feat: infer child and attribute steps from schema content models

[CaiJimmy]

Child and attribute steps from a schema-typed element or document now infer the nodes that the schema declares, instead of element()* or attribute()*.

Based on #1889, which gives validate results and typed node tests their schema types.

Examples

From static-typing/schema-path-steps/, where $order := validate { <p:order>...</p:order> } (types from PathSteps.xsd):

Expression	Inferred type	Why
$order/p:item	element(p:item, p:Item)*	members of the substitution group, such as p:special, can take its place
$order/p:item/p:price	element(p:price, xs:decimal)*	so sum(...) is xs:decimal
$order/p:item/@id	attribute(id, xs:integer)*	a required attribute
$item/@note	attribute(note, xs:string)?	an optional attribute
$order/p:paid	element(p:paid, xs:boolean)?	a choice between p:paid and p:due
$order/p:customer/p:extra	element(p:extra, xs:integer)?	only the derived p:Extended declares it, which an instance may select with xsi:type
$order/p:item/*	(element(p:price, xs:decimal) | element(p:tag, xs:string))*	different declarations stay a union
$order/*/p:name	element(p:name, xs:string)*	steps from a union still use the schema
$order/p:meta/*	element()*	an xs:any wildcard may match any element from another namespace
$library/l:library/l:book	element(l:book, <anonymous>)*	a document matching document-node(E) has one child, which matches E

A step that the schema rules out, such as $order/p:missing or $order/p:meta/p:x (the wildcard is ##other), is statically empty and raises XPST0005, like other empty steps.

How it works

XmlSchemaCatalog.getStepType resolves the context's schema type, including anonymous ones, and walks the content model of the type and of every type derived from it:

• Element declarations count once each, members of their substitution group included and abstract declarations excluded.
• A sequence adds the counts of its particles, a choice takes the range over its branches, and minOccurs/maxOccurs multiply. The resulting bounds become the step's cardinality.
• An attribute is exactly one if required, otherwise optional.
• A nillable context element adds the empty sequence for children, since a nilled element has none.
• Element declarations become typed node types with their typed value, so atomizing the result uses the schema too (sum($order/p:item/p:price)).

If the context is a union, every member must be schema-typed, and the selections of the members are combined.

Where it falls back

The step keeps the previous inference when the schema does not describe every node it can select:

• a wildcard in the content model that allows the name,
• an xs:anyType context,
• @*, since instances also carry undeclared xsi: attributes,
• partial wildcards (p:*, *:name) and typed tests (element(n, T), schema-element(E)),
• other axes, which schema-descendant-steps covers for descendants.

Anonymous schema types now display as <anonymous>, for example element(p:meta, <anonymous>), instead of their internal names.

[github-actions]

QT3 Test Results - Parser: jsoniq

Regression baseline: run 37854736992

🎉 No regressions found for this build.

✅ Improvement summary: 12 test(s) previously failing, errored, or skipped now pass.

Suite	Test file	Test case
fn		fn/avg.xml:K-SeqAVGFunc-31
fn		fn/avg.xml:K-SeqAVGFunc-32
fn		fn/avg.xml:K-SeqAVGFunc-34
fn		fn/avg.xml:fn-avg-mix-args-029
fn		fn/avg.xml:fn-avg-mix-args-030
fn		fn/avg.xml:fn-avg-mix-args-031
fn		fn/sum.xml:K-SeqSUMFunc-23
fn		fn/sum.xml:K-SeqSUMFunc-24
fn		fn/sum.xml:fn-sum-10
fn		fn/sum.xml:fn-sum-9
prod		prod/SwitchExpr.xml:switch-901
prod		prod/SwitchExpr.xml:switch-902

Summary of passed tests for jsoniq

Test Suite	Passing	Failing	Errors	Skipped	Missing features	Total
app	2056	27	69	5	5	2157
array	208	0	1	0	0	209
fn	10321	258	169	387	315	11135
map	220	0	0	0	0	220
math	149	0	0	0	0	149
misc	765	34	20	52	18	871
op	4351	3	12	0	0	4366
prod	11507	326	166	208	7	12207
ser	332	11	0	0	0	343
xs	155	0	0	9	0	164
Total	30064	659	437	661	345	31821

Full analysis report: see analysis-jsoniq in artifacts.

Download detailed test results

[github-actions]

QT3 Test Results - Parser: xquery

Regression baseline: run 37854736992

🎉 No regressions found for this build.

✅ Improvement summary: 12 test(s) previously failing, errored, or skipped now pass.

Suite	Test file	Test case
fn		fn/avg.xml:K-SeqAVGFunc-31
fn		fn/avg.xml:K-SeqAVGFunc-32
fn		fn/avg.xml:K-SeqAVGFunc-34
fn		fn/avg.xml:fn-avg-mix-args-029
fn		fn/avg.xml:fn-avg-mix-args-030
fn		fn/avg.xml:fn-avg-mix-args-031
fn		fn/sum.xml:K-SeqSUMFunc-23
fn		fn/sum.xml:K-SeqSUMFunc-24
fn		fn/sum.xml:fn-sum-10
fn		fn/sum.xml:fn-sum-9
prod		prod/SwitchExpr.xml:switch-901
prod		prod/SwitchExpr.xml:switch-902

Summary of passed tests for xquery

Test Suite	Passing	Failing	Errors	Skipped	Missing features	Total
app	2063	26	63	5	5	2157
array	209	0	0	0	0	209
fn	10343	237	168	387	315	11135
map	220	0	0	0	0	220
math	149	0	0	0	0	149
misc	769	30	20	52	18	871
op	4351	3	12	0	0	4366
prod	11590	277	132	208	7	12207
ser	343	0	0	0	0	343
xs	155	0	0	9	0	164
Total	30192	573	395	661	345	31821

Full analysis report: see analysis-xquery in artifacts.

Download detailed test results