pr-prism 1.1.3 → 1.1.6

package/scripts/lib/sanitize.ts

@@ -0,0 +1,152 @@
+export const KNOWN_BOTS = [
+  "github-actions",
+  "dependabot",
+  "coderabbitai",
+  "changeset-bot",
+  "codeantai",
+];
+
+export const NOISE_DOMAINS = [
+  "twitter.com/intent",
+  "x.com/intent",
+  "reddit.com/submit",
+  "linkedin.com/sharing",
+  "app.codeant.ai",
+  "codeant.ai/feedback",
+  "cubic.dev",
+  "qodo.ai/wp-content",
+  "dashboard.gitguardian.com",
+  "blog.gitguardian.com",
+  "docs.gitguardian.com",
+];
+
+export function isBot(login: string): boolean {
+  const l = login.toLowerCase();
+  return l.endsWith("[bot]") || KNOWN_BOTS.some((b) => l.includes(b));
+}
+
+export function renderSuggestions(body: string): string {
+  return body.replace(/```suggestion\r?\n([\s\S]*?)```/g, (_, code: string) => {
+    const lines = code
+      .trimEnd()
+      .split(/\r?\n/)
+      .map((l: string) => `+ ${l}`)
+      .join("\n");
+    return `\n**SUGGESTED CHANGE:**\n\`\`\`diff\n${lines}\n\`\`\`\n`;
+  });
+}
+
+const NOISE_LINE_PATTERNS = [
+  /thanks for using/i,
+  /free for open-source/i,
+  /help us grow by sharing/i,
+  /Generated by.*CodeAnt/i,
+  /CodeAnt AI (?:is reviewing|finished reviewing)/i,
+  /Check back in a few minutes/i,
+  /AI review agent is analyzing/i,
+  /Looking for bugs\?/i,
+  /Validate the correctness of the flagged issue/i,
+  /This is a comment left during a code review/i,
+  /implement it and please make it concise/i,
+];
+
+export function stripNoise(body: string): string {
+  let result = body;
+
+  result = result.replace(/<!--[\s\S]*?-->/g, "");
+  result = result.replace(/<picture>[\s\S]*?<\/picture>/gi, "");
+  result = result.replace(/<sub>[\s\S]*?<\/sub>/gi, "");
+  result = result.replace(/<sup>[\s\S]*?<\/sup>/gi, "");
+  result = result.replace(/<s>[\s\S]*?<\/s>/gi, "");
+  result = result.replace(/<img\s[^>]*\/?>/gi, "");
+  result = result.replace(/<source\s[^>]*\/?>/gi, "");
+
+  result = result.replace(/```mermaid[\s\S]*?```/g, "");
+
+  result = result.replace(/^.*\u24D8.*$/gm, "");
+
+  result = result.replace(
+    /<a\s[^>]*href=['"]([^'"]+)['"][^>]*>[\s\S]*?<\/a>/gi,
+    (match, url: string) => {
+      if (NOISE_DOMAINS.some((d) => url.includes(d))) return "";
+      const innerText = match.replace(/<[^>]*>/g, "").trim();
+      if (!innerText) return "";
+      return innerText;
+    },
+  );
+
+  result = result.replace(/<\/?(details|summary|pre|table|thead|tbody)\b[^>]*>/gi, "");
+  result = result.replace(/<tr\b[^>]*>/gi, "");
+  result = result.replace(/<\/tr>/gi, "\n");
+  result = result.replace(/<\/?(td|th)\b[^>]*>/gi, " ");
+  result = result.replace(/<\/?(h[1-6])\b[^>]*>/gi, "");
+  result = result.replace(/<br\s*\/?>/gi, "\n");
+  result = result.replace(/<hr\s*\/?>/gi, "");
+  result = result.replace(/<\/?(b|i|strong|em|ins)\s*>/gi, "");
+  result = result.replace(/<code>([\s\S]*?)<\/code>/gi, "`$1`");
+
+  result = result.replace(/<\/?file context>/gi, "");
+  result = result.replace(/<\/?comment>/gi, "");
+  result = result.replace(/<violation\b[^>]*>/gi, "");
+  result = result.replace(/<\/violation>/gi, "");
+  result = result.replace(/<file\b[^>]*>/gi, "");
+  result = result.replace(/<\/file>/gi, "");
+
+  // Decode HTML entities
+  result = result.replace(/&amp;/g, "&");
+  result = result.replace(/&lt;/g, "<");
+  result = result.replace(/&gt;/g, ">");
+  result = result.replace(/&quot;/g, '"');
+  result = result.replace(/&nbsp;/g, " ");
+
+  // Strip block quote markers
+  result = result.replace(/^>\s?/gm, "");
+
+  // Strip emoji
+  result = result.replace(/[\p{Extended_Pictographic}\u200D\uFE0F]/gu, "");
+
+  // Strip GitHub blob/diff/files links — path is already in comment header
+  result = result.replace(
+    /`?\[([^\]]*)\]\(https:\/\/github\.com\/[^)]*(?:\/files#diff-|\/blob\/|\/pull\/)[^)]*\)`?/g,
+    "$1",
+  );
+
+  // Strip noise-domain markdown links
+  result = result.replace(
+    /\[([^\]]*)\]\((https?:\/\/[^)]+)\)/g,
+    (match, _text: string, url: string) =>
+      NOISE_DOMAINS.some((d) => url.includes(d)) ? "" : match,
+  );
+
+  // Strip promotional/status noise lines
+  result = result.replace(/^.*$/gm, (line) =>
+    NOISE_LINE_PATTERNS.some((p) => p.test(line)) ? "" : line,
+  );
+
+  // Strip standalone section headers (noise when content follows)
+  result = result.replace(
+    /^\s*(?:Agent prompt|Walkthroughs|View more|Prompt for AI agents?|Prompt for AI Agent|Review Summary by Qodo|Code Review by Qodo|Nitpicks|Recommended areas for review)\s*(?:\([^)]*\))?\s*$/gim,
+    "",
+  );
+
+  result = result.replace(/^[\s\u00B7|\u2014\-]+$/gm, "");
+  result = result.replace(/\n{3,}/g, "\n\n");
+
+  return result.trim();
+}
+
+// ~4 chars per token is the standard LLM approximation
+const CHARS_PER_TOKEN = 4;
+
+export function estimateTokens(text: string): number {
+  return Math.ceil(text.length / CHARS_PER_TOKEN);
+}
+
+export function formatTokenSummary(rawChars: number, cleanedChars: number): string {
+  const rawTokens = Math.ceil(rawChars / CHARS_PER_TOKEN);
+  const cleanedTokens = Math.ceil(cleanedChars / CHARS_PER_TOKEN);
+  const removedTokens = Math.max(0, rawTokens - cleanedTokens);
+  const pctSaved = rawTokens > 0 ? Math.round((removedTokens / rawTokens) * 100) : 0;
+
+  return `Tokens: ${rawTokens.toLocaleString()} raw > ${cleanedTokens.toLocaleString()} clean (${pctSaved}% saved)`;
+}

package/scripts/lib/scrape-issues.ts

@@ -0,0 +1,204 @@
+import { mkdirSync } from "node:fs";
+import { writeFileSync } from "node:fs";
+import { join } from "node:path";
+import { defineWizard, runWizard } from "grimoire-wizard";
+import { detectRepo, parseGhUrl, loadCache, saveCache, fetchGraphQL, ensureGitignore, run } from "./shared.js";
+import { isBot, stripNoise, formatTokenSummary } from "./sanitize.js";
+
+const OUT_DIR = "issues";
+const CACHE_FILE = join(OUT_DIR, ".scraped-ids.json");
+
+const ISSUE_GRAPHQL = `
+query($owner: String!, $repo: String!, $issueNumber: Int!) {
+  repository(owner: $owner, name: $repo) {
+    issue(number: $issueNumber) {
+      title
+      body
+      state
+      author { login }
+      labels(first: 20) { nodes { name } }
+      comments(first: 100) {
+        nodes { databaseId author { login } body }
+      }
+    }
+  }
+}`.trim();
+
+interface IssueComment { databaseId: number; author: { login: string }; body: string; }
+interface IssueLabel { name: string; }
+
+interface IssuePayload {
+  data: {
+    repository: {
+      issue: {
+        title: string;
+        body: string;
+        state: string;
+        author: { login: string };
+        labels: { nodes: IssueLabel[] };
+        comments: { nodes: IssueComment[] };
+      };
+    };
+  };
+}
+
+interface IssueListItem { number: number; title: string; author: { login: string }; labels: { name: string }[]; }
+
+const VALID_STATES = new Set(["open", "closed", "all"]);
+
+function listIssues(owner: string, repo: string, state: string): IssueListItem[] {
+  if (!VALID_STATES.has(state)) {
+    console.error(`Invalid --state "${state}". Must be: open, closed, or all`);
+    process.exit(1);
+  }
+  const out = run(`gh issue list --repo ${owner}/${repo} --state ${state} --json number,title,author,labels --limit 50`);
+  return JSON.parse(out) as IssueListItem[];
+}
+
+async function selectIssue(issues: IssueListItem[]): Promise<number> {
+  if (issues.length === 0) { console.log("No issues found."); process.exit(0); }
+
+  const config = defineWizard({
+    meta: { name: "pr-prism" },
+    steps: [
+      {
+        id: "issue",
+        type: "select" as const,
+        message: "Select an issue:",
+        options: issues.map((i) => {
+          const labels = i.labels.map((l) => l.name).join(", ");
+          const labelSuffix = labels ? `  [${labels}]` : "";
+          return {
+            value: String(i.number),
+            label: `#${i.number}  ${i.title}  (${i.author.login})${labelSuffix}`,
+          };
+        }),
+      },
+    ],
+  });
+
+  const answers = await runWizard(config, { onCancel: () => process.exit(0) });
+  const selected = answers?.issue;
+  if (!selected) process.exit(0);
+  return parseInt(selected as string, 10);
+}
+
+interface AppendResult { text: string; rawChars: number; cleanedChars: number; }
+
+function appendBody(
+  out: string,
+  body: string,
+  login: string,
+  issueNumber: number,
+  labels: IssueLabel[],
+): AppendResult {
+  const rawChars = body.length;
+  const cleaned = stripNoise(body).trim();
+  if (!cleaned) return { text: out, rawChars, cleanedChars: 0 };
+  const labelStr = labels.length > 0
+    ? " · labels: " + labels.map((l) => `\`${l.name}\``).join(", ")
+    : "";
+  const labelTag = labels.length > 0
+    ? " [" + labels.map((l) => l.name).join(", ") + "]"
+    : "";
+  const appended = `ISSUE #${issueNumber} ${login}${labelTag}\n${cleaned}\n\n---\n\n`;
+  return { text: out + appended, rawChars, cleanedChars: cleaned.length };
+}
+
+function appendComment(out: string, c: IssueComment): AppendResult {
+  const rawChars = c.body.length;
+  const cleaned = stripNoise(c.body).trim();
+  if (!cleaned) return { text: out, rawChars, cleanedChars: 0 };
+  const appended = `${c.author.login} #${c.databaseId}\n${cleaned}\n\n---\n\n`;
+  return { text: out + appended, rawChars, cleanedChars: cleaned.length };
+}
+
+function parseRepoFlag(flagArgs: string[]): { owner: string; repo: string } | null {
+  const idx = flagArgs.indexOf("--repo");
+  if (idx === -1 || !flagArgs[idx + 1]) return null;
+  const parts = flagArgs[idx + 1].split("/");
+  if (parts.length !== 2) return null;
+  return { owner: parts[0], repo: parts[1] };
+}
+
+export async function scrapeIssue(args: string[]): Promise<void> {
+  const repoFlag = parseRepoFlag(args);
+  const stateIdx = args.indexOf("--state");
+  const state = stateIdx !== -1 && args[stateIdx + 1] ? args[stateIdx + 1] : "open";
+  const skipFlags = new Set(["--state", "--repo"]);
+  const positional = args.filter((a, i) => !skipFlags.has(a) && !skipFlags.has(args[i - 1]));
+
+  let owner: string, repo: string, issueNumber: number;
+
+  const firstArg = positional[0];
+  if (firstArg?.startsWith("http")) {
+    const parsed = parseGhUrl(firstArg);
+    if (parsed.type !== "issues") {
+      console.error(`Expected an issue URL, got a PR URL: ${firstArg}`);
+      process.exit(1);
+    }
+    owner = parsed.owner;
+    repo = parsed.repo;
+    issueNumber = parsed.number;
+  } else if (firstArg && /^\d+$/.test(firstArg)) {
+    ({ owner, repo } = repoFlag ?? detectRepo());
+    issueNumber = parseInt(firstArg, 10);
+  } else {
+    ({ owner, repo } = repoFlag ?? detectRepo());
+    const issues = listIssues(owner, repo, state);
+    console.log(`\nFound ${issues.length} ${state} issue(s) in ${owner}/${repo}\n`);
+    issueNumber = await selectIssue(issues);
+  }
+
+  console.log(`\nFetching issue #${issueNumber} from ${owner}/${repo}…`);
+  const payload = fetchGraphQL<IssuePayload>(ISSUE_GRAPHQL, { owner, repo, issueNumber });
+  const issue = payload.data.repository.issue;
+  if (!issue) {
+    console.error(`Issue #${issueNumber} not found in ${owner}/${repo}.`);
+    process.exit(1);
+  }
+
+  mkdirSync(OUT_DIR, { recursive: true });
+  ensureGitignore(["issues/new-*.md"]);
+
+  const cache = loadCache(CACHE_FILE);
+  const header = `Issue #${issueNumber} -- ${owner}/${repo}\n\n`;
+  let body = "";
+  let count = 0;
+  let totalRawChars = 0;
+  let totalCleanedChars = 0;
+
+  const bodyKey = `issue-body-${issueNumber}`;
+  if (!cache.has(bodyKey) && issue.body?.trim()) {
+    const result = appendBody(body, issue.body, issue.author.login, issueNumber, issue.labels.nodes);
+    body = result.text;
+    totalRawChars += result.rawChars;
+    totalCleanedChars += result.cleanedChars;
+    cache.add(bodyKey);
+    if (result.cleanedChars > 0) count++;
+  }
+
+  for (const c of issue.comments.nodes) {
+    const key = String(c.databaseId);
+    if (isBot(c.author.login) || !c.body.trim()) { cache.add(key); continue; }
+    if (cache.has(key)) continue;
+    const result = appendComment(body, c);
+    if (result.cleanedChars === 0) { cache.add(key); continue; }
+    body = result.text;
+    totalRawChars += result.rawChars;
+    totalCleanedChars += result.cleanedChars;
+    cache.add(key);
+    count++;
+  }
+
+  const tokenSummary = count > 0
+    ? formatTokenSummary(totalRawChars, totalCleanedChars) + "\n\n"
+    : "";
+  const output = header + tokenSummary + body;
+
+  saveCache(CACHE_FILE, cache);
+  const outFile = join(OUT_DIR, `new-${new Date().toISOString().replace(/[:.]/g, "-")}.md`);
+  writeFileSync(outFile, output, "utf-8");
+
+  console.log(count > 0 ? `\n✅ ${count} new entry(s) → ${outFile}` : `\n✅ No new comments since last run. → ${outFile}`);
+}

package/scripts/lib/shared.test.ts

@@ -0,0 +1,20 @@
+import { describe, it, expect } from "vitest";
+import { parseGhUrl } from "./shared.js";
+
+describe("parseGhUrl", () => {
+  it("parses PR URLs", () => {
+    const result = parseGhUrl("https://github.com/octocat/repo/pull/42");
+    expect(result).toEqual({ owner: "octocat", repo: "repo", type: "pull", number: 42 });
+  });
+
+  it("parses issue URLs", () => {
+    const result = parseGhUrl("https://github.com/YosefHayim/extensions/issues/15");
+    expect(result).toEqual({ owner: "YosefHayim", repo: "extensions", type: "issues", number: 15 });
+  });
+
+  it("handles URLs with trailing content", () => {
+    const result = parseGhUrl("https://github.com/org/project/issues/99#issuecomment-123");
+    expect(result.number).toBe(99);
+    expect(result.type).toBe("issues");
+  });
+});

package/scripts/lib/shared.ts

@@ -0,0 +1,74 @@
+import { execSync } from "node:child_process";
+import { readFileSync, writeFileSync, existsSync, unlinkSync } from "node:fs";
+import { join } from "node:path";
+import { tmpdir } from "node:os";
+
+export interface IdCache { seen: string[]; }
+
+export function run(cmd: string): string {
+  return execSync(cmd, { encoding: "utf-8", stdio: "pipe" }).trim();
+}
+
+export function detectRepo(): { owner: string; repo: string } {
+  try {
+    const remote = run("git remote get-url origin");
+    const m = remote.match(/github\.com[:/]([^/]+)\/([^/\s]+?)(?:\.git)?$/);
+    if (!m) throw new Error();
+    return { owner: m[1], repo: m[2] };
+  } catch {
+    console.error("Could not detect GitHub repo. Run from a GitHub repo or pass a URL.");
+    process.exit(1);
+  }
+}
+
+export function parseGhUrl(url: string): { owner: string; repo: string; type: "pull" | "issues"; number: number } {
+  const m = url.match(/github\.com\/([^/]+)\/([^/]+)\/(pull|issues)\/(\d+)/);
+  if (!m) { console.error(`Invalid GitHub URL: ${url}`); process.exit(1); }
+  return { owner: m[1], repo: m[2], type: m[3] as "pull" | "issues", number: parseInt(m[4], 10) };
+}
+
+export function loadCache(cacheFile: string): Set<string> {
+  if (!existsSync(cacheFile)) return new Set();
+  try { return new Set((JSON.parse(readFileSync(cacheFile, "utf-8")) as IdCache).seen); } catch { return new Set(); }
+}
+
+export function saveCache(cacheFile: string, seen: Set<string>): void {
+  writeFileSync(cacheFile, JSON.stringify({ seen: [...seen] }, null, 2), "utf-8");
+}
+
+export function fetchGraphQL<T>(query: string, variables: Record<string, unknown>): T {
+  const reqFile = join(tmpdir(), `.pr-prism-req-${Date.now()}.json`);
+  writeFileSync(reqFile, JSON.stringify({ query, variables }), "utf-8");
+  try {
+    return JSON.parse(run(`gh api https://api.github.com/graphql --input "${reqFile}"`)) as T;
+  } catch (err) {
+    console.error("gh API failed. Is gh authenticated? Run: gh auth login");
+    console.error((err as Error).message);
+    process.exit(1);
+  } finally {
+    unlinkSync(reqFile);
+  }
+}
+
+export function ensureGitignore(entries: string[]): void {
+  const GITIGNORE = ".gitignore";
+  const MARKER = "# pr-prism";
+  const BLOCK = ["", MARKER, ...entries].join("\n");
+
+  if (!existsSync(GITIGNORE)) {
+    writeFileSync(GITIGNORE, BLOCK.trimStart() + "\n", "utf-8");
+    return;
+  }
+
+  const content = readFileSync(GITIGNORE, "utf-8");
+  if (content.includes(MARKER)) {
+    const missing = entries.filter((entry) => !content.includes(entry));
+    if (missing.length === 0) return;
+    const tail = content.endsWith("\n") ? "" : "\n";
+    writeFileSync(GITIGNORE, content + tail + missing.join("\n") + "\n", "utf-8");
+    return;
+  }
+
+  const suffix = content.endsWith("\n") ? "" : "\n";
+  writeFileSync(GITIGNORE, content + suffix + BLOCK + "\n", "utf-8");
+}