render_application.py

#!/usr/bin/env python3
"""
render_application.py — deterministic ATS-clean PDF renderer.

Takes the structured content produced by lib/generate-application.js and builds
two PDFs: the tailored résumé and the cover letter. The LLM never controls
layout — this renderer does, so formatting stays consistent and ATS-safe.

Usage:  python3 render_application.py <content.json> <output_dir>

Requires: python-docx  (pip3 install python-docx)
          LibreOffice   (soffice on PATH, or /Applications/LibreOffice.app)

ATS rules honoured: single column, standard headings, real bullet lists,
no tables/text-boxes/headers/footers, standard font, dates as MMM YYYY.
"""

import json
import os
import subprocess
import sys
import tempfile

from docx import Document
from docx.shared import Pt, RGBColor
from docx.enum.text import WD_ALIGN_PARAGRAPH

FONT = "Calibri"
DARK = RGBColor(0x1a, 0x1a, 0x1a)


def _base(doc):
    style = doc.styles["Normal"]
    style.font.name = FONT
    style.font.size = Pt(10.5)
    style.font.color.rgb = DARK
    for section in doc.sections:
        section.top_margin = section.bottom_margin = Pt(40)
        section.left_margin = section.right_margin = Pt(54)


def _heading(doc, text):
    p = doc.add_paragraph()
    p.space_before = Pt(8)
    run = p.add_run(text.upper())
    run.bold = True
    run.font.size = Pt(11)
    p.paragraph_format.space_before = Pt(10)
    p.paragraph_format.space_after = Pt(2)
    return p


def _line(doc, text, size=10.5, bold=False, align=None, after=2, before=0):
    p = doc.add_paragraph()
    r = p.add_run(text)
    r.bold = bold
    r.font.size = Pt(size)
    if align is not None:
        p.alignment = align
    p.paragraph_format.space_after = Pt(after)
    p.paragraph_format.space_before = Pt(before)
    return p


def _bullet(doc, text):
    p = doc.add_paragraph(style="List Bullet")
    r = p.add_run(text)
    r.font.size = Pt(10.5)
    p.paragraph_format.space_after = Pt(2)
    return p


def build_resume(content, path):
    doc = Document()
    _base(doc)
    cand = content.get("candidate", {})
    r = content.get("resume", {})

    # Name + positioning + contact (plain text, never a header)
    _line(doc, cand.get("name", ""), size=18, bold=True, align=WD_ALIGN_PARAGRAPH.CENTER, after=1)
    if r.get("title"):
        _line(doc, r["title"], size=11, bold=False, align=WD_ALIGN_PARAGRAPH.CENTER, after=1)
    if cand.get("contact"):
        _line(doc, cand["contact"], size=9.5, align=WD_ALIGN_PARAGRAPH.CENTER, after=4)

    if r.get("summary"):
        _heading(doc, "Professional Summary")
        _line(doc, r["summary"])

    if r.get("competencies"):
        _heading(doc, "Core Competencies")
        _line(doc, " · ".join(r["competencies"]))

    if r.get("achievements"):
        _heading(doc, "Key Achievements")
        for a in r["achievements"]:
            _bullet(doc, a)

    if r.get("experience"):
        _heading(doc, "Professional Experience")
        for e in r["experience"]:
            _line(doc, e.get("titleLine", ""), bold=True, after=0, before=6)
            org = e.get("org", "")
            dates = e.get("dates", "")
            meta = " · ".join([x for x in [org, dates] if x])
            if meta:
                _line(doc, meta, size=9.5, after=2)
            for b in e.get("bullets", []):
                _bullet(doc, b)

    if r.get("education"):
        _heading(doc, "Education")
        for ed in r["education"]:
            _line(doc, ed)

    if r.get("certifications"):
        _heading(doc, "Certifications")
        for c in r["certifications"]:
            _bullet(doc, c)

    if r.get("languages"):
        _heading(doc, "Languages")
        _line(doc, r["languages"])

    doc.save(path)


def build_cover(content, path):
    doc = Document()
    _base(doc)
    cand = content.get("candidate", {})
    c = content.get("cover", {})
    meta = content.get("meta", {})

    _line(doc, cand.get("name", ""), size=14, bold=True, after=0)
    if cand.get("contact"):
        _line(doc, cand["contact"], size=9.5, after=8)

    from datetime import date
    _line(doc, date.today().strftime("%B %d, %Y"), size=10.5, after=8)

    company = meta.get("company", "")
    if company:
        _line(doc, f"Re: {meta.get('role','')} — {company}", bold=True, after=8)

    if c.get("greeting"):
        _line(doc, c["greeting"], after=6)

    for para in c.get("paragraphs", []):
        _line(doc, para, after=8)

    if c.get("close"):
        _line(doc, c["close"], after=8, before=2)

    _line(doc, "Sincerely,", after=2)
    _line(doc, c.get("signature", cand.get("name", "")), after=0)

    doc.save(path)


def _soffice():
    for cand in ("soffice",
                 "/Applications/LibreOffice.app/Contents/MacOS/soffice",
                 "/usr/bin/soffice"):
        try:
            subprocess.run([cand, "--version"], capture_output=True, timeout=15)
            return cand
        except Exception:
            continue
    raise RuntimeError("LibreOffice (soffice) not found. Install it, or `brew install --cask libreoffice`.")


def to_pdf(docx_path, out_dir, final_name):
    soffice = _soffice()
    subprocess.run([soffice, "--headless", "--convert-to", "pdf", "--outdir", out_dir, docx_path],
                   capture_output=True, timeout=120, check=True)
    produced = os.path.join(out_dir, os.path.splitext(os.path.basename(docx_path))[0] + ".pdf")
    final = os.path.join(out_dir, final_name)
    if produced != final and os.path.exists(produced):
        os.replace(produced, final)
    return final


def main():
    if len(sys.argv) < 3:
        print("usage: render_application.py <content.json> <output_dir>", file=sys.stderr)
        sys.exit(2)

    content_path, out_dir = sys.argv[1], sys.argv[2]
    with open(content_path, "r", encoding="utf-8") as fh:
        content = json.load(fh)

    os.makedirs(out_dir, exist_ok=True)
    files = content.get("files", {})
    resume_name = files.get("resume", "Resume.pdf")
    cover_name = files.get("cover", "Cover.pdf")

    with tempfile.TemporaryDirectory() as tmp:
        r_docx = os.path.join(tmp, os.path.splitext(resume_name)[0] + ".docx")
        c_docx = os.path.join(tmp, os.path.splitext(cover_name)[0] + ".docx")
        build_resume(content, r_docx)
        build_cover(content, c_docx)
        to_pdf(r_docx, out_dir, resume_name)
        to_pdf(c_docx, out_dir, cover_name)

    print(f"OK {resume_name} {cover_name}")


if __name__ == "__main__":
    main()