Merge pull request #770 from yoouyeon/main

[engine] git log 파싱에 사용되는 separator 개선

packages/analysis-engine/src/index.ts

@@ -75,7 +75,7 @@ export class AnalysisEngine {
     if (this.isDebugMode) console.log("stemDict: ", stemDict);
     const csmDict = buildCSMDict(commitDict, stemDict, this.baseBranchName, pullRequests);
     if (this.isDebugMode) console.log("csmDict: ", csmDict);
-    const nodes = stemDict.get(this.baseBranchName)?.nodes?.map(({commit}) => commit);
+    const nodes = stemDict.get(this.baseBranchName)?.nodes?.map(({ commit }) => commit);
     const geminiCommitSummary = await getSummary(nodes ? nodes?.slice(-10) : []);
     if (this.isDebugMode) console.log("GeminiCommitSummary: ", geminiCommitSummary);
 
@@ -92,4 +92,3 @@ export class AnalysisEngine {
 }
 
 export default AnalysisEngine;
-export { COMMIT_SEPARATOR, GIT_LOG_SEPARATOR } from "./constant";

packages/analysis-engine/src/parser.spec.ts

@@ -1,5 +1,4 @@
 import { getCommitMessageType } from "./commit.util";
-import { COMMIT_SEPARATOR, GIT_LOG_SEPARATOR } from "./constant";
 import getCommitRaws from "./parser";
 import type { CommitRaw, DifferenceStatistic } from "./types";
 
@@ -36,11 +35,15 @@ describe("commit message type", () => {
 });
 
 describe("getCommitRaws", () => {
-  const fakeAuthorAndCommitter = `${GIT_LOG_SEPARATOR}John Park${GIT_LOG_SEPARATOR}[email protected]${GIT_LOG_SEPARATOR}Sun Sep 4 20:17:59 2022 +0900${GIT_LOG_SEPARATOR}John Park 2${GIT_LOG_SEPARATOR}[email protected]${GIT_LOG_SEPARATOR}Sun Sep 5 20:17:59 2022 +0900`;
-  const fakeCommitMessage = `${GIT_LOG_SEPARATOR}commit message${GIT_LOG_SEPARATOR}`;
-  const fakeCommitMessageAndBody = `${GIT_LOG_SEPARATOR}commit message title\n\ncommit message body${GIT_LOG_SEPARATOR}`;
-  const fakeCommitHash = `a${GIT_LOG_SEPARATOR}b`;
-  const fakeCommitRef = `${GIT_LOG_SEPARATOR}HEAD`;
+  const FRONT_NEW_LINE = "\n\n";
+  const INDENTATION = "    ";
+
+  const fakeAuthor = "John Park\[email protected]\nSun Sep 4 20:17:59 2022 +0900";
+  const fakeCommitter = `John Park 2\[email protected]\nSun Sep 5 20:17:59 2022 +0900`;
+  const fakeCommitMessage = `commit message\n${INDENTATION}`;
+  const fakeCommitMessageAndBody = `commit message title\n${INDENTATION}\n${INDENTATION}commit message body`;
+  const fakeCommitHash = "a\nb";
+  const fakeCommitRef = "HEAD";
   const fakeCommitFileChange = "10\t0\ta.ts\n1\t0\tREADME.md";
 
   const commonExpectatedResult: CommitRaw = {
@@ -73,23 +76,23 @@ describe("getCommitRaws", () => {
 
   it.each([
     [
-      `${COMMIT_SEPARATOR}${`a${GIT_LOG_SEPARATOR}`}${fakeCommitRef}${fakeAuthorAndCommitter}${fakeCommitMessage}`,
+      `${FRONT_NEW_LINE}${"a\n"}\n${fakeCommitRef}\n${fakeAuthor}\n${fakeCommitter}\n${fakeCommitMessage}`,
       {
         ...commonExpectatedResult,
         id: "a",
-        parents: [""],
+        parents: [],
       },
     ],
     [
-      `${COMMIT_SEPARATOR}${`c${GIT_LOG_SEPARATOR}b`}${fakeCommitRef}${fakeAuthorAndCommitter}${fakeCommitMessage}`,
+      `${FRONT_NEW_LINE}${"c\nd"}\n${fakeCommitRef}\n${fakeAuthor}\n${fakeCommitter}\n${fakeCommitMessage}`,
       {
         ...commonExpectatedResult,
         id: "c",
-        parents: ["b"],
+        parents: ["d"],
       },
     ],
     [
-      `${COMMIT_SEPARATOR}${`d${GIT_LOG_SEPARATOR}e f`}${fakeCommitRef}${fakeAuthorAndCommitter}${fakeCommitMessage}`,
+      `${FRONT_NEW_LINE}${"d\ne f"}\n${fakeCommitRef}\n${fakeAuthor}\n${fakeCommitter}\n${fakeCommitMessage}`,
       {
         ...commonExpectatedResult,
         id: "d",
@@ -103,47 +106,47 @@ describe("getCommitRaws", () => {
 
   it.each([
     [
-      `${COMMIT_SEPARATOR}${fakeCommitHash}${`${GIT_LOG_SEPARATOR}HEAD`}${fakeAuthorAndCommitter}${fakeCommitMessage}`,
+      `${FRONT_NEW_LINE}${fakeCommitHash}\n${"HEAD"}\n${fakeAuthor}\n${fakeCommitter}\n${fakeCommitMessage}`,
       {
         ...commonExpectatedResult,
         branches: ["HEAD"],
         tags: [],
       },
     ],
     [
-      `${COMMIT_SEPARATOR}${fakeCommitHash}${`${GIT_LOG_SEPARATOR}HEAD -> main, origin/main, origin/HEAD`}${fakeAuthorAndCommitter}${fakeCommitMessage}`,
+      `${FRONT_NEW_LINE}${fakeCommitHash}\n${"HEAD -> main, origin/main, origin/HEAD"}\n${fakeAuthor}\n${fakeCommitter}\n${fakeCommitMessage}`,
       {
         ...commonExpectatedResult,
         branches: ["HEAD", "main", "origin/main", "origin/HEAD"],
         tags: [],
       },
     ],
     [
-      `${COMMIT_SEPARATOR}${fakeCommitHash}${`${GIT_LOG_SEPARATOR}HEAD, tag: v1.0.0`}${fakeAuthorAndCommitter}${fakeCommitMessage}`,
+      `${FRONT_NEW_LINE}${fakeCommitHash}\n${"HEAD, tag: v1.0.0"}\n${fakeAuthor}\n${fakeCommitter}$\n${fakeCommitMessage}`,
       {
         ...commonExpectatedResult,
         branches: ["HEAD"],
         tags: ["v1.0.0"],
       },
     ],
     [
-      `${COMMIT_SEPARATOR}${fakeCommitHash}${`${GIT_LOG_SEPARATOR}HEAD -> main, origin/main, origin/HEAD, tag: v2.0.0`}${fakeAuthorAndCommitter}${fakeCommitMessage}`,
+      `${FRONT_NEW_LINE}${fakeCommitHash}\n${"HEAD -> main, origin/main, origin/HEAD, tag: v2.0.0"}\n${fakeAuthor}\n${fakeCommitter}\n${fakeCommitMessage}`,
       {
         ...commonExpectatedResult,
         branches: ["HEAD", "main", "origin/main", "origin/HEAD"],
         tags: ["v2.0.0"],
       },
     ],
     [
-      `${COMMIT_SEPARATOR}${fakeCommitHash}${`${GIT_LOG_SEPARATOR}HEAD, tag: v2.0.0, tag: v1.4`}${fakeAuthorAndCommitter}${fakeCommitMessage}`,
+      `${FRONT_NEW_LINE}${fakeCommitHash}\n${"HEAD, tag: v2.0.0, tag: v1.4"}\n${fakeAuthor}\n${fakeCommitter}\n${fakeCommitMessage}`,
       {
         ...commonExpectatedResult,
         branches: ["HEAD"],
         tags: ["v2.0.0", "v1.4"],
       },
     ],
     [
-      `${COMMIT_SEPARATOR}${fakeCommitHash}${GIT_LOG_SEPARATOR}${fakeAuthorAndCommitter}${fakeCommitMessage}`,
+      `${FRONT_NEW_LINE}${fakeCommitHash}\n${""}\n${fakeAuthor}\n${fakeCommitter}\n${fakeCommitMessage}`,
       {
         ...commonExpectatedResult,
         branches: [],
@@ -157,7 +160,7 @@ describe("getCommitRaws", () => {
 
   it.each([
     [
-      `${COMMIT_SEPARATOR}${fakeCommitHash}${fakeCommitRef}${fakeAuthorAndCommitter}${fakeCommitMessage}\n${"10\t0\ta.ts\n1\t0\tREADME.md"}`,
+      `${FRONT_NEW_LINE}${fakeCommitHash}\n${fakeCommitRef}\n${fakeAuthor}\n${fakeCommitter}\n${fakeCommitMessage}\n${"10\t0\ta.ts\n1\t0\tREADME.md"}`,
       {
         ...commonExpectatedResult,
         differenceStatistic: {
@@ -171,7 +174,7 @@ describe("getCommitRaws", () => {
       },
     ],
     [
-      `${COMMIT_SEPARATOR}${fakeCommitHash}${fakeCommitRef}${fakeAuthorAndCommitter}${fakeCommitMessage}\n${"3\t3\ta.ts"}`,
+      `${FRONT_NEW_LINE}${fakeCommitHash}\n${fakeCommitRef}\n${fakeAuthor}\n${fakeCommitter}\n${fakeCommitMessage}\n${"3\t3\ta.ts"}`,
       {
         ...commonExpectatedResult,
         differenceStatistic: {
@@ -182,7 +185,7 @@ describe("getCommitRaws", () => {
       },
     ],
     [
-      `${COMMIT_SEPARATOR}${fakeCommitHash}${fakeCommitRef}${fakeAuthorAndCommitter}${fakeCommitMessage}\n${"4\t0\ta.ts"}`,
+      `${FRONT_NEW_LINE}${fakeCommitHash}\n${fakeCommitRef}\n${fakeAuthor}\n${fakeCommitter}\n${fakeCommitMessage}\n${"4\t0\ta.ts"}`,
       {
         ...commonExpectatedResult,
         differenceStatistic: {
@@ -193,7 +196,7 @@ describe("getCommitRaws", () => {
       },
     ],
     [
-      `${COMMIT_SEPARATOR}${fakeCommitHash}${fakeCommitRef}${fakeAuthorAndCommitter}${fakeCommitMessage}\n${"0\t6\ta.ts\n2\t0\tb.ts\n3\t3\tc.ts"}`,
+      `${FRONT_NEW_LINE}${fakeCommitHash}\n${fakeCommitRef}\n${fakeAuthor}\n${fakeCommitter}\n${fakeCommitMessage}\n${"0\t6\ta.ts\n2\t0\tb.ts\n3\t3\tc.ts"}`,
       {
         ...commonExpectatedResult,
         differenceStatistic: {
@@ -213,7 +216,7 @@ describe("getCommitRaws", () => {
   });
 
   it(`should parse gitlog to commitRaw(multiple commits)`, () => {
-    const mockLog = `${COMMIT_SEPARATOR}${fakeCommitHash}${fakeCommitRef}${fakeAuthorAndCommitter}${fakeCommitMessage}\n${fakeCommitFileChange}${COMMIT_SEPARATOR}${fakeCommitHash}${fakeCommitRef}${fakeAuthorAndCommitter}${fakeCommitMessage}`;
+    const mockLog = `${FRONT_NEW_LINE}${fakeCommitHash}\n${fakeCommitRef}\n${fakeAuthor}\n${fakeCommitter}\n${fakeCommitMessage}\n${fakeCommitFileChange}\n\n\n\n${fakeCommitHash}\n${fakeCommitRef}\n${fakeAuthor}\n${fakeCommitter}\n${fakeCommitMessage}`;
     const result = getCommitRaws(mockLog);
     const expectedResult = [
       { ...commonExpectatedResult, differenceStatistic: expectedFileChange },
@@ -225,23 +228,23 @@ describe("getCommitRaws", () => {
 
   it.each([
     [
-      `${COMMIT_SEPARATOR}${fakeCommitHash}${fakeCommitRef}${fakeAuthorAndCommitter}${`${GIT_LOG_SEPARATOR}commit message title${GIT_LOG_SEPARATOR}`}`,
+      `${FRONT_NEW_LINE}${fakeCommitHash}\n${fakeCommitRef}\n${fakeAuthor}\n${fakeCommitter}\n${"commit message title"}\n${INDENTATION}`,
       { ...commonExpectatedResult, message: "commit message title" },
     ],
     [
-      `${COMMIT_SEPARATOR}${fakeCommitHash}${fakeCommitRef}${fakeAuthorAndCommitter}${`${GIT_LOG_SEPARATOR}commit message title\ncommit message${GIT_LOG_SEPARATOR}`}`,
-      { ...commonExpectatedResult, message: "commit message title\ncommit message" },
+      `${FRONT_NEW_LINE}${fakeCommitHash}\n${fakeCommitRef}\n${fakeAuthor}\n${fakeCommitter}\n${"commit message title"}\n${INDENTATION}${"commit message body"}`,
+      { ...commonExpectatedResult, message: "commit message title\ncommit message body" },
     ],
     [
-      `${COMMIT_SEPARATOR}${fakeCommitHash}${fakeCommitRef}${fakeAuthorAndCommitter}${`${GIT_LOG_SEPARATOR}commit message title\n\ncommit message body${GIT_LOG_SEPARATOR}`}`,
+      `${FRONT_NEW_LINE}${fakeCommitHash}\n${fakeCommitRef}\n${fakeAuthor}\n${fakeCommitter}\n${"commit message title"}\n${INDENTATION}\n${INDENTATION}${"commit message body"}`,
       { ...commonExpectatedResult, message: "commit message title\n\ncommit message body" },
     ],
     [
-      `${COMMIT_SEPARATOR}${fakeCommitHash}${fakeCommitRef}${fakeAuthorAndCommitter}${`${GIT_LOG_SEPARATOR}commit message title\n\n\ncommit message body${GIT_LOG_SEPARATOR}`}`,
+      `${FRONT_NEW_LINE}${fakeCommitHash}\n${fakeCommitRef}\n${fakeAuthor}\n${fakeCommitter}\n${"commit message title"}\n${INDENTATION}\n${INDENTATION}\n${INDENTATION}${"commit message body"}`,
       { ...commonExpectatedResult, message: "commit message title\n\n\ncommit message body" },
     ],
     [
-      `${COMMIT_SEPARATOR}${fakeCommitHash}${fakeCommitRef}${fakeAuthorAndCommitter}${`${GIT_LOG_SEPARATOR}${GIT_LOG_SEPARATOR}`}`,
+      `${FRONT_NEW_LINE}${fakeCommitHash}\n${fakeCommitRef}\n${fakeAuthor}\n${fakeCommitter}\n\n${INDENTATION}`,
       { ...commonExpectatedResult, message: "" },
     ],
   ])("should parse gitlog to commitRaw(commit message)", (mockLog, expectedResult) => {
@@ -250,7 +253,7 @@ describe("getCommitRaws", () => {
   });
 
   it(`should parse gitlog to commitRaw(commit message body and file change)`, () => {
-    const mockLog = `${COMMIT_SEPARATOR}${fakeCommitHash}${fakeCommitRef}${fakeAuthorAndCommitter}${fakeCommitMessageAndBody}\n${fakeCommitFileChange}`;
+    const mockLog = `${FRONT_NEW_LINE}${fakeCommitHash}\n${fakeCommitRef}\n${fakeAuthor}\n${fakeCommitter}\n${fakeCommitMessageAndBody}\n${fakeCommitFileChange}`;
     const result = getCommitRaws(mockLog);
     const expectedResult = {
       ...commonExpectatedResult,

packages/analysis-engine/src/parser.ts

@@ -1,18 +1,18 @@
 import { getCommitMessageType } from "./commit.util";
-import { COMMIT_SEPARATOR, GIT_LOG_SEPARATOR } from "./constant";
-import type { CommitRaw } from "./types";
+import type { CommitRaw, DifferenceStatistic } from "./types";
 
 export default function getCommitRaws(log: string) {
   if (!log) return [];
   const EOL_REGEX = /\r?\n/;
+  const COMMIT_SEPARATOR = new RegExp(`${EOL_REGEX.source}{4}`);
+  const INDENTATION = "    ";
 
   // step 0: Split log into commits
-  const commits = log.split(COMMIT_SEPARATOR);
+  const commits = log.substring(2).split(COMMIT_SEPARATOR);
   const commitRaws: CommitRaw[] = [];
-  // skip the first empty element
-  for (let commitIdx = 1; commitIdx < commits.length; commitIdx += 1) {
-    // step 1: Extract commitData from the first line of the commit
-    const commitData = commits[commitIdx].split(GIT_LOG_SEPARATOR);
+  for (let commitIdx = 0; commitIdx < commits.length; commitIdx += 1) {
+    // step 1: Extract commitData
+    const commitData = commits[commitIdx].split(EOL_REGEX);
     const [
       id,
       parents,
@@ -23,10 +23,9 @@ export default function getCommitRaws(log: string) {
       committerName,
       committerEmail,
       committerDate,
-      message,
-      diffStats,
+      ...messageAndDiffStats
     ] = commitData;
-    // Extract branch and tag data from refs
+    // step 2: Extract branch and tag data from refs
     const refsArray = refs.replace(" -> ", ", ").split(", ");
     const [branches, tags]: string[][] = refsArray.reduce(
       ([branches, tags], ref) => {
@@ -41,11 +40,45 @@ export default function getCommitRaws(log: string) {
       [new Array<string>(), new Array<string>()]
     );
 
-    // make base commitRaw object
+    // step 3: Extract message and diffStats
+    let messageSubject = "";
+    let messageBody = "";
+    const diffStats: DifferenceStatistic = {
+      totalInsertionCount: 0,
+      totalDeletionCount: 0,
+      fileDictionary: {},
+    };
+    for (let idx = 0; idx < messageAndDiffStats.length; idx++) {
+      const line = messageAndDiffStats[idx];
+      if (idx === 0)
+        // message subject
+        messageSubject = line;
+      else if (line.startsWith(INDENTATION)) {
+        // message body (add newline if not first line)
+        messageBody += idx === 1 ? line.trim() : `\n${line.trim()}`;
+      } else if (line === "")
+        // pass empty line
+        continue;
+      else {
+        // diffStats
+        const [insertions, deletions, path] = line.split("\t");
+        const numberedInsertions = insertions === "-" ? 0 : Number(insertions);
+        const numberedDeletions = deletions === "-" ? 0 : Number(deletions);
+        diffStats.totalInsertionCount += numberedInsertions;
+        diffStats.totalDeletionCount += numberedDeletions;
+        diffStats.fileDictionary[path] = {
+          insertionCount: numberedInsertions,
+          deletionCount: numberedDeletions,
+        };
+      }
+    }
+
+    const message = messageBody === "" ? messageSubject : `${messageSubject}\n${messageBody}`;
+    // step 4: Construct commitRaw
     const commitRaw: CommitRaw = {
-      sequence: commitIdx - 1,
+      sequence: commitIdx,
       id,
-      parents: parents.split(" "),
+      parents: parents.length === 0 ? [] : parents.split(" "),
       branches,
       tags,
       author: {
@@ -60,32 +93,8 @@ export default function getCommitRaws(log: string) {
       committerDate: new Date(committerDate),
       message,
       commitMessageType: getCommitMessageType(message),
-      differenceStatistic: {
-        totalInsertionCount: 0,
-        totalDeletionCount: 0,
-        fileDictionary: {},
-      },
+      differenceStatistic: diffStats,
     };
-
-    // step 2: Extract diffStats from the rest of the commit
-    if (!diffStats) {
-      commitRaws.push(commitRaw);
-      continue;
-    }
-    const diffStatsArray = diffStats.split(EOL_REGEX);
-    // pass the first empty element
-    for (let diffIdx = 1; diffIdx < diffStatsArray.length; diffIdx += 1) {
-      if (diffStatsArray[diffIdx] === "") continue;
-      const [insertions, deletions, path] = diffStatsArray[diffIdx].split("\t");
-      const numberedInsertions = insertions === "-" ? 0 : Number(insertions);
-      const numberedDeletions = deletions === "-" ? 0 : Number(deletions);
-      commitRaw.differenceStatistic.totalInsertionCount += numberedInsertions;
-      commitRaw.differenceStatistic.totalDeletionCount += numberedDeletions;
-      commitRaw.differenceStatistic.fileDictionary[path] = {
-        insertionCount: numberedInsertions,
-        deletionCount: numberedDeletions,
-      };
-    }
     commitRaws.push(commitRaw);
   }
 

packages/vscode/src/utils/git.util.ts

@@ -1,4 +1,3 @@
-import { COMMIT_SEPARATOR, GIT_LOG_SEPARATOR } from "@githru-vscode-ext/analysis-engine";
 import * as cp from "child_process";
 import * as fs from "fs";
 import * as path from "path";
@@ -156,20 +155,20 @@ export async function getGitExecutableFromPaths(paths: string[]): Promise<GitExe
 export async function getGitLog(gitPath: string, currentWorkspacePath: string): Promise<string> {
   return new Promise((resolve, reject) => {
     const gitLogFormat =
-      COMMIT_SEPARATOR +
+      "%n%n" +
       [
         "%H", // commit hash (id)
         "%P", // parent hashes
         "%D", // ref names (branches, tags)
         "%an", // author name
         "%ae", // author email
         "%ad", // author date
-        "%cn",
-        "%ce",
-        "%cd", // committer name, committer email and committer date
-        "%B", // commit message  (subject and body)
-      ].join(GIT_LOG_SEPARATOR) +
-      GIT_LOG_SEPARATOR;
+        "%cn", // committer name
+        "%ce", // committer email
+        "%cd", // committer date
+        "%w(0,0,4)%s", // commit message subject
+        "%b", // commit message body
+      ].join("%n");
     const args = [
       "--no-pager",
       "log",