rm unpaper

Author: ericphanson

Project.toml

@@ -13,28 +13,26 @@ ProgressMeter = "92933f4c-e287-5a05-a399-4b506db050ca"
 Random = "9a3f8284-a2c9-5f02-9a11-845980a1fd5c"
 Scratch = "6c6a2e73-6563-6170-7368-637461726353"
 Tesseract_jll = "efd95c89-babc-5260-8753-618084eaf9d7"
-unpaper_jll = "d52248c9-e08a-51c2-9066-05d0bf3e6245"
 
 [compat]
 Aqua = "0.8"
 CSV = "0.10"
 DocOpt = "0.5.0"
 OutputCollectors = "0.1.1"
+Pkg = "1"
 Poppler_jll = "21.9"
 ProgressMeter = "1.5"
+Random = "1"
 Scratch = "1"
 Tesseract_jll = "4.1.100"
-julia = "1.10"
-unpaper_jll = "6.1.100"
 Test = "1"
-Random = "1"
-Pkg = "1"
+julia = "1.10"
+
+[apps.searchable-pdf]
 
 [extras]
 Aqua = "4c88cf16-eb10-579e-8560-4a9242c79595"
 Test = "8dfed614-e22c-5e08-85e1-65c5234f0b40"
 
 [targets]
 test = ["Aqua", "Test"]
-
-[apps.searchable-pdf]

src/SearchablePDFs.jl

@@ -11,7 +11,6 @@ using CSV
 using DocOpt
 
 using Poppler_jll
-using unpaper_jll
 using Tesseract_jll
 
 export ocr
@@ -117,14 +116,6 @@ function get_images(pdf, page_range::UnitRange{Int}, tmp, total_pages; exit_on_e
     return paths, (; binary="pdftoppm", logs...)
 end
 
-# Clean up an image with unpaper
-function unpaper(img; exit_on_error)
-    img_base, img_ext = splitext(img)
-    img_unpaper = img_base * "_unpaper" * img_ext
-    logs = run_and_collect_logs(`$(unpaper_jll.unpaper()) $img $img_unpaper`; exit_on_error)
-    return (; img_unpaper, logs=(; binary="unpaper", logs...))
-end
-
 #####
 ##### Step 2: Use tesseract to generate a one-page searchable PDF from an image
 #####
@@ -178,7 +169,7 @@ end
 #####
 
 """
-    ocr(pdf, output_path=string(splitext(pdf)[1], "_OCR", ".pdf"); apply_unpaper=false,
+    ocr(pdf, output_path=string(splitext(pdf)[1], "_OCR", ".pdf");
              ntasks=Sys.CPU_THREADS - 1, tesseract_nthreads=1, pages=num_pages(pdf),
              cleanup_after=true, cleanup_at_exit=true, tmp=get_scratch_dir(pdf),
              verbose=true)
@@ -189,14 +180,13 @@ Keyword arguments:
 
 * `ntasks`: how many parallel tasks to use for launching `tesseract` and `pdftoppm`.
 * `tesseract_nthreads`: how many threads to direct Tesseract to use
-* `apply_unpaper`: whether or not to apply `unpaper` to try to improve the image quality
 * `tmp`: a directory to store intermediate files. This directory is deleted at the end of the function if `cleanup_after` is set to `true`, and when the Julia session is ended if `cleanup_at_exit` is set to `true`.
 * `pages=nothing`: the number of pages of the PDF to process; the default of `nothing` indicates all pages in the PDF. It can help in debugging to set this to something small.
 * `verbose`: show a progress bar for each step of the process.
 
 Set `ENV["JULIA_DEBUG"] = SearchablePDFs` to see (many) debug messages.
 """
-function ocr(pdf, output_path=string(splitext(pdf)[1], "_OCR", ".pdf"); apply_unpaper=false,
+function ocr(pdf, output_path=string(splitext(pdf)[1], "_OCR", ".pdf");
              ntasks=Sys.CPU_THREADS - 1, tesseract_nthreads=1, pages=nothing,
              cleanup_after=true, cleanup_at_exit=true, tmp=get_scratch_dir(pdf),
              verbose=true, force=false, max_files_per_unite=100,
@@ -240,10 +230,6 @@ function ocr(pdf, output_path=string(splitext(pdf)[1], "_OCR", ".pdf"); apply_un
     ocr_prog = Progress(pages; desc="(2/3) OCRing: ", enabled=verbose)
     pdfs = asyncmap(enumerate(img_paths); ntasks) do (page, img)
         @debug "img" page img
-        if apply_unpaper
-            img, unpaper_logs = unpaper(img; exit_on_error)
-            put!(all_logs, (; page, unpaper_logs...))
-        end
         pdf, tesseract_logs = make_pdf(img; tesseract_nthreads, exit_on_error)
         put!(all_logs, (; page, tesseract_logs...))
         next!(ocr_prog)
@@ -272,19 +258,16 @@ end
 ##### CLI interface
 #####
 
-CAN_USE_UNPAPER::Bool = unpaper_jll.is_available() && Sys.ARCH != :aarch64
-
 doc::String = """Searchable PDFs (OCR).
 
 Usage:
-  searchable-pdf <input_pdf> [<output_path>] [--apply_unpaper] [--keep_intermediates] [--quiet] [--force] [--logfile=<logfile>] [--tmp=<tmp>] [-n=<ntasks>] [-t=<tesseract_nthreads>]
+  searchable-pdf <input_pdf> [<output_path>] [--keep_intermediates] [--quiet] [--force] [--logfile=<logfile>] [--tmp=<tmp>] [-n=<ntasks>] [-t=<tesseract_nthreads>]
   searchable-pdf -h | --help
   searchable-pdf --version
 
 Options:
   -h --help        Show this screen.
   --version        Show version.
-  --apply_unpaper  todo
   --keep_intermediates  xyz
   --quiet           todo
   --force           todo
@@ -310,7 +293,6 @@ function main(args=ARGS)
 
     tesseract_nthreads = parse(Int, parsed["--tesseract_nthreads"])
     result = _main(input_pdf, output_path;
-                   apply_unpaper=parsed["--apply_unpaper"],
                    ntasks,
                    tesseract_nthreads,
                    keep_intermediates=parsed["--keep_intermediates"],
@@ -334,19 +316,12 @@ end
 
 function _main(input_pdf::String,
                output_path::String=string(splitext(input_pdf)[1], "_OCR",
-                                          ".pdf"); apply_unpaper::Bool=false,
+                                          ".pdf");
                ntasks::Int=Sys.CPU_THREADS - 1, tesseract_nthreads::Int=1,
                keep_intermediates::Bool=false,
                tmp::String=get_scratch_dir(input_pdf), quiet::Bool=false,
                logfile::Union{Nothing,String}=nothing, force::Bool=false,
                exit_on_error=false)
-    if apply_unpaper && !CAN_USE_UNPAPER
-        if Sys.ARCH == :aarch64
-            argument_error("Cannot use `unpaper` on `aarch64` systems"; exit_on_error)
-        else
-            argument_error("`unpaper` is not available on this system"; exit_on_error)
-        end
-    end
     # some of these are redundant with checks inside `ocr`; that's because we want to do them before the "Starting to ocr" message,
     # and we want them to exit if they fail in a non-interactive context, instead of printing a stacktracee.
     isfile(input_pdf) ||
@@ -361,7 +336,7 @@ function _main(input_pdf::String,
     verbose = !quiet
     verbose &&
         println("Starting to ocr `$(input_pdf)`; result will be located at `$(output_path)`.")
-    result = ocr(input_pdf, output_path; apply_unpaper, ntasks, tesseract_nthreads,
+    result = ocr(input_pdf, output_path; ntasks, tesseract_nthreads,
                  cleanup_after=!keep_intermediates, cleanup_at_exit=!keep_intermediates,
                  tmp, verbose, force, exit_on_error)
     verbose && println("\nOutput is located at `$(output_path)`.")

test/runtests.jl

@@ -7,16 +7,12 @@ using Aqua
 TEST_PDF_PATH = joinpath(@__DIR__, "test.pdf")
 TEST_PDF_RASTERIZED_PATH = joinpath(@__DIR__, "test_rasterized.pdf")
 
-unpaper_settings = SearchablePDFs.CAN_USE_UNPAPER ? (true, false) : (false,)
-
 @testset "SearchablePDFs.jl" begin
     @test SearchablePDFs.num_pages(TEST_PDF_PATH; exit_on_error=false) == 3
 
-    @testset "verbose=$verbose apply_unpaper=$apply_unpaper f=$f opt=$opt" for verbose in
+    @testset "verbose=$verbose f=$f opt=$opt" for verbose in
                                                                                (false,
                                                                                 true),
-                                                                               apply_unpaper in
-                                                                               unpaper_settings,
                                                                                f in
                                                                                (_main,
                                                                                 ocr),
@@ -28,7 +24,7 @@ unpaper_settings = SearchablePDFs.CAN_USE_UNPAPER ? (true, false) : (false,)
                   keep_intermediates=opt) :
                  (; verbose=verbose, max_files_per_unite=opt ? 2 : 100)
 
-        result = f(TEST_PDF_RASTERIZED_PATH, joinpath(@__DIR__, "out.pdf"); apply_unpaper,
+        result = f(TEST_PDF_RASTERIZED_PATH, joinpath(@__DIR__, "out.pdf");
                    kwargs...)
         # make sure we delete the generated files eventually, even if the tests throw
         atexit(() -> rm(result.output_path; force=true))