Unify tesseract and sandwich renderer paths
Since the new weaving method copies the font and content stream from the Tesseract PDF, it doesn't matter if Tesseract happens to have an image or not. If Tesseract is text-only capable we use that feature for efficiency, but ignore the image either way.
This commit is contained in:
James R. Barlow
+403
-420
@@ -4,470 +4,453 @@
|
||||
<!-- Generated by graphviz version 2.40.1 (20161225.0304)
|
||||
-->
|
||||
<!-- Title: Pipeline: Pages: 1 -->
|
||||
<svg width="1932pt" height="1277pt"
|
||||
viewBox="0.00 0.00 1932.00 1277.21" 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 1273.2069)">
|
||||
<svg width="1660pt" height="1270pt"
|
||||
viewBox="0.00 0.00 1660.00 1270.29" 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 1266.2885)">
|
||||
<title>Pipeline:</title>
|
||||
<polygon fill="#ffffff" stroke="transparent" points="-4,4 -4,-1273.2069 1928,-1273.2069 1928,4 -4,4"/>
|
||||
<polygon fill="#ffffff" stroke="transparent" points="-4,4 -4,-1266.2885 1656,-1266.2885 1656,4 -4,4"/>
|
||||
<g id="clust1" class="cluster">
|
||||
<title>clustertasks</title>
|
||||
<polygon fill="none" stroke="#000000" points="8,-8 8,-1261.2069 1916,-1261.2069 1916,-8 8,-8"/>
|
||||
<text text-anchor="middle" x="962" y="-1233.2069" font-family="Times,serif" font-size="30.00" fill="#ff3232">Pipeline:</text>
|
||||
<polygon fill="none" stroke="#000000" points="8,-8 8,-1254.2885 1644,-1254.2885 1644,-8 8,-8"/>
|
||||
<text text-anchor="middle" x="826" y="-1226.2885" 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="#000000" points="1810.7053,-1215.2069 1601.2947,-1215.2069 1597.2947,-1211.2069 1597.2947,-1179.2069 1806.7053,-1179.2069 1810.7053,-1183.2069 1810.7053,-1215.2069"/>
|
||||
<polyline fill="none" stroke="#000000" points="1806.7053,-1211.2069 1597.2947,-1211.2069 "/>
|
||||
<polyline fill="none" stroke="#000000" points="1806.7053,-1211.2069 1806.7053,-1179.2069 "/>
|
||||
<polyline fill="none" stroke="#000000" points="1806.7053,-1211.2069 1810.7053,-1215.2069 "/>
|
||||
<text text-anchor="middle" x="1704" y="-1191.2069" font-family="Times,serif" font-size="20.00" fill="#000000">ocrmypdf.pipeline.triage</text>
|
||||
<polygon fill="#efa03b" stroke="#000000" points="293.7053,-1208.2885 84.2947,-1208.2885 80.2947,-1204.2885 80.2947,-1172.2885 289.7053,-1172.2885 293.7053,-1176.2885 293.7053,-1208.2885"/>
|
||||
<polyline fill="none" stroke="#000000" points="289.7053,-1204.2885 80.2947,-1204.2885 "/>
|
||||
<polyline fill="none" stroke="#000000" points="289.7053,-1204.2885 289.7053,-1172.2885 "/>
|
||||
<polyline fill="none" stroke="#000000" points="289.7053,-1204.2885 293.7053,-1208.2885 "/>
|
||||
<text text-anchor="middle" x="187" y="-1184.2885" font-family="Times,serif" font-size="20.00" fill="#000000">ocrmypdf.pipeline.triage</text>
|
||||
</g>
|
||||
<!-- t1 -->
|
||||
<g id="node2" class="node">
|
||||
<title>t1</title>
|
||||
<polygon fill="#efa03b" stroke="#000000" points="1875.0433,-1157.2069 1536.9567,-1157.2069 1532.9567,-1153.2069 1532.9567,-1121.2069 1871.0433,-1121.2069 1875.0433,-1125.2069 1875.0433,-1157.2069"/>
|
||||
<polyline fill="none" stroke="#000000" points="1871.0433,-1153.2069 1532.9567,-1153.2069 "/>
|
||||
<polyline fill="none" stroke="#000000" points="1871.0433,-1153.2069 1871.0433,-1121.2069 "/>
|
||||
<polyline fill="none" stroke="#000000" points="1871.0433,-1153.2069 1875.0433,-1157.2069 "/>
|
||||
<text text-anchor="middle" x="1704" y="-1133.2069" font-family="Times,serif" font-size="20.00" fill="#000000">ocrmypdf.pipeline.repair_and_parse_pdf</text>
|
||||
<polygon fill="#efa03b" stroke="#000000" points="358.0433,-1150.2885 19.9567,-1150.2885 15.9567,-1146.2885 15.9567,-1114.2885 354.0433,-1114.2885 358.0433,-1118.2885 358.0433,-1150.2885"/>
|
||||
<polyline fill="none" stroke="#000000" points="354.0433,-1146.2885 15.9567,-1146.2885 "/>
|
||||
<polyline fill="none" stroke="#000000" points="354.0433,-1146.2885 354.0433,-1114.2885 "/>
|
||||
<polyline fill="none" stroke="#000000" points="354.0433,-1146.2885 358.0433,-1150.2885 "/>
|
||||
<text text-anchor="middle" x="187" y="-1126.2885" font-family="Times,serif" font-size="20.00" fill="#000000">ocrmypdf.pipeline.repair_and_parse_pdf</text>
|
||||
</g>
|
||||
<!-- t0->t1 -->
|
||||
<g id="edge1" class="edge">
|
||||
<title>t0->t1</title>
|
||||
<path fill="none" stroke="#0044a0" d="M1704,-1179.1733C1704,-1175.4987 1704,-1171.5487 1704,-1167.6132"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="1707.5001,-1167.4291 1704,-1157.4291 1700.5001,-1167.4292 1707.5001,-1167.4291"/>
|
||||
<path fill="none" stroke="#0044a0" d="M187,-1172.2548C187,-1168.5803 187,-1164.6302 187,-1160.6948"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="190.5001,-1160.5107 187,-1150.5107 183.5001,-1160.5108 190.5001,-1160.5107"/>
|
||||
</g>
|
||||
<!-- t2 -->
|
||||
<g id="node3" class="node">
|
||||
<title>t2</title>
|
||||
<polygon fill="#efa03b" stroke="#000000" points="1656.9447,-1078.283 1400,-1095.1824 1143.0553,-1078.283 1143.2951,-1050.9391 1656.7049,-1050.9391 1656.9447,-1078.283"/>
|
||||
<polygon fill="none" stroke="#000000" points="1660.9814,-1082.0236 1400,-1099.1885 1139.0186,-1082.0236 1139.3263,-1046.9417 1660.6737,-1046.9417 1660.9814,-1082.0236"/>
|
||||
<text text-anchor="middle" x="1400" y="-1064.7253" font-family="Times,serif" font-size="20.00" fill="#000000">ocrmypdf.pipeline.pre_split_pages</text>
|
||||
<polygon fill="#efa03b" stroke="#000000" points="686.0926,-1071.3639 441,-1088.2639 195.9074,-1071.3639 196.1362,-1044.0193 685.8638,-1044.0193 686.0926,-1071.3639"/>
|
||||
<polygon fill="none" stroke="#000000" points="690.0px49,-1075.0929 441,-1092.2709 191.8751,-1075.0929 192.1685,-1040.0218 689.8315,-1040.0218 690.0px49,-1075.0929"/>
|
||||
<text text-anchor="middle" x="441" y="-1057.8061" font-family="Times,serif" font-size="20.00" fill="#000000">ocrmypdf.pipeline.marker_pages</text>
|
||||
</g>
|
||||
<!-- t1->t2 -->
|
||||
<g id="edge2" class="edge">
|
||||
<title>t1->t2</title>
|
||||
<path fill="none" stroke="#0044a0" d="M1623.7299,-1121.1246C1588.2849,-1113.14 1546.0878,-1103.6343 1507.9425,-1095.0414"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="1508.3459,-1091.5446 1497.8212,-1092.7613 1506.8075,-1098.3735 1508.3459,-1091.5446"/>
|
||||
</g>
|
||||
<!-- t20 -->
|
||||
<g id="node22" class="node">
|
||||
<title>t20</title>
|
||||
<polygon fill="#efa03b" stroke="#000000" points="1887.5156,-1020.2438 1524.4844,-1020.2438 1520.4844,-1016.2438 1520.4844,-984.2438 1883.5156,-984.2438 1887.5156,-988.2438 1887.5156,-1020.2438"/>
|
||||
<polyline fill="none" stroke="#000000" points="1883.5156,-1016.2438 1520.4844,-1016.2438 "/>
|
||||
<polyline fill="none" stroke="#000000" points="1883.5156,-1016.2438 1883.5156,-984.2438 "/>
|
||||
<polyline fill="none" stroke="#000000" points="1883.5156,-1016.2438 1887.5156,-1020.2438 "/>
|
||||
<text text-anchor="middle" x="1704" y="-996.2438" font-family="Times,serif" font-size="20.00" fill="#000000">ocrmypdf.pipeline.generate_postscript_stub</text>
|
||||
</g>
|
||||
<!-- t1->t20 -->
|
||||
<g id="edge31" class="edge">
|
||||
<title>t1->t20</title>
|
||||
<path fill="none" stroke="#0044a0" d="M1704,-1121.0193C1704,-1097.893 1704,-1057.7125 1704,-1030.5603"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="1707.5001,-1030.5066 1704,-1020.5066 1700.5001,-1030.5067 1707.5001,-1030.5066"/>
|
||||
</g>
|
||||
<!-- t22 -->
|
||||
<g id="node23" class="node">
|
||||
<title>t22</title>
|
||||
<polygon fill="#efa03b" stroke="#000000" points="1542.1753,-156.4327 1723,-129.1104 1903.8247,-156.4327 1903.6559,-200.6411 1542.3441,-200.6411 1542.1753,-156.4327"/>
|
||||
<polygon fill="none" stroke="#000000" points="1538.1584,-152.99 1723,-125.0608 1907.8416,-152.99 1907.6444,-204.6453 1538.3556,-204.6453 1538.1584,-152.99"/>
|
||||
<text text-anchor="middle" x="1723" y="-162.6516" font-family="Times,serif" font-size="20.00" fill="#000000">ocrmypdf.pipeline.merge_pages</text>
|
||||
</g>
|
||||
<!-- t1->t22 -->
|
||||
<g id="edge37" class="edge">
|
||||
<title>t1->t22</title>
|
||||
<path fill="none" stroke="#0044a0" d="M1779.013,-1121.1962C1836.8401,-1102.3174 1907,-1066.2281 1907,-1002.2438 1907,-1002.2438 1907,-1002.2438 1907,-335.7367 1907,-285.0344 1892.7171,-268.0813 1855,-234.197 1844.8914,-225.1156 1833.2959,-216.9812 1821.2655,-209.7918"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="1822.8581,-206.6703 1812.4436,-204.7237 1819.3711,-212.74 1822.8581,-206.6703"/>
|
||||
</g>
|
||||
<!-- t3 -->
|
||||
<g id="node4" class="node">
|
||||
<title>t3</title>
|
||||
<polygon fill="#efa03b" stroke="#000000" points="1502.3851,-1020.2438 1255.6149,-1020.2438 1251.6149,-1016.2438 1251.6149,-984.2438 1498.3851,-984.2438 1502.3851,-988.2438 1502.3851,-1020.2438"/>
|
||||
<polyline fill="none" stroke="#000000" points="1498.3851,-1016.2438 1251.6149,-1016.2438 "/>
|
||||
<polyline fill="none" stroke="#000000" points="1498.3851,-1016.2438 1498.3851,-984.2438 "/>
|
||||
<polyline fill="none" stroke="#000000" points="1498.3851,-1016.2438 1502.3851,-1020.2438 "/>
|
||||
<text text-anchor="middle" x="1377" y="-996.2438" font-family="Times,serif" font-size="20.00" fill="#000000">ocrmypdf.pipeline.split_page</text>
|
||||
</g>
|
||||
<!-- t2->t3 -->
|
||||
<g id="edge3" class="edge">
|
||||
<title>t2->t3</title>
|
||||
<path fill="none" stroke="#0044a0" d="M1391.9557,-1046.7738C1390.1421,-1041.3739 1388.2021,-1035.5977 1386.3405,-1030.0546"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="1389.62,-1028.826 1383.1183,-1020.4607 1382.9843,-1031.0547 1389.62,-1028.826"/>
|
||||
</g>
|
||||
<!-- t4 -->
|
||||
<g id="node5" class="node">
|
||||
<title>t4</title>
|
||||
<polygon fill="#efa03b" stroke="#000000" points="1607.5194,-941.3183 1377,-958.219 1146.4806,-941.3183 1146.6958,-913.9725 1607.3042,-913.9725 1607.5194,-941.3183"/>
|
||||
<polygon fill="none" stroke="#000000" points="1611.5491,-945.0313 1377,-962.2274 1142.4509,-945.0313 1142.7267,-909.9748 1611.2733,-909.9748 1611.5491,-945.0313"/>
|
||||
<text text-anchor="middle" x="1377" y="-927.7601" font-family="Times,serif" font-size="20.00" fill="#000000">ocrmypdf.pipeline.ocr_or_skip</text>
|
||||
</g>
|
||||
<!-- t3->t4 -->
|
||||
<g id="edge4" class="edge">
|
||||
<title>t3->t4</title>
|
||||
<path fill="none" stroke="#0044a0" d="M1377,-983.8914C1377,-980.3974 1377,-976.6197 1377,-972.7598"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="1380.5001,-972.4886 1377,-962.4886 1373.5001,-972.4887 1380.5001,-972.4886"/>
|
||||
</g>
|
||||
<!-- t5 -->
|
||||
<g id="node6" class="node">
|
||||
<title>t5</title>
|
||||
<polygon fill="#efa03b" stroke="#000000" points="1405.3166,-883.2765 1098.6834,-883.2765 1094.6834,-879.2765 1094.6834,-847.2765 1401.3166,-847.2765 1405.3166,-851.2765 1405.3166,-883.2765"/>
|
||||
<polyline fill="none" stroke="#000000" points="1401.3166,-879.2765 1094.6834,-879.2765 "/>
|
||||
<polyline fill="none" stroke="#000000" points="1401.3166,-879.2765 1401.3166,-847.2765 "/>
|
||||
<polyline fill="none" stroke="#000000" points="1401.3166,-879.2765 1405.3166,-883.2765 "/>
|
||||
<text text-anchor="middle" x="1250" y="-859.2765" font-family="Times,serif" font-size="20.00" fill="#000000">ocrmypdf.pipeline.rasterize_preview</text>
|
||||
</g>
|
||||
<!-- t4->t5 -->
|
||||
<g id="edge5" class="edge">
|
||||
<title>t4->t5</title>
|
||||
<path fill="none" stroke="#0044a0" d="M1332.5816,-909.8079C1319.5732,-902.7932 1305.3872,-895.1435 1292.5011,-888.1948"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="1293.8464,-884.9439 1283.3833,-883.2781 1290.5239,-891.1052 1293.8464,-884.9439"/>
|
||||
</g>
|
||||
<!-- t6 -->
|
||||
<g id="node7" class="node">
|
||||
<title>t6</title>
|
||||
<polygon fill="#efa03b" stroke="#000000" points="1475.0824,-825.2765 1216.9176,-825.2765 1212.9176,-821.2765 1212.9176,-789.2765 1471.0824,-789.2765 1475.0824,-793.2765 1475.0824,-825.2765"/>
|
||||
<polyline fill="none" stroke="#000000" points="1471.0824,-821.2765 1212.9176,-821.2765 "/>
|
||||
<polyline fill="none" stroke="#000000" points="1471.0824,-821.2765 1471.0824,-789.2765 "/>
|
||||
<polyline fill="none" stroke="#000000" points="1471.0824,-821.2765 1475.0824,-825.2765 "/>
|
||||
<text text-anchor="middle" x="1344" y="-801.2765" font-family="Times,serif" font-size="20.00" fill="#000000">ocrmypdf.pipeline.orient_page</text>
|
||||
</g>
|
||||
<!-- t4->t6 -->
|
||||
<g id="edge7" class="edge">
|
||||
<title>t4->t6</title>
|
||||
<path fill="none" stroke="#0044a0" d="M1399.1144,-909.8909C1412.803,-892.0931 1425.7216,-867.4749 1414,-847.2765 1410.4598,-841.1761 1405.6012,-835.9275 1400.0802,-831.4317"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="1401.7661,-828.3348 1391.594,-825.372 1397.6983,-834.0316 1401.7661,-828.3348"/>
|
||||
</g>
|
||||
<!-- t5->t6 -->
|
||||
<g id="edge6" class="edge">
|
||||
<title>t5->t6</title>
|
||||
<path fill="none" stroke="#0044a0" d="M1279.2269,-847.2428C1287.7153,-842.0053 1297.1107,-836.2082 1306.0518,-830.6913"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="1308.1187,-833.5287 1314.7912,-825.2989 1304.4429,-827.5714 1308.1187,-833.5287"/>
|
||||
</g>
|
||||
<!-- t7 -->
|
||||
<g id="node8" class="node">
|
||||
<title>t7</title>
|
||||
<polygon fill="#efa03b" stroke="#000000" points="1533.7325,-767.2765 1158.2675,-767.2765 1154.2675,-763.2765 1154.2675,-731.2765 1529.7325,-731.2765 1533.7325,-735.2765 1533.7325,-767.2765"/>
|
||||
<polyline fill="none" stroke="#000000" points="1529.7325,-763.2765 1154.2675,-763.2765 "/>
|
||||
<polyline fill="none" stroke="#000000" points="1529.7325,-763.2765 1529.7325,-731.2765 "/>
|
||||
<polyline fill="none" stroke="#000000" points="1529.7325,-763.2765 1533.7325,-767.2765 "/>
|
||||
<text text-anchor="middle" x="1344" y="-743.2765" font-family="Times,serif" font-size="20.00" fill="#000000">ocrmypdf.pipeline.rasterize_with_ghostscript</text>
|
||||
</g>
|
||||
<!-- t6->t7 -->
|
||||
<g id="edge8" class="edge">
|
||||
<title>t6->t7</title>
|
||||
<path fill="none" stroke="#0044a0" d="M1344,-789.2428C1344,-785.5683 1344,-781.6182 1344,-777.6828"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="1347.5001,-777.4987 1344,-767.4987 1340.5001,-777.4987 1347.5001,-777.4987"/>
|
||||
<path fill="none" stroke="#0044a0" d="M254.0678,-1114.2059C282.4495,-1106.5538 316.0126,-1097.5046 346.8117,-1089.2007"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="347.9601,-1092.5161 356.7042,-1086.5335 346.1379,-1085.7575 347.9601,-1092.5161"/>
|
||||
</g>
|
||||
<!-- t17 -->
|
||||
<g id="node15" class="node">
|
||||
<title>t17</title>
|
||||
<polygon fill="#ff69b4" stroke="#000000" points="1064.2696,-447.2765 683.7304,-447.2765 679.7304,-443.2765 679.7304,-411.2765 1060.2696,-411.2765 1064.2696,-415.2765 1064.2696,-447.2765"/>
|
||||
<polyline fill="none" stroke="#000000" points="1060.2696,-443.2765 679.7304,-443.2765 "/>
|
||||
<polyline fill="none" stroke="#000000" points="1060.2696,-443.2765 1060.2696,-411.2765 "/>
|
||||
<polyline fill="none" stroke="#000000" points="1060.2696,-443.2765 1064.2696,-447.2765 "/>
|
||||
<text text-anchor="middle" x="872" y="-423.2765" font-family="Times,serif" font-size="20.00" fill="#000000">ocrmypdf.pipeline.ocr_tesseract_textonly_pdf</text>
|
||||
</g>
|
||||
<!-- t6->t17 -->
|
||||
<g id="edge16" class="edge">
|
||||
<title>t6->t17</title>
|
||||
<path fill="none" stroke="#0044a0" d="M1212.9072,-798.053C1014.8011,-782.3275 662,-746.5615 662,-691.2765 662,-691.2765 662,-691.2765 662,-575.2765 662,-532.3789 643.882,-511.6736 672,-479.2765 682.6302,-467.0285 710.7979,-457.1862 742.7543,-449.5814"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="743.8081,-452.9309 752.7782,-447.2929 742.25,-446.1065 743.8081,-452.9309"/>
|
||||
</g>
|
||||
<!-- t14 -->
|
||||
<g id="node17" class="node">
|
||||
<title>t14</title>
|
||||
<polygon fill="#00cc66" stroke="#000000" points="339,-457.2765 16.4141,-429.2765 339,-401.2765 661.5859,-429.2765 339,-457.2765"/>
|
||||
<text text-anchor="middle" x="339" y="-423.2765" font-family="Times,serif" font-size="20.00" fill="#000000">ocrmypdf.pipeline.select_image_layer</text>
|
||||
<title>t17</title>
|
||||
<polygon fill="#00cc66" stroke="#000000" points="1578.5059,-295.7315 1305.4941,-295.7315 1301.4941,-291.7315 1301.4941,-259.7315 1574.5059,-259.7315 1578.5059,-263.7315 1578.5059,-295.7315"/>
|
||||
<polyline fill="none" stroke="#000000" points="1574.5059,-291.7315 1301.4941,-291.7315 "/>
|
||||
<polyline fill="none" stroke="#000000" points="1574.5059,-291.7315 1574.5059,-259.7315 "/>
|
||||
<polyline fill="none" stroke="#000000" points="1574.5059,-291.7315 1578.5059,-295.7315 "/>
|
||||
<text text-anchor="middle" x="1440" y="-271.7315" font-family="Times,serif" font-size="20.00" fill="#000000">ocrmypdf.pipeline.weave_layers</text>
|
||||
</g>
|
||||
<!-- t6->t14 -->
|
||||
<g id="edge22" class="edge">
|
||||
<title>t6->t14</title>
|
||||
<path fill="none" stroke="#0044a0" d="M1212.7923,-802.9902C970.4193,-793.3094 477,-764.5974 477,-691.2765 477,-691.2765 477,-691.2765 477,-575.2765 477,-524.368 432.1386,-484.0495 393.5722,-458.7944"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="395.2709,-455.7271 384.952,-453.3247 391.5205,-461.6377 395.2709,-455.7271"/>
|
||||
</g>
|
||||
<!-- t21 -->
|
||||
<g id="node20" class="node">
|
||||
<title>t21</title>
|
||||
<polygon fill="#efa03b" stroke="#000000" points="1867.7716,-709.2765 1622.2284,-709.2765 1618.2284,-705.2765 1618.2284,-673.2765 1863.7716,-673.2765 1867.7716,-677.2765 1867.7716,-709.2765"/>
|
||||
<polyline fill="none" stroke="#000000" points="1863.7716,-705.2765 1618.2284,-705.2765 "/>
|
||||
<polyline fill="none" stroke="#000000" points="1863.7716,-705.2765 1863.7716,-673.2765 "/>
|
||||
<polyline fill="none" stroke="#000000" points="1863.7716,-705.2765 1867.7716,-709.2765 "/>
|
||||
<text text-anchor="middle" x="1743" y="-685.2765" font-family="Times,serif" font-size="20.00" fill="#000000">ocrmypdf.pipeline.skip_page</text>
|
||||
</g>
|
||||
<!-- t6->t21 -->
|
||||
<g id="edge28" class="edge">
|
||||
<title>t6->t21</title>
|
||||
<path fill="none" stroke="#0044a0" d="M1475.3964,-796.0838C1517.2386,-790.1439 1563.2301,-781.082 1604,-767.2765 1641.4476,-754.5959 1680.6677,-732.1928 1707.8621,-714.9842"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="1710.0816,-717.7185 1716.6036,-709.369 1706.2983,-711.8289 1710.0816,-717.7185"/>
|
||||
</g>
|
||||
<!-- t19 -->
|
||||
<g id="node21" class="node">
|
||||
<title>t19</title>
|
||||
<polygon fill="#66ccff" stroke="#000000" points="1847.6739,-447.2765 1442.3261,-447.2765 1438.3261,-443.2765 1438.3261,-411.2765 1843.6739,-411.2765 1847.6739,-415.2765 1847.6739,-447.2765"/>
|
||||
<polyline fill="none" stroke="#000000" points="1843.6739,-443.2765 1438.3261,-443.2765 "/>
|
||||
<polyline fill="none" stroke="#000000" points="1843.6739,-443.2765 1843.6739,-411.2765 "/>
|
||||
<polyline fill="none" stroke="#000000" points="1843.6739,-443.2765 1847.6739,-447.2765 "/>
|
||||
<text text-anchor="middle" x="1643" y="-423.2765" font-family="Times,serif" font-size="20.00" fill="#000000">ocrmypdf.pipeline.ocr_tesseract_and_render_pdf</text>
|
||||
</g>
|
||||
<!-- t6->t19 -->
|
||||
<g id="edge30" class="edge">
|
||||
<title>t6->t19</title>
|
||||
<path fill="none" stroke="#0044a0" d="M1471.3274,-789.2627C1502.1321,-783.2141 1529.9438,-775.8026 1543,-767.2765 1585.741,-739.3653 1573.921,-710.3614 1609,-673.2765 1677.0666,-601.3175 1740.7796,-624.4008 1784,-535.2765 1794.8601,-512.8819 1798.3601,-499.6049 1784,-479.2765 1776.4259,-468.5545 1752.8487,-458.504 1727.0874,-450.2665"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="1728.1106,-446.9194 1717.5232,-447.3158 1726.0469,-453.6083 1728.1106,-446.9194"/>
|
||||
</g>
|
||||
<!-- t8 -->
|
||||
<g id="node9" class="node">
|
||||
<title>t8</title>
|
||||
<polygon fill="#efa03b" stroke="#000000" points="1418.0801,-709.2765 993.9199,-709.2765 989.9199,-705.2765 989.9199,-673.2765 1414.0801,-673.2765 1418.0801,-677.2765 1418.0801,-709.2765"/>
|
||||
<polyline fill="none" stroke="#000000" points="1414.0801,-705.2765 989.9199,-705.2765 "/>
|
||||
<polyline fill="none" stroke="#000000" points="1414.0801,-705.2765 1414.0801,-673.2765 "/>
|
||||
<polyline fill="none" stroke="#000000" points="1414.0801,-705.2765 1418.0801,-709.2765 "/>
|
||||
<text text-anchor="middle" x="1204" y="-685.2765" font-family="Times,serif" font-size="20.00" fill="#000000">ocrmypdf.pipeline.preprocess_remove_background</text>
|
||||
</g>
|
||||
<!-- t7->t8 -->
|
||||
<g id="edge9" class="edge">
|
||||
<title>t7->t8</title>
|
||||
<path fill="none" stroke="#0044a0" d="M1300.4705,-731.2428C1286.6431,-725.5143 1271.1996,-719.1163 1256.7928,-713.1478"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="1258.0807,-709.8929 1247.5025,-709.2989 1255.4015,-716.3599 1258.0807,-709.8929"/>
|
||||
</g>
|
||||
<!-- t13 -->
|
||||
<g id="node16" class="node">
|
||||
<title>t13</title>
|
||||
<polygon fill="#efa03b" stroke="#000000" points="1391,-535.2765 1007.1874,-507.2765 1391,-479.2765 1774.8126,-507.2765 1391,-535.2765"/>
|
||||
<text text-anchor="middle" x="1391" y="-501.2765" font-family="Times,serif" font-size="20.00" fill="#000000">ocrmypdf.pipeline.select_visible_page_image</text>
|
||||
</g>
|
||||
<!-- t7->t13 -->
|
||||
<g id="edge20" class="edge">
|
||||
<title>t7->t13</title>
|
||||
<path fill="none" stroke="#0044a0" d="M1403.1818,-731.1462C1412.6461,-725.627 1421.2143,-718.4872 1427,-709.2765 1459.1275,-658.1296 1432.9036,-586.12 1411.5309,-543.3218"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="1414.6214,-541.6783 1406.9295,-534.3924 1408.399,-544.8848 1414.6214,-541.6783"/>
|
||||
</g>
|
||||
<!-- t9 -->
|
||||
<g id="node10" class="node">
|
||||
<title>t9</title>
|
||||
<polygon fill="#efa03b" stroke="#000000" points="1365.6759,-651.2765 1046.3241,-651.2765 1042.3241,-647.2765 1042.3241,-615.2765 1361.6759,-615.2765 1365.6759,-619.2765 1365.6759,-651.2765"/>
|
||||
<polyline fill="none" stroke="#000000" points="1361.6759,-647.2765 1042.3241,-647.2765 "/>
|
||||
<polyline fill="none" stroke="#000000" points="1361.6759,-647.2765 1361.6759,-615.2765 "/>
|
||||
<polyline fill="none" stroke="#000000" points="1361.6759,-647.2765 1365.6759,-651.2765 "/>
|
||||
<text text-anchor="middle" x="1204" y="-627.2765" font-family="Times,serif" font-size="20.00" fill="#000000">ocrmypdf.pipeline.preprocess_deskew</text>
|
||||
</g>
|
||||
<!-- t8->t9 -->
|
||||
<g id="edge10" class="edge">
|
||||
<title>t8->t9</title>
|
||||
<path fill="none" stroke="#0044a0" d="M1204,-673.2428C1204,-669.5683 1204,-665.6182 1204,-661.6828"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="1207.5001,-661.4987 1204,-651.4987 1200.5001,-661.4987 1207.5001,-661.4987"/>
|
||||
</g>
|
||||
<!-- t8->t13 -->
|
||||
<g id="edge19" class="edge">
|
||||
<title>t8->t13</title>
|
||||
<path fill="none" stroke="#0044a0" d="M1324.7334,-673.2711C1347.6618,-667.5415 1367.0063,-660.2841 1375,-651.2765 1400.3856,-622.6711 1401.2608,-577.249 1397.7857,-545.0765"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="1401.2555,-544.6171 1396.5301,-535.1345 1394.3107,-545.4942 1401.2555,-544.6171"/>
|
||||
</g>
|
||||
<!-- t10 -->
|
||||
<g id="node11" class="node">
|
||||
<title>t10</title>
|
||||
<polygon fill="#efa03b" stroke="#000000" points="1351.8828,-593.2765 1050.1172,-593.2765 1046.1172,-589.2765 1046.1172,-557.2765 1347.8828,-557.2765 1351.8828,-561.2765 1351.8828,-593.2765"/>
|
||||
<polyline fill="none" stroke="#000000" points="1347.8828,-589.2765 1046.1172,-589.2765 "/>
|
||||
<polyline fill="none" stroke="#000000" points="1347.8828,-589.2765 1347.8828,-557.2765 "/>
|
||||
<polyline fill="none" stroke="#000000" points="1347.8828,-589.2765 1351.8828,-593.2765 "/>
|
||||
<text text-anchor="middle" x="1199" y="-569.2765" font-family="Times,serif" font-size="20.00" fill="#000000">ocrmypdf.pipeline.preprocess_clean</text>
|
||||
</g>
|
||||
<!-- t9->t10 -->
|
||||
<g id="edge11" class="edge">
|
||||
<title>t9->t10</title>
|
||||
<path fill="none" stroke="#0044a0" d="M1202.4454,-615.2428C1202.1286,-611.5683 1201.7881,-607.6182 1201.4488,-603.6828"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="1204.9169,-603.1611 1200.5709,-593.4987 1197.9428,-603.7624 1204.9169,-603.1611"/>
|
||||
</g>
|
||||
<!-- t9->t13 -->
|
||||
<g id="edge18" class="edge">
|
||||
<title>t9->t13</title>
|
||||
<path fill="none" stroke="#0044a0" d="M1305.4912,-615.2209C1329.9901,-609.1504 1351.7763,-601.7378 1361,-593.2765 1374.4133,-580.9719 1381.8295,-562.3451 1385.93,-545.5694"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="1389.4281,-545.945 1388.0785,-535.4364 1382.5803,-544.4929 1389.4281,-545.945"/>
|
||||
</g>
|
||||
<!-- t11 -->
|
||||
<g id="node12" class="node">
|
||||
<title>t11</title>
|
||||
<polygon fill="#efa03b" stroke="#000000" points="989.1094,-525.2765 684.8906,-525.2765 680.8906,-521.2765 680.8906,-489.2765 985.1094,-489.2765 989.1094,-493.2765 989.1094,-525.2765"/>
|
||||
<polyline fill="none" stroke="#000000" points="985.1094,-521.2765 680.8906,-521.2765 "/>
|
||||
<polyline fill="none" stroke="#000000" points="985.1094,-521.2765 985.1094,-489.2765 "/>
|
||||
<polyline fill="none" stroke="#000000" points="985.1094,-521.2765 989.1094,-525.2765 "/>
|
||||
<text text-anchor="middle" x="835" y="-501.2765" font-family="Times,serif" font-size="20.00" fill="#000000">ocrmypdf.pipeline.select_ocr_image</text>
|
||||
</g>
|
||||
<!-- t10->t11 -->
|
||||
<g id="edge12" class="edge">
|
||||
<title>t10->t11</title>
|
||||
<path fill="none" stroke="#0044a0" d="M1102.4103,-557.2322C1052.8491,-547.9736 992.3597,-536.6733 941.4558,-527.1638"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="941.8816,-523.6829 931.4089,-525.2869 940.5961,-530.5638 941.8816,-523.6829"/>
|
||||
</g>
|
||||
<!-- t10->t13 -->
|
||||
<g id="edge17" class="edge">
|
||||
<title>t10->t13</title>
|
||||
<path fill="none" stroke="#0044a0" d="M1249.9484,-557.2322C1270.0775,-550.1032 1293.6242,-541.7637 1315.6488,-533.9633"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="1317.0528,-537.1792 1325.3106,-530.5415 1314.7158,-530.5808 1317.0528,-537.1792"/>
|
||||
</g>
|
||||
<!-- t12 -->
|
||||
<g id="node13" class="node">
|
||||
<title>t12</title>
|
||||
<polygon fill="#00cc66" stroke="#000000" points="1400.0996,-447.2765 1085.9004,-447.2765 1081.9004,-443.2765 1081.9004,-411.2765 1396.0996,-411.2765 1400.0996,-415.2765 1400.0996,-447.2765"/>
|
||||
<polyline fill="none" stroke="#000000" points="1396.0996,-443.2765 1081.9004,-443.2765 "/>
|
||||
<polyline fill="none" stroke="#000000" points="1396.0996,-443.2765 1396.0996,-411.2765 "/>
|
||||
<polyline fill="none" stroke="#000000" points="1396.0996,-443.2765 1400.0996,-447.2765 "/>
|
||||
<text text-anchor="middle" x="1241" y="-423.2765" font-family="Times,serif" font-size="20.00" fill="#000000">ocrmypdf.pipeline.ocr_tesseract_hocr</text>
|
||||
</g>
|
||||
<!-- t11->t12 -->
|
||||
<g id="edge13" class="edge">
|
||||
<title>t11->t12</title>
|
||||
<path fill="none" stroke="#0044a0" d="M928.9096,-489.2347C991.1602,-477.2753 1072.8804,-461.5753 1137.0821,-449.241"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="1138.07,-452.6153 1147.23,-447.2914 1136.7492,-445.741 1138.07,-452.6153"/>
|
||||
</g>
|
||||
<!-- t11->t17 -->
|
||||
<g id="edge15" class="edge">
|
||||
<title>t11->t17</title>
|
||||
<path fill="none" stroke="#0044a0" d="M843.5815,-489.1856C848.1717,-479.5091 853.9184,-467.3944 859.0408,-456.5958"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="862.2132,-458.0745 863.3368,-447.5394 855.8887,-455.0743 862.2132,-458.0745"/>
|
||||
</g>
|
||||
<!-- t15 -->
|
||||
<g id="node14" class="node">
|
||||
<title>t15</title>
|
||||
<polygon fill="#00cc66" stroke="#000000" points="1398.0433,-353.7367 1089.9567,-353.7367 1085.9567,-349.7367 1085.9567,-317.7367 1394.0433,-317.7367 1398.0433,-321.7367 1398.0433,-353.7367"/>
|
||||
<polyline fill="none" stroke="#000000" points="1394.0433,-349.7367 1085.9567,-349.7367 "/>
|
||||
<polyline fill="none" stroke="#000000" points="1394.0433,-349.7367 1394.0433,-317.7367 "/>
|
||||
<polyline fill="none" stroke="#000000" points="1394.0433,-349.7367 1398.0433,-353.7367 "/>
|
||||
<text text-anchor="middle" x="1242" y="-329.7367" font-family="Times,serif" font-size="20.00" fill="#000000">ocrmypdf.pipeline.render_hocr_page</text>
|
||||
</g>
|
||||
<!-- t12->t15 -->
|
||||
<g id="edge14" class="edge">
|
||||
<title>t12->t15</title>
|
||||
<path fill="none" stroke="#0044a0" d="M1241.1928,-411.2411C1241.3358,-397.8693 1241.5334,-379.3811 1241.6973,-364.0466"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="1245.1972,-364.0728 1241.8044,-354.0359 1238.1976,-363.9979 1245.1972,-364.0728"/>
|
||||
</g>
|
||||
<!-- t16 -->
|
||||
<g id="node19" class="node">
|
||||
<title>t16</title>
|
||||
<polygon fill="#00cc66" stroke="#000000" points="1787.9202,-353.7367 1420.0798,-353.7367 1416.0798,-349.7367 1416.0798,-317.7367 1783.9202,-317.7367 1787.9202,-321.7367 1787.9202,-353.7367"/>
|
||||
<polyline fill="none" stroke="#000000" points="1783.9202,-349.7367 1416.0798,-349.7367 "/>
|
||||
<polyline fill="none" stroke="#000000" points="1783.9202,-349.7367 1783.9202,-317.7367 "/>
|
||||
<polyline fill="none" stroke="#000000" points="1783.9202,-349.7367 1787.9202,-353.7367 "/>
|
||||
<text text-anchor="middle" x="1602" y="-329.7367" font-family="Times,serif" font-size="20.00" fill="#000000">ocrmypdf.pipeline.render_hocr_debug_page</text>
|
||||
</g>
|
||||
<!-- t12->t16 -->
|
||||
<g id="edge27" class="edge">
|
||||
<title>t12->t16</title>
|
||||
<path fill="none" stroke="#0044a0" d="M1310.6044,-411.2411C1371.1859,-395.5436 1458.9787,-372.7954 1522.3491,-356.3753"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="1523.4996,-359.6928 1532.302,-353.7964 1521.7438,-352.9166 1523.4996,-359.6928"/>
|
||||
</g>
|
||||
<!-- t23 -->
|
||||
<g id="node25" class="node">
|
||||
<title>t23</title>
|
||||
<polygon fill="#efa03b" stroke="#000000" points="677.7205,-323.5194 871,-296.2006 1064.2795,-323.5194 1064.0991,-367.7221 677.9009,-367.7221 677.7205,-323.5194"/>
|
||||
<polygon fill="none" stroke="#000000" points="673.7033,-320.0438 871,-292.1572 1068.2967,-320.0438 1068.0858,-371.7257 673.9142,-371.7257 673.7033,-320.0438"/>
|
||||
<text text-anchor="middle" x="871" y="-329.7367" font-family="Times,serif" font-size="20.00" fill="#000000">ocrmypdf.pipeline.merge_sidecars</text>
|
||||
</g>
|
||||
<!-- t12->t23 -->
|
||||
<g id="edge41" class="edge">
|
||||
<title>t12->t23</title>
|
||||
<path fill="none" stroke="#0044a0" d="M1169.6603,-411.2411C1128.034,-400.7175 1073.8731,-387.0251 1023.1228,-374.1949"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="1023.9099,-370.7839 1013.3571,-371.726 1022.1942,-377.5703 1023.9099,-370.7839"/>
|
||||
<!-- t1->t17 -->
|
||||
<g id="edge25" class="edge">
|
||||
<title>t1->t17</title>
|
||||
<path fill="none" stroke="#0044a0" d="M358.0758,-1123.6249C741.2403,-1103.0857 1635,-1048.3676 1635,-984.84 1635,-984.84 1635,-984.84 1635,-480.3564 1635,-418.9654 1656.1556,-391.3705 1618,-343.2769 1605.0678,-326.9763 1561.5297,-310.7971 1520.7529,-298.5915"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="1521.5972,-295.1917 1511.0163,-295.7346 1519.6263,-301.9086 1521.5972,-295.1917"/>
|
||||
</g>
|
||||
<!-- t18 -->
|
||||
<g id="node18" class="node">
|
||||
<title>t18</title>
|
||||
<polygon fill="#00cc66" stroke="#000000" points="1096.2989,-270.197 805.7011,-270.197 801.7011,-266.197 801.7011,-234.197 1092.2989,-234.197 1096.2989,-238.197 1096.2989,-270.197"/>
|
||||
<polyline fill="none" stroke="#000000" points="1092.2989,-266.197 801.7011,-266.197 "/>
|
||||
<polyline fill="none" stroke="#000000" points="1092.2989,-266.197 1092.2989,-234.197 "/>
|
||||
<polyline fill="none" stroke="#000000" points="1092.2989,-266.197 1096.2989,-270.197 "/>
|
||||
<text text-anchor="middle" x="949" y="-246.197" font-family="Times,serif" font-size="20.00" fill="#000000">ocrmypdf.pipeline.combine_layers</text>
|
||||
<polygon fill="#efa03b" stroke="#000000" points="403.5156,-404.8166 40.4844,-404.8166 36.4844,-400.8166 36.4844,-368.8166 399.5156,-368.8166 403.5156,-372.8166 403.5156,-404.8166"/>
|
||||
<polyline fill="none" stroke="#000000" points="399.5156,-400.8166 36.4844,-400.8166 "/>
|
||||
<polyline fill="none" stroke="#000000" points="399.5156,-400.8166 399.5156,-368.8166 "/>
|
||||
<polyline fill="none" stroke="#000000" points="399.5156,-400.8166 403.5156,-404.8166 "/>
|
||||
<text text-anchor="middle" x="220" y="-380.8166" font-family="Times,serif" font-size="20.00" fill="#000000">ocrmypdf.pipeline.generate_postscript_stub</text>
|
||||
</g>
|
||||
<!-- t15->t18 -->
|
||||
<g id="edge24" class="edge">
|
||||
<title>t15->t18</title>
|
||||
<path fill="none" stroke="#0044a0" d="M1178.7841,-317.7127C1132.7483,-304.587 1070.2826,-286.7769 1022.1,-273.0391"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="1022.9314,-269.6368 1012.355,-270.2607 1021.012,-276.3685 1022.9314,-269.6368"/>
|
||||
</g>
|
||||
<!-- t17->t18 -->
|
||||
<g id="edge23" class="edge">
|
||||
<title>t17->t18</title>
|
||||
<path fill="none" stroke="#0044a0" d="M742.5275,-411.2563C707.334,-403.5731 675.8911,-393.1114 665,-379.2765 641.0609,-348.8665 639.6136,-321.4095 665,-292.197 674.2623,-281.5388 731.0775,-272.4723 791.4852,-265.6796"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="792.0426,-269.1394 801.5997,-264.5664 791.2768,-262.1814 792.0426,-269.1394"/>
|
||||
</g>
|
||||
<!-- t17->t23 -->
|
||||
<g id="edge40" class="edge">
|
||||
<title>t17->t23</title>
|
||||
<path fill="none" stroke="#0044a0" d="M871.8072,-411.2411C871.7178,-402.8795 871.607,-392.5175 871.4965,-382.1769"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="874.9936,-381.8803 871.3868,-371.9184 867.994,-381.9552 874.9936,-381.8803"/>
|
||||
</g>
|
||||
<!-- t13->t14 -->
|
||||
<g id="edge21" class="edge">
|
||||
<title>t13->t14</title>
|
||||
<path fill="none" stroke="#0044a0" d="M1197.8658,-493.3637C1134.2194,-488.8196 1063.0826,-483.7842 998,-479.2765 852.6863,-469.2117 816.2452,-468.2864 671,-457.2765 618.6842,-453.3108 561.7627,-448.6403 510.0667,-444.2623"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="510.0603,-440.7492 499.8001,-443.3908 509.4682,-447.7241 510.0603,-440.7492"/>
|
||||
</g>
|
||||
<!-- t13->t16 -->
|
||||
<!-- t1->t18 -->
|
||||
<g id="edge26" class="edge">
|
||||
<title>t13->t16</title>
|
||||
<path fill="none" stroke="#0044a0" d="M1400.9154,-479.7728C1403.5471,-472.4624 1406.3842,-464.5715 1409,-457.2765 1417.9204,-432.3989 1410.4916,-420.142 1429,-401.2765 1448.8322,-381.0616 1475.5255,-366.9968 1502.0009,-357.2438"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="1503.4471,-360.446 1511.7321,-353.8422 1501.1373,-353.8381 1503.4471,-360.446"/>
|
||||
<title>t1->t18</title>
|
||||
<path fill="none" stroke="#0044a0" d="M176.8904,-1114.035C173.5882,-1107.3677 170.2063,-1099.6531 168,-1092.2885 154.1622,-1046.0988 153,-1033.058 153,-984.84 153,-984.84 153,-984.84 153,-558.3564 153,-510.7759 149.2243,-496.0756 168,-452.3564 174.2264,-437.8581 184.5322,-423.8627 194.334,-412.554"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="197.2096,-414.5921 201.3166,-404.8257 192.0155,-409.8993 197.2096,-414.5921"/>
|
||||
</g>
|
||||
<!-- t13->t19 -->
|
||||
<!-- t20 -->
|
||||
<g id="node19" class="node">
|
||||
<title>t20</title>
|
||||
<polygon fill="#efa03b" stroke="#000000" points="27.189,-156.4288 220,-129.1099 412.811,-156.4288 412.631,-200.6317 27.369,-200.6317 27.189,-156.4288"/>
|
||||
<polygon fill="none" stroke="#000000" points="23.1715,-152.9543 220,-125.0662 416.8285,-152.9543 416.6181,-204.6353 23.3819,-204.6353 23.1715,-152.9543"/>
|
||||
<text text-anchor="middle" x="220" y="-162.6461" font-family="Times,serif" font-size="20.00" fill="#000000">ocrmypdf.pipeline.metadata_fixup</text>
|
||||
</g>
|
||||
<!-- t1->t20 -->
|
||||
<g id="edge29" class="edge">
|
||||
<title>t13->t19</title>
|
||||
<path fill="none" stroke="#0044a0" d="M1464.2273,-484.6109C1499.1449,-473.803 1540.6621,-460.9525 1574.8943,-450.3568"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="1576.2874,-453.5895 1584.8054,-447.2891 1574.2176,-446.9025 1576.2874,-453.5895"/>
|
||||
<title>t1->t20</title>
|
||||
<path fill="none" stroke="#0044a0" d="M138.2848,-1114.2775C88.0501,-1092.1806 17,-1049.4987 17,-984.84 17,-984.84 17,-984.84 17,-386.8166 17,-309.4325 86.2972,-247.9982 143.9157,-210.2921"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="146.1535,-213.0152 152.6859,-204.6739 142.3775,-207.1209 146.1535,-213.0152"/>
|
||||
</g>
|
||||
<!-- t14->t18 -->
|
||||
<g id="edge25" class="edge">
|
||||
<title>t14->t18</title>
|
||||
<path fill="none" stroke="#0044a0" d="M382.1114,-404.8262C439.5148,-373.4449 545.0952,-319.7698 642,-292.197 689.7605,-278.6074 743.0314,-269.6169 791.6153,-263.6758"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="792.0696,-267.1466 801.5866,-262.4909 791.2434,-260.1955 792.0696,-267.1466"/>
|
||||
<!-- t21 -->
|
||||
<g id="node22" class="node">
|
||||
<title>t21</title>
|
||||
<polygon fill="#efa03b" stroke="#000000" points="255.1753,-265.5126 436,-238.1903 616.8247,-265.5126 616.6559,-309.721 255.3441,-309.721 255.1753,-265.5126"/>
|
||||
<polygon fill="none" stroke="#000000" points="251.1584,-262.0699 436,-234.1406 620.8416,-262.0699 620.6444,-313.7252 251.3556,-313.7252 251.1584,-262.0699"/>
|
||||
<text text-anchor="middle" x="436" y="-271.7315" font-family="Times,serif" font-size="20.00" fill="#000000">ocrmypdf.pipeline.merge_pages</text>
|
||||
</g>
|
||||
<!-- t18->t22 -->
|
||||
<g id="edge35" class="edge">
|
||||
<title>t18->t22</title>
|
||||
<path fill="none" stroke="#0044a0" d="M1096.4053,-236.2861C1218.8926,-223.0648 1394.1726,-204.1451 1528.216,-189.6765"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="1528.6338,-193.1519 1538.2004,-188.5988 1527.8825,-186.1923 1528.6338,-193.1519"/>
|
||||
<!-- t1->t21 -->
|
||||
<g id="edge36" class="edge">
|
||||
<title>t1->t21</title>
|
||||
<path fill="none" stroke="#0044a0" d="M184.1669,-1114.1813C180.076,-1086.421 173,-1031.6392 173,-984.84 173,-984.84 173,-984.84 173,-558.3564 173,-534.7746 161.7149,-469.412 178,-452.3564 214.2214,-414.4213 374.884,-466.3873 413,-430.3564 416.9098,-426.6604 424.9678,-367.5181 430.4294,-323.9622"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="433.9293,-324.1792 431.6894,-313.8239 426.9827,-323.3158 433.9293,-324.1792"/>
|
||||
</g>
|
||||
<!-- t3 -->
|
||||
<g id="node4" class="node">
|
||||
<title>t3</title>
|
||||
<polygon fill="#efa03b" stroke="#000000" points="671.5194,-992.3982 441,-1009.2989 210.4806,-992.3982 210.6958,-965.0524 671.3042,-965.0524 671.5194,-992.3982"/>
|
||||
<polygon fill="none" stroke="#000000" points="675.5491,-996.1112 441,-1013.3073 206.4509,-996.1112 206.7267,-961.0547 675.2733,-961.0547 675.5491,-996.1112"/>
|
||||
<text text-anchor="middle" x="441" y="-978.84" font-family="Times,serif" font-size="20.00" fill="#000000">ocrmypdf.pipeline.ocr_or_skip</text>
|
||||
</g>
|
||||
<!-- t2->t3 -->
|
||||
<g id="edge3" class="edge">
|
||||
<title>t2->t3</title>
|
||||
<path fill="none" stroke="#0044a0" d="M441,-1039.6846C441,-1034.6543 441,-1029.2298 441,-1023.8181"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="444.5001,-1023.5335 441,-1013.5335 437.5001,-1023.5336 444.5001,-1023.5335"/>
|
||||
</g>
|
||||
<!-- t4 -->
|
||||
<g id="node5" class="node">
|
||||
<title>t4</title>
|
||||
<polygon fill="#efa03b" stroke="#000000" points="550.3166,-934.3564 243.6834,-934.3564 239.6834,-930.3564 239.6834,-898.3564 546.3166,-898.3564 550.3166,-902.3564 550.3166,-934.3564"/>
|
||||
<polyline fill="none" stroke="#000000" points="546.3166,-930.3564 239.6834,-930.3564 "/>
|
||||
<polyline fill="none" stroke="#000000" points="546.3166,-930.3564 546.3166,-898.3564 "/>
|
||||
<polyline fill="none" stroke="#000000" points="546.3166,-930.3564 550.3166,-934.3564 "/>
|
||||
<text text-anchor="middle" x="395" y="-910.3564" font-family="Times,serif" font-size="20.00" fill="#000000">ocrmypdf.pipeline.rasterize_preview</text>
|
||||
</g>
|
||||
<!-- t3->t4 -->
|
||||
<g id="edge4" class="edge">
|
||||
<title>t3->t4</title>
|
||||
<path fill="none" stroke="#0044a0" d="M424.9114,-960.8877C421.0122,-955.0827 416.8209,-948.8427 412.846,-942.925"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="415.7179,-940.9235 407.2365,-934.5738 409.907,-944.8266 415.7179,-940.9235"/>
|
||||
</g>
|
||||
<!-- t5 -->
|
||||
<g id="node6" class="node">
|
||||
<title>t5</title>
|
||||
<polygon fill="#efa03b" stroke="#000000" points="697.0824,-876.3564 438.9176,-876.3564 434.9176,-872.3564 434.9176,-840.3564 693.0824,-840.3564 697.0824,-844.3564 697.0824,-876.3564"/>
|
||||
<polyline fill="none" stroke="#000000" points="693.0824,-872.3564 434.9176,-872.3564 "/>
|
||||
<polyline fill="none" stroke="#000000" points="693.0824,-872.3564 693.0824,-840.3564 "/>
|
||||
<polyline fill="none" stroke="#000000" points="693.0824,-872.3564 697.0824,-876.3564 "/>
|
||||
<text text-anchor="middle" x="566" y="-852.3564" font-family="Times,serif" font-size="20.00" fill="#000000">ocrmypdf.pipeline.orient_page</text>
|
||||
</g>
|
||||
<!-- t3->t5 -->
|
||||
<g id="edge6" class="edge">
|
||||
<title>t3->t5</title>
|
||||
<path fill="none" stroke="#0044a0" d="M527.5821,-961.0486C539.7567,-954.2462 550.87,-945.5258 559,-934.3564 568.8835,-920.7777 570.696,-902.0144 570.0069,-886.5291"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="573.4858,-886.1249 569.1985,-876.4363 566.5081,-886.6838 573.4858,-886.1249"/>
|
||||
</g>
|
||||
<!-- t4->t5 -->
|
||||
<g id="edge5" class="edge">
|
||||
<title>t4->t5</title>
|
||||
<path fill="none" stroke="#0044a0" d="M448.1681,-898.3227C465.5399,-892.4305 485,-885.83 503.0225,-879.7171"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="504.5189,-882.9055 512.8648,-876.3788 502.2704,-876.2764 504.5189,-882.9055"/>
|
||||
</g>
|
||||
<!-- t6 -->
|
||||
<g id="node7" class="node">
|
||||
<title>t6</title>
|
||||
<polygon fill="#efa03b" stroke="#000000" points="800.7325,-818.3564 425.2675,-818.3564 421.2675,-814.3564 421.2675,-782.3564 796.7325,-782.3564 800.7325,-786.3564 800.7325,-818.3564"/>
|
||||
<polyline fill="none" stroke="#000000" points="796.7325,-814.3564 421.2675,-814.3564 "/>
|
||||
<polyline fill="none" stroke="#000000" points="796.7325,-814.3564 796.7325,-782.3564 "/>
|
||||
<polyline fill="none" stroke="#000000" points="796.7325,-814.3564 800.7325,-818.3564 "/>
|
||||
<text text-anchor="middle" x="611" y="-794.3564" font-family="Times,serif" font-size="20.00" fill="#000000">ocrmypdf.pipeline.rasterize_with_ghostscript</text>
|
||||
</g>
|
||||
<!-- t5->t6 -->
|
||||
<g id="edge7" class="edge">
|
||||
<title>t5->t6</title>
|
||||
<path fill="none" stroke="#0044a0" d="M579.9916,-840.3227C583.3457,-835.9997 586.9956,-831.2954 590.5748,-826.6821"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="593.4974,-828.625 596.8621,-818.5786 587.9667,-824.334 593.4974,-828.625"/>
|
||||
</g>
|
||||
<!-- t16 -->
|
||||
<g id="node14" class="node">
|
||||
<title>t16</title>
|
||||
<polygon fill="#ff69b4" stroke="#000000" points="596.2696,-498.3564 215.7304,-498.3564 211.7304,-494.3564 211.7304,-462.3564 592.2696,-462.3564 596.2696,-466.3564 596.2696,-498.3564"/>
|
||||
<polyline fill="none" stroke="#000000" points="592.2696,-494.3564 211.7304,-494.3564 "/>
|
||||
<polyline fill="none" stroke="#000000" points="592.2696,-494.3564 592.2696,-462.3564 "/>
|
||||
<polyline fill="none" stroke="#000000" points="592.2696,-494.3564 596.2696,-498.3564 "/>
|
||||
<text text-anchor="middle" x="404" y="-474.3564" font-family="Times,serif" font-size="20.00" fill="#000000">ocrmypdf.pipeline.ocr_tesseract_textonly_pdf</text>
|
||||
</g>
|
||||
<!-- t5->t16 -->
|
||||
<g id="edge15" class="edge">
|
||||
<title>t5->t16</title>
|
||||
<path fill="none" stroke="#0044a0" d="M436.0597,-840.313C335.4079,-822.5625 213,-790.9799 213,-742.3564 213,-742.3564 213,-742.3564 213,-626.3564 213,-582.8621 202.6196,-562.4273 232,-530.3564 242.3778,-519.0283 270.3671,-508.9962 300.8272,-500.9473"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="301.9614,-504.2701 310.7845,-498.4046 300.2294,-497.4877 301.9614,-504.2701"/>
|
||||
</g>
|
||||
<!-- t13 -->
|
||||
<g id="node16" class="node">
|
||||
<title>t13</title>
|
||||
<polygon fill="#00cc66" stroke="#000000" points="1293,-508.3564 970.4141,-480.3564 1293,-452.3564 1615.5859,-480.3564 1293,-508.3564"/>
|
||||
<text text-anchor="middle" x="1293" y="-474.3564" font-family="Times,serif" font-size="20.00" fill="#000000">ocrmypdf.pipeline.select_image_layer</text>
|
||||
</g>
|
||||
<!-- t5->t13 -->
|
||||
<g id="edge21" class="edge">
|
||||
<title>t5->t13</title>
|
||||
<path fill="none" stroke="#0044a0" d="M697.4498,-852.2263C921.648,-839.9951 1354,-807.9097 1354,-742.3564 1354,-742.3564 1354,-742.3564 1354,-626.3564 1354,-583.4588 1362.6414,-568.9918 1344,-530.3564 1341.0314,-524.2037 1336.9784,-518.3835 1332.4454,-513.0377"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="1334.8902,-510.5263 1325.5307,-505.5613 1329.7511,-515.2793 1334.8902,-510.5263"/>
|
||||
</g>
|
||||
<!-- t19 -->
|
||||
<g id="node21" class="node">
|
||||
<title>t19</title>
|
||||
<polygon fill="#efa03b" stroke="#000000" points="502.7716,-760.3564 257.2284,-760.3564 253.2284,-756.3564 253.2284,-724.3564 498.7716,-724.3564 502.7716,-728.3564 502.7716,-760.3564"/>
|
||||
<polyline fill="none" stroke="#000000" points="498.7716,-756.3564 253.2284,-756.3564 "/>
|
||||
<polyline fill="none" stroke="#000000" points="498.7716,-756.3564 498.7716,-724.3564 "/>
|
||||
<polyline fill="none" stroke="#000000" points="498.7716,-756.3564 502.7716,-760.3564 "/>
|
||||
<text text-anchor="middle" x="378" y="-736.3564" font-family="Times,serif" font-size="20.00" fill="#000000">ocrmypdf.pipeline.skip_page</text>
|
||||
</g>
|
||||
<!-- t5->t19 -->
|
||||
<g id="edge32" class="edge">
|
||||
<title>t5->t19</title>
|
||||
<path fill="none" stroke="#0044a0" d="M469.724,-840.3464C444.565,-834.1182 421.7539,-826.6075 412,-818.3564 397.6062,-806.1802 389.0537,-786.6831 384.1266,-770.4858"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="387.3923,-769.1572 381.3874,-760.4284 380.6384,-770.9967 387.3923,-769.1572"/>
|
||||
</g>
|
||||
<!-- t7 -->
|
||||
<g id="node8" class="node">
|
||||
<title>t7</title>
|
||||
<polygon fill="#efa03b" stroke="#000000" points="1058.0801,-760.3564 633.9199,-760.3564 629.9199,-756.3564 629.9199,-724.3564 1054.0801,-724.3564 1058.0801,-728.3564 1058.0801,-760.3564"/>
|
||||
<polyline fill="none" stroke="#000000" points="1054.0801,-756.3564 629.9199,-756.3564 "/>
|
||||
<polyline fill="none" stroke="#000000" points="1054.0801,-756.3564 1054.0801,-724.3564 "/>
|
||||
<polyline fill="none" stroke="#000000" points="1054.0801,-756.3564 1058.0801,-760.3564 "/>
|
||||
<text text-anchor="middle" x="844" y="-736.3564" font-family="Times,serif" font-size="20.00" fill="#000000">ocrmypdf.pipeline.preprocess_remove_background</text>
|
||||
</g>
|
||||
<!-- t6->t7 -->
|
||||
<g id="edge8" class="edge">
|
||||
<title>t6->t7</title>
|
||||
<path fill="none" stroke="#0044a0" d="M683.4454,-782.3227C708.2116,-776.1578 736.0929,-769.2174 761.5862,-762.8714"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="762.741,-766.1908 771.5994,-760.3788 761.05,-759.3981 762.741,-766.1908"/>
|
||||
</g>
|
||||
<!-- t12 -->
|
||||
<g id="node15" class="node">
|
||||
<title>t12</title>
|
||||
<polygon fill="#efa03b" stroke="#000000" points="951,-586.3564 567.1874,-558.3564 951,-530.3564 1334.8126,-558.3564 951,-586.3564"/>
|
||||
<text text-anchor="middle" x="951" y="-552.3564" font-family="Times,serif" font-size="20.00" fill="#000000">ocrmypdf.pipeline.select_visible_page_image</text>
|
||||
</g>
|
||||
<!-- t6->t12 -->
|
||||
<g id="edge19" class="edge">
|
||||
<title>t6->t12</title>
|
||||
<path fill="none" stroke="#0044a0" d="M606.5536,-782.0806C600.8758,-753.9975 594.6738,-699.964 621,-666.3564 648.4442,-631.3215 754.0103,-600.7729 838.4514,-581.1905"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="839.4102,-584.5615 848.3759,-578.9164 837.8468,-577.7384 839.4102,-584.5615"/>
|
||||
</g>
|
||||
<!-- t8 -->
|
||||
<g id="node9" class="node">
|
||||
<title>t8</title>
|
||||
<polygon fill="#efa03b" stroke="#000000" points="953.6759,-702.3564 634.3241,-702.3564 630.3241,-698.3564 630.3241,-666.3564 949.6759,-666.3564 953.6759,-670.3564 953.6759,-702.3564"/>
|
||||
<polyline fill="none" stroke="#000000" points="949.6759,-698.3564 630.3241,-698.3564 "/>
|
||||
<polyline fill="none" stroke="#000000" points="949.6759,-698.3564 949.6759,-666.3564 "/>
|
||||
<polyline fill="none" stroke="#000000" points="949.6759,-698.3564 953.6759,-702.3564 "/>
|
||||
<text text-anchor="middle" x="792" y="-678.3564" font-family="Times,serif" font-size="20.00" fill="#000000">ocrmypdf.pipeline.preprocess_deskew</text>
|
||||
</g>
|
||||
<!-- t7->t8 -->
|
||||
<g id="edge9" class="edge">
|
||||
<title>t7->t8</title>
|
||||
<path fill="none" stroke="#0044a0" d="M827.8319,-724.3227C823.8592,-719.8916 819.5274,-715.06 815.2924,-710.3363"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="817.6187,-707.6879 808.3372,-702.5786 812.4067,-712.3607 817.6187,-707.6879"/>
|
||||
</g>
|
||||
<!-- t7->t12 -->
|
||||
<g id="edge18" class="edge">
|
||||
<title>t7->t12</title>
|
||||
<path fill="none" stroke="#0044a0" d="M920.4182,-724.2656C939.9479,-718.0339 957.1161,-710.5415 963,-702.3564 985.1975,-671.477 976.8981,-626.6742 966.4363,-595.266"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="969.6404,-593.8291 962.9995,-585.5738 963.0428,-596.1686 969.6404,-593.8291"/>
|
||||
</g>
|
||||
<!-- t9 -->
|
||||
<g id="node10" class="node">
|
||||
<title>t9</title>
|
||||
<polygon fill="#efa03b" stroke="#000000" points="591.8828,-644.3564 290.1172,-644.3564 286.1172,-640.3564 286.1172,-608.3564 587.8828,-608.3564 591.8828,-612.3564 591.8828,-644.3564"/>
|
||||
<polyline fill="none" stroke="#000000" points="587.8828,-640.3564 286.1172,-640.3564 "/>
|
||||
<polyline fill="none" stroke="#000000" points="587.8828,-640.3564 587.8828,-608.3564 "/>
|
||||
<polyline fill="none" stroke="#000000" points="587.8828,-640.3564 591.8828,-644.3564 "/>
|
||||
<text text-anchor="middle" x="439" y="-620.3564" font-family="Times,serif" font-size="20.00" fill="#000000">ocrmypdf.pipeline.preprocess_clean</text>
|
||||
</g>
|
||||
<!-- t8->t9 -->
|
||||
<g id="edge10" class="edge">
|
||||
<title>t8->t9</title>
|
||||
<path fill="none" stroke="#0044a0" d="M682.2436,-666.3227C643.0621,-659.885 598.7342,-652.6016 558.7536,-646.0326"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="559.1237,-642.5465 548.6885,-644.3788 557.9887,-649.4539 559.1237,-642.5465"/>
|
||||
</g>
|
||||
<!-- t8->t12 -->
|
||||
<g id="edge17" class="edge">
|
||||
<title>t8->t12</title>
|
||||
<path fill="none" stroke="#0044a0" d="M814.7808,-666.3036C839.7684,-646.5022 880.3545,-614.3396 910.7367,-590.2631"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="912.9552,-592.9709 918.6188,-584.0169 908.6076,-587.4846 912.9552,-592.9709"/>
|
||||
</g>
|
||||
<!-- t10 -->
|
||||
<g id="node11" class="node">
|
||||
<title>t10</title>
|
||||
<polygon fill="#efa03b" stroke="#000000" points="549.1094,-576.3564 244.8906,-576.3564 240.8906,-572.3564 240.8906,-540.3564 545.1094,-540.3564 549.1094,-544.3564 549.1094,-576.3564"/>
|
||||
<polyline fill="none" stroke="#000000" points="545.1094,-572.3564 240.8906,-572.3564 "/>
|
||||
<polyline fill="none" stroke="#000000" points="545.1094,-572.3564 545.1094,-540.3564 "/>
|
||||
<polyline fill="none" stroke="#000000" points="545.1094,-572.3564 549.1094,-576.3564 "/>
|
||||
<text text-anchor="middle" x="395" y="-552.3564" font-family="Times,serif" font-size="20.00" fill="#000000">ocrmypdf.pipeline.select_ocr_image</text>
|
||||
</g>
|
||||
<!-- t9->t10 -->
|
||||
<g id="edge11" class="edge">
|
||||
<title>t9->t10</title>
|
||||
<path fill="none" stroke="#0044a0" d="M427.2088,-608.1336C422.6713,-601.1211 417.3894,-592.9581 412.4313,-585.2957"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="415.1948,-583.1237 406.8237,-576.6294 409.3178,-586.9265 415.1948,-583.1237"/>
|
||||
</g>
|
||||
<!-- t9->t12 -->
|
||||
<g id="edge16" class="edge">
|
||||
<title>t9->t12</title>
|
||||
<path fill="none" stroke="#0044a0" d="M574.8624,-608.3121C645.599,-598.9174 732.1622,-587.4207 804.4069,-577.8258"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="805.0506,-581.2711 814.5027,-576.4849 804.1289,-574.332 805.0506,-581.2711"/>
|
||||
</g>
|
||||
<!-- t11 -->
|
||||
<g id="node12" class="node">
|
||||
<title>t11</title>
|
||||
<polygon fill="#00cc66" stroke="#000000" points="932.0996,-498.3564 617.9004,-498.3564 613.9004,-494.3564 613.9004,-462.3564 928.0996,-462.3564 932.0996,-466.3564 932.0996,-498.3564"/>
|
||||
<polyline fill="none" stroke="#000000" points="928.0996,-494.3564 613.9004,-494.3564 "/>
|
||||
<polyline fill="none" stroke="#000000" points="928.0996,-494.3564 928.0996,-462.3564 "/>
|
||||
<polyline fill="none" stroke="#000000" points="928.0996,-494.3564 932.0996,-498.3564 "/>
|
||||
<text text-anchor="middle" x="773" y="-474.3564" font-family="Times,serif" font-size="20.00" fill="#000000">ocrmypdf.pipeline.ocr_tesseract_hocr</text>
|
||||
</g>
|
||||
<!-- t10->t11 -->
|
||||
<g id="edge12" class="edge">
|
||||
<title>t10->t11</title>
|
||||
<path fill="none" stroke="#0044a0" d="M482.6709,-540.2655C540.4141,-528.3503 616.0731,-512.7381 675.6782,-500.4386"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="676.3914,-503.8653 685.4778,-498.4165 674.9767,-497.0097 676.3914,-503.8653"/>
|
||||
</g>
|
||||
<!-- t10->t16 -->
|
||||
<g id="edge14" class="edge">
|
||||
<title>t10->t16</title>
|
||||
<path fill="none" stroke="#0044a0" d="M397.0874,-540.2655C398.1701,-530.8822 399.5173,-519.2062 400.7341,-508.6609"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="404.2233,-508.9546 401.8927,-498.6193 397.2695,-508.1521 404.2233,-508.9546"/>
|
||||
</g>
|
||||
<!-- t14 -->
|
||||
<g id="node13" class="node">
|
||||
<title>t14</title>
|
||||
<polygon fill="#00cc66" stroke="#000000" points="1596.0433,-404.8166 1287.9567,-404.8166 1283.9567,-400.8166 1283.9567,-368.8166 1592.0433,-368.8166 1596.0433,-372.8166 1596.0433,-404.8166"/>
|
||||
<polyline fill="none" stroke="#000000" points="1592.0433,-400.8166 1283.9567,-400.8166 "/>
|
||||
<polyline fill="none" stroke="#000000" points="1592.0433,-400.8166 1592.0433,-368.8166 "/>
|
||||
<polyline fill="none" stroke="#000000" points="1592.0433,-400.8166 1596.0433,-404.8166 "/>
|
||||
<text text-anchor="middle" x="1440" y="-380.8166" font-family="Times,serif" font-size="20.00" fill="#000000">ocrmypdf.pipeline.render_hocr_page</text>
|
||||
</g>
|
||||
<!-- t11->t14 -->
|
||||
<g id="edge13" class="edge">
|
||||
<title>t11->t14</title>
|
||||
<path fill="none" stroke="#0044a0" d="M884.1541,-462.3371C909.372,-458.6572 936.0694,-455.0821 961,-452.3564 1100.069,-437.1517 1136.9788,-453.1931 1275,-430.3564 1307.5629,-424.9685 1343.0324,-416.0244 1372.7118,-407.6396"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="1373.8931,-410.9421 1382.5425,-404.8235 1371.9653,-404.2128 1373.8931,-410.9421"/>
|
||||
</g>
|
||||
<!-- t15 -->
|
||||
<g id="node20" class="node">
|
||||
<title>t15</title>
|
||||
<polygon fill="#00cc66" stroke="#000000" points="1265.9202,-404.8166 898.0798,-404.8166 894.0798,-400.8166 894.0798,-368.8166 1261.9202,-368.8166 1265.9202,-372.8166 1265.9202,-404.8166"/>
|
||||
<polyline fill="none" stroke="#000000" points="1261.9202,-400.8166 894.0798,-400.8166 "/>
|
||||
<polyline fill="none" stroke="#000000" points="1261.9202,-400.8166 1261.9202,-368.8166 "/>
|
||||
<polyline fill="none" stroke="#000000" points="1261.9202,-400.8166 1265.9202,-404.8166 "/>
|
||||
<text text-anchor="middle" x="1080" y="-380.8166" font-family="Times,serif" font-size="20.00" fill="#000000">ocrmypdf.pipeline.render_hocr_debug_page</text>
|
||||
</g>
|
||||
<!-- t11->t15 -->
|
||||
<g id="edge31" class="edge">
|
||||
<title>t11->t15</title>
|
||||
<path fill="none" stroke="#0044a0" d="M832.1927,-462.321C883.2881,-446.7527 957.1451,-424.2492 1010.9288,-407.8619"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="1012.1821,-411.139 1020.7277,-404.8763 1010.1418,-404.4429 1012.1821,-411.139"/>
|
||||
</g>
|
||||
<!-- t22 -->
|
||||
<g id="node24" class="node">
|
||||
<title>t22</title>
|
||||
<polygon fill="#efa03b" stroke="#000000" points="485.7205,-374.5993 679,-347.2805 872.2795,-374.5993 872.0991,-418.802 485.9009,-418.802 485.7205,-374.5993"/>
|
||||
<polygon fill="none" stroke="#000000" points="481.7033,-371.1237 679,-343.2371 876.2967,-371.1237 876.0858,-422.8056 481.9142,-422.8056 481.7033,-371.1237"/>
|
||||
<text text-anchor="middle" x="679" y="-380.8166" font-family="Times,serif" font-size="20.00" fill="#000000">ocrmypdf.pipeline.merge_sidecars</text>
|
||||
</g>
|
||||
<!-- t11->t22 -->
|
||||
<g id="edge40" class="edge">
|
||||
<title>t11->t22</title>
|
||||
<path fill="none" stroke="#0044a0" d="M754.8759,-462.321C745.6854,-453.1755 734.09,-441.6369 722.7483,-430.3507"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="724.9169,-427.571 715.3597,-422.9982 719.9793,-432.5329 724.9169,-427.571"/>
|
||||
</g>
|
||||
<!-- t14->t17 -->
|
||||
<g id="edge23" class="edge">
|
||||
<title>t14->t17</title>
|
||||
<path fill="none" stroke="#0044a0" d="M1440,-368.8034C1440,-351.6708 1440,-325.6387 1440,-305.7879"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="1443.5001,-305.7714 1440,-295.7714 1436.5001,-305.7715 1443.5001,-305.7714"/>
|
||||
</g>
|
||||
<!-- t16->t17 -->
|
||||
<g id="edge22" class="edge">
|
||||
<title>t16->t17</title>
|
||||
<path fill="none" stroke="#0044a0" d="M423.7551,-461.9454C432.2558,-453.1007 441.6731,-441.9207 448,-430.3564 467.326,-395.032 441.6768,-368.5782 473,-343.2769 488.6951,-330.5991 1023.9298,-300.1127 1291.2068,-285.6289"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="1291.5864,-289.1135 1301.3826,-285.0782 1291.2081,-282.1238 1291.5864,-289.1135"/>
|
||||
</g>
|
||||
<!-- t16->t22 -->
|
||||
<g id="edge36" class="edge">
|
||||
<title>t16->t22</title>
|
||||
<path fill="none" stroke="#0044a0" d="M1615.1795,-317.5376C1633.1921,-292.6646 1666.0831,-247.2464 1690.8186,-213.0898"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="1693.887,-214.8201 1696.9176,-204.668 1688.2175,-210.7143 1693.887,-214.8201"/>
|
||||
</g>
|
||||
<!-- t21->t22 -->
|
||||
<g id="edge32" class="edge">
|
||||
<title>t21->t22</title>
|
||||
<path fill="none" stroke="#0044a0" d="M1784.6352,-673.2357C1820.6585,-654.4909 1867,-621.4113 1867,-575.2765 1867,-575.2765 1867,-575.2765 1867,-335.7367 1867,-284.2548 1827.2878,-240.7654 1789.5729,-211.0717"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="1791.6387,-208.2453 1781.5703,-204.9469 1787.3842,-213.8041 1791.6387,-208.2453"/>
|
||||
</g>
|
||||
<!-- t19->t22 -->
|
||||
<g id="edge34" class="edge">
|
||||
<title>t19->t22</title>
|
||||
<path fill="none" stroke="#0044a0" d="M1727.2665,-411.2415C1758.1998,-402.7736 1787.9112,-391.7833 1797,-379.2765 1833.7864,-328.6561 1794.7398,-258.1105 1760.8796,-212.906"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="1763.5015,-210.5735 1754.637,-204.7708 1757.9481,-214.835 1763.5015,-210.5735"/>
|
||||
</g>
|
||||
<!-- t19->t23 -->
|
||||
<g id="edge39" class="edge">
|
||||
<title>t19->t23</title>
|
||||
<path fill="none" stroke="#0044a0" d="M1502.6864,-411.2619C1471.854,-407.6457 1439.3509,-404.0959 1409,-401.2765 1261.7548,-387.598 1223.4889,-399.5063 1077,-379.2765 1064.9997,-377.6192 1052.6541,-375.6614 1040.2537,-373.5086"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="1040.7604,-370.0439 1030.3026,-371.7423 1039.537,-376.9362 1040.7604,-370.0439"/>
|
||||
<title>t16->t22</title>
|
||||
<path fill="none" stroke="#0044a0" d="M457.0228,-462.321C487.2537,-452.0381 526.3796,-438.7296 563.3441,-426.1563"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="564.8539,-429.3398 573.1941,-422.8059 562.5996,-422.7127 564.8539,-429.3398"/>
|
||||
</g>
|
||||
<!-- t20->t22 -->
|
||||
<g id="edge33" class="edge">
|
||||
<title>t20->t22</title>
|
||||
<path fill="none" stroke="#0044a0" d="M1765.9555,-984.2118C1819.0452,-964.5026 1887,-927.1837 1887,-865.2765 1887,-865.2765 1887,-865.2765 1887,-335.7367 1887,-284.8302 1870.8252,-269.3449 1834,-234.197 1824.9302,-225.5403 1814.5141,-217.5334 1803.7754,-210.3065"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="1805.4063,-207.1923 1795.1135,-204.6804 1801.5933,-213.0627 1805.4063,-207.1923"/>
|
||||
<!-- t12->t13 -->
|
||||
<g id="edge20" class="edge">
|
||||
<title>t12->t13</title>
|
||||
<path fill="none" stroke="#0044a0" d="M1044.453,-537.0425C1090.7976,-526.4727 1146.7836,-513.704 1193.8777,-502.9632"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="1194.8141,-506.3396 1203.7855,-500.7035 1193.2575,-499.5148 1194.8141,-506.3396"/>
|
||||
</g>
|
||||
<!-- t24 -->
|
||||
<g id="node24" class="node">
|
||||
<title>t24</title>
|
||||
<polygon fill="#efa03b" stroke="#000000" points="1556.0652,-47.332 1723,-20.0049 1889.9348,-47.332 1889.779,-91.5483 1556.221,-91.5483 1556.0652,-47.332"/>
|
||||
<polygon fill="none" stroke="#000000" points="1552.0538,-43.9306 1723,-15.9468 1893.9462,-43.9306 1893.7643,-95.5532 1552.2357,-95.5532 1552.0538,-43.9306"/>
|
||||
<text text-anchor="middle" x="1723" y="-53.5531" font-family="Times,serif" font-size="20.00" fill="#000000">ocrmypdf.pipeline.copy_final</text>
|
||||
<!-- t12->t15 -->
|
||||
<g id="edge30" class="edge">
|
||||
<title>t12->t15</title>
|
||||
<path fill="none" stroke="#0044a0" d="M946.8614,-530.3506C944.9813,-507.514 945.8899,-475.3833 961,-452.3564 973.5116,-433.2894 993.5483,-419.2759 1013.6406,-409.2185"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="1015.3,-412.3056 1022.8493,-404.872 1012.3121,-405.9753 1015.3,-412.3056"/>
|
||||
</g>
|
||||
<!-- t22->t24 -->
|
||||
<!-- t13->t17 -->
|
||||
<g id="edge24" class="edge">
|
||||
<title>t13->t17</title>
|
||||
<path fill="none" stroke="#0044a0" d="M1441.7799,-465.2142C1516.5232,-456.0688 1593.6782,-443.637 1605,-430.3564 1630.1081,-400.9042 1626.7355,-375.2989 1605,-343.2769 1591.383,-323.2155 1570.5555,-309.2942 1548.2343,-299.6337"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="1549.5211,-296.3788 1538.9381,-295.8785 1546.8992,-302.8693 1549.5211,-296.3788"/>
|
||||
</g>
|
||||
<!-- t17->t20 -->
|
||||
<g id="edge28" class="edge">
|
||||
<title>t17->t20</title>
|
||||
<path fill="none" stroke="#0044a0" d="M1301.2383,-265.3242C1087.4111,-246.205 676.0608,-209.4245 426.976,-187.1527"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="426.9845,-183.6396 416.7125,-186.235 426.361,-190.6118 426.9845,-183.6396"/>
|
||||
</g>
|
||||
<!-- t18->t20 -->
|
||||
<g id="edge27" class="edge">
|
||||
<title>t18->t20</title>
|
||||
<path fill="none" stroke="#0044a0" d="M220,-368.5908C220,-335.1014 220,-262.8295 220,-214.9755"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="223.5001,-214.8609 220,-204.861 216.5001,-214.861 223.5001,-214.8609"/>
|
||||
</g>
|
||||
<!-- t18->t21 -->
|
||||
<g id="edge34" class="edge">
|
||||
<title>t18->t21</title>
|
||||
<path fill="none" stroke="#0044a0" d="M255.668,-368.8034C282.7268,-355.1381 320.9967,-335.8109 355.3245,-318.4745"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="357.0735,-321.5123 364.422,-313.8801 353.9179,-315.2639 357.0735,-321.5123"/>
|
||||
</g>
|
||||
<!-- t23 -->
|
||||
<g id="node23" class="node">
|
||||
<title>t23</title>
|
||||
<polygon fill="#efa03b" stroke="#000000" points="161.0652,-47.332 328,-20.0049 494.9348,-47.332 494.779,-91.5483 161.221,-91.5483 161.0652,-47.332"/>
|
||||
<polygon fill="none" stroke="#000000" points="157.0538,-43.9306 328,-15.9468 498.9462,-43.9306 498.7643,-95.5532 157.2357,-95.5532 157.0538,-43.9306"/>
|
||||
<text text-anchor="middle" x="328" y="-53.5531" font-family="Times,serif" font-size="20.00" fill="#000000">ocrmypdf.pipeline.copy_final</text>
|
||||
</g>
|
||||
<!-- t20->t23 -->
|
||||
<g id="edge38" class="edge">
|
||||
<title>t22->t24</title>
|
||||
<path fill="none" stroke="#0044a0" d="M1723,-124.9057C1723,-118.6774 1723,-112.2535 1723,-105.9763"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="1726.5001,-105.6115 1723,-95.6115 1719.5001,-105.6115 1726.5001,-105.6115"/>
|
||||
<title>t20->t23</title>
|
||||
<path fill="none" stroke="#0044a0" d="M258.0768,-130.184C266.7982,-121.3743 276.1282,-111.9499 285.0392,-102.9487"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="287.5282,-105.4094 292.0763,-95.8404 282.5535,-100.4846 287.5282,-105.4094"/>
|
||||
</g>
|
||||
<!-- t15->t21 -->
|
||||
<g id="edge35" class="edge">
|
||||
<title>t15->t21</title>
|
||||
<path fill="none" stroke="#0044a0" d="M1005.2626,-368.7554C969.0731,-360.3621 924.9121,-350.6352 885,-343.2769 802.0689,-327.9873 709.9855,-314.0238 630.9044,-302.9508"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="631.2182,-299.4608 620.8306,-301.546 630.2513,-306.3937 631.2182,-299.4608"/>
|
||||
</g>
|
||||
<!-- t19->t21 -->
|
||||
<g id="edge33" class="edge">
|
||||
<title>t19->t21</title>
|
||||
<path fill="none" stroke="#0044a0" d="M351.7514,-724.1105C285.3188,-675.8338 121.4207,-542.3014 203,-452.3564 236.7512,-415.1441 393.2861,-466.6723 428,-430.3564 442.1375,-415.5664 442.1449,-363.2074 439.9161,-323.9111"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="443.3962,-323.4852 439.2735,-313.7255 436.4101,-323.926 443.3962,-323.4852"/>
|
||||
</g>
|
||||
<!-- t21->t23 -->
|
||||
<g id="edge37" class="edge">
|
||||
<title>t21->t23</title>
|
||||
<path fill="none" stroke="#0044a0" d="M442.8749,-235.0071C445.8908,-202.6847 445.1767,-158.3414 426,-125.1062 421.1919,-116.7732 414.846,-109.2905 407.6891,-102.6266"/>
|
||||
<polygon fill="#0044a0" stroke="#0044a0" points="409.5064,-99.5739 399.6391,-95.7155 404.9466,-104.8851 409.5064,-99.5739"/>
|
||||
</g>
|
||||
</g>
|
||||
</svg>
|
||||
|
||||
|
Before Width: | Height: | Size: 31 KiB After Width: | Height: | Size: 30 KiB |
@@ -316,7 +316,7 @@ def generate_pdf(*, input_image, skip_pdf, output_pdf, output_text,
|
||||
if pagesegmode is not None:
|
||||
args_tesseract.extend([psm(), str(pagesegmode)])
|
||||
|
||||
if text_only:
|
||||
if text_only and has_textonly_pdf():
|
||||
args_tesseract.extend(['-c', 'textonly_pdf=1'])
|
||||
|
||||
if user_words:
|
||||
|
||||
@@ -939,40 +939,6 @@ def weave_layers(
|
||||
# pdf_output.write(out)
|
||||
|
||||
|
||||
def ocr_tesseract_and_render_pdf(
|
||||
infiles,
|
||||
outfiles,
|
||||
log,
|
||||
context):
|
||||
options = context.get_options()
|
||||
input_image = next((ii for ii in infiles if ii.endswith('.image')), '')
|
||||
input_pdf = next((ii for ii in infiles if ii.endswith('.pdf')))
|
||||
output_pdf = next((ii for ii in outfiles if ii.endswith('.pdf')))
|
||||
output_text = next((ii for ii in outfiles if ii.endswith('.txt')))
|
||||
|
||||
if not input_image:
|
||||
# Skipping this page
|
||||
re_symlink(input_pdf, output_pdf, log)
|
||||
with open(output_text, 'w') as f:
|
||||
f.write('[skipped page]')
|
||||
return
|
||||
|
||||
tesseract.generate_pdf(
|
||||
input_image=input_image,
|
||||
skip_pdf=input_pdf,
|
||||
output_pdf=output_pdf,
|
||||
output_text=output_text,
|
||||
language=options.language,
|
||||
engine_mode=options.tesseract_oem,
|
||||
text_only=False,
|
||||
tessconfig=options.tesseract_config,
|
||||
timeout=options.tesseract_timeout,
|
||||
pagesegmode=options.tesseract_pagesegmode,
|
||||
user_words=options.user_words,
|
||||
user_patterns=options.user_patterns,
|
||||
log=log)
|
||||
|
||||
|
||||
def ocr_tesseract_textonly_pdf(
|
||||
infiles,
|
||||
outfiles,
|
||||
@@ -1025,9 +991,7 @@ def get_pdfmark(base_pdf, options):
|
||||
if options.subject:
|
||||
pdfmark['/Subject'] = options.subject
|
||||
|
||||
if options.pdf_renderer == 'tesseract':
|
||||
renderer_tag = 'OCR+PDF'
|
||||
elif options.pdf_renderer == 'sandwich':
|
||||
if options.pdf_renderer == 'sandwich':
|
||||
renderer_tag = 'OCR-PDF'
|
||||
else:
|
||||
renderer_tag = 'OCR'
|
||||
@@ -1438,7 +1402,8 @@ def build_pipeline(options, work_folder, log, context):
|
||||
os.path.join(work_folder, r'\1.text.txt')],
|
||||
extras=[log, context])
|
||||
task_ocr_tesseract_textonly_pdf.graphviz(fillcolor='"#ff69b4"')
|
||||
task_ocr_tesseract_textonly_pdf.active_if(options.pdf_renderer == 'sandwich')
|
||||
task_ocr_tesseract_textonly_pdf.active_if(
|
||||
options.pdf_renderer == 'tesseract' or options.pdf_renderer == 'sandwich')
|
||||
|
||||
task_weave_layers = main_pipeline.collate(
|
||||
task_func=weave_layers,
|
||||
@@ -1451,19 +1416,6 @@ def build_pipeline(options, work_folder, log, context):
|
||||
output=os.path.join(work_folder, r'layers.rendered.pdf'),
|
||||
extras=[log, context])
|
||||
task_weave_layers.graphviz(fillcolor='"#00cc66"')
|
||||
task_weave_layers.active_if(options.pdf_renderer == 'hocr' or
|
||||
options.pdf_renderer == 'sandwich')
|
||||
|
||||
# Tesseract OCR+PDF
|
||||
task_ocr_tesseract_and_render_pdf = main_pipeline.collate(
|
||||
task_func=ocr_tesseract_and_render_pdf,
|
||||
input=[task_select_visible_page_image, task_orient_page],
|
||||
filter=regex(r".*/(\d{6})(?:\.image|\.ocr\.oriented\.pdf)"),
|
||||
output=[os.path.join(work_folder, r'\1.rendered.pdf'),
|
||||
os.path.join(work_folder, r'\1.rendered.txt')],
|
||||
extras=[log, context])
|
||||
task_ocr_tesseract_and_render_pdf.graphviz(fillcolor='"#66ccff"')
|
||||
task_ocr_tesseract_and_render_pdf.active_if(options.pdf_renderer == 'tesseract')
|
||||
|
||||
# PDF/A
|
||||
task_generate_postscript_stub = main_pipeline.transform(
|
||||
@@ -1498,7 +1450,6 @@ def build_pipeline(options, work_folder, log, context):
|
||||
input=[task_repair_and_parse_pdf,
|
||||
task_render_hocr_debug_page,
|
||||
task_skip_page,
|
||||
task_ocr_tesseract_and_render_pdf,
|
||||
task_generate_postscript_stub],
|
||||
output=os.path.join(work_folder, 'merged.pdf'),
|
||||
extras=[log, context])
|
||||
@@ -1507,7 +1458,6 @@ def build_pipeline(options, work_folder, log, context):
|
||||
task_merge_sidecars = main_pipeline.merge(
|
||||
task_func=merge_sidecars,
|
||||
input=[task_ocr_tesseract_hocr,
|
||||
task_ocr_tesseract_and_render_pdf,
|
||||
task_ocr_tesseract_textonly_pdf],
|
||||
output=options.sidecar,
|
||||
extras=[log, context])
|
||||
|
||||
Reference in New Issue
Block a user