Merge branch 'master' into readme-updates

bin/cml.js

@@ -10,24 +10,7 @@ const yargs = require('yargs');
 const CML = require('../src/cml').default;
 const { jitsuEventPayload, send } = require('../src/analytics');
 
-const setupOpts = (opts) => {
-  const legacyEnvironmentVariables = {
-    TB_CREDENTIALS: 'CML_TENSORBOARD_DEV_CREDENTIALS',
-    DOCKER_MACHINE: 'CML_RUNNER_DOCKER_MACHINE',
-    RUNNER_IDLE_TIMEOUT: 'CML_RUNNER_IDLE_TIMEOUT',
-    RUNNER_LABELS: 'CML_RUNNER_LABELS',
-    RUNNER_SINGLE: 'CML_RUNNER_SINGLE',
-    RUNNER_REUSE: 'CML_RUNNER_REUSE',
-    RUNNER_NO_RETRY: 'CML_RUNNER_NO_RETRY',
-    RUNNER_DRIVER: 'CML_RUNNER_DRIVER',
-    RUNNER_REPO: 'CML_RUNNER_REPO',
-    RUNNER_PATH: 'CML_RUNNER_PATH'
-  };
-
-  for (const [oldName, newName] of Object.entries(legacyEnvironmentVariables)) {
-    if (process.env[oldName]) process.env[newName] = process.env[oldName];
-  }
-
+const aliasLegacyEnvironmentVariables = () => {
   const legacyEnvironmentPrefixes = {
     CML_CI: 'CML_REPO',
     CML_PUBLISH: 'CML_ASSET',
@@ -46,6 +29,25 @@ const setupOpts = (opts) => {
     }
   }
 
+  const legacyEnvironmentVariables = {
+    TB_CREDENTIALS: 'CML_TENSORBOARD_CREDENTIALS',
+    DOCKER_MACHINE: 'CML_RUNNER_DOCKER_MACHINE',
+    RUNNER_IDLE_TIMEOUT: 'CML_RUNNER_IDLE_TIMEOUT',
+    RUNNER_LABELS: 'CML_RUNNER_LABELS',
+    RUNNER_SINGLE: 'CML_RUNNER_SINGLE',
+    RUNNER_REUSE: 'CML_RUNNER_REUSE',
+    RUNNER_NO_RETRY: 'CML_RUNNER_NO_RETRY',
+    RUNNER_DRIVER: 'CML_RUNNER_DRIVER',
+    RUNNER_REPO: 'CML_RUNNER_REPO',
+    RUNNER_PATH: 'CML_RUNNER_PATH'
+  };
+
+  for (const [oldName, newName] of Object.entries(legacyEnvironmentVariables)) {
+    if (process.env[oldName]) process.env[newName] = process.env[oldName];
+  }
+};
+
+const setupOpts = (opts) => {
   const { markdownfile } = opts;
   opts.markdownFile = markdownfile;
   opts.cml = new CML(opts);
@@ -117,6 +119,7 @@ const handleError = (message, error) => {
 };
 
 (async () => {
+  aliasLegacyEnvironmentVariables();
   setupLogger({ log: 'debug' });
 
   try {

bin/cml/comment/create.js

@@ -68,6 +68,13 @@ exports.options = kebabcaseKeys({
       'Avoid watermark; CML needs a watermark to be able to distinguish CML comments from others',
     hidden: true,
     telemetryData: 'name'
+  },
+  watermarkTitle: {
+    type: 'string',
+    description:
+      'Hidden comment marker (used for targeting in subsequent `cml comment update`); "{workflow}" & "{run}" are auto-replaced',
+    default: '',
+    conflicts: ['rmWatermark']
   }
 });
 exports.DOCSURL = DOCSURL;

bin/cml/comment/create.test.js

@@ -3,7 +3,6 @@ const { exec } = require('../../../src/utils');
 describe('Comment integration tests', () => {
   test('cml send-comment --help', async () => {
     const output = await exec(`node ./bin/cml.js send-comment --help`);
-
     expect(output).toMatchInlineSnapshot(`
       "cml.js send-comment <markdown file>
 
@@ -30,7 +29,10 @@ describe('Comment integration tests', () => {
                                            [string] [default: \\"https://asset.cml.dev\\"]
         --publish-native, --native  Uses driver's native capabilities to upload assets
                                     instead of CML's storage; not available on GitHub
-                                                                             [boolean]"
+                                                                             [boolean]
+        --watermark-title           Hidden comment marker (used for targeting in
+                                    subsequent \`cml comment update\`); \\"{workflow}\\" &
+                                    \\"{run}\\" are auto-replaced   [string] [default: \\"\\"]"
     `);
   });
 });

bin/cml/repo/prepare.js

@@ -20,10 +20,15 @@ exports.builder = (yargs) =>
     .options(exports.options);
 
 exports.options = kebabcaseKeys({
+  fetchDepth: {
+    type: 'number',
+    description: 'Number of commits to fetch (use `0` for all branches & tags)'
+  },
   unshallow: {
     type: 'boolean',
     description:
-      'Fetch as much as possible, converting a shallow repository to a complete one'
+      'Fetch as much as possible, converting a shallow repository to a complete one',
+    hidden: true
   },
   userEmail: {
     type: 'string',

bin/cml/repo/prepare.test.js

@@ -20,10 +20,10 @@ describe('CML e2e', () => {
         --help    Show help                                                  [boolean]
 
       Options:
-        --unshallow   Fetch as much as possible, converting a shallow repository to a
-                      complete one                                           [boolean]
-        --user-email  Git user email         [string] [default: \\"[email protected]\\"]
-        --user-name   Git user name                  [string] [default: \\"Olivaw[bot]\\"]"
+        --fetch-depth  Number of commits to fetch (use \`0\` for all branches & tags)
+                                                                              [number]
+        --user-email   Git user email        [string] [default: \\"[email protected]\\"]
+        --user-name    Git user name                 [string] [default: \\"Olivaw[bot]\\"]"
     `);
   });
 });

bin/cml/runner/launch.js

@@ -495,7 +495,8 @@ exports.options = kebabcaseKeys({
   cloudType: {
     type: 'string',
     description:
-      'Instance type. Choices: [m, l, xl]. Also supports native types like i.e. t2.micro'
+      'Instance type. Choices: [m, l, xl]. Also supports native types like i.e. t2.micro',
+    telemetryData: 'full'
   },
   cloudPermissionSet: {
     type: 'string',
@@ -518,7 +519,8 @@ exports.options = kebabcaseKeys({
     type: 'string',
     description:
       'GPU type. Choices: k80, v100, or native types e.g. nvidia-tesla-t4',
-    coerce: (val) => (val === 'nogpu' ? undefined : val)
+    coerce: (val) => (val === 'nogpu' ? undefined : val),
+    telemetryData: 'full'
   },
   cloudHddSize: {
     type: 'number',

src/cml.js

@@ -23,6 +23,7 @@ const {
 
 const { GITHUB_REPOSITORY, CI_PROJECT_URL, BITBUCKET_REPO_UUID } = process.env;
 
+const WATERMARK_IMAGE = 'https://cml.dev/watermark.png';
 const GIT_USER_NAME = 'Olivaw[bot]';
 const GIT_USER_EMAIL = '[email protected]';
 const GIT_REMOTE = 'origin';
@@ -164,26 +165,55 @@ class CML {
     const triggerSha = await this.triggerSha();
     const {
       commitSha: inCommitSha = triggerSha,
-      rmWatermark,
-      update,
+      markdownFile,
       pr,
       publish,
       publishUrl,
-      markdownFile,
       report: testReport,
+      rmWatermark,
+      triggerFile,
+      update,
       watch,
-      triggerFile
+      watermarkTitle
     } = opts;
 
+    const drv = this.getDriver();
+
     const commitSha =
       (await this.revParse({ ref: inCommitSha })) || inCommitSha;
 
     if (rmWatermark && update)
       throw new Error('watermarks are mandatory for updateable comments');
 
+    // Create the watermark.
+    const genWatermark = (opts = {}) => {
+      const { label = '', workflow, run } = opts;
+      // Replace {workflow} and {run} placeholders in label with actual values.
+      const lbl = label.replace('{workflow}', workflow).replace('{run}', run);
+
+      let title = `CML watermark ${lbl}`.trim();
+      // Github appears to escape underscores and asterisks in markdown content.
+      // Without escaping them, the watermark content in comments retrieved
+      // from github will not match the input.
+      const patterns = [
+        [/_/g, '\\_'], // underscore
+        [/\*/g, '\\*'], // asterisk
+        [/\[/g, '\\['], // opening square bracket
+        [/</g, '\\<'] // opening angle bracket
+      ];
+      title = patterns.reduce(
+        (label, pattern) => label.replace(pattern[0], pattern[1]),
+        title
+      );
+      return `![](${WATERMARK_IMAGE} "${title}")`;
+    };
     const watermark = rmWatermark
       ? ''
-      : '![CML watermark](https://raw.githubusercontent.com/iterative/cml/master/assets/watermark.svg)';
+      : genWatermark({
+          label: watermarkTitle,
+          workflow: drv.workflowId,
+          run: drv.runId
+        });
 
     let userReport = testReport;
     try {
@@ -195,15 +225,17 @@ class CML {
     }
 
     let report = `${userReport}\n\n${watermark}`;
-    const drv = this.getDriver();
 
     const publishLocalFiles = async (tree) => {
       const nodes = [];
 
       visit(tree, ['definition', 'image', 'link'], (node) => nodes.push(node));
 
+      const isWatermark = (node) => {
+        return node.title && node.title.startsWith('CML watermark');
+      };
       const visitor = async (node) => {
-        if (node.url && node.alt !== 'CML watermark') {
+        if (node.url && !isWatermark(node)) {
           const absolutePath = path.resolve(
             path.dirname(markdownFile),
             node.url
@@ -264,7 +296,7 @@ class CML {
     let comment;
     const updatableComment = (comments) => {
       return comments.reverse().find(({ body }) => {
-        return body.includes('watermark.svg');
+        return body.includes(watermark);
       });
     };
 
@@ -469,15 +501,19 @@ class CML {
       userName = GIT_USER_NAME,
       remote = GIT_REMOTE
     } = opts;
+    const { fetchDepth = unshallow ? 0 : undefined } = opts;
 
     const driver = this.getDriver();
     await exec(await driver.updateGitConfig({ userName, userEmail, remote }));
-    if (unshallow) {
-      if ((await exec('git rev-parse --is-shallow-repository')) === 'true') {
-        await exec('git fetch --unshallow');
+    if (fetchDepth !== undefined) {
+      if (fetchDepth <= 0) {
+        if ((await exec('git rev-parse --is-shallow-repository')) === 'true') {
+          return await exec('git fetch --all --unshallow');
+        }
+      } else {
+        return await exec(`git fetch --all --depth=${fetchDepth}`);
       }
     }
-    await exec('git fetch --all');
   }
 
   async prCreate(opts = {}) {

src/drivers/bitbucket_cloud.js

@@ -8,8 +8,12 @@ const ProxyAgent = require('proxy-agent');
 
 const { fetchUploadData, exec, gpuPresent, sleep } = require('../utils');
 
-const { BITBUCKET_COMMIT, BITBUCKET_BRANCH, BITBUCKET_PIPELINE_UUID } =
-  process.env;
+const {
+  BITBUCKET_COMMIT,
+  BITBUCKET_BRANCH,
+  BITBUCKET_PIPELINE_UUID,
+  BITBUCKET_BUILD_NUMBER
+} = process.env;
 
 class BitbucketCloud {
   constructor(opts = {}) {
@@ -421,6 +425,14 @@ class BitbucketCloud {
     return command;
   }
 
+  get workflowId() {
+    return BITBUCKET_PIPELINE_UUID;
+  }
+
+  get runId() {
+    return BITBUCKET_BUILD_NUMBER;
+  }
+
   get sha() {
     return BITBUCKET_COMMIT;
   }

src/drivers/github.js

@@ -18,14 +18,15 @@ const CHECK_TITLE = 'CML Report';
 process.env.RUNNER_ALLOW_RUNASROOT = 1;
 
 const {
-  GITHUB_REPOSITORY,
-  GITHUB_SHA,
-  GITHUB_REF,
-  GITHUB_HEAD_REF,
+  CI,
   GITHUB_EVENT_NAME,
+  GITHUB_HEAD_REF,
+  GITHUB_REF,
+  GITHUB_REPOSITORY,
   GITHUB_RUN_ID,
+  GITHUB_SHA,
   GITHUB_TOKEN,
-  CI,
+  GITHUB_WORKFLOW,
   TPI_TASK
 } = process.env;
 
@@ -705,6 +706,14 @@ class Github {
     return command;
   }
 
+  get workflowId() {
+    return GITHUB_WORKFLOW;
+  }
+
+  get runId() {
+    return GITHUB_RUN_ID;
+  }
+
   warn(message) {
     console.error(`::warning::${message}`);
   }

src/drivers/gitlab.js

@@ -11,7 +11,8 @@ const winston = require('winston');
 
 const { fetchUploadData, download, gpuPresent } = require('../utils');
 
-const { IN_DOCKER, CI_PIPELINE_ID } = process.env;
+const { CI_JOB_ID, CI_PIPELINE_ID, IN_DOCKER } = process.env;
+
 const API_VER = 'v4';
 class Gitlab {
   constructor(opts = {}) {
@@ -444,6 +445,14 @@ class Gitlab {
     return command;
   }
 
+  get workflowId() {
+    return CI_PIPELINE_ID;
+  }
+
+  get runId() {
+    return CI_JOB_ID;
+  }
+
   get sha() {
     return process.env.CI_COMMIT_SHA;
   }

src/utils.js

@@ -136,14 +136,14 @@ const isProcRunning = async (opts) => {
 };
 
 const watermarkUri = ({ uri, type } = {}) => {
-  return uriParmam({ uri, param: 'cml', value: type });
+  return uriParam({ uri, param: 'cml', value: type });
 };
 
 const preventcacheUri = ({ uri } = {}) => {
-  return uriParmam({ uri, param: 'cache-bypass', value: uuid.v4() });
+  return uriParam({ uri, param: 'cache-bypass', value: uuid.v4() });
 };
 
-const uriParmam = (opts = {}) => {
+const uriParam = (opts = {}) => {
   const { uri, param, value } = opts;
   const url = new URL(uri);
   url.searchParams.set(param, value);