Python dot.split_field函数代码示例

    def select(self, selectList, fromPath, varName, sourceVar):
        path = split_field(fromPath)
        is_deep = len(path) > 1
        heads = []
        list = []
        for s in selectList:
            if is_deep:
                if s.value and isKeyword(s.value):
                    shortForm = self._translate(s.value)
                    list.append("Value2Pipe(" + shortForm + ")\n")
                else:
                    Log.error("do not know how to handle yet")
            else:
                if s.value and isKeyword(s.value):
                    list.append("Value2Pipe(getDocValue(" + value2MVEL(s.value) + "))\n")
                elif s.value:
                    shortForm = self._translate(s.value)
                    list.append("Value2Pipe(" + shortForm + ")\n")
                else:
                    code, decode = self.Parts2Term(s.domain)
                    heads.append(code.head)
                    list.append("Value2Pipe(" + code.body + ")\n")


        if len(split_field(fromPath)) > 1:
            output = 'if (' + varName + ' != "") ' + varName + '+="|";\n' + varName + '+=' + '+"|"+'.join(["Value2Pipe("+v+")\n" for v in list]) + ';\n'
        else:
            output = varName + ' = ' + '+"|"+'.join(["Value2Pipe("+v+")\n" for v in list]) + ';\n'

        return Dict(
            head="".join(heads),
            body=output
        )

def tuple(data, field_name):
    """
    RETURN LIST  OF TUPLES
    """
    if isinstance(data, Cube):
        Log.error("not supported yet")

    if isinstance(data, FlatList):
        Log.error("not supported yet")

    if isinstance(field_name, Mapping) and "value" in field_name:
        # SIMPLIFY {"value":value} AS STRING
        field_name = field_name["value"]

    # SIMPLE PYTHON ITERABLE ASSUMED
    if isinstance(field_name, basestring):
        if len(split_field(field_name)) == 1:
            return [(d[field_name],) for d in data]
        else:
            path = split_field(field_name)
            output = []
            flat_list._tuple1(data, path, 0, output)
            return output
    elif isinstance(field_name, list):
        paths = [_select_a_field(f) for f in field_name]
        output = DictList()
        _tuple((), unwrap(data), paths, 0, output)
        return output
    else:
        paths = [_select_a_field(field_name)]
        output = DictList()
        _tuple((), data, paths, 0, output)
        return output

    def _get_columns(self, table=None):
        # TODO: HANDLE MORE THEN ONE ES, MAP TABLE SHORT_NAME TO ES INSTANCE
        alias_done = set()
        index = split_field(table)[0]
        query_path = split_field(table)[1:]
        metadata = self.default_es.get_metadata(index=index)
        for index, meta in qb.sort(metadata.indices.items(), {"value": 0, "sort": -1}):
            for _, properties in meta.mappings.items():
                columns = _elasticsearch.parse_properties(index, None, properties.properties)
                columns = columns.filter(lambda r: not r.abs_name.startswith("other.") and not r.abs_name.startswith("previous_values.cf_"))  # TODO: REMOVE WHEN jobs PROPERTY EXPLOSION IS CONTAINED
                with Timer("upserting {{num}} columns", {"num": len(columns)}, debug=DEBUG):
                    with self.columns.locker:
                        for c in columns:
                            # ABSOLUTE
                            c.table = join_field([index]+query_path)
                            self.upsert_column(c)

                            for alias in meta.aliases:
                                # ONLY THE LATEST ALIAS IS CHOSEN TO GET COLUMNS
                                if alias in alias_done:
                                    continue
                                alias_done.add(alias)
                                c = copy(c)
                                c.table = join_field([alias]+query_path)
                                self.upsert_column(c)

def _select_a_field(field):
    if isinstance(field, basestring):
        return wrap({"name": field, "value": split_field(field)})
    elif isinstance(wrap(field).value, basestring):
        field = wrap(field)
        return wrap({"name": field.name, "value": split_field(field.value)})
    else:
        return wrap({"name": field.name, "value": field.value})

    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 add_column(c, query_path):
                c.last_updated = Date.now()
                c.table = join_field([c.es_index]+split_field(query_path[0]))

                with self.meta.columns.locker:
                    self._upsert_column(c)
                    for alias in meta.aliases:
                        c = copy(c)
                        c.table = join_field([alias]+split_field(query_path[0]))
                        self._upsert_column(c)

                def defParent(name):
                    # DO NOT MAKE THE SAME PARENT TWICE
                    if name in parentVarNames:
                        return
                    parentVarNames.add(name)

                    if len(split_field(name)) == 1:
                        contextVariables.append("Map " + name + " = new HashMap();\n")
                    else:
                        defParent(join_field(split_field(name)[0:-1]))
                        contextVariables.append(name + " = new HashMap();\n")

def _get_nested_path(field, schema):
    if not INDEX_CACHE:
        _late_import()

    if is_keyword(field):
        field = join_field([schema.es.alias] + split_field(field))
        for i, f in reverse(enumerate(split_field(field))):
            path = join_field(split_field(field)[0:i + 1:])
            if path in INDEX_CACHE:
                return join_field(split_field(path)[1::])
    return None

    def getFrameVariables(self, body):
        contextVariables = []
        columns = self.fromData.columns

        parentVarNames = set()    # ALL PARENTS OF VARIABLES WITH "." IN NAME
        body = body.replace(".?", ".")

        for i, c in enumerate(columns):
            j = body.find(c.name, 0)
            while j >= 0:
                s = j
                j = body.find(c.name, s + 1)

                test0 = body[s - 1: s + len(c.name) + 1:]
                test3 = body[s - 8: s + len(c.name):]

                if test0[:-1] == "\"" + c.name:
                    continue
                if test3 == "_source." + c.name:
                    continue

                def defParent(name):
                    # DO NOT MAKE THE SAME PARENT TWICE
                    if name in parentVarNames:
                        return
                    parentVarNames.add(name)

                    if len(split_field(name)) == 1:
                        contextVariables.append("Map " + name + " = new HashMap();\n")
                    else:
                        defParent(join_field(split_field(name)[0:-1]))
                        contextVariables.append(name + " = new HashMap();\n")

                body = body.replace(c.name, "-"*len(c.name))

                if self.isLean or c.useSource:
                    if len(split_field(c.name)) > 1:
                        defParent(join_field(split_field(c.name)[0:-1]))
                        contextVariables.append(c.name + " = getSourceValue(\"" + c.name + "\");\n")
                    else:
                        contextVariables.append(c.name + " = _source[\"" + c.name + "\"];\n")
                else:
                    if len(split_field(c.name)) > 1:
                        defParent(join_field(split_field(c.name)[0:-1]))
                        contextVariables.append(c.name + " = getDocValue(\"" + c.name + "\");\n")
                    else:
                        contextVariables.append(c.name + " = getDocValue(\"" + c.name + "\");\n")
                break

        return "".join(contextVariables)

    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)

    def get_columns(self, table_name, column_name=None, force=False):
        """
        RETURN METADATA COLUMNS
        """
        try:
            # LAST TIME WE GOT INFO FOR THIS TABLE
            short_name = join_field(split_field(table_name)[0:1])
            table = self.get_table(short_name)[0]

            if not table:
                table = Table(
                    name=short_name,
                    url=None,
                    query_path=None,
                    timestamp=Date.now()
                )
                with self.meta.tables.locker:
                    self.meta.tables.add(table)
                self._get_columns(table=short_name)
            elif force or table.timestamp == None or table.timestamp < Date.now() - MAX_COLUMN_METADATA_AGE:
                table.timestamp = Date.now()
                self._get_columns(table=short_name)

            with self.meta.columns.locker:
                columns = self.meta.columns.find(table_name, column_name)
            if columns:
                columns = jx.sort(columns, "name")
                # AT LEAST WAIT FOR THE COLUMNS TO UPDATE
                while len(self.todo) and not all(columns.get("last_updated")):
                    Log.note("waiting for columns to update {{columns|json}}", columns=[c.table+"."+c.es_column for c in columns if not c.last_updated])
                    Thread.sleep(seconds=1)
                return columns
        except Exception, e:
            Log.error("Not expected", cause=e)

def wrap_from(frum, schema=None):
    """
    :param frum:
    :param schema:
    :return:
    """
    if not type2container:
        _delayed_imports()

    frum = wrap(frum)

    if isinstance(frum, basestring):
        if not config.default.settings:
            Log.error("expecting pyLibrary.queries.query.config.default.settings to contain default elasticsearch connection info")

        settings = set_default(
            {
                "index": split_field(frum)[0],
                "name": frum,
            },
            config.default.settings
        )
        settings.type = None  # WE DO NOT WANT TO INFLUENCE THE TYPE BECAUSE NONE IS IN THE frum STRING ANYWAY
        return type2container["elasticsearch"](settings)
    elif isinstance(frum, Mapping) and frum.type and type2container[frum.type]:
        # TODO: Ensure the frum.name is set, so we capture the deep queries
        if not frum.type:
            Log.error("Expecting from clause to have a 'type' property")
        return type2container[frum.type](frum.settings)
    elif isinstance(frum, Mapping) and (frum["from"] or isinstance(frum["from"], (list, set))):
        from pyLibrary.queries.query import Query
        return Query(frum, schema=schema)
    else:
        return frum

def es_query_template(path):
    """
    RETURN TEMPLATE AND PATH-TO-FILTER AS A 2-TUPLE
    :param path:
    :return:
    """
    sub_path = split_field(path)[1:]

    if sub_path:
        f0 = {}
        f1 = {}
        output = wrap(
            {
                "filter": {
                    "and": [
                        f0,
                        {"nested": {"path": join_field(sub_path), "filter": f1, "inner_hits": {"size": 100000}}},
                    ]
                },
                "from": 0,
                "size": 0,
                "sort": [],
            }
        )
        return output, wrap([f0, f1])
    else:
        f0 = {}
        output = wrap({"query": {"filtered": {"filter": f0}}, "from": 0, "size": 0, "sort": []})
        return output, wrap([f0])

    def parse_field(fieldname, data, depth):
        """
        RETURN (first, rest) OF fieldname
        """
        col = split_field(fieldname)
        d = data
        for i, c in enumerate(col):
            try:
                d = d[c]
            except Exception, e:
                Log.error("{{name}} does not exist", name=fieldname)
            if isinstance(d, list) and len(col) > 1:
                if len(primary_column) <= depth + i:
                    primary_nested.append(True)
                    primary_column.append(c)
                    primary_branch.append(d)
                elif primary_nested[depth] and primary_column[depth + i] != c:
                    Log.error("only one branch of tree allowed")
                else:
                    primary_nested[depth + i] = True
                    primary_column[depth + i] = c
                    primary_branch[depth + i] = d

                return c, join_field(col[i + 1 :])
            else:
                if len(primary_column) <= depth + i:
                    primary_nested.append(False)
                    primary_column.append(c)
                    primary_branch.append([d])

    def __setitem__(self, key, value):
        if key == "":
            from pyLibrary.debugs.logs import Log

            Log.error("key is empty string.  Probably a bad idea")
        if isinstance(key, str):
            key = key.decode("utf8")
        d=self
        try:
            value = unwrap(value)
            if key.find(".") == -1:
                if value is None:
                    dict.pop(d, key, None)
                else:
                    dict.__setitem__(d, key, value)
                return self

            seq = split_field(key)
            for k in seq[:-1]:
                d = _getdefault(d, k)
            if value == None:
                dict.pop(d, seq[-1], None)
            else:
                dict.__setitem__(d, seq[-1], value)
            return self
        except Exception, e:
            raise e