| | <<<Back 1 day (to 2018/03/06) | 20180307 |
| enometh | from tor's comment on https://bugs.ghostscript.com/show_bug.cgi?id=699079 it looks like the x11 regression may not be fixed? | 05:31.53 |
| | all the "new improvements" in text extraction just do not work they undo all decades of work done before | 05:45.25 |
| | but unlike gs, this isnt an easy revert | 05:53.07 |
| | 626ea2ea77 | 05:53.22 |
| tor8 | Robin_Watts: two bug fix commits on tor master that could do with a review | 11:36.10 |
| | no need to bother with the other ones just yet | 11:36.30 |
| Robin_Watts | ok... | 11:37.04 |
| | so the font/cidfont one, and the missing transformation? | 11:39.02 |
| | both lgtm. | 11:39.04 |
| tor8 | the scoping bug in png_write_band one too | 11:44.52 |
| enometh | tor, in pdfapp_oncopy, after int x0 = app->selr.x0... , after pdfapp_viewctm(&ctm, app), ctm has to be inverted and the points x0,x1.. have to be transformed back to the coordinates in which the text was extracted | 11:48.55 |
| Robin_Watts | tor8: lgtm too. | 11:49.26 |
| tor8 | enometh: yes, I've got a fix for that just cooked up and ready to commit. | 11:49.29 |
| enometh | ah ok | 11:49.34 |
| tor8 | enometh: and it's pushed :) | 11:49.47 |
| Robin_Watts | enometh: That's the "missing transformation" one :) | 11:49.49 |
| sebras | tor8: in 699094 we lose /Filter /JPXDecode in the image dictionary, but we do not appear to have decompressed the image. | 14:22.34 |
| Robin_Watts | sebras: That seems bad :) | 14:36.36 |
| sebras | Robin_Watts: yes, in expandstream() we appera to pdf_load_stream_truncated(), but I'm thinking that this loads the compressed stream. | 14:39.44 |
| | but I haven't debugged it yet. | 14:39.52 |
| | Robin_Watts: oh, so this is because JPX is special cased in the image loading code. | 15:32.50 |
| Robin_Watts | Ah. | 15:33.02 |
| sebras | tor8: I think I might have something that fixes 699094 | 16:41.07 |
| | tor8: there is something on sebras/master, the first one ought to be a no-brainer. | 17:14.50 |
| | tor8: the second one is not pretty, but I think it gets things done. | 17:15.03 |
| | tor8: I wonder if I have to take softmasks and decode arrays into account in the same way as being done in pdf_load_jpx(). I don't think that would be necessary because the softmask and decode array would remain in the imag edictionary. | 17:27.33 |
| n35xdxb0 | is there a config file for mupdf? i'd like to change some of the keybindings. is there a way of doing this? | 23:02.47 |
| | Forward 1 day (to 2018/03/08)>>> | |
Log of #mupdf at irc.freenode.net.