feat(tracing): add masking i/o docs (#310)

Will be released in [email protected], [email protected]

Author: dqbd

versioned_docs/version-2.0/how_to_guides/index.md

@@ -53,7 +53,7 @@ Get started with LangSmith's tracing features to start adding observability to y
   - [Manually provide token counts](./how_to_guides/tracing/log_llm_trace#manually-provide-token-counts)
   - [Instruct-style models](./how_to_guides/tracing/log_llm_trace#instruct-style-models)
 - [Prevent logging of sensitive data in traces](./how_to_guides/tracing/mask_inputs_outputs)
-  - [Example masking UUIDs and emails in inputs and outputs](./how_to_guides/tracing/mask_inputs_outputs#example-masking-uuids-and-emails-in-inputs-and-outputs)
+  - [Rule-based masking of inputs and outputs](./how_to_guides/tracing/mask_inputs_outputs#rule-based-masking-of-inputs-and-outputs)
 - [Export traces](./how_to_guides/tracing/export_traces)
   - [Use filter arguments](./how_to_guides/tracing/export_traces#use-filter-arguments)
   - [Use filter query language](./how_to_guides/tracing/export_traces#use-filter-query-language)

versioned_docs/version-2.0/how_to_guides/tracing/mask_inputs_outputs.mdx

@@ -6,6 +6,8 @@ import {
   CodeTabs,
   PythonBlock,
   TypeScriptBlock,
+  python,
+  typescript,
 } from "@site/src/components/InstructionsWithCode";
 
 # Prevent logging of sensitive data in traces
@@ -84,18 +86,79 @@ await openaiClient.chat.completions.create({
   groupId="client-language"
 />
 
-## Example masking UUIDs and emails in inputs and outputs
+## Rule-based masking of inputs and outputs
 
-You can also customize the `hide_inputs` and `hide_outputs` functions to mask specific data in the inputs and outputs. For example, you can mask UUIDs and emails in the inputs and outputs.
+:::info
+This feature is available in the following LangSmith SDK versions:
+
+- Python: 0.1.81 and above
+- TypeScript: 0.1.33 and above
+
+:::
+
+To mask specific data in inputs and outputs, you can use the `create_anonymizer` / `createAnonymizer` function and pass the newly created anonymizer when instantiating the client. The anonymizer can be either constructed from a list of regex patterns and the replacement values or from a function that accepts and returns a string value.
+
+The anonymizer will be skipped for inputs if `LANGCHAIN_HIDE_INPUTS = true`. Same applies for outputs if `LANGCHAIN_HIDE_OUTPUTS = true`.
+
+However, if inputs or outputs are to be sent to client, the `anonymizer` method will take precedence over functions found in `hide_inputs` and `hide_outputs`. By default, the `create_anonymizer` will only look at maximum of 10 nesting levels deep, which can be configured via the `max_depth` parameter.
+
+<CodeTabs
+  tabs={[
+    python`
+        from langsmith.anonymizer import create_anonymizer
+        from langsmith import Client, traceable
+        
+        # create anonymizer from list of regex patterns and replacement values
+        anonymizer = create_anonymizer([
+            { "pattern": r"[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\.[a-zA-Z]{2,}", "replace": "<email>" },
+            { "pattern": r"[0-9a-fA-F]{8}-[0-9a-fA-F]{4}-[0-9a-fA-F]{4}-[0-9a-fA-F]{4}-[0-9a-fA-F]{12}", "replace": "<uuid>" }
+        ])
+        
+        # or create anonymizer from a function
+        anonymizer = create_anonymizer(lambda text: r"...".sub("[value]", text))
+        
+        client = Client(anonymizer=anonymizer)
+        
+        @traceable(client=client)
+        def main(inputs: dict) -> dict:
+            ...
+        `,
+    typescript`
+        import { createAnonymizer } from "langsmith/anonymizer"
+        import { traceable } from "langsmith/traceable"
+        import { Client } from "langsmith"
+        
+        // create anonymizer from list of regex patterns and replacement values
+        const anonymizer = createAnonymizer([
+            { pattern: /[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\.[a-zA-Z]{2,}/g, replace: "<email>" },
+            { pattern: /[0-9a-fA-F]{8}-[0-9a-fA-F]{4}-[0-9a-fA-F]{4}-[0-9a-fA-F]{4}-[0-9a-fA-F]{12}/g, replace: "<uuid>" }
+        ])
+        
+        // or create anonymizer from a function
+        const anonymizer = createAnonymizer((value) => value.replace("...", "<value>"))
+        
+        const client = new Client({ anonymizer })
+        
+        const main = traceable(async (inputs: any) => {
+            // ...
+        }, { client })
+    `,
+  ]}
+  groupId="client-language"
+/>
+
+Please note, that using the anonymizer might incur a performance hit with complex regular expressions or large payloads, as the anonymizer serializes the payload to JSON before processing.
 
 :::note
 
-We are in the process of adding more built-in filters for common data types. If you have a specific use case that you would like us to support, please contact [email protected].
+Improving the performance of `anonymizer` API is on our roadmap! If you are encountering performance issues, please contact us at [email protected].
 
 :::
 
 ![](../static/hide_inputs_outputs.png)
 
+Older versions of LangSmith SDKs can use the `hide_inputs` and `hide_outputs` parameters to achieve the same effect. You can also use these parameters to process the inputs and outputs more efficiently as well.
+
 <CodeTabs
   tabs={[
     PythonBlock(`import re