Snake case naming optimization (with vendor code from `@azure-tools/c…

…odegen`) (#2810)

* init

* debug for vitest

* fix

* regenerate

* fix test

* fix

* fix ci

.chronus/changes/snake-case-naming-optimization-inline-2024-8-3-11-0-30.md

@@ -0,0 +1,7 @@
+---
+changeKind: feature
+packages:
+  - "@azure-tools/typespec-python"
+---
+
+Optimize snake-case naming rule

.vscode/launch.json

@@ -52,6 +52,15 @@
             ]
         },
         {
+          "type": "node",
+          "request": "launch",
+          "name": "Debug Current Test File",
+          "autoAttachChildProcesses": true,
+          "skipFiles": ["<node_internals>/**"],
+          "program": "${workspaceFolder}/packages/typespec-python/node_modules/vitest/vitest.mjs",
+          "args": ["run", "${relativeFile}"],
+          "smartStep": true
+        },
           "name": "npm: regenerate",
           "request": "launch",
           "type": "node",

packages/typespec-python/package.json

@@ -36,7 +36,7 @@
     "install": "tsx ./scripts/run-python3.ts ./scripts/install.py",
     "prepare": "tsx ./scripts/run-python3.ts ./scripts/prepare.py",
     "regenerate": "tsx ./scripts/eng/regenerate.ts",
-    "test": "tsx ./scripts/eng/run-tests.ts"
+    "test": "npx vitest run ./test && tsx ./scripts/eng/run-tests.ts"
   },
   "files": [
     "dist/**",
@@ -74,8 +74,9 @@
     "@typespec/eslint-config-typespec": "~0.55.0",
     "@typespec/openapi": "~0.59.0",
     "c8": "~7.13.0",
+    "vitest": "^2.0.4",
     "rimraf": "~5.0.0",
-    "typescript": "~5.1.3",
+    "typescript": "~5.5.4",
     "@azure-tools/typespec-azure-core": "~0.45.0",
     "@azure-tools/typespec-client-generator-core": "0.45.4",
     "@typespec/compiler": "~0.59.1",

packages/typespec-python/src/utils.ts

@@ -13,15 +13,80 @@ import { getSimpleTypeResult, getType } from "./types.js";
 import { getNamespaceFullName } from "@typespec/compiler";
 import { PythonSdkContext } from "./lib.js";
 
+function IsFullyUpperCase(identifier: string, maxUppercasePreserve: number) {
+    const len = identifier.length;
+    if (len > 1) {
+        if (len <= maxUppercasePreserve && identifier === identifier.toUpperCase()) {
+            return true;
+        }
+
+        if (len <= maxUppercasePreserve + 1 && identifier.endsWith("s")) {
+            const i = identifier.substring(0, len - 1);
+            if (i.toUpperCase() === i) {
+                return true;
+            }
+        }
+    }
+    return false;
+}
+
+function deconstruct(identifier: string | Array<string>, maxUppercasePreserve: number): Array<string> {
+    if (Array.isArray(identifier)) {
+        return [...identifier.flatMap((each) => deconstruct(each, maxUppercasePreserve))];
+    }
+
+    return `${identifier}`
+        .replace(/([a-z]+)([A-Z])/g, "$1 $2") // Add a space in between camelCase words(e.g. fooBar => foo Bar)
+        .replace(/(\d+)/g, " $1 ") // Adds a space after numbers(e.g. foo123 => foo123 bar)
+        .replace(/\b([A-Z]+)([A-Z])s([^a-z])(.*)/g, "$1$2« $3$4") // Add a space after a plural uppper cased word(e.g. MBsFoo => MBs Foo)
+        .replace(/\b([A-Z]+)([A-Z])([a-z]+)/g, "$1 $2$3") // Add a space between an upper case word(2 char+) and the last captial case.(e.g. SQLConnection -> SQL Connection)
+        .replace(/«/g, "s")
+        .trim()
+        .split(/[\W|_]+/)
+        .map((each) => (IsFullyUpperCase(each, maxUppercasePreserve) ? each : each.toLowerCase()));
+}
+
+function isEqual(s1: string, s2: string): boolean {
+    // when s2 is undefined and s1 is the string 'undefined', it returns 0, making this true.
+    // To prevent that, first we need to check if s2 is undefined.
+    return s2 !== undefined && !!s1 && !s1.localeCompare(s2, undefined, { sensitivity: "base" });
+}
+
+function removeSequentialDuplicates(identifier: Iterable<string>) {
+    const ids = [...identifier].filter((each) => !!each);
+    for (let i = 0; i < ids.length; i++) {
+        while (isEqual(ids[i], ids[i - 1])) {
+            ids.splice(i, 1);
+        }
+        while (isEqual(ids[i], ids[i - 2]) && isEqual(ids[i + 1], ids[i - 1])) {
+            ids.splice(i, 2);
+        }
+    }
+
+    return ids;
+}
+
+function normalize(
+    identifier: string | Array<string>,
+    removeDuplicates = true,
+    maxUppercasePreserve = 0,
+): Array<string> {
+    if (!identifier || identifier.length === 0) {
+        return [""];
+    }
+    return typeof identifier === "string"
+        ? normalize(deconstruct(identifier, maxUppercasePreserve), removeDuplicates, maxUppercasePreserve)
+        : removeDuplicates
+        ? removeSequentialDuplicates(identifier)
+        : identifier;
+}
+
 export function camelToSnakeCase(name: string): string {
     if (!name) return name;
-    const camelToSnakeCaseRe = (str: string) =>
-        str
-            .replace(/[^a-zA-Z0-9]/g, "_")
-            .replace(/[A-Z]/g, (letter) => `_${letter.toLowerCase()}`)
-            .replace(/_+/g, "_");
-
-    return camelToSnakeCaseRe(name[0].toLowerCase() + name.slice(1));
+    const words = normalize(name, false, 6);
+    const result = words.join("_").toLowerCase();
+    const result_final = result.replace(/([^\d])_(\d+)/g, "$1$2");
+    return result_final;
 }
 
 export function removeUnderscoresFromNamespace(name?: string): string {

packages/typespec-python/test/azure/generated/encode-bytes/apiview_mapping_python.json

@@ -7,25 +7,25 @@
         "encode.bytes.models.DefaultBytesProperty": "Encode.Bytes.DefaultBytesProperty",
         "encode.bytes.BytesClient.query.default": "Encode.Bytes.Query.default",
         "encode.bytes.BytesClient.query.base64": "Encode.Bytes.Query.base64",
-        "encode.bytes.BytesClient.query.base64url": "Encode.Bytes.Query.base64url",
-        "encode.bytes.BytesClient.query.base64url_array": "Encode.Bytes.Query.base64urlArray",
+        "encode.bytes.BytesClient.query.base64_url": "Encode.Bytes.Query.base64url",
+        "encode.bytes.BytesClient.query.base64_url_array": "Encode.Bytes.Query.base64urlArray",
         "encode.bytes.BytesClient.property.default": "Encode.Bytes.Property.default",
         "encode.bytes.BytesClient.property.base64": "Encode.Bytes.Property.base64",
-        "encode.bytes.BytesClient.property.base64url": "Encode.Bytes.Property.base64url",
-        "encode.bytes.BytesClient.property.base64url_array": "Encode.Bytes.Property.base64urlArray",
+        "encode.bytes.BytesClient.property.base64_url": "Encode.Bytes.Property.base64url",
+        "encode.bytes.BytesClient.property.base64_url_array": "Encode.Bytes.Property.base64urlArray",
         "encode.bytes.BytesClient.header.default": "Encode.Bytes.Header.default",
         "encode.bytes.BytesClient.header.base64": "Encode.Bytes.Header.base64",
-        "encode.bytes.BytesClient.header.base64url": "Encode.Bytes.Header.base64url",
-        "encode.bytes.BytesClient.header.base64url_array": "Encode.Bytes.Header.base64urlArray",
+        "encode.bytes.BytesClient.header.base64_url": "Encode.Bytes.Header.base64url",
+        "encode.bytes.BytesClient.header.base64_url_array": "Encode.Bytes.Header.base64urlArray",
         "encode.bytes.BytesClient.request_body.default": "Encode.Bytes.RequestBody.default",
         "encode.bytes.BytesClient.request_body.octet_stream": "Encode.Bytes.RequestBody.octetStream",
         "encode.bytes.BytesClient.request_body.custom_content_type": "Encode.Bytes.RequestBody.customContentType",
         "encode.bytes.BytesClient.request_body.base64": "Encode.Bytes.RequestBody.base64",
-        "encode.bytes.BytesClient.request_body.base64url": "Encode.Bytes.RequestBody.base64url",
+        "encode.bytes.BytesClient.request_body.base64_url": "Encode.Bytes.RequestBody.base64url",
         "encode.bytes.BytesClient.response_body.default": "Encode.Bytes.ResponseBody.default",
         "encode.bytes.BytesClient.response_body.octet_stream": "Encode.Bytes.ResponseBody.octetStream",
         "encode.bytes.BytesClient.response_body.custom_content_type": "Encode.Bytes.ResponseBody.customContentType",
         "encode.bytes.BytesClient.response_body.base64": "Encode.Bytes.ResponseBody.base64",
-        "encode.bytes.BytesClient.response_body.base64url": "Encode.Bytes.ResponseBody.base64url"
+        "encode.bytes.BytesClient.response_body.base64_url": "Encode.Bytes.ResponseBody.base64url"
     }
 }