j90 p S r SSKJr SSKrSSKJrJr SSKJr SSK J
r
J r J r J
r
JrJr \
( a SSKJr SSKJr SS KJr SS
KJrJr SS KJr SS KJ r J r \\ \ \ \ 4 r " S
S\ 5 r " S S\ 5 r! " S S\ 5 r" " S S\ 5 r#\ " S S5 5 r$S S jr%S S jr&S S jr'S S jr(S S jr)g)!u The output contract a `Parser` hands back: the document plus the signals
docling produces alongside it.
`result.document` is the DoclingDocument. Four more fields ride with it. The
fusion step reads only `pages` and `table_provenance`:
- `confidence` — docling's own confidence report, taken straight from
`ConversionResult.confidence`. It is held in memory only. `save` does not
write it and `load` leaves it `None`.
- `page_scores` — the per-page parse/layout/table/ocr scores copied out of that
report. They are written to `.confidence.json` and reloaded, but no code
outside this module reads them.
- `pages` — the parsed-page cells (`text`, top-left box, `from_ocr`,
`confidence`), one list per page. docling frees these after assembling the
document unless `generate_parsed_pages=True`; capturing them is what makes the
OCR-vs-native signal available downstream.
- `table_provenance` — derived once per table from the cells under its box: the
fraction read by OCR, the mean OCR confidence, and a native/ocr/empty verdict.
A table whose region carries no native text layer (verdict `ocr`, or `empty`
when Camelot also finds nothing) is a table rendered as an image.
Parsed cells are stored top-left in PDF points. A table's box stays in the
DoclingDocument in docling's own coordinate origin. `_table_provenance` flips it
to top-left only while comparing it with the cells under it.
) annotationsN) dataclassfield)Path)
TYPE_CHECKINGDictListLiteralOptionalTuple)ConfidenceReport)ConversionResult) CoordOrigin)DoclingDocument TableItem)BoundingRectangle) BaseModelFieldc h \ rS rSr% SrS\S' \" SS9rS\S' \" S S9rS
\S ' \" S S
S9r S\S' Sr
g)
ParsedCell3 z6One parsed-page text cell, box in top-left PDF points.strtextz-(x0, y0, x1, y1), top-left origin, PDF points descriptionBoxboxz1True if docling read this cell off the page imageboolfrom_ocr z%docling's OCR confidence for the cell)defaultr float
confidence N) __name__
__module__ __qualname____firstlineno____doc____annotations__r r r r# __static_attributes__r$ 7/home/mande/repo/quber/src/quber/core/parsers/result.pyr r 3 s< @
I!PQCQ'Z[Hd[ c7^_J_r, r c N \ rS rSr% SrS\S' S\S' S\S' \" \S9rS \S
' S r g )
PageParse< zEThe parsed-page cells for one page, with the page's point dimensions.intpage_nor" widthheightdefault_factoryzList[ParsedCell]cellsr$ N)
r% r& r' r( r) r* r listr7 r+ r$ r, r- r/ r/ < s$ O
L L M #D 9E 9r, r/ c L \ rS rSr% SrS\S' S\S' S\S' S\S' S\S ' S
rg ) PageScoreE zGdocling's per-page confidence scores (NaN where docling did not score).r1 r2 r" parse_score layout_score table_score ocr_scorer$ N)r% r& r' r( r) r* r+ r$ r, r- r: r: E s" Q
Lr, r: c \ rS rSr% Sr\" SS9rS\S' S\S' \" S S9rS\S
' \" S S9r S \S
' \" SS9r
S \S' \" SS9r S\S' Sr g)TableProvenanceO zJOCR-vs-native verdict for one docling table, from the cells under its box.z0The TableItem.self_ref this provenance describesr r self_refr1 r2 z5Parsed-page cells whose center falls in the table box
cell_countz+Fraction of those cells docling read by OCRr" from_ocr_fractionz'Mean confidence over the OCR-read cellsmean_ocr_confidenceznative: the table box has a real text layer. ocr: docling read it off the page image (a table rendered as an image). empty: no cells fall in the box at all.z!Literal['native', 'ocr', 'empty']verdictr$ N)
r% r& r' r( r) r rC r* rD rE rF rG r+ r$ r, r- rA rA O sc T&XYHcY
L(_`J` $1^ _u_!&3\!]]16
2G
. r, rA c \ rS rSr% SrS\S' SrS\S' \" \S9r S \S
' \" \S9r
S \S ' \" \S9r S
\S' SS jr SS jr
\SS j5 r\SS j5 rSrg) ParseResult` zEverything one parse produced: the document and docling's side signals.
`document` is the canonical IR. The rest is docling's confidence report, the
per-page scores, the parsed-page cells, and the per-table OCR/native verdict
derived from those cells.
r documentNzOptional[ConfidenceReport]r# r5 zList[PageScore] page_scoresList[PageParse]pagesList[TableProvenance]table_provenancec Z U R Vs0 s H oR U_M sn$ s snf )z3Per-table provenance keyed by `TableItem.self_ref`.)rP rC )selfps r- provenance_by_ref ParseResult.provenance_by_refo s( '+'<'<='.docling.json` — the DoclingDocument.
- `.confidence.json` — per-page scores + per-table provenance.
- `.cells.json` — the parsed-page cells.
Returns the written paths keyed by `document` / `confidence` / `cells`.
T)parentsexist_ok
.docling.json.confidence.json .cells.json)rL rP )indentr! zutf-8)encoding)rK r# r7 ) mkdirrK save_as_json
write_textjsondumpsrL
model_dumprP r rN )rR directorybase
document_pathconfidence_path
cells_pathsrS s r- saveParseResult.saves s t4!fM$::
#.>&?? 6!55
""=1"" JJ<@T2TU **iF:J2K&K%V%V%XY/>/B/B=RT/UV/U!y~1~/U V:I:M:MN`bd:ef:eQO 0a 0:efZZv[-A!A L L NO
)34A Q 4 # -
Wf4s C)C.C3c J UR n/ nUR bz [ UR R R 5 5 HI u pEUR [
UUR UR UR UR S95 MK / nUR H nUR ( a UR R OSnUR ( a UR R OSn UR V
s/ s HI n
[ U
R [# U
R$ 5 U
R& U
R =( d SS9PMK n n
UR [) UR* XU S95 M [- X&5 n U " UUR UUU S9$ s sn
f )aV Build a ParseResult from a docling ConversionResult.
Reads the per-page scores and the parsed-page cells off the result, then
derives the per-table provenance from the cells under each table box.
Requires the converter to have run with `generate_parsed_pages=True`;
without it `result.pages[i].cells` is empty and every table reads `empty`.
)r2 r< r= r> r? r )r r r r# )r2 r3 r4 r7 rn )rK r# sortedrN itemsappendr: r< r= r> r? sizer3 r4 r7 r r _cell_box_top_leftrectr r/ r2 derive_table_provenance)
rs resultrK rL r2 scoresrN pager3 r4 cellr7
provenances
r- from_conversionParseResult.from_conversion sg ??') (#)&*;*;*A*A*G*G*I#J"" '$*$6$6%+%8%8$*$6$6"("2"2 $K "$LLD'+yyDIIOOcE)-TYY%%F !JJ 'D
*4995 !]] # 5# '
LL4<K > "4 8E?8.3D.I+I>!_F
* .
.
r, rI c U Vs0 s H o"R U_M nnU R Vs/ s H oDR ( d M [ XC5 PM" sn$ s snf s snf )aV The OCR-versus-native verdict for every table in `document`.
Purely geometric: each table's box is compared against the parsed-page cells
it encloses. It runs once, in `ParseResult.from_conversion`, on the document
docling just produced. Fusion looks the stored verdicts up by each table's
parse-time `self_ref` through `provenance_by_ref`.
)r2 tablesprov_table_provenance)rK rN rS pages_by_notables r- r r sN *//A99a<K/?G ]eR\R\ 1 e 1 ]] 0 ]s AAAc U R U R U R U R /nU R U R
U R U R /n[ U5 [ U5 [ U5 [ U5 4$ )zMA parsed-cell rectangle (four corner points) -> a top-left (x0,y0,x1,y1) box.)
r_x0r_x1r_x2r_x3r_y0r_y1r_y2r_y3minmax)r~ xsyss r- r} r} s]
))TYY 499 5B
))TYY 499 5B GSWc"gs2w //r, c 0 U R S R nUR [ R :X a. UR
UR UR UR 4$ UR
XR -
UR XR -
4$ )zA table's provenance bbox -> a top-left (x0,y0,x1,y1) box in PDF points.
docling table boxes are bottom-left origin; flip y against the page height so
they share the parsed cells' top-left frame.
r ) r bbox coord_originr TOPLEFTltrb)r page_heightr s r- _table_box_top_leftr so ::a= D K / //// FFK&&($&&+2F GGr, c ( U S U S - S- U S U S - S- p2[ US US 5 [ US US 5 [ US US 5 [ US US 5 4u pEpgXBs=:* =( a U:* Os =( a XSs=:* =( a U:* $ s $ )Nr r\ )r r )r regioncxcyx0y0x1y1s r-
_center_inr s !fs1vo
"SVc!f_$9 F1Ivay! F1Ivay! F1Ivay! F1Ivay! NBB >>r> ,bnn"n,n,r, c p U R S R nUR U5 nU R nUc [ UUSSSSS9$ [ XR 5 nUR Vs/ s H" n[ UR U5 ( d M UPM$ nnU( d [ UUSSSSS9$ U Vs/ s H ofR ( d M UPM nn[ U5 [ U5 - n U( a [ S U 5 5 [ U5 - OSn
U S: a SOSn [ UU[ U5 U U
U S9$ s snf s snf ) zIDerive the OCR/native verdict for one table from the cells under its box.r r empty)rC r2 rD rE rF rG c 3 8 # U H oR v M g 7f)N)r# ).0cs r- $_table_provenance..& s 59a\\9s g ?ocrnative)
r r2 rr rC rA r r4 r7 r r r lensum) r r r2 r rC r r under ocr_cellsrE mean_confrG s r- r r s/ jjm##G ??7 #D~~H | ! #
!
4F
@
1j&?Q
E @ ! #
"0EqZZEI0IU3KT5955IFZ]I:Kc:QW_G u:+ %
A 1s + D.D./D3D3)rK r rN rM r rO )r~ r r r )r r r r" r r )r r r r r r )r r r zDict[int, PageParse]r rA )*r)
__future__r rb dataclassesr r pathlibr typingr r r r
r r docling.datamodel.base_modelsr
docling.datamodel.documentr docling_core.types.doc.baser docling_core.types.doc.documentr r docling_core.types.doc.pager pydanticr r r" r r r/ r: rA rI r r} r r r r$ r, r- r s 4 # ( G F>; 3 F 9 % E5% &'` `: : i " z
z
z
z ^0 H-'r,