|
| 1 | +import avro.schema |
| 2 | +import requests |
| 3 | + |
| 4 | +from avro.errors import AvroException |
| 5 | +from avro.io import BinaryDecoder, BinaryEncoder, DatumReader, DatumWriter |
| 6 | +from io import BytesIO |
| 7 | +from nypl_py_utils.functions.log_helper import create_log |
| 8 | +from requests.adapters import HTTPAdapter, Retry |
| 9 | +from requests.exceptions import JSONDecodeError |
| 10 | + |
| 11 | + |
| 12 | +class AvroClient: |
| 13 | + """ |
| 14 | + Base class for Avro schema interaction. Takes as input the |
| 15 | + Platform API endpoint from which to fetch the schema in JSON format. |
| 16 | + """ |
| 17 | + |
| 18 | + def __init__(self, platform_schema_url): |
| 19 | + self.logger = create_log("avro_client") |
| 20 | + retry_policy = Retry(total=3, backoff_factor=45, |
| 21 | + status_forcelist=[500, 502, 503, 504], |
| 22 | + allowed_methods=frozenset(['GET'])) |
| 23 | + self.session = requests.Session() |
| 24 | + self.session.mount("https://", |
| 25 | + HTTPAdapter(max_retries=retry_policy)) |
| 26 | + self.schema = avro.schema.parse( |
| 27 | + self.get_json_schema(platform_schema_url)) |
| 28 | + |
| 29 | + def get_json_schema(self, platform_schema_url): |
| 30 | + """ |
| 31 | + Fetches a JSON response from the input Platform API endpoint and |
| 32 | + interprets it as an Avro schema. |
| 33 | + """ |
| 34 | + self.logger.info( |
| 35 | + "Fetching Avro schema from {}".format(platform_schema_url)) |
| 36 | + try: |
| 37 | + |
| 38 | + response = self.session.get(url=platform_schema_url, |
| 39 | + timeout=60) |
| 40 | + response.raise_for_status() |
| 41 | + except Exception as e: |
| 42 | + self.logger.error( |
| 43 | + "Failed to retrieve schema from {url}: {error}".format( |
| 44 | + url=platform_schema_url, error=e |
| 45 | + ) |
| 46 | + ) |
| 47 | + raise AvroClientError( |
| 48 | + "Failed to retrieve schema from {url}: {error}".format( |
| 49 | + url=platform_schema_url, error=e |
| 50 | + ) |
| 51 | + ) |
| 52 | + |
| 53 | + try: |
| 54 | + json_response = response.json() |
| 55 | + return json_response["data"]["schema"] |
| 56 | + except (JSONDecodeError, KeyError) as e: |
| 57 | + self.logger.error( |
| 58 | + "Retrieved schema is malformed: {errorType} {errorMessage}" |
| 59 | + .format(errorType=type(e), errorMessage=e) |
| 60 | + ) |
| 61 | + raise AvroClientError( |
| 62 | + "Retrieved schema is malformed: {errorType} {errorMessage}" |
| 63 | + .format(errorType=type(e), errorMessage=e) |
| 64 | + ) from None |
| 65 | + |
| 66 | + |
| 67 | +class AvroEncoder(AvroClient): |
| 68 | + """ |
| 69 | + Class for encoding records using an Avro schema. Takes as input the |
| 70 | + Platform API endpoint from which to fetch the schema in JSON format. |
| 71 | + """ |
| 72 | + |
| 73 | + def encode_record(self, record): |
| 74 | + """ |
| 75 | + Encodes a single JSON record using the given Avro schema. |
| 76 | +
|
| 77 | + Returns the encoded record as a byte string. |
| 78 | + """ |
| 79 | + self.logger.debug( |
| 80 | + "Encoding record using {schema} schema".format( |
| 81 | + schema=self.schema.name) |
| 82 | + ) |
| 83 | + datum_writer = DatumWriter(self.schema) |
| 84 | + with BytesIO() as output_stream: |
| 85 | + encoder = BinaryEncoder(output_stream) |
| 86 | + try: |
| 87 | + datum_writer.write(record, encoder) |
| 88 | + return output_stream.getvalue() |
| 89 | + except AvroException as e: |
| 90 | + self.logger.error("Failed to encode record: {}".format(e)) |
| 91 | + raise AvroClientError( |
| 92 | + "Failed to encode record: {}".format(e)) from None |
| 93 | + |
| 94 | + def encode_batch(self, record_list): |
| 95 | + """ |
| 96 | + Encodes a list of JSON records using the given Avro schema. |
| 97 | +
|
| 98 | + Returns a list of byte strings where each string is an encoded record. |
| 99 | + """ |
| 100 | + self.logger.info( |
| 101 | + "Encoding ({num_rec}) records using {schema} schema".format( |
| 102 | + num_rec=len(record_list), schema=self.schema.name |
| 103 | + ) |
| 104 | + ) |
| 105 | + encoded_records = [] |
| 106 | + datum_writer = DatumWriter(self.schema) |
| 107 | + with BytesIO() as output_stream: |
| 108 | + encoder = BinaryEncoder(output_stream) |
| 109 | + for record in record_list: |
| 110 | + try: |
| 111 | + datum_writer.write(record, encoder) |
| 112 | + encoded_records.append(output_stream.getvalue()) |
| 113 | + output_stream.seek(0) |
| 114 | + output_stream.truncate(0) |
| 115 | + except AvroException as e: |
| 116 | + self.logger.error("Failed to encode record: {}".format(e)) |
| 117 | + raise AvroClientError( |
| 118 | + "Failed to encode record: {}".format(e) |
| 119 | + ) from None |
| 120 | + return encoded_records |
| 121 | + |
| 122 | + |
| 123 | +class AvroDecoder(AvroClient): |
| 124 | + """ |
| 125 | + Class for decoding records using an Avro schema. Takes as input the |
| 126 | + Platform API endpoint from which to fetch the schema in JSON format. |
| 127 | + """ |
| 128 | + |
| 129 | + def decode_record(self, record): |
| 130 | + """ |
| 131 | + Decodes a single record represented using the given Avro |
| 132 | + schema. Input must be a bytes-like object. |
| 133 | +
|
| 134 | + Returns a dictionary where each key is a field in the schema. |
| 135 | + """ |
| 136 | + self.logger.info( |
| 137 | + "Decoding {rec} using {schema} schema".format( |
| 138 | + rec=record, schema=self.schema.name |
| 139 | + ) |
| 140 | + ) |
| 141 | + datum_reader = DatumReader(self.schema) |
| 142 | + with BytesIO(record) as input_stream: |
| 143 | + decoder = BinaryDecoder(input_stream) |
| 144 | + try: |
| 145 | + return datum_reader.read(decoder) |
| 146 | + except Exception as e: |
| 147 | + self.logger.error("Failed to decode record: {}".format(e)) |
| 148 | + raise AvroClientError( |
| 149 | + "Failed to decode record: {}".format(e)) from None |
| 150 | + |
| 151 | + def decode_batch(self, record_list): |
| 152 | + """ |
| 153 | + Decodes a list of JSON records using the given Avro schema. Input |
| 154 | + must be a list of bytes-like objects. |
| 155 | +
|
| 156 | + Returns a list of strings where each string is a decoded record. |
| 157 | + """ |
| 158 | + self.logger.info( |
| 159 | + "Decoding ({num_rec}) records using {schema} schema".format( |
| 160 | + num_rec=len(record_list), schema=self.schema.name |
| 161 | + ) |
| 162 | + ) |
| 163 | + decoded_records = [] |
| 164 | + for record in record_list: |
| 165 | + decoded_record = self.decode_record(record) |
| 166 | + decoded_records.append(decoded_record) |
| 167 | + return decoded_records |
| 168 | + |
| 169 | + |
| 170 | +class AvroClientError(Exception): |
| 171 | + def __init__(self, message=None): |
| 172 | + self.message = message |
0 commit comments