An out of memory issue occurs when loading and retrieving images from the 2 GB document

[CCMKarthik123]

We are loading the 2 GB document and retrieving images from it. In this case, an out of memory issue occurs. We have attached a simple sample; please check and provide a solution to resolve this issue.

Run the attached sample, load the 2 GB document using the open option, and click the RenderPage button to reproduce the issue.

SimpleSample.zip

[makingbloke]

This isn’t the group for PDFium support (You need to try one of the official groups). I’ve had similar issues but you can write a custom load method which only loads the parts of the PDF it needs. In my case it loaded it from a C# stream. Unfortunately, I cannot give you the source code or an example. Cheers

…

On Wed, 19 Mar 2025 at 14:55, CCMKarthik ***@***.***> wrote: We are loading the 2 GB document and retrieving images from it. In this case, an out of memory issue occurs. We have attached a simple sample; please check and provide a solution to resolve this issue. Run the attached sample, load the 2 GB document using the open option, and click the RenderPage button to reproduce the issue. SimpleSample.zip <https://github.com/user-attachments/files/19343231/SimpleSample.zip> image.png (view on web) <https://github.com/user-attachments/assets/35ad9a76-9b43-4308-a82e-9fa28909a1ee> — Reply to this email directly, view it on GitHub <#196>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABPMCPKK2DYXR2AG6CGJF432VGAO5AVCNFSM6AAAAABZLCA2V6VHI2DSMVQWIX3LMV43ASLTON2WKOZSHEZTEMJQHE4DMMA> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***> [image: CCMKarthik123]*CCMKarthik123* created an issue (bblanchon/pdfium-binaries#196) <#196> We are loading the 2 GB document and retrieving images from it. In this case, an out of memory issue occurs. We have attached a simple sample; please check and provide a solution to resolve this issue. Run the attached sample, load the 2 GB document using the open option, and click the RenderPage button to reproduce the issue. SimpleSample.zip <https://github.com/user-attachments/files/19343231/SimpleSample.zip> image.png (view on web) <https://github.com/user-attachments/assets/35ad9a76-9b43-4308-a82e-9fa28909a1ee> — Reply to this email directly, view it on GitHub <#196>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABPMCPKK2DYXR2AG6CGJF432VGAO5AVCNFSM6AAAAABZLCA2V6VHI2DSMVQWIX3LMV43ASLTON2WKOZSHEZTEMJQHE4DMMA> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

[bblanchon]

Our WASM build uses a 32-bit address space, so it is limited to 2GB.

@jerbob92, do you think we should include a 64-bit build (see MEMORY64)?

[jerbob92]

@bblanchon Yeah a 64 bit build would make sense, but @makingbloke is correct as well, it's not a good idea to fully load a 2GB PDF into memory.

[CCMKarthik123]

Thank you for the update. We have generated the pdfium.js and pdfium.wasm files. We were able to load the 2 GB document without any errors; however, when rendering the image, it shows a blank page after the 105th page. We have provided the updated sample below.

Sample: SimpleSampleUpdated.zip

Below are the changes we made in the build.sh file to generate the pdfium.js and pdfium.wasm files to load the 2 GB document. Please review them and provide a solution to address this issue.

SOURCE=${PDFium_SOURCE_DIR:-pdfium}
BUILD_DIR=${PDFium_BUILD_DIR:-$SOURCE/out}
TARGET_CPU=${PDFium_TARGET_CPU:?}
IS_DEBUG=${PDFium_IS_DEBUG:-false}
ninja -C "$BUILD_DIR" pdfium
if [ "$TARGET_CPU" == "wasm" ]; then
LIBPDFIUMA="$BUILD_DIR/obj/libpdfium.a"
EXPORTED_FUNCTIONS=$(llvm-nm $LIBPDFIUMA --format=just-symbols | grep "^FPDF|^FSDK|^FORM|^IFSDK" | sed 's/^/_/' | paste -sd "," -)
EMCC_ARGS=(
-s ALLOW_MEMORY_GROWTH=1
-s ALLOW_TABLE_GROWTH=1
-s EXPORTED_FUNCTIONS="$EXPORTED_FUNCTIONS,_free,_malloc,_calloc,_realloc"
-s EXPORTED_RUNTIME_METHODS="ccall,cwrap,addFunction,setValue,removeFunction"
-s LLD_REPORT_UNDEFINED
-s WASM=1
-s WASM_MEM_MAX=4294967296
-o "$BUILD_DIR/pdfium.html"
"$LIBPDFIUMA"
--no-entry
)
if [[ "$IS_DEBUG" == "true" ]]; then
EMCC_ARGS+=(
--profile
-g
)
else
# O3 does not work! Strips out too much!
EMCC_ARGS+=(
-O2
)
fi
em++ "${EMCC_ARGS[@]}"
fi

[VenkadaSubramanian-Durai]

Hi team,

Any update on this issue?

[CCMKarthik123]

Hi team,

Any solution for this issue?

[jerbob92]

@CCMKarthik123 did you try implementing the file reader so that you dont have to load the full file at once?

[CCMKarthik123]

@jerbob92 , yes, I am using the FileReader to open the PDF file. However, while retrieving the images, they appear blank after the 105th page. Please refer to the sample below.

Sample: SimpleSampleUpdated.zip

In this sample, when loading a 2 GB document, the images appear blank after the 105th page during retrieval. Kindly check the output images after the 105th page in the console.

[jerbob92]

@jerbob92 , yes, I am using the FileReader to open the PDF file. However, while retrieving the images, they appear blank after the 105th page. Please refer to the sample below.

Sample: SimpleSampleUpdated.zip

In this sample, when loading a 2 GB document, the images appear blank after the 105th page during retrieval. Kindly check the output images after the 105th page in the console.

No idea really, did you try compiling with MEMORY64?