mirror of
https://github.com/qurator-spk/dinglehopper.git
synced 2025-06-07 19:05:13 +02:00
Merge pull request #50 from b2m/fix-table-extraction
Fix the extraction of text from Page with TableRegion
This commit is contained in:
GitHub
commit
691ce371ca
7 changed files with 724 additions and 17 deletions
|
|
@ -70,24 +70,11 @@ def page_extract(tree, *, textequiv_level="region"):
|
||||||
if reading_order is not None:
|
if reading_order is not None:
|
||||||
for group in reading_order.iterfind("./*", namespaces=nsmap):
|
for group in reading_order.iterfind("./*", namespaces=nsmap):
|
||||||
if ET.QName(group.tag).localname == "OrderedGroup":
|
if ET.QName(group.tag).localname == "OrderedGroup":
|
||||||
region_ref_indexeds = group.findall(
|
regions.extend(
|
||||||
"./page:RegionRefIndexed", namespaces=nsmap
|
extract_texts_from_reading_order_group(
|
||||||
)
|
group, tree, nsmap, textequiv_level
|
||||||
for region_ref_indexed in sorted(
|
|
||||||
region_ref_indexeds, key=lambda r: int(r.attrib["index"])
|
|
||||||
):
|
|
||||||
region_id = region_ref_indexed.attrib["regionRef"]
|
|
||||||
region = tree.find(
|
|
||||||
'.//page:TextRegion[@id="%s"]' % region_id, namespaces=nsmap
|
|
||||||
)
|
)
|
||||||
if region is not None:
|
)
|
||||||
regions.append(
|
|
||||||
ExtractedText.from_text_segment(
|
|
||||||
region, nsmap, textequiv_level=textequiv_level
|
|
||||||
)
|
|
||||||
)
|
|
||||||
else:
|
|
||||||
pass # Not a TextRegion
|
|
||||||
else:
|
else:
|
||||||
raise NotImplementedError
|
raise NotImplementedError
|
||||||
else:
|
else:
|
||||||
|
|
@ -104,6 +91,35 @@ def page_extract(tree, *, textequiv_level="region"):
|
||||||
return ExtractedText(None, regions, "\n", None)
|
return ExtractedText(None, regions, "\n", None)
|
||||||
|
|
||||||
|
|
||||||
|
def extract_texts_from_reading_order_group(group, tree, nsmap, textequiv_level):
|
||||||
|
"""Recursive function to extract the texts from TextRegions in ReadingOrder."""
|
||||||
|
regions = []
|
||||||
|
ro_children = group.findall("./page:RegionRefIndexed", namespaces=nsmap)
|
||||||
|
ro_children.extend(group.findall("./page:OrderedGroupIndexed", namespaces=nsmap))
|
||||||
|
ro_children = filter(lambda child: "index" in child.attrib.keys(), ro_children)
|
||||||
|
for ro_child in sorted(ro_children, key=lambda child: int(child.attrib["index"])):
|
||||||
|
if ET.QName(ro_child.tag).localname == "OrderedGroupIndexed":
|
||||||
|
regions.extend(
|
||||||
|
extract_texts_from_reading_order_group(
|
||||||
|
ro_child, tree, nsmap, textequiv_level
|
||||||
|
)
|
||||||
|
)
|
||||||
|
else:
|
||||||
|
region_id = ro_child.attrib["regionRef"]
|
||||||
|
region = tree.find(
|
||||||
|
'.//page:TextRegion[@id="%s"]' % region_id, namespaces=nsmap
|
||||||
|
)
|
||||||
|
if region is not None:
|
||||||
|
regions.append(
|
||||||
|
ExtractedText.from_text_segment(
|
||||||
|
region, nsmap, textequiv_level=textequiv_level
|
||||||
|
)
|
||||||
|
)
|
||||||
|
else:
|
||||||
|
pass # Not a TextRegion
|
||||||
|
return regions
|
||||||
|
|
||||||
|
|
||||||
def page_text(tree, *, textequiv_level="region"):
|
def page_text(tree, *, textequiv_level="region"):
|
||||||
return page_extract(tree, textequiv_level=textequiv_level).text
|
return page_extract(tree, textequiv_level=textequiv_level).text
|
||||||
|
|
||||||
|
|
|
||||||
|
|
@ -0,0 +1,121 @@
|
||||||
|
<?xml version="1.0" encoding="UTF-8" standalone="no"?>
|
||||||
|
<PcGts xmlns="http://schema.primaresearch.org/PAGE/gts/pagecontent/2019-07-15"
|
||||||
|
xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
|
||||||
|
xsi:schemaLocation="http://schema.primaresearch.org/PAGE/gts/pagecontent/2019-07-15 http://schema.primaresearch.org/PAGE/gts/pagecontent/2019-07-15/pagecontent.xsd">
|
||||||
|
<Metadata>
|
||||||
|
<Creator/>
|
||||||
|
<Created>2020-10-28T08:43:47</Created>
|
||||||
|
<LastChange>1970-01-01T00:00:00</LastChange>
|
||||||
|
<Comments/>
|
||||||
|
</Metadata>
|
||||||
|
<Page imageFilename="0001.png" imageHeight="1123" imageWidth="794">
|
||||||
|
<TextRegion id="r1" orientation="0.0">
|
||||||
|
<Coords points="315,437 315,407 339,407 339,437"/>
|
||||||
|
<TextLine id="l1">
|
||||||
|
<Coords points="318,434 318,409 337,409 337,434"/>
|
||||||
|
<TextEquiv index="0">
|
||||||
|
<Unicode>5</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextLine>
|
||||||
|
<TextEquiv>
|
||||||
|
<Unicode>5</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextRegion>
|
||||||
|
<TextRegion id="r2" orientation="0.0">
|
||||||
|
<Coords points="425,436 425,406 450,406 450,436"/>
|
||||||
|
<TextLine id="l2">
|
||||||
|
<Coords points="429,434 429,410 446,410 446,434"/>
|
||||||
|
<TextEquiv index="0">
|
||||||
|
<Unicode>6</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextLine>
|
||||||
|
<TextEquiv>
|
||||||
|
<Unicode>6</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextRegion>
|
||||||
|
<TextRegion id="r3" orientation="0.0">
|
||||||
|
<Coords points="233,499 233,467 262,467 262,499"/>
|
||||||
|
<TextLine id="l3">
|
||||||
|
<Coords points="237,496 237,468 258,468 258,496"/>
|
||||||
|
<TextEquiv index="0">
|
||||||
|
<Unicode>7</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextLine>
|
||||||
|
<TextEquiv>
|
||||||
|
<Unicode>7</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextRegion>
|
||||||
|
<TextRegion id="r4" orientation="0.0">
|
||||||
|
<Coords points="316,497 316,470 340,470 340,497"/>
|
||||||
|
<TextLine id="l4">
|
||||||
|
<Coords points="319,494 319,472 337,472 337,494"/>
|
||||||
|
<TextEquiv index="0">
|
||||||
|
<Unicode>8</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextLine>
|
||||||
|
<TextEquiv>
|
||||||
|
<Unicode>8</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextRegion>
|
||||||
|
<TextRegion id="r5" orientation="0.0">
|
||||||
|
<Coords points="423,501 423,468 451,468 451,501"/>
|
||||||
|
<TextLine id="l5">
|
||||||
|
<Coords points="427,497 427,470 447,470 447,497"/>
|
||||||
|
<TextEquiv index="0">
|
||||||
|
<Unicode>9</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextLine>
|
||||||
|
<TextEquiv>
|
||||||
|
<Unicode>9</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextRegion>
|
||||||
|
<TextRegion id="r6" orientation="0.0">
|
||||||
|
<Coords points="237,373 237,347 259,347 259,373"/>
|
||||||
|
<TextLine id="l6">
|
||||||
|
<Coords points="240,372 240,349 256,349 256,372"/>
|
||||||
|
<TextEquiv index="0">
|
||||||
|
<Unicode>1</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextLine>
|
||||||
|
<TextEquiv>
|
||||||
|
<Unicode>1</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextRegion>
|
||||||
|
<TextRegion id="r7" orientation="0.0">
|
||||||
|
<Coords points="312,373 312,347 341,347 341,373"/>
|
||||||
|
<TextLine id="l7">
|
||||||
|
<Coords points="318,372 318,350 338,350 338,372"/>
|
||||||
|
<TextEquiv index="0">
|
||||||
|
<Unicode>2</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextLine>
|
||||||
|
<TextEquiv>
|
||||||
|
<Unicode>2</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextRegion>
|
||||||
|
<TextRegion id="r8" orientation="0.0">
|
||||||
|
<Coords points="428,373 428,349 448,349 448,373"/>
|
||||||
|
<TextLine id="l8">
|
||||||
|
<Coords points="430,373 430,349 445,349 445,373"/>
|
||||||
|
<TextEquiv index="0">
|
||||||
|
<Unicode>3</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextLine>
|
||||||
|
<TextEquiv>
|
||||||
|
<Unicode>3</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextRegion>
|
||||||
|
<TextRegion id="r9" orientation="0.0">
|
||||||
|
<Coords points="236,438 236,406 261,406 261,438"/>
|
||||||
|
<TextLine id="l9">
|
||||||
|
<Coords points="238,436 238,408 258,408 258,436"/>
|
||||||
|
<TextEquiv index="0">
|
||||||
|
<Unicode>4</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextLine>
|
||||||
|
<TextEquiv>
|
||||||
|
<Unicode>4</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextRegion>
|
||||||
|
</Page>
|
||||||
|
</PcGts>
|
||||||
134
qurator/dinglehopper/tests/data/table-order/table-order-0001.xml
Normal file
134
qurator/dinglehopper/tests/data/table-order/table-order-0001.xml
Normal file
|
|
@ -0,0 +1,134 @@
|
||||||
|
<?xml version="1.0" encoding="UTF-8" standalone="no"?>
|
||||||
|
<PcGts xmlns="http://schema.primaresearch.org/PAGE/gts/pagecontent/2019-07-15"
|
||||||
|
xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
|
||||||
|
xsi:schemaLocation="http://schema.primaresearch.org/PAGE/gts/pagecontent/2019-07-15 http://schema.primaresearch.org/PAGE/gts/pagecontent/2019-07-15/pagecontent.xsd">
|
||||||
|
<Metadata>
|
||||||
|
<Creator/>
|
||||||
|
<Created>2020-10-28T08:43:47</Created>
|
||||||
|
<LastChange>1970-01-01T00:00:00</LastChange>
|
||||||
|
<Comments/>
|
||||||
|
</Metadata>
|
||||||
|
<Page imageFilename="0001.png" imageHeight="1123" imageWidth="794">
|
||||||
|
<ReadingOrder>
|
||||||
|
<OrderedGroup id="g1">
|
||||||
|
<RegionRefIndexed index="0" regionRef="r6"/>
|
||||||
|
<RegionRefIndexed index="1" regionRef="r7"/>
|
||||||
|
<RegionRefIndexed index="2" regionRef="r8"/>
|
||||||
|
<RegionRefIndexed index="3" regionRef="r9"/>
|
||||||
|
<RegionRefIndexed index="4" regionRef="r1"/>
|
||||||
|
<RegionRefIndexed index="5" regionRef="r2"/>
|
||||||
|
<RegionRefIndexed index="6" regionRef="r3"/>
|
||||||
|
<RegionRefIndexed index="7" regionRef="r4"/>
|
||||||
|
<RegionRefIndexed index="8" regionRef="r5"/>
|
||||||
|
</OrderedGroup>
|
||||||
|
</ReadingOrder>
|
||||||
|
<TextRegion id="r1" orientation="0.0">
|
||||||
|
<Coords points="315,437 315,407 339,407 339,437"/>
|
||||||
|
<TextLine id="l1">
|
||||||
|
<Coords points="318,434 318,409 337,409 337,434"/>
|
||||||
|
<TextEquiv index="0">
|
||||||
|
<Unicode>5</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextLine>
|
||||||
|
<TextEquiv>
|
||||||
|
<Unicode>5</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextRegion>
|
||||||
|
<TextRegion id="r2" orientation="0.0">
|
||||||
|
<Coords points="425,436 425,406 450,406 450,436"/>
|
||||||
|
<TextLine id="l2">
|
||||||
|
<Coords points="429,434 429,410 446,410 446,434"/>
|
||||||
|
<TextEquiv index="0">
|
||||||
|
<Unicode>6</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextLine>
|
||||||
|
<TextEquiv>
|
||||||
|
<Unicode>6</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextRegion>
|
||||||
|
<TextRegion id="r3" orientation="0.0">
|
||||||
|
<Coords points="233,499 233,467 262,467 262,499"/>
|
||||||
|
<TextLine id="l3">
|
||||||
|
<Coords points="237,496 237,468 258,468 258,496"/>
|
||||||
|
<TextEquiv index="0">
|
||||||
|
<Unicode>7</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextLine>
|
||||||
|
<TextEquiv>
|
||||||
|
<Unicode>7</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextRegion>
|
||||||
|
<TextRegion id="r4" orientation="0.0">
|
||||||
|
<Coords points="316,497 316,470 340,470 340,497"/>
|
||||||
|
<TextLine id="l4">
|
||||||
|
<Coords points="319,494 319,472 337,472 337,494"/>
|
||||||
|
<TextEquiv index="0">
|
||||||
|
<Unicode>8</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextLine>
|
||||||
|
<TextEquiv>
|
||||||
|
<Unicode>8</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextRegion>
|
||||||
|
<TextRegion id="r5" orientation="0.0">
|
||||||
|
<Coords points="423,501 423,468 451,468 451,501"/>
|
||||||
|
<TextLine id="l5">
|
||||||
|
<Coords points="427,497 427,470 447,470 447,497"/>
|
||||||
|
<TextEquiv index="0">
|
||||||
|
<Unicode>9</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextLine>
|
||||||
|
<TextEquiv>
|
||||||
|
<Unicode>9</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextRegion>
|
||||||
|
<TextRegion id="r6" orientation="0.0">
|
||||||
|
<Coords points="237,373 237,347 259,347 259,373"/>
|
||||||
|
<TextLine id="l6">
|
||||||
|
<Coords points="240,372 240,349 256,349 256,372"/>
|
||||||
|
<TextEquiv index="0">
|
||||||
|
<Unicode>1</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextLine>
|
||||||
|
<TextEquiv>
|
||||||
|
<Unicode>1</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextRegion>
|
||||||
|
<TextRegion id="r7" orientation="0.0">
|
||||||
|
<Coords points="312,373 312,347 341,347 341,373"/>
|
||||||
|
<TextLine id="l7">
|
||||||
|
<Coords points="318,372 318,350 338,350 338,372"/>
|
||||||
|
<TextEquiv index="0">
|
||||||
|
<Unicode>2</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextLine>
|
||||||
|
<TextEquiv>
|
||||||
|
<Unicode>2</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextRegion>
|
||||||
|
<TextRegion id="r8" orientation="0.0">
|
||||||
|
<Coords points="428,373 428,349 448,349 448,373"/>
|
||||||
|
<TextLine id="l8">
|
||||||
|
<Coords points="430,373 430,349 445,349 445,373"/>
|
||||||
|
<TextEquiv index="0">
|
||||||
|
<Unicode>3</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextLine>
|
||||||
|
<TextEquiv>
|
||||||
|
<Unicode>3</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextRegion>
|
||||||
|
<TextRegion id="r9" orientation="0.0">
|
||||||
|
<Coords points="236,438 236,406 261,406 261,438"/>
|
||||||
|
<TextLine id="l9">
|
||||||
|
<Coords points="238,436 238,408 258,408 258,436"/>
|
||||||
|
<TextEquiv index="0">
|
||||||
|
<Unicode>4</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextLine>
|
||||||
|
<TextEquiv>
|
||||||
|
<Unicode>4</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextRegion>
|
||||||
|
</Page>
|
||||||
|
</PcGts>
|
||||||
134
qurator/dinglehopper/tests/data/table-order/table-order-0002.xml
Normal file
134
qurator/dinglehopper/tests/data/table-order/table-order-0002.xml
Normal file
|
|
@ -0,0 +1,134 @@
|
||||||
|
<?xml version="1.0" encoding="UTF-8" standalone="no"?>
|
||||||
|
<PcGts xmlns="http://schema.primaresearch.org/PAGE/gts/pagecontent/2019-07-15"
|
||||||
|
xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
|
||||||
|
xsi:schemaLocation="http://schema.primaresearch.org/PAGE/gts/pagecontent/2019-07-15 http://schema.primaresearch.org/PAGE/gts/pagecontent/2019-07-15/pagecontent.xsd">
|
||||||
|
<Metadata>
|
||||||
|
<Creator/>
|
||||||
|
<Created>2020-10-28T08:43:47</Created>
|
||||||
|
<LastChange>1970-01-01T00:00:00</LastChange>
|
||||||
|
<Comments/>
|
||||||
|
</Metadata>
|
||||||
|
<Page imageFilename="0002.png" imageHeight="1123" imageWidth="794">
|
||||||
|
<ReadingOrder>
|
||||||
|
<OrderedGroup id="g1">
|
||||||
|
<RegionRefIndexed index="0" regionRef="r6"/>
|
||||||
|
<RegionRefIndexed index="1" regionRef="r9"/>
|
||||||
|
<RegionRefIndexed index="2" regionRef="r3"/>
|
||||||
|
<RegionRefIndexed index="3" regionRef="r7"/>
|
||||||
|
<RegionRefIndexed index="4" regionRef="r1"/>
|
||||||
|
<RegionRefIndexed index="5" regionRef="r4"/>
|
||||||
|
<RegionRefIndexed index="6" regionRef="r8"/>
|
||||||
|
<RegionRefIndexed index="7" regionRef="r2"/>
|
||||||
|
<RegionRefIndexed index="8" regionRef="r5"/>
|
||||||
|
</OrderedGroup>
|
||||||
|
</ReadingOrder>
|
||||||
|
<TextRegion id="r1" orientation="0.0">
|
||||||
|
<Coords points="315,437 315,407 339,407 339,437"/>
|
||||||
|
<TextLine id="l1">
|
||||||
|
<Coords points="318,434 318,409 337,409 337,434"/>
|
||||||
|
<TextEquiv index="0">
|
||||||
|
<Unicode>5</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextLine>
|
||||||
|
<TextEquiv>
|
||||||
|
<Unicode>5</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextRegion>
|
||||||
|
<TextRegion id="r2" orientation="0.0">
|
||||||
|
<Coords points="425,436 425,406 450,406 450,436"/>
|
||||||
|
<TextLine id="l2">
|
||||||
|
<Coords points="429,434 429,410 446,410 446,434"/>
|
||||||
|
<TextEquiv index="0">
|
||||||
|
<Unicode>6</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextLine>
|
||||||
|
<TextEquiv>
|
||||||
|
<Unicode>6</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextRegion>
|
||||||
|
<TextRegion id="r3" orientation="0.0">
|
||||||
|
<Coords points="233,499 233,467 262,467 262,499"/>
|
||||||
|
<TextLine id="l3">
|
||||||
|
<Coords points="237,496 237,468 258,468 258,496"/>
|
||||||
|
<TextEquiv index="0">
|
||||||
|
<Unicode>7</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextLine>
|
||||||
|
<TextEquiv>
|
||||||
|
<Unicode>7</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextRegion>
|
||||||
|
<TextRegion id="r4" orientation="0.0">
|
||||||
|
<Coords points="316,497 316,470 340,470 340,497"/>
|
||||||
|
<TextLine id="l4">
|
||||||
|
<Coords points="319,494 319,472 337,472 337,494"/>
|
||||||
|
<TextEquiv index="0">
|
||||||
|
<Unicode>8</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextLine>
|
||||||
|
<TextEquiv>
|
||||||
|
<Unicode>8</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextRegion>
|
||||||
|
<TextRegion id="r5" orientation="0.0">
|
||||||
|
<Coords points="423,501 423,468 451,468 451,501"/>
|
||||||
|
<TextLine id="l5">
|
||||||
|
<Coords points="427,497 427,470 447,470 447,497"/>
|
||||||
|
<TextEquiv index="0">
|
||||||
|
<Unicode>9</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextLine>
|
||||||
|
<TextEquiv>
|
||||||
|
<Unicode>9</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextRegion>
|
||||||
|
<TextRegion id="r6" orientation="0.0">
|
||||||
|
<Coords points="237,373 237,347 259,347 259,373"/>
|
||||||
|
<TextLine id="l6">
|
||||||
|
<Coords points="240,372 240,349 256,349 256,372"/>
|
||||||
|
<TextEquiv index="0">
|
||||||
|
<Unicode>1</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextLine>
|
||||||
|
<TextEquiv>
|
||||||
|
<Unicode>1</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextRegion>
|
||||||
|
<TextRegion id="r7" orientation="0.0">
|
||||||
|
<Coords points="312,373 312,347 341,347 341,373"/>
|
||||||
|
<TextLine id="l7">
|
||||||
|
<Coords points="318,372 318,350 338,350 338,372"/>
|
||||||
|
<TextEquiv index="0">
|
||||||
|
<Unicode>2</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextLine>
|
||||||
|
<TextEquiv>
|
||||||
|
<Unicode>2</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextRegion>
|
||||||
|
<TextRegion id="r8" orientation="0.0">
|
||||||
|
<Coords points="428,373 428,349 448,349 448,373"/>
|
||||||
|
<TextLine id="l8">
|
||||||
|
<Coords points="430,373 430,349 445,349 445,373"/>
|
||||||
|
<TextEquiv index="0">
|
||||||
|
<Unicode>3</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextLine>
|
||||||
|
<TextEquiv>
|
||||||
|
<Unicode>3</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextRegion>
|
||||||
|
<TextRegion id="r9" orientation="0.0">
|
||||||
|
<Coords points="236,438 236,406 261,406 261,438"/>
|
||||||
|
<TextLine id="l9">
|
||||||
|
<Coords points="238,436 238,408 258,408 258,436"/>
|
||||||
|
<TextEquiv index="0">
|
||||||
|
<Unicode>4</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextLine>
|
||||||
|
<TextEquiv>
|
||||||
|
<Unicode>4</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextRegion>
|
||||||
|
</Page>
|
||||||
|
</PcGts>
|
||||||
139
qurator/dinglehopper/tests/data/table-order/table-region.xml
Normal file
139
qurator/dinglehopper/tests/data/table-order/table-region.xml
Normal file
|
|
@ -0,0 +1,139 @@
|
||||||
|
<?xml version="1.0" encoding="UTF-8" standalone="no"?>
|
||||||
|
<PcGts xmlns="http://schema.primaresearch.org/PAGE/gts/pagecontent/2019-07-15"
|
||||||
|
xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
|
||||||
|
xsi:schemaLocation="http://schema.primaresearch.org/PAGE/gts/pagecontent/2019-07-15 http://schema.primaresearch.org/PAGE/gts/pagecontent/2019-07-15/pagecontent.xsd">
|
||||||
|
<Metadata>
|
||||||
|
<Creator/>
|
||||||
|
<Created>2020-10-28T08:43:47</Created>
|
||||||
|
<LastChange>1970-01-01T00:00:00</LastChange>
|
||||||
|
<Comments/>
|
||||||
|
</Metadata>
|
||||||
|
<Page imageFilename="0001.png" imageHeight="1123" imageWidth="794">
|
||||||
|
<ReadingOrder>
|
||||||
|
<OrderedGroup id="g1">
|
||||||
|
<OrderedGroupIndexed id="r0_order" regionRef="r0" index="0">
|
||||||
|
<RegionRefIndexed index="0" regionRef="r6"/>
|
||||||
|
<RegionRefIndexed index="1" regionRef="r7"/>
|
||||||
|
<RegionRefIndexed index="2" regionRef="r8"/>
|
||||||
|
<RegionRefIndexed index="3" regionRef="r9"/>
|
||||||
|
<RegionRefIndexed index="4" regionRef="r1"/>
|
||||||
|
<RegionRefIndexed index="5" regionRef="r2"/>
|
||||||
|
<RegionRefIndexed index="6" regionRef="r3"/>
|
||||||
|
<RegionRefIndexed index="7" regionRef="r4"/>
|
||||||
|
<RegionRefIndexed index="8" regionRef="r5"/>
|
||||||
|
</OrderedGroupIndexed>
|
||||||
|
</OrderedGroup>
|
||||||
|
</ReadingOrder>
|
||||||
|
<TableRegion id="r0">
|
||||||
|
<Coords points="230,530 230,330 460,330 460,530"/>
|
||||||
|
<TextRegion id="r1" orientation="0.0">
|
||||||
|
<Coords points="315,437 315,407 339,407 339,437"/>
|
||||||
|
<TextLine id="l1">
|
||||||
|
<Coords points="318,434 318,409 337,409 337,434"/>
|
||||||
|
<TextEquiv index="0">
|
||||||
|
<Unicode>5</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextLine>
|
||||||
|
<TextEquiv>
|
||||||
|
<Unicode>5</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextRegion>
|
||||||
|
<TextRegion id="r2" orientation="0.0">
|
||||||
|
<Coords points="425,436 425,406 450,406 450,436"/>
|
||||||
|
<TextLine id="l2">
|
||||||
|
<Coords points="429,434 429,410 446,410 446,434"/>
|
||||||
|
<TextEquiv index="0">
|
||||||
|
<Unicode>6</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextLine>
|
||||||
|
<TextEquiv>
|
||||||
|
<Unicode>6</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextRegion>
|
||||||
|
<TextRegion id="r3" orientation="0.0">
|
||||||
|
<Coords points="233,499 233,467 262,467 262,499"/>
|
||||||
|
<TextLine id="l3">
|
||||||
|
<Coords points="237,496 237,468 258,468 258,496"/>
|
||||||
|
<TextEquiv index="0">
|
||||||
|
<Unicode>7</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextLine>
|
||||||
|
<TextEquiv>
|
||||||
|
<Unicode>7</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextRegion>
|
||||||
|
<TextRegion id="r4" orientation="0.0">
|
||||||
|
<Coords points="316,497 316,470 340,470 340,497"/>
|
||||||
|
<TextLine id="l4">
|
||||||
|
<Coords points="319,494 319,472 337,472 337,494"/>
|
||||||
|
<TextEquiv index="0">
|
||||||
|
<Unicode>8</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextLine>
|
||||||
|
<TextEquiv>
|
||||||
|
<Unicode>8</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextRegion>
|
||||||
|
<TextRegion id="r5" orientation="0.0">
|
||||||
|
<Coords points="423,501 423,468 451,468 451,501"/>
|
||||||
|
<TextLine id="l5">
|
||||||
|
<Coords points="427,497 427,470 447,470 447,497"/>
|
||||||
|
<TextEquiv index="0">
|
||||||
|
<Unicode>9</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextLine>
|
||||||
|
<TextEquiv>
|
||||||
|
<Unicode>9</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextRegion>
|
||||||
|
<TextRegion id="r6" orientation="0.0">
|
||||||
|
<Coords points="237,373 237,347 259,347 259,373"/>
|
||||||
|
<TextLine id="l6">
|
||||||
|
<Coords points="240,372 240,349 256,349 256,372"/>
|
||||||
|
<TextEquiv index="0">
|
||||||
|
<Unicode>1</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextLine>
|
||||||
|
<TextEquiv>
|
||||||
|
<Unicode>1</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextRegion>
|
||||||
|
<TextRegion id="r7" orientation="0.0">
|
||||||
|
<Coords points="312,373 312,347 341,347 341,373"/>
|
||||||
|
<TextLine id="l7">
|
||||||
|
<Coords points="318,372 318,350 338,350 338,372"/>
|
||||||
|
<TextEquiv index="0">
|
||||||
|
<Unicode>2</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextLine>
|
||||||
|
<TextEquiv>
|
||||||
|
<Unicode>2</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextRegion>
|
||||||
|
<TextRegion id="r8" orientation="0.0">
|
||||||
|
<Coords points="428,373 428,349 448,349 448,373"/>
|
||||||
|
<TextLine id="l8">
|
||||||
|
<Coords points="430,373 430,349 445,349 445,373"/>
|
||||||
|
<TextEquiv index="0">
|
||||||
|
<Unicode>3</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextLine>
|
||||||
|
<TextEquiv>
|
||||||
|
<Unicode>3</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextRegion>
|
||||||
|
<TextRegion id="r9" orientation="0.0">
|
||||||
|
<Coords points="236,438 236,406 261,406 261,438"/>
|
||||||
|
<TextLine id="l9">
|
||||||
|
<Coords points="238,436 238,408 258,408 258,436"/>
|
||||||
|
<TextEquiv index="0">
|
||||||
|
<Unicode>4</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextLine>
|
||||||
|
<TextEquiv>
|
||||||
|
<Unicode>4</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextRegion>
|
||||||
|
</TableRegion>
|
||||||
|
</Page>
|
||||||
|
</PcGts>
|
||||||
134
qurator/dinglehopper/tests/data/table-order/table-unordered.xml
Normal file
134
qurator/dinglehopper/tests/data/table-order/table-unordered.xml
Normal file
|
|
@ -0,0 +1,134 @@
|
||||||
|
<?xml version="1.0" encoding="UTF-8" standalone="no"?>
|
||||||
|
<PcGts xmlns="http://schema.primaresearch.org/PAGE/gts/pagecontent/2019-07-15"
|
||||||
|
xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
|
||||||
|
xsi:schemaLocation="http://schema.primaresearch.org/PAGE/gts/pagecontent/2019-07-15 http://schema.primaresearch.org/PAGE/gts/pagecontent/2019-07-15/pagecontent.xsd">
|
||||||
|
<Metadata>
|
||||||
|
<Creator/>
|
||||||
|
<Created>2020-10-28T08:43:47</Created>
|
||||||
|
<LastChange>1970-01-01T00:00:00</LastChange>
|
||||||
|
<Comments/>
|
||||||
|
</Metadata>
|
||||||
|
<Page imageFilename="0001.png" imageHeight="1123" imageWidth="794">
|
||||||
|
<ReadingOrder>
|
||||||
|
<UnorderedGroup id="g1">
|
||||||
|
<RegionRef regionRef="r6"/>
|
||||||
|
<RegionRef regionRef="r7"/>
|
||||||
|
<RegionRef regionRef="r8"/>
|
||||||
|
<RegionRef regionRef="r9"/>
|
||||||
|
<RegionRef regionRef="r1"/>
|
||||||
|
<RegionRef regionRef="r2"/>
|
||||||
|
<RegionRef regionRef="r3"/>
|
||||||
|
<RegionRef regionRef="r4"/>
|
||||||
|
<RegionRef regionRef="r5"/>
|
||||||
|
</UnorderedGroup>
|
||||||
|
</ReadingOrder>
|
||||||
|
<TextRegion id="r1" orientation="0.0">
|
||||||
|
<Coords points="315,437 315,407 339,407 339,437"/>
|
||||||
|
<TextLine id="l1">
|
||||||
|
<Coords points="318,434 318,409 337,409 337,434"/>
|
||||||
|
<TextEquiv index="0">
|
||||||
|
<Unicode>5</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextLine>
|
||||||
|
<TextEquiv>
|
||||||
|
<Unicode>5</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextRegion>
|
||||||
|
<TextRegion id="r2" orientation="0.0">
|
||||||
|
<Coords points="425,436 425,406 450,406 450,436"/>
|
||||||
|
<TextLine id="l2">
|
||||||
|
<Coords points="429,434 429,410 446,410 446,434"/>
|
||||||
|
<TextEquiv index="0">
|
||||||
|
<Unicode>6</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextLine>
|
||||||
|
<TextEquiv>
|
||||||
|
<Unicode>6</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextRegion>
|
||||||
|
<TextRegion id="r3" orientation="0.0">
|
||||||
|
<Coords points="233,499 233,467 262,467 262,499"/>
|
||||||
|
<TextLine id="l3">
|
||||||
|
<Coords points="237,496 237,468 258,468 258,496"/>
|
||||||
|
<TextEquiv index="0">
|
||||||
|
<Unicode>7</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextLine>
|
||||||
|
<TextEquiv>
|
||||||
|
<Unicode>7</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextRegion>
|
||||||
|
<TextRegion id="r4" orientation="0.0">
|
||||||
|
<Coords points="316,497 316,470 340,470 340,497"/>
|
||||||
|
<TextLine id="l4">
|
||||||
|
<Coords points="319,494 319,472 337,472 337,494"/>
|
||||||
|
<TextEquiv index="0">
|
||||||
|
<Unicode>8</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextLine>
|
||||||
|
<TextEquiv>
|
||||||
|
<Unicode>8</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextRegion>
|
||||||
|
<TextRegion id="r5" orientation="0.0">
|
||||||
|
<Coords points="423,501 423,468 451,468 451,501"/>
|
||||||
|
<TextLine id="l5">
|
||||||
|
<Coords points="427,497 427,470 447,470 447,497"/>
|
||||||
|
<TextEquiv index="0">
|
||||||
|
<Unicode>9</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextLine>
|
||||||
|
<TextEquiv>
|
||||||
|
<Unicode>9</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextRegion>
|
||||||
|
<TextRegion id="r6" orientation="0.0">
|
||||||
|
<Coords points="237,373 237,347 259,347 259,373"/>
|
||||||
|
<TextLine id="l6">
|
||||||
|
<Coords points="240,372 240,349 256,349 256,372"/>
|
||||||
|
<TextEquiv index="0">
|
||||||
|
<Unicode>1</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextLine>
|
||||||
|
<TextEquiv>
|
||||||
|
<Unicode>1</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextRegion>
|
||||||
|
<TextRegion id="r7" orientation="0.0">
|
||||||
|
<Coords points="312,373 312,347 341,347 341,373"/>
|
||||||
|
<TextLine id="l7">
|
||||||
|
<Coords points="318,372 318,350 338,350 338,372"/>
|
||||||
|
<TextEquiv index="0">
|
||||||
|
<Unicode>2</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextLine>
|
||||||
|
<TextEquiv>
|
||||||
|
<Unicode>2</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextRegion>
|
||||||
|
<TextRegion id="r8" orientation="0.0">
|
||||||
|
<Coords points="428,373 428,349 448,349 448,373"/>
|
||||||
|
<TextLine id="l8">
|
||||||
|
<Coords points="430,373 430,349 445,349 445,373"/>
|
||||||
|
<TextEquiv index="0">
|
||||||
|
<Unicode>3</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextLine>
|
||||||
|
<TextEquiv>
|
||||||
|
<Unicode>3</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextRegion>
|
||||||
|
<TextRegion id="r9" orientation="0.0">
|
||||||
|
<Coords points="236,438 236,406 261,406 261,438"/>
|
||||||
|
<TextLine id="l9">
|
||||||
|
<Coords points="238,436 238,408 258,408 258,436"/>
|
||||||
|
<TextEquiv index="0">
|
||||||
|
<Unicode>4</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextLine>
|
||||||
|
<TextEquiv>
|
||||||
|
<Unicode>4</Unicode>
|
||||||
|
</TextEquiv>
|
||||||
|
</TextRegion>
|
||||||
|
</Page>
|
||||||
|
</PcGts>
|
||||||
29
qurator/dinglehopper/tests/test_integ_table_extraction.py
Normal file
29
qurator/dinglehopper/tests/test_integ_table_extraction.py
Normal file
|
|
@ -0,0 +1,29 @@
|
||||||
|
import os
|
||||||
|
|
||||||
|
import pytest
|
||||||
|
from lxml import etree as ET
|
||||||
|
|
||||||
|
from .. import page_text
|
||||||
|
|
||||||
|
|
||||||
|
@pytest.mark.parametrize(
|
||||||
|
"file,expected_text",
|
||||||
|
[
|
||||||
|
("table-order-0001.xml", "1\n2\n3\n4\n5\n6\n7\n8\n9"),
|
||||||
|
("table-order-0002.xml", "1\n4\n7\n2\n5\n8\n3\n6\n9"),
|
||||||
|
("table-region.xml", "1\n2\n3\n4\n5\n6\n7\n8\n9"),
|
||||||
|
("table-no-reading-order.xml", "5\n6\n7\n8\n9\n1\n2\n3\n4"),
|
||||||
|
("table-unordered.xml", "5\n6\n7\n8\n9\n1\n2\n3\n4"),
|
||||||
|
],
|
||||||
|
)
|
||||||
|
@pytest.mark.integration
|
||||||
|
def test_reading_order_settings(file, expected_text):
|
||||||
|
data_dir = os.path.join(
|
||||||
|
os.path.dirname(os.path.abspath(__file__)), "data", "table-order"
|
||||||
|
)
|
||||||
|
if "table-unordered.xml" == file:
|
||||||
|
with pytest.raises(NotImplementedError):
|
||||||
|
page_text(ET.parse(os.path.join(data_dir, file)))
|
||||||
|
else:
|
||||||
|
ocr = page_text(ET.parse(os.path.join(data_dir, file)))
|
||||||
|
assert ocr == expected_text
|
||||||
Loading…
Add table
Add a link
Reference in a new issue