docrafter

Unified Rust library for PDF and DOCX documents — ergonomic API, strong typing, snapshot-tested output.

Status

Phase	Scope	Status
0	Workspace, core types, Hello PDF, snapshots	✅ Done
0.1	Table, Spacer, PageBreak, Image, word wrap, TTF/Cyrillic	✅ Done
0.2	PDF read / merge (PdfReader, merge)	✅ Done
0.3	DOCX read/write (runs, tables, images, lists, styles)	✅ Done
0.4	ODT / LibreOffice (OdtDocument, shared office model)	✅ Done
1.0	Multi-format export (OfficeDocument → PDF / DOCX / ODT)	✅ Done
1.1	PDF multi-run (reportlab), PdfReader::extract_text (pypdf)	✅ Done
1.2	PDF read: lopdf-compatible ToUnicode, OCR (Tesseract CLI)	✅ Done
1.3	In-repo OCR (docrafter-ocr) + PDF render (docrafter-pdf-render), from_bytes	✅ Done
1.4	Templates ({{vars}}), ReportBuilder, export_bundle	✅ Done
1.5	PDF manipulate: extract/split pages, rotate, metadata (pypdf)	✅ Done
1.6	Bookmarks, watermark, decrypt, Office/DOCX/ODT append	✅ Done
1.7	PDF links, replace text, compress, PdfExportOptions, office::open	✅ Done
1.8	HTML → office/PDF, PDF header/footer ({page} / {pages})	✅ Done
2.0	CLI (docrafter), OpenDocument::open	✅ Done
3.0	CLI metadata/bookmark/extract, HTML style=, OCR CI, publish docs	✅ Done
4.0	cargo deny, CHANGELOG, CLI info/replace/compress, license notes	✅ Done
5.0	Publish prep, ocr feature flag (AGPL isolation), CLI template/links/batch, release workflow	✅ Done
6.0	GitHub Release CLI binaries (Linux/macOS/Windows), HTML text-align/div, docs/RELEASE.md	✅ Done
7.0	HTML a/hr, CLI pdf add-link, examples/sample.html	✅ Done
8.0	pdf render→PNG, HTML blockquote/line-height, layout line-height, Linux arm64 release, docs/ZENPDF.md	✅ Done
9.0	Underline PDF/DOCX/ODT/HTML, CLI batch --to pdf	✅ Done
10.0	Strikethrough all formats, HTML colspan, docs/FINAL_CHECKLIST.md	✅ Done
11.0	Sup/sub, HTML rowspan, DOCX comments, PDF text fields, docs/API.md	✅ Done
12.0	crates.io + GitHub release, docs/RELEASE_GUIDE_RU.md	✅ Done

Quick start

use docrafter::prelude::*;

let mut doc = PdfDocument::new();
doc.push(Paragraph::new("Hello, docrafter!"));
doc.save("hello.pdf")?;

Merge and read PDFs (pypdf-style):

use docrafter::prelude::*;

let mut out = PdfReader::open("cover.pdf")?;
assert!(out.page_count() >= 1);
out.merge(&PdfReader::open("appendix.pdf")?)?;
out.save("merged.pdf")?;

// Text extraction (works on docrafter-generated PDFs too)
let text = PdfReader::open("merged.pdf")?.extract_text()?;

// Page tools (pypdf-style)
use docrafter::pdf::{PdfMetadata, Rotate, WatermarkOptions};
let mut pdf = PdfReader::open("merged.pdf")?;
pdf.set_metadata(&PdfMetadata { title: Some("Merged".into()), ..Default::default() })?;
pdf.rotate(Some(&[1]), Rotate::Clockwise90)?;
let page2 = pdf.with_pages(&[2])?;
let parts = pdf.split()?;
pdf.add_bookmark("Chapter 1", 1, None)?;
pdf.add_watermark(None, &WatermarkOptions { text: "DRAFT".into(), ..Default::default() })?;
let unlocked = PdfReader::open_with_password("secret.pdf", "pass")?;

// Scanned PDFs: pure-Rust OCR (no tesseract/pdftoppm on PATH)

OCR (models + release builds)

OCR model weights are not in git. Download once:

./scripts/fetch-ocr-models.sh   # → crates/docrafter-ocr/models/*.rten

Detection and recognition run in-repo (docrafter-ocr, docrafter-pdf-render). For production OCR use release (debug builds of ocrs/rten are very slow):

cargo build --release -p docrafter-cli
./target/release/docrafter pdf text scan.pdf --ocr -o text.txt

use docrafter::pdf::{OcrOptions, TextExtractMode};
let bytes = std::fs::read("scan.pdf")?;
let ocr = PdfReader::from_bytes(&bytes)?
    .extract_text_mode(TextExtractMode::Ocr(OcrOptions::default()))?;

The default RTen models are Latin-focused (HierText). Cyrillic in generated PDFs uses embedded DejaVu fonts (extract_text()). For scanned Cyrillic, OCR quality depends on the model; train or swap recognition weights via docrafter-ocr model paths.

PDF edit on docrafter PDFs: replace_text / replace_text_all (DejaVu CID). Encrypt on save: reader.encrypt(&EncryptOptions::user("pass"))? then save (revision 2, compatible with open_with_password).

use docrafter::pdf::{EncryptOptions, PdfReader};
let mut pdf = PdfReader::open("report.pdf")?;
pdf.encrypt(&EncryptOptions::user("secret"))?;
pdf.save("locked.pdf")?;

PDF paragraphs support multiple runs (like python-docx / reportlab inline styles):

use docrafter::prelude::*;

let mut doc = PdfDocument::new();
doc.push(
    Paragraph::new("Hello ")
        .run("world", Style::new().bold().color_value(Color::rgb(200, 0, 0)?)),
);
doc.save("styled.pdf")?;

DOCX:

use docrafter::docx::{DocxDocument, Image, List, Paragraph, Table};
use docrafter::prelude::Style;

let mut doc = DocxDocument::new();
doc.push(
    Paragraph::new("Отчёт ")
        .run("2026", Style::new().bold()),
);
doc.push_table(
    Table::professional()
        .columns(["Name", "Hours"])
        .row(["Anna", "40"]),
);
doc.push_list(List::new().item("First").item("Second"));
doc.push_image(Image::from_path("logo.png")?.size(48.0, 48.0));
doc.save("report.docx")?;

// Roundtrip
let loaded = DocxDocument::open("report.docx")?;

ODT (LibreOffice Writer):

use docrafter::odt::{OdtDocument, Paragraph};

let mut doc = OdtDocument::new();
doc.push(Paragraph::new("Привет, LibreOffice!"));
doc.save("hello.odt")?;  // OpenDocument 1.2 package

One model → PDF + Word + LibreOffice:

use docrafter::office::{OfficeDocument, Paragraph, Table};
use docrafter::prelude::*;

let mut doc = OfficeDocument::new();
doc.push(Paragraph::new("Отчёт").style(Style::heading1()));
doc.push_table(
    Table::professional()
        .columns(["Name", "Hours"])
        .row(["Anna", "40"]),
);

export_save_auto(&doc, "report.pdf")?;
export_save_auto(&doc, "report.docx")?;
export_save_auto(&doc, "report.odt")?;
// or all three at once:
export_bundle(&doc, "report")?;

Templates and report builder (Phase 1.4):

use docrafter::prelude::*;

let doc = ReportBuilder::new()
    .title("Report for {{month}}")
    .table_professional(["Name", "Hours"], &[vec!["Anna".into(), "40".into()]])
    .build(&Context::new().with("month", "May"))?;

export_bundle(&doc, "report")?;

Combine sections (Phase 1.6):

use docrafter::office::{OfficeDocument, Paragraph};

let mut report = OfficeDocument::new();
report.append(&intro);
report.append(&body);
export_bundle(&report, "full_report")?;

PDF export layout and links (Phase 1.7):

use docrafter::export::{PdfExportOptions, export_bytes_with_pdf_options, OutputFormat};
use docrafter::office::open;
use docrafter_core::PageSize;

let pdf = export_bytes_with_pdf_options(&doc, OutputFormat::Pdf, &PdfExportOptions {
    page_size: PageSize::letter(),
    ..Default::default()
})?;

let doc = open("report.docx")?; // or .odt

HTML → PDF / DOCX / ODT (Phase 1.8):

use docrafter::prelude::*;

let doc = html_to_office(
    "<h1>Report</h1><p>Hello <b>world</b></p><table><tr><th>A</th></tr><tr><td>1</td></tr></table>",
)?;
export_save_auto(&doc, "from_html.pdf")?;

let mut pdf = PdfDocument::new().with_header_footer(
    PageHeaderFooter::new().header("DRAFT").page_numbers(),
);
pdf.push(Paragraph::new("Content"));
pdf.save("with_footer.pdf")?;

CLI (Phase 2):

# Install binary into ~/.cargo/bin
./scripts/install-cli.sh
# or: cargo install --path crates/docrafter-cli --locked

docrafter pdf merge out.pdf a.pdf b.pdf
docrafter pdf text scan.pdf --ocr -o text.txt
docrafter pdf rotate in.pdf -o rot.pdf --angle 90
docrafter pdf split in.pdf -o ./pages/
docrafter pdf watermark in.pdf -o draft.pdf --text DRAFT
docrafter pdf extract in.pdf -o part.pdf --pages 1,3
docrafter pdf metadata in.pdf -o out.pdf --title "Report"
docrafter pdf bookmark in.pdf -o out.pdf --title "Start" --page 1
docrafter pdf encrypt in.pdf -o locked.pdf --password secret
docrafter pdf info report.pdf
docrafter pdf replace in.pdf -o out.pdf --from world --to docrafter
docrafter pdf compress big.pdf -o small.pdf
docrafter export report.docx -o report --bundle   # → report.pdf, .docx, .odt
docrafter convert report.docx -o report.pdf
docrafter html page.html -o page.pdf
docrafter template render examples/template-spec.json -o report --bundle
docrafter pdf links scan.pdf
docrafter pdf add-link doc.pdf -o linked.pdf --page 1 --rect 72,72,300,100 --uri https://example.com
docrafter pdf text-batch a.pdf b.pdf -o texts/ --ocr
docrafter html examples/sample.html -o sample.pdf
docrafter pdf render scan.pdf -o page1.png --page 1 --dpi 150
docrafter pdf render scan.pdf -o pages/ --all
docrafter batch report.html invoice.docx -o out/ --to pdf
docrafter pdf add-field form.pdf -o filled.pdf --page 1 --name email --rect 72,72,280,90 --value user@example.com

Rust API summary: docs/API.md.

AGPL-free builds (no OCR / zenpdf): depend on docrafter with default-features = false; enable only what you need. OCR needs feature ocr (default on the CLI crate).

From git (when published on crates.io: cargo install docrafter-cli):

cargo install --git https://github.com/pathofharmony/dockrafter --locked docrafter-cli

Публикация и безопасность: docs/RELEASE_GUIDE_RU.md (лицензии, секреты, crates.io, GitHub). Кратко: docs/PUBLISH.md, docs/FINAL_CHECKLIST.md.

Releases: tag v0.1.0 → CI builds CLI archives (docs/RELEASE.md).

License

MIT OR Apache-2.0. PDF rendering for OCR uses zenpdf (AGPL-3.0) — see LICENSE-NOTES.md.

Quality gate: ./scripts/check.sh (fmt, clippy, test, doc, cargo deny when installed).

Unified open API:

use docrafter::OpenDocument;

let doc = OpenDocument::open("report.docx")?;
match doc {
    OpenDocument::Office(o) => { /* export, edit blocks */ }
    OpenDocument::Pdf(p) => println!("pages: {}", p.page_count()),
}

cargo run --example hello_docx
cargo run --example hello_odt
cargo run --example hello_pdf
cargo run --example report_multi
cargo run --example report_template
./scripts/check.sh          # fmt + clippy + test + doc (как в CI)
./scripts/update-snapshots.sh  # обновить PDF fingerprints после смены вёрстки

Workspace crates

• docrafter — public facade, prelude, OpenDocument
• docrafter-cli — docrafter binary (cargo install --path crates/docrafter-cli)
• docrafter-core — Error, Style, Color, Length, PageSize
• docrafter-pdf-write — PDF generation
• docrafter-pdf-read — PDF open, page count, merge
• docrafter-office — shared Paragraph, Table, Image, List (DOCX + ODT)
• docrafter-docx — DOCX OOXML read/write (Word + LibreOffice via filter)
• docrafter-odt — ODT OpenDocument read/write (native LibreOffice)
• docrafter-layout — flow layout engine
• docrafter-font — embedded TrueType (DejaVu Sans)
• docrafter-template — {{variable}} substitution and ReportBuilder
• docrafter-html — minimal HTML → OfficeDocument
• docrafter-pdf-render — pure-Rust PDF page rasterization
• docrafter-ocr — pure-Rust OCR engine
• docrafter-testing — snapshot and structural assertions

License

MIT OR Apache-2.0