⚗️ Group similar commits together (#50)

.gitignore

@@ -61,3 +61,4 @@ typings/
 .next
 
 CHANGELOG.json
+.vscode

package.json

@@ -20,6 +20,7 @@
   "devDependencies": {
     "eslint": "^5.4.0",
     "eslint-config-airbnb-base": "^13.1.0",
+    "eslint-plugin-import": "^2.14.0",
     "jest": "^23.5.0",
     "lerna": "^2.11.0"
   },

packages/gitmoji-changelog-cli/src/cli.js

@@ -29,6 +29,12 @@ async function main(options = {}) {
     }
   } catch (e) { /* ignore error */ }
 
+  if (options.groupSimilarCommits) {
+    logger.warn('⚗️  You are using a beta feature - may not working as expected')
+    logger.warn('Feel free to open issues or PR into gitmoji-changelog')
+    logger.warn('\t> https://github.com/frinyvonnick/gitmoji-changelog')
+  }
+
   try {
     const changelog = await generateChangelog(options)
 

packages/gitmoji-changelog-cli/src/index.js

@@ -48,6 +48,7 @@ yargs
 
   .option('format', { default: 'markdown', desc: 'changelog format (markdown, json)' })
   .option('output', { desc: 'output changelog file' })
+  .option('group-similar-commits', { desc: '[⚗️  - beta] try to group similar commits', default: false })
   .option('author', { default: false, desc: 'add the author in changelog lines' })
 
   .help('help')

packages/gitmoji-changelog-core/package.json

@@ -18,6 +18,7 @@
   "homepage": "https://github.com/frinyvonnick/gitmoji-changelog#readme",
   "dependencies": {
     "concat-stream": "^1.6.2",
+    "fast-levenshtein": "^2.0.6",
     "get-pkg-repo": "^2.0.0",
     "git-raw-commits": "^2.0.0",
     "git-remote-origin-url": "^2.0.0",

packages/gitmoji-changelog-core/src/index.js

@@ -11,6 +11,7 @@ const { parseCommit } = require('./parser')
 const { getPackageInfo, getRepoInfo } = require('./metaInfo')
 const groupMapping = require('./groupMapping')
 const logger = require('./logger')
+const { groupSentencesByDistance } = require('./utils')
 
 const gitSemverTagsAsync = promisify(gitSemverTags)
 
@@ -62,24 +63,39 @@ function filterCommits(commits) {
     .filter(commit => commit.group !== 'useless')
 }
 
-async function generateVersion({ from, to, version }) {
-  const commits = filterCommits(await getCommits(from, to))
-  const lastCommitDate = getLastCommitDate(commits)
+async function generateVersion(options) {
+  const {
+    from,
+    to,
+    version,
+    groupSimilarCommits,
+  } = options
+  let commits = filterCommits(await getCommits(from, to))
+
+  if (groupSimilarCommits) {
+    commits = groupSentencesByDistance(commits.map(commit => commit.message))
+      .map(indexes => indexes.map(index => commits[index]))
+      .map(([first, ...siblings]) => ({
+        ...first,
+        siblings,
+      }))
+  }
 
   return {
     version,
-    date: version !== 'next' ? lastCommitDate : undefined,
+    date: version !== 'next' ? getLastCommitDate(commits) : undefined,
     groups: makeGroups(commits),
   }
 }
 
-async function generateVersions(tags) {
+async function generateVersions({ tags, groupSimilarCommits }) {
   let nextTag = ''
 
   return Promise.all(
     [...tags, '']
       .map(tag => {
         const params = {
+          groupSimilarCommits,
           from: tag,
           to: nextTag,
           version: nextTag ? sanitizeVersion(nextTag) : 'next',
@@ -97,7 +113,7 @@ async function generateVersions(tags) {
 }
 
 async function generateChangelog(options = {}) {
-  const { mode, release } = options
+  const { mode, release, groupSimilarCommits } = options
 
   const packageInfo = await getPackageInfo()
 
@@ -116,9 +132,10 @@ async function generateChangelog(options = {}) {
   const lastTag = head(tags)
 
   if (mode === 'init') {
-    changes = await generateVersions(tags)
+    changes = await generateVersions({ tags, groupSimilarCommits })
   } else {
     const lastChanges = await generateVersion({
+      groupSimilarCommits,
       from: lastTag,
       version,
     })

packages/gitmoji-changelog-core/src/index.spec.js

@@ -19,24 +19,26 @@ const sparklesCommit = {
   hash: 'c40ee8669ba7ea5151adc2942fa8a7fc98d9e23a',
   author: 'John Doe',
   date: '2018-08-28T10:06:00+02:00',
-  subject: ':sparkles: Upgrade brand new feature',
+  subject: ':sparkles: Add a brand new feature',
   body: 'Waouh this is awesome 2',
   emoji: '✨',
   emojiCode: 'sparkles',
   group: 'added',
-  message: 'Upgrade brand new feature',
+  message: 'Add a brand new feature',
+  siblings: [],
 }
 
 const recycleCommit = {
   hash: 'c40ee8669ba7ea5151adc2942fa8a7fc98d9e23c',
   author: 'John Doe',
   date: '2018-08-01T10:07:00+02:00',
-  subject: ':recycle: Upgrade brand new feature',
+  subject: ':recycle: Make some reworking on code',
   body: 'Waouh this is awesome 3',
   emoji: '♻️',
   emojiCode: 'recycle',
   group: 'changed',
-  message: 'Upgrade brand new feature',
+  message: 'Make some reworking on code',
+  siblings: [],
 }
 
 const secondRecycleCommit = {
@@ -49,6 +51,7 @@ const secondRecycleCommit = {
   emojiCode: 'recycle',
   group: 'changed',
   message: 'Upgrade another brand new feature',
+  siblings: [],
 }
 
 const lipstickCommit = {
@@ -61,6 +64,7 @@ const lipstickCommit = {
   emojiCode: 'lipstick',
   group: 'changed',
   message: 'Change graphics for a feature',
+  siblings: [],
 }
 
 const secondLipstickCommit = {
@@ -73,6 +77,7 @@ const secondLipstickCommit = {
   emojiCode: 'lipstick',
   group: 'changed',
   message: 'Change more graphics for a feature',
+  siblings: [],
 }
 
 describe('changelog', () => {
@@ -111,7 +116,12 @@ describe('changelog', () => {
           {
             group: 'changed',
             label: 'Changed',
-            commits: [lipstickCommit, secondLipstickCommit, recycleCommit, secondRecycleCommit],
+            commits: [
+              lipstickCommit,
+              secondLipstickCommit,
+              recycleCommit,
+              secondRecycleCommit,
+            ],
           },
         ],
       },
@@ -129,6 +139,23 @@ describe('changelog', () => {
     ])
   })
 
+  it('should group similar commits', async () => {
+    mockGroups()
+
+    gitSemverTags.mockImplementation(cb => cb(null, ['v1.0.0']))
+
+    const { changes } = await generateChangelog({ mode: 'init', groupSimilarCommits: true })
+
+    expect(changes[0].groups[0].commits).toEqual([
+      {
+        ...lipstickCommit,
+        siblings: [secondLipstickCommit],
+      },
+      recycleCommit,
+      secondRecycleCommit,
+    ])
+  })
+
   it('should filter some commits out', async () => {
     gitRawCommits.mockReset()
     mockGroup([uselessCommit, lipstickCommit])

packages/gitmoji-changelog-core/src/parser.js

@@ -47,6 +47,7 @@ function parseCommit(commit) {
     emoji,
     message,
     group,
+    siblings: [],
     body: body.join('\n'),
   }
 }

packages/gitmoji-changelog-core/src/utils.js

@@ -0,0 +1,68 @@
+const { deburr } = require('lodash')
+const levenshtein = require('fast-levenshtein')
+
+// this is a magic number, this comes from various testing
+// feel free to tweak it
+const MAX_DISTANCE_PERCENT = 0.30
+
+function groupSentencesByDistance(texts = []) {
+  const textsWithSortedWords = texts
+    .map(text => (
+      // to basic latin characters
+      deburr(text)
+        // replace specials characters by a filler
+        .replace(/[^\w\s]/gi, '▩')
+        // split words
+        .split(' ')
+        // little words are replaces by fillers
+        // this way -> we remove useless word like (a, of, etc)
+        // we keep the string length for the algorithm
+        .map(word => word.length < 4 ? Array.from({ length: word.length }).join('▩') : word)
+        // we sort words
+        .sort()
+        // we make them a sentence
+        .join('')
+    ))
+
+  const alreadyProcessedWords = new Set()
+  const keyGroups = []
+
+  for (
+    let indexesFromStart = 0;
+    indexesFromStart < textsWithSortedWords.length;
+    indexesFromStart += 1
+  ) {
+    if (!alreadyProcessedWords.has(indexesFromStart)) {
+      alreadyProcessedWords.add(indexesFromStart)
+      const group = [indexesFromStart]
+      keyGroups.push(group)
+
+      for (
+        let indexesFromNext = indexesFromStart + 1;
+        indexesFromNext < textsWithSortedWords.length;
+        indexesFromNext += 1
+      ) {
+        const textA = textsWithSortedWords[indexesFromStart]
+        const textB = textsWithSortedWords[indexesFromNext]
+        const distance = levenshtein.get(textA, textB)
+        const textAverageLength = (textA.length + textB.length) / 2
+
+        if (
+          // close distance
+          (textAverageLength * MAX_DISTANCE_PERCENT) >= distance
+          // not already in a group
+          && !alreadyProcessedWords.has(indexesFromNext)
+        ) {
+          group.push(indexesFromNext)
+          alreadyProcessedWords.add(indexesFromNext)
+        }
+      }
+    }
+  }
+
+  return keyGroups
+}
+
+module.exports = {
+  groupSentencesByDistance,
+}

packages/gitmoji-changelog-core/src/utils.spec.js

@@ -0,0 +1,18 @@
+const { groupSentencesByDistance } = require('./utils')
+
+describe('utils', () => {
+  describe('groupSentencesByDistance', () => {
+    it('should group values together', () => {
+      const messages = [
+        'add levenshtein', // 0 - group1
+        'fix a bug about failures graph', // 1 - group2
+        'levenshtein', // 2 - group1
+        'fix levenshtein', // 3 - group1
+        'nothing to group with me',
+        'fix a graph of failures bug', // 5 - group2
+      ]
+
+      expect(groupSentencesByDistance(messages)).toEqual([[0, 2, 3], [1, 5], [4]])
+    })
+  })
+})

packages/gitmoji-changelog-markdown/src/index.js

@@ -15,19 +15,24 @@ function buildMarkdownFile(changelog = {}, options = {}) {
   return markdownIncremental(changelog, options)
 }
 
-function toMarkdown({ meta, changes }, { author }) {
-  const template = fs.readFileSync(MARKDOWN_TEMPLATE, 'utf-8')
-
-  const compileTemplate = handlebars.compile(template)
+function mapCommit(meta, options) {
+  const { author } = options
 
-  const changelog = update(changes, '[:].groups[:].commits[:]', commit => ({
+  return commit => ({
     ...commit,
     hash: getShortHash(commit.hash, meta.repository),
     subject: autolink(commit.subject, meta.repository),
     message: autolink(commit.message, meta.repository),
     body: autolink(commit.body, meta.repository),
     author: author ? commit.author : null,
-  }))
+    siblings: commit.siblings.map(mapCommit(meta, options)),
+  })
+}
+
+function toMarkdown({ meta, changes }, options) {
+  const template = fs.readFileSync(MARKDOWN_TEMPLATE, 'utf-8')
+  const compileTemplate = handlebars.compile(template)
+  const changelog = update(changes, '[:].groups[:].commits[:]', mapCommit(meta, options))
 
   return compileTemplate({ changelog })
 }

packages/gitmoji-changelog-markdown/src/index.spec.js

@@ -40,6 +40,7 @@ describe('Markdown converter', () => {
                   emoji: '♻️',
                   message: 'Upgrade brand new feature',
                   body: 'Waouh this is awesome 3',
+                  siblings: [],
                 },
               ],
             },
@@ -61,6 +62,7 @@ describe('Markdown converter', () => {
                   emoji: '✨',
                   message: 'Upgrade brand new feature',
                   body: 'Waouh this is awesome 2',
+                  siblings: [],
                 },
               ],
             },
@@ -124,6 +126,7 @@ describe('Markdown converter', () => {
                   emoji: '♻️',
                   message: 'Upgrade brand new feature',
                   body: 'Waouh this is awesome 3',
+                  siblings: [],
                 },
               ],
             },
@@ -208,6 +211,7 @@ I am the last version
                   emoji: '♻️',
                   message: 'Upgrade brand new feature',
                   body: 'Waouh this is awesome 3',
+                  siblings: [],
                 },
               ],
             },

packages/gitmoji-changelog-markdown/src/template.md

@@ -7,6 +7,9 @@
 
 {{#each commits}}
 - {{emoji}} {{message}} [{{hash}}]{{#if author}} (by {{author}}){{/if}}
+  {{#each siblings}}
+    * {{emoji}} {{message}} ({{hash}})
+  {{/each}}
 {{/each}}
 
 {{/each}}