@fitlab-ai/agent-infra 0.9.14 → 0.10.0

package/lib/task/sections.ts

@@ -1,4 +1,5 @@
 import type { FrontmatterScalar } from './frontmatter.ts';
+import { scanVisibleMarkdown } from './markdown.ts';
 
 function escapeRegExp(value: string): string {
   return value.replace(/[.*+?^${}()|[\]\\]/g, '\\$&');
@@ -93,12 +94,14 @@ function parseTable(
     throw new DocumentMutationError('TASK_DOCUMENT_INVALID', 'table section is ambiguous');
   }
   const section = sections[0]!;
+  const visibleStarts = new Set(scanVisibleMarkdown(content).lines.map((line) => line.start));
   const sectionLines = linesOf(content).filter(
     (line) => line.start >= section.bodyStart && line.start < section.end
   );
   const firstContent = sectionLines.findIndex((line) => line.text.trim() !== '');
   const tables: number[] = [];
   for (let index = 0; index + 1 < sectionLines.length; index += 1) {
+    if (!visibleStarts.has(sectionLines[index]!.start) || !visibleStarts.has(sectionLines[index + 1]!.start)) continue;
     const header = splitTableCells(sectionLines[index]!.text);
     const separator = splitTableCells(sectionLines[index + 1]!.text);
     if (!header || !separator || header.cells.length !== input.columns.length) continue;
@@ -173,24 +176,24 @@ function documentEol(content: string): '\n' | '\r\n' {
 
 function matchingSections(content: string, aliases: readonly string[]) {
   const allowed = new Set(aliases);
-  return linesOf(content)
-    .map((line, index, lines) => {
-      const match = /^(#{2,3})\s+(.+?)\s*$/.exec(line.text);
-      if (!match?.[2] || !allowed.has(match[2])) return null;
-      const level = match[1]!.length;
-      let end = content.length;
-      for (let i = index + 1; i < lines.length; i += 1) {
-        const nextHeading = /^(#{2,3})\s+/.exec(lines[i]!.text);
-        if (nextHeading && nextHeading[1]!.length <= level) {
-          end = lines[i]!.start;
-          break;
-        }
-      }
-      return { heading: match[2]!, line, bodyStart: line.end, end };
+  const lines = new Map(linesOf(content).map((line) => [line.start, line]));
+  const headings = scanVisibleMarkdown(content).headings.filter((heading) => heading.level === 2 || heading.level === 3);
+  return headings
+    .map((heading, index) => {
+      if (!allowed.has(heading.text)) return null;
+      const next = headings.slice(index + 1).find((candidate) => candidate.level <= heading.level);
+      const line = lines.get(heading.start)!;
+      return { heading: heading.text, level: heading.level, line, bodyStart: line.end, end: next?.start ?? content.length };
     })
     .filter((match): match is NonNullable<typeof match> => match !== null);
 }
 
+// Select the first visible section at the requested level, in document order.
+function findSectionRange(content: string, aliases: readonly string[], level: 2 | 3 = 2) {
+  const section = matchingSections(content, aliases).find((candidate) => candidate.level === level);
+  return section ? { start: section.line.start, bodyStart: section.bodyStart, end: section.end, heading: section.heading } : null;
+}
+
 function validateAliases(aliases: readonly string[], name: string): void {
   if (!Array.isArray(aliases) || aliases.length === 0 || aliases.some((alias) => !alias || /[\r\n]/.test(alias))) {
     throw new DocumentMutationError('MUTATION_INVALID', `${name} must contain headings`);
@@ -351,11 +354,13 @@ function mutateTableRow(content: string, mutation: TableRowMutation): TableMutat
     );
   }
   const section = sections[0]!;
+  const visibleStarts = new Set(scanVisibleMarkdown(content).lines.map((line) => line.start));
   const sectionLines = linesOf(content).filter(
     (line) => line.start >= section.bodyStart && line.start < section.end
   );
   const tables: { headerIndex: number; cells: string[] }[] = [];
   for (let i = 0; i + 1 < sectionLines.length; i += 1) {
+    if (!visibleStarts.has(sectionLines[i]!.start) || !visibleStarts.has(sectionLines[i + 1]!.start)) continue;
     const header = splitTableCells(sectionLines[i]!.text);
     const separator = splitTableCells(sectionLines[i + 1]!.text);
     if (!header || !separator || header.cells.length !== mutation.columns.length) continue;
@@ -477,24 +482,11 @@ function mutateTableRow(content: string, mutation: TableRowMutation): TableMutat
  * lines are trimmed. Returns '' when no alias heading is present.
  */
 function extractSection(content: string, aliases: string[]): string {
-  const lines = content.split('\n');
-  let start = -1;
-  for (let i = 0; i < lines.length; i += 1) {
-    const line = lines[i]!.trim();
-    if (aliases.some((alias) => new RegExp(`^##\\s+${escapeRegExp(alias)}\\s*$`).test(line))) {
-      start = i + 1;
-      break;
-    }
-  }
-  if (start === -1) return '';
-  let end = lines.length;
-  for (let i = start; i < lines.length; i += 1) {
-    if (/^##\s+/.test(lines[i]!)) {
-      end = i;
-      break;
-    }
-  }
-  return lines.slice(start, end).join('\n').replace(/^\n+/, '').replace(/\n+$/, '');
+  const headings = scanVisibleMarkdown(content).headings.filter((heading) => heading.level === 2);
+  const index = headings.findIndex((heading) => aliases.includes(heading.text));
+  if (index === -1) return '';
+  return content.slice(headings[index]!.end + 1, headings[index + 1]?.start ?? content.length)
+    .replace(/^(?:\r?\n)+|(?:\r?\n)+$/g, '');
 }
 
 /**
@@ -503,10 +495,8 @@ function extractSection(content: string, aliases: string[]): string {
  * alias when none is present.
  */
 function findSectionHeading(content: string, aliases: string[]): string {
-  for (const alias of aliases) {
-    if (new RegExp(`^##\\s+${escapeRegExp(alias)}\\s*$`, 'm').test(content)) return alias;
-  }
-  return aliases[0]!;
+  const headings = scanVisibleMarkdown(content).headings.filter((heading) => heading.level === 2);
+  return aliases.find((alias) => headings.some((heading) => heading.text === alias)) ?? aliases[0]!;
 }
 
 /**
@@ -518,27 +508,16 @@ function findSectionHeading(content: string, aliases: string[]): string {
  * and trailing blank lines are trimmed. Returns '' when no match is present.
  */
 function extractSubSection(content: string, headingPrefix: string): string {
-  const lines = content.split('\n');
-  const headRe = new RegExp(`^###\\s+${escapeRegExp(headingPrefix)}(?![\\w-])`);
-  let start = -1;
-  for (let i = 0; i < lines.length; i += 1) {
-    if (headRe.test(lines[i]!.trim())) {
-      start = i;
-      break;
-    }
-  }
-  if (start === -1) return '';
-  let end = lines.length;
-  for (let i = start + 1; i < lines.length; i += 1) {
-    if (/^###?\s+/.test(lines[i]!)) {
-      end = i;
-      break;
-    }
-  }
-  return lines.slice(start, end).join('\n').replace(/^\n+/, '').replace(/\n+$/, '');
+  const headings = scanVisibleMarkdown(content).headings.filter((heading) => heading.level === 2 || heading.level === 3);
+  const headRe = new RegExp(`^${escapeRegExp(headingPrefix)}(?![\\w-])`);
+  const index = headings.findIndex((heading) => heading.level === 3 && headRe.test(heading.text));
+  if (index === -1) return '';
+  return content.slice(headings[index]!.start, headings[index + 1]?.start ?? content.length)
+    .replace(/^(?:\r?\n)+|(?:\r?\n)+$/g, '');
 }
 
 export {
+  findSectionRange,
   extractSection,
   findSectionHeading,
   extractSubSection,

package/lib/task/short-id.ts

@@ -329,14 +329,18 @@ function mutateShortIdRegistryAt(
   activeDir: string,
   width: number,
   taskId: string,
-  requested: ShortIdMutationEffect
+  requested: ShortIdMutationEffect,
+  onlyReleaseWhenInactive = false,
+  pruneOrphans = true
 ): ShortIdMutationResult {
   const registryPath = path.join(activeDir, REGISTRY_NAME);
   return withRegistryLock(activeDir, () => {
     const registry = readRegistryStrict(registryPath, width);
     const original = JSON.stringify(registry.ids);
-    for (const [key, candidate] of Object.entries(registry.ids)) {
-      if (candidate !== taskId && !fs.existsSync(path.join(activeDir, candidate, 'task.md'))) delete registry.ids[key];
+    if (pruneOrphans) {
+      for (const [key, candidate] of Object.entries(registry.ids)) {
+        if (candidate !== taskId && !fs.existsSync(path.join(activeDir, candidate, 'task.md'))) delete registry.ids[key];
+      }
     }
     const existing = Object.entries(registry.ids).find(([, candidate]) => candidate === taskId);
     let result: ShortIdMutationResult;
@@ -356,8 +360,13 @@ function mutateShortIdRegistryAt(
         result = { effect: 'allocated', shortId: key, changed: true };
       }
     } else if (requested === 'release') {
-      if (!existing) result = { effect: 'unchanged', shortId: null, changed: false };
-      else { delete registry.ids[existing[0]]; result = { effect: 'released', shortId: existing[0], changed: true }; }
+      const taskIsActive = fs.existsSync(path.join(activeDir, taskId, 'task.md'));
+      if (!existing || (onlyReleaseWhenInactive && taskIsActive)) {
+        result = { effect: 'unchanged', shortId: existing?.[0] ?? null, changed: false };
+      } else {
+        delete registry.ids[existing[0]];
+        result = { effect: 'released', shortId: existing[0], changed: true };
+      }
     } else result = { effect: 'unchanged', shortId: existing ? existing[0] : null, changed: false };
     const changed = original !== JSON.stringify(registry.ids);
     if (changed) writeRegistry(registryPath, registry);
@@ -365,6 +374,18 @@ function mutateShortIdRegistryAt(
   });
 }
 
+function releaseStaleShortIdRegistry(repoRoot: string, taskId: string): ShortIdMutationResult {
+  const activeDir = path.join(repoRoot, '.agents', 'workspace', 'active');
+  return mutateShortIdRegistryAt(
+    activeDir,
+    configuredShortIdLength(repoRoot),
+    taskId,
+    'release',
+    true,
+    false
+  );
+}
+
 function executeShortIdCommand(request: ShortIdCommandRequest): ShortIdCommandResult {
   try {
     if (!Number.isInteger(request.shortIdLength) || request.shortIdLength < 1) {
@@ -550,6 +571,7 @@ export {
   inspectShortIdRegistry,
   configuredShortIdLength,
   mutateShortIdRegistry,
+  releaseStaleShortIdRegistry,
   executeShortIdCommand
 };
 export type {

package/lib/task/task-execution-lock.ts

@@ -293,7 +293,7 @@ function withTaskExecutionLock<T>(
     return value;
   } catch (error) {
     if (error instanceof TaskExecutionLockError) throw error;
-    if (error instanceof Error && error.name === 'OrchestrationStateError') throw error;
+    if (error instanceof Error && ['OrchestrationStateError', 'ArtifactRecoveryError'].includes(error.name)) throw error;
     throw new TaskExecutionLockError(
       'ORCHESTRATION_LOCK_FAILED',
       'lifecycle task lock operation failed',