ci(github): commit parity check based on similarity

Primary Changes --------------- 1. Added a commit-pr similarity ratio to lessen the check strictness 2. Added an env var to the workflow to easily control the similarity ratio Changes required to incorporate 1) --------------------------------- 3. Added string-similarity-js package in package.json Fixes hyperledger-cacti#3470 Signed-off-by: jagpreetsinghsasan <[email protected]>
jagpreetsinghsasan · Sep 1, 2024 · 9bbb846 · 9bbb846
1 parent 7e7bb44
commit 9bbb846
Show file tree

Hide file tree

Showing 4 changed files with 45 additions and 13 deletions.
diff --git a/.github/workflows/pr-commit-parity.yaml b/.github/workflows/pr-commit-parity.yaml
@@ -15,11 +15,13 @@ jobs:
   pr-commit-parity:
     name: PR and Commit messages Parity
     runs-on: ubuntu-22.04
+    env:
+      ACCEPTABLE_SIMILARITY_RATIO: 0.9
     steps:
     - uses: actions/[email protected]
     - name: Use Node.js v18.18.2
       uses: actions/[email protected]
       with:
         node-version: v18.19.0
     - name: Execute pr-commit-parity script
-      run: node tools/pr-commit-parity.js ${{ github.event.pull_request.url }}
+      run: node tools/pr-commit-parity.js ${{ github.event.pull_request.url }} $ACCEPTABLE_SIMILARITY_RATIO
diff --git a/package.json b/package.json
@@ -202,6 +202,7 @@
     "sort-package-json": "1.57.0",
     "source-map-loader": "4.0.1",
     "stream-browserify": "3.0.0",
+    "string-similarity-js": "2.1.4",
     "tap": "16.3.8",
     "tape": "5.6.6",
     "tape-promise": "4.0.0",

diff --git a/tools/pr-commit-parity.js b/tools/pr-commit-parity.js
@@ -1,20 +1,30 @@
+import { stringSimilarity } from "string-similarity-js";
+
 export async function fetchJsonFromUrl(url) {
   const fetchResponse = await fetch(url);
   return fetchResponse.json();
 }
 
-//regex expressions
+// regex expressions
 const PULL_REQ_REQUIREMENTS_REGEX = /\*\*Pull\sRequest\sRequirements(.|\n)*/gim;
 const FIXES_OR_DEPENDS_REGEX = /(Fixes|Depends)(.|\n)*/gim;
 const SIGNED_OFF_REGEX = /(")*Signed-off-by:(.|\s)*/gim;
+const COMMIT_TITLE_REGEX = /.*\n/m;
 const HYPHEN_REGEX = /(-)+/gm;
 const BACKTICK_REGEX = /`+/gm;
-const SPACE_REGEX = / +/gm;
 const COMMIT_TO_BE_REVIEWED_REGEX = /("#*\s*Commit\sto\sbe\sreviewed)/gim;
-const WHITESPACES_HARDCODED_REGEX = /(\r\n|\n|\r|\\r|\\n)/gm;
+const WHITESPACES_HARDCODED_REGEX = /(\r\n|\\r)/gm;
+const NEWLINE_HARDCODED_REGEX = /\\n/gm;
 
 const args = process.argv.slice(2);
+
+// The following 2 lines should be commented to test this script locally
 const pullReqUrl = args[0];
+const ACCEPTABLE_SIMILARITY_RATIO = parseFloat(args[1]);
+
+// These following 2 lines should be un-commented to test this script locally
+// const pullReqUrl = "https://api.github.com/repos/hyperledger/cactus/pulls/3338";
+// const ACCEPTABLE_SIMILARITY_RATIO = 0.9;
 
 const prMetadata = await fetchJsonFromUrl(pullReqUrl);
 const prBodyRaw = JSON.stringify(prMetadata.body);
@@ -29,33 +39,44 @@ commitMessagesMetadata.forEach((commitMessageMetadata) => {
       .replace(HYPHEN_REGEX, "")
       .replace(BACKTICK_REGEX, "")
       .replace(WHITESPACES_HARDCODED_REGEX, "")
-      .replace(FIXES_OR_DEPENDS_REGEX, "")
-      .replace(SPACE_REGEX, ""),
+      .replace(FIXES_OR_DEPENDS_REGEX, ""),
   );
 });
 
 let prBodyStriped = prBodyRaw
   .replace(PULL_REQ_REQUIREMENTS_REGEX, "")
   .replace(FIXES_OR_DEPENDS_REGEX, "")
-  .replace(WHITESPACES_HARDCODED_REGEX, "")
+  .replace(WHITESPACES_HARDCODED_REGEX, "\n")
   .replace(SIGNED_OFF_REGEX, "")
   .replace(HYPHEN_REGEX, "")
   .replace(BACKTICK_REGEX, "")
   .replace(COMMIT_TO_BE_REVIEWED_REGEX, "")
-  .replace(SPACE_REGEX, "");
+  .replace(NEWLINE_HARDCODED_REGEX, "");
 
-let PR_BODY_IN_COMMIT_MESSAGES = false;
+let PR_COMMIT_PARITY = false;
 for (let commitMessageListIndex in commitMessageList) {
   let commitMessage = commitMessageList[commitMessageListIndex];
-  if (commitMessage == prBodyStriped) PR_BODY_IN_COMMIT_MESSAGES = true;
+  if (
+    stringSimilarity(commitMessage, prBodyStriped) >=
+      ACCEPTABLE_SIMILARITY_RATIO ||
+    stringSimilarity(
+      commitMessage.replace(COMMIT_TITLE_REGEX, ""),
+      prBodyStriped,
+    ) >= ACCEPTABLE_SIMILARITY_RATIO
+  )
+    PR_COMMIT_PARITY = true;
 }
 
-if (!PR_BODY_IN_COMMIT_MESSAGES) {
+if (!PR_COMMIT_PARITY) {
   console.error(
-    "PR Body does not match any existing commit message\n" +
-      "Please make sure that the PR Body matches to minimum one of the commit messages\n" +
+    "PR message and commit message are not similar. A general solution for this is to have PR message exactly same as the commit message\n" +
       "Please refer the following PR for reference: https://github.com/hyperledger/cacti/pull/3338\n" +
       "And the commit message here: https://github.com/hyperledger/cacti/pull/3338/commits/47ebdec442d30fa48c8518b876c47c38097cf028\n",
+      "-----------------------------------------------\n\n",
+    "Commit Message List (ignore extra white spaces and new lines)\n" +
+      commitMessageList +
+      "\n----------------------------------------------\nRelevant Pr Description (ignore extra white spaces and new lines)\n" +
+      prBodyStriped,
   );
   process.exit(-1);
 }
diff --git a/yarn.lock b/yarn.lock
@@ -11103,6 +11103,7 @@ __metadata:
     sort-package-json: "npm:1.57.0"
     source-map-loader: "npm:4.0.1"
     stream-browserify: "npm:3.0.0"
+    string-similarity-js: "npm:2.1.4"
     tap: "npm:16.3.8"
     tape: "npm:5.6.6"
     tape-promise: "npm:4.0.0"
@@ -49340,6 +49341,13 @@ __metadata:
   languageName: node
   linkType: hard
 
+"string-similarity-js@npm:2.1.4":
+  version: 2.1.4
+  resolution: "string-similarity-js@npm:2.1.4"
+  checksum: 10/55063198f233aca8833d8429626d9fafdeb164e0fc6a66364a4d8f509df17d49f8a8a42e4096da60c53fe4586b1bace7098c0265da590fe4ec96c3a8ecb90f69
+  languageName: node
+  linkType: hard
+
 "string-similarity@npm:^4.0.1":
   version: 4.0.4
   resolution: "string-similarity@npm:4.0.4"