Exis.PdfEditor for Python

Comprehensive PDF toolkit for Python — find/replace, merge, split, form filling, redaction, image editing, watermark, Bates stamping, page editing, encryption, optimization, digital signatures, PDF/A compliance, XMP + /Info metadata, OCR for scanned PDFs, and more.

Powered by the Exis.PdfEditor .NET library, compiled to a native binary via .NET Native AOT. No .NET runtime required.

Installation

pip install exis-pdfeditor

Platform-specific wheels are available for:

• Windows x64 (win_amd64)
• Linux x64 (manylinux_2_17_x86_64)
• macOS ARM64 / Apple Silicon (macosx_11_0_arm64)

Quick start

import exis_pdfeditor

# Optional: set a license key (or use EXIS_PDF_LICENSE_KEY env var).
# Without a key, a 14-day trial is activated automatically.
exis_pdfeditor.initialize("XXXX-XXXX-XXXX-XXXX")

# Inspect a PDF (free — no license required)
info = exis_pdfeditor.inspect("document.pdf")
print(f"{info.pageCount} pages, title: {info.title}")

# Find and replace text
result = exis_pdfeditor.find_replace(
    "input.pdf", "output.pdf",
    "old text", "new text",
    case_sensitive=False,
)
print(f"{result.totalReplacements} replacements made")

Features

Feature	Function
Inspect	inspect(path) — metadata, fonts, pages, encryption status
Text extraction	extract_text(path), extract_text_structured(path)
Find & replace	find_replace(...) — regex, case-insensitive, whole-word, wrap, styling
Merge	merge(paths, output)
Split	split(path, output_dir) — per page, or by_bookmarks=True
Bookmarks	bookmarks(path) — list outline / TOC
Extract pages	extract_pages(path, output, pages=[1,3,5])
Form fields	list_fields(...), fill_form(...)
Signature images	list_signature_fields(path), fill_signature(...)
Redaction	redact(path, output, redactions)
Region edit	erase_region(...), replace_region(...)
Vector text	find_vector_text(...), replace_vector_text(...)
Vector graphics	find_vector_graphics(...), replace_vector_graphic(...)
Watermark	watermark(path, output, "DRAFT", position="across")
Stamp	stamp(path, output, stamp_pdf, mode="overlay")
Bates stamping	bates_stamp(path, output, prefix="ABC", digits=6) — optional qr_code={...} beside the number
QR codes / Code 128	qr_stamp(path, output, "https://…/{Page}"), qr_placeholders(...) (mail merge), qr_validate(...)
Build PDFs	PdfBuilder() (absolute positions), DocumentBuilder() (auto layout) — text, images, shapes, QR codes
Optimize	optimize(path, output, downsample_images=True)
Encrypt / Decrypt	encrypt(...), decrypt(...)
Page editing	rotate(...), crop(...), reorder(...), delete_pages(...), insert_blank_pages(...)
Images	find_images(path), replace_image(...)
Digital signatures	sign(...), verify(...)
PDF/A	pdfa_validate(path), pdfa_convert(path, output)
Metadata (XMP + /Info)	get_metadata(path), set_xmp(...), set_info(...), remove_xmp(...), remove_info(...), remove_metadata(...)
Page classification	analyze_pages(path) — pre-OCR triage (free, no license)
OCR — searchable PDF (Windows)	make_searchable_pdf(path, output, languages=..., progress=...)
OCR — de-identify scans (Windows)	redact_scanned_pdf(path, output, terms=..., visible_replacement=...)

---

Platform support

Function group	Windows	Linux	macOS
All non-OCR functions (find/replace, merge, forms, redaction, watermark, optimization, signatures, PDF/A, metadata, …)	yes	yes	yes
analyze_pages — page classification	yes	yes	yes
make_searchable_pdf, redact_scanned_pdf — OCR	yes	not yet — raises OcrNotSupportedError	not yet — raises OcrNotSupportedError

OCR is currently Windows-only because the page rasterizer (Exis.PdfOcr.Windows) depends on the WinRT Windows.Data.Pdf API and System.Drawing.Common. When a cross-platform rasterizer ships, the Python side will pick it up automatically.

---

Inspect

Retrieve metadata, page info, fonts, encryption status, and form field counts. No license required.

info = exis_pdfeditor.inspect("document.pdf")

print(f"Version:    {info.version}")
print(f"Pages:      {info.pageCount}")
print(f"Title:      {info.title}")
print(f"Author:     {info.author}")
print(f"Encrypted:  {info.isEncrypted}")
print(f"Has forms:  {info.hasFormFields} ({info.formFieldCount} fields)")
print(f"Fonts:      {info.fontsUsed}")

for page in info.pages:
    print(f"  Page {page.pageNumber}: {page.widthInPoints}x{page.heightInPoints} pt, "
          f"{page.characterCount} chars")

Text extraction

Plain text

result = exis_pdfeditor.extract_text("document.pdf")
print(result.fullText)

for page in result.pages:
    print(f"Page {page.pageNumber}: {page.text[:100]}...")

Specific pages only

result = exis_pdfeditor.extract_text("document.pdf", pages=[1, 3, 5])

Structured text (with positions and font data)

result = exis_pdfeditor.extract_text_structured("document.pdf")

for page in result.pages:
    for block in page.textBlocks:
        print(f"  [{block.x:.0f}, {block.y:.0f}] {block.fontName} {block.fontSize}pt: "
              f"{block.text}")

Find & replace

Single pair

result = exis_pdfeditor.find_replace(
    "input.pdf", "output.pdf",
    "old text", "new text",
)
print(f"{result.totalReplacements} replacements across {result.pagesModified} pages")

Multiple pairs

result = exis_pdfeditor.find_replace(
    "input.pdf", "output.pdf",
    pairs=[
        {"search": "foo", "replace": "bar"},
        {"search": "hello", "replace": "world"},
    ],
)

Regex

result = exis_pdfeditor.find_replace(
    "input.pdf", "output.pdf",
    pairs=[
        {"search": r"\d{3}-\d{2}-\d{4}", "replace": "XXX-XX-XXXX", "isRegex": True},
    ],
)

Multi-line wrap

result = exis_pdfeditor.find_replace(
    "input.pdf", "output.pdf",
    "Short label", "Longer label that\nwraps to two lines",
    wrap="explicit",
    max_lines=3,
    line_spacing=1.2,
)

All options

result = exis_pdfeditor.find_replace(
    "input.pdf", "output.pdf",
    "confidential", "[REDACTED]",
    case_sensitive=False,
    whole_word=True,
    use_regex=False,
    page_range=[1, 2, 3],
    text_fitting="adaptive",         # none, preserve_width, fit_to_page, adaptive
    min_horizontal_scale=70,
    max_font_size_reduction=1.5,
    replacement_text_color={"r": 1, "g": 0, "b": 0},
    replacement_highlight_color={"r": 1, "g": 1, "b": 0},
    replacement_bold=True,
    replacement_underline=False,
    replacement_strikethrough=False,
    preserve_form_fields=True,
    use_incremental_update=True,
)

for detail in result.details:
    print(f"  Page {detail.pageNumber}: '{detail.originalText}' -> '{detail.replacementText}'")

Merge

exis_pdfeditor.merge(["part1.pdf", "part2.pdf", "part3.pdf"], "merged.pdf")

Split

Split a PDF into one file per page:

result = exis_pdfeditor.split("document.pdf", "output_folder/")
print(f"Split into {result.pageCount} files")
for path in result.files:
    print(f"  {path}")

Split by bookmarks (outline / TOC)

result = exis_pdfeditor.split(
    "book.pdf", "chapters/",
    by_bookmarks=True,
    prefix="ch_",
    max_depth=1,
)
print(f"{result.sectionCount} sections")
for s in result.sections:
    print(f"  {s.title}: pages {s.startPage}-{s.endPage} -> {s.file}")

Bookmarks

toc = exis_pdfeditor.bookmarks("document.pdf")
for b in toc.bookmarks:
    print(f"{'  ' * b.level}{b.title} (page {b.startPage})")

Extract pages

Extract specific pages into a new PDF:

exis_pdfeditor.extract_pages("document.pdf", "pages_1_and_3.pdf", pages=[1, 3])

Form fields

Each field has a PDF name, optional displayName (smart label from nearby page text), type, value, options (choice fields), isReadOnly, and hasDuplicateWidgets (true when one logical field has multiple widgets at different positions — e.g. two-up receipts). Pass split_duplicate_widgets=True to list_fields / fill_form to use per-widget keys name_1, name_2, … (page-reading order).

List all fields

fields = exis_pdfeditor.list_fields("form.pdf")

for field in fields:
    print(f"  {field.name} ({field.type}): {field.value}")
    if field.displayName:
        print(f"    Label: {field.displayName}")
    if field.options:
        print(f"    Options: {field.options}")
    if field.hasDuplicateWidgets:
        print("    (duplicate widgets — use split_duplicate_widgets=True for per-widget names)")

List with duplicate-widget split (two-up / carbon copy)

fields = exis_pdfeditor.list_fields("two-up-receipt.pdf", split_duplicate_widgets=True)
# e.g. Address_1, Address_2, date_1, date_2, ...

Fill a form

result = exis_pdfeditor.fill_form(
    "form.pdf", "filled.pdf",
    fields={
        "FirstName": "John",
        "LastName": "Doe",
        "Email": "john@example.com",
        "AgreeToTerms": "Yes",
    },
)
print(f"Filled {result.fieldsFilled} fields, {result.fieldsNotFound} not found")

Force text alignment (default honors each field's PDF /Q)

result = exis_pdfeditor.fill_form(
    "form.pdf", "filled_centered.pdf",
    fields={"Comments": "Centered text"},
    text_alignment="center",  # "auto" | "left" | "center" | "right"
)

Fill duplicate-widget fields (different value per widget; auto-flattens)

result = exis_pdfeditor.fill_form(
    "two-up-receipt.pdf", "filled.pdf",
    fields={
        "Address_1": "Mr. David G Cruz",
        "date_1": "10-Jan-67",
        "Address_2": "Ms. Claudia Morales",
        "date_2": "22-Feb-26",
    },
    split_duplicate_widgets=True,
)

Fill and flatten (make fields non-editable)

result = exis_pdfeditor.fill_form(
    "form.pdf", "filled_flat.pdf",
    fields={"Name": "Jane Doe"},
    flatten=True,
)

Signature image fields

fields = exis_pdfeditor.list_signature_fields("contract.pdf")
for f in fields:
    print(f.name, f.pageNumber)

exis_pdfeditor.fill_signature(
    "contract.pdf", "signed_image.pdf",
    image_path="signature.png",
    field="Signature1",
    scale="fit",
    flatten=True,
)

Redaction

Redact text

result = exis_pdfeditor.redact(
    "document.pdf", "redacted.pdf",
    redactions=[
        {"text": "John Doe", "replaceWith": "[NAME REDACTED]"},
        {"text": r"\d{3}-\d{2}-\d{4}", "isRegex": True, "replaceWith": "[SSN]"},
    ],
)
print(f"{result.redactionsApplied} redactions applied")

Redact a specific area on a page

result = exis_pdfeditor.redact(
    "document.pdf", "redacted.pdf",
    redactions=[
        {
            "area": {"x": 100, "y": 700, "width": 200, "height": 20},
            "pageNumber": 1,
        },
    ],
)

Combined text + area redaction

result = exis_pdfeditor.redact(
    "document.pdf", "redacted.pdf",
    redactions=[
        {"text": "confidential", "caseSensitive": False},
        {"area": {"x": 50, "y": 50, "width": 500, "height": 30}, "pageNumber": 2},
    ],
)

Region erase / replace

Erase content inside a rectangle, or replace it with an image and/or overlay text:

exis_pdfeditor.erase_region(
    "page.pdf", "erased.pdf",
    page=1,
    rect=(72, 400, 200, 40),  # x, y, w, h in PDF points
)

exis_pdfeditor.replace_region(
    "page.pdf", "patched.pdf",
    page=1,
    rect={"x": 72, "y": 400, "width": 200, "height": 40},
    image="logo.png",
    image_scale="fit",
    background="#FFFFFF",
    text="ACME Corp",
    align="center",
)

Vector text & graphics

Find outlined (path-drawn) text clusters or logos. The AOT CLI ships a blob recognizer (ocr="none") for listing clusters; phrase-matching replace needs a real OCR engine via the .NET Exis.PdfOcr API.

hits = exis_pdfeditor.find_vector_text("outlined.pdf", ocr="none")
for h in hits:
    print(h)

graphics = exis_pdfeditor.find_vector_graphics(
    "branded.pdf",
    thumbnails_dir="thumbs/",
)
for g in graphics:
    print(g.index, g.pageNumber)

exis_pdfeditor.replace_vector_graphic(
    "branded.pdf", "rebranded.pdf",
    image_path="new-logo.png",
    index=graphics[0].index,
    scale="fit",
)

Page classification (pre-OCR triage)

analyze_pages classifies every page as Digital, Scanned, AlreadyOcrd, or Empty purely from the content stream — no rendering and no OCR. Use it to predict which pages will need OCR before paying for a run. Free, no license required, available on every wheel.

pages = exis_pdfeditor.analyze_pages("mixed.pdf")
for p in pages:
    print(f"  page {p.pageNumber}: {p.kind}  "
          f"(text {p.textCoverageRatio:.0%}, images {p.imageCoverageRatio:.0%})")

scanned = [p.pageNumber for p in pages if p.kind == "Scanned"]
if scanned:
    print(f"{len(scanned)} pages need OCR: {scanned}")

Each entry has pageNumber, kind, textCharCount, textCoverageRatio, imageCoverageRatio, and hasInvisibleTextLayer (signature of an already-OCR'd page).

Searchable PDFs (OCR) — Windows only

make_searchable_pdf adds an invisible, selectable text layer over scanned pages so find/replace, extraction, and redaction work on the result like any digital PDF. Born-digital and already-OCR'd pages are passed through untouched — only true scans are rasterized and recognized, which is fast on mixed documents.

result = exis_pdfeditor.make_searchable_pdf(
    "scan.pdf", "searchable.pdf",
    languages=("eng",),                   # default
    progress=lambda p: print(f"  page {p.page}/{p.total}: {p.phase}"),
)

print(f"OCR'd {result.pagesProcessed}, skipped {result.pagesSkipped}, "
      f"avg confidence {result.averageConfidence:.0%}")

# Confidence is never swallowed — flag low-confidence pages for human review.
for page in result.pages:
    if page.wasOcrd and page.confidence < 0.80:
        print(f"  page {page.pageNumber} needs review (conf {page.confidence:.0%})")

Additional languages

The wheel ships English (eng.traineddata). To add languages, drop the matching xxx.traineddata files into the wheel's bundled tessdata/ folder, or place them in your own folder and point at it:

result = exis_pdfeditor.make_searchable_pdf(
    "scan.pdf", "searchable.pdf",
    languages=("eng", "spa", "vie"),
    tessdata_path=r"C:\my-app\tessdata",
)

The invisible text layer embeds a Unicode font (DejaVu Sans), so accented and non-Latin names (e.g. Nguyễn, Peña) survive find/replace intact rather than being mangled.

Options

Parameter	Default	Notes
languages	("eng",)	Tesseract language codes
dpi	300	Rendering DPI for non-image pages (full-page scans use their native resolution)
apply_deskew	True	Pre-process hint (engine-dependent)
apply_denoise	False	Pre-process hint (engine-dependent)
min_confidence_to_include	0.0	Words below this are still included (dropping them would risk leaking names you need to match) but counted in per-page confidence
tessdata_path	bundled	Override the trained-data folder
progress	None	Callback (page, total, phase, confidence)

De-identify scanned pages — Windows only

redact_scanned_pdf finds terms on scanned pages via OCR and burns opaque boxes into the actual page image. Unlike OCR-layer redaction (which only changes selectable text), this destroys the underlying pixels — the term is gone from the visible scan and unrecoverable, appropriate for sending de-identified scans externally.

r = exis_pdfeditor.redact_scanned_pdf(
    "intake_scan.pdf", "intake_redacted.pdf",
    terms=["Jane Doe", "555-12-3456"],
    visible_replacement="[REDACTED]",     # drawn into the raster
    fill="white",                         # or "black"
)

print(f"Redacted {r.occurrencesRedacted} occurrences on {r.pagesAffected} pages")
if r.pagesSkipped:
    print(f"WARNING: {r.pagesSkipped} pages could not be safely burned — "
          "review them before sending the output.")

for occ in r.occurrences:
    print(f"  page {occ.pageNumber}: {occ.term!r} (conf {occ.confidence:.0%})")

Scope: this handles image-based pages. Born-digital text pages are NOT modified — use redact(...) for those.

Pages that are image-based but not a single full-page image are reported in pagesSkipped and left unchanged. A non-zero value means the output is not safe to send as-is.

The output is rewritten via the optimizer to drop the superseded original image objects, so the un-redacted pixels are not recoverable with PDF forensics.

Watermark

Basic diagonal watermark

result = exis_pdfeditor.watermark("input.pdf", "watermarked.pdf", "DRAFT")
print(f"Watermarked {result.pagesWatermarked} of {result.totalPages} pages")

All watermark options

result = exis_pdfeditor.watermark(
    "input.pdf", "watermarked.pdf",
    "CONFIDENTIAL",
    position="across",          # top, bottom, center, across
    font_size=72,
    text_color={"r": 1, "g": 0, "b": 0},
    opacity=0.15,
    page_range=[1, 2],
)

Stamp (PDF overlay / underlay)

Overlay a letterhead on top

result = exis_pdfeditor.stamp(
    "document.pdf", "stamped.pdf",
    "letterhead.pdf",
    mode="overlay",
)
print(f"Stamped {result.pagesStamped} pages")

Underlay a background behind content

result = exis_pdfeditor.stamp(
    "document.pdf", "stamped.pdf",
    "background.pdf",
    mode="underlay",
    opacity=0.5,
    page_range=[1],
)

Bates stamping

Sequential page numbering for legal production and discovery workflows. Each page receives a zero-padded identifier (e.g. ABC000001, ABC000002) rendered in a chosen corner of the visual page. Placement is relative to the page's /Rotate orientation, so mixed-rotation documents render consistently. An XMP audit block recording the range, digit width, and prefix/suffix is written to the document catalog by default.

Defaults (number starts at 1, 6 digits, bottom-right corner)

result = exis_pdfeditor.bates_stamp("input.pdf", "stamped.pdf")
print(f"Stamped pages {result.firstNumber}-{result.lastNumber}")
print(f"Digits used: {result.digitsUsed}")

Prefix, custom position, color, confidentiality label

result = exis_pdfeditor.bates_stamp(
    "input.pdf", "stamped.pdf",
    prefix="ABC",
    start_number=1,
    digits=6,                                 # -> "ABC000001"
    position="bottom_right",                  # top_left, top_center, top_right,
                                              # bottom_left, bottom_center, bottom_right
    font_size=10,
    text_color={"r": 0, "g": 0, "b": 0},
    background_color={"r": 1, "g": 1, "b": 1},  # opaque box behind stamp
    margin_inches=0.5,
    confidentiality_label="CONFIDENTIAL",     # stacked above the Bates number
)

Continuous numbering across a batch

Thread lastNumber + 1 into the next call:

next_n = 1
for path in docs:
    r = exis_pdfeditor.bates_stamp(
        path, path + ".stamped.pdf",
        prefix="ABC", start_number=next_n,
    )
    next_n = r.lastNumber + 1

Skip the cover page (legal convention)

The cover is ABC000001 in the production log even if physically unstamped:

exis_pdfeditor.bates_stamp(
    "input.pdf", "stamped.pdf",
    prefix="ABC",
    skip_first_page=True,
    counter_advances_on_skipped_pages=True,   # default
)

Stamp only selected pages

exis_pdfeditor.bates_stamp(
    "input.pdf", "stamped.pdf",
    page_range=[2, 3, 5],                     # 1-based
)

Signed input

Stamping invalidates signatures — opt in explicitly:

result = exis_pdfeditor.bates_stamp(
    "signed.pdf", "stamped.pdf",
    allow_signed_input=True,                  # result.warnings will record it
)
for w in result.warnings:
    print(f"  warning: {w}")

Suppress the XMP audit block

exis_pdfeditor.bates_stamp(
    "input.pdf", "stamped.pdf",
    write_xmp_metadata=False,
)

All options at a glance

Parameter	Default	Purpose
prefix / suffix	""	Text bracketing the number
start_number	1	First Bates number (for batch continuation)
digits	6	Minimum zero-padded width; auto-expands if needed
position	"bottom_right"	Corner/edge on the visual page
font_size	10.0	Point size (Helvetica)
text_color	black	RGB dict {"r": 0, "g": 0, "b": 0}
background_color	none	Opaque rectangle behind stamp for legibility
margin_inches	0.5	Distance from trimmed page edge
confidentiality_label	none	e.g. "CONFIDENTIAL"
confidentiality_position	same corner	Override to place label elsewhere
confidentiality_font_size	matches font_size	Label point size
page_range	all pages	1-based page numbers to stamp
skip_first_page	False	Skip page 1 (cover sheets)
counter_advances_on_skipped_pages	True	Advance counter on skipped pages
allow_signed_input	False	Stamp PDFs with digital signatures
write_xmp_metadata	True	Write an XMP audit block to the catalog
qr_code	none	Dict of qr_stamp options: a QR code in the same corner, number beside it

Result fields: firstNumber, lastNumber, pagesStamped, digitsUsed (equals digits unless auto-expanded), warnings (non-fatal diagnostics: digit expansion, signed-input stamped anyway, etc.).

Bates number + QR code

exis_pdfeditor.bates_stamp(
    "production.pdf", "stamped.pdf",
    prefix="ABC",
    qr_code={
        "content": "{Bates}|{DocId}",          # default "{Bates}"
        "values": {"DocId": "MATTER-17"},
        "size_points": 54,
    },
)

QR codes & barcodes

Codes are vector paths on an opaque background: sharp at any zoom, no fonts or images added, PDF/A files stay conformant. Placement follows the page as displayed (CropBox and /Rotate), and the code is always upright.

One code per page, with tokens

r = exis_pdfeditor.qr_stamp(
    "input.pdf", "output.pdf",
    "https://example.com/docs/{FileNameNoExt}?page={Page}",
    position="bottom_right",   # top_left, top_center, top_right, bottom_left, bottom_center, center, custom
    size_points=72,            # includes the quiet zone
    error_correction="M",      # L, M, Q, H
    caption="{FileName} p.{Page}",
)
for code in r.codes:
    print(code.pageNumber, code.content, code.version, code.moduleSizeMm)

Tokens: {FileName}, {FileNameNoExt}, {Page}, {PageCount}, {Date}, {InputSha256} (SHA-256 of the original input file), {Bates} (with bates={...}), and any key of values.

One code per document, Bates, Code 128, encrypted input

# First page only, with your own values and the input hash
exis_pdfeditor.qr_stamp("contract.pdf", "out.pdf", "{InvoiceId}|{InputSha256}",
                        values={"InvoiceId": "INV-42"}, pages="first", position="top_right")

# {Bates} follows the same numbering rules as bates_stamp
exis_pdfeditor.qr_stamp("in.pdf", "out.pdf", "{Bates}", bates={"prefix": "ABC", "start_number": 1001})

# Code 128 (ASCII only)
exis_pdfeditor.qr_stamp("in.pdf", "out.pdf", "SHIP-000123", barcode_type="code128",
                        size_points=180, bar_height_points=36)

# Encrypted input is re-encrypted with the same password and permissions
exis_pdfeditor.qr_stamp("locked.pdf", "out.pdf", "x", password="secret")

Will it scan?

v = exis_pdfeditor.qr_validate("https://example.com/a/long/link", size_points=36)
if not v.isValid:
    print(v.message)              # explains the problem and the minimum size
print(v.minimumSizePoints)

A module smaller than 0.5 mm (QR) or 0.25 mm (Code 128) is an error unless allow_undersized=True. Colors must be dark on light with at least 4.5:1 contrast.

Mail merge: {{QR:Column}} placeholders

Put {{QR:TicketId}} in the template; each token is replaced by a code whose top-left corner sits where the token began:

import csv

with open("attendees.csv", newline="", encoding="utf-8") as f:
    for row in csv.DictReader(f):
        exis_pdfeditor.qr_placeholders(
            "ticket_template.pdf", f"ticket_{row['TicketId']}.pdf",
            values=row,
            require_values=True,   # a misspelt column is an error
            size_points=90,
        )

Build PDFs from scratch

b = exis_pdfeditor.PdfBuilder(title="Ticket")
page = b.add_page()                                   # Letter; add_page(width, height) in points
page.add_text("Admit one", 72, 700, 24, bold=True)
page.add_rectangle(60, 480, 300, 260, stroke_color="#333333")
page.add_qr_code("T-0001", 72, 500, 144, caption="T-0001")
b.build("ticket.pdf")

d = exis_pdfeditor.DocumentBuilder(page_size="a4", margin=54, title="Report")
d.add_paragraph("Quarterly report", 18, bold=True)
d.add_paragraph("Scan for the online version:")
d.add_qr_code("https://example.com/r/q1", 108, align="center")
d.add_horizontal_rule()
d.build("report.pdf")

Optimize

Default optimization (compress + deduplicate)

result = exis_pdfeditor.optimize("large.pdf", "smaller.pdf")
print(f"Reduced {result.originalSize:,} -> {result.optimizedSize:,} bytes "
      f"({result.reductionPercent:.1f}% smaller)")
print(f"  Streams compressed: {result.streamsCompressed}")
print(f"  Duplicates removed: {result.duplicatesRemoved}")

With image downsampling

result = exis_pdfeditor.optimize(
    "large.pdf", "smaller.pdf",
    downsample_images=True,
    max_image_dpi=150,
    remove_metadata=True,
)
print(f"Images downsampled: {result.imagesDownsampled}")

Encrypt & decrypt

Encrypt with a password

exis_pdfeditor.encrypt(
    "document.pdf", "protected.pdf",
    user_password="openme",
    owner_password="secret",
    permissions=["Print", "CopyText"],
)

Available permissions: Print, ModifyContents, CopyText, AddAnnotations, FillForms, PrintHighQuality, All.

Decrypt

exis_pdfeditor.decrypt("protected.pdf", "unlocked.pdf", password="openme")

Page editing

Rotate pages

result = exis_pdfeditor.rotate("input.pdf", "rotated.pdf", angle=90)
print(f"Rotated {result.pagesModified} pages")

# Rotate only specific pages
result = exis_pdfeditor.rotate("input.pdf", "rotated.pdf", angle=180, pages=[2, 4])

Crop pages

result = exis_pdfeditor.crop(
    "input.pdf", "cropped.pdf",
    rect={"x": 50, "y": 50, "width": 500, "height": 700},
)

Reorder pages

exis_pdfeditor.reorder("input.pdf", "reordered.pdf", order=[3, 1, 2])

Delete pages

exis_pdfeditor.delete_pages("input.pdf", "trimmed.pdf", pages=[2, 4])

Insert blank pages

Each insertion specifies a 0-based "after page" anchor (0 = before page 1) and an optional page size. Supported sizes: letter (default), legal, a4, a3, a5, tabloid.

exis_pdfeditor.insert_blank_pages(
    "input.pdf", "with_blanks.pdf",
    insertions=[
        {"afterPage": 0, "size": "a4"},       # before page 1
        {"afterPage": 3, "size": "letter"},   # after page 3
    ],
)

Images

Find all images

result = exis_pdfeditor.find_images("document.pdf")
print(f"Found {result.totalImages} images across {result.pagesSearched} pages")

for img in result.images:
    print(f"  Image {img.index}: {img.pixelWidth}x{img.pixelHeight} "
          f"{img.colorSpace} {img.format}")
    print(f"    Pages: {img.pageNumbers}")

Find images and save to disk

result = exis_pdfeditor.find_images("document.pdf", output_dir="extracted_images/")

Replace all images

result = exis_pdfeditor.replace_image(
    "document.pdf", "replaced.pdf",
    "new_logo.png",
)
print(f"Replaced {result.imagesReplaced} of {result.imagesFound} images")

Replace specific images by index or page

result = exis_pdfeditor.replace_image(
    "document.pdf", "replaced.pdf",
    "new_logo.jpg",
    image_indices=[0, 2],
    page_range=[1],
    scale_mode="scale_to_fit",  # match_original_size, preserve_aspect_ratio, scale_to_fit
)

Digital signatures

Sign a PDF

exis_pdfeditor.sign(
    "document.pdf", "signed.pdf",
    cert_path="certificate.pfx",
    cert_password="certpass",
    reason="Approved",
    location="New York, NY",
    signer_name="John Doe",
)

Sign with a visible signature

exis_pdfeditor.sign(
    "document.pdf", "signed.pdf",
    cert_path="certificate.pfx",
    cert_password="certpass",
    visible=True,
    page=1,
    rect={"x": 50, "y": 50, "width": 200, "height": 60},
    reason="Reviewed and approved",
)

Verify signatures

info = exis_pdfeditor.verify("signed.pdf")
print(f"Signed: {info.isSigned}")
print(f"Signer: {info.signerName}")
print(f"Valid:  {info.isValid}")
print(f"Reason: {info.reason}")
print(f"Date:   {info.signDate}")

# Verify all signatures in a multi-signed document
signatures = exis_pdfeditor.verify("multi_signed.pdf", all_signatures=True)
for sig in signatures:
    print(f"  {sig.signerName}: valid={sig.isValid}")

PDF/A compliance

Validate

result = exis_pdfeditor.pdfa_validate("document.pdf", level="2b")
print(f"Compliant: {result.isCompliant}")

if not result.isCompliant:
    for v in result.violations:
        print(f"  [{v.code}] {v.message} (auto-fixable: {v.canAutoFix})")

Convert to PDF/A

exis_pdfeditor.pdfa_convert("document.pdf", "archive.pdf", level="2b")

Supported levels: 1b, 2b, 2u, 3b, 3u.

XMP + /Info metadata

PDFs carry document metadata in two places:

• XMP — the modern RDF/XML packet referenced from the Catalog /Metadata entry.
• /Info — the legacy trailer dictionary (Title, Author, Subject, Keywords, Creator, Producer, CreationDate, ModDate, plus custom keys).

get_metadata() is free and works on encrypted PDFs (XMP only). All mutation calls (set_xmp, set_info, remove_xmp, remove_info, remove_metadata) require a license and reject encrypted input — decrypt first with decrypt().

Read metadata

meta = exis_pdfeditor.get_metadata("document.pdf")

if meta.hasXmp:
    print(f"XMP packet: {meta.xmpByteSize} bytes")
    print(meta.xmpXml)          # Full <?xpacket ... ?> payload as UTF-8 text

if meta.hasInfo:
    info = meta.info
    print(f"Title:    {info.title}")
    print(f"Author:   {info.author}")
    print(f"Subject:  {info.subject}")
    print(f"Keywords: {info.keywords}")
    print(f"Creator:  {info.creator}")
    print(f"Producer: {info.producer}")
    print(f"Created:  {info.creationDate}")        # ISO 8601 string or None
    print(f"Modified: {info.modificationDate}")    # same

    # Non-standard /Info keys (e.g. "Company", custom producer fields)
    for key, value in vars(info.custom).items():
        print(f"  [custom] {key} = {value}")

Replace the XMP packet

Pass a full <?xpacket ...?>-wrapped RDF/XML document as a string:

xmp = """<?xpacket begin=""?>
<x:xmpmeta xmlns:x="adobe:ns:meta/">
  <rdf:RDF xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#">
    <rdf:Description xmlns:dc="http://purl.org/dc/elements/1.1/">
      <dc:title><rdf:Alt><rdf:li xml:lang="x-default">Quarterly Report</rdf:li></rdf:Alt></dc:title>
      <dc:creator><rdf:Seq><rdf:li>Finance Team</rdf:li></rdf:Seq></dc:creator>
    </rdf:Description>
  </rdf:RDF>
</x:xmpmeta>
<?xpacket end="w"?>"""

exis_pdfeditor.set_xmp("input.pdf", "output.pdf", xmp)

Passing an empty string writes an empty packet (still present, no data). To drop the /Metadata reference entirely, use remove_xmp().

Replace the /Info dictionary

The PDF must already have an /Info reference in its trailer; otherwise this raises PdfEditorError. Fields you omit are written as missing (not empty strings) so "not set" semantics are preserved.

from datetime import datetime, timezone

exis_pdfeditor.set_info(
    "input.pdf", "output.pdf",
    info={
        "title":    "Quarterly Report",
        "author":   "Finance Team",
        "subject":  "Q1 2026 earnings summary",
        "keywords": "earnings, Q1, 2026",
        "creator":  "exis-pdfeditor",
        "producer": "Exis.PdfEditor 3.6",
        "creationDate":     datetime(2026, 4, 1, 9, 0, tzinfo=timezone.utc),
        "modificationDate": datetime.now(timezone.utc),
        # Arbitrary non-standard /Info keys
        "custom": {
            "Company":  "Exis LLC",
            "Revision": "v3.6.4",
        },
    },
)

Dates can be provided as datetime / date objects or ISO 8601 strings (e.g. "2026-04-20T09:00:00Z"). Snake_case aliases creation_date and modification_date are also accepted.

Remove metadata (wipe for privacy)

# Drop the XMP packet only
exis_pdfeditor.remove_xmp("input.pdf", "stripped-xmp.pdf")

# Empty the /Info dict (trailer reference kept, contents cleared)
exis_pdfeditor.remove_info("input.pdf", "stripped-info.pdf")

# Drop both in one incremental update (smaller output than calling both)
exis_pdfeditor.remove_metadata("input.pdf", "stripped.pdf")

Diagnostic structure dump

When a PDF fails to process and you can't share the file, dump_structure() produces a self-contained report you can paste into a bug report. It walks every object in the file, tallies filter chains and font subtypes, lists encryption details, and records any streams that fail to decode — without needing the original file. No license required.

Human-readable report

dump = exis_pdfeditor.dump_structure("problem.pdf")

# The dump has a built-in text report — just print it
# (or paste it into a support email / bug report)
print(dump)

Individual fields

dump = exis_pdfeditor.dump_structure("problem.pdf")

print(f"PDF version:    {dump.version}")
print(f"Pages:          {dump.pageCount}")
print(f"Total objects:  {dump.totalObjects}")
print(f"Stream objects: {dump.streamObjectCount}")
print(f"Xref entries:   {dump.xrefEntryCount}")
print(f"Xref streams:   {dump.usesXrefStreams}")
print(f"Producer:       {dump.producer}")
print(f"Creator:        {dump.creator}")

# Encryption details
print(f"Encrypted:      {dump.isEncrypted}")
if dump.isEncrypted:
    print(f"  Version:      {dump.encryptionVersion}")
    print(f"  Revision:     {dump.encryptionRevision}")
    print(f"  Key length:   {dump.encryptionKeyLengthBits} bits")

# Catalog flags
print(f"AcroForm:       {dump.hasAcroForm}")
print(f"Signed:         {dump.hasDigitalSignature}")
print(f"Embedded files: {dump.hasEmbeddedFiles}")

# Filter chains — which compression methods are used and how often
for chain in dump.filterChains:
    print(f"  {chain}: {dump.filterChains[chain]} streams")

# Font subtypes
for subtype in dump.fontSubtypes:
    print(f"  /{subtype}: {dump.fontSubtypes[subtype]}")

# Streams that failed to decode (capped at 50)
for bad in dump.unsupportedStreams:
    print(f"  obj {bad.objectNumber} [{bad.filterChain}]: {bad.error}")

# Free-form notes about parse anomalies
for note in dump.notes:
    print(f"  {note}")

---

Licensing

• Free: inspect(), pdfa_validate(), dump_structure(), and get_metadata() work without a license.
• Trial: Call exis_pdfeditor.initialize() with no key for a 14-day full-feature trial.
• Licensed: Pass your key to exis_pdfeditor.initialize("XXXX-XXXX-XXXX-XXXX") or set the EXIS_PDF_LICENSE_KEY environment variable.
• Evaluation: After trial expiry, all features work on documents up to 3 pages.

Purchase a license at pdfbatcheditor.com/developers.

Requirements

• Python 3.9+
• No external dependencies — the native binary is bundled in the wheel.