PDF Resource

The PDF resource generates a PDF file from HTML or Markdown content using one of three rendering backends. It can be used as a primary resource or as an inline resource inside before / after blocks.

Basic Usage

yaml

apiVersion: kdeps.io/v1
kind: Resource

metadata:
  actionId: generateReport
  name: Generate Report

run:
  pdf:
    content: "<h1>Hello, World!</h1><p>This is a PDF report.</p>"
    contentType: html
    backend: wkhtmltopdf

Configuration Options

Option	Type	Description
`content`	string	Required. The HTML or Markdown source to render. Supports expressions.
`contentType`	string	`html` (default) or `markdown`.
`backend`	string	`wkhtmltopdf` (default), `pandoc`, or `weasyprint`.
`outputFile`	string	Explicit output path for the PDF. If omitted, a path under `/tmp/kdeps-pdf/` is auto-generated.
`options`	list	Extra CLI flags passed directly to the backend binary.
`timeoutDuration`	string	Maximum time allowed for the backend to run (e.g. `120s`, `2m`). Default: `60s`.

Backends

wkhtmltopdf (default)

Renders HTML to PDF using the wkhtmltopdf CLI. When contentType: markdown, the Markdown is wrapped in a minimal HTML skeleton before rendering.

Install:

bash

# macOS
brew install wkhtmltopdf

# Debian/Ubuntu
apt install wkhtmltopdf

yaml

run:
  pdf:
    content: "<h1>Report</h1><p>Generated by KDeps.</p>"
    contentType: html
    backend: wkhtmltopdf
    options:
      - "--page-size"
      - "A4"
      - "--margin-top"
      - "10mm"

pandoc

Converts HTML or Markdown to PDF via pandoc + a LaTeX engine (e.g. pdflatex or xelatex). Best choice for Markdown-heavy documents and academic/technical reports.

Install:

bash

# macOS
brew install pandoc mactex

# Debian/Ubuntu
apt install pandoc texlive-xetex

yaml

run:
  pdf:
    content: |
      # My Report

      This is a **Markdown** report with _formatting_.

      - Item one
      - Item two
    contentType: markdown
    backend: pandoc
    options:
      - "--pdf-engine=xelatex"
      - "--variable"
      - "geometry:margin=1in"

weasyprint

Renders HTML/CSS to PDF using the WeasyPrint Python library. Excellent CSS support; good for styled reports and invoices.

Install:

bash

pip install weasyprint

yaml

run:
  pdf:
    content: |
      <html>
      <head><style>
        body { font-family: Arial, sans-serif; }
        h1 { color: #333; }
      </style></head>
      <body><h1>Styled Report</h1><p>Beautiful CSS-rendered PDF.</p></body>
      </html>
    contentType: html
    backend: weasyprint

Accessing the Output File

After the PDF resource runs, the result map is available via get('actionId'):

Key	Type	Description
`success`	bool	`true` if the PDF was generated successfully.
`outputFile`	string	Absolute path to the generated PDF file.
`backend`	string	Backend used (e.g. `wkhtmltopdf`).
`contentType`	string	Content type used (`html` or `markdown`).
`sizeBytes`	int	File size of the generated PDF in bytes.

yaml

resources:
  - metadata:
      actionId: generateReport
    run:
      pdf:
        content: "<h1>Report</h1>"
        backend: wkhtmltopdf

  - metadata:
      actionId: sendReport
      requires: [generateReport]
    run:
      apiResponse:
        success: true
        response:
          pdfPath: "{{ get('generateReport').outputFile }}"
          sizeBytes: "{{ get('generateReport').sizeBytes }}"

PDF as an Inline Resource

PDF generation can run before or after the main resource action:

yaml

run:
  before:
    - pdf:
        content: "<h1>Pre-flight doc</h1>"
        backend: wkhtmltopdf

  chat:
    model: llama3
    prompt: "{{ input() }}"

  after:
    - pdf:
        content: "<h1>Summary</h1><p>{{ get('chat') }}</p>"
        backend: wkhtmltopdf
        outputFile: /tmp/summary.pdf

Using Expressions in `content`

The content field supports KDeps expressions:

yaml

run:
  pdf:
    content: |
      <h1>Report for {{ get("name") }}</h1>
      <p>Score: <strong>{{ get("score") }}</strong> / 100</p>
      <p>Generated: {{ info("timestamp") }}</p>
    contentType: html
    backend: wkhtmltopdf

Example: CV/JD Match Report

Full pipeline: scrape a job description, score the match, generate a PDF report.

yaml

# resources/scrapeJD.yaml
metadata:
  actionId: scrapeJD
run:
  scraper:
    url: "{{ get('jd_url') }}"

# resources/scoreMatch.yaml
metadata:
  actionId: scoreMatch
  requires: [scrapeJD]
run:
  chat:
    model: gpt-4o
    prompt: |
      CV: {{ get('cv_text') }}
      Job Description: {{ get('scrapeJD') }}
      Rate the match from 0-100 and explain the top 3 strengths and gaps. Output JSON.

# resources/generateReport.yaml
metadata:
  actionId: generateReport
  requires: [scoreMatch]
run:
  pdf:
    content: |
      <html>
      <head><style>
        body { font-family: Arial, sans-serif; max-width: 800px; margin: auto; padding: 2em; }
        h1 { color: #1a1a2e; }
        .score { font-size: 2em; font-weight: bold; color: #16213e; }
      </style></head>
      <body>
        <h1>Match Report</h1>
        <p class="score">Score: {{ get('scoreMatch').score }} / 100</p>
        <h2>Strengths</h2>
        <ul>{{ get('scoreMatch').strengths }}</ul>
        <h2>Gaps</h2>
        <ul>{{ get('scoreMatch').gaps }}</ul>
      </body>
      </html>
    contentType: html
    backend: wkhtmltopdf
    outputFile: /tmp/match-report.pdf

Example: Motivation Letter

Generate a personalized motivation letter as a PDF using an LLM:

yaml

# resources/writeLetter.yaml
metadata:
  actionId: writeLetter
run:
  chat:
    model: gpt-4o
    prompt: |
      Write a professional motivation letter for this candidate applying to this role.
      CV: {{ get('cv_text') }}
      Job: {{ get('jd_text') }}
      Format as clean HTML with <h1>, <p> tags only.

# resources/letterPDF.yaml
metadata:
  actionId: letterPDF
  requires: [writeLetter]
run:
  pdf:
    content: "{{ get('writeLetter') }}"
    contentType: html
    backend: weasyprint
    outputFile: /tmp/motivation-letter.pdf

Backend Comparison

Feature	wkhtmltopdf	pandoc	weasyprint
HTML input	Yes	Yes	Yes
Markdown input	Basic (via skeleton)	Native	Basic (via skeleton)
CSS support	Good	Limited	Excellent
LaTeX required	No	Yes	No
JavaScript	Yes	No	No
Install size	Medium	Large	Small (pip)
Best for	HTML reports, invoices	Academic/tech docs	Styled reports

Validator Rules

The pdf resource is validated before execution:

content must not be empty.
contentType must be html or markdown (if set).
backend must be wkhtmltopdf, pandoc, or weasyprint (if set).
A resource may only declare one primary executor (pdf, chat, exec, etc.).

Next Steps

Scraper Resource - Extract content from web pages and documents
LLM Resource - Generate content with AI models
Exec Resource - Run shell commands
TTS Resource - Text-to-Speech synthesis
API Response - Return data to the HTTP caller

PDF Resource ​

Basic Usage ​

Configuration Options ​

Backends ​

wkhtmltopdf (default) ​

pandoc ​

weasyprint ​

Accessing the Output File ​

PDF as an Inline Resource ​

Using Expressions in content ​

Example: CV/JD Match Report ​

Example: Motivation Letter ​

Backend Comparison ​

Validator Rules ​

Next Steps ​

PDF Resource

Basic Usage

Configuration Options

Backends

wkhtmltopdf (default)

pandoc

weasyprint

Accessing the Output File

PDF as an Inline Resource

Using Expressions in `content`

Example: CV/JD Match Report

Example: Motivation Letter

Backend Comparison

Validator Rules

Next Steps