Commit Graph
111 Commits
Author SHA1 Message Date
James R. Barlow c725bf79da flake8 delinting 2021-09-21 16:37:03 -07:00
James R. Barlow 79fd8d01a5 info: fix incorrect handling of inline images and other typing fixes 2021-09-15 00:26:15 -07:00
James R. Barlow 95d9e8d91a info: inconsistent types used in ContentsInfo.name_index
This broke PyPy but CPython is fine with it.
2021-08-28 00:18:14 -07:00
James R. Barlow 3764ee872a typing: refactor namedtuples in info 2021-08-27 00:51:11 -07:00
James R. Barlow 9b81e76ed4 typing: fix issues; avoid some magic literals 2021-08-26 23:48:27 -07:00
James R. Barlow 4eca0a165b pre-commit: pyupgrade modernizing 2021-08-26 18:04:38 -07:00
James R. Barlow 5112e9e857 Redo dpi calc to avoid 'math.ulp' 2021-04-16 00:54:41 -07:00
James R. Barlow d673126994 Fix ZeroDivisionError on files containing images drawn at scale 0
Fixes #761
2021-04-15 23:26:14 -07:00
James R. Barlow a4e1f8e1f3 Merge branch 'feature/lambda' 2021-04-01 16:36:22 -07:00
James R. Barlow d274d88929 Refactor to eliminate global state in _concurrent 2021-01-30 17:36:30 -08:00
James R. Barlow ef1e7a814e Delinting 2021-01-26 01:45:04 -08:00
James R. Barlow 6953f32465 pdfinfo: remove some messy concurrency handling
We can cut down on the use of global variables and save opening
an extra copy of the Pdf when threaded.
2021-01-24 23:46:33 -08:00
James R. Barlow 26b4d9bb4b Refactor concurrency so that it is pluggable
However, this may not be the best idea because it involves global
state that could be overridden by a parallel call to ocrmypdf.ocr.
2021-01-24 23:46:29 -08:00
James R. Barlow 0b3a526049 Partial fix crash on 'userunit' None (#700)
Our method of getting data from pdfminer would silently consume a StopIteration
if pdfminer returned no processed pages, leading to odd error message.

We improve an error from pdfminer properly, and returning a more
descriptive error of our own.

It would be possible for ocrmypdf to repair the file before sending it to
pdfminer, but this seems to be rare enough that we won't do that yet.
2021-01-01 01:11:32 -08:00
James R. Barlow 416df803d4 pdfinfo: stricter typing 2020-12-24 22:39:00 -08:00
James R. Barlow 037b96ca16 pdfinfo: refactor to eliminate RawPageInfo 2020-12-24 02:57:44 -08:00
James R. Barlow bb258fc99c pdfinfo: Refactor pageinfo dictionary into a class 2020-12-24 01:47:53 -08:00
James R. Barlow b8aa89e1ec Fix log message queue flooding on certain files
Fixes #692
2020-12-11 14:14:21 -08:00
James R. Barlow 3707af3b74 Change pdf.root to pdf.Root 2020-11-03 01:30:31 -08:00
James R. Barlow d1e0c81eda Ensure worker_pdf is closed after gathering info in a thread
This is hacky, uses global state, but it does improve the situation for now.
2020-10-22 00:38:24 -07:00
James R. Barlow aa0ec40102 Change license of all GPLv3 files to MPL-2.0
https://github.com/jbarlow83/OCRmyPDF/issues/600
2020-08-05 00:44:42 -07:00
James R. Barlow d6128e6937 Fix support for older versions of pdfminer.six (boxes_flow error) 2020-07-26 21:51:25 -07:00
James R. Barlow 642437e804 Merge branch 'master' of github.com:jbarlow83/OCRmyPDF 2020-07-22 00:34:33 -07:00
James R. Barlow a672422b0b Enable pikepdf mmap in other contexts 2020-07-22 00:20:07 -07:00
fcatusandGitHub d80d963cea pdfinfo: Replace list comp with gen expr'n 2020-07-20 02:21:58 -07:00
James R. Barlow dc42beb6a8 More typing improvements
Typing fixes bugs.
2020-06-30 15:02:30 -07:00
James R. Barlow 378f543619 TextPositionTracker: set boxes_flow=None
We don't care about the order of lines in our analysis, and this is an
expensive calculation in pdfminer.
2020-06-30 04:20:58 -07:00
James R. Barlow c9bd87254e A few minor typing issues 2020-06-22 02:31:53 -07:00
James R. Barlow 7caf1e85ff info: change "Scan" message 2020-06-10 12:11:37 -07:00
James R. Barlow f59a757e8b info: tidy handling of content streams 2020-06-10 12:09:24 -07:00
James R. Barlow 872bafad4b Reinstate quick test for text/no text
Partial revert of commit 991db17
2020-06-10 12:00:52 -07:00
James R. Barlow 8599400445 Only do page analysis on pages we will do OCR on 2020-06-10 11:33:27 -07:00
James R. Barlow b6eebadf05 Use pikepdf.open with block to manage PdfInfo 2020-06-10 11:32:46 -07:00
James R. Barlow fe156db41d Merge branch 'release/v10' into trialmerge 2020-06-09 15:12:56 -07:00
James R. Barlow d118132fa6 layout: look for text in XObjects too 2020-06-03 13:16:55 -07:00
James R. Barlow a2d3e0b53e Convert remaining imports to absolute 2020-05-12 02:12:08 -07:00
James R. Barlow 977665d2b6 Delint some tests 2020-05-08 03:49:33 -07:00
James R. Barlow 85cbf94a6e Convert many uses of str paths to Path 2020-05-06 02:53:47 -07:00
James R. Barlow c85278b31d Delinting 2020-05-03 00:53:29 -07:00
James R. Barlow 18c4aa10bf Adjust number of workers for concurrent page scanning 2020-04-26 04:21:15 -07:00
James R. Barlow 991db17fde Remove Ghostscript-based text extraction
While faster than Python based methods, we've outgrown the limited
amount of information Ghostscript provides with this feature, and it
repeats an analysis we have to do anyway to learn what images are
present.
2020-04-26 04:02:07 -07:00
James R. Barlow 27a3b80376 Use once-per-worker pikepdf init 2020-04-26 03:49:20 -07:00
James R. Barlow 8c381a0227 Replace task_initargs with use of partial() 2020-04-26 03:49:20 -07:00
James R. Barlow 86145a8c76 Some wrong with forking worker_pdf, just open it once per page for now 2020-04-26 03:49:20 -07:00
James R. Barlow 7513f5425c Fix some broken tests 2020-04-26 03:49:20 -07:00
James R. Barlow af3c3c6466 Further refactoring of concurrency concerns 2020-04-26 03:49:20 -07:00
James R. Barlow db3e75e33e Refactor multiprocessing pool 2020-04-26 03:49:13 -07:00
James R. Barlow ce49fc26dd Do pikepdf.open() once instead of per worker 2020-04-26 03:42:13 -07:00
James R. Barlow d0d0a98dca First cut at concurrent page scan
Improvement appears on 168 page file. Needs refactoring
2020-04-26 03:42:13 -07:00
James R. Barlow 94c52a6fa3 Refactor 'xyres' into Resolution 2020-04-24 04:12:05 -07:00