A real Czech filing

One public filing, two official sources: the scanned annual report of Strojírny Prostějov s.r.o. for 2024 and the XML statements filed with it. The scan is read with the open-source Tesseract OCR, rows are anchored on their printed row codes, and the accounting identities of the vyhláška 500 form decide which readings to re-check. The XML is used only to score the result, never to choose a value.

Source

Company: Strojírny Prostějov s.r.o., IČO 05516269
Registry: Sbírka listin, veřejný rejstřík (or.justice.cz) — filing record
Scan: Výroční zpráva s výrokem auditora 2024 (komprimovaný).pdf (PDF), embedded page images at about 72 dpi
XML: UZ-462100029424-20250627-112807.xml (XML), schema version 06.03, amounts in thousands of CZK

Both files are pinned by SHA-256 in data/manifests/czech_filings.yaml. DealBench has no affiliation with the company or the registry; only the statement tables of public pages are shown, without the signature boxes. Takedown requests are honoured.

Rows anchored

198 / 198

169 exact code, 17 near, 12 by position

Cells matching the XML

173 → 532 of 550

page OCR → after identity repair

Page-OCR misreads

373 of 550

all 373 among the 464 cells an identity flagged; 366 fixed, 0 introduced

Identities holding on the scan

14 → 230 of 239

page OCR → repaired; the XML: 238

Sources disagree

cells; all 11 agree with the value read by eye

Honest reading. The page OCR alone gets 32.2 % of the cells right (31.5 % match the XML), mostly because a lone 0 at 72 dpi is rarely read: 279 cells have no reading at all. A misread counts as caught when an identity pointed at its cell, and that is a low bar here: with 193 of 239 identities unreadable on the page OCR, identities flagged 464 of the 550 cells, 91 of them already right. So "all 373 misreads caught" mostly says that 84.4 % of the cells were re-read. After the repair 96.7 % match the XML; 7 cells are still misread; all 7 sit in an identity that still fails, which is where the page says it is unsure. The remaining 11 differences are not OCR errors: the scan's own identities hold on them. Right means equal to the reference: the XML, except in the 11 cells where the sources disagree, where it is the printed value read by eye.

The sources disagree

The scan and the XML give different prior-year values in 11 asset cells. Within the XML itself, prior-year total assets (223 620) and total liabilities (228 364) differ; on the scan both are 228 364. Every difference traces to 3 rows:

017 Stavby: The XML's prior-year value equals its own current-year value for this row.
019 Oceňovací rozdíl k nabytému majetku: The XML's prior-year value equals its own current-year value for this row.
066 Dohadné účty aktivní: The XML has no value for this row; the scan prints one.

The other 8 are subtotals that carry those differences up to total assets. Wording rule: the sources disagree; this page does not say which one is right.

Cells where the scan and the XML disagree
Řád	Položka	Scan, OCR	XML	Why
001	AKTIVA CELKEM	228 364	223 620	A subtotal: the difference of 4 744 is the sum of the differences in rows 003, 037.
003	Stálá aktiva	111 144	106 428	A subtotal: the difference of 4 716 is the sum of the differences in row 014.
014	Dlouhodobý hmotný majetek	98 983	94 267	A subtotal: the difference of 4 716 is the sum of the differences in rows 015, 019.
015	Pozemky a stavby	44 805	42 750	A subtotal: the difference of 2 055 is the sum of the differences in row 017. The XML's prior-year value equals its own current-year value for this row.
017	Stavby	40 682	38 627	The XML's prior-year value equals its own current-year value for this row.
019	Oceňovací rozdíl k nabytému majetku	39 910	37 249	The XML's prior-year value equals its own current-year value for this row.
037	Oběžná aktiva	116 838	116 810	A subtotal: the difference of 28 is the sum of the differences in row 046.
046	Pohledávky	50 079	50 051	A subtotal: the difference of 28 is the sum of the differences in row 057.
057	Krátkodobé pohledávky	50 079	50 051	A subtotal: the difference of 28 is the sum of the differences in row 061.
061	Pohledávky – ostatní	2 744	2 716	A subtotal: the difference of 28 is the sum of the differences in row 066.
066	Dohadné účty aktivní	28	absent	The XML has no value for this row; the scan prints one.

Identity checks

239 identities generated from the row tree

Identity status by kind and reading
Identity	Page OCR: holds / fails / unreadable	Repaired	XML
Subtotals (parent = Σ children, per column)	7 / 21 / 130	151 / 7 / 0	158 / 0 / 0
Netto = brutto + korekce (assets)	6 / 9 / 62	75 / 2 / 0	77 / 0 / 0
Aktiva = Pasiva (per period)	0 / 2 / 0	2 / 0 / 0	1 / 1 / 0
Rozvaha 098 = VZZ 55 (result for the period)	1 / 0 / 1	2 / 0 / 0	2 / 0 / 0

Identities that still fail on either source

Failing identities after repair and on the XML
Identity	Scan, repaired	XML
038 = 039 + 040 + 041 + 044 + 045 (netto)	fails 45 613 vs 45 612	holds
041 = 042 + 043 (korekce)	fails −553 vs −533	holds
122 = 123 + 126 + 127 + 128 + 129 + 130 + 131 + 132 (prior)	fails 87 768 vs 81 768	holds
132 = 133 + 134 + 135 + 136 + 137 + 138 + 139 (prior)	fails 22 118 vs 22 028	holds
14 = 15 + 18 + 19 (current)	fails 10 031 vs 11 101	holds
30 = 01 + 02 - 03 - 07 - 08 - 09 - 14 + 20 - 24 (current)	fails 13 138 vs 2 325 690	holds
56 = 01 + 02 (current)	fails 256 959 vs 2 569 511	holds
042: netto = brutto + korekce	fails 14 233 vs 14 253	holds
045: netto = brutto + korekce	fails 25 vs 26	holds
AKTIVA CELKEM 001 = PASIVA CELKEM 078 (prior year)	holds	fails 223 620 vs 228 364

An identity failing on the repaired scan marks a cell the OCR could not settle; all 7 remaining OCR errors are inside these.

Every row: scan, OCR and XML

77 rows

Column Show

Statement rows with the scan, both OCR readings and the XML
Řád	Položka	Page OCR	Repaired	XML	Verdict and OCR passes
001	AKTIVA CELKEM	228	228 364(corrected by the repair)(re-read)	223 620	sources disagree OCR passes Row code read as “001)”; placed on an exact code Page OCR: “228” Re-read x4, x5, x6, x8, x4, binarised at 160, x6, binarised at 160: “228364” flagged; replaced by the re-reads
002	A. Pohledávky za upsaný základní kapitál	0	0(re-read)	0	matches XML OCR passes Row code read as “oe”, re-read “002”; placed on an exact code Page OCR: “0” Re-read x4, x5, x6: “” Re-read x8, x4, binarised at 160, x6, binarised at 160: “0” flagged; the re-reads kept the page OCR value
003	B. Stálá aktiva	111 144	111 144(re-read)	106 428	sources disagree OCR passes Row code read as “003\|”; placed on an exact code Page OCR: “111 144]” Re-read x4, x5, x8, x6, binarised at 160: “111144” Re-read x6: “” Re-read x4, binarised at 160: “111143” flagged; the re-reads kept the page OCR value
004	B.I. Dlouhodobý nehmotný majetek	9 501	1 959(corrected by the repair)(re-read)	1 959	matches XML OCR passes Row code read as “004)”; placed on an exact code Page OCR: “950] 1” Re-read x4, x6, x8, x4, binarised at 160, x6, binarised at 160: “1959” Re-read x5: “” flagged; replaced by the re-reads
005	B.I.1 Nehmotné výsledky vývoje	—	0(corrected by the repair)(re-read)	0	matches XML OCR passes Row code read as “005”; placed on an exact code Page OCR: “o” Re-read x4, x5, x6, x8: “0” Re-read x4, binarised at 160: “c” Re-read x6, binarised at 160: “C” flagged; replaced by the re-reads
006	B.I.2 Ocenitelná práva	1 959	1 959(re-read)	1 959	matches XML OCR passes Row code read as “006”; placed on an exact code Page OCR: “1 959” Re-read x4, x5, x6, x8: “” Re-read x4, binarised at 160: “195” Re-read x6, binarised at 160: “1959” flagged; re-reads disagreed, the identities chose
007	B.I.2.1 Software	9 551	1 959(corrected by the repair)(re-read)	1 959	matches XML OCR passes Row code read as “oo7\|”; placed on an exact code Page OCR: “955] 1” Re-read x4, x5: “-” Re-read x6, x8: “” Re-read x4, binarised at 160: “1959” Re-read x6, binarised at 160: “155” flagged; re-reads disagreed, the identities chose
008	B.I.2.2 Ostatní ocenitelná práva	—	0(corrected by the repair)(re-read)	0	matches XML OCR passes Row code read as “008”; placed on an exact code Page OCR: “of” Re-read x4, x5, x6, x8, x6, binarised at 160: “0” Re-read x4, binarised at 160: “6” flagged; replaced by the re-reads
009	B.I.3 Goodwill (+/-)	—	0(corrected by the repair)(re-read)	0	matches XML OCR passes Row code read as “00”; placed on a near code Page OCR: “of” Re-read x4, x5, x6, x8, x4, binarised at 160, x6, binarised at 160: “0” flagged; replaced by the re-reads
010	B.I.4 Ostatní dlouhodobý nehmotný majetek	—	0(corrected by the repair)(re-read)	0	matches XML OCR passes Row code read as “010”; placed on an exact code Page OCR: “of” Re-read x4, x5, x8, x4, binarised at 160, x6, binarised at 160: “0” Re-read x6: “.” flagged; replaced by the re-reads
011	B.I.5 Poskytnuté zálohy na dlouhodobý nehmotný majetek a nedokončený dlouhodobý nehmotný majetek	—	0(corrected by the repair)(re-read)	0	matches XML OCR passes Row code read as “011”; placed on an exact code Page OCR: “o” Re-read x4, x5, x6, x8, x4, binarised at 160, x6, binarised at 160: “0” flagged; replaced by the re-reads
012	B.I.5.1 Poskytnuté zálohy na dlouhodobý nehmotný majetek	—	0(corrected by the repair)(re-read)	0	matches XML OCR passes Row code read as “niž”, re-read “012”; placed on an exact code Page OCR: “ol” Re-read x4, x5, x6, x8, x4, binarised at 160, x6, binarised at 160: “0” flagged; replaced by the re-reads
013	B.I.5.2 Nedokončený dlouhodobý nehmotný majetek	—	0(corrected by the repair)(re-read)	0	matches XML OCR passes Row code read as “013”; placed on an exact code Page OCR: “ol” Re-read x4, x5, x6, x8, x4, binarised at 160, x6, binarised at 160: “0” flagged; replaced by the re-reads
014	B.II. Dlouhodobý hmotný majetek	963	98 983(corrected by the repair)(re-read)	94 267	sources disagree OCR passes Row code read as “014]”; placed on an exact code Page OCR: “se 963)” Re-read x4, x6, x8: “” Re-read x5, x6, binarised at 160: “989683” Re-read x4, binarised at 160: “98983” flagged; re-reads disagreed, the identities chose
015	B.II.1 Pozemky a stavby	44 008	44 805(corrected by the repair)(re-read)	42 750	sources disagree OCR passes Row code read as “015]”; placed on an exact code Page OCR: “44008)” Re-read x4, x5, x6, x8, x4, binarised at 160, x6, binarised at 160: “44805” flagged; replaced by the re-reads
016	B.II.1.1 Pozemky	4 123	4 123(re-read)	4 123	matches XML OCR passes Row code read as “016”; placed on an exact code Page OCR: “4123” Re-read x4, x5, x6, x8, x4, binarised at 160, x6, binarised at 160: “4123” flagged; the re-reads kept the page OCR value
017	B.II.1.2 Stavby	40 682	40 682(re-read)	38 627	sources disagree OCR passes Row code read as “017]”; placed on an exact code Page OCR: “40 682” Re-read x4, x5, x6, x8, x4, binarised at 160, x6, binarised at 160: “40682” flagged; the re-reads kept the page OCR value
018	B.II.2 Hmotné movité věci a jejich soubory	14 100	14 100(re-read)	14 100	matches XML OCR passes Row code read as “018]”; placed on an exact code Page OCR: “14 100” Re-read x4, x5, x6, x8, x4, binarised at 160: “14100” Re-read x6, binarised at 160: “1410” flagged; the re-reads kept the page OCR value
019	B.II.3 Oceňovací rozdíl k nabytému majetku	3	39 910(corrected by the repair)(re-read)	37 249	sources disagree OCR passes Row code read as “v1]”, re-read “019”; placed on an exact code Page OCR: “3am” Re-read x4, x5, x6, x8, x4, binarised at 160, x6, binarised at 160: “39910” flagged; replaced by the re-reads
020	B.II.4 Ostatní dlouhodobý hmotný majetek	—	0(corrected by the repair)(re-read)	0	matches XML OCR passes Row code read as “020”; placed on an exact code Page OCR: “of” Re-read x4, x5, x6, x8, x4, binarised at 160, x6, binarised at 160: “0” flagged; replaced by the re-reads
021	B.II.4.1 Pěstitelské celky trvalých porostů	—	0(corrected by the repair)(re-read)	0	matches XML OCR passes Row code read as “021”; placed on an exact code Page OCR: “of” Re-read x4, x5, x6, x8, x4, binarised at 160, x6, binarised at 160: “0” flagged; replaced by the re-reads
022	B.II.4.2 Dospělá zvířata a jejich skupiny	—	0(corrected by the repair)(re-read)	0	matches XML OCR passes Row code read as “022”; placed on an exact code Page OCR: “of” Re-read x4, x4, binarised at 160, x6, binarised at 160: “0” Re-read x5: “°” Re-read x6, x8: “c” flagged; replaced by the re-reads
023	B.II.4.3 Jiný dlouhodobý hmotný majetek	—	0(corrected by the repair)(re-read)	0	matches XML OCR passes Row code read as “023”; placed on an exact code Page OCR: “qf” Re-read x4, x5, x6, x8, x4, binarised at 160, x6, binarised at 160: “0” flagged; replaced by the re-reads
024	B.II.5 Poskytnuté zálohy na dlouhodobý hmotný majetek a nedokončený dlouhodobý hmotný majetek	—	168(corrected by the repair)(re-read)	168	matches XML OCR passes Row code read as “ry”, re-read “024”; placed on an exact code Page OCR: “ee]” Re-read x4, x6, x8: “” Re-read x5, x4, binarised at 160, x6, binarised at 160: “168” flagged; replaced by the re-reads
025	B.II.5.1 Poskytnuté zálohy na dlouhodobý hmotný majetek	—	0(corrected by the repair)(re-read)	0	matches XML OCR passes Row code read as “025”; placed on an exact code Page OCR: “of” Re-read x4, x5, x6, x8, x4, binarised at 160, x6, binarised at 160: “0” flagged; replaced by the re-reads
026	B.II.5.2 Nedokončený dlouhodobý hmotný majetek	188	168(corrected by the repair)(re-read)	168	matches XML OCR passes Row code read as “026”; placed on an exact code Page OCR: “188]” Re-read x4: “” Re-read x5, x6, x8, x4, binarised at 160, x6, binarised at 160: “168” flagged; replaced by the re-reads
027	B.III. Dlouhodobý finanční majetek	10 202	10 202(re-read)	10 202	matches XML OCR passes Row code read as “027\|”; placed on an exact code Page OCR: “10202” Re-read x4, x5, x6, x8, x4, binarised at 160, x6, binarised at 160: “10202” flagged; the re-reads kept the page OCR value
028	B.III.1 Podíly – ovládaná nebo ovládající osoba	202	202	202	matches XML OCR passes Row code read as “028”; placed on an exact code Page OCR: “202]” page OCR, no identity pointed at it
029	B.III.2 Zápůjčky a úvěry – ovládaná nebo ovládající osoba	—	0(corrected by the repair)(re-read)	0	matches XML OCR passes Row code read as “axa”, re-read “029”; placed on an exact code Page OCR: “of” Re-read x4, x5, x6, x8, x4, binarised at 160, x6, binarised at 160: “0” flagged; replaced by the re-reads
030	B.III.3 Podíly – podstatný vliv	—	0(corrected by the repair)(re-read)	0	matches XML OCR passes Row code read as “030”; placed on an exact code Page OCR: “of” Re-read x4, x5, x6, x8, x4, binarised at 160, x6, binarised at 160: “0” flagged; replaced by the re-reads
031	B.III.4 Zápůjčky a úvěry – podstatný vliv	—	0(corrected by the repair)(re-read)	0	matches XML OCR passes Row code read as “031”; placed on an exact code Page OCR: “of” Re-read x4, x5, x6, x8, x4, binarised at 160, x6, binarised at 160: “0” flagged; replaced by the re-reads
032	B.III.5 Ostatní dlouhodobé cenné papíry a podíly	—	0(corrected by the repair)(re-read)	0	matches XML OCR passes Row code read as “032”; placed on an exact code Page OCR: “of” Re-read x4, x5, x6, x8, x4, binarised at 160, x6, binarised at 160: “0” flagged; replaced by the re-reads
033	B.III.6 Zápůjčky a úvěry – ostatní	—	0(corrected by the repair)(re-read)	0	matches XML OCR passes Row code read as “033”; placed on an exact code Page OCR: “of” Re-read x4, x5, x6, x8, x4, binarised at 160, x6, binarised at 160: “0” flagged; replaced by the re-reads
034	B.III.7 Ostatní dlouhodobý finanční majetek	10 000	10 000(re-read)	10 000	matches XML OCR passes Row code read as “oa4\|”, re-read “034”; placed on an exact code Page OCR: “10000]” Re-read x4, x5, x6, x8, x4, binarised at 160, x6, binarised at 160: “10000” flagged; the re-reads kept the page OCR value
035	B.III.7.1 Jiný dlouhodobý finanční majetek	10	10 000(corrected by the repair)(re-read)	10 000	matches XML OCR passes Row code read as “035]”; placed on an exact code Page OCR: “10” Re-read x4: “1000” Re-read x5, x6, x8: “10000” Re-read x4, binarised at 160, x6, binarised at 160: “100” flagged; replaced by the re-reads
036	B.III.7.2 Poskytnuté zálohy na dlouhodobý finanční majetek	—	0(corrected by the repair)(re-read)	0	matches XML OCR passes Row code read as “036”; placed on an exact code Page OCR: “=” Re-read x4, x5, x6, x8, x4, binarised at 160, x6, binarised at 160: “0” flagged; replaced by the re-reads
037	C. Oběžná aktiva	—	116 838(corrected by the repair)(re-read)	116 810	sources disagree OCR passes Row code read as “037\|”; placed on an exact code Page OCR: “3a) 116” Re-read x4, x8: “” Re-read x5, x6, x6, binarised at 160: “116838” Re-read x4, binarised at 160: “116834” flagged; replaced by the re-reads
038	C.I. Zásoby	56 057	56 057(re-read)	56 057	matches XML OCR passes Row code read as “038]”; placed on an exact code Page OCR: “56057” Re-read x4, x5, x6, x8, x4, binarised at 160: “56057” Re-read x6, binarised at 160: “36057” flagged; the re-reads kept the page OCR value
039	C.I.1 Materiál	15 048	15 048	15 048	matches XML OCR passes Row code read as “p39]”; placed on a near code Page OCR: “15 048}” page OCR, no identity pointed at it
040	C.I.2 Nedokončená výroba a polotovary	—	15 768(corrected by the repair)(re-read)	15 768	matches XML OCR passes Row code read as “pen\|”, re-read “040”; placed on an exact code Page OCR: “” Re-read x4, x5, x6, x8, x4, binarised at 160, x6, binarised at 160: “15768” flagged; replaced by the re-reads
041	C.I.3 Výrobky a zboží	—	25 241(corrected by the repair)(re-read)	25 241	matches XML OCR passes Row code read as “pet]”, re-read “041”; placed on an exact code Page OCR: “i 2s2an” Re-read x4, x5, x6, x8, x4, binarised at 160, x6, binarised at 160: “25241” flagged; replaced by the re-reads
042	C.I.3.1 Výrobky	2 521	25 241(corrected by the repair)(re-read)	25 241	matches XML OCR passes Row code read as “o42\|”; placed on an exact code Page OCR: “\| 2521)” Re-read x4, x5, x6, x8, x4, binarised at 160, x6, binarised at 160: “25241” flagged; replaced by the re-reads
043	C.I.3.2 Zboží	—	0(corrected by the repair)(re-read)	0	matches XML OCR passes Row code read as “043”; placed on an exact code Page OCR: “” Re-read x4, x5, x6, x8: “0” Re-read x4, binarised at 160: “¢” Re-read x6, binarised at 160: “C” flagged; replaced by the re-reads
044	C.I.4 Mladá a ostatní zvířata a jejich skupiny	—	0(corrected by the repair)(re-read)	0	matches XML OCR passes Row code read as “044”; placed on an exact code Page OCR: “” Re-read x4, x5, x6, x8: “0” Re-read x4, binarised at 160: “<” Re-read x6, binarised at 160: “¢” flagged; replaced by the re-reads
045	C.I.5 Poskytnuté zálohy na zásoby	—	0(corrected by the repair)(re-read)	0	matches XML OCR passes Row code read as “045”; placed on an exact code Page OCR: “i]” Re-read x4, x5, x6, x8, x4, binarised at 160, x6, binarised at 160: “0” flagged; replaced by the re-reads
046	C.II. Pohledávky	—	50 079(corrected by the repair)(re-read)	50 051	sources disagree OCR passes Row code read as “oas\|”, re-read “212”; placed by position between anchored rows Page OCR: “soo” Re-read x4, x5, x6, x8: “50079” Re-read x4, binarised at 160, x6, binarised at 160: “0079” flagged; replaced by the re-reads
047	C.II.1 Dlouhodobé pohledávky	—	0(corrected by the repair)(re-read)	0	matches XML OCR passes Row code read as “naz\|”, re-read “047”; placed on an exact code Page OCR: “ol” Re-read x4, x5, x6, x8, x4, binarised at 160, x6, binarised at 160: “0” flagged; replaced by the re-reads
048	C.II.1.1 Pohledávky z obchodních vztahů	—	0(corrected by the repair)(re-read)	0	matches XML OCR passes Row code read as “04a”; placed on a near code Page OCR: “of” Re-read x4, x5, x6, x8, x4, binarised at 160, x6, binarised at 160: “0” flagged; replaced by the re-reads
049	C.II.1.2 Pohledávky – ovládaná nebo ovládající osoba	—	0(corrected by the repair)(re-read)	0	matches XML OCR passes Row code read as “pas]”, re-read “049”; placed on an exact code Page OCR: “a” Re-read x4, x5, x6, x8, x4, binarised at 160, x6, binarised at 160: “0” flagged; replaced by the re-reads
050	C.II.1.3 Pohledávky – podstatný vliv	—	0(corrected by the repair)(re-read)	0	matches XML OCR passes Row code read as “050”; placed on an exact code Page OCR: “ol” Re-read x4, x5, x6, x8, x4, binarised at 160, x6, binarised at 160: “0” flagged; replaced by the re-reads
051	C.II.1.4 Odložená daňová pohledávka	—	0(corrected by the repair)(re-read)	0	matches XML OCR passes Row code read as “051”; placed on an exact code Page OCR: “ol” Re-read x4, x5, x6, x8, x4, binarised at 160, x6, binarised at 160: “0” flagged; replaced by the re-reads
052	C.II.1.5 Pohledávky – ostatní	—	0(corrected by the repair)(re-read)	0	matches XML OCR passes Row code read as “052”; placed on an exact code Page OCR: “of” Re-read x4, x5, x6, x8, x4, binarised at 160, x6, binarised at 160: “0” flagged; replaced by the re-reads
053	C.II.1.5.1 Pohledávky za společníky	—	0(corrected by the repair)(re-read)	0	matches XML OCR passes Row code read as “053”; placed on an exact code Page OCR: “” Re-read x4, x5, x6, x8, x4, binarised at 160, x6, binarised at 160: “0” flagged; replaced by the re-reads
054	C.II.1.5.2 Dlouhodobé poskytnuté zálohy	—	0(corrected by the repair)(re-read)	0	matches XML OCR passes Row code read as “054”; placed on an exact code Page OCR: “o” Re-read x4, x5, x6, x8, x4, binarised at 160, x6, binarised at 160: “0” flagged; replaced by the re-reads
055	C.II.1.5.3 Dohadné účty aktivní	0	0	0	matches XML OCR passes Row code read as “055”; placed on an exact code Page OCR: “0” page OCR, no identity pointed at it
056	C.II.1.5.4 Jiné pohledávky	—	0(corrected by the repair)(re-read)	0	matches XML OCR passes Row code read as “056”; placed on an exact code Page OCR: “o” Re-read x4, x5, x6, x8, x4, binarised at 160, x6, binarised at 160: “0” flagged; replaced by the re-reads
057	C.II.2 Krátkodobé pohledávky	50 079	50 079	50 051	sources disagree OCR passes Row code read as “gs7\|”, re-read “057”; placed on an exact code Page OCR: “50 079}” page OCR, no identity pointed at it
058	C.II.2.1 Pohledávky z obchodních vztahů	36 765	36 765	36 765	matches XML OCR passes Row code read as “osa\|”; placed by position between anchored rows Page OCR: “36765” page OCR, no identity pointed at it
059	C.II.2.2 Pohledávky – ovládaná nebo ovládající osoba	10 570	10 570	10 570	matches XML OCR passes Row code read as “059”; placed on an exact code Page OCR: “10 570” page OCR, no identity pointed at it
060	C.II.2.3 Pohledávky – podstatný vliv	0	0	0	matches XML OCR passes Row code read as “060”; placed on an exact code Page OCR: “0” page OCR, no identity pointed at it
061	C.II.2.4 Pohledávky – ostatní	2 744	2 744(re-read)	2 716	sources disagree OCR passes Row code read as “061”; placed on an exact code Page OCR: “2744” Re-read x4, x5, x6, x8, x4, binarised at 160, x6, binarised at 160: “2744” flagged; the re-reads kept the page OCR value
062	C.II.2.4.1 Pohledávky za společníky	—	0(corrected by the repair)(re-read)	0	matches XML OCR passes Row code read as “0&2”; placed on a near code Page OCR: “of” Re-read x4, x5, x6, x8, x4, binarised at 160, x6, binarised at 160: “0” flagged; replaced by the re-reads
063	C.II.2.4.2 Sociální zabezpečení a zdravotní pojištění	—	0(corrected by the repair)(re-read)	0	matches XML OCR passes Row code read as “ex)”, re-read “063”; placed on an exact code Page OCR: “o\|” Re-read x4, x5, x6, x8, x4, binarised at 160, x6, binarised at 160: “0” flagged; replaced by the re-reads
064	C.II.2.4.3 Stát – daňové pohledávky	—	2 245(corrected by the repair)(re-read)	2 245	matches XML OCR passes Row code read as “084”; placed on a near code Page OCR: “” Re-read x4, x5, x6, x8, x4, binarised at 160, x6, binarised at 160: “2245” flagged; replaced by the re-reads
065	C.II.2.4.4 Krátkodobé poskytnuté zálohy	22 155	195(corrected by the repair)(re-read)	195	matches XML OCR passes Row code read as “065”; placed on an exact code Page OCR: “22 155” Re-read x4, x5, x6, x8, x4, binarised at 160, x6, binarised at 160: “195” flagged; replaced by the re-reads
066	C.II.2.4.5 Dohadné účty aktivní	—	28(corrected by the repair)(re-read)	0	sources disagree OCR passes Row code read as “DES”, re-read “056”; placed on a near code Page OCR: “za]” Re-read x4, x5, x6, x8, x4, binarised at 160: “28” Re-read x6, binarised at 160: “” flagged; replaced by the re-reads
067	C.II.2.4.6 Jiné pohledávky	278	276(corrected by the repair)(re-read)	276	matches XML OCR passes Row code read as “067”; placed on an exact code Page OCR: “278\|” Re-read x4, x5, x6, x8, x4, binarised at 160, x6, binarised at 160: “276” flagged; replaced by the re-reads
068	C.III. Krátkodobý finanční majetek	—	0(corrected by the repair)(re-read)	0	matches XML OCR passes Row code read as “AA”; placed by position between anchored rows Page OCR: “” Re-read x4, x5, x6, x8, x4, binarised at 160, x6, binarised at 160: “0” flagged; replaced by the re-reads
069	C.III.1 Podíly – ovládaná nebo ovládající osoba	—	0(corrected by the repair)(re-read)	0	matches XML OCR passes Row code read as “Dag”; placed by position between anchored rows Page OCR: “” Re-read x4, x5, x6, x8, x4, binarised at 160, x6, binarised at 160: “0” flagged; replaced by the re-reads
070	C.III.2 Ostatní krátkodobý finanční majetek	—	0(corrected by the repair)(re-read)	0	matches XML OCR passes Row code read as “070”; placed on an exact code Page OCR: “\|]” Re-read x4, x5, x6, x8, x4, binarised at 160, x6, binarised at 160: “0” flagged; replaced by the re-reads
071	C.IV. Peněžní prostředky	10	10 702(corrected by the repair)(re-read)	10 702	matches XML OCR passes Row code read as “oT”, re-read “071”; placed on an exact code Page OCR: “10” Re-read x4, x5, x6, x8, x4, binarised at 160, x6, binarised at 160: “10702” flagged; replaced by the re-reads
072	C.IV.1 Peněžní prostředky v pokladně	—	38(corrected by the repair)(re-read)	38	matches XML OCR passes Row code read as “úřž”, re-read “072”; placed on an exact code Page OCR: “” Re-read x4: “KJ” Re-read x5: “38” Re-read x6, x8: “3” Re-read x4, binarised at 160: “3e” Re-read x6, binarised at 160: “K2” flagged; re-reads disagreed, the identities chose
073	C.IV.2 Peněžní prostředky na účtech	10	10 664(corrected by the repair)(re-read)	10 664	matches XML OCR passes Row code read as “ga”, re-read “073”; placed on an exact code Page OCR: “M 10” Re-read x4, x5, x6, x8, x4, binarised at 160, x6, binarised at 160: “10664” flagged; replaced by the re-reads
074	D. Časové rozlišení aktiv	2	382(corrected by the repair)(re-read)	382	matches XML OCR passes Row code read as “E”, re-read “074”; placed on an exact code Page OCR: “a2]” Re-read x4, x5, x6, x4, binarised at 160, x6, binarised at 160: “382” Re-read x8: “” flagged; replaced by the re-reads
075	D.1 Náklady příštích období	—	382(corrected by the repair)(re-read)	382	matches XML OCR passes Row code read as “075]”; placed on an exact code Page OCR: “” Re-read x4, x5, x8, x4, binarised at 160, x6, binarised at 160: “382” Re-read x6: “” flagged; replaced by the re-reads
076	D.2 Komplexní náklady příštích období	—	0(corrected by the repair)(re-read)	0	matches XML OCR passes Row code read as “076”; placed on an exact code Page OCR: “r” Re-read x4, x5, x6, x8, x4, binarised at 160, x6, binarised at 160: “0” flagged; replaced by the re-reads
077	D.3 Příjmy příštích období	—	0(corrected by the repair)(re-read)	0	matches XML OCR passes Row code read as “077”; placed on an exact code Page OCR: “” Re-read x4, x5, x6, x8, x4, binarised at 160, x6, binarised at 160: “0” flagged; replaced by the re-reads

Red: differs from the reference (the XML, or the printed value where the sources disagree). Green with ✓: corrected by the repair. Italic XML values are rows the XML omits (it leaves out zero rows), read as 0. * the cell was re-read because an identity pointed at it; open “OCR passes” in a row for every reading of the cell and of its row code.

Method

Page OCR: Tesseract psm 3, ces+eng on the page rendered at 300 dpi (the same parser as the synthetic room).
Re-reads: crops of the embedded image upscaled x4; x5; x6; x8; x4, binarised at 160; x6, binarised at 160, one line of digits each. The scan is about 72 dpi, so rendering the PDF at a higher resolution would add nothing.
Pages are found by matching each page's OCR text against the statements' row labels and kept only if they carry the statement's ruled form.
Rows are anchored on the printed row code (řád): the page OCR's code when it reads cleanly, otherwise a re-read of the code cell, aligned to the vyhláška 500 row order.
Every value is first read by the page OCR. Cells that a failing or unreadable identity points at are re-read from crops of the embedded page image, repeated while new cells are pointed at; disagreeing re-reads are settled by the identities, choosing only among values some OCR pass produced.
Korekce is printed as a deduction, so a re-read that lost its minus sign is negated; no other sign is changed.
The XML is never used to choose a value. A reading that differs from the XML counts as a source disagreement only if every identity through it holds on the scan's own values and two OCR passes or two identities back it; otherwise it counts as an OCR error.
No vision model and no API was used for this page: everything runs offline with Tesseract and is cached, so make czech reproduces it.

Also checked: XML aktiva row 78 has no printed counterpart (brutto 0, korekce 0, netto 0, netto_prior 0). XML pasiva row 67 has no printed counterpart (current 0, prior 0).

Source: Sbírka listin, veřejný rejstřík (or.justice.cz), filing record https://or.justice.cz/ias/ui/vypis-sl-detail?dokument=87344310&subjektId=949174&spis=1403435. Public documents of Strojírny Prostějov s.r.o., shown for research. No affiliation with the company; synthetic documents are never attached to it. Takedown requests are honoured.