Filing rules checked¶

The readers record the EBA Filing Rules (version 5.8, February 2026) that a report breaks and that the file alone can show, in Report.filing. RULE_TITLES names each rule; filing rules and coverage explains the approach.

Both formats¶

Rule	Checks
1.5, 2.2, 2.3	One absolute taxonomy entry point.
2.8, 3.6	The reporting subject: scheme (LEI or reporting-subject), suffix such as `.CON` or `.IND` from 31 December 2022 (unless the module is a `_con` or `_ind` module), upper case, and LEI check digits (ISO 17442).
File naming	The file name against `ReportSubject_Country_FrameworkCodeModuleVersion_Module_ReferenceDate_CreationTimestamp`, including the reporting subject, module and reference date. Warnings.

XBRL-XML¶

Rule	Checks
1.4, 1.13	UTF-8, and no standalone declaration.
1.6, 1.6.1	Filing indicators present, unique, and in a context without segment or scenario.
1.14, 1.15, 2.1, 2.4, 2.25	No `xsi:schemaLocation`, XInclude, `xml:base`, `link:linkbaseRef` or footnotes.
2.11, 2.13, 2.14, 2.15	No forever or duration periods, segments, or non-dimensional scenario content.
2.9, 2.10, 2.13	One entity and one reference date, as plain dates.
2.16, 2.16.1	No duplicate facts, and none differing only by unit.
2.17, 2.18	`decimals` rather than `precision`, and present on numeric facts.
2.19	No nil or empty facts.
2.7, 2.21, 2.22	No unused or repeated contexts and units.
3.1	One reporting currency (a), and currency-of-denomination facts in the currency their CUS or CUA member names (c). Rule (c) binds only facts under (b), in contexts with CCA `eba_CA:x1` or qAEA `eba_qCA:qx2000`, so a CUS or CUA member in other contexts is not compared with the unit.
3.2 (a)	`xbrli:pure` for other numeric facts.
3.4, 3.9, 3.10	Namespace prefixes declared once, on the document element, and used.

xBRL-CSV¶

Rule	Checks
Package	`META-INF/reportPackage.json`, `parameters.csv` with `refPeriod`, and `FilingIndicators.csv`.
File naming	The root folder named as the ZIP.
1.5	No `documentInfo.taxonomy`.
2.10	A plain-date `refPeriod`.
1.6.1	Unique filing indicators.
xBRL-CSV 6	No table file for a template indicated as not reported.
2.19, xBRL-CSV 5	No special values in fact or key columns: `#nil`, `#empty` and any other value starting with `#`.
xBRL-CSV 1	No decimals suffix, such as `78920030d-4`.
3.11	Key values without leading or trailing spaces.
xBRL-CSV 4	A value in the key columns of each reported fact: without metadata every key column but `unit`, which only some data points use; with metadata, the columns each data point refers to.
xBRL-CSV 2	With resolved metadata, every column the table's metadata defines in its header.
Package	No files besides `META-INF/reportPackage.json` and, in `reports`, `report.json`, `parameters.csv`, `FilingIndicators.csv` and the tables (`unexpectedFile`, a warning: the filing rules describe the layout without a MUST).

read_xbrl_csv_frames checks special fact values only; the decimals suffix, spaces around key values, special key values and empty key values are left to the caller.

With taxonomy packages¶

Given taxonomy_packages, both readers also check the rules that need the taxonomy, and only then:

Rule	Checks
1.5	The entry point is one the packages' `taxonomyPackage.xml` lists (an XSD, or the JSON beside it), and the reference date is within the module's `FromReferenceDate` and `toReferenceDate`. The EBA's placeholder 9999-12-31 counts as no date.
1.6, 1.6.3	A filing indicator, true or false, for each template of the module, and only the module's codes.
2.18	Decimals of at least -4 for monetary facts (-6 in the FP, ESG, Pillar 3 and REM_DBM modules), at least 4 for percentages and exactly 0 for integers, or INF, by the metric's schema type (`accuracyRequirements`). The -3 that applied to submissions before April 2025 is not checked: the file does not show when it is sent.
3.2 (b)	Percentages, by the metric's `percentItemType`, as decimal fractions: a value above 50 is a warning, since only the value suggests a percentage multiplied by 100.
3.5	XML only: each namespace prefix as the namespace's schema in the packages binds it, or as the XBRL specifications use it (`xbrli`, `link`, `xlink`, `xbrldi`, `xbrldt`, `xsi`, `iso4217`, `find`).
xBRL-CSV 2	Header columns, as the packages resolve the metadata.
Package	Only the module's table files in `reports`.

What each mode checks¶

Mode	Rules
`read_xbrl_xml`	1.4, 1.5, 1.6, 1.6.1, 1.13 to 1.15, 2.1 to 2.4, 2.7 to 2.11, 2.13 to 2.19, 2.21, 2.22, 2.25, 3.1, 3.2 (a), 3.4, 3.6, 3.9, 3.10, file naming
`read_xbrl_csv`	1.5, 1.6, 1.6.1, 2.2, 2.3, 2.8, 2.10, 2.19, 3.6, 3.11 (keys), xBRL-CSV 1, 2 (with metadata), 4, 5, 6, package, file naming
`read_xbrl_csv_frames`	As `read_xbrl_csv`, but of the values only special fact values (2.19, xBRL-CSV 5)
Any, with `taxonomy_packages`	Adds 1.5 (published entry point, reference dates), 1.6 (each template), 1.6.3, 2.18, 3.2 (b), and 3.5 for XML; the frames reader, which builds no facts, adds neither 2.18 nor 3.2

checked_rules gives the same, exactly, as (rule, code) pairs for one mode, so that callers who check rules themselves can skip the ones the reader already reports:

from dpmxbrl.filing import checked_rules

xml = checked_rules("xbrl-xml", taxonomy=True) - checked_rules("xbrl-xml")
frames = checked_rules("xbrl-csv", taxonomy=True, columnar=True)
print(sorted(xml))
print(("2.18", "accuracyRequirements") in frames)

[('1.6.3', 'invalidFilingIndicatorValue'), ('2.18', 'accuracyRequirements'), ('3.2', 'useDecimalFractions'), ('3.5', 'notRecommendedNamespacePrefix')]
False

Left to callers¶

The other rules that need the taxonomy are left to the caller, which can build FilingFindings and merge them with FilingLog; see add your own filing findings:

Rule	Needs
1.7.1	Facts outside the reported templates.
3.11	Fact values without leading or trailing spaces, for string data points only (`SPACES`).
(none)	`baseCurrency` when monetary facts use it.

Not checked¶

Units against the XBRL unit registry (2.23), generating-software information (2.26) and comments (2.5).