airbyte-cdk 6.18.2__tar.gz → 6.19.0__tar.gz

{airbyte_cdk-6.18.2 → airbyte_cdk-6.19.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.3
 Name: airbyte-cdk
-Version: 6.18.2
+Version: 6.19.0
 Summary: A framework for writing Airbyte Connectors.
 License: MIT
 Keywords: airbyte,connector-development-kit,cdk

{airbyte_cdk-6.18.2 → airbyte_cdk-6.19.0}/airbyte_cdk/sources/declarative/declarative_component_schema.yaml

@@ -678,7 +678,7 @@ definitions:
     properties:
       type:
         type: string
-        enum: [ CustomSchemaNormalization ]
+        enum: [CustomSchemaNormalization]
       class_name:
         title: Class Name
         description: Fully-qualified name of the class that will be implementing the custom normalization. The format is `source_<name>.<package>.<class_name>`.
@@ -2886,6 +2886,7 @@ definitions:
       parser:
         anyOf:
           - "$ref": "#/definitions/GzipParser"
+          - "$ref": "#/definitions/JsonParser"
           - "$ref": "#/definitions/JsonLineParser"
           - "$ref": "#/definitions/CsvParser"
   #    PARSERS
@@ -2902,6 +2903,20 @@ definitions:
         anyOf:
           - "$ref": "#/definitions/JsonLineParser"
           - "$ref": "#/definitions/CsvParser"
+          - "$ref": "#/definitions/JsonParser"
+  JsonParser:
+    title: JsonParser
+    description: Parser used for parsing str, bytes, or bytearray data and returning data in a dictionary format.
+    type: object
+    required:
+      - type
+    properties:
+      type:
+        type: string
+        enum: [JsonParser]
+      encoding:
+        type: string
+        default: utf-8
   JsonLineParser:
     type: object
     required:

{airbyte_cdk-6.18.2 → airbyte_cdk-6.19.0}/airbyte_cdk/sources/declarative/decoders/composite_raw_decoder.py

@@ -7,9 +7,12 @@ from dataclasses import dataclass
 from io import BufferedIOBase, TextIOWrapper
 from typing import Any, Generator, MutableMapping, Optional
 
+import orjson
 import requests
 
+from airbyte_cdk.models import FailureType
 from airbyte_cdk.sources.declarative.decoders.decoder import Decoder
+from airbyte_cdk.utils import AirbyteTracedException
 
 logger = logging.getLogger("airbyte")
 
@@ -42,6 +45,46 @@ class GzipParser(Parser):
             yield from self.inner_parser.parse(gzipobj)
 
 
+@dataclass
+class JsonParser(Parser):
+    encoding: str = "utf-8"
+
+    def parse(self, data: BufferedIOBase) -> Generator[MutableMapping[str, Any], None, None]:
+        """
+        Attempts to deserialize data using orjson library. As an extra layer of safety we fallback on the json library to deserialize the data.
+        """
+        raw_data = data.read()
+        body_json = self._parse_orjson(raw_data) or self._parse_json(raw_data)
+
+        if body_json is None:
+            raise AirbyteTracedException(
+                message="Response JSON data failed to be parsed. See logs for more information.",
+                internal_message=f"Response JSON data failed to be parsed.",
+                failure_type=FailureType.system_error,
+            )
+
+        if isinstance(body_json, list):
+            yield from body_json
+        else:
+            yield from [body_json]
+
+    def _parse_orjson(self, raw_data: bytes) -> Optional[Any]:
+        try:
+            return orjson.loads(raw_data.decode(self.encoding))
+        except Exception as exc:
+            logger.debug(
+                f"Failed to parse JSON data using orjson library. Falling back to json library. {exc}"
+            )
+            return None
+
+    def _parse_json(self, raw_data: bytes) -> Optional[Any]:
+        try:
+            return json.loads(raw_data.decode(self.encoding))
+        except Exception as exc:
+            logger.error(f"Failed to parse JSON data using json library. {exc}")
+            return None
+
+
 @dataclass
 class JsonLineParser(Parser):
     encoding: Optional[str] = "utf-8"

{airbyte_cdk-6.18.2 → airbyte_cdk-6.19.0}/airbyte_cdk/sources/declarative/models/declarative_component_schema.py

@@ -1201,6 +1201,14 @@ class LegacySessionTokenAuthenticator(BaseModel):
     parameters: Optional[Dict[str, Any]] = Field(None, alias="$parameters")
 
 
+class JsonParser(BaseModel):
+    class Config:
+        extra = Extra.allow
+
+    type: Literal["JsonParser"]
+    encoding: Optional[str] = "utf-8"
+
+
 class JsonLineParser(BaseModel):
     type: Literal["JsonLineParser"]
     encoding: Optional[str] = "utf-8"
@@ -1599,7 +1607,7 @@ class RecordSelector(BaseModel):
 
 class GzipParser(BaseModel):
     type: Literal["GzipParser"]
-    inner_parser: Union[JsonLineParser, CsvParser]
+    inner_parser: Union[JsonLineParser, CsvParser, JsonParser]
 
 
 class Spec(BaseModel):
@@ -1634,7 +1642,7 @@ class CompositeErrorHandler(BaseModel):
 
 class CompositeRawDecoder(BaseModel):
     type: Literal["CompositeRawDecoder"]
-    parser: Union[GzipParser, JsonLineParser, CsvParser]
+    parser: Union[GzipParser, JsonParser, JsonLineParser, CsvParser]
 
 
 class DeclarativeSource1(BaseModel):

{airbyte_cdk-6.18.2 → airbyte_cdk-6.19.0}/airbyte_cdk/sources/declarative/parsers/model_to_component_factory.py

@@ -72,6 +72,8 @@ from airbyte_cdk.sources.declarative.decoders.composite_raw_decoder import (
     CsvParser,
     GzipParser,
     JsonLineParser,
+    JsonParser,
+    Parser,
 )
 from airbyte_cdk.sources.declarative.extractors import (
     DpathExtractor,
@@ -247,6 +249,9 @@ from airbyte_cdk.sources.declarative.models.declarative_component_schema import
 from airbyte_cdk.sources.declarative.models.declarative_component_schema import (
     JsonLineParser as JsonLineParserModel,
 )
+from airbyte_cdk.sources.declarative.models.declarative_component_schema import (
+    JsonParser as JsonParserModel,
+)
 from airbyte_cdk.sources.declarative.models.declarative_component_schema import (
     JwtAuthenticator as JwtAuthenticatorModel,
 )
@@ -522,6 +527,7 @@ class ModelToComponentFactory:
             JsonDecoderModel: self.create_json_decoder,
             JsonlDecoderModel: self.create_jsonl_decoder,
             JsonLineParserModel: self.create_json_line_parser,
+            JsonParserModel: self.create_json_parser,
             GzipJsonDecoderModel: self.create_gzipjson_decoder,
             GzipParserModel: self.create_gzip_parser,
             KeysToLowerModel: self.create_keys_to_lower_transformation,
@@ -1032,17 +1038,17 @@ class ModelToComponentFactory:
         self, model: CursorPaginationModel, config: Config, decoder: Decoder, **kwargs: Any
     ) -> CursorPaginationStrategy:
         if isinstance(decoder, PaginationDecoderDecorator):
-            if not isinstance(decoder.decoder, (JsonDecoder, XmlDecoder)):
-                raise ValueError(
-                    f"Provided decoder of {type(decoder.decoder)=} is not supported. Please set JsonDecoder or XmlDecoder instead."
-                )
+            inner_decoder = decoder.decoder
+        else:
+            inner_decoder = decoder
+            decoder = PaginationDecoderDecorator(decoder=decoder)
+
+        if self._is_supported_decoder_for_pagination(inner_decoder):
             decoder_to_use = decoder
         else:
-            if not isinstance(decoder, (JsonDecoder, XmlDecoder)):
-                raise ValueError(
-                    f"Provided decoder of {type(decoder)=} is not supported. Please set JsonDecoder or XmlDecoder instead."
-                )
-            decoder_to_use = PaginationDecoderDecorator(decoder=decoder)
+            raise ValueError(
+                self._UNSUPPORTED_DECODER_ERROR.format(decoder_type=type(inner_decoder))
+            )
 
         return CursorPaginationStrategy(
             cursor_value=model.cursor_value,
@@ -1515,11 +1521,10 @@ class ModelToComponentFactory:
         cursor_used_for_stop_condition: Optional[DeclarativeCursor] = None,
     ) -> Union[DefaultPaginator, PaginatorTestReadDecorator]:
         if decoder:
-            if not isinstance(decoder, (JsonDecoder, XmlDecoder)):
-                raise ValueError(
-                    f"Provided decoder of {type(decoder)=} is not supported. Please set JsonDecoder or XmlDecoder instead."
-                )
-            decoder_to_use = PaginationDecoderDecorator(decoder=decoder)
+            if self._is_supported_decoder_for_pagination(decoder):
+                decoder_to_use = PaginationDecoderDecorator(decoder=decoder)
+            else:
+                raise ValueError(self._UNSUPPORTED_DECODER_ERROR.format(decoder_type=type(decoder)))
         else:
             decoder_to_use = PaginationDecoderDecorator(decoder=JsonDecoder(parameters={}))
         page_size_option = (
@@ -1748,6 +1753,11 @@ class ModelToComponentFactory:
     def create_json_decoder(model: JsonDecoderModel, config: Config, **kwargs: Any) -> JsonDecoder:
         return JsonDecoder(parameters={})
 
+    @staticmethod
+    def create_json_parser(model: JsonParserModel, config: Config, **kwargs: Any) -> JsonParser:
+        encoding = model.encoding if model.encoding else "utf-8"
+        return JsonParser(encoding=encoding)
+
     @staticmethod
     def create_jsonl_decoder(
         model: JsonlDecoderModel, config: Config, **kwargs: Any
@@ -1940,22 +1950,22 @@ class ModelToComponentFactory:
             message_repository=self._message_repository,
         )
 
-    @staticmethod
     def create_offset_increment(
-        model: OffsetIncrementModel, config: Config, decoder: Decoder, **kwargs: Any
+        self, model: OffsetIncrementModel, config: Config, decoder: Decoder, **kwargs: Any
     ) -> OffsetIncrement:
         if isinstance(decoder, PaginationDecoderDecorator):
-            if not isinstance(decoder.decoder, (JsonDecoder, XmlDecoder)):
-                raise ValueError(
-                    f"Provided decoder of {type(decoder.decoder)=} is not supported. Please set JsonDecoder or XmlDecoder instead."
-                )
+            inner_decoder = decoder.decoder
+        else:
+            inner_decoder = decoder
+            decoder = PaginationDecoderDecorator(decoder=decoder)
+
+        if self._is_supported_decoder_for_pagination(inner_decoder):
             decoder_to_use = decoder
         else:
-            if not isinstance(decoder, (JsonDecoder, XmlDecoder)):
-                raise ValueError(
-                    f"Provided decoder of {type(decoder)=} is not supported. Please set JsonDecoder or XmlDecoder instead."
-                )
-            decoder_to_use = PaginationDecoderDecorator(decoder=decoder)
+            raise ValueError(
+                self._UNSUPPORTED_DECODER_ERROR.format(decoder_type=type(inner_decoder))
+            )
+
         return OffsetIncrement(
             page_size=model.page_size,
             config=config,
@@ -2555,3 +2565,25 @@ class ModelToComponentFactory:
             components_mapping=components_mapping,
             parameters=model.parameters or {},
         )
+
+    _UNSUPPORTED_DECODER_ERROR = (
+        "Specified decoder of {decoder_type} is not supported for pagination."
+        "Please set as `JsonDecoder`, `XmlDecoder`, or a `CompositeRawDecoder` with an inner_parser of `JsonParser` or `GzipParser` instead."
+        "If using `GzipParser`, please ensure that the lowest level inner_parser is a `JsonParser`."
+    )
+
+    def _is_supported_decoder_for_pagination(self, decoder: Decoder) -> bool:
+        if isinstance(decoder, (JsonDecoder, XmlDecoder)):
+            return True
+        elif isinstance(decoder, CompositeRawDecoder):
+            return self._is_supported_parser_for_pagination(decoder.parser)
+        else:
+            return False
+
+    def _is_supported_parser_for_pagination(self, parser: Parser) -> bool:
+        if isinstance(parser, JsonParser):
+            return True
+        elif isinstance(parser, GzipParser):
+            return isinstance(parser.inner_parser, JsonParser)
+        else:
+            return False

{airbyte_cdk-6.18.2 → airbyte_cdk-6.19.0}/pyproject.toml

@@ -23,7 +23,7 @@ classifiers = [
 keywords = ["airbyte", "connector-development-kit", "cdk"]
 
 # Python CDK uses dynamic versioning: https://github.com/mtkennerly/poetry-dynamic-versioning
-version = "6.18.2"  # Version will be calculated dynamically.
+version = "6.19.0"  # Version will be calculated dynamically.
 
 [tool.poetry-dynamic-versioning]
 enable = false