Split knowledge watch state derivation

packages/elf-service/src/knowledge/watch.rs

@@ -2,6 +2,7 @@ mod candidates;
 mod inputs;
 mod outcomes;
 mod outputs;
+mod states;
 mod summary;
 
 pub(super) use self::{
@@ -14,10 +15,8 @@ pub(super) use self::{
 		rebuild_request_from_page,
 	},
 	outcomes::{blocked_watch_rebuild, successful_watch_rebuild},
-	outputs::{
-		blocked_outputs, blocked_section_states, candidate_reasons_by_section, rebuild_outputs,
-		successful_rebuild_state, successful_section_states,
-	},
+	outputs::{blocked_outputs, candidate_reasons_by_section, rebuild_outputs},
+	states::{blocked_section_states, successful_rebuild_state, successful_section_states},
 	summary::{page_operator_summary, watch_operator_summary, watch_rebuild_summary},
 };
 

packages/elf-service/src/knowledge/watch/outputs.rs

@@ -1,7 +1,7 @@
 use crate::knowledge::watch::{
 	BTreeMap, BTreeSet, KnowledgePageChangedSource, KnowledgePageRebuildOutput,
-	KnowledgePageSection, KnowledgePageSectionRebuildState, KnowledgePageSectionResponse,
-	KnowledgePageSourceRef, KnowledgeSourceKind, LintDraft, Uuid, Value, serde_json,
+	KnowledgePageSection, KnowledgePageSourceRef, KnowledgeSourceKind, LintDraft, Uuid, Value,
+	serde_json,
 };
 
 pub(in crate::knowledge) fn rebuild_outputs(
@@ -159,107 +159,6 @@ pub(in crate::knowledge) fn conflict_outputs(
 		.collect()
 }
 
-pub(in crate::knowledge) fn successful_section_states(
-	before_sections: &[KnowledgePageSection],
-	rebuilt_sections: &[KnowledgePageSectionResponse],
-	outputs: &[KnowledgePageRebuildOutput],
-) -> Vec<KnowledgePageSectionRebuildState> {
-	let output_map = outputs_by_section(outputs);
-	let before_by_key = before_sections
-		.iter()
-		.map(|section| (section.section_key.as_str(), section))
-		.collect::<BTreeMap<_, _>>();
-
-	rebuilt_sections
-		.iter()
-		.map(|section| {
-			let output_types =
-				output_map.get(section.section_key.as_str()).cloned().unwrap_or_default();
-			let lint_finding_types = lint_finding_types_for_outputs(&output_types);
-			let state = section_state(
-				before_by_key.get(section.section_key.as_str()).copied(),
-				section,
-				&output_types,
-			);
-
-			KnowledgePageSectionRebuildState {
-				section_key: section.section_key.clone(),
-				heading: section.heading.clone(),
-				state,
-				output_types,
-				lint_finding_types,
-			}
-		})
-		.collect()
-}
-
-pub(in crate::knowledge) fn blocked_section_states(
-	sections: &[KnowledgePageSection],
-	outputs: &[KnowledgePageRebuildOutput],
-) -> Vec<KnowledgePageSectionRebuildState> {
-	let output_map = outputs_by_section(outputs);
-
-	sections
-		.iter()
-		.map(|section| {
-			let output_types =
-				output_map.get(section.section_key.as_str()).cloned().unwrap_or_default();
-			let lint_finding_types = lint_finding_types_for_outputs(&output_types);
-			let state = if output_types.iter().any(|kind| kind == "missing_citation") {
-				"blocked"
-			} else if output_types.iter().any(|kind| kind == "stale_section") {
-				"stale"
-			} else {
-				"blocked"
-			};
-
-			KnowledgePageSectionRebuildState {
-				section_key: section.section_key.clone(),
-				heading: section.heading.clone(),
-				state: state.to_string(),
-				output_types,
-				lint_finding_types,
-			}
-		})
-		.collect()
-}
-
-pub(in crate::knowledge) fn section_state(
-	before: Option<&KnowledgePageSection>,
-	after: &KnowledgePageSectionResponse,
-	output_types: &[String],
-) -> String {
-	if output_types.iter().any(|kind| kind == "missing_citation") {
-		return "blocked".to_string();
-	}
-	if before.is_some_and(|section| section.content_hash != after.content_hash)
-		|| output_types.iter().any(|kind| kind == "changed_claim" || kind == "conflict")
-	{
-		return "changed".to_string();
-	}
-
-	if output_types.iter().any(|kind| kind == "stale_section") {
-		return "stale".to_string();
-	}
-
-	"unchanged".to_string()
-}
-
-pub(in crate::knowledge) fn successful_rebuild_state(
-	diff: Option<&Value>,
-	outputs: &[KnowledgePageRebuildOutput],
-) -> String {
-	if diff_content_changed(diff) {
-		return "changed".to_string();
-	}
-
-	if outputs.iter().any(|output| output.output_type == "stale_section") {
-		return "stale".to_string();
-	}
-
-	"unchanged".to_string()
-}
-
 pub(in crate::knowledge) fn section_lookup(
 	sections: &[KnowledgePageSection],
 ) -> BTreeMap<Uuid, (String, String)> {
@@ -276,13 +175,6 @@ pub(in crate::knowledge) fn diff_section_keys(diff: Option<&Value>, key: &str) -
 		.unwrap_or_default()
 }
 
-pub(in crate::knowledge) fn diff_content_changed(diff: Option<&Value>) -> bool {
-	diff.and_then(|value| value.get("content_changed")).and_then(Value::as_bool).unwrap_or(false)
-		|| !diff_section_keys(diff, "added_section_keys").is_empty()
-		|| !diff_section_keys(diff, "removed_section_keys").is_empty()
-		|| !diff_section_keys(diff, "changed_section_keys").is_empty()
-}
-
 pub(in crate::knowledge) fn changed_source_set(
 	changed_sources: &[KnowledgePageChangedSource],
 ) -> BTreeSet<(String, Uuid)> {
@@ -303,42 +195,6 @@ pub(in crate::knowledge) fn output_section_keys(
 		.collect()
 }
 
-pub(in crate::knowledge) fn outputs_by_section(
-	outputs: &[KnowledgePageRebuildOutput],
-) -> BTreeMap<&str, Vec<String>> {
-	let mut map = BTreeMap::<&str, Vec<String>>::new();
-
-	for output in outputs {
-		let Some(section_key) = output.section_key.as_deref() else {
-			continue;
-		};
-
-		map.entry(section_key).or_default().push(output.output_type.clone());
-	}
-	for values in map.values_mut() {
-		values.sort();
-		values.dedup();
-	}
-
-	map
-}
-
-pub(in crate::knowledge) fn lint_finding_types_for_outputs(output_types: &[String]) -> Vec<String> {
-	let mut out = output_types
-		.iter()
-		.filter_map(|output_type| match output_type.as_str() {
-			"stale_section" => Some("stale_source_ref".to_string()),
-			"missing_citation" => Some("missing_citation".to_string()),
-			_ => None,
-		})
-		.collect::<Vec<_>>();
-
-	out.sort();
-	out.dedup();
-
-	out
-}
-
 pub(in crate::knowledge) fn candidate_reasons_by_section(
 	outputs: &[KnowledgePageRebuildOutput],
 ) -> BTreeMap<&str, String> {

packages/elf-service/src/knowledge/watch/states.rs

@@ -0,0 +1,146 @@
+use crate::knowledge::watch::{
+	BTreeMap, KnowledgePageRebuildOutput, KnowledgePageSection, KnowledgePageSectionRebuildState,
+	KnowledgePageSectionResponse, Value, outputs,
+};
+
+pub(in crate::knowledge) fn successful_section_states(
+	before_sections: &[KnowledgePageSection],
+	rebuilt_sections: &[KnowledgePageSectionResponse],
+	outputs: &[KnowledgePageRebuildOutput],
+) -> Vec<KnowledgePageSectionRebuildState> {
+	let output_map = outputs_by_section(outputs);
+	let before_by_key = before_sections
+		.iter()
+		.map(|section| (section.section_key.as_str(), section))
+		.collect::<BTreeMap<_, _>>();
+
+	rebuilt_sections
+		.iter()
+		.map(|section| {
+			let output_types =
+				output_map.get(section.section_key.as_str()).cloned().unwrap_or_default();
+			let lint_finding_types = lint_finding_types_for_outputs(&output_types);
+			let state = section_state(
+				before_by_key.get(section.section_key.as_str()).copied(),
+				section,
+				&output_types,
+			);
+
+			KnowledgePageSectionRebuildState {
+				section_key: section.section_key.clone(),
+				heading: section.heading.clone(),
+				state,
+				output_types,
+				lint_finding_types,
+			}
+		})
+		.collect()
+}
+
+pub(in crate::knowledge) fn blocked_section_states(
+	sections: &[KnowledgePageSection],
+	outputs: &[KnowledgePageRebuildOutput],
+) -> Vec<KnowledgePageSectionRebuildState> {
+	let output_map = outputs_by_section(outputs);
+
+	sections
+		.iter()
+		.map(|section| {
+			let output_types =
+				output_map.get(section.section_key.as_str()).cloned().unwrap_or_default();
+			let lint_finding_types = lint_finding_types_for_outputs(&output_types);
+			let state = if output_types.iter().any(|kind| kind == "missing_citation") {
+				"blocked"
+			} else if output_types.iter().any(|kind| kind == "stale_section") {
+				"stale"
+			} else {
+				"blocked"
+			};
+
+			KnowledgePageSectionRebuildState {
+				section_key: section.section_key.clone(),
+				heading: section.heading.clone(),
+				state: state.to_string(),
+				output_types,
+				lint_finding_types,
+			}
+		})
+		.collect()
+}
+
+pub(in crate::knowledge) fn successful_rebuild_state(
+	diff: Option<&Value>,
+	outputs: &[KnowledgePageRebuildOutput],
+) -> String {
+	if diff_content_changed(diff) {
+		return "changed".to_string();
+	}
+
+	if outputs.iter().any(|output| output.output_type == "stale_section") {
+		return "stale".to_string();
+	}
+
+	"unchanged".to_string()
+}
+
+fn section_state(
+	before: Option<&KnowledgePageSection>,
+	after: &KnowledgePageSectionResponse,
+	output_types: &[String],
+) -> String {
+	if output_types.iter().any(|kind| kind == "missing_citation") {
+		return "blocked".to_string();
+	}
+	if before.is_some_and(|section| section.content_hash != after.content_hash)
+		|| output_types.iter().any(|kind| kind == "changed_claim" || kind == "conflict")
+	{
+		return "changed".to_string();
+	}
+
+	if output_types.iter().any(|kind| kind == "stale_section") {
+		return "stale".to_string();
+	}
+
+	"unchanged".to_string()
+}
+
+fn diff_content_changed(diff: Option<&Value>) -> bool {
+	diff.and_then(|value| value.get("content_changed")).and_then(Value::as_bool).unwrap_or(false)
+		|| !outputs::diff_section_keys(diff, "added_section_keys").is_empty()
+		|| !outputs::diff_section_keys(diff, "removed_section_keys").is_empty()
+		|| !outputs::diff_section_keys(diff, "changed_section_keys").is_empty()
+}
+
+fn outputs_by_section(outputs: &[KnowledgePageRebuildOutput]) -> BTreeMap<&str, Vec<String>> {
+	let mut map = BTreeMap::<&str, Vec<String>>::new();
+
+	for output in outputs {
+		let Some(section_key) = output.section_key.as_deref() else {
+			continue;
+		};
+
+		map.entry(section_key).or_default().push(output.output_type.clone());
+	}
+	for values in map.values_mut() {
+		values.sort();
+		values.dedup();
+	}
+
+	map
+}
+
+fn lint_finding_types_for_outputs(output_types: &[String]) -> Vec<String> {
+	let mut out = output_types
+		.iter()
+		.filter_map(|output_type| match output_type.as_str() {
+			"stale_section" => Some("stale_source_ref".to_string()),
+			"missing_citation" => Some("missing_citation".to_string()),
+			_ => None,
+		})
+		.collect::<Vec<_>>();
+
+	out.sort();
+	out.dedup();
+
+	out
+}