Hyperlink Path Garbled in PDF Using fitz.LINK_LAUNCH

[bagwechien]

The program opens input.pdf, adds a piece of text (PDF) and a hyperlink, then saves it as output.pdf.

The program runs without errors, but when I open output.pdf and click on the hyperlink, the path appears as:
E:冥衄秂æ−•å¤§å, which is clearly garbled and different from the path I specified, making it impossible to open the target file.

I tried using uri, which works for opening external links, but I need a relative path, so I had to switch back to using fitz.LINK_LAUNCH.

Does anyone know how to resolve this issue?

import subprocess
import fitz
import chardet

def add_text_with_link(pdf_path, output_path, text, link_path):
try:
doc = fitz.open(pdf_path)
except Exception as e:
print("無法開啟 PDF: ", e)
return

fontsize = 20
color = (0, 0, 1) ）

page = doc[0]  
text_rect = fitz.Rect(50, 100, 400, 150)  # 設定文字區域 (x1, y1, x2, y2)

page.insert_textbox(
    text_rect, text,
    fontsize=fontsize,
    color=color,
    fontname="china-t"
)


page.insert_link({
    "kind": fitz.LINK_LAUNCH,  # 使用 LINK_LAUNCH 開啟本地檔案
    "from": text_rect,  # 超連結區域
    "file": link_path  # 指定本地 PDF 檔案
})

# 儲存新的 PDF
doc.save(output_path)
doc.close()
print("PDF 已儲存至", output_path)

input_pdf = "input.pdf" # 輸入的 PDF
output_pdf = "output.pdf" # 輸出的 PDF
link_path = r"E:\健行科技大學\智慧機械概論\期末報告\期末報告-4CLI.PDF" # 本地超連結檔案路徑

bom_text = "PDF"
if bom_text:
# 在 PDF 中加入 BOM 文字與超連結
add_text_with_link(input_pdf, output_pdf, bom_text, link_path)

[JorjMcKie]

You could try using this:

link_path = "E:/健行科技大學/智慧機械概論/期末報告/期末報告-4CLI.PDF"

Instead of this:

link_path = r"E:\健行科技大學\智慧機械概論\期末報告\期末報告-4CLI.PDF"

[bagwechien]

I tried, but same problem

[JorjMcKie]

But maybe the deeper problem are the Chinese letters ...

[JorjMcKie]

You can convert strings into PDF-compatible text via pymupdf.get_pdf_str():

pymupdf.get_pdf_str("E:/健行科技大學/智慧機械概論/期末報告/期末報告-4CLI.PDF")
'<feff0045003a002f5065884c79d1628059275b78002f667a61676a5f68b069828ad6002f671f672b5831544a002f671f672b5831544a002d00340043004c0049002e005000440046>'

This might solve it - although I'm not sure about the "<>" brackets ...

[JorjMcKie]

No, just tried it: doesn't work either.
I am afraid we need to modify the method itself.

[JorjMcKie]

Ah, sometimes AI chatbots are good for something:

If the destination name contains characters outside the ASCII range, they must be converted first in the following way:

text="E:/健行科技大學/智慧機械概論/期末報告/期末報告-4CLI.PDF"
link_text = "%".join([hex(b)[2:] for b in text.encode()])

The result looks like '45%3a%2f%e5%81%a5%e8%a1%8c%e7%a7%91%e6%8a%80%e5%a4%a7%e5%ad%b8%2f%e6%99%ba%e6%85%a7%e6%a9%9f%e6%a2%b0%e6%a6%82%e8%ab%96%2f%e6%9c%9f%e6%9c%ab%e5%a0%b1%e5%91%8a%2f%e6%9c%9f%e6%9c%ab%e5%a0%b1%e5%91%8a%2d%34%43%4c%49%2e%50%44%46'.
This string must be used as destination ... just confirmed that it works!

[JorjMcKie]

IAW, we must convert the destination string to UTF8 (always the case anyway), then make a bytes version of it and finally join the hexadecimal strings of each byte using "%" delimiters

[bagwechien]

I recorded an animation demonstrating two methods you provided for generating a PDF and clicking hyperlinks. First, I attempted to open "E:\健行科技大學\智慧機械概論\期末報告\期末報告-4CLI.PDF" to verify the file's existence. I then executed the two methods alternately, with comments explaining each step.

Unfortunately, neither method successfully opened the correct PDF.

Currently, I am using relative paths to open files and ignoring filenames containing Chinese characters. Although this is not ideal, it is the only solution that works for now.

Thank you for your assistance.