Merge pull request #680 from yoouyeon/main

[engine] getCommitRaws 함수 개선

packages/analysis-engine/src/constant.ts

@@ -0,0 +1,2 @@
+export const COMMIT_SEPARATOR = "4itc2s8hH-oA64s08h19";
+export const GIT_LOG_SEPARATOR = "I9M-0XOzvHlYPegVPpzb";

packages/analysis-engine/src/index.ts

@@ -92,3 +92,4 @@ export class AnalysisEngine {
 }
 
 export default AnalysisEngine;
+export { COMMIT_SEPARATOR, GIT_LOG_SEPARATOR } from "./constant";

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

@@ -1,5 +1,6 @@
 import { getCommitMessageType } from "./commit.util";
-import getCommitRaws from "./parser"; 
+import { COMMIT_SEPARATOR, GIT_LOG_SEPARATOR } from "./constant";
+import getCommitRaws from "./parser";
 import type { CommitRaw, DifferenceStatistic } from "./types";
 
 describe("commit message type", () => {
@@ -34,113 +35,114 @@ describe("commit message type", () => {
   });
 });
 
-describe('getCommitRaws', () => {
-  const testCommitLines = [
-    "commit a b (HEAD)",
-    "commit a b (HEAD -> main, origin/main, origin/HEAD)",
-    "commit a b (HEAD, tag: v1.0.0)",
-    "commit a b (HEAD -> main, origin/main, origin/HEAD, tag: v2.0.0)",
-    "commit a b (HEAD, tag: v2.0.0, tag: v1.4)"
+describe("getCommitRaws", () => {
+  const testAuthorCommitter = [
+    "John Park",
+    "[email protected]",
+    "Sun Sep 4 20:17:59 2022 +0900",
+    "John Park 2",
+    "[email protected]",
+    "Sun Sep 5 20:17:59 2022 +0900",
   ];
 
-  const expectedBranches = [
-    ['HEAD'],
-    ['HEAD', 'main', 'origin/main', 'origin/HEAD'],
-    ['HEAD'],
-    ['HEAD', 'main', 'origin/main', 'origin/HEAD'],
-    ['HEAD']
+  const testRefs = [
+    "HEAD",
+    "HEAD -> main, origin/main, origin/HEAD",
+    "HEAD, tag: v1.0.0",
+    "HEAD -> main, origin/main, origin/HEAD, tag: v2.0.0",
+    "HEAD, tag: v2.0.0, tag: v1.4",
   ];
 
-  const expectedTags = [
-    [],
-    [],
-    ['v1.0.0'],
-    ['v2.0.0'],
-    ['v2.0.0', 'v1.4']
+  const testCommitHash = ["a", "b"];
+
+  const testCommitMessage = "commit message";
+
+  const testCommitLines = testRefs.map((ref) =>
+    [...testCommitHash, ref, ...testAuthorCommitter, testCommitMessage].join(GIT_LOG_SEPARATOR)
+  );
+
+  const expectedBranches = [
+    ["HEAD"],
+    ["HEAD", "main", "origin/main", "origin/HEAD"],
+    ["HEAD"],
+    ["HEAD", "main", "origin/main", "origin/HEAD"],
+    ["HEAD"],
   ];
 
+  const expectedTags = [[], [], ["v1.0.0"], ["v2.0.0"], ["v2.0.0", "v1.4"]];
+
   const testCommitFileChanges = [
     "10\t0\ta.ts\n1\t0\tREADME.md",
     "3\t3\ta.ts",
     "4\t0\ta.ts",
-    "0\t6\ta.ts\n2\t0\tb.ts\n3\t3\tc.ts"
+    "0\t6\ta.ts\n2\t0\tb.ts\n3\t3\tc.ts",
   ];
 
-  const expectedFileChanged:DifferenceStatistic[] = [
+  const expectedFileChanged: DifferenceStatistic[] = [
     {
       totalInsertionCount: 11,
       totalDeletionCount: 0,
       fileDictionary: {
-        'a.ts': { insertionCount: 10, deletionCount: 0 },
-        'README.md': { insertionCount: 1, deletionCount: 0 },
-      }
+        "a.ts": { insertionCount: 10, deletionCount: 0 },
+        "README.md": { insertionCount: 1, deletionCount: 0 },
+      },
     },
     {
       totalInsertionCount: 3,
       totalDeletionCount: 3,
-      fileDictionary: { 'a.ts': { insertionCount: 3, deletionCount: 3 } }
+      fileDictionary: { "a.ts": { insertionCount: 3, deletionCount: 3 } },
     },
     {
       totalInsertionCount: 4,
       totalDeletionCount: 0,
-      fileDictionary: { 'a.ts': { insertionCount: 4, deletionCount: 0 } }
+      fileDictionary: { "a.ts": { insertionCount: 4, deletionCount: 0 } },
     },
     {
       totalInsertionCount: 5,
       totalDeletionCount: 9,
       fileDictionary: {
-        'a.ts': { insertionCount: 0, deletionCount: 6 },
-        'b.ts': { insertionCount: 2, deletionCount: 0 },
-        'c.ts': { insertionCount: 3, deletionCount: 3 },
-      }
-    }
+        "a.ts": { insertionCount: 0, deletionCount: 6 },
+        "b.ts": { insertionCount: 2, deletionCount: 0 },
+        "c.ts": { insertionCount: 3, deletionCount: 3 },
+      },
+    },
   ];
 
-  const commonExpectatedResult: CommitRaw={
+  const commonExpectatedResult: CommitRaw = {
     sequence: 0,
-    id: 'a',
-    parents: ['b'],
-    branches: ['HEAD'],
+    id: "a",
+    parents: ["b"],
+    branches: ["HEAD"],
     tags: [],
-    author: { name: 'John Park', email: '[email protected]' },
-    authorDate: new Date('Sun Sep 4 20:17:59 2022 +0900'),
-    committer: { name: 'John Park', email: '[email protected]' },
-    committerDate: new Date('Sun Sep 4 20:17:59 2022 +0900'),
-    message: 'commit message',
+    author: { name: testAuthorCommitter[0], email: testAuthorCommitter[1] },
+    authorDate: new Date(testAuthorCommitter[2]),
+    committer: { name: testAuthorCommitter[3], email: testAuthorCommitter[4] },
+    committerDate: new Date(testAuthorCommitter[5]),
+    message: testCommitMessage,
     differenceStatistic: {
       totalInsertionCount: 0,
       totalDeletionCount: 0,
       fileDictionary: {},
     },
-    commitMessageType: ""
+    commitMessageType: "",
   };
 
   testCommitLines.forEach((mockLog, index) => {
     it(`should parse gitlog to commitRaw(branch, tag)`, () => {
-      const mock = `${mockLog}
-Author: John Park <[email protected]>
-AuthorDate: Sun Sep 4 20:17:59 2022 +0900
-Commit: John Park <[email protected]>
-CommitDate: Sun Sep 4 20:17:59 2022 +0900
-\n\tcommit message
-`;
-      const result = getCommitRaws(mock);
-      const expectedResult = { ...commonExpectatedResult, branches: expectedBranches[index], tags: expectedTags[index] };
-
+      const result = getCommitRaws(COMMIT_SEPARATOR + mockLog);
+      const expectedResult = {
+        ...commonExpectatedResult,
+        branches: expectedBranches[index],
+        tags: expectedTags[index],
+      };
+
       expect(result).toEqual([expectedResult]);
     });
   });
 
   testCommitFileChanges.forEach((mockLog, index) => {
     it(`should parse gitlog to commitRaw(file changed)`, () => {
-      const mock = `commit a b (HEAD)
-Author: John Park <[email protected]>
-AuthorDate: Sun Sep 4 20:17:59 2022 +0900
-Commit: John Park <[email protected]>
-CommitDate: Sun Sep 4 20:17:59 2022 +0900
-\n\tcommit message
-\n${mockLog}
-`;
+      const mock = `${COMMIT_SEPARATOR}${testCommitLines[0]}\n${mockLog}`;
       const result = getCommitRaws(mock);
       const expectedResult = { ...commonExpectatedResult, differenceStatistic: expectedFileChanged[index] };
 

packages/analysis-engine/src/parser.ts

@@ -1,121 +1,74 @@
 import { getCommitMessageType } from "./commit.util";
-import type { CommitMessageType, CommitRaw, DifferenceStatistic, GitUser } from "./types";
-
-function getNameAndEmail(category: GitUser[], preParsedInfo: string) {
-  category.push({
-    name: preParsedInfo.split(": ")[1].split("<")[0].trim(),
-    email: preParsedInfo.split(": ")[1].split("<")[1].split(">")[0].trim(),
-  });
-}
+import { COMMIT_SEPARATOR, GIT_LOG_SEPARATOR } from "./constant";
+import type { CommitRaw } from "./types";
 
 export default function getCommitRaws(log: string) {
   if (!log) return [];
+  const EOL_REGEX = /\r?\n/;
 
-  // line 별로 분리하기
-  const splitByNewLine = log.split(/\r?\n/);
-
-  // 분리한 것들을 쭉 돌면서 각 카테고리별로 담을 예정
-  type Refs = string[];
-
-  const ids: string[] = [];
-  const parentsMatrix: string[][] = [];
-  const branchesMatrix: Refs[] = [];
-  const tagsMatrix: Refs[] = [];
-  const authors: GitUser[] = [];
-  const authorDates: Date[] = [];
-  const committers: GitUser[] = [];
-  const commitDates: Date[] = [];
-  const messages: string[] = [];
-  const commitTypes: CommitMessageType[] = [];
-  const differenceStatistics: DifferenceStatistic[] = [];
-
-  // commit별 fileChanged를 분리시키기 위한 임시 index
-  let commitIdx = -1;
+  // step 0: Split log into commits
+  const commits = log.split(COMMIT_SEPARATOR);
+  const commitRaws: CommitRaw[] = [];
+  // skip the first empty element
+  for (let commitIdx = 1; commitIdx < commits.length; commitIdx += 1) {
+    const commitLines = commits[commitIdx].split(EOL_REGEX);
 
-  if (splitByNewLine) {
-    splitByNewLine.forEach((str, idx) => {
-      if (str.startsWith("commit")) {
-        commitIdx += 1;
-        tagsMatrix.push([]);
-        branchesMatrix.push([]);
-        differenceStatistics.push({
-          totalInsertionCount: 0,
-          totalDeletionCount: 0,
-          fileDictionary: {},
-        });
-        const splitedCommitLine = str.split("(");
-        const commitInfos = splitedCommitLine[0]
-          .replace("commit ", "")
-          .split(" ")
-          .filter((e) => e);
-        ids.push(commitInfos[0]);
-        commitInfos.splice(0, 1);
-        parentsMatrix.push(commitInfos);
-        const branchAndTagInfos = splitedCommitLine[1]?.replace(")", "").replace(" -> ", ", ").split(", ");
-        if (branchAndTagInfos) {
-          branchAndTagInfos.forEach((branchAndTagInfo) => {
-            if (branchAndTagInfo.startsWith("tag:"))
-              return tagsMatrix[commitIdx].push(branchAndTagInfo.replace("tag: ", ""));
-            return branchesMatrix[commitIdx].push(branchAndTagInfo);
-          });
-        }
-        return false;
-      }
-      if (str.startsWith("Author:")) return getNameAndEmail(authors, str);
-      if (str.startsWith("AuthorDate")) return authorDates.push(new Date(str.split(": ")[1].trim()));
-      if (str.startsWith("Commit:")) return getNameAndEmail(committers, str);
-      if (str.startsWith("CommitDate")) {
-        let indexCheckFileChanged = idx + 2;
-        let eachCommitMessage = "";
-        while (splitByNewLine[indexCheckFileChanged] !== "") {
-          if (eachCommitMessage !== "") {
-            eachCommitMessage += "\n";
-          }
-          eachCommitMessage += splitByNewLine[indexCheckFileChanged].trim();
-          indexCheckFileChanged += 1;
+    // step 1: Extract commitData from the first line of the commit
+    const commitData = commitLines[0].split(GIT_LOG_SEPARATOR);
+    // Extract branch and tag data from commitData[2]
+    const refs = commitData[2].replace(" -> ", ", ").split(", ");
+    const [branches, tags]: string[][] = refs.reduce(
+      ([branches, tags], ref) => {
+        if (ref.startsWith("tag: ")) {
+          tags.push(ref.replace("tag: ", ""));
+        } else {
+          branches.push(ref);
         }
-        commitDates.push(new Date(str.split(": ")[1].trim()));
-        messages.push(eachCommitMessage);
-        commitTypes.push(getCommitMessageType(eachCommitMessage));
-      }
-      if (/^\d/.test(str) || /^-/.test(str)) {
-        const [addition, deletion, path] = str
-          .split(" ")
-          .filter((e) => e)[0]
-          .split("\t");
+        return [branches, tags];
+      },
+      [new Array<string>(), new Array<string>()]
+    );
 
-        const numberedAddition = addition === "-" ? 0 : Number(addition);
-        const numberedDeletion = deletion === "-" ? 0 : Number(deletion);
-        differenceStatistics[commitIdx].totalInsertionCount += numberedAddition;
-        differenceStatistics[commitIdx].totalDeletionCount += numberedDeletion;
-        differenceStatistics[commitIdx].fileDictionary[path] = {
-          insertionCount: numberedAddition,
-          deletionCount: numberedDeletion,
-        };
-      }
-      return false;
-    });
-  }
-
-  // 각 카테고리로 담은 다음 다시 JSON으로 변환하기 위함
-  const commitRaws: CommitRaw[] = [];
+    // make base commitRaw object
+    const commitRaw: CommitRaw = {
+      sequence: commitIdx - 1,
+      id: commitData[0],
+      parents: commitData[1].split(" "),
+      branches, // commitData[2] is already split into branches and tags
+      tags,
+      author: {
+        name: commitData[3],
+        email: commitData[4],
+      },
+      authorDate: new Date(commitData[5]),
+      committer: {
+        name: commitData[6],
+        email: commitData[7],
+      },
+      committerDate: new Date(commitData[8]),
+      message: commitData[9],
+      commitMessageType: getCommitMessageType(commitData[9]),
+      differenceStatistic: {
+        totalInsertionCount: 0,
+        totalDeletionCount: 0,
+        fileDictionary: {},
+      },
+    };
 
-  // 카테고리 별로 담은 것을 JSON화 시키기
-  for (let i = 0; i < ids.length; i += 1) {
-    commitRaws.push({
-      sequence: i,
-      id: ids[i],
-      parents: parentsMatrix[i],
-      branches: branchesMatrix[i],
-      tags: tagsMatrix[i],
-      author: authors[i],
-      authorDate: authorDates[i],
-      committer: committers[i],
-      committerDate: commitDates[i],
-      message: messages[i],
-      commitMessageType: commitTypes[i],
-      differenceStatistic: differenceStatistics[i],
-    });
+    // step 2: Extract diffStats from the rest of the commit
+    for (let diffIdx = 1; diffIdx < commitLines.length; diffIdx += 1) {
+      if (commitLines[diffIdx] === "") continue;
+      const [insertions, deletions, path] = commitLines[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);
   }
 
   return commitRaws;