@jsonstudio/llms 0.6.753 → 0.6.802

package/dist/tools/tool-registry.js

@@ -1,5 +1,9 @@
 // Tool registry and validator (single source of truth)
-import { buildStructuredPatch, isStructuredApplyPatchPayload, StructuredApplyPatchError } from './apply-patch-structured.js';
+import { parseToolArgsJson } from './args-json.js';
+import { validateApplyPatchArgs } from './apply-patch/validator.js';
+import { captureApplyPatchRegression } from './patch-regression-capturer.js';
+import { validateExecCommandArgs } from './exec-command/validator.js';
+import { captureExecCommandRegression } from './exec-command/regression-capturer.js';
 const isRecord = (value) => typeof value === 'object' && value !== null && !Array.isArray(value);
 const asString = (value) => {
     if (typeof value === 'string' && value.trim().length > 0) {
@@ -27,471 +31,7 @@ const readNumber = (value) => {
     }
     return value;
 };
-const stripArgKeyArtifacts = (input) => {
-    try {
-        return String(input || '')
-            .replace(/<\/?\s*tool_call[^>]*>/gi, '')
-            .replace(/<\/?\s*arg_key\s*>/gi, '')
-            .replace(/<\/?\s*arg_value\s*>/gi, '');
-    }
-    catch {
-        return input;
-    }
-};
-const coercePrimitive = (raw) => {
-    const trimmed = raw.trim();
-    if (!trimmed)
-        return '';
-    if (/^(true|false)$/i.test(trimmed))
-        return /^true$/i.test(trimmed);
-    if (/^-?\d+(?:\.\d+)?$/.test(trimmed))
-        return Number(trimmed);
-    if ((trimmed.startsWith('{') && trimmed.endsWith('}')) ||
-        (trimmed.startsWith('[') && trimmed.endsWith(']'))) {
-        try {
-            return JSON.parse(trimmed);
-        }
-        catch {
-            return trimmed;
-        }
-    }
-    return trimmed;
-};
-const extractInjectedArgPairs = (raw) => {
-    const delimiter = '</arg_key><arg_value>';
-    if (typeof raw !== 'string' || !raw.includes(delimiter))
-        return null;
-    const parts = raw.split(delimiter);
-    if (parts.length < 2)
-        return null;
-    const looksLikeKey = (s) => /^[A-Za-z_][A-Za-z0-9_-]*$/.test(s.trim());
-    const pairs = [];
-    let baseValue = parts[0] ?? '';
-    if (parts.length === 2) {
-        const k = (parts[0] ?? '').trim();
-        const v = (parts[1] ?? '').trim();
-        if (looksLikeKey(k) && v.length > 0) {
-            baseValue = '';
-            pairs.push({ key: k, value: coercePrimitive(v) });
-        }
-        return pairs.length ? { baseValue, pairs } : null;
-    }
-    for (let i = 1; i + 1 < parts.length; i += 2) {
-        const key = (parts[i] ?? '').trim();
-        const rawValue = (parts[i + 1] ?? '').trim();
-        if (!looksLikeKey(key)) {
-            continue;
-        }
-        if (rawValue.length === 0) {
-            continue;
-        }
-        pairs.push({ key, value: coercePrimitive(rawValue) });
-    }
-    if (!pairs.length)
-        return null;
-    return { baseValue, pairs };
-};
-const repairArgKeyArtifactsInObject = (value) => {
-    const visit = (node) => {
-        if (Array.isArray(node)) {
-            for (const entry of node) {
-                visit(entry);
-            }
-            return;
-        }
-        if (!isRecord(node))
-            return;
-        for (const [k, v] of Object.entries(node)) {
-            if (typeof v === 'string') {
-                const injected = extractInjectedArgPairs(v);
-                if (injected) {
-                    if (injected.baseValue !== '') {
-                        node[k] = injected.baseValue;
-                    }
-                    for (const pair of injected.pairs) {
-                        if (!Object.prototype.hasOwnProperty.call(node, pair.key)) {
-                            node[pair.key] = pair.value;
-                        }
-                    }
-                }
-            }
-            visit(node[k]);
-        }
-    };
-    visit(value);
-};
-const tryParseJson = (input) => {
-    const raw = typeof input === 'string' ? input : '';
-    if (!raw.trim())
-        return {};
-    try {
-        const parsed = JSON.parse(raw);
-        repairArgKeyArtifactsInObject(parsed);
-        return parsed;
-    }
-    catch {
-        // try stripping common tool-call markup artifacts and parsing again
-        try {
-            const stripped = stripArgKeyArtifacts(raw).trim();
-            if (stripped && stripped !== raw) {
-                const parsed = JSON.parse(stripped);
-                repairArgKeyArtifactsInObject(parsed);
-                return parsed;
-            }
-        }
-        catch {
-            // continue
-        }
-        // attempt to parse the first JSON container substring
-        try {
-            const candidate = raw.match(/\{[\s\S]*\}|\[[\s\S]*\]/)?.[0];
-            if (candidate) {
-                const strippedCandidate = stripArgKeyArtifacts(candidate).trim();
-                const parsed = JSON.parse(strippedCandidate);
-                repairArgKeyArtifactsInObject(parsed);
-                return parsed;
-            }
-        }
-        catch {
-            // ignore
-        }
-        return {};
-    }
-};
-const stripCodeFences = (text) => {
-    const trimmed = text.trim();
-    // Only treat the entire payload as fenced when it *starts* with a code fence.
-    // Patch bodies (especially added Markdown files) may legitimately contain ``` blocks;
-    // we must not strip those.
-    if (!trimmed.startsWith('```')) {
-        return text;
-    }
-    const fenceRe = /^```(?:diff|patch|apply_patch|text|json)?[ \t]*\n([\s\S]*?)\n```/gmi;
-    const candidates = [];
-    let match = null;
-    while ((match = fenceRe.exec(trimmed))) {
-        if (match[1]) {
-            candidates.push(match[1].trim());
-        }
-    }
-    if (!candidates.length) {
-        return text;
-    }
-    for (const candidate of candidates) {
-        if (candidate.includes('*** Begin Patch') ||
-            candidate.includes('*** Update File:') ||
-            candidate.includes('diff --git')) {
-            return candidate;
-        }
-    }
-    return candidates[0] ?? text;
-};
-const looksLikePatch = (text) => {
-    if (!text)
-        return false;
-    const t = text.trim();
-    if (!t)
-        return false;
-    return (t.includes('*** Begin Patch') ||
-        t.includes('*** Update File:') ||
-        t.includes('*** Add File:') ||
-        t.includes('*** Delete File:') ||
-        t.includes('diff --git') ||
-        /^(?:@@|\+\+\+\s|---\s)/m.test(t));
-};
-const convertGitDiffToApplyPatch = (text) => {
-    const lines = text.replace(/\r\n/g, '\n').split('\n');
-    const files = [];
-    let current = null;
-    let sawDiff = false;
-    const flush = () => {
-        if (!current)
-            return;
-        if (current.path && current.kind === 'delete') {
-            files.push(current);
-            current = null;
-            return;
-        }
-        if (!current.path || (current.kind === 'update' && current.lines.length === 0)) {
-            current = null;
-            return;
-        }
-        files.push(current);
-        current = null;
-    };
-    for (const raw of lines) {
-        const line = raw;
-        const diffMatch = line.match(/^diff --git a\/(.+?) b\/(.+)$/);
-        if (diffMatch) {
-            sawDiff = true;
-            flush();
-            const path = diffMatch[2] || diffMatch[1];
-            current = { path, kind: 'update', lines: [] };
-            continue;
-        }
-        if (!current) {
-            continue;
-        }
-        if (line.startsWith('new file mode')) {
-            current.kind = 'add';
-            continue;
-        }
-        if (line.startsWith('deleted file mode')) {
-            current.kind = 'delete';
-            continue;
-        }
-        if (line.startsWith('index ') || line.startsWith('old mode') || line.startsWith('new mode')) {
-            continue;
-        }
-        if (line.startsWith('--- ')) {
-            if (line.includes('/dev/null')) {
-                current.kind = 'add';
-            }
-            continue;
-        }
-        if (line.startsWith('+++ ')) {
-            if (line.includes('/dev/null')) {
-                current.kind = 'delete';
-            }
-            else {
-                const plusMatch = line.match(/^\+\+\+\s+(?:b\/)?(.+)$/);
-                if (plusMatch && plusMatch[1]) {
-                    current.path = plusMatch[1];
-                }
-            }
-            continue;
-        }
-        if (current.kind === 'delete') {
-            continue;
-        }
-        if (current.kind === 'add') {
-            if (line.startsWith('+') && !line.startsWith('+++')) {
-                current.lines.push(line);
-            }
-            else if (line.startsWith('\\')) {
-                current.lines.push(`+${line}`);
-            }
-            continue;
-        }
-        if (line.startsWith('@@') ||
-            line.startsWith('+') ||
-            line.startsWith('-') ||
-            line.startsWith(' ') ||
-            line.startsWith('\\')) {
-            current.lines.push(line.startsWith('\\') ? ` ${line}` : line);
-        }
-    }
-    flush();
-    if (!sawDiff || files.length === 0) {
-        return null;
-    }
-    const output = ['*** Begin Patch'];
-    for (const file of files) {
-        if (!file.path)
-            continue;
-        if (file.kind === 'add') {
-            output.push(`*** Add File: ${file.path}`);
-            for (const line of file.lines) {
-                if (line.startsWith('+')) {
-                    output.push(line);
-                }
-                else {
-                    output.push(`+${line}`);
-                }
-            }
-            continue;
-        }
-        if (file.kind === 'delete') {
-            output.push(`*** Delete File: ${file.path}`);
-            continue;
-        }
-        output.push(`*** Update File: ${file.path}`);
-        for (const line of file.lines) {
-            if (line.startsWith('@@') || line.startsWith('+') || line.startsWith('-') || line.startsWith(' ')) {
-                output.push(line);
-            }
-            else {
-                output.push(` ${line}`);
-            }
-        }
-    }
-    output.push('*** End Patch');
-    return output.join('\n');
-};
-const decodeEscapedNewlinesIfNeeded = (value) => {
-    if (!value)
-        return value;
-    if (value.includes('\n'))
-        return value;
-    if (!value.includes('\\n') && !value.includes('\\r') && !value.toLowerCase().includes('\\u000a') && !value.toLowerCase().includes('\\u000d')) {
-        return value;
-    }
-    // Only decode when the patch appears to be newline-escaped into a single line.
-    // This avoids corrupting legitimate file content that contains "\n" sequences.
-    let out = value;
-    out = out.replace(/\\r\\n/g, '\n');
-    out = out.replace(/\\n/g, '\n');
-    out = out.replace(/\\r/g, '\n');
-    out = out.replace(/\\u000a/gi, '\n');
-    out = out.replace(/\\u000d/gi, '\n');
-    return out;
-};
-const normalizeApplyPatchText = (raw) => {
-    if (!raw)
-        return raw;
-    let text = raw.replace(/\r\n/g, '\n');
-    text = decodeEscapedNewlinesIfNeeded(text);
-    text = stripCodeFences(text);
-    text = text.trim();
-    if (!text) {
-        return raw;
-    }
-    if (!text.includes('*** Begin Patch') && text.includes('diff --git')) {
-        const converted = convertGitDiffToApplyPatch(text);
-        if (converted) {
-            text = converted;
-        }
-    }
-    else if (!text.includes('*** Begin Patch') && !text.includes('diff --git')) {
-        // 兼容只包含 --- / +++ / @@ 的 Git diff 片段（无 diff --git 头）
-        const minusMatch = text.match(/^---\s+(.*)$/m);
-        const plusMatch = text.match(/^\+\+\+\s+(.*)$/m);
-        if (minusMatch && plusMatch) {
-            const rawPlus = plusMatch[1] || '';
-            const pathMatch = rawPlus.match(/^(?:b\/)?(.+)$/);
-            const path = (pathMatch && pathMatch[1] ? pathMatch[1] : rawPlus).trim();
-            if (path) {
-                const synthetic = `diff --git a/${path} b/${path}\n${text}`;
-                const converted = convertGitDiffToApplyPatch(synthetic);
-                if (converted) {
-                    text = converted;
-                }
-            }
-        }
-    }
-    if (text.includes('*** Create File:')) {
-        text = text.replace(/\*\*\* Create File:/g, '*** Add File:');
-    }
-    let hasBegin = text.includes('*** Begin Patch');
-    const hasEnd = text.includes('*** End Patch');
-    if (hasBegin && !hasEnd) {
-        text = `${text}\n*** End Patch`;
-    }
-    if (!hasBegin && /^\*\*\* (Add|Update|Delete) File:/m.test(text)) {
-        text = `*** Begin Patch\n${text}\n*** End Patch`;
-        hasBegin = true;
-    }
-    if (!text.includes('*** Begin Patch')) {
-        return text;
-    }
-    // 容错处理：如果补丁前后包裹了多余的 JSON / 文本（例如 {"patch":"*** Begin Patch ... *** End Patch"}），
-    // 则截断到第一个 '*** Begin Patch' 与 '*** End Patch' 之间的主体，恢复为标准 apply_patch 片段。
-    const beginIndex = text.indexOf('*** Begin Patch');
-    if (beginIndex > 0) {
-        text = text.slice(beginIndex);
-    }
-    const endMarker = '*** End Patch';
-    const firstEndIndex = text.indexOf(endMarker);
-    const concatSignatures = [
-        `${endMarker}","input":"*** Begin Patch`,
-        `${endMarker}","patch":"*** Begin Patch`,
-        `${endMarker}\\",\\"input\\":\\"*** Begin Patch`,
-        `${endMarker}\\",\\"patch\\":\\"*** Begin Patch`
-    ];
-    const hasConcatenationSignal = concatSignatures.some((sig) => text.includes(sig));
-    if (hasConcatenationSignal && firstEndIndex >= 0) {
-        text = text.slice(0, firstEndIndex + endMarker.length);
-    }
-    else {
-        const lastEndIndex = text.lastIndexOf(endMarker);
-        if (lastEndIndex >= 0) {
-            const afterEnd = text.slice(lastEndIndex + endMarker.length);
-            if (afterEnd.trim().length > 0) {
-                text = text.slice(0, lastEndIndex + endMarker.length);
-            }
-        }
-    }
-    const lines = text.split('\n');
-    const output = [];
-    let inUpdateSection = false;
-    let afterUpdateHeader = false;
-    for (const line of lines) {
-        if (line.startsWith('*** Begin Patch')) {
-            output.push(line);
-            inUpdateSection = false;
-            afterUpdateHeader = false;
-            continue;
-        }
-        if (line.startsWith('*** End Patch')) {
-            output.push(line);
-            inUpdateSection = false;
-            afterUpdateHeader = false;
-            continue;
-        }
-        if (line.startsWith('*** Update File:')) {
-            output.push(line);
-            inUpdateSection = true;
-            afterUpdateHeader = true;
-            continue;
-        }
-        if (line.startsWith('*** Add File:') || line.startsWith('*** Delete File:')) {
-            output.push(line);
-            inUpdateSection = false;
-            afterUpdateHeader = false;
-            continue;
-        }
-        if (inUpdateSection) {
-            if (afterUpdateHeader && line.trim() === '') {
-                continue;
-            }
-            afterUpdateHeader = false;
-            if (line.startsWith('@@') || line.startsWith('+') || line.startsWith('-') || line.startsWith(' ')) {
-                output.push(line);
-            }
-            else {
-                output.push(` ${line}`);
-            }
-            continue;
-        }
-        output.push(line);
-    }
-    return output.join('\n');
-};
-const buildSingleChangePayload = (record) => {
-    const kindRaw = asString(record.kind);
-    if (!kindRaw)
-        return undefined;
-    const change = {
-        kind: kindRaw.toLowerCase(),
-        lines: record.lines ?? record.text ?? record.body,
-        target: asString(record.target) ?? undefined,
-        anchor: asString(record.anchor) ?? undefined
-    };
-    if (typeof record.use_anchor_indent === 'boolean') {
-        change.use_anchor_indent = record.use_anchor_indent;
-    }
-    const changeFile = asString(record.file);
-    if (changeFile) {
-        change.file = changeFile;
-    }
-    return {
-        file: changeFile,
-        changes: [change]
-    };
-};
-const coerceStructuredPayload = (record) => {
-    if (isStructuredApplyPatchPayload(record)) {
-        return record;
-    }
-    if (Array.isArray(record.changes) && record.changes.length === 0) {
-        return undefined;
-    }
-    const single = buildSingleChangePayload(record);
-    if (single) {
-        return single;
-    }
-    return undefined;
-};
+// JSON parsing/repair helpers are shared via tools/args-json.ts
 const toJson = (value) => {
     try {
         return JSON.stringify(value ?? {});
@@ -500,62 +40,6 @@ const toJson = (value) => {
         return '{}';
     }
 };
-const extractApplyPatchText = (argsString, rawArgs) => {
-    const rawTrimmed = typeof argsString === 'string' ? argsString.trim() : '';
-    const looksJsonContainer = rawTrimmed.startsWith('{') || rawTrimmed.startsWith('[');
-    if (!looksJsonContainer && looksLikePatch(rawTrimmed)) {
-        return { patchText: normalizeApplyPatchText(rawTrimmed) };
-    }
-    if (isRecord(rawArgs)) {
-        // Special case: argsString claims to be JSON but isn't parseable (tryParseJson fallback returns {}),
-        // and the raw text looks like a patch. Treat it as a patch string to avoid dropping edits.
-        if (looksJsonContainer && Object.keys(rawArgs).length === 0 && looksLikePatch(rawTrimmed)) {
-            return { patchText: normalizeApplyPatchText(rawTrimmed) };
-        }
-        const patchField = asString(rawArgs.patch);
-        if (patchField && looksLikePatch(patchField)) {
-            return { patchText: normalizeApplyPatchText(patchField) };
-        }
-        const inputField = asString(rawArgs.input);
-        if (inputField && looksLikePatch(inputField)) {
-            return { patchText: normalizeApplyPatchText(inputField) };
-        }
-        const payload = coerceStructuredPayload(rawArgs);
-        if (payload) {
-            try {
-                return { patchText: buildStructuredPatch(payload) };
-            }
-            catch (error) {
-                if (!(error instanceof StructuredApplyPatchError)) {
-                    throw error;
-                }
-                return { failureReason: error.reason || 'structured_apply_patch_error' };
-            }
-        }
-    }
-    else if (Array.isArray(rawArgs) && rawArgs.length > 0) {
-        // 兼容形态：arguments 直接是单元素数组 [{ file, changes: [...] }]
-        const first = rawArgs.find((entry) => isRecord(entry));
-        if (first) {
-            const payload = coerceStructuredPayload(first);
-            if (payload) {
-                try {
-                    return { patchText: buildStructuredPatch(payload) };
-                }
-                catch (error) {
-                    if (!(error instanceof StructuredApplyPatchError)) {
-                        throw error;
-                    }
-                    return { failureReason: error.reason || 'structured_apply_patch_error' };
-                }
-            }
-        }
-    }
-    else if (typeof rawArgs === 'string' && looksLikePatch(rawArgs)) {
-        return { patchText: normalizeApplyPatchText(rawArgs) };
-    }
-    return {};
-};
 const detectForbiddenWrite = (script) => {
     const normalized = script.toLowerCase();
     if (!normalized) {
@@ -588,6 +72,7 @@ const isImagePath = (value) => {
 export function getAllowedToolNames() {
     return [
         'shell',
+        'exec_command',
         'apply_patch',
         'update_plan',
         'view_image',
@@ -605,27 +90,37 @@ export function validateToolCall(name, argsString) {
     if (!allowed.has(normalizedName)) {
         return { ok: false, reason: 'unknown_tool' };
     }
-    const rawArgsAny = tryParseJson(typeof argsString === 'string' ? argsString : '{}');
+    const rawArgsAny = parseToolArgsJson(typeof argsString === 'string' ? argsString : '{}');
     switch (normalizedName) {
+        case 'exec_command': {
+            const validation = validateExecCommandArgs(argsString, rawArgsAny);
+            if (!validation.ok) {
+                const reason = validation.reason || 'unknown';
+                captureExecCommandRegression({
+                    errorType: reason,
+                    originalArgs: typeof argsString === 'string' ? argsString : String(argsString ?? ''),
+                    normalizedArgs: typeof argsString === 'string' ? argsString : String(argsString ?? ''),
+                    validationError: reason,
+                    source: 'tool-registry.validateToolCall'
+                });
+                return { ok: false, reason };
+            }
+            return { ok: true, normalizedArgs: validation.normalizedArgs };
+        }
         case 'apply_patch': {
-            const extracted = extractApplyPatchText(argsString, rawArgsAny);
-            const patchText = extracted.patchText;
-            if (!patchText) {
-                const rawTrimmed = typeof argsString === 'string' ? argsString.trim() : '';
-                const looksJsonContainer = rawTrimmed.startsWith('{') || rawTrimmed.startsWith('[');
-                if (extracted.failureReason) {
-                    return { ok: false, reason: extracted.failureReason };
-                }
-                if (looksJsonContainer &&
-                    isRecord(rawArgsAny) &&
-                    Object.keys(rawArgsAny).length === 0 &&
-                    rawTrimmed.length > 0 &&
-                    !looksLikePatch(rawTrimmed)) {
-                    return { ok: false, reason: 'invalid_json' };
-                }
-                return { ok: false, reason: 'missing_changes' };
-            }
-            return { ok: true, normalizedArgs: toJson({ patch: patchText, input: patchText }) };
+            const validation = validateApplyPatchArgs(argsString, rawArgsAny);
+            if (!validation.ok) {
+                const reason = validation.reason || 'unknown';
+                captureApplyPatchRegression({
+                    errorType: reason,
+                    originalArgs: typeof argsString === 'string' ? argsString : String(argsString ?? ''),
+                    normalizedArgs: typeof argsString === 'string' ? argsString : String(argsString ?? ''),
+                    validationError: reason,
+                    source: 'tool-registry.validateToolCall'
+                });
+                return { ok: false, reason };
+            }
+            return { ok: true, normalizedArgs: validation.normalizedArgs };
         }
         case 'shell': {
             const rawArgs = isRecord(rawArgsAny) ? rawArgsAny : {};

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@jsonstudio/llms",
-  "version": "0.6.753",
+  "version": "0.6.802",
   "type": "module",
   "main": "dist/index.js",
   "module": "dist/index.js",