[BUG] writeToAzureSearch fails when the index has custom analyzers or tokenizers since 0.11.0

DBundred-cfc commented 10 months ago

SynapseML version

0.11.0-1.0.2

System information

Python Standalone - python 3.11.3, scala 2.12, Spark 3.4.1, SynapseML 1.0.2
Azure Synapse 3.3, SynapseML 0.11.4-spark3.3

Describe the problem

When a version of Synapse ML is used to load data into an index that has a custom analyzer or tokenizer (and possibly other custom objects but they haven't neem tested) it fails with the following error : -

Py4JJavaError: An error occurred while calling z:com.microsoft.azure.synapse.ml.services.search.AzureSearchWriter.write.
: spray.json.DeserializationException: Expected String as JsString, but got {"@odata.type":"#Microsoft.Azure.Search.CustomAnalyzer","charFilters":[],"name":"keyword_analyzer","tokenFilters":["lowercase"],"tokenizer":"keyword_v2"}

This happens with all apiVersions set and seemingly any version greater than 0.11.0. It works correctly against the same index when run using a Spark 3.2 Azure Synapse cluster, which uses Synapse ML version 0.10.2

Code to reproduce issue

Create an index with a custom analyzer

This needs to be done through the API: - POST https://{{service-name}}.search.windows.net/indexes?api-version={{api-version}}

{
  "name": "{{index-name}}",
  "defaultScoringProfile": null,
  "fields": [
    {
      "name": "Id",
      "type": "Edm.String",
      "searchable": false,
      "filterable": false,
      "retrievable": true,
      "sortable": false,
      "facetable": false,
      "key": true,
      "indexAnalyzer": null,
      "searchAnalyzer": null,
      "analyzer": null,
      "synonymMaps": []
    }
  ],
  "analyzers": [
    {
    "@odata.type": "#Microsoft.Azure.Search.CustomAnalyzer",
    "name": "keyword_analyzer",
    "tokenizer": "keyword_v2",
    "tokenFilters": ["lowercase"]
    }
    ]
}

Try and load the index

Run the following pyspark on a spark 3.4

from pyspark.sql import SparkSession
spark = SparkSession.builder.appName("MyApp") \
    .config("spark.jars.packages", "com.microsoft.azure:synapseml_2.12:1.0.2") \
    .getOrCreate()

import synapse.ml
from synapse.ml.services import writeToAzureSearch
from pyspark.sql.functions import lit, col

df = spark.range(10) \
    .withColumn("Id", col("id").cast("string")) \
    .withColumn("action", lit("upload"))

x = writeToAzureSearch(df, 
        subscriptionKey=admin_key,
        actionCol="action",
        serviceName=search_service,
        indexName=search_index,
        keyCol="Id")

You can also run the same code (without the spark creation) on Azure Synapse 3.3 and get the same result. I imagine this will happen on Databricks, and Synapse 3.4 but haven't tested it.

Other info / logs

---------------------------------------------------------------------------
Py4JJavaError                             Traceback (most recent call last)
c:\Users\dbundred\Documents\Projects\OSM\OtherStuff\synapseMlTest.ipynb Cell 2 line 1
     11 from pyspark.sql.functions import lit, col
     14 df = spark.range(10) \
     15     .withColumn("Id", col("id").cast("string")) \
     16     .withColumn("action", lit("upload"))
---> 18 x = writeToAzureSearch(df, 
     19         subscriptionKey=admin_key,
     20         actionCol="action",
     21         serviceName=search_service,
     22         indexName=search_index,
     23         keyCol="Id")

File ~\AppData\Local\Temp\spark-c09947f2-255d-45b5-a241-6a7165bbac06\userFiles-b3991b65-4cea-4012-980d-108b02406dcc\com.microsoft.azure_synapseml-cognitive_2.12-1.0.2.jar\synapse\ml\services\search\AzureSearchWriter.py:28, in writeToAzureSearch(df, **options)
     26 jvm = SparkContext.getOrCreate()._jvm
     27 writer = jvm.com.microsoft.azure.synapse.ml.services.search.AzureSearchWriter
---> 28 writer.write(df._jdf, options)

File c:\Users\dbundred\AppData\Local\Programs\Python\Python311\Lib\site-packages\py4j\java_gateway.py:1322, in JavaMember.__call__(self, *args)
   1316 command = proto.CALL_COMMAND_NAME +\
   1317     self.command_header +\
   1318     args_command +\
   1319     proto.END_COMMAND_PART
   1321 answer = self.gateway_client.send_command(command)
-> 1322 return_value = get_return_value(
   1323     answer, self.gateway_client, self.target_id, self.name)
   1325 for temp_arg in temp_args:
   1326     if hasattr(temp_arg, "_detach"):

File c:\Users\dbundred\AppData\Local\Programs\Python\Python311\Lib\site-packages\pyspark\errors\exceptions\captured.py:169, in capture_sql_exception.<locals>.deco(*a, **kw)
    167 def deco(*a: Any, **kw: Any) -> Any:
    168     try:
--> 169         return f(*a, **kw)
    170     except Py4JJavaError as e:
    171         converted = convert_exception(e.java_exception)

File c:\Users\dbundred\AppData\Local\Programs\Python\Python311\Lib\site-packages\py4j\protocol.py:326, in get_return_value(answer, gateway_client, target_id, name)
    324 value = OUTPUT_CONVERTER[type](answer[2:], gateway_client)
    325 if answer[1] == REFERENCE_TYPE:
--> 326     raise Py4JJavaError(
    327         "An error occurred while calling {0}{1}{2}.\n".
    328         format(target_id, ".", name), value)
    329 else:
    330     raise Py4JError(
    331         "An error occurred while calling {0}{1}{2}. Trace:\n{3}\n".
    332         format(target_id, ".", name, value))

Py4JJavaError: An error occurred while calling z:com.microsoft.azure.synapse.ml.services.search.AzureSearchWriter.write.
: spray.json.DeserializationException: Expected String as JsString, but got {"@odata.type":"#Microsoft.Azure.Search.CustomAnalyzer","charFilters":[],"name":"keyword_analyzer","tokenFilters":["lowercase"],"tokenizer":"keyword_v2"}
    at spray.json.package$.deserializationError(package.scala:23)
    at spray.json.ProductFormats.fromField(ProductFormats.scala:63)
    at spray.json.ProductFormats.fromField$(ProductFormats.scala:51)
    at com.microsoft.azure.synapse.ml.services.search.AzureSearchProtocol$.fromField(AzureSearchSchemas.scala:67)
    at spray.json.ProductFormatsInstances$$anon$11.read(ProductFormatsInstances.scala:341)
    at spray.json.ProductFormatsInstances$$anon$11.read(ProductFormatsInstances.scala:319)
    at spray.json.JsValue.convertTo(JsValue.scala:33)
    at com.microsoft.azure.synapse.ml.services.search.IndexParser.parseIndexJson(AzureSearchAPI.scala:25)
    at com.microsoft.azure.synapse.ml.services.search.IndexParser.parseIndexJson$(AzureSearchAPI.scala:24)
    at com.microsoft.azure.synapse.ml.services.search.AzureSearchWriter$.parseIndexJson(AzureSearch.scala:147)
    at com.microsoft.azure.synapse.ml.services.search.AzureSearchWriter$.getVectorColConf(AzureSearch.scala:325)
    at com.microsoft.azure.synapse.ml.services.search.AzureSearchWriter$.prepareDF(AzureSearch.scala:269)
    at com.microsoft.azure.synapse.ml.services.search.AzureSearchWriter$.write(AzureSearch.scala:432)
    at com.microsoft.azure.synapse.ml.services.search.AzureSearchWriter$.write(AzureSearch.scala:440)
    at com.microsoft.azure.synapse.ml.services.search.AzureSearchWriter.write(AzureSearch.scala)
    at java.base/jdk.internal.reflect.NativeMethodAccessorImpl.invoke0(Native Method)
    at java.base/jdk.internal.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java:77)
    at java.base/jdk.internal.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43)
    at java.base/java.lang.reflect.Method.invoke(Method.java:568)
    at py4j.reflection.MethodInvoker.invoke(MethodInvoker.java:244)
    at py4j.reflection.ReflectionEngine.invoke(ReflectionEngine.java:374)
    at py4j.Gateway.invoke(Gateway.java:282)
    at py4j.commands.AbstractCommand.invokeMethod(AbstractCommand.java:132)
    at py4j.commands.CallCommand.execute(CallCommand.java:79)
    at py4j.ClientServerConnection.waitForCommands(ClientServerConnection.java:182)
    at py4j.ClientServerConnection.run(ClientServerConnection.java:106)
    at java.base/java.lang.Thread.run(Thread.java:833)

What component(s) does this bug affect?

[X] area/cognitive: Cognitive project
[ ] area/core: Core project
[ ] area/deep-learning: DeepLearning project
[ ] area/lightgbm: Lightgbm project
[ ] area/opencv: Opencv project
[ ] area/vw: VW project
[ ] area/website: Website
[ ] area/build: Project build system
[ ] area/notebooks: Samples under notebooks folder
[ ] area/docker: Docker usage
[ ] area/models: models related issue

What language(s) does this bug affect?

[ ] language/scala: Scala source code
[X] language/python: Pyspark APIs
[ ] language/r: R APIs
[ ] language/csharp: .NET APIs
[ ] language/new: Proposals for new client languages

What integration(s) does this bug affect?

[X] integrations/synapse: Azure Synapse integrations
[ ] integrations/azureml: Azure ML integrations
[ ] integrations/databricks: Databricks integrations

github-actions[bot] commented 10 months ago

Hey @DBundred-cfc :wave:! Thank you so much for reporting the issue/feature request :rotating_light:. Someone from SynapseML Team will be looking to triage this issue soon. We appreciate your patience.

jelmewcrv commented 2 weeks ago

It is still present in the current version.

Spray seems to have a lot of issues with serializing/deserializing jsons. For example: spray.json.DeserializationException: Expected String as JsString, but got {"flags":null,"lowercase":true,"name":"patternAnalyzer","pattern":"\\s+","stopwords":[]}

microsoft / SynapseML