S/ISO/PDF

PDF stands for Portable Document Format, which was created by Adobe[1], and currently maintained by the International Organization for Standardization (ISO) as an open source international standard [2].

Some commonly used specialized PDF types include:

  • ISO 14289/PDF/UA for accessible PDF documents and processors (extends PDF/A conformance level A)

  • ISO 15930/PDF/X for printing

  • ISO 19005/PDF/A for long-term archiving [3]

    • Sub-parts:

      • ISO 19005-1:2005/PDF/A-1 (based on PDF v1.4)

      • ISO 19005-2:2011/PDF/A-2 (based on PDF v1.7)

      • ISO 19005-3:2012/PDF/A-3 (add file)

      • ISO 19005-4:2020/PDF/A-4 (based on PDF v2.0)

    • Not allow: audio, video, 3d objects, JS, certain actions, encryption, non-standard metadata

    • Require: embedding font with proper license

  • ISO 24517/PDF/E for representing engineering documents (CAD, etc.).

For regulatory submission, FDA currently support “PDF versions 1.4 through 1.7, PDF/A-1 and PDF/A-2”[4]. Steps for creating and validating PDF/A files can be found in reference [5],[6].

The module stdiso.pdfsummary depends on package pypdf [7]. The module stdiso.pdfsummary include functions for creating summaries about a specified PDF file.

PDF File Summary

To use mtbp3.stdiso:

from mtbp3.stdiso.pdfsummary import pdfSummary

pfr = pdfSummary(path="")
print(pfr.get_summary_string())

File name: attention.pdf
File size (byte): : 2215244
File creation date: 2024-04-10 21:11:43+00:00
File modification date: 2024-04-10 21:11:43+00:00
PDF version: 1.5
Number of pages: : 15
Number of images in individual pages: : [0, 0, 1, 2, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]
Total number of images: : 3

If the path left as empty, an example pdf file will be loaded for illustration. More details about the example pdf can be found here: https://arxiv.org/abs/1706.03762.

To view the outline tree:

print(pfr.show_outline_tree())

    attention.pdf
    ├── Introduction
    ├── Background
    ├── Model Architecture
    │   ├── Encoder and Decoder Stacks
    │   ├── Attention
    │   │   ├── Scaled Dot-Product Attention
    │   │   ├── Multi-Head Attention
    │   │   └── Applications of Attention in our Model
    │   ├── Position-wise Feed-Forward Networks
    │   ├── Embeddings and Softmax
    │   └── Positional Encoding
    ├── Why Self-Attention
    ├── Training
    │   ├── Training Data and Batching
    │   ├── Hardware and Schedule
    │   ├── Optimizer
    │   └── Regularization
    ├── Results
    │   ├── Machine Translation
    │   ├── Model Variations
    │   └── English Constituency Parsing
    └── Conclusion

Work with Images

We can see that there is one image in the 3rd page from the summary above. To extract the first image on the 3rd page:

img = pfr.get_image(page_index=2, image_index=0, outfolder='')
print(type(img))
print(img.size)

<class 'PIL.PngImagePlugin.PngImageFile'>
(1520, 2239)

display(img.resize((300, int(300*img.size[1]/img.size[0]))))
_images/8691dc1ee94c67631c3d83e9e0c7bfeac1048cb3bd155daf14e6ff5a08f94e51.png

The resize() function above resized the figure before displaying. Use display(img) in Jupyter if resizing is not required.

To save the 2nd image on the 4th page to a file, add an existing folder path using outfolder='add_path_here':

img_path = pfr.get_image(page_index=3, image_index=1, outfolder='.')

The function get_image() returns a file path instead of the image when the outfolder option is not an empty string. To read and display the saved image file in Jupyter:

from IPython.display import Image 

img = Image(filename=img_path, width=300)
display(img)
_images/2973dff4b153b750c37e9aa3fb18981b1ae9455832391354ce31f681e242b511.png

Reference