Codegen: Improve Formatter for better field name handling (#5534)

CLOSE
https://linear.app/sourcegraph/issue/CODY-3696/codegen-generated-invalid-field-name-causing-jetbrains-build-to-fail

This PR fixes the issue where codegen does not escape all the invalid
characters for field names.

- Replace non-alphanumeric characters with underscores for Kotlin field
names
- Add helper function `getEscapedValue` to handle various field name
formatting rules

Also removed the bindings from the
agent/bindings/kotlin/lib/bin/main/com/sourcegraph/cody/agent/protocol_generated
directory

## Test plan

<!-- Required. See
https://docs-legacy.sourcegraph.com/dev/background-information/testing_principles.
-->

Run the generate agent bindings command to confirm everything works as
expected.
Some before and after comparison:
before:
const val `recently used` = "recently used"
const val `https__/example.com` = "https://example.com"

after:
const val `recently-used` = "recently used"
const val `https-example-com` = "https://example.com"


## Changelog

<!-- OPTIONAL; info at
https://www.notion.so/sourcegraph/Writing-a-changelog-entry-dd997f411d524caabf0d8d38a24a878c
-->

agent/bindings/kotlin/lib/src/main/kotlin/com/sourcegraph/cody/agent/protocol_generated/Constants.kt

@@ -60,7 +60,7 @@ object Constants {
   const val openctx = "openctx"
   const val power = "power"
   const val pro = "pro"
-  const val `recently used` = "recently used"
+  const val `recently-used` = "recently used"
   const val recommended = "recommended"
   const val `rename-file` = "rename-file"
   const val replace = "replace"

agent/src/cli/scip-codegen/Formatter.ts

@@ -265,10 +265,13 @@ export class Formatter {
         if (this.language === TargetLanguage.Kotlin) {
             const isKeyword = this.kotlinKeywords.has(escaped)
             const needsBacktick = isKeyword || !/^[a-zA-Z0-9_]+$/.test(escaped)
-            return needsBacktick ? `\`${escaped}\`` : escaped
+            // Replace all non-alphanumeric characters with underscores
+            const fieldName = getEscapedValue(escaped, '-')
+            return needsBacktick ? `\`${fieldName}\`` : fieldName
         }
+        // CSharp
         if (this.language === TargetLanguage.CSharp) {
-            return escaped
+            return getEscapedValue(escaped)
                 .split('_')
                 .map(part => part.charAt(0).toUpperCase() + part.slice(1))
                 .join('')
@@ -297,3 +300,14 @@ export class Formatter {
         return `${capitalize(name)}Enum`
     }
 }
+
+function getEscapedValue(name: string, replacer: '_' | '-' = '_'): string {
+    const nonAlphanumericRegex = replacer === '-' ? /[^a-zA-Z0-9]+/g : /[^a-zA-Z0-9]/g
+    const repeatedReplacerRegex = new RegExp(`${replacer}+`, 'g')
+    const trimReplacerRegex = new RegExp(`^${replacer}|${replacer}$`, 'g')
+
+    return name
+        .replace(nonAlphanumericRegex, replacer)
+        .replace(repeatedReplacerRegex, replacer)
+        .replace(trimReplacerRegex, '')
+}