!dj | S r SSKJr SSKrSSKJrJr SSKJr " S S\5 r SS jr
\R " S \R \R - 5 r\R " S
\R \R - 5 r\R " S \R \R - 5 rSS jrSS
jrSS jrSS jrg)a Chart selection from an ADE parse response.
`select_charts` takes the raw parse response dict and returns one record per
chart the parse recognized, carrying the page, the grounding box, the
description text, and the reconstructed data table when one came back.
The two API generations mark charts differently. A v1 response carries a
flat `chunks` list; figures are chunks of type `figure`, and a figure is
counted as a chart when our figure prompt made the model emit an HTML data
table inside it. A v2 response marks charts affirmatively: the markdown
wraps each figure in a `` element, and only elements
typed CHART are selected.
) annotationsN)AnyOptional) BaseModelc ` \ rS rSr% SrSrS\S' SrS\S' SrS \S
' Sr S \S ' Sr
S \S
' Sr g) ChartRecord z One chart the parse recognized.Nz
Optional[int]pagezOptional[dict[str, float]]box str descriptionz
Optional[str]
table_html source_id ) __name__
__module__ __qualname____firstlineno____doc__r
__annotations__r r r r __static_attributes__r z2]*\btype=\"CHART\"[^>]*>(.*?)z (.*?)c / nU H nUR S5 S:w a M UR S5 =( d Sn[ R U5 nU( d MR UR S5 =( d 0 nUR S5 nUR [ Ub US- OSUR S 5 [ R SU5 R
5 UR S
5 UR S 5 S 95 M U$ )
zNFigures whose prompted response includes a data table (v1 has no CHART label).typefiguremarkdownr groundingr
N r r idr
r r r r )r TABLE_REsearchappendr substripgroup)r recordschunkmdtabler( r
s r r r 2 s !#G 99V (
YYz
"
(b# IIk*0b }}V$ "&"2dQh MM%( $LLR 0 6 6 8 ;;q> ))D/
$ Nr c ^ [ U R S5 =( d 0 5 n/ n[ R U R S5 =( d S5 GH mTR S5 n[
R
U5 n[ R
U5 n[ U4S jU 5 S5 nU=( d 0 R S5 =( d 0 nUR [ UR S5 UR S 5 U( a UR S5 R 5 OSU( a UR S
5 OSU=( d 0 R S 5 S 95 GM U$ )
a Figure elements the v2 parse typed CHART in the response markdown.
Page and box come from the structure tree: every figure node's grounding
carries a character range into the markdown, so each CHART element is
matched to the figure node whose range contains its start offset.
structurer' r r) c 3 j ># U H( n[ UTR 5 5 ( d M$ Uv M* g 7f)N)_range_containsstart).0fmatchs r $_from_v2_markdown..W s" M1?1ekkm+LQQs #3 3Nr( r
r r r* r+ ) _v2_figure_nodesr V2_FIGURE_REfinditerr1 V2_DESCRIPTION_REr- r, nextr. r r0 ) r" figuresr2 bodydescr5 noder( r= s @r r! r! J s sww{ 3 9r:G!#G &&swwz':'@bA{{1~ ''-%MMtTZR$$[17R ]]6* MM%(6:TZZ] 0 0 2.3EKKN :2**40
B Nr c 0 ^^ / mSUU4S jjmT" U 5 T$ )z:All figure nodes in the structure tree, in document order.c > U R S5 S:X a TR U 5 U R S5 =( d / H nT" U5 M
g )Nr% r& children)r r. )rH childfoundwalks r rN _v2_figure_nodes..walki sB 88F x ' LL XXj)/R/E K 0r )rH dict[str, Any]returnNoner )r7 rM rN s @@r r@ r@ e s "$E O Lr c U R S5 =( d 0 R S5 =( d 0 nUR S5 UR S5 pCUSL=( a USL=( a X1s=:* =( a U: $ s $ )zBTrue when the figure's markdown character range covers ``offset``.r( ranger: endN)r )r& offsetrngr: rU s r r9 r9 s sg ::k " (b
-
-g
6
<"C!3775>3 JD JU5J5Js5JJ5JJr )r" rP rQ list[ChartRecord])r list[dict[str, Any]]rQ rX )r7 rP rQ rY )r& rP rV intrQ bool)r
__future__r retypingr r pydanticr r r# compileDOTALL
IGNORECASEr, rA rC r r! r@ r9 r r r rc s # $) $ " ::,bii"--.G HzzOQSQZQZ]_]j]jQjk JJ BBIIPRP]P]D]^ 06 Kr