Package: gscan2pdf Version: 1.1.3
OCRopus is not longer actively distributing the recognition scripts that gscan2pdf was using. (the orcoscript directory; they are in the OCRopus's archive, but the scripts do not run cleanly). gscan2pdf should update Ocropus.pm to use the current OCRopus command line interfaces. Attached is a prototype for a new version of the Gscan2pdf::Ocropus.pm module. Good news: It works. At least, it works on my system. Bad news: There are a number of known deficiencies. Meta issue 1) I approximated the coding style of the rest of the application. There are presumably style differences I unintentionally introduced. Meta issue 2) Perl is my weakest language in terms coding. I'm sure there are some coding defects I don't know, or better ways of doing things. 1) I do not have OCRopus using Tesseract for recognition working. Yet. 2) Determining success or failure of a recognition step is fragile, and based on whether there is output to stderr (most cases) or if the output to stderr is of a known/desired form. 3) There is/was no way I saw in existing code to pass configuration from the application to the modules. This means there are a number of configurable values that are hard coded, when parameterizing them is the right thing to do. (number of processes OCRopus should start/parallelization, saving or destroying intermediate files, choosing OCRopus modules, using tesseract, or any other OCR, for character/word recognition, and others) 4) While the code will tell OCRopus to run the recognition steps in parallel, the module itself is single-threaded and does not dispatch recognition against multiple pages simultaneously. 5) In order to facilitate the 'keep intermediate files' functionality, the module uses a distinct tmp directory from the rest of the application. ---------------------------------------------------------------------- That's the news from the Mystic River, where all the alliums are strong, all the degu are good looking, and all the stuffed animals are above average. "May the ducks of your life quack ever harmoniously" - A. Yelton [email protected] [email protected] [email protected] & others
Ocropus.pm
Description: Binary data
