Merge branch 'feature/pypdf-page-merge' into develop
This commit is contained in:
James R. Barlow
+66
-8
@@ -11,6 +11,7 @@ import warnings
|
||||
import multiprocessing
|
||||
import atexit
|
||||
import textwrap
|
||||
import img2pdf
|
||||
|
||||
import PyPDF2 as pypdf
|
||||
from PIL import Image
|
||||
@@ -247,6 +248,10 @@ if options.clean and not options.clean_final \
|
||||
"Tesseract PDF renderer cannot render --clean pages without "
|
||||
"also performing --clean-final, so --clean-final is assumed.")
|
||||
|
||||
lossless_reconstruction = False
|
||||
if options.pdf_renderer == 'hocr':
|
||||
if not options.deskew and not options.clean_final and not options.force_ocr:
|
||||
lossless_reconstruction = True
|
||||
|
||||
# ----------
|
||||
# Logging
|
||||
@@ -567,24 +572,48 @@ def select_image_for_pdf(
|
||||
|
||||
@active_if(options.pdf_renderer == 'hocr')
|
||||
@collate(
|
||||
input=[select_image_for_pdf, ocr_tesseract_hocr],
|
||||
filter=regex(r".*/(\d{6})(?:\.image|\.hocr)"),
|
||||
output=os.path.join(work_folder, r'\1.rendered.pdf'),
|
||||
input=[select_image_for_pdf, split_pages],
|
||||
filter=regex(r".*/(\d{6})(?:\.image|\.ocr\.page\.pdf)"),
|
||||
output=os.path.join(work_folder, r'\1.image-layer.pdf'),
|
||||
extras=[_log, _pdfinfo, _pdfinfo_lock])
|
||||
def render_hocr_page(
|
||||
def select_image_layer(
|
||||
infiles,
|
||||
output_file,
|
||||
log,
|
||||
pdfinfo,
|
||||
pdfinfo_lock):
|
||||
hocr = next(ii for ii in infiles if ii.endswith('.hocr'))
|
||||
|
||||
page_pdf = next(ii for ii in infiles if ii.endswith('.page.pdf'))
|
||||
image = next(ii for ii in infiles if ii.endswith('.image'))
|
||||
|
||||
pageinfo = get_pageinfo(image, pdfinfo, pdfinfo_lock)
|
||||
if lossless_reconstruction:
|
||||
re_symlink(page_pdf, output_file)
|
||||
else:
|
||||
pageinfo = get_pageinfo(image, pdfinfo, pdfinfo_lock)
|
||||
dpi = round(max(pageinfo['xres'], pageinfo['yres'], options.oversample))
|
||||
pdf_bytes = img2pdf.convert([image], dpi=dpi)
|
||||
with open(output_file, 'wb') as pdf:
|
||||
pdf.write(pdf_bytes)
|
||||
|
||||
|
||||
@active_if(options.pdf_renderer == 'hocr')
|
||||
@transform(
|
||||
input=ocr_tesseract_hocr,
|
||||
filter=suffix('.hocr'),
|
||||
output='.hocr.pdf',
|
||||
extras=[_log, _pdfinfo, _pdfinfo_lock])
|
||||
def render_hocr_page(
|
||||
input_file,
|
||||
output_file,
|
||||
log,
|
||||
pdfinfo,
|
||||
pdfinfo_lock):
|
||||
hocr = input_file
|
||||
pageinfo = get_pageinfo(hocr, pdfinfo, pdfinfo_lock)
|
||||
dpi = round(max(pageinfo['xres'], pageinfo['yres'], options.oversample))
|
||||
|
||||
hocrtransform = HocrTransform(hocr, dpi)
|
||||
hocrtransform.to_pdf(output_file, imageFileName=image,
|
||||
hocrtransform.to_pdf(output_file, imageFileName=None,
|
||||
showBoundingboxes=False, invisibleText=True)
|
||||
|
||||
|
||||
@@ -612,6 +641,35 @@ def render_hocr_debug_page(
|
||||
showBoundingboxes=True, invisibleText=False)
|
||||
|
||||
|
||||
@active_if(options.pdf_renderer == 'hocr')
|
||||
@collate(
|
||||
input=[render_hocr_page, select_image_layer],
|
||||
filter=regex(r".*/(\d{6})(?:\.hocr\.pdf|\.image-layer\.pdf)"),
|
||||
output=os.path.join(work_folder, r'\1.rendered.pdf'),
|
||||
extras=[_log, _pdfinfo, _pdfinfo_lock])
|
||||
def add_text_layer(
|
||||
infiles,
|
||||
output_file,
|
||||
log,
|
||||
pdfinfo,
|
||||
pdfinfo_lock):
|
||||
text = next(ii for ii in infiles if ii.endswith('.hocr.pdf'))
|
||||
image = next(ii for ii in infiles if ii.endswith('.image-layer.pdf'))
|
||||
|
||||
pdf_output = pypdf.PdfFileWriter()
|
||||
|
||||
pdf_text = pypdf.PdfFileReader(open(text, "rb"))
|
||||
pdf_image = pypdf.PdfFileReader(open(image, "rb"))
|
||||
|
||||
page = pdf_text.getPage(0)
|
||||
page.mergePage(pdf_image.getPage(0))
|
||||
|
||||
pdf_output.addPage(page)
|
||||
|
||||
with open(output_file, "wb") as out:
|
||||
pdf_output.write(out)
|
||||
|
||||
|
||||
@active_if(options.pdf_renderer == 'tesseract')
|
||||
@collate(
|
||||
input=[preprocess_clean, split_pages],
|
||||
@@ -703,7 +761,7 @@ def skip_page(
|
||||
|
||||
|
||||
@merge(
|
||||
input=[render_hocr_page, render_hocr_debug_page, skip_page,
|
||||
input=[add_text_layer, render_hocr_debug_page, skip_page,
|
||||
tesseract_ocr_and_render_pdf, generate_postscript_stub],
|
||||
output=os.path.join(work_folder, 'merged.pdf'),
|
||||
extras=[_log, _pdfinfo, _pdfinfo_lock])
|
||||
|
||||
+198
-167
@@ -4,232 +4,263 @@
|
||||
<!-- Generated by graphviz version 2.38.0 (20140413.2041)
|
||||
-->
|
||||
<!-- Title: Pipeline: Pages: 1 -->
|
||||
<svg width="728pt" height="651pt"
|
||||
viewBox="0.00 0.00 728.00 650.53" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink">
|
||||
<g id="graph0" class="graph" transform="scale(1 1) rotate(0) translate(4 646.53)">
|
||||
<svg width="1132pt" height="708pt"
|
||||
viewBox="0.00 0.00 1132.00 708.08" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink">
|
||||
<g id="graph0" class="graph" transform="scale(1 1) rotate(0) translate(4 704.083)">
|
||||
<title>Pipeline:</title>
|
||||
<polygon fill="white" stroke="none" points="-4,4 -4,-646.53 724,-646.53 724,4 -4,4"/>
|
||||
<polygon fill="white" stroke="none" points="-4,4 -4,-704.083 1128,-704.083 1128,4 -4,4"/>
|
||||
<g id="clust1" class="cluster"><title>clustertasks</title>
|
||||
<polygon fill="none" stroke="black" points="8,-8 8,-634.53 712,-634.53 712,-8 8,-8"/>
|
||||
<text text-anchor="middle" x="360" y="-606.53" font-family="Times,serif" font-size="30.00" fill="#ff3232">Pipeline:</text>
|
||||
<polygon fill="none" stroke="black" points="8,-8 8,-692.083 1116,-692.083 1116,-8 8,-8"/>
|
||||
<text text-anchor="middle" x="562" y="-664.083" font-family="Times,serif" font-size="30.00" fill="#ff3232">Pipeline:</text>
|
||||
</g>
|
||||
<!-- t0 -->
|
||||
<g id="node1" class="node"><title>t0</title>
|
||||
<polygon fill="#efa03b" stroke="#006000" points="481.791,-588.53 386.209,-588.53 382.209,-584.53 382.209,-552.53 477.791,-552.53 481.791,-556.53 481.791,-588.53"/>
|
||||
<polyline fill="none" stroke="#006000" points="477.791,-584.53 382.209,-584.53 "/>
|
||||
<polyline fill="none" stroke="#006000" points="477.791,-584.53 477.791,-552.53 "/>
|
||||
<polyline fill="none" stroke="#006000" points="477.791,-584.53 481.791,-588.53 "/>
|
||||
<text text-anchor="middle" x="432" y="-564.53" font-family="Times,serif" font-size="20.00" fill="#006000">repair_pdf</text>
|
||||
<polygon fill="#efa03b" stroke="black" points="936.535,-646.083 713.465,-646.083 709.465,-642.083 709.465,-610.083 932.535,-610.083 936.535,-614.083 936.535,-646.083"/>
|
||||
<polyline fill="none" stroke="black" points="932.535,-642.083 709.465,-642.083 "/>
|
||||
<polyline fill="none" stroke="black" points="932.535,-642.083 932.535,-610.083 "/>
|
||||
<polyline fill="none" stroke="black" points="932.535,-642.083 936.535,-646.083 "/>
|
||||
<text text-anchor="middle" x="823" y="-622.083" font-family="Times,serif" font-size="20.00">repair_pdf</text>
|
||||
</g>
|
||||
<!-- t1 -->
|
||||
<g id="node2" class="node"><title>t1</title>
|
||||
<polygon fill="#efa03b" stroke="black" points="466.782,-509.564 374,-526.497 281.218,-509.564 281.304,-482.165 466.696,-482.165 466.782,-509.564"/>
|
||||
<polygon fill="none" stroke="black" points="470.799,-512.902 374,-530.569 277.201,-512.902 277.311,-478.159 470.689,-478.159 470.799,-512.902"/>
|
||||
<text text-anchor="middle" x="374" y="-495.991" font-family="Times,serif" font-size="20.00">split_pages</text>
|
||||
<polygon fill="#efa03b" stroke="black" points="914.112,-567.155 710,-584.057 505.888,-567.155 506.078,-539.806 913.922,-539.806 914.112,-567.155"/>
|
||||
<polygon fill="none" stroke="black" points="918.134,-570.834 710,-588.069 501.866,-570.834 502.11,-535.808 917.89,-535.808 918.134,-570.834"/>
|
||||
<text text-anchor="middle" x="710" y="-553.596" font-family="Times,serif" font-size="20.00">split_pages</text>
|
||||
</g>
|
||||
<!-- t0->t1 -->
|
||||
<g id="edge1" class="edge"><title>t0->t1</title>
|
||||
<path fill="none" stroke="gray" d="M417.064,-552.394C412.296,-546.925 406.86,-540.689 401.493,-534.532"/>
|
||||
<polygon fill="gray" stroke="gray" points="403.996,-532.077 394.787,-526.838 398.719,-536.676 403.996,-532.077"/>
|
||||
<path fill="none" stroke="#0044a0" d="M793.9,-609.961C783.582,-603.89 771.656,-596.873 760.092,-590.069"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="761.747,-586.982 751.353,-584.927 758.197,-593.015 761.747,-586.982"/>
|
||||
</g>
|
||||
<!-- t10 -->
|
||||
<g id="node12" class="node"><title>t10</title>
|
||||
<polygon fill="#efa03b" stroke="#006000" points="704.338,-451.452 493.662,-451.452 489.662,-447.452 489.662,-415.452 700.338,-415.452 704.338,-419.452 704.338,-451.452"/>
|
||||
<polyline fill="none" stroke="#006000" points="700.338,-447.452 489.662,-447.452 "/>
|
||||
<polyline fill="none" stroke="#006000" points="700.338,-447.452 700.338,-415.452 "/>
|
||||
<polyline fill="none" stroke="#006000" points="700.338,-447.452 704.338,-451.452 "/>
|
||||
<text text-anchor="middle" x="597" y="-427.452" font-family="Times,serif" font-size="20.00" fill="#006000">generate_postscript_stub</text>
|
||||
<!-- t12 -->
|
||||
<g id="node14" class="node"><title>t12</title>
|
||||
<polygon fill="#efa03b" stroke="black" points="1108.08,-509.109 769.918,-509.109 765.918,-505.109 765.918,-473.109 1104.08,-473.109 1108.08,-477.109 1108.08,-509.109"/>
|
||||
<polyline fill="none" stroke="black" points="1104.08,-505.109 765.918,-505.109 "/>
|
||||
<polyline fill="none" stroke="black" points="1104.08,-505.109 1104.08,-473.109 "/>
|
||||
<polyline fill="none" stroke="black" points="1104.08,-505.109 1108.08,-509.109 "/>
|
||||
<text text-anchor="middle" x="937" y="-485.109" font-family="Times,serif" font-size="20.00">generate_postscript_stub</text>
|
||||
</g>
|
||||
<!-- t0->t10 -->
|
||||
<g id="edge16" class="edge"><title>t0->t10</title>
|
||||
<path fill="none" stroke="gray" d="M453.048,-552.468C461.441,-545.654 471.185,-537.73 480,-530.53 510.227,-505.84 544.753,-477.467 568.421,-457.99"/>
|
||||
<polygon fill="gray" stroke="gray" points="570.818,-460.55 576.315,-451.493 566.369,-455.146 570.818,-460.55"/>
|
||||
<!-- t0->t12 -->
|
||||
<g id="edge19" class="edge"><title>t0->t12</title>
|
||||
<path fill="none" stroke="#0044a0" d="M899.32,-610.004C909.979,-604.592 919.748,-597.466 927,-588.083 941.916,-568.78 943.099,-540.385 941.345,-519.486"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="944.8,-518.881 940.218,-509.328 937.843,-519.653 944.8,-518.881"/>
|
||||
</g>
|
||||
<!-- t2 -->
|
||||
<g id="node3" class="node"><title>t2</title>
|
||||
<polygon fill="#efa03b" stroke="black" points="451.555,-451.452 228.445,-451.452 224.445,-447.452 224.445,-415.452 447.555,-415.452 451.555,-419.452 451.555,-451.452"/>
|
||||
<polyline fill="none" stroke="black" points="447.555,-447.452 224.445,-447.452 "/>
|
||||
<polyline fill="none" stroke="black" points="447.555,-447.452 447.555,-415.452 "/>
|
||||
<polyline fill="none" stroke="black" points="447.555,-447.452 451.555,-451.452 "/>
|
||||
<text text-anchor="middle" x="338" y="-427.452" font-family="Times,serif" font-size="20.00">rasterize_with_ghostscript</text>
|
||||
<polygon fill="#efa03b" stroke="black" points="592.299,-509.109 241.701,-509.109 237.701,-505.109 237.701,-473.109 588.299,-473.109 592.299,-477.109 592.299,-509.109"/>
|
||||
<polyline fill="none" stroke="black" points="588.299,-505.109 237.701,-505.109 "/>
|
||||
<polyline fill="none" stroke="black" points="588.299,-505.109 588.299,-473.109 "/>
|
||||
<polyline fill="none" stroke="black" points="588.299,-505.109 592.299,-509.109 "/>
|
||||
<text text-anchor="middle" x="415" y="-485.109" font-family="Times,serif" font-size="20.00">rasterize_with_ghostscript</text>
|
||||
</g>
|
||||
<!-- t1->t2 -->
|
||||
<g id="edge2" class="edge"><title>t1->t2</title>
|
||||
<path fill="none" stroke="#0044a0" d="M361.611,-478.092C358.514,-472.369 355.171,-466.19 352.003,-460.333"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="355.064,-458.636 347.227,-451.506 348.907,-461.967 355.064,-458.636"/>
|
||||
<path fill="none" stroke="#0044a0" d="M608.89,-535.808C573.672,-527.87 534.524,-519.048 500.679,-511.42"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="501.345,-507.982 490.82,-509.198 499.806,-514.811 501.345,-507.982"/>
|
||||
</g>
|
||||
<!-- t7 -->
|
||||
<g id="node9" class="node"><title>t7</title>
|
||||
<polygon fill="#efa03b" stroke="black" points="314.109,-277.109 19.8906,-277.109 15.8906,-273.109 15.8906,-241.109 310.109,-241.109 314.109,-245.109 314.109,-277.109"/>
|
||||
<polyline fill="none" stroke="black" points="310.109,-273.109 15.8906,-273.109 "/>
|
||||
<polyline fill="none" stroke="black" points="310.109,-273.109 310.109,-241.109 "/>
|
||||
<polyline fill="none" stroke="black" points="310.109,-273.109 314.109,-277.109 "/>
|
||||
<text text-anchor="middle" x="165" y="-253.109" font-family="Times,serif" font-size="20.00">select_image_layer</text>
|
||||
</g>
|
||||
<!-- t1->t7 -->
|
||||
<g id="edge11" class="edge"><title>t1->t7</title>
|
||||
<path fill="none" stroke="#0044a0" d="M501.985,-548.028C416.651,-540.897 317.351,-528.985 229,-509.109 131.492,-487.174 17,-534.054 17,-434.109 17,-434.109 17,-434.109 17,-374.109 17,-340.481 4.97908,-324.525 27,-299.109 32.8004,-292.415 39.6539,-286.828 47.1559,-282.17"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="49.1993,-285.038 56.2361,-277.118 45.7959,-278.921 49.1993,-285.038"/>
|
||||
</g>
|
||||
<!-- t13 -->
|
||||
<g id="node12" class="node"><title>t13</title>
|
||||
<polygon fill="#efa03b" stroke="black" points="1029.34,-451.109 808.662,-451.109 804.662,-447.109 804.662,-415.109 1025.34,-415.109 1029.34,-419.109 1029.34,-451.109"/>
|
||||
<polyline fill="none" stroke="black" points="1025.34,-447.109 804.662,-447.109 "/>
|
||||
<polyline fill="none" stroke="black" points="1025.34,-447.109 1025.34,-415.109 "/>
|
||||
<polyline fill="none" stroke="black" points="1025.34,-447.109 1029.34,-451.109 "/>
|
||||
<text text-anchor="middle" x="917" y="-427.109" font-family="Times,serif" font-size="20.00">skip_page</text>
|
||||
</g>
|
||||
<!-- t1->t13 -->
|
||||
<g id="edge16" class="edge"><title>t1->t13</title>
|
||||
<path fill="none" stroke="#0044a0" d="M716.849,-535.484C723.825,-515.965 736.562,-488.763 757,-473.109 768.342,-464.422 781.352,-457.641 794.947,-452.352"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="796.223,-455.612 804.44,-448.923 793.845,-449.029 796.223,-455.612"/>
|
||||
</g>
|
||||
<!-- t11 -->
|
||||
<g id="node10" class="node"><title>t11</title>
|
||||
<polygon fill="#efa03b" stroke="black" points="644.594,-393.452 551.406,-393.452 547.406,-389.452 547.406,-357.452 640.594,-357.452 644.594,-361.452 644.594,-393.452"/>
|
||||
<polyline fill="none" stroke="black" points="640.594,-389.452 547.406,-389.452 "/>
|
||||
<polyline fill="none" stroke="black" points="640.594,-389.452 640.594,-357.452 "/>
|
||||
<polyline fill="none" stroke="black" points="640.594,-389.452 644.594,-393.452 "/>
|
||||
<text text-anchor="middle" x="596" y="-369.452" font-family="Times,serif" font-size="20.00">skip_page</text>
|
||||
<g id="node13" class="node"><title>t11</title>
|
||||
<polygon fill="#efa03b" stroke="black" points="1068.24,-335.109 687.76,-335.109 683.76,-331.109 683.76,-299.109 1064.24,-299.109 1068.24,-303.109 1068.24,-335.109"/>
|
||||
<polyline fill="none" stroke="black" points="1064.24,-331.109 683.76,-331.109 "/>
|
||||
<polyline fill="none" stroke="black" points="1064.24,-331.109 1064.24,-299.109 "/>
|
||||
<polyline fill="none" stroke="black" points="1064.24,-331.109 1068.24,-335.109 "/>
|
||||
<text text-anchor="middle" x="876" y="-311.109" font-family="Times,serif" font-size="20.00">tesseract_ocr_and_render_pdf</text>
|
||||
</g>
|
||||
<!-- t1->t11 -->
|
||||
<g id="edge13" class="edge"><title>t1->t11</title>
|
||||
<path fill="none" stroke="#0044a0" d="M425.139,-478.007C437.834,-470.731 450.731,-461.83 461,-451.452 473.874,-438.442 466.946,-427.178 481,-415.452 490.212,-407.766 513.965,-399.205 537.508,-392.067"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="538.592,-395.396 547.189,-389.203 536.607,-388.683 538.592,-395.396"/>
|
||||
</g>
|
||||
<!-- t9 -->
|
||||
<g id="node11" class="node"><title>t9</title>
|
||||
<polygon fill="#efa03b" stroke="black" points="272.496,-277.452 19.5039,-277.452 15.5039,-273.452 15.5039,-241.452 268.496,-241.452 272.496,-245.452 272.496,-277.452"/>
|
||||
<polyline fill="none" stroke="black" points="268.496,-273.452 15.5039,-273.452 "/>
|
||||
<polyline fill="none" stroke="black" points="268.496,-273.452 268.496,-241.452 "/>
|
||||
<polyline fill="none" stroke="black" points="268.496,-273.452 272.496,-277.452 "/>
|
||||
<text text-anchor="middle" x="144" y="-253.452" font-family="Times,serif" font-size="20.00">tesseract_ocr_and_render_pdf</text>
|
||||
</g>
|
||||
<!-- t1->t9 -->
|
||||
<g id="edge15" class="edge"><title>t1->t9</title>
|
||||
<path fill="none" stroke="#0044a0" d="M277.341,-486.044C254.708,-478.702 232.21,-467.759 215,-451.452 168.207,-407.114 152.056,-329.127 146.633,-287.855"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="150.084,-287.235 145.422,-277.721 143.133,-288.065 150.084,-287.235"/>
|
||||
<g id="edge18" class="edge"><title>t1->t11</title>
|
||||
<path fill="none" stroke="#0044a0" d="M710.717,-535.657C712.098,-517.744 715.879,-492.693 726,-473.109 754.651,-417.673 809.56,-368.731 844.403,-341.318"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="846.607,-344.038 852.372,-335.148 842.322,-338.503 846.607,-344.038"/>
|
||||
</g>
|
||||
<!-- t3 -->
|
||||
<g id="node4" class="node"><title>t3</title>
|
||||
<polygon fill="#efa03b" stroke="black" points="458.999,-393.452 291.001,-393.452 287.001,-389.452 287.001,-357.452 454.999,-357.452 458.999,-361.452 458.999,-393.452"/>
|
||||
<polyline fill="none" stroke="black" points="454.999,-389.452 287.001,-389.452 "/>
|
||||
<polyline fill="none" stroke="black" points="454.999,-389.452 454.999,-357.452 "/>
|
||||
<polyline fill="none" stroke="black" points="454.999,-389.452 458.999,-393.452 "/>
|
||||
<text text-anchor="middle" x="373" y="-369.452" font-family="Times,serif" font-size="20.00">preprocess_deskew</text>
|
||||
<polygon fill="#efa03b" stroke="black" points="564.742,-451.109 269.258,-451.109 265.258,-447.109 265.258,-415.109 560.742,-415.109 564.742,-419.109 564.742,-451.109"/>
|
||||
<polyline fill="none" stroke="black" points="560.742,-447.109 265.258,-447.109 "/>
|
||||
<polyline fill="none" stroke="black" points="560.742,-447.109 560.742,-415.109 "/>
|
||||
<polyline fill="none" stroke="black" points="560.742,-447.109 564.742,-451.109 "/>
|
||||
<text text-anchor="middle" x="415" y="-427.109" font-family="Times,serif" font-size="20.00">preprocess_deskew</text>
|
||||
</g>
|
||||
<!-- t2->t3 -->
|
||||
<g id="edge3" class="edge"><title>t2->t3</title>
|
||||
<path fill="none" stroke="#0044a0" d="M348.691,-415.346C351.337,-411.112 354.23,-406.485 357.066,-401.946"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="360.042,-403.788 362.374,-393.453 354.106,-400.078 360.042,-403.788"/>
|
||||
<path fill="none" stroke="#0044a0" d="M415,-473.003C415,-469.312 415,-465.322 415,-461.352"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="418.5,-461.111 415,-451.111 411.5,-461.111 418.5,-461.111"/>
|
||||
</g>
|
||||
<!-- t6 -->
|
||||
<g id="node6" class="node"><title>t6</title>
|
||||
<polygon fill="#efa03b" stroke="black" points="664.375,-277.452 477.625,-277.452 473.625,-273.452 473.625,-241.452 660.375,-241.452 664.375,-245.452 664.375,-277.452"/>
|
||||
<polyline fill="none" stroke="black" points="660.375,-273.452 473.625,-273.452 "/>
|
||||
<polyline fill="none" stroke="black" points="660.375,-273.452 660.375,-241.452 "/>
|
||||
<polyline fill="none" stroke="black" points="660.375,-273.452 664.375,-277.452 "/>
|
||||
<text text-anchor="middle" x="569" y="-253.452" font-family="Times,serif" font-size="20.00">select_image_for_pdf</text>
|
||||
<g id="node8" class="node"><title>t6</title>
|
||||
<polygon fill="#efa03b" stroke="black" points="354.119,-335.109 39.8808,-335.109 35.8808,-331.109 35.8808,-299.109 350.119,-299.109 354.119,-303.109 354.119,-335.109"/>
|
||||
<polyline fill="none" stroke="black" points="350.119,-331.109 35.8808,-331.109 "/>
|
||||
<polyline fill="none" stroke="black" points="350.119,-331.109 350.119,-299.109 "/>
|
||||
<polyline fill="none" stroke="black" points="350.119,-331.109 354.119,-335.109 "/>
|
||||
<text text-anchor="middle" x="195" y="-311.109" font-family="Times,serif" font-size="20.00">select_image_for_pdf</text>
|
||||
</g>
|
||||
<!-- t2->t6 -->
|
||||
<g id="edge7" class="edge"><title>t2->t6</title>
|
||||
<path fill="none" stroke="#0044a0" d="M423.032,-415.374C438.812,-409.925 454.543,-402.785 468,-393.452 508.201,-365.571 539.287,-316.662 555.808,-286.558"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="559.024,-287.965 560.653,-277.496 552.851,-284.664 559.024,-287.965"/>
|
||||
<g id="edge9" class="edge"><title>t2->t6</title>
|
||||
<path fill="none" stroke="#0044a0" d="M294.315,-473.06C280.461,-467.585 267.283,-460.434 256,-451.109 223.245,-424.039 207.251,-375.559 200.099,-345.207"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="203.486,-344.31 197.925,-335.292 196.648,-345.809 203.486,-344.31"/>
|
||||
</g>
|
||||
<!-- t4 -->
|
||||
<g id="node5" class="node"><title>t4</title>
|
||||
<polygon fill="#efa03b" stroke="black" points="449.705,-335.452 300.295,-335.452 296.295,-331.452 296.295,-299.452 445.705,-299.452 449.705,-303.452 449.705,-335.452"/>
|
||||
<polyline fill="none" stroke="black" points="445.705,-331.452 296.295,-331.452 "/>
|
||||
<polyline fill="none" stroke="black" points="445.705,-331.452 445.705,-299.452 "/>
|
||||
<polyline fill="none" stroke="black" points="445.705,-331.452 449.705,-335.452 "/>
|
||||
<text text-anchor="middle" x="373" y="-311.452" font-family="Times,serif" font-size="20.00">preprocess_clean</text>
|
||||
<polygon fill="#efa03b" stroke="black" points="587.95,-393.109 310.05,-393.109 306.05,-389.109 306.05,-357.109 583.95,-357.109 587.95,-361.109 587.95,-393.109"/>
|
||||
<polyline fill="none" stroke="black" points="583.95,-389.109 306.05,-389.109 "/>
|
||||
<polyline fill="none" stroke="black" points="583.95,-389.109 583.95,-357.109 "/>
|
||||
<polyline fill="none" stroke="black" points="583.95,-389.109 587.95,-393.109 "/>
|
||||
<text text-anchor="middle" x="447" y="-369.109" font-family="Times,serif" font-size="20.00">preprocess_clean</text>
|
||||
</g>
|
||||
<!-- t3->t4 -->
|
||||
<g id="edge4" class="edge"><title>t3->t4</title>
|
||||
<path fill="none" stroke="#0044a0" d="M373,-357.346C373,-353.655 373,-349.665 373,-345.695"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="376.5,-345.453 373,-335.453 369.5,-345.453 376.5,-345.453"/>
|
||||
<path fill="none" stroke="#0044a0" d="M424.775,-415.003C427.132,-410.878 429.703,-406.379 432.232,-401.952"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="435.362,-403.53 437.285,-393.111 429.285,-400.057 435.362,-403.53"/>
|
||||
</g>
|
||||
<!-- t3->t6 -->
|
||||
<g id="edge6" class="edge"><title>t3->t6</title>
|
||||
<path fill="none" stroke="#0044a0" d="M414.399,-357.412C428.785,-351.039 444.852,-343.405 459,-335.452 466.909,-331.007 506.49,-303.805 535.921,-283.431"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="538.073,-286.198 544.299,-277.625 534.086,-280.444 538.073,-286.198"/>
|
||||
</g>
|
||||
<!-- t4->t6 -->
|
||||
<g id="edge5" class="edge"><title>t4->t6</title>
|
||||
<path fill="none" stroke="#0044a0" d="M432.605,-299.422C453.701,-293.395 477.617,-286.561 499.465,-280.319"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="500.687,-283.61 509.341,-277.497 498.764,-276.879 500.687,-283.61"/>
|
||||
<g id="edge8" class="edge"><title>t3->t6</title>
|
||||
<path fill="none" stroke="#0044a0" d="M351.041,-415.02C333.086,-409.151 313.871,-401.82 297,-393.109 269.828,-379.08 242.065,-358.149 222.377,-341.96"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="224.378,-339.071 214.46,-335.347 219.891,-344.444 224.378,-339.071"/>
|
||||
</g>
|
||||
<!-- t5 -->
|
||||
<g id="node7" class="node"><title>t5</title>
|
||||
<polygon fill="#efa03b" stroke="black" points="455.922,-277.452 294.078,-277.452 290.078,-273.452 290.078,-241.452 451.922,-241.452 455.922,-245.452 455.922,-277.452"/>
|
||||
<polyline fill="none" stroke="black" points="451.922,-273.452 290.078,-273.452 "/>
|
||||
<polyline fill="none" stroke="black" points="451.922,-273.452 451.922,-241.452 "/>
|
||||
<polyline fill="none" stroke="black" points="451.922,-273.452 455.922,-277.452 "/>
|
||||
<text text-anchor="middle" x="373" y="-253.452" font-family="Times,serif" font-size="20.00">ocr_tesseract_hocr</text>
|
||||
<g id="node6" class="node"><title>t5</title>
|
||||
<polygon fill="#efa03b" stroke="black" points="665.666,-335.109 376.334,-335.109 372.334,-331.109 372.334,-299.109 661.666,-299.109 665.666,-303.109 665.666,-335.109"/>
|
||||
<polyline fill="none" stroke="black" points="661.666,-331.109 372.334,-331.109 "/>
|
||||
<polyline fill="none" stroke="black" points="661.666,-331.109 661.666,-299.109 "/>
|
||||
<polyline fill="none" stroke="black" points="661.666,-331.109 665.666,-335.109 "/>
|
||||
<text text-anchor="middle" x="519" y="-311.109" font-family="Times,serif" font-size="20.00">ocr_tesseract_hocr</text>
|
||||
</g>
|
||||
<!-- t4->t5 -->
|
||||
<g id="edge8" class="edge"><title>t4->t5</title>
|
||||
<path fill="none" stroke="#0044a0" d="M373,-299.346C373,-295.655 373,-291.665 373,-287.695"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="376.5,-287.453 373,-277.453 369.5,-287.453 376.5,-287.453"/>
|
||||
<g id="edge5" class="edge"><title>t4->t5</title>
|
||||
<path fill="none" stroke="#0044a0" d="M468.994,-357.003C475.274,-352.118 482.229,-346.709 488.905,-341.516"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="491.396,-344.013 497.141,-335.111 487.099,-338.487 491.396,-344.013"/>
|
||||
</g>
|
||||
<!-- t4->t9 -->
|
||||
<g id="edge14" class="edge"><title>t4->t9</title>
|
||||
<path fill="none" stroke="#0044a0" d="M303.36,-299.422C278.157,-293.259 249.509,-286.253 223.521,-279.898"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="224.249,-276.473 213.704,-277.497 222.586,-283.273 224.249,-276.473"/>
|
||||
<!-- t4->t6 -->
|
||||
<g id="edge7" class="edge"><title>t4->t6</title>
|
||||
<path fill="none" stroke="#0044a0" d="M370.365,-357.079C342.386,-350.862 310.55,-343.787 281.752,-337.388"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="282.225,-333.907 271.704,-335.155 280.707,-340.741 282.225,-333.907"/>
|
||||
</g>
|
||||
<!-- t7 -->
|
||||
<g id="node8" class="node"><title>t7</title>
|
||||
<polygon fill="#efa03b" stroke="black" points="426.366,-219.452 269.634,-219.452 265.634,-215.452 265.634,-183.452 422.366,-183.452 426.366,-187.452 426.366,-219.452"/>
|
||||
<polyline fill="none" stroke="black" points="422.366,-215.452 265.634,-215.452 "/>
|
||||
<polyline fill="none" stroke="black" points="422.366,-215.452 422.366,-183.452 "/>
|
||||
<polyline fill="none" stroke="black" points="422.366,-215.452 426.366,-219.452 "/>
|
||||
<text text-anchor="middle" x="346" y="-195.452" font-family="Times,serif" font-size="20.00">render_hocr_page</text>
|
||||
</g>
|
||||
<!-- t6->t7 -->
|
||||
<g id="edge9" class="edge"><title>t6->t7</title>
|
||||
<path fill="none" stroke="#0044a0" d="M501.184,-241.422C476.75,-235.286 448.99,-228.315 423.772,-221.982"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="424.429,-218.538 413.877,-219.497 422.724,-225.328 424.429,-218.538"/>
|
||||
<!-- t4->t11 -->
|
||||
<g id="edge17" class="edge"><title>t4->t11</title>
|
||||
<path fill="none" stroke="#0044a0" d="M577.461,-357.079C627.38,-350.563 684.512,-343.106 735.342,-336.47"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="735.957,-339.92 745.42,-335.155 735.051,-332.979 735.957,-339.92"/>
|
||||
</g>
|
||||
<!-- t8 -->
|
||||
<g id="node9" class="node"><title>t8</title>
|
||||
<polygon fill="#efa03b" stroke="black" points="663.742,-219.452 448.258,-219.452 444.258,-215.452 444.258,-183.452 659.742,-183.452 663.742,-187.452 663.742,-219.452"/>
|
||||
<polyline fill="none" stroke="black" points="659.742,-215.452 444.258,-215.452 "/>
|
||||
<polyline fill="none" stroke="black" points="659.742,-215.452 659.742,-183.452 "/>
|
||||
<polyline fill="none" stroke="black" points="659.742,-215.452 663.742,-219.452 "/>
|
||||
<text text-anchor="middle" x="554" y="-195.452" font-family="Times,serif" font-size="20.00">render_hocr_debug_page</text>
|
||||
</g>
|
||||
<!-- t6->t8 -->
|
||||
<g id="edge11" class="edge"><title>t6->t8</title>
|
||||
<path fill="none" stroke="#0044a0" d="M564.418,-241.346C563.4,-237.546 562.298,-233.43 561.203,-229.345"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="564.522,-228.207 558.554,-219.453 557.761,-230.018 564.522,-228.207"/>
|
||||
</g>
|
||||
<!-- t5->t7 -->
|
||||
<g id="edge10" class="edge"><title>t5->t7</title>
|
||||
<path fill="none" stroke="#0044a0" d="M364.752,-241.346C362.816,-237.329 360.708,-232.958 358.629,-228.645"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="361.693,-226.941 354.197,-219.453 355.387,-229.981 361.693,-226.941"/>
|
||||
<g id="node7" class="node"><title>t8</title>
|
||||
<polygon fill="#efa03b" stroke="black" points="986.109,-277.109 701.891,-277.109 697.891,-273.109 697.891,-241.109 982.109,-241.109 986.109,-245.109 986.109,-277.109"/>
|
||||
<polyline fill="none" stroke="black" points="982.109,-273.109 697.891,-273.109 "/>
|
||||
<polyline fill="none" stroke="black" points="982.109,-273.109 982.109,-241.109 "/>
|
||||
<polyline fill="none" stroke="black" points="982.109,-273.109 986.109,-277.109 "/>
|
||||
<text text-anchor="middle" x="842" y="-253.109" font-family="Times,serif" font-size="20.00">render_hocr_page</text>
|
||||
</g>
|
||||
<!-- t5->t8 -->
|
||||
<g id="edge12" class="edge"><title>t5->t8</title>
|
||||
<path fill="none" stroke="#0044a0" d="M428.289,-241.346C447.587,-235.375 469.416,-228.622 489.404,-222.437"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="490.53,-225.753 499.049,-219.453 488.461,-219.065 490.53,-225.753"/>
|
||||
<g id="edge6" class="edge"><title>t5->t8</title>
|
||||
<path fill="none" stroke="#0044a0" d="M617.226,-299.079C654.028,-292.699 696.036,-285.416 733.699,-278.886"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="734.429,-282.312 743.685,-277.155 733.234,-275.415 734.429,-282.312"/>
|
||||
</g>
|
||||
<!-- t12 -->
|
||||
<g id="node13" class="node"><title>t12</title>
|
||||
<polygon fill="#efa03b" stroke="black" points="473.845,-105.456 554,-78.0208 634.155,-105.456 634.08,-149.848 473.92,-149.848 473.845,-105.456"/>
|
||||
<polygon fill="none" stroke="black" points="469.836,-102.581 554,-73.7729 638.164,-102.581 638.078,-153.869 469.922,-153.869 469.836,-102.581"/>
|
||||
<text text-anchor="middle" x="554" y="-111.726" font-family="Times,serif" font-size="20.00">merge_pages</text>
|
||||
<!-- t9 -->
|
||||
<g id="node11" class="node"><title>t9</title>
|
||||
<polygon fill="#efa03b" stroke="black" points="679.486,-277.109 336.514,-277.109 332.514,-273.109 332.514,-241.109 675.486,-241.109 679.486,-245.109 679.486,-277.109"/>
|
||||
<polyline fill="none" stroke="black" points="675.486,-273.109 332.514,-273.109 "/>
|
||||
<polyline fill="none" stroke="black" points="675.486,-273.109 675.486,-241.109 "/>
|
||||
<polyline fill="none" stroke="black" points="675.486,-273.109 679.486,-277.109 "/>
|
||||
<text text-anchor="middle" x="506" y="-253.109" font-family="Times,serif" font-size="20.00">render_hocr_debug_page</text>
|
||||
</g>
|
||||
<!-- t7->t12 -->
|
||||
<g id="edge20" class="edge"><title>t7->t12</title>
|
||||
<path fill="none" stroke="#0044a0" d="M389.35,-183.419C409.937,-175.33 435.398,-165.326 460.092,-155.624"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="461.479,-158.839 469.507,-151.924 458.919,-152.324 461.479,-158.839"/>
|
||||
<!-- t5->t9 -->
|
||||
<g id="edge15" class="edge"><title>t5->t9</title>
|
||||
<path fill="none" stroke="#0044a0" d="M515.029,-299.003C514.147,-295.204 513.191,-291.087 512.243,-287.002"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="515.617,-286.06 509.947,-277.111 508.799,-287.643 515.617,-286.06"/>
|
||||
</g>
|
||||
<!-- t8->t12 -->
|
||||
<g id="edge19" class="edge"><title>t8->t12</title>
|
||||
<path fill="none" stroke="#0044a0" d="M554,-183.12C554,-177.585 554,-171.177 554,-164.592"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="557.5,-164.201 554,-154.201 550.5,-164.201 557.5,-164.201"/>
|
||||
<!-- t10 -->
|
||||
<g id="node10" class="node"><title>t10</title>
|
||||
<polygon fill="#efa03b" stroke="black" points="973.082,-219.109 714.918,-219.109 710.918,-215.109 710.918,-183.109 969.082,-183.109 973.082,-187.109 973.082,-219.109"/>
|
||||
<polyline fill="none" stroke="black" points="969.082,-215.109 710.918,-215.109 "/>
|
||||
<polyline fill="none" stroke="black" points="969.082,-215.109 969.082,-183.109 "/>
|
||||
<polyline fill="none" stroke="black" points="969.082,-215.109 973.082,-219.109 "/>
|
||||
<text text-anchor="middle" x="842" y="-195.109" font-family="Times,serif" font-size="20.00">add_text_layer</text>
|
||||
</g>
|
||||
<!-- t11->t12 -->
|
||||
<g id="edge17" class="edge"><title>t11->t12</title>
|
||||
<path fill="none" stroke="#0044a0" d="M615.939,-357.13C634.767,-339.333 661.745,-309.733 673,-277.452 686.754,-238.003 694.347,-219.364 673,-183.452 666.345,-172.256 656.999,-162.875 646.44,-155.047"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="648.378,-152.132 638.152,-149.36 644.417,-157.904 648.378,-152.132"/>
|
||||
<!-- t8->t10 -->
|
||||
<g id="edge12" class="edge"><title>t8->t10</title>
|
||||
<path fill="none" stroke="#0044a0" d="M842,-241.003C842,-237.312 842,-233.322 842,-229.352"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="845.5,-229.111 842,-219.111 838.5,-229.111 845.5,-229.111"/>
|
||||
</g>
|
||||
<!-- t9->t12 -->
|
||||
<g id="edge18" class="edge"><title>t9->t12</title>
|
||||
<path fill="none" stroke="#0044a0" d="M164.711,-241.235C186.394,-224.08 222.088,-198.224 257,-183.452 321.895,-155.994 399.893,-139.571 459.67,-130.141"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="460.504,-133.554 469.855,-128.574 459.439,-126.635 460.504,-133.554"/>
|
||||
<!-- t6->t7 -->
|
||||
<g id="edge10" class="edge"><title>t6->t7</title>
|
||||
<path fill="none" stroke="#0044a0" d="M185.836,-299.003C183.626,-294.878 181.216,-290.379 178.845,-285.952"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="181.915,-284.273 174.108,-277.111 175.745,-287.578 181.915,-284.273"/>
|
||||
</g>
|
||||
<!-- t10->t12 -->
|
||||
<g id="edge22" class="edge"><title>t10->t12</title>
|
||||
<path fill="none" stroke="gray" d="M629.379,-415.337C660.552,-396.255 703,-362.233 703,-318.452 703,-318.452 703,-318.452 703,-258.452 703,-224.066 705.581,-210.223 684,-183.452 673.976,-171.017 660.952,-160.773 647.026,-152.396"/>
|
||||
<polygon fill="gray" stroke="gray" points="648.682,-149.312 638.258,-147.419 645.227,-155.4 648.682,-149.312"/>
|
||||
<!-- t6->t9 -->
|
||||
<g id="edge14" class="edge"><title>t6->t9</title>
|
||||
<path fill="none" stroke="#0044a0" d="M289.577,-299.079C324.86,-292.726 365.115,-285.478 401.259,-278.969"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="402.116,-282.372 411.337,-277.155 400.875,-275.482 402.116,-282.372"/>
|
||||
</g>
|
||||
<!-- t13 -->
|
||||
<g id="node14" class="node"><title>t13</title>
|
||||
<polygon fill="#efa03b" stroke="black" points="616.338,-52 495.662,-52 491.662,-48 491.662,-16 612.338,-16 616.338,-20 616.338,-52"/>
|
||||
<polyline fill="none" stroke="black" points="612.338,-48 491.662,-48 "/>
|
||||
<polyline fill="none" stroke="black" points="612.338,-48 612.338,-16 "/>
|
||||
<polyline fill="none" stroke="black" points="612.338,-48 616.338,-52 "/>
|
||||
<text text-anchor="middle" x="554" y="-28" font-family="Times,serif" font-size="20.00">validate_pdfa</text>
|
||||
<!-- t7->t10 -->
|
||||
<g id="edge13" class="edge"><title>t7->t10</title>
|
||||
<path fill="none" stroke="#0044a0" d="M314.366,-242.002C317.606,-241.697 320.821,-241.399 324,-241.109 451.228,-229.521 596.262,-218.816 700.44,-211.565"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="700.976,-215.036 710.71,-210.852 700.491,-208.053 700.976,-215.036"/>
|
||||
</g>
|
||||
<!-- t12->t13 -->
|
||||
<g id="edge21" class="edge"><title>t12->t13</title>
|
||||
<path fill="none" stroke="#0044a0" d="M554,-73.9482C554,-69.9654 554,-66.007 554,-62.2247"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="557.5,-62.1573 554,-52.1573 550.5,-62.1574 557.5,-62.1573"/>
|
||||
<!-- t14 -->
|
||||
<g id="node15" class="node"><title>t14</title>
|
||||
<polygon fill="#efa03b" stroke="black" points="774.472,-105.333 939,-78.005 1103.53,-105.333 1103.37,-149.551 774.625,-149.551 774.472,-105.333"/>
|
||||
<polygon fill="none" stroke="black" points="770.46,-101.94 939,-73.9453 1107.54,-101.94 1107.36,-153.556 770.639,-153.556 770.46,-101.94"/>
|
||||
<text text-anchor="middle" x="939" y="-111.555" font-family="Times,serif" font-size="20.00">merge_pages</text>
|
||||
</g>
|
||||
<!-- t10->t14 -->
|
||||
<g id="edge23" class="edge"><title>t10->t14</title>
|
||||
<path fill="none" stroke="#0044a0" d="M862.571,-182.814C870.479,-176.165 879.888,-168.254 889.356,-160.293"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="891.89,-162.736 897.292,-153.622 887.385,-157.378 891.89,-162.736"/>
|
||||
</g>
|
||||
<!-- t9->t14 -->
|
||||
<g id="edge24" class="edge"><title>t9->t14</title>
|
||||
<path fill="none" stroke="#0044a0" d="M546.87,-241.103C586.226,-225.054 647.626,-200.869 702,-183.109 730.59,-173.771 761.415,-164.679 791.074,-156.405"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="792.38,-159.675 801.082,-153.632 790.511,-152.93 792.38,-159.675"/>
|
||||
</g>
|
||||
<!-- t13->t14 -->
|
||||
<g id="edge20" class="edge"><title>t13->t14</title>
|
||||
<path fill="none" stroke="#0044a0" d="M986.636,-414.956C1033.58,-398.748 1087,-369.103 1087,-318.109 1087,-318.109 1087,-318.109 1087,-258.109 1087,-215.948 1054.81,-182.761 1020.46,-159.346"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="1021.98,-156.158 1011.7,-153.604 1018.15,-162.013 1021.98,-156.158"/>
|
||||
</g>
|
||||
<!-- t11->t14 -->
|
||||
<g id="edge22" class="edge"><title>t11->t14</title>
|
||||
<path fill="none" stroke="#0044a0" d="M969.056,-299.033C979.177,-293.579 988.239,-286.44 995,-277.109 1020.03,-242.568 998.243,-195.803 974.929,-162.021"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="977.661,-159.825 968.996,-153.728 971.968,-163.897 977.661,-159.825"/>
|
||||
</g>
|
||||
<!-- t12->t14 -->
|
||||
<g id="edge21" class="edge"><title>t12->t14</title>
|
||||
<path fill="none" stroke="#0044a0" d="M1006.64,-472.956C1053.58,-456.748 1107,-427.103 1107,-376.109 1107,-376.109 1107,-376.109 1107,-258.109 1107,-220.742 1095.53,-209.426 1069,-183.109 1059.8,-173.988 1049.06,-165.927 1037.76,-158.869"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="1039.41,-155.777 1029.03,-153.665 1035.83,-161.791 1039.41,-155.777"/>
|
||||
</g>
|
||||
<!-- t15 -->
|
||||
<g id="node16" class="node"><title>t15</title>
|
||||
<polygon fill="#efa03b" stroke="black" points="1053.18,-52 828.822,-52 824.822,-48 824.822,-16 1049.18,-16 1053.18,-20 1053.18,-52"/>
|
||||
<polyline fill="none" stroke="black" points="1049.18,-48 824.822,-48 "/>
|
||||
<polyline fill="none" stroke="black" points="1049.18,-48 1049.18,-16 "/>
|
||||
<polyline fill="none" stroke="black" points="1049.18,-48 1053.18,-52 "/>
|
||||
<text text-anchor="middle" x="939" y="-28" font-family="Times,serif" font-size="20.00">copy_final</text>
|
||||
</g>
|
||||
<!-- t14->t15 -->
|
||||
<g id="edge25" class="edge"><title>t14->t15</title>
|
||||
<path fill="none" stroke="#0044a0" d="M939,-73.8665C939,-69.8921 939,-65.942 939,-62.1676"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="942.5,-62.1213 939,-52.1214 935.5,-62.1214 942.5,-62.1213"/>
|
||||
</g>
|
||||
</g>
|
||||
</svg>
|
||||
|
||||
|
Before Width: | Height: | Size: 16 KiB After Width: | Height: | Size: 18 KiB |
+2
-1
@@ -1,4 +1,5 @@
|
||||
ruffus>=2.6.3
|
||||
Pillow>=2.4.0
|
||||
reportlab>=3.1.44
|
||||
PyPDF2>=1.25.1
|
||||
PyPDF2>=1.25.1
|
||||
img2pdf>=0.1.5
|
||||
@@ -1,2 +1 @@
|
||||
img2pdf>=0.1.5
|
||||
pytest>=2.7.2
|
||||
@@ -199,6 +199,7 @@ def test_override_metadata(spoof_tesseract_noop):
|
||||
def test_oversample(spoof_tesseract_cache, renderer):
|
||||
oversampled_pdf = check_ocrmypdf(
|
||||
'skew.pdf', 'test_oversample_%s.pdf' % renderer, '--oversample', '300',
|
||||
'-f',
|
||||
'--pdf-renderer', renderer, env=spoof_tesseract_cache)
|
||||
|
||||
pdfinfo = pdf_get_all_pageinfo(oversampled_pdf)
|
||||
|
||||
Reference in New Issue
Block a user