patram 0.2.0 → 0.3.0

package/lib/build-graph-identity.js

@@ -27,14 +27,14 @@ export function collectDocumentEntityKeys(mappings, claims) {
     const source_path = normalizeRepoRelativePath(claim.origin.path);
     const entity_map_key = getDocumentEntityMapKey(
       source_path,
-      mapping_definition.node.kind,
+      mapping_definition.node.class,
     );
     const entity_key = getStringClaimValue(claim);
     const existing_entity_key = document_entity_keys.get(entity_map_key);
 
     if (existing_entity_key && existing_entity_key !== entity_key) {
       throw new Error(
-        `Document "${source_path}" defines multiple ${mapping_definition.node.kind} ids.`,
+        `Document "${source_path}" defines multiple ${mapping_definition.node.class} ids.`,
       );
     }
 
@@ -47,7 +47,7 @@ export function collectDocumentEntityKeys(mappings, claims) {
 /**
  * Resolve the node key for one mapped claim.
  *
- * @param {{ field: string, key?: 'path' | 'value', kind: string }} node_mapping
+ * @param {{ field: string, key?: 'path' | 'value', class: string }} node_mapping
  * @param {PatramClaim} claim
  * @param {Map<string, string>} document_entity_keys
  * @returns {string}
@@ -55,7 +55,7 @@ export function collectDocumentEntityKeys(mappings, claims) {
 export function resolveNodeKey(node_mapping, claim, document_entity_keys) {
   const source_key = normalizeRepoRelativePath(claim.origin.path);
 
-  if (node_mapping.kind === 'document') {
+  if (node_mapping.class === 'document') {
     return source_key;
   }
 
@@ -65,7 +65,7 @@ export function resolveNodeKey(node_mapping, claim, document_entity_keys) {
 
   return (
     document_entity_keys.get(
-      getDocumentEntityMapKey(source_key, node_mapping.kind),
+      getDocumentEntityMapKey(source_key, node_mapping.class),
     ) ?? source_key
   );
 }
@@ -73,7 +73,7 @@ export function resolveNodeKey(node_mapping, claim, document_entity_keys) {
 /**
  * Resolve one edge target key and canonical path.
  *
- * @param {string} target_kind
+ * @param {string} target_class
  * @param {'path' | 'value'} target_type
  * @param {PatramClaim} claim
  * @param {Map<string, string>} document_entity_keys
@@ -81,18 +81,18 @@ export function resolveNodeKey(node_mapping, claim, document_entity_keys) {
  * @returns {{ key: string, path?: string }}
  */
 export function resolveTargetReference(
-  target_kind,
+  target_class,
   target_type,
   claim,
   document_entity_keys,
   document_paths,
 ) {
   if (target_type === 'value') {
-    return resolveValueTargetReference(target_kind, claim);
+    return resolveValueTargetReference(target_class, claim);
   }
 
   return resolvePathTargetReference(
-    target_kind,
+    target_class,
     claim,
     document_entity_keys,
     document_paths,
@@ -106,16 +106,17 @@ export function resolveTargetReference(
  * @param {string | undefined} source_path
  */
 export function setNonDocumentPath(graph_node, source_path) {
-  if (!source_path || graph_node.kind === 'document') {
+  if (!source_path || graph_node.$class === 'document') {
     return;
   }
 
-  if (!graph_node.path) {
+  if (!graph_node.$path) {
+    graph_node.$path = source_path;
     graph_node.path = source_path;
     return;
   }
 
-  if (graph_node.path !== source_path) {
+  if (graph_node.$path !== source_path) {
     throw new Error(
       `Node "${graph_node.id}" maps to multiple canonical paths.`,
     );
@@ -180,14 +181,14 @@ function resolveValueTargetReference(target_kind, claim) {
 }
 
 /**
- * @param {string} target_kind
+ * @param {string} target_class
  * @param {PatramClaim} claim
  * @param {Map<string, string>} document_entity_keys
  * @param {Set<string>} document_paths
  * @returns {{ key: string, path?: string }}
  */
 function resolvePathTargetReference(
-  target_kind,
+  target_class,
   claim,
   document_entity_keys,
   document_paths,
@@ -199,7 +200,7 @@ function resolvePathTargetReference(
     document_paths,
   );
 
-  if (target_kind === 'document') {
+  if (target_class === 'document') {
     return {
       key: target_path,
       path: target_path,
@@ -207,7 +208,7 @@ function resolvePathTargetReference(
   }
 
   const semantic_target_key = document_entity_keys.get(
-    getDocumentEntityMapKey(target_path, target_kind),
+    getDocumentEntityMapKey(target_path, target_class),
   );
 
   return {
@@ -262,9 +263,9 @@ function getStringClaimValue(claim) {
 
 /**
  * @param {string} document_path
- * @param {string} kind_name
+ * @param {string} class_name
  * @returns {string}
  */
-function getDocumentEntityMapKey(document_path, kind_name) {
-  return `${kind_name}:${document_path}`;
+function getDocumentEntityMapKey(document_path, class_name) {
+  return `${class_name}:${document_path}`;
 }

package/lib/build-graph.js

@@ -1,9 +1,13 @@
+/* eslint-disable max-lines */
 /**
  * @import { BuildGraphResult, GraphEdge, GraphNode } from './build-graph.types.ts';
  * @import { PatramClaim } from './parse-claims.types.ts';
+ * @import { MetadataFieldConfig } from './load-patram-config.types.ts';
  * @import { MappingDefinition, PatramConfig } from './patram-config.types.ts';
  */
 
+import { posix } from 'node:path';
+
 import {
   collectDocumentEntityKeys,
   normalizeRepoRelativePath,
@@ -25,7 +29,7 @@ import {
  * Uses Term: ../docs/reference/terms/graph.md
  * Uses Term: ../docs/reference/terms/mapping.md
  * Uses Term: ../docs/reference/terms/relation.md
- * Tracked in: ../docs/plans/v0/source-anchor-dogfooding.md
+ * Tracked in: ../docs/plans/v0/source-anchor-dogfood.md
  * Decided by: ../docs/decisions/graph-materialization.md
  * Implements: ../docs/tasks/v0/materialize-graph.md
  * @patram
@@ -33,6 +37,11 @@ import {
  * @see {@link ../docs/decisions/graph-materialization.md}
  */
 
+const STRUCTURAL_FIELD_NAMES = new Set(['$class', '$id', '$path']);
+const DETERMINISTIC_LOCALE = 'en';
+const DERIVED_TITLE_PRIORITY = 1;
+const EXPLICIT_TITLE_PRIORITY = 2;
+
 /**
  * Build a Patram graph from semantic config and parsed claims.
  *
@@ -51,13 +60,16 @@ export function buildGraph(patram_config, claims) {
   const document_paths = new Set(
     claims.map((claim) => normalizeRepoRelativePath(claim.origin.path)),
   );
+  /** @type {Map<string, number>} */
+  const title_priorities = new Map();
 
   createDocumentNodes(graph_nodes, claims);
   applyNodeMappings(
     graph_nodes,
-    patram_config.mappings,
+    patram_config,
     claims,
     document_entity_keys,
+    title_priorities,
   );
   const graph_edges = createGraphEdges(
     graph_nodes,
@@ -66,6 +78,7 @@ export function buildGraph(patram_config, claims) {
     document_entity_keys,
     document_paths,
   );
+  applyFallbackTitles(graph_nodes, title_priorities);
 
   return {
     edges: graph_edges,
@@ -117,18 +130,23 @@ function createDocumentNodes(graph_nodes, claims) {
 
 /**
  * @param {Map<string, GraphNode>} graph_nodes
- * @param {Record<string, MappingDefinition>} mappings
+ * @param {PatramConfig} patram_config
  * @param {PatramClaim[]} claims
  * @param {Map<string, string>} document_entity_keys
+ * @param {Map<string, number>} title_priorities
  */
 function applyNodeMappings(
   graph_nodes,
-  mappings,
+  patram_config,
   claims,
   document_entity_keys,
+  title_priorities,
 ) {
   for (const claim of claims) {
-    const mapping_definition = resolveMappingDefinition(mappings, claim);
+    const mapping_definition = resolveMappingDefinition(
+      patram_config.mappings,
+      claim,
+    );
 
     if (!mapping_definition?.node) {
       continue;
@@ -136,9 +154,11 @@ function applyNodeMappings(
 
     applyNodeMapping(
       graph_nodes,
+      patram_config,
       mapping_definition.node,
       claim,
       document_entity_keys,
+      title_priorities,
     );
   }
 }
@@ -175,7 +195,7 @@ function createGraphEdges(
       normalizeRepoRelativePath(claim.origin.path),
     );
     const target_reference = resolveTargetReference(
-      mapping_definition.emit.target_kind,
+      mapping_definition.emit.target_class,
       mapping_definition.emit.target,
       claim,
       document_entity_keys,
@@ -183,7 +203,7 @@ function createGraphEdges(
     );
     const target_node = upsertNode(
       graph_nodes,
-      mapping_definition.emit.target_kind,
+      mapping_definition.emit.target_class,
       target_reference.key,
     );
 
@@ -204,23 +224,117 @@ function createGraphEdges(
 
 /**
  * @param {Map<string, GraphNode>} graph_nodes
- * @param {{ field: string, key?: 'path' | 'value', kind: string }} node_mapping
+ * @param {PatramConfig} patram_config
+ * @param {{ field: string, key?: 'path' | 'value', class: string }} node_mapping
  * @param {PatramClaim} claim
  * @param {Map<string, string>} document_entity_keys
+ * @param {Map<string, number>} title_priorities
  */
 function applyNodeMapping(
   graph_nodes,
+  patram_config,
   node_mapping,
   claim,
   document_entity_keys,
+  title_priorities,
 ) {
   const source_key = normalizeRepoRelativePath(claim.origin.path);
   const node_key = resolveNodeKey(node_mapping, claim, document_entity_keys);
-  const graph_node = upsertNode(graph_nodes, node_mapping.kind, node_key);
+  const graph_node = upsertNode(graph_nodes, node_mapping.class, node_key);
+  const field_value = getNodeFieldValue(claim);
 
   setNonDocumentPath(graph_node, source_key);
+  validateNodeFieldMapping(
+    patram_config,
+    node_mapping.class,
+    node_mapping.field,
+  );
+
+  if (node_mapping.field === 'title') {
+    setNodeTitle(
+      graph_node,
+      title_priorities,
+      field_value,
+      claim.type === 'document.title'
+        ? DERIVED_TITLE_PRIORITY
+        : EXPLICIT_TITLE_PRIORITY,
+    );
+    return;
+  }
+
+  setNodeFieldValue(
+    graph_node,
+    node_mapping.field,
+    field_value,
+    getFieldDefinition(patram_config, node_mapping.field),
+  );
+}
+
+/**
+ * Validate one mapped node field against the configured field model.
+ *
+ * @param {PatramConfig} patram_config
+ * @param {string} node_class
+ * @param {string} field_name
+ */
+function validateNodeFieldMapping(patram_config, node_class, field_name) {
+  const validation_error = getNodeFieldValidationError(
+    patram_config,
+    node_class,
+    field_name,
+  );
+
+  if (validation_error) {
+    throw new Error(validation_error);
+  }
+}
+
+/**
+ * @param {PatramConfig} patram_config
+ * @param {string} node_class
+ * @param {string} field_name
+ * @returns {string | null}
+ */
+function getNodeFieldValidationError(patram_config, node_class, field_name) {
+  if (isStructuralFieldName(field_name) || field_name === 'title') {
+    return null;
+  }
+
+  const field_definition = getFieldDefinition(patram_config, field_name);
+
+  if (!field_definition) {
+    return `Node class "${node_class}" maps to unknown field "${field_name}".`;
+  }
+
+  const class_schema = getClassSchema(patram_config, node_class);
+  const class_field_rule = class_schema?.fields?.[field_name];
+
+  if (isForbiddenClassField(class_field_rule)) {
+    return `Field "${field_name}" is forbidden for class "${node_class}".`;
+  }
+
+  if (isUndeclaredClassField(class_schema, class_field_rule)) {
+    return `Field "${field_name}" is not declared for class "${node_class}".`;
+  }
+
+  return null;
+}
+
+/**
+ * @param {{ presence: 'required' | 'optional' | 'forbidden' } | undefined} class_field_rule
+ * @returns {boolean}
+ */
+function isForbiddenClassField(class_field_rule) {
+  return class_field_rule?.presence === 'forbidden';
+}
 
-  graph_node[node_mapping.field] = getNodeFieldValue(claim);
+/**
+ * @param {{ fields?: Record<string, { presence: 'required' | 'optional' | 'forbidden' }>, unknown_fields?: 'ignore' | 'error' } | undefined} class_schema
+ * @param {{ presence: 'required' | 'optional' | 'forbidden' } | undefined} class_field_rule
+ * @returns {boolean}
+ */
+function isUndeclaredClassField(class_schema, class_field_rule) {
+  return class_schema?.unknown_fields === 'error' && !class_field_rule;
 }
 
 /**
@@ -246,23 +360,26 @@ function upsertNode(graph_nodes, kind_name, node_key) {
 
 /**
  * @param {string} node_id
- * @param {string} kind_name
+ * @param {string} class_name
  * @param {string} node_key
  * @returns {GraphNode}
  */
-function createNode(node_id, kind_name, node_key) {
-  if (kind_name === 'document') {
+function createNode(node_id, class_name, node_key) {
+  if (class_name === 'document') {
     return {
+      $class: class_name,
+      $id: node_id,
+      $path: node_key,
       id: node_id,
-      kind: kind_name,
       path: node_key,
     };
   }
 
   return {
+    $class: class_name,
+    $id: node_id,
     id: node_id,
     key: node_key,
-    kind: kind_name,
   };
 }
 
@@ -291,11 +408,247 @@ function getNodeFieldValue(claim) {
   throw new Error(`Claim "${claim.id}" does not carry a string value.`);
 }
 
+/**
+ * @param {Map<string, GraphNode>} graph_nodes
+ * @param {Map<string, number>} title_priorities
+ */
+function applyFallbackTitles(graph_nodes, title_priorities) {
+  for (const graph_node of graph_nodes.values()) {
+    if (graph_node.title !== undefined) {
+      continue;
+    }
+
+    const fallback_title = getFallbackTitle(graph_node);
+
+    graph_node.title = fallback_title;
+    title_priorities.set(graph_node.id, 0);
+  }
+}
+
+/**
+ * @param {GraphNode} graph_node
+ * @returns {string}
+ */
+function getFallbackTitle(graph_node) {
+  if (graph_node.$path) {
+    return posix.basename(graph_node.$path);
+  }
+
+  return getNodeIdKey(graph_node.$id ?? graph_node.id);
+}
+
+/**
+ * @param {GraphNode} graph_node
+ * @param {Map<string, number>} title_priorities
+ * @param {string} title_value
+ * @param {number} source_priority
+ */
+function setNodeTitle(
+  graph_node,
+  title_priorities,
+  title_value,
+  source_priority,
+) {
+  const current_priority = title_priorities.get(graph_node.id);
+
+  if (current_priority === undefined) {
+    graph_node.title = title_value;
+    title_priorities.set(graph_node.id, source_priority);
+    return;
+  }
+
+  if (source_priority > current_priority) {
+    graph_node.title = title_value;
+    title_priorities.set(graph_node.id, source_priority);
+    return;
+  }
+
+  if (
+    source_priority === current_priority &&
+    graph_node.title !== title_value
+  ) {
+    throw new Error(
+      `Node "${graph_node.id}" has conflicting title values "${graph_node.title}" and "${title_value}".`,
+    );
+  }
+}
+
+/**
+ * @param {GraphNode} graph_node
+ * @param {string} field_name
+ * @param {string} field_value
+ * @param {MetadataFieldConfig | undefined} field_definition
+ */
+function setNodeFieldValue(
+  graph_node,
+  field_name,
+  field_value,
+  field_definition,
+) {
+  if (
+    field_name === '$id' ||
+    field_name === '$class' ||
+    field_name === '$path'
+  ) {
+    setStructuralFieldValue(graph_node, field_name, field_value);
+    return;
+  }
+
+  if (!field_definition || field_definition.multiple !== true) {
+    setSingleValueField(graph_node, field_name, field_value);
+    return;
+  }
+
+  setMultiValueField(graph_node, field_name, field_value);
+}
+
+/**
+ * @param {GraphNode} graph_node
+ * @param {'$id' | '$class' | '$path'} field_name
+ * @param {string} field_value
+ */
+function setStructuralFieldValue(graph_node, field_name, field_value) {
+  const current_value = graph_node[field_name];
+
+  if (current_value === undefined) {
+    assignStructuralFieldValue(graph_node, field_name, field_value);
+    return;
+  }
+
+  if (
+    field_name === '$class' &&
+    graph_node.id.startsWith('doc:') &&
+    current_value === 'document'
+  ) {
+    assignStructuralFieldValue(graph_node, field_name, field_value);
+    return;
+  }
+
+  if (current_value !== field_value) {
+    throw new Error(
+      `Node "${graph_node.id}" has conflicting structural values for "${field_name}".`,
+    );
+  }
+}
+
+/**
+ * Keep legacy mirrors in sync while the rest of the codebase still reads them.
+ *
+ * @param {GraphNode} graph_node
+ * @param {'$id' | '$class' | '$path'} field_name
+ * @param {string} field_value
+ */
+function assignStructuralFieldValue(graph_node, field_name, field_value) {
+  graph_node[field_name] = field_value;
+
+  if (field_name === '$path') {
+    graph_node.path = field_value;
+  }
+}
+
+/**
+ * @param {GraphNode} graph_node
+ * @param {string} field_name
+ * @param {string} field_value
+ */
+function setSingleValueField(graph_node, field_name, field_value) {
+  const current_value = graph_node[field_name];
+
+  if (current_value === undefined) {
+    graph_node[field_name] = field_value;
+    return;
+  }
+
+  if (current_value !== field_value) {
+    throw new Error(
+      `Node "${graph_node.id}" has conflicting values for field "${field_name}": "${current_value}" and "${field_value}".`,
+    );
+  }
+}
+
+/**
+ * @param {GraphNode} graph_node
+ * @param {string} field_name
+ * @param {string} field_value
+ */
+function setMultiValueField(graph_node, field_name, field_value) {
+  const current_value = graph_node[field_name];
+
+  if (current_value === undefined) {
+    graph_node[field_name] = [field_value];
+    return;
+  }
+
+  if (Array.isArray(current_value)) {
+    if (current_value.includes(field_value)) {
+      return;
+    }
+
+    graph_node[field_name] = [...current_value, field_value].sort(
+      compareFieldValues,
+    );
+    return;
+  }
+
+  graph_node[field_name] = [current_value, field_value].sort(
+    compareFieldValues,
+  );
+}
+
+/**
+ * @param {string} left_value
+ * @param {string} right_value
+ * @returns {number}
+ */
+function compareFieldValues(left_value, right_value) {
+  return left_value.localeCompare(right_value, DETERMINISTIC_LOCALE);
+}
+
+/**
+ * @param {PatramConfig} patram_config
+ * @param {string} field_name
+ * @returns {MetadataFieldConfig | undefined}
+ */
+function getFieldDefinition(patram_config, field_name) {
+  return patram_config.fields?.[field_name];
+}
+
+/**
+ * @param {PatramConfig} patram_config
+ * @param {string} class_name
+ * @returns {{ fields?: Record<string, { presence: 'required' | 'optional' | 'forbidden' }>, unknown_fields?: 'ignore' | 'error' } | undefined}
+ */
+function getClassSchema(patram_config, class_name) {
+  return patram_config.class_schemas?.[class_name];
+}
+
+/**
+ * @param {string} field_name
+ * @returns {field_name is '$class' | '$id' | '$path'}
+ */
+function isStructuralFieldName(field_name) {
+  return STRUCTURAL_FIELD_NAMES.has(field_name);
+}
+
 /**
  * @param {[string, GraphNode]} left_entry
  * @param {[string, GraphNode]} right_entry
  * @returns {number}
  */
 function compareNodeEntries(left_entry, right_entry) {
-  return left_entry[0].localeCompare(right_entry[0], 'en');
+  return left_entry[0].localeCompare(right_entry[0], DETERMINISTIC_LOCALE);
+}
+
+/**
+ * @param {string} node_id
+ * @returns {string}
+ */
+function getNodeIdKey(node_id) {
+  const separator_index = node_id.indexOf(':');
+
+  if (separator_index < 0) {
+    return node_id;
+  }
+
+  return node_id.slice(separator_index + 1);
 }

package/lib/build-graph.types.ts

@@ -1,12 +1,15 @@
 import type { ClaimOrigin } from './parse-claims.types.ts';
 
 export interface GraphNode {
+  $class?: string;
+  $id?: string;
+  $path?: string;
   id: string;
-  kind: string;
+  kind?: string;
   key?: string;
   path?: string;
   title?: string;
-  [field: string]: string | undefined;
+  [field: string]: string | string[] | undefined;
 }
 
 export interface GraphEdge {