30 lines
842 B
Protocol Buffer
30 lines
842 B
Protocol Buffer
syntax = "proto3";
|
|
|
|
message AllDocumentTextData {
|
|
|
|
repeated DocumentTextData documentTextData = 1;
|
|
}
|
|
|
|
message DocumentTextData {
|
|
// Identifier of the text block.
|
|
int64 id = 1;
|
|
|
|
// The page the text block occurs on.
|
|
int64 page = 2;
|
|
|
|
// The text of the text block.
|
|
string searchText = 3;
|
|
|
|
// Each text block is assigned a number on a page, starting from 0.
|
|
int32 numberOnPage = 4;
|
|
|
|
// The text blocks are ordered, this number represents the start of the text block as a string offset.
|
|
int32 start = 5;
|
|
|
|
// The text blocks are ordered, this number represents the end of the text block as a string offset.
|
|
int32 end = 6;
|
|
|
|
// The line breaks in the text of this semantic node in string offsets. They are exclusive end. At the end of each semantic node there is an implicit linebreak.
|
|
repeated int32 lineBreaks = 7;
|
|
}
|