Merge pull request #142 from juanjux/compat

V1 compatibility layer

Author: juanjux

bblfsh/compat.py

@@ -0,0 +1,335 @@
+"""
+This file provides a compatibility layer with the old UAST V1 (or client-python
+v2) API. You can see a summary of that API here:
+
+https://github.com/bblfsh/client-python/blob/d485273f457a174b40b820ad71195a739db04197/README.md
+
+Note that this won't translate the XPath queries from the old projection to the new use;
+even when using this module you're expected to use expressions matching the new
+projection.
+
+Note that since this is a pure Python translation layer, some performance
+impact is to be expected.
+"""
+import os
+import sys
+from typing import Union, List, Any, Optional
+
+import grpc
+
+import bblfsh.client as newbbl
+from bblfsh import role_id, role_name
+from bblfsh.node import Node
+from bblfsh.node_iterator import NodeIterator
+from bblfsh.result_context import ResultContext
+from bblfsh.aliases import (
+    ParseRequest, ParseResponse, DriverStub, ProtocolServiceStub,
+    VersionRequest, SupportedLanguagesRequest, ModeType,
+    Mode, VersionResponse, DESCRIPTOR
+)
+from bblfsh.pyuast import uast, iterator as native_iterator
+from bblfsh.tree_order import TreeOrder
+
+print("Warning: using deprecated bblfsh v1 compatibility layer.", file=sys.stderr)
+
+
+class WrongTypeException(Exception):
+    """
+    This exception is raised when the API receives an unexpected type
+    """
+    pass
+
+
+class CompatParseResponse:
+    """
+    This class emulates the API of the old ParseResponse object.
+    """
+    def __init__(self, ctx: ResultContext, filename: str = "") -> None:
+        self._res_context = ctx
+        self._filename = filename
+
+    @property
+    def uast(self) -> Node:
+        """
+        Returns the root Node.
+        """
+        return self._res_context.uast
+
+    @property
+    def ast(self) -> Node:
+        """
+        Returns the root Node. This is provided for compatibility, but
+        since the type of result is now expecified using CompatBblfshClient.parse
+        or parse_native, it'll return the same as uast().
+        """
+        return self._res_context.ast
+
+    @property
+    def ctx(self) -> ResultContext:
+        """
+        Returns the ResultContext of the response.
+        """
+        return self._res_context
+
+    @property
+    def elapsed(self) -> int:
+        """
+        Provided for compatibility, but since the new API's ParseResponse doesn't
+        provide an elapsed time it'll always return -1.
+        """
+        # FIXME(juanjux): check if the caller can get this, or measure it ourselves.
+        return -1
+
+    @property
+    def language(self) -> str:
+        """
+        Returns the language used for the request.
+        """
+        return self._res_context.language
+
+    @property
+    def filename(self) -> str:
+        """
+        Returns the filename used for the request.
+        """
+        return self._filename
+
+    @property
+    def DESCRIPTOR(self) -> Any:
+        """
+        Returns the gRPC context descriptor.
+        """
+        return self._res_context.ctx.DESCRIPTOR
+
+    @property
+    def errors(selfs) -> List:
+        """
+        Provided for compatibility. Since the new API will raise exceptions on errors,
+        this just returns and empty array.
+        """
+        # ParseResponse would have raised an exception on errors
+        return []
+
+
+class CompatBblfshClient:
+    """
+    This emulates the methods and properties of the old BblfshClient.
+    """
+    def __init__(self, endpoint: Union[str, grpc.Channel]) -> None:
+        """
+        Connects to the specified grpc endpoint which can be specified either as
+        a grpc Channel object or a connection string (like "0.0.0.0:6432").
+        """
+        self._bblfsh_cli = newbbl.BblfshClient(endpoint)
+
+        self._channel = self._bblfsh_cli._channel
+        self._stub_v1 = self._bblfsh_cli._stub_v1
+        self._stub_v2 = self._bblfsh_cli._stub_v2
+
+    def _parse(self, filename: str, language: str = None, contents: str = None,
+               timeout: float = None,
+               mode: ModeType = Mode.Value('ANNOTATED')) -> CompatParseResponse:
+
+        if timeout is not None:
+            timeout = int(timeout)
+
+        res = self._bblfsh_cli.parse(filename, language, contents,
+                                     mode=mode, timeout=timeout)
+        return CompatParseResponse(res, filename)
+
+    def parse(self, filename: str, language: str = None, contents: str = None,
+              timeout: float = None) -> CompatParseResponse:
+
+        """
+        Parse the specified filename or contents and return a CompatParseResponse.
+        """
+
+        return self._parse(filename, language, contents, timeout,
+                           Mode.Value('ANNOTATED'))
+
+    def native_parse(self, filename: str, language: str = None,
+                     contents: str = None,
+                     timeout: float = None) -> CompatParseResponse:
+        """
+        Same as parse() but the returned response will include only the native
+        (non annotated) AST.
+        """
+
+        return self._parse(filename, language, contents, timeout,
+                           Mode.Value('NATIVE'))
+
+    def supported_languages(self) -> List[str]:
+        """
+        Return a list of the languages that can be parsed by the connected
+        endpoint (driver or bblfsh daemon).
+        """
+        return self._bblfsh_cli.supported_languages()
+
+    def version(self) -> VersionResponse:
+        """
+        Returns the connected endpoint version.
+        """
+        return self._bblfsh_cli.version()
+
+    def close(self) -> None:
+        """
+        Closes the connection to the endpoint.
+        """
+        return self._bblfsh_cli.close()
+
+
+class CompatNodeIterator:
+    """
+    This emulates the API of the pre-v3 iterators.
+    """
+    def __init__(self, nodeit: NodeIterator, only_nodes: bool = False) -> None:
+        """
+        Creates a CompatNodeIterator compatibility object using a NodeIterator
+        from the post-v3 API. If the only_nodes parameter is set to true,
+        scalars and strings won't be included in the results.
+        """
+        self._nodeit = nodeit
+        self._only_nodes = only_nodes
+        # Used to forward calls of the old Node object
+        self._last_node: Optional[Node] = None
+
+    def __iter__(self) -> 'CompatNodeIterator':
+        return self
+
+    def __next__(self) -> Node:
+        next_val = next(self._nodeit)
+
+        is_node = isinstance(next_val, Node)
+        val = next_val.internal_node if is_node else next_val
+
+        # Skip positions and non dicts/lists, the later if only_nodes = True
+        skip = False
+        if isinstance(val, dict):
+            if "@type" not in val or val["@type"] == "uast:Positions":
+                skip = True
+        elif self._only_nodes:
+            skip = True
+
+        if skip:
+            val = self.__next__().internal_node
+
+        ret_val = next_val if is_node else Node(value=val)
+        self._last_node = ret_val
+        return ret_val
+
+    def filter(self, query: str) -> Optional['CompatNodeIterator']:
+        """
+        Further filter the results using this iterator as base.
+        """
+        if not self._last_node:
+            return None
+
+        return filter(self._last_node, query)
+
+    @property
+    def properties(self) -> dict:
+        """
+        Returns the properties of the current node in the iteration.
+        """
+        if isinstance(self._last_node, dict):
+            return self._last_node.keys()
+        else:
+            return {}
+
+
+def iterator(n: Union[Node, CompatNodeIterator, dict],
+        order: TreeOrder = TreeOrder.PRE_ORDER) -> CompatNodeIterator:
+    """
+    This function has the same signature as the pre-v3 iterator()
+    call returning a compatibility CompatNodeIterator.
+    """
+
+    if isinstance(n, CompatNodeIterator):
+        return CompatNodeIterator(n._nodeit.iterate(order), only_nodes=True)
+    elif isinstance(n, Node):
+        nat_it = native_iterator(n.internal_node, order)
+        return CompatNodeIterator(NodeIterator(nat_it), only_nodes=True)
+    elif isinstance(n, dict):
+        nat_it = native_iterator(n, order)
+        return CompatNodeIterator(NodeIterator(nat_it, uast()), only_nodes=True)
+    else:
+        raise WrongTypeException(
+            "iterator on non node or iterator type (%s)" % str(type(n))
+        )
+
+
+def filter(n: Node, query: str) -> CompatNodeIterator:
+    """
+    This function has the same signature as the pre-v3 filter() returning a
+    compatibility CompatNodeIterator.
+    """
+    ctx = uast()
+    return CompatNodeIterator(NodeIterator(ctx.filter(query, n.internal_node), ctx))
+
+
+def filter_nodes(n: Node, query: str) -> CompatNodeIterator:
+    """
+    Utility function. Same as filter() but will only filter for nodes (i. e.
+    it will exclude scalars and positions).
+    """
+    return CompatNodeIterator(filter(n, query)._nodeit, only_nodes=True)
+
+
+class TypedQueryException(Exception):
+    """
+    This exception will be raised when a query for a specific type (str, int, float...)
+    returns a different type of more than one result.
+    """
+    pass
+
+
+def _scalariter2item(n: Node, query: str, wanted_type: type) -> Any:
+    rlist = list(filter(n, query))
+
+    if len(rlist) > 1:
+        raise TypedQueryException("More than one result for %s typed query" % str(type))
+
+    value = rlist[0]
+    if isinstance(value, Node):
+        value = value.internal_node
+
+    value_type = type(value)
+    if wanted_type == float and value_type == int:
+        value = float(value)
+
+    if not isinstance(value, wanted_type):
+        raise TypedQueryException("Typed query for type %s returned type %s instead"
+                        % (str(wanted_type), str(type(value))))
+
+    return wanted_type(value)
+
+
+def filter_string(n: Node, query: str) -> str:
+    """
+    Filter and ensure that the returned value is of string type.
+    """
+    return _scalariter2item(n, query, str)
+
+
+def filter_bool(n: Node, query: str) -> bool:
+    """
+    Filter and ensure that the returned value is of type bool.
+    """
+    return _scalariter2item(n, query, bool)
+
+
+def filter_int(n: Node, query: str) -> int:
+    """
+    Filter and ensure that the returned value is of type int.
+    """
+    return _scalariter2item(n, query, int)
+
+
+def filter_float(n: Node, query: str) -> float:
+    """
+    Filter and ensure that the returned value is of type int.
+    """
+    return _scalariter2item(n, query, float)
+
+
+filter_number = filter_float