feat(core): Add organization-level activity fetching and RSS parsing

This commit is contained in:
Philipp Kunz 2024-12-13 19:51:42 +01:00
parent c1e15ab47c
commit 76c662356e
6 changed files with 163 additions and 53 deletions

View File

@ -1,5 +1,13 @@
# Changelog
## 2024-12-13 - 1.2.0 - feat(core)
Add organization-level activity fetching and RSS parsing
- Integrated smartxml package for XML parsing.
- Implemented fetching of all organizations within a Gitea instance.
- Added functionality to check new activities in organization RSS feeds.
- Enhanced fetching logic to include repository commits and tags.
## 2024-12-13 - 1.1.0 - feat(core)
Add tracking of commits published on npm

View File

@ -23,7 +23,8 @@
},
"dependencies": {
"@push.rocks/qenv": "^6.1.0",
"@push.rocks/smartnpm": "^2.0.4"
"@push.rocks/smartnpm": "^2.0.4",
"@push.rocks/smartxml": "^1.0.8"
},
"repository": {
"type": "git",

3
pnpm-lock.yaml generated
View File

@ -14,6 +14,9 @@ importers:
'@push.rocks/smartnpm':
specifier: ^2.0.4
version: 2.0.4
'@push.rocks/smartxml':
specifier: ^1.0.8
version: 1.0.8
devDependencies:
'@git.zone/tsbuild':
specifier: ^2.1.25

View File

@ -3,6 +3,6 @@
*/
export const commitinfo = {
name: '@foss.global/codefeed',
version: '1.1.0',
version: '1.2.0',
description: 'a module for creating feeds for code development'
}

View File

@ -1,8 +1,10 @@
// @push.rocks
import * as qenv from '@push.rocks/qenv'
import * as smartnpm from '@push.rocks/smartnpm'
import * as qenv from '@push.rocks/qenv';
import * as smartnpm from '@push.rocks/smartnpm';
import * as smartxml from '@push.rocks/smartxml';
export {
qenv,
smartnpm,
smartxml,
}

View File

@ -47,14 +47,81 @@ interface CommitResult {
export class CodeFeed {
private baseUrl: string;
private token?: string;
private npmRegistry = new plugins.smartnpm.NpmRegistry();
private npmRegistry = new plugins.smartnpm.NpmRegistry();
private smartxmlInstance = new plugins.smartxml.SmartXml();
private lastRunTimestamp: string;
constructor(baseUrl: string, token?: string) {
constructor(baseUrl: string, token?: string, lastRunTimestamp?: string) {
this.baseUrl = baseUrl;
this.token = token;
console.log('CodeFeed initialized');
this.lastRunTimestamp = lastRunTimestamp || new Date(Date.now() - 24 * 60 * 60 * 1000).toISOString();
console.log('CodeFeed initialized with last run timestamp:', this.lastRunTimestamp);
}
/**
* Fetch all organizations from the Gitea instance.
*/
private async fetchAllOrganizations(): Promise<string[]> {
const url = `${this.baseUrl}/api/v1/orgs`;
const response = await fetch(url, {
headers: this.token ? { Authorization: `token ${this.token}` } : {},
});
if (!response.ok) {
throw new Error(`Failed to fetch organizations: ${response.statusText}`);
}
const data: { username: string }[] = await response.json();
return data.map((org) => org.username);
}
/**
* Fetch organization-level activity RSS feed.
*/
private async fetchOrgRssFeed(optionsArg: {
orgName: string,
repoName?: string,
}): Promise<any[]> {
let rssUrl: string
if (optionsArg.orgName && !optionsArg.repoName) {
rssUrl = `${this.baseUrl}/${optionsArg.orgName}.atom`;
} else if (optionsArg.orgName && optionsArg.repoName) {
rssUrl = `${this.baseUrl}/${optionsArg.orgName}/${optionsArg.repoName}.atom`;
}
const response = await fetch(rssUrl);
if (!response.ok) {
throw new Error(`Failed to fetch RSS feed for organization ${optionsArg.orgName}/${optionsArg.repoName}: ${response.statusText}`);
}
const rssText = await response.text();
// Parse the Atom feed using fast-xml-parser
const rssData = this.smartxmlInstance.parseXmlToObject(rssText);
// Return the <entry> elements from the feed
return rssData.feed.entry || [];
}
/**
* Check if the organization's RSS feed has any new activities since the last run.
*/
private async hasNewActivity(optionsArg: {
orgName: string,
repoName?: string,
}): Promise<boolean> {
const entries = await this.fetchOrgRssFeed(optionsArg);
// Filter entries to find new activities since the last run
return entries.some((entry: any) => {
const updated = new Date(entry.updated);
return updated > new Date(this.lastRunTimestamp);
});
}
/**
* Fetch all repositories accessible to the token/user.
*/
private async fetchAllRepositories(): Promise<Repository[]> {
let page = 1;
const allRepos: Repository[] = [];
@ -65,7 +132,7 @@ export class CodeFeed {
url.searchParams.set('page', page.toString());
const resp = await fetch(url.href, {
headers: this.token ? { 'Authorization': `token ${this.token}` } : {}
headers: this.token ? { 'Authorization': `token ${this.token}` } : {},
});
if (!resp.ok) {
@ -84,6 +151,9 @@ export class CodeFeed {
return allRepos;
}
/**
* Fetch all tags for a given repository.
*/
private async fetchTags(owner: string, repo: string): Promise<Set<string>> {
let page = 1;
const tags: Tag[] = [];
@ -94,9 +164,9 @@ export class CodeFeed {
url.searchParams.set('page', page.toString());
const resp = await fetch(url.href, {
headers: this.token ? { 'Authorization': `token ${this.token}` } : {}
headers: this.token ? { 'Authorization': `token ${this.token}` } : {},
});
if (!resp.ok) {
console.error(`Failed to fetch tags for ${owner}/${repo}: ${resp.status} ${resp.statusText} at ${url.href}`);
throw new Error(`Failed to fetch tags for ${owner}/${repo}: ${resp.statusText}`);
@ -121,6 +191,9 @@ export class CodeFeed {
return taggedCommitShas;
}
/**
* Fetch commits from the last 24 hours for a repository.
*/
private async fetchRecentCommitsForRepo(owner: string, repo: string): Promise<Commit[]> {
const twentyFourHoursAgo = new Date(Date.now() - 24 * 60 * 60 * 1000);
let page = 1;
@ -128,12 +201,13 @@ export class CodeFeed {
while (true) {
const url = new URL(`${this.baseUrl}/api/v1/repos/${owner}/${repo}/commits`);
url.searchParams.set('limit', '1');
url.searchParams.set('limit', '50');
url.searchParams.set('page', page.toString());
const resp = await fetch(url.href, {
headers: this.token ? { 'Authorization': `token ${this.token}` } : {}
headers: this.token ? { 'Authorization': `token ${this.token}` } : {},
});
if (!resp.ok) {
console.error(`Failed to fetch commits for ${owner}/${repo}: ${resp.status} ${resp.statusText} at ${url.href}`);
throw new Error(`Failed to fetch commits for ${owner}/${repo}: ${resp.statusText}`);
@ -149,7 +223,6 @@ export class CodeFeed {
if (commitDate > twentyFourHoursAgo) {
recentCommits.push(commit);
} else {
// If we encounter a commit older than 24 hours, we can stop fetching more pages
return recentCommits;
}
}
@ -160,24 +233,56 @@ export class CodeFeed {
return recentCommits;
}
/**
* Fetch all commits by querying all organizations.
*/
public async fetchAllCommitsFromInstance(): Promise<CommitResult[]> {
const repos = await this.fetchAllRepositories();
const skippedRepos: string[] = [];
console.log(`Found ${repos.length} repositories`);
const orgs = await this.fetchAllOrganizations();
console.log(`Found ${orgs.length} organizations`);
let allCommits: CommitResult[] = [];
for (const r of repos) {
const org = r.owner.login;
const repo = r.name;
console.log(`Processing repository ${org}/${repo}`);
for (const orgName of orgs) {
console.log(`Checking activity for organization: ${orgName}`);
try {
const taggedCommitShas = await this.fetchTags(org, repo);
const commits = await this.fetchRecentCommitsForRepo(org, repo);
console.log(`${org}/${repo} -> Found ${commits.length} commits`);
const commitResults: CommitResult[] = [];
for (const c of commits) {
const commit: CommitResult = {
const hasActivity = await this.hasNewActivity({
orgName,
});
if (!hasActivity) {
console.log(`No new activity for organization: ${orgName}`);
continue;
}
} catch (error) {
console.error(`Error fetching activity for organization ${orgName}:`, error.message);
continue;
}
console.log(`New activity detected for organization: ${orgName}. Processing repositories...`);
const repos = await this.fetchAllRepositories();
for (const r of repos.filter((repo) => repo.owner.login === orgName)) {
try {
const hasActivity = await this.hasNewActivity({
orgName,
repoName: r.name,
});
if (!hasActivity) {
console.log(`No new activity for repository: ${orgName}/${r.name}`);
continue;
}
} catch (error) {
console.error(`Error fetching activity for repository ${orgName}/${r.name}:`, error.message);
continue;
}
const org = r.owner.login;
const repo = r.name;
console.log(`Processing repository ${org}/${repo}`);
try {
const taggedCommitShas = await this.fetchTags(org, repo);
const commits = await this.fetchRecentCommitsForRepo(org, repo);
const commitResults = commits.map((c) => ({
baseUrl: this.baseUrl,
org,
repo,
@ -186,40 +291,32 @@ export class CodeFeed {
commitMessage: c.commit.message,
tagged: taggedCommitShas.has(c.sha),
publishedOnNpm: false,
}
commitResults.push(commit);
}
}));
if (commitResults.length > 0) {
try {
const packageInfo = await this.npmRegistry.getPackageInfo(`@${org}/${repo}`);
for (const commit of commitResults.filter(c => c.tagged)) {
const correspondingVersion = packageInfo.allVersions.find(versionArg => {
return versionArg.version === commit.commitMessage.replace('\n', '');
});
if (correspondingVersion) {
commit.publishedOnNpm = true;
if (commitResults.length > 0) {
try {
const packageInfo = await this.npmRegistry.getPackageInfo(`@${org}/${repo}`);
for (const commit of commitResults.filter((c) => c.tagged)) {
const correspondingVersion = packageInfo.allVersions.find((versionArg) => {
return versionArg.version === commit.commitMessage.replace('\n', '');
});
if (correspondingVersion) {
commit.publishedOnNpm = true;
}
}
} catch (error) {
console.error(`Failed to fetch package info for ${org}/${repo}:`, error.message);
}
} catch (error: any) {
console.error(`Failed to fetch package info for ${org}/${repo}:`, error.message);
continue;
}
}
allCommits.push(...commitResults);
} catch (error: any) {
skippedRepos.push(`${org}/${repo}`);
console.error(`Skipping repository ${org}/${repo} due to error:`, error.message);
continue;
allCommits.push(...commitResults);
} catch (error) {
console.error(`Skipping repository ${org}/${repo} due to error:`, error.message);
}
}
}
console.log(`Found ${allCommits.length} relevant commits`);
console.log(`Skipped ${skippedRepos.length} repositories due to errors`);
for (const s of skippedRepos) {
console.log(`Skipped ${s}`);
}
console.log(`Processed ${allCommits.length} commits in total.`);
for (const c of allCommits) {
console.log(`______________________________________________________
Commit ${c.hash} by ${c.org}/${c.repo} at ${c.timestamp}
@ -227,7 +324,6 @@ export class CodeFeed {
Published on npm: ${c.publishedOnNpm}
`);
}
return allCommits;
}
}