zlei6/ragflow
1
0
Fork 0
mirror of https://github.com/infiniflow/ragflow.git synced 2026-07-22 16:36:47 +08:00
Code Issues Packages Projects Releases Wiki Activity

Fix centered PDF title reading order (#17009)

Browse Source
This commit is contained in:
zcxGGmu
2026-07-18 01:31:27 +08:00
committed by GitHub
parent 982b9c7b25
commit 1c0432a816
3 changed files with 116 additions and 8 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
internal/deepdoc/parser/pdf/parser.go
View File

@@ -456,9 +456,8 @@ func (p *Parser) buildLayout(ctx context.Context,
boxes = lyt.TextMerge(boxes, medianHeights)
result.Metrics.BoxesTextMerge = len(boxes)
// FinalReadingOrderMerge sorts page → column (ColID) → top → x0, which
// is the correct multi-column reading order. NaiveVerticalMerge preserves
// this column-major order (it buckets by ColID before merging).
// Preserve column-major content order while NaiveVerticalMerge promotes a
// page-leading title isolated in its own column.
boxes = lyt.FinalReadingOrderMerge(boxes)
boxes = lyt.NaiveVerticalMerge(boxes, medianHeights, medianWidths, pageEnglish)