@gafj/gafj 0.1.19 → 0.1.20

package/http/guard.js

@@ -91,7 +91,7 @@ function readJson(req, cap = BODY_CAP) {
     const chunks = [];
     req.on("data", (c) => {
       size += c.length;
-      if (size > cap) { reject(Object.assign(new Error("body larger than 1 MB"), { status: 413 })); req.destroy(); return; }
+      if (size > cap) { req.removeAllListeners("data"); req.resume(); reject(Object.assign(new Error(`body larger than ${Math.round(cap / 1024 / 1024)} MB; paste less`), { status: 413 })); return; }
       chunks.push(c);
     });
     req.on("end", () => {

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@gafj/gafj",
-  "version": "0.1.19",
+  "version": "0.1.20",
   "description": "GAF-J: a local campaign engine for job search. One SQLite file, one ingest door, your own AI subscription.",
   "homepage": "https://gaf-j.com",
   "repository": {

package/ui/lib.js

@@ -90,3 +90,38 @@ export function useInstall() {
   const standalone = matchMedia("(display-mode: standalone)").matches || navigator.standalone === true;
   return { canInstall: !!installer.event, installed: installer.installed || standalone, install: async () => { if (!installer.event) return; installer.event.prompt(); await installer.event.userChoice; installer.event = null; } };
 }
+
+/**
+ * A job posting as pasted: markdown links become their text, bare URLs and tracking lines go, blank runs
+ * collapse. A LinkedIn copy carries a wall of overlay URLs that says nothing about the job.
+ */
+export function cleanPosting(text) {
+  return String(text || "")
+    .replace(/!\[[^\]]*\]\([^)]*\)/g, "")
+    .replace(/\[([^\]]*)\]\((?:https?:)?\/\/[^)\s]*(?:\s+"[^"]*")?\)/g, "$1")
+    .replace(/\((?:https?:)?\/\/[^)\s]{20,}\)/g, "")
+    .split(/\r?\n/)
+    .map((l) => l.replace(/\s+$/, ""))
+    .filter((l, i, all) => !/^\s*(?:https?:)?\/\/\S+\s*$/.test(l) || (/\/(jobs?|careers?|positions?|openings?)\//i.test(l) && all.findIndex((x) => /^\s*(?:https?:)?\/\/\S+\s*$/.test(x) && /\/(jobs?|careers?|positions?|openings?)\//i.test(x)) === i))
+    .join("\n")
+    .replace(/[ \t]{3,}/g, "  ")
+    .replace(/\n{3,}/g, "\n\n")
+    .trim();
+}
+
+const NOISE = /\b(ago|hours?|days?|weeks?|applicants?|promoted|easy apply|reposted|save|share|show more|about the job|job details|united states|remote|hybrid|on-site|full-time|part-time|contract|tailor my resume|help me stand out|create cover letter)\b/i;
+const TITLE = /\b(director|manager|engineer|analyst|lead|head|vp|vice president|specialist|coordinator|supervisor|associate|consultant|architect|designer|developer|scientist|officer|president|chief|planner|buyer|controller|administrator|technician|representative|recruiter|nurse|teacher|intern|accountant|assistant|generalist|partner|strategist|writer|editor)\b/i;
+/** Title and company guessed from a posting's first lines; both are the person's to correct. */
+export function guessPosting(text) {
+  const lines = text.split(/\r?\n/).map((l) => l.trim()).filter(Boolean).slice(0, 20);
+  const plain = (l) => l.length <= 90 && !/^https?:/i.test(l) && !/^(at|company|location|posted|apply|share)\b/i.test(l) && !NOISE.test(l);
+  const title = lines.find((l) => plain(l) && TITLE.test(l) && !/[.:;]$/.test(l) && !/\b(company|inc|llc|corp|group|ltd)\.?$/i.test(l)) || lines.find(plain) || "";
+  let company = "";
+  for (const l of lines) { const m = l.match(/^(?:at|company[:\s]+|employer[:\s]+)\s*([^|·,]{2,60})/i); if (m) { company = m[1].trim(); break; } }
+  const i = lines.findIndex((l) => l === title);
+  const nameish = (l) => { const c = l.split(/\s[·|]\s/)[0].replace(/^[·|\-\s]+/, "").trim(); return c && c !== title && c.length <= 60 && !/^https?:/i.test(c) && !/\d{4}/.test(c) && !NOISE.test(c) && !/[.:;]$/.test(c) && c.split(/\s+/).length <= 8 ? c : ""; };
+  // the line after the title ("Company · City"), else a short name-like line above it, else one soon after
+  if (!company) company = [lines[i + 1] || ""].map(nameish).find(Boolean) || lines.slice(0, i).map(nameish).find(Boolean) || lines.slice(i + 2, i + 6).map(nameish).find(Boolean) || "";
+  const url = lines.find((l) => /^https?:\/\//i.test(l)) || "";
+  return { title, company, url };
+}

package/ui/screens/postings.js

@@ -1,5 +1,5 @@
 import { html, useState, useEffect } from "../vendor/preact.mjs";
-import { api } from "../lib.js";
+import { cleanPosting, guessPosting, api } from "../lib.js";
 
 /** Screen 2: paste a posting, store and scan locally, edit the proposed categories, freeze, log the application. */
 
@@ -39,21 +39,13 @@ export function Postings({ live, id }) {
   const [form, setForm] = useState({ company: "", title: "", source_url: "", location: "", comp_text: "", raw_text: "" });
   const [guessed, setGuessed] = useState({ company: true, title: true });
   // the paste is all a person should have to give; company and title are guessed from its first lines and can be corrected
-  const onPaste = (text) => {
-    const lines = text.split(/\r?\n/).map((l) => l.trim()).filter(Boolean).slice(0, 12);
+  const onPaste = (raw) => {
+    const text = cleanPosting(raw);
+    const g = guessPosting(text);
     const next = { ...form, raw_text: text };
-    if (guessed.title) {
-      const t = lines.find((l) => l.length <= 90 && !/^https?:/i.test(l) && !/^(at|company|location|posted|apply|share)\b/i.test(l));
-      next.title = t || "";
-    }
-    if (guessed.company) {
-      let c = "";
-      for (const l of lines) { const m = l.match(/^(?:at|company[:\s]+|employer[:\s]+)\s*([^|·,]{2,60})/i); if (m) { c = m[1].trim(); break; } }
-      if (!c) { const i = lines.findIndex((l) => l === next.title); const cand = lines[i + 1] || ""; if (cand && cand.length <= 50 && !/^https?:/i.test(cand) && !/\d{4}/.test(cand)) c = cand.replace(/^[·|\-\s]+/, ""); }
-      next.company = c;
-    }
-    const url = lines.find((l) => /^https?:\/\//i.test(l));
-    if (url && !form.source_url) next.source_url = url;
+    if (guessed.title) next.title = g.title;
+    if (guessed.company) next.company = g.company;
+    if (g.url && !form.source_url) next.source_url = g.url;
     setForm(next);
   };
   const [err, setErr] = useState("");

package/ui/screens/welcome.js

@@ -1,5 +1,5 @@
 import { html, useState, useEffect } from "../vendor/preact.mjs";
-import { api } from "../lib.js";
+import { api, cleanPosting, guessPosting } from "../lib.js";
 import { Onboarding, DraftForm, fromDraft, Positions, ContactForm } from "./kb.js";
 import { PacketModal } from "./packet.js";
 
@@ -64,13 +64,12 @@ function StepPosting({ reload }) {
   const [title, setTitle] = useState("");
   const [g, setG] = useState({ company: true, title: true });
   const [err, setErr] = useState("");
-  const onPaste = (t) => {
+  const onPaste = (raw) => {
+    const t = cleanPosting(raw);
     setText(t);
-    const lines = t.split(/\r?\n/).map((l) => l.trim()).filter(Boolean).slice(0, 12);
-    let ti = title, co = company;
-    if (g.title) ti = lines.find((l) => l.length <= 90 && !/^https?:/i.test(l) && !/^(at|company|location|posted|apply|share)\b/i.test(l)) || "";
-    if (g.company) { co = ""; for (const l of lines) { const m = l.match(/^(?:at|company[:\s]+|employer[:\s]+)\s*([^|·,]{2,60})/i); if (m) { co = m[1].trim(); break; } } if (!co) { const i = lines.findIndex((l) => l === ti); const c = lines[i + 1] || ""; if (c && c.length <= 50 && !/^https?:/i.test(c) && !/\d{4}/.test(c)) co = c.replace(/^[·|\-\s]+/, ""); } }
-    setTitle(ti); setCompany(co);
+    const gu = guessPosting(t);
+    if (g.title) setTitle(gu.title);
+    if (g.company) setCompany(gu.company);
   };
   const store = async (e) => {
     e.preventDefault(); setErr("");