performance improvements #60
Conversation
With this speedup, when you have various pages with png images, the improvements is significant. As a example, I used an old dual core machine, with 104 png images, to generate a pdf, 1 image per page. The time consuming before was 1 minute and 28 seconds. After this patch, the time is 6 seconds!!! Yes, 6 seconds, 14,7x faster!
fpdf/fpdf.py
Outdated
| if PY3K: | ||
| if not isinstance(s, bytes): | ||
| s = s.encode('latin1') | ||
| s += b"\n" |
There was a problem hiding this comment.
According to the readme, this supports Python 2.5, which I suspect would choke compiling this syntax
There was a problem hiding this comment.
I had no idea about 2.5 here. I have app using this in py2.7 and py3.4. Do you have a suggestion?
There was a problem hiding this comment.
If I am right about s always being str at this point, change it to
s += "\n"
if(self.state == 2):
self.pages[self.page]["content"] += s
else:
if PY3K:
s = s.encode('latin1')
self.stream.write(s)Otherwise, maybe you could use the b() function from the py3k module.
There was a problem hiding this comment.
you are right!
|
The general idea sounds good to me. Writing into BytesIO or similar should be well optimized, and concatenating strings in Python performs badly in this sort of scenario. |
|
@reingart: This patch is good to be merged? |
|
Thank you for this performance work. Swapping in this version reduced the build time for a 112MB, 228 page PDF containing over 4500 images from 40 minutes to 30 seconds - an awesome improvement. |
|
This offers a huge performance benefit, any chance this will get merged? |
|
@alallier you may want to check out #171 and also check out this link to see that the fpdf2 fork uses a bytearray instead of a string (and is otherwise maintained): https://github.com/PyFPDF/fpdf2/blob/master/fpdf/fpdf.py#L178 |
|
Thanks for the link. I noticed shortly after that this project was abandoned which is unfortunate |
With this speedup, when you have various pages with png images, the
improvements is significant.
As a example, I used an old dual core machine, with 104 png images (16MB per image), to
generate a pdf, 1 image per page. The time consuming before was 1 minute
and 28 seconds. After this patch, the time is 6 seconds!!!
Yes, 6 seconds, 14,7x faster!