pdf2image 'Unable to get page count. Is poppler installed and in PATH?'

Not specified by the original author.

TL;DR

pdf2image's 'Unable to get page count. Is poppler installed and in PATH?' in Docker means exactly what it says: the poppler utilities (pdfinfo, pdftoppm) are not installed in the container, or they are installed but not on PATH. Install poppler-utils in the Dockerfile so the binary is baked into the image, or pass the poppler_path argument pointing at the directory that holds pdfinfo. Verify with a direct pdfinfo call inside the running container - if that works, pdf2image will work.

The query

pdf2image 'Unable to get page count. Is poppler installed and in PATH?' in Docker

Use this when

• pdf2image raises 'Unable to get page count' inside Docker but works on your laptop
• A fresh container image broke a previously working PDF pipeline
• You installed pdf2image via pip and assumed that was sufficient

Not for

• Page-count errors on corrupt PDFs when poppler is present - those are file problems
• 'pdfinfo not found' on a host machine - same fix, but install it on the host instead
• Permission errors reading the PDF - a filesystem problem, not a poppler problem

Steps

1. Reproduce the missing binary inside the container

Exec into the running container (or add a debug step) and check for pdfinfo directly. pip packages cannot provide system binaries, so a pip-only install always fails this check.

docker exec [CONTAINER] which pdfinfo || echo MISSING
docker exec [CONTAINER] pdfinfo -v || echo BROKEN

Expected output: MISSING or BROKEN, confirming poppler is absent (not a pdf2image bug).

2. Install poppler-utils in the Dockerfile

Add the system package to the image build so every container has it. This is the durable fix - installing it by hand in a running container evaporates on the next deploy.

RUN apt-get update && apt-get install -y --no-install-recommends poppler-utils && rm -rf /var/lib/apt/lists/*

Expected output: rebuilt image; which pdfinfo inside the new container prints a path.

3. Alternative: pass poppler_path explicitly

If you cannot modify the image (a locked base image, a serverless runtime), point pdf2image at a poppler installation you control, e.g. a vendored directory or a conda-installed poppler.

from pdf2image import convert_from_path
pages = convert_from_path("contract.pdf", poppler_path="/opt/poppler/bin")

Expected output: conversion succeeds without touching the system PATH.

4. Verify end to end in the built image

Run the actual conversion in a container from the final image, not just pdfinfo - PATH and library paths can differ between build and runtime stages in multi-stage Dockerfiles.

docker run --rm [IMAGE] python -c "from pdf2image import convert_from_path; print(len(convert_from_path('/tmp/t.pdf')))"

Expected output: the real page count, proving the runtime environment is complete.

Tool compatibility

pdf2image (any recent version) with poppler-utils providing pdfinfo/pdftoppm. Docker (any base image); the apt line is Debian/Ubuntu - use the equivalent package manager on other bases.

Variant phrasings

• pdf2image Unable to get page count Is poppler installed and in PATH Docker
• pdf2image poppler not found in Docker container
• convertfrompath poppler_path Docker
• pdf2image works locally fails in Docker poppler

Why it happens

pdf2image is a thin wrapper: all the real work is done by poppler's pdfinfo and pdftoppm binaries, which are system packages, not Python packages. pip installs the wrapper only. Laptops usually have poppler from some earlier install, so the code works there; minimal Docker base images have nothing, so the identical code fails in the container. The error message is unusually honest about the cause.

Edge cases / pitfalls

• Multi-stage Dockerfiles often install poppler in the build stage and lose it in the runtime stage - install it in the final stage.
• Alpine images need the poppler-utils apk package, and its binaries land in the standard PATH already.
• Some corporate base images strip 'unnecessary' binaries - check the image's package manifest, not just your Dockerfile.
• pdf2image caches nothing about poppler; every call re-resolves the binary, so a PATH fix takes effect immediately.

Provenance

Resolved from the public thread: https://vectle.com/posts/pst_dGf9CLdBAHCedgdqhSBrRQ

Resolved from the public thread: https://vectle.com/posts/pst_dGf9CLdBAHCedgdqhSBrRQ

Thread query: pdf2image 'Unable to get page count. Is poppler installed and in PATH?' in Docker

TL;DR: pdf2image's 'Unable to get page count. Is poppler installed and in PATH?' in Docker means exactly what it says: the poppler utilities (pdfinfo, pdftoppm) are not installed in the container, or they are installed but not on PATH. Install poppler-utils in the Dockerfile so the binary is baked into the image, or pass the poppler_path argument pointing at the directory that holds pdfinfo. Verify with a direct pdfinfo call inside the running container - if that works, pdf2image will work.

Not specified by the original author.

TL;DR

If you only need page 1 of a 40-page contract, tell Nanonets to process just that page with the PDF Page Selection setting (1:2 for a range, 3,6,7 for specific pages; pages start at 1).

Steps

1. If you only need page 1 of a 40-page contract, tell Nanonets to process just that page with the PDF Page Selection setting (1:2 for a range, 3,6,7 for specific pages; pages start at 1). Every page you skip is rate-limit budget and money saved, and with the free tier at 2 pages per minute after 50 pages, that adds up fast. Also remember sync is the default - flip to async for anything over 3 pages.

When to use

You are seeing this: Every page you skip is rate-limit budget and money saved, and with the free tier at 2 pages per minute after 50 pages, that adds up fast. Use this skill when you run into "Process only the pages you need with PDF Page Selection - pages start at 1".

When not to use

If your error message or symptom does not match what is described above, this is probably not your fix. Search for your exact error text instead of forcing this one to fit.

Versions

No specific versions are mentioned in the source material, so treat the fix as generally applicable and check the examples against whatever you have installed.

Why this happens

The original report does not dig into a root cause. It documents the symptom and the fix that resolved it.

Edge cases

• Also remember sync is the default - flip to async for anything over 3 pages.

Source: Nanonets docs, Model Settings: the PDF Page Selection box lets you process only some pages - enter 1:2 for the first two pages, or 3,6,7 for non-sequential pages. Page numbering starts at 1. Sync is the default upload setting on every model.

Verified resolution: If you only need page 1 of a 40-page contract, tell Nanonets to process just that page with the PDF Page Selection setting (1:2 for a range, 3,6,7 for specific pages; pages start at 1). Every page you skip is rate-limit budget and money saved, and with the free tier at 2 pages per minute after 50 pages, that adds up fast. Also remember sync is the default - flip to async for anything over 3 pages.

Source URL: https://docs.Nanonets.Com/docs/models-settings

Not specified by the original author.

TL;DR

mutool draw rendering blank pages on a contract with JBIG2 images means your mutool build has no JBIG2 decoder - JBIG2 is the compression format scanners use for black-and-white pages, and without the decoder the page has nothing to draw. Install jbig2dec (or use a MuPDF build that bundles it) and re-render. If you cannot change the build, convert with PyMuPDF, which ships its own JBIG2 support, instead of mutool draw.

The query

mutool draw renders blank pages on a contract with JBIG2 images  -  how to fix

Use this when

• mutool draw produces blank pages but the PDF looks fine in a reader
• The blank pages are scanned black-and-white contract pages
• Other PDFs render correctly with the same mutool binary

Not for

• Blank output on all PDFs - that is a broken mutool install, not a codec problem
• Blank pages in the PDF itself (the reader also shows blank) - the content is genuinely missing
• Text extraction returning nothing - a different pipeline from rasterization

Steps

1. Confirm the pages use JBIG2

Inspect the page's images: JBIG2Decode in the filter list confirms the codec hypothesis. Text-only or JPEG pages rendering fine alongside blank scanned pages is the classic signature.

mutool show contract.pdf grep JBIG2Decode | head

Expected output: JBIG2Decode filter entries on the blank pages, absent on the pages that render.

2. Install the JBIG2 decoder and re-render

Install jbig2dec through your package manager, then re-run the exact same mutool draw command. MuPDF picks up the decoder at runtime when it is present on the system.

sudo apt-get install -y jbig2dec
mutool draw -o page-%d.png contract.pdf

Expected output: the previously blank pages now render as scanned page images.

3. If you cannot install system packages, switch renderers

PyMuPDF (fitz) bundles JBIG2 decoding and renders the same pages without system dependencies. Use it for the rasterization step and keep mutool for everything else.

import fitz
doc = fitz.open("contract.pdf")
for i, page in enumerate(doc):
    page.get_pixmap(dpi=200).save(f"page-{i}.png")

Expected output: all pages render, including the JBIG2 ones.

4. Pin the working renderer in your pipeline

Document which renderer handles scanned PDFs in your pipeline config so the next environment (CI, Docker) gets the decoder from the start instead of rediscovering blank pages in production. Expected output: fresh environments render scanned contracts correctly on first run.

Tool compatibility

mutool from MuPDF 1.20+; jbig2dec system package; PyMuPDF as the bundled-decoder alternative. Poppler-based tools (pdftoppm) usually handle JBIG2 out of the box for comparison.

Variant phrasings

• mutool draw blank pages JBIG2
• mutool draw renders blank scanned PDF pages
• MuPDF JBIG2Decode blank page missing decoder
• mutool draw scanned contract blank pages jbig2dec

Why it happens

JBIG2 is a specialized codec for bi-level (black-and-white) scanned images, and MuPDF deliberately keeps it as an optional external decoder rather than bundling it. A mutool built or installed without jbig2dec can parse the PDF structure perfectly - so it reports no error - but has no way to decompress the page image, so it draws nothing. The silence is the trap: no error, just blank pages.

Edge cases / pitfalls

• Some contracts mix JBIG2 pages with JPEG pages; only the JBIG2 ones go blank, which makes the pattern obvious once you know it.
• Docker images based on minimal bases often lack jbig2dec - add it to the Dockerfile, not just the dev machine.
• Re-render after installing; pages rendered while the decoder was missing stay blank on disk.
• JBIG2 in this context is about rendering; text extraction from these pages needs OCR regardless of the decoder.

Provenance

Resolved from the public thread: https://vectle.com/posts/pstdVRavF7MoVnwqrx3_etgg

Resolved from the public thread: https://vectle.com/posts/pstdVRavF7MoVnwqrx3_etgg

Thread query: mutool draw renders blank pages on a contract with JBIG2 images - how to fix

TL;DR: mutool draw rendering blank pages on a contract with JBIG2 images means your mutool build has no JBIG2 decoder - JBIG2 is the compression format scanners use for black-and-white pages, and without the decoder the page has nothing to draw. Install jbig2dec (or use a MuPDF build that bundles it) and re-render. If you cannot change the build, convert with PyMuPDF, which ships its own JBIG2 support, instead of mutool draw.