menekis/OCRmyPDF
Public Access
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
1,817 Commits 5 Branches 303 Tags
ff41fbf67347a9d7261f3a65825d655f7f1ee832
Commit Graph
294 Commits
Author SHA1 Message Date
James R. Barlow ff41fbf673 Add pdfminer based layout analysis 2018-10-25 12:42:35 -07:00
James R. Barlow 2435cd23ce Move pdfinfo into a package 2018-10-25 00:37:38 -07:00
James R. Barlow a063cff720 Rename/expose strip_invisible_text 2018-10-24 21:53:24 -07:00
James R. Barlow 0d396e1ac0 option check: Remove always-True condition 
Both renderers are now lossless reconstruction-capable. (Have
been since 7.0)
 2018-10-22 22:13:59 -07:00
James R. Barlow d11c428407 Redo OCR: disallow in cases that will damage the output PDF 2018-10-20 01:14:33 -07:00
James R. Barlow 6182b1f53e Merge branch 'feature/remove-vectors' into feature/redo-ocr 2018-10-20 01:13:24 -07:00
James R. Barlow 16af753206 Add functional "redo OCR" feature 
Needs argument validation and some other changes. Needs testing
with mixed-content PDFs.

Only really works for pure invisible text at the moment.
 2018-10-19 00:02:19 -07:00
James R. Barlow fa48205bb8 Add feature to remove vector graphics objects 2018-10-18 21:46:08 -07:00
James R. Barlow f7dbf94071 pipeline: if vector graphic objects exist, ensure the DPI is reasonable 2018-10-18 01:23:31 -07:00
James R. Barlow b18e66e2ca pdfinfo: learn to detect vector graphic objects 2018-10-18 01:21:51 -07:00
James R. Barlow 7a5504dfa5 pdfinfo: fix terminology (operands, command) -> (operands, operator) 2018-10-18 01:18:30 -07:00
James R. Barlow 68bac1b177 Fix compatibility with pikepdf 0.3.5 API change 2018-10-11 15:51:34 -07:00
James R. Barlow 1495b78330 Remove cruft to support leptonica < 1.72 in test suite 2018-10-11 01:37:32 -07:00
James R. Barlow c1e18bb825 optimize: Exclude soft masks (SMasks) from optimization 
Soft masks are only allowed to be of colorspace DeviceGray so we
shouldn't use pngquant on them. For now, avoid this exceptional
case by excluded soft masks from optimization.
 2018-10-05 01:23:26 -07:00
James R. Barlow 58282ea0fb optimize: more refactoring 
Now properly generalized/specialized where it should be
 2018-10-04 13:44:51 -07:00
James R. Barlow 891da7834c optimize: refactor image extraction 2018-10-04 12:34:22 -07:00
James R. Barlow 5c229d48d5 optimize: Reorganize so JBIG2 can be performed on images reduced to 1bpp 
Closes #297
 2018-10-04 11:53:11 -07:00
James R. Barlow 5b84549716 Change JBIG2 lossy mode to require --jbig2-lossy 2018-10-04 01:20:49 -07:00
James R. Barlow c74f2ee6e8 Refactor the detailed error messages 2018-10-04 00:10:59 -07:00
James R. Barlow b32dd9f9d3 Fix lossless JBIG2 when there are multiple JBIG2 images on a single page 2018-10-03 17:40:26 -07:00
James R. Barlow fb8b161f6c Fix suppression of tesseract config error messages 2018-10-03 17:39:50 -07:00
James R. Barlow 6f554c6ae8 tesseract: account for behavior changes when params are missing 
Tesseract 4.0-rc1 now accepts invalid parameters in config and
won't return an error anymore. We prefer to raise an error if this
occurs.

See: https://github.com/tesseract-ocr/tesseract/commit/741ea00d7059d8ff7c55797ffc525a461d7f3ced
 2018-10-03 15:11:34 -07:00
James R. Barlow 72156b5653 Degrade more gracefully when --optimize is set but JBIG2 is not present 2018-10-03 14:24:20 -07:00
James R. Barlow 9a8ec4b210 optimize: only enable lossy JBIG2 for -O3 2018-10-03 00:38:58 -07:00
James R. Barlow 75aad4cc79 optimize: Refactor convert_to_jbig2 2018-10-02 23:42:12 -07:00
James R. Barlow 4b27feca98 optimize: Disable JBIG2 lossy mode, use lossless instead 2018-10-01 12:28:54 -07:00
James R. Barlow 45522cd15f weave: clarify comment about garbage data in ToC 2018-09-27 13:48:35 -07:00
James R. Barlow 677d9a4e76 Remove some unhelpful lambdas 2018-09-27 13:48:12 -07:00
James R. Barlow efa7ea4fde Fix log.error where log is None 2018-09-19 23:01:27 -07:00
James R. Barlow 137a6e45f5 ghostscript: fix missing fspath for py3.5 2018-09-19 22:57:20 -07:00
James R. Barlow cfd4f8a850 Improve error handling for improvements to Ghostscript text extraction 2018-09-19 20:29:18 -07:00
James R. Barlow ef70e538f7 Improve error message on handling KeyboardInterrupt 
Closes #301
 2018-09-19 01:40:26 -07:00
James R. Barlow b7b912e56a Fix test suite and blank pages 2018-09-17 01:12:58 -07:00
James R. Barlow 4615cf2f1e First cut at improving text extraction speed 2018-09-16 23:34:18 -07:00
James R. Barlow 96ba75eabd Ghostscript: fix issues in strict ASCII implementation 2018-09-16 15:41:54 -07:00
James R. Barlow fdfe52c1ad main: add debug option to force threads 2018-09-15 00:01:45 -07:00
James R. Barlow 932b2e2a29 main: print Ghostscript version too 2018-09-14 23:58:06 -07:00
James R. Barlow 57e489c957 main: Cleanup; support overriding sys.args in run_pipeline 2018-09-14 23:57:35 -07:00
James R. Barlow 17a3fa671c ghostscript: API docs update 2018-09-14 23:51:52 -07:00
James R. Barlow 7392115507 Blacklist Ghostscript 9.24 due to regressions 
As per issue #291. Forced push to remove a copyrighted test file that was
accidentally included.
 2018-09-14 20:41:13 -07:00
James R. Barlow 686207ab7f Check for and reject Adobe LiveCycle Designer PDFs 
These are the ones that display a "Please wait..." message.

Closes #296
 2018-09-13 21:50:51 -07:00
James R. Barlow 517b385fe5 Work around loss of Unicode DOCINFO in Ghostscript 9.24+ 
Ghostscript no longer supports UTF-16-BE-hex strings as a way of
supplying Unicode data in pdfmark so we have lost this functionality too:
http://git.ghostscript.com/?p=ghostpdl.git;a=commit;h=e997c6836d243ab37fe3a5f0d57974af95eb5eac

For users this means setting --title, --author, etc. will not work if gs
9.24 is installed, but if the file has existing metadata it might work.

For now we enforce police-state-strict ASCII, until there's time to
implement proper metadata editing. Relevant tests set to xfail.
 2018-09-13 21:33:39 -07:00
James R. Barlow 795019b0c1 Work around invalid TOC entries 
Kodak Capture Desktop and probably other software creates a
/Outlines entry with /First being set to an invalid indirect reference to
an object that hasn't been created. This is legal in the PDF spec but
problematic for qpdf. The objgen will be (max valid object ID + 1, 0).
Because we create new objects in _weave, some TOC entries will end
up assigned to new objects we create. Typically /ProcSet.

We solve the issue by refactoring page traversal and then doing it
twice, once to resolve all references (eliminating the null
reference problem) and a second pass to make our changes.
 2018-09-11 14:44:16 -07:00
James R. Barlow 3127a73822 Ghostscript: no need to specify ProcessColorModel when ColorConversionStrategy 2018-09-11 11:56:05 -07:00
James R. Barlow 069ee6c91f ghostscript: fix for 9.24 having jpeg passthrough available 2018-09-10 23:09:51 -07:00
James R. Barlow a96710aa7b leptonica: update comments 2018-09-10 11:47:38 -07:00
James R. Barlow edcc58826a pdfinfo: remove some dead code 2018-09-10 11:47:00 -07:00
James R. Barlow 4d5c9b8cdf Fix error in optimize.py on PNGs at -O2 
Error was
TypeError: unsupported operand type(s) for -: 'tuple' and 'int'
 2018-08-20 15:45:34 -07:00
James R. Barlow 1e23ea5364 Remove pikepdf < 0.3 compatibility shims since > 0.3.1 is now required 2018-08-10 17:01:03 -07:00
James R. Barlow 8013fd50da Draw preview image at full resolution 
As reported in #281 and confirmed by the test file in #279, downsampling
the preview adversely affects quality of image rotation especially for
small font sizes and marginal scans.

Full size gets rotation accuracy. This makes rotation a little inefficient
since it rasterizes twice - to be addressed later.
 2018-08-03 13:32:10 -07:00