urn:showpapers:reading:1JSON Schema 2020-12Download JSON

readings/<id>.json: the app's on-device reading of one original (OCR pages, blocks, lines, optional words, normalized regions and confidences) and the reader's result JSON. Text limits are UTF-8 bytes: page 262144, block 65536, line 8192, word 2048. Page indexes are unique. Together with the paper record it must fit the apps' analysis store (50,000 nodes, 2 MiB). sdkAnalysis is ShowPapers-internal: copy it unchanged or set it to null; never edit it.

Top Level

MemberTypeRules
readingVersion required1
extractorVersion requiredintegermin 1; max 100000
concerns requiredarray of one of "no-text", "unrecognized-document", "conflicting-document-types", "no-supported-fields", "field-limit-reached"max items 5; unique
pages requiredarray of pagemax items 200
sdkAnalysis requirednull or object

Definitions

unit

number · min 0; max 1

region

Normalized to the rendered page (0..1). box is [left, top, right, bottom] with left <= right and top <= bottom.

MemberTypeRules
box requirednull or array of unit
polygon requiredarray of array of unitmax items 8

word

MemberTypeRules
text requiredstringmin length 0; max length 2048
region requiredregion
confidence requirednull or unit

line

MemberTypeRules
text requiredstringmin length 0; max length 8192
region requiredregion
confidence requirednull or unit
words requiredarray of wordmax items 2048

block

MemberTypeRules
text requiredstringmin length 0; max length 65536
region requiredregion
lines requiredarray of linemax items 10000

page

MemberTypeRules
index requiredintegermin 0; max 199
width requiredintegermin 1; max 20000
height requiredintegermin 1; max 20000
engine requiredstringmin length 1; max length 128; pattern ^[\x20-\x7e]+$
text requiredstringmin length 0; max length 262144
blocks requiredarray of blockmax items 2000