how to use insert_link

[emi0000]

Hi,

I'm trying to insert a link on page 0 of a pdf to page 1 of the same pdf. I tried also non-matplotlib pdfs or file with different pages. I always get an error message. My code looks as follows

import fitz as fz
import matplotlib.pyplot as plt

#https://app.readthedocs.org/projects/pymupdf/downloads/pdf/latest/

# make pdf file from matplotlib
plt.plot([-2,2],[1,1])
plt.ylim(-1,2)
plt.text(1,1.5,"mytext")
plt.text(1,0.5,"mytext")
plt.text(0.5,0.5,"text")
plt.savefig('mlf.pdf')  



filename = "mlf.pdf"
doc1 = fz.open(filename)
doc2 = fz.open(filename)
doc1.insert_pdf(doc2)
findtext="mytext"
page = doc1[0]
rec=page.search_for(findtext,quads=True)
page.add_highlight_annot(rec)
for rc in rec:
  page.insert_link({"kind": fz.LINK_GOTO, "page": 1, "from": rc,"to": fz.Point(0.,0.)})
doc1.save("both_linked.pdf")

the rror message looks as follows

File "pdft_test.py", line 28, in <module>
    page.insert_link({"kind": fz.LINK_GOTO, "page": 1, "from": rc,"to": fz.Point(0.,0.)})
  File "....\MyApps\Anaconda3\envs\env1\lib\site-packages\pymupdf\utils.py", line 2244, in insert_link
    annot = getLinkText(page, lnk)
  File "....\MyApps\Anaconda3\envs\env1\lib\site-packages\pymupdf\utils.py", line 2140, in getLinkText
    rect = _format_g(tuple(r * ictm))
  File "....\MyApps\Anaconda3\envs\env1\lib\site-packages\pymupdf\__init__.py", line 412, in _format_g
    ret += _format_g(v, fmt=fmt)
  File "....\MyApps\Anaconda3\envs\env1\lib\site-packages\pymupdf\__init__.py", line 416, in _format_g
    return mupdf.fz_format_double(fmt, value)
  File "....\MyApps\Anaconda3\envs\env1\lib\site-packages\pymupdf\mupdf.py", line 43478, in fz_format_double
    return _mupdf.fz_format_double(fmt, value)
TypeError: in method 'fz_format_double', argument 2 of type 'double'

Actually I wanted to make a link from "mytext" on page 0 to "text" on page 1.
Any ideas? thanks

[JorjMcKie]

You were almost there! This works:

import pymupdf as fz  # do not use the deprecated name fitz
import matplotlib.pyplot as plt

filename = "mlf.pdf"
plt.plot([-2, 2], [1, 1])
plt.ylim(-1, 2)
plt.text(1, 1.5, "mytext")
plt.text(1, 0.5, "mytext")
plt.text(0.5, 0.5, "text")
plt.savefig(filename)

doc1 = fz.open(filename)
doc2 = fz.open(filename)
doc1.insert_pdf(doc2)
findtext = "mytext"
page = doc1[0]
# you need to set quads=False (default) to get the rects of the text!
rec = page.search_for(findtext, quads=False)
page.add_highlight_annot(rec)
for rc in rec:
    page.insert_link(
        {"kind": fz.LINK_GOTO, "page": 1, "from": rc, "to": fz.Point(0.0, 0.0)}
    )
doc1.ez_save("both_linked.pdf")  # use best compression

A bit unrelated, but here is an alternative that only requires opening 1 document:

filename = "mlf.pdf"
plt.plot([-2, 2], [1, 1])
plt.ylim(-1, 2)
plt.text(1, 1.5, "mytext")
plt.text(1, 0.5, "mytext")
plt.text(0.5, 0.5, "text")
plt.savefig(filename)

doc = fz.open(filename)
findtext = "mytext"
doc.fullcopy_page(0)  # full copy of page 0
page = doc[0]
# you need to set quads=False (default) to get the rects of the text!
rec = page.search_for(findtext, quads=False)
page.add_highlight_annot(rec)
for rc in rec:
    page.insert_link(
        {"kind": fz.LINK_GOTO, "page": 1, "from": rc, "to": fz.Point(0.0, 0.0)}
    )
doc.ez_save("both_linked1.pdf")  # use best compression

[emi0000]

may I add a small question here, sorry

is it also possible to search_for words containing german umlaute like ä ö ü .....

[JorjMcKie]

Yes, that works. However, other than for ASCII characters, a search for "Österreich" will not find "österreich" if you see the point ...

[emi0000]

I see, but all I tried like "hätte" or "häßlich" was not found. Is there some optional coding I have to add somewhere?

[JorjMcKie]

No, it should work normally - see the following snippet:

But beware, other things may go wrong too. If the characters weren't stored in reading sequence for instance, nothing will be found at all.
And of course, the search algorithm doesn't know the heck about "ß = ss".
Another question is how the font has stored these umlauts: when you specify the needle text, Python will use UTF-8 encoding, and so will MuPDF ultimately, which does the search.
You can attach an example here and I can look at what the problem.

[emi0000]

Do you have any idea what going on (see example below).Is there a way to see what happens in _fitz.TextPage_search(self, needle, hit_max, quads)?

[emi0000]

I'm generating the pdfs from matplotlib. I tried to make a small example out of what I'm doing.
For me "häßlich1" is not found, the other 2 häßlich are found. Seems that the prop's are making the problems. I tried to change to Arial and removed weight, same result. But, removing äß in häßlich1 works.

import fitz as fz
import matplotlib.pyplot as plt
import numpy as np


# make pdf file from matplotlib
plt.plot([-2,2],[1,1])
plt.ylim(-1,2)
prop={'color': 'black', 'ha' : 'center', 'va': 'center', 'fontname': 'Corbel', 'fontsize' : 11, 'weight': 'bold', 'linespacing': 0.8} 
plt.annotate("häßlich1",(1,1.5),**prop)
plt.text(1,0.5,"häßlich2")
plt.annotate("häßlich3",(-0.5,0.5))
plt.savefig('mlf.pdf')  


# load different existing pdfs, here the same
filename = "mlf.pdf"
doc1 = fz.open(filename)
doc2 = fz.open(filename)
doc1.insert_pdf(doc2)

#text to find
findtext="häßlich1"
findtext="häßlich2"
findtext="häßlich3"

page = doc1[0]
rec=page.search_for(findtext,quads=False)
page.add_highlight_annot(rec)
for rc in rec:
  page.insert_link({"kind": fz.LINK_GOTO, "page": 1, "from": rc,"to": fz.Point(100.,100.)})
doc1.save("both_linked.pdf")

[emi0000]

thank you so much for your detailed explanation, I appreciate that.
Do you know the trick how finally firefox or UR find häßlich1 in the pdf attached?
mlf.pdf

[JorjMcKie]

No

[JorjMcKie]

Some PDF viewers simply do some more or some less guesswork when encountering mess like delivered by matplotlib.

[JorjMcKie]

Ich bin Deutscher übrigens. Natürlich ist uns klar, dass "häßlich" orthografisch falsch ist: es müsste "hässlich" geschrieben werden ... 😉

[emi0000]

Dann müssen wir nur noch die nächste Rechtschreibreform abwarten .... :-)