mixdog 0.9.93 → 0.9.95

package/src/runtime/agent/orchestrator/tools/builtin/read-tool.mjs

@@ -113,6 +113,7 @@ export async function executeReadTool(args, workDir, readStateScope, executeChil
         _rangeHashesFromRenderedReadText,
         _readEntryLineWindow,
         _recordReadSnapshot,
+        READ_MAX_OUTPUT_BYTES,
     } = helpers;
     // CC `file_path` alias — official SDK schema uses `file_path`;
     // mixdog has historically used `path`. Honor `file_path` so a
@@ -310,6 +311,21 @@ export async function executeReadTool(args, workDir, readStateScope, executeChil
             if (a.offset !== b.offset) return a.offset - b.offset;
             return a.index - b.index;
         });
+        // A path[] Read shares one byte ceiling. Reserve room for headers and
+        // split the remaining body budget across distinct disk windows.
+        const _readCallBudget = Number(options?.readOutputBudgetBytes) > 0
+            ? Math.min(READ_MAX_OUTPUT_BYTES, Math.trunc(Number(options.readOutputBudgetBytes)))
+            : READ_MAX_OUTPUT_BYTES;
+        const _headerReserve = Math.min(
+            Math.max(2_048, entries.reduce((sum, entry) => (
+                sum + Buffer.byteLength(String(normalizeOutputPath(entry?.path || '(missing-path)')), 'utf8') + 128
+            ), 0)),
+            Math.floor(_readCallBudget / 2),
+        );
+        const _perTaskOutputBudget = Math.max(
+            256,
+            Math.floor((_readCallBudget - _headerReserve) / Math.max(1, tasks.length)),
+        );
         const results = new Array(entries.length);
         const readChains = new Map();
         await Promise.all(tasks.map(({ entry, index }) => {
@@ -319,9 +335,15 @@ export async function executeReadTool(args, workDir, readStateScope, executeChil
             }
             const run = async () => {
                 const _diskWin = readEntryCoalescedDiskWindow(entry);
-                const readEntry = _diskWin
+                let readEntry = _diskWin
                     ? { ...entry, offset: _diskWin.offset, limit: _diskWin.limit }
                     : entry;
+                if ((!readEntry.mode || readEntry.mode === 'full')
+                    && readEntry.offset == null
+                    && readEntry.limit == null
+                    && readEntry.full !== true) {
+                    readEntry = { ...readEntry, offset: 0, limit: 2000 };
+                }
                 // Full image children retain their rich blocks; the aggregate
                 // assembler below flattens them without stringification.
                 // Other media (PDF/notebook) remains text-only in a batch so
@@ -332,6 +354,9 @@ export async function executeReadTool(args, workDir, readStateScope, executeChil
                     suppressReadUnchangedStub: true,
                     mediaTextOnly: !richImage,
                     _skipReachPreflight: true,
+                    forceReadRangeStream: true,
+                    readOutputBudgetBytes: _perTaskOutputBudget,
+                    toolOutputMaxBytes: _perTaskOutputBudget,
                 });
                 results[index] = { path: entry.path, mode: entry.mode || 'full', n: entry.n, body };
             };
@@ -470,11 +495,13 @@ export async function executeReadTool(args, workDir, readStateScope, executeChil
                 const r = orderedResults[i];
                 const path = normalizeOutputPath(r.path);
                 const mode = r.n !== undefined ? `${r.mode} n=${r.n}` : r.mode;
+                // Default full mode carries no information — tag only non-default modes.
+                const modeTag = mode && mode !== 'full' ? ` [${mode}]` : '';
                 const status = bodyFailed(r.body) ? 'error' : 'ok';
                 const textBody = bodyTextFor(r.body);
                 const match = /\[TRUNCATED (?:—|-) file is (\d+) lines \/ (\d+) KB\./.exec(textBody);
                 const suffix = match ? ` (truncated ${match[1]}L/${match[2]}KB)` : '';
-                const entryHeader = `${path} [${mode}] [${status}]${suffix}`;
+                const entryHeader = `${path}${modeTag} [${status}]${suffix}`;
                 const richParts = richPartsFor(r.body);
                 let priorIdx;
                 if (richParts) {
@@ -513,16 +540,17 @@ export async function executeReadTool(args, workDir, readStateScope, executeChil
         const body = orderedResults.map((r, _i) => {
             const path = normalizeOutputPath(r.path);
             const mode = r.n !== undefined ? `${r.mode} n=${r.n}` : r.mode;
+            const modeTag = mode && mode !== 'full' ? ` [${mode}]` : '';
             const status = classifyResultKind(String(r.body || '')) === 'error' ? 'error' : 'ok';
             const dupKey = JSON.stringify([path, mode, r.body || '']);
             const priorIdx = _seenEntryBody.get(dupKey);
             if (priorIdx !== undefined) {
-                return `${path} [${mode}] [${status}] [= entry #${priorIdx + 1}, identical result omitted]`;
+                return `${path}${modeTag} [${status}] [= entry #${priorIdx + 1}, identical result omitted]`;
             }
             _seenEntryBody.set(dupKey, _i);
             const match = /\[TRUNCATED (?:—|-) file is (\d+) lines \/ (\d+) KB\./.exec(r.body || '');
             const suffix = match ? ` (truncated ${match[1]}L/${match[2]}KB)` : '';
-            return `${path} [${mode}] [${status}]${suffix}\n${r.body}`;
+            return `${path}${modeTag} [${status}]${suffix}\n${r.body}`;
         }).join('\n\n');
         return `${header}\n\n${body}${args._batchCapNote ? `\n\n${args._batchCapNote}` : ''}`;
     }

package/src/runtime/agent/orchestrator/tools/builtin/rg-runner.mjs

@@ -302,7 +302,7 @@ export async function rgSupportsPcre2() {
 }
 
 // Cap rg stdout accumulation so a runaway producer (huge repo, accidental
-// match-all) cannot balloon the JS string heap. Mirrors CC's ripgrep.ts cap.
+// match-all) cannot balloon the JS string heap.
 const MAX_RG_STDOUT_BYTES = 20 * 1024 * 1024; // 20MB
 
 // SIGTERM → grace → force kill; hard Promise settle if 'close' never fires

package/src/runtime/agent/orchestrator/tools/builtin/search-builders.mjs

@@ -31,7 +31,9 @@ export function buildGrepCacheKey(parts) {
         onlyMatching,
         pcre2,
         withFilename,
+        contextCharBudget = 0,
         patternCapTotal = 0,
+        candidatesKey = '',
     } = parts;
     return [
         'grep',
@@ -51,10 +53,15 @@ export function buildGrepCacheKey(parts) {
         Array.isArray(fileType) ? fileType.join('\x01') : (fileType || ''),
         pcre2 ? 'p1' : 'p0',
         withFilename ? 'H1' : 'H0',
+        'cb' + String(contextCharBudget || 0),
         // Cap total keeps a capped request (first-N of M patterns, carrying the
         // "[capped at N of M]" notice) from colliding with an exact N-pattern
         // request or with a differently-capped one (of 15 vs of 20).
         'pc' + String(patternCapTotal || 0),
+        // Candidate-file scoping (fan-out prefilter) changes which files rg
+        // visits and their order; scoped results get their own cache slot so
+        // they never collide with an unscoped run of the same pattern.
+        'cd' + String(candidatesKey || ''),
     ].join('|');
 }
 
@@ -75,6 +82,7 @@ export function buildGrepRgArgs(parts) {
         fixedStrings = false,
         pcre2 = false,
         withFilename = false,
+        candidateFiles = null,
     } = parts;
     // `--hidden`: search dotfiles/dot-dirs (.github, .mixdog) that
     // rg skips by default. The DEFAULT_IGNORE_GLOBS below still exclude .git and
@@ -127,7 +135,14 @@ export function buildGrepRgArgs(parts) {
     // `--type` is treated as a positional path, not parsed as an rg
     // option. Patterns already use `-e`, so the separator only needs to
     // guard the trailing path operand.
-    rgArgs.push('--', searchPath);
+    if (Array.isArray(candidateFiles) && candidateFiles.length > 0) {
+        // Fan-out prefilter scoping: search ONLY the candidate files (paths
+        // relative to the spawn cwd) instead of walking searchPath — explicit
+        // file operands skip the directory walk entirely.
+        rgArgs.push('--', ...candidateFiles);
+    } else {
+        rgArgs.push('--', searchPath);
+    }
     return rgArgs;
 }