._routing.path),
                "query": {
                    "filtered": {"query": {"match_all": {}}, "filter": jx_expression(command.where).to_esfilter()}
                },
                "size": 200000,
            }
        )

        # SCRIPT IS SAME FOR ALL (CAN ONLY HANDLE ASSIGNMENT TO CONSTANT)
        scripts = DictList()
        for k, v in command.set.items():
            if not is_keyword(k):
                Log.error("Only support simple paths for now")
            if isinstance(v, Mapping) and v.doc:
                scripts.append({"doc": v.doc})
            else:
                scripts.append({"script": "ctx._source." + k + " = " + jx_expression(v).to_ruby()})

        if results.hits.hits:
            updates = []
            for h in results.hits.hits:
                for s in scripts:
                    updates.append(
                        {
                            "update": {
                                "_id": h._id,
                                "_routing": unwraplist(h.fields[literal_field(schema._routing.path)]),
                            }
                        }
                    )
                    updates.append(s)
            content = ("\n".join(convert.value2json(c) for c in updates) + "\n").encode("utf-8")
            response = self._es.cluster.post(
                self._es.path + "/_bulk", data=content, headers={"Content-Type": "application/json"}
            )
            if response.errors:
                Log.error(
                    "could not update: {{error}}",
                    error=[e.error for i in response["items"] for e in i.values() if e.status not in (200, 201)],
                )

    def _aggop(self, query):
        """
        SINGLE ROW RETURNED WITH AGGREGATES
        """
        if isinstance(query.select, list):
            # RETURN SINGLE OBJECT WITH AGGREGATES
            for s in query.select:
                if s.aggregate not in aggregates:
                    Log.error("Expecting all columns to have an aggregate: {{select}}", select=s)

            selects = DictList()
            for s in query.select:
                selects.append(aggregates[s.aggregate].replace("{{code}}", s.value) + " AS " + self.db.quote_column(s.name))

            sql = expand_template("""
                SELECT
                    {{selects}}
                FROM
                    {{table}}
                {{where}}
            """, {
                "selects": SQL(",\n".join(selects)),
                "table": self._subquery(query["from"])[0],
                "where": self._where2sql(query.filter)
            })

            return sql, lambda sql: self.db.column(sql)[0]  # RETURNING SINGLE OBJECT WITH AGGREGATE VALUES
        else:
            # RETURN SINGLE VALUE
            s0 = query.select
            if s0.aggregate not in aggregates:
                Log.error("Expecting all columns to have an aggregate: {{select}}", select=s0)

            select = aggregates[s0.aggregate].replace("{{code}}", s0.value) + " AS " + self.db.quote_column(s0.name)

            sql = expand_template("""
                SELECT
                    {{selects}}
                FROM
                    {{table}}
                {{where}}
            """, {
                "selects": SQL(select),
                "table": self._subquery(query["from"])[0],
                "where": self._where2sql(query.where)
            })

            def post(sql):
                result = self.db.column_query(sql)
                return result[0][0]

            return sql, post  # RETURN SINGLE VALUE

class DefaultDomain(Domain):
    """
    DOMAIN IS A LIST OF OBJECTS, EACH WITH A value PROPERTY
    """

    __slots__ = ["NULL", "partitions", "map", "limit"]

    def __init__(self, **desc):
        Domain.__init__(self, **desc)

        self.NULL = Null
        self.partitions = DictList()
        self.map = dict()
        self.map[None] = self.NULL
        self.limit = desc.get('limit')

    def compare(self, a, b):
        return value_compare(a.value, b.value)

    def getCanonicalPart(self, part):
        return self.getPartByKey(part.value)

    def getPartByKey(self, key):
        canonical = self.map.get(key)
        if canonical:
            return canonical

        canonical = Dict(name=key, value=key)

        self.partitions.append(canonical)
        self.map[key] = canonical
        return canonical

    # def getIndexByKey(self, key):
    #     return self.map.get(key).dataIndex;

    def getKey(self, part):
        return part.value

    def getEnd(self, part):
        return part.value

    def getLabel(self, part):
        return part.value

    def as_dict(self):
        output = Domain.as_dict(self)
        output.partitions = self.partitions
        output.limit = self.limit
        return output

def _normalize_sort(sort=None):
    """
    CONVERT SORT PARAMETERS TO A NORMAL FORM SO EASIER TO USE
    """

    if not sort:
        return DictList.EMPTY

    output = DictList()
    for s in listwrap(sort):
        if isinstance(s, basestring) or Math.is_integer(s):
            output.append({"field": s, "sort": 1})
        else:
            output.append({"field": coalesce(s.field, s.value), "sort": coalesce(sort_direction[s.sort], 1)})
    return wrap(output)

    def __init__(self, **desc):
        Domain.__init__(self, **desc)

        self.NULL = Null
        self.partitions = DictList()
        self.map = dict()
        self.map[None] = self.NULL

def _getAllEdges(facetEdges, edgeDepth):
    """
    RETURN ALL PARTITION COMBINATIONS:  A LIST OF ORDERED TUPLES
    """
    if edgeDepth == len(facetEdges):
        return [()]
    edge = facetEdges[edgeDepth]

    deeper = _getAllEdges(facetEdges, edgeDepth + 1)

    output = DictList()
    partitions = edge.domain.partitions
    for part in partitions:
        for deep in deeper:
            output.append((part,) + deep)
    return output

 def _iter():
     g = 0
     out = DictList()
     try:
         for i, d in enumerate(data):
             out.append(d)
             if (i + 1) % max_size == 0:
                 yield g, out
                 g += 1
                 out = DictList()
         if out:
             yield g, out
     except Exception, e:
         if out:
             # AT LEAST TRY TO RETURN WHAT HAS BEEN PROCESSED SO FAR
             yield g, out
         Log.error("Problem inside qb.groupby", e)

def _tuple(template, data, fields, depth, output):
    deep_path = None
    deep_fields = DictList()
    for d in data:
        record = template
        for f in fields:
            index, children, record = _tuple_deep(d, f, depth, record)
            if index:
                path = f.value[0:index:]
                deep_fields.append(f)
                if deep_path and path != deep_path:
                    Log.error("Dangerous to select into more than one branch at time")
        if not children:
            output.append(record)
        else:
            _tuple(record, children, deep_fields, depth + 1, output)

    return output

def _where_terms(master, where, schema):
    """
    USE THE SCHEMA TO CONVERT DIMENSION NAMES TO ES FILTERS
    master - TOP LEVEL WHERE (FOR PLACING NESTED FILTERS)
    """
    if isinstance(where, Mapping):
        if where.term:
            # MAP TERM
            try:
                output = _map_term_using_schema(master, [], where.term, schema.edges)
                return output
            except Exception, e:
                Log.error("programmer problem?", e)
        elif where.terms:
            # MAP TERM
            output = DictList()
            for k, v in where.terms.items():
                if not isinstance(v, (list, set)):
                    Log.error("terms filter expects list of values")
                edge = schema.edges[k]
                if not edge:
                    output.append({"terms": {k: v}})
                else:
                    if isinstance(edge, basestring):
                        # DIRECT FIELD REFERENCE
                        return {"terms": {edge: v}}
                    try:
                        domain = edge.getDomain()
                    except Exception, e:
                        Log.error("programmer error", e)
                    fields = domain.dimension.fields
                    if isinstance(fields, Mapping):
                        or_agg = []
                        for vv in v:
                            and_agg = []
                            for local_field, es_field in fields.items():
                                vvv = vv[local_field]
                                if vvv != None:
                                    and_agg.append({"term": {es_field: vvv}})
                            or_agg.append({"and": and_agg})
                        output.append({"or": or_agg})
                    elif isinstance(fields, list) and len(fields) == 1 and is_keyword(fields[0]):
                        output.append({"terms": {fields[0]: v}})
                    elif domain.partitions:
                        output.append({"or": [domain.getPartByKey(vv).esfilter for vv in v]})
            return {"and": output}

    def getDomain(self, **kwargs):
        # kwargs.depth IS MEANT TO REACH INTO SUB-PARTITIONS
        kwargs = wrap(kwargs)
        kwargs.depth = coalesce(kwargs.depth, len(self.fields)-1 if isinstance(self.fields, list) else None)

        if not self.partitions and self.edges:
            # USE EACH EDGE AS A PARTITION, BUT isFacet==True SO IT ALLOWS THE OVERLAP
            partitions = [
                {
                    "name": v.name,
                    "value": v.name,
                    "where": v.where,
                    "style": v.style,
                    "weight": v.weight  # YO! WHAT DO WE *NOT* COPY?
                }
                for i, v in enumerate(self.edges)
                if i < coalesce(self.limit, DEFAULT_QUERY_LIMIT) and v.where
            ]
            self.isFacet = True
        elif kwargs.depth == None:  # ASSUME self.fields IS A dict
            partitions = DictList()
            for i, part in enumerate(self.partitions):
                if i >= coalesce(self.limit, DEFAULT_QUERY_LIMIT):
                    break
                partitions.append({
                    "name":part.name,
                    "value":part.value,
                    "where":part.where,
                    "style":coalesce(part.style, part.parent.style),
                    "weight":part.weight   # YO!  WHAT DO WE *NOT* COPY?
                })
        elif kwargs.depth == 0:
            partitions = [
                {
                    "name":v.name,
                    "value":v.value,
                    "where":v.where,
                    "style":v.style,
                    "weight":v.weight   # YO!  WHAT DO WE *NOT* COPY?
                }
                for i, v in enumerate(self.partitions)
                if i < coalesce(self.limit, DEFAULT_QUERY_LIMIT)]
        elif kwargs.depth == 1:
            partitions = DictList()
            rownum = 0
            for i, part in enumerate(self.partitions):
                if i >= coalesce(self.limit, DEFAULT_QUERY_LIMIT):
                    continue
                rownum += 1
                try:
                    for j, subpart in enumerate(part.partitions):
                        partitions.append({
                            "name":join_field(split_field(subpart.parent.name) + [subpart.name]),
                            "value":subpart.value,
                            "where":subpart.where,
                            "style":coalesce(subpart.style, subpart.parent.style),
                            "weight":subpart.weight   # YO!  WHAT DO WE *NOT* COPY?
                        })
                except Exception, e:
                    Log.error("", e)