extrait 0.1.2 → 0.2.0

package/dist/index.cjs

@@ -298,6 +298,9 @@ function isOnlyWhitespace(value) {
 }
 
 // src/extract.ts
+var RE_EMPTY_OBJECT = /^\{\s*\}$/;
+var RE_EMPTY_ARRAY = /^\[\s*\]$/;
+var RE_BOUNDARY_CHAR = /[\s,.;:!?`"'()\[\]{}<>]/;
 var DEFAULT_EXTRACTION_HEURISTICS = {
   firstPassMin: 12,
   firstPassCap: 30,
@@ -522,7 +525,7 @@ function jsonShapeScore(content, acceptArrays) {
   const commaCount = countChar(trimmed, ",");
   const quoteCount = countChar(trimmed, '"');
   if (root === "{") {
-    if (/^\{\s*\}$/.test(trimmed)) {
+    if (RE_EMPTY_OBJECT.test(trimmed)) {
       score += 12;
     } else if (colonCount > 0) {
       score += 22;
@@ -533,7 +536,7 @@ function jsonShapeScore(content, acceptArrays) {
       score += quoteCount % 2 === 0 ? 8 : -8;
     }
   } else {
-    score += /^\[\s*\]$/.test(trimmed) ? 8 : 4;
+    score += RE_EMPTY_ARRAY.test(trimmed) ? 8 : 4;
     if (colonCount > 0) {
       score += 4;
     }
@@ -566,7 +569,7 @@ function isBoundary(char) {
   if (!char) {
     return true;
   }
-  return /[\s,.;:!?`"'()[\]{}<>]/.test(char);
+  return RE_BOUNDARY_CHAR.test(char);
 }
 function lengthScore(length) {
   return Math.min(120, Math.floor(Math.sqrt(length) * 6));
@@ -755,6 +758,8 @@ function clamp(value, min, max) {
   return Math.max(min, Math.min(max, Math.floor(value)));
 }
 // src/schema.ts
+var RE_SIMPLE_IDENTIFIER = /^[A-Za-z_$][A-Za-z0-9_$]*$/;
+var RE_WHITESPACE = /\s+/g;
 function formatZodSchemaLikeTypeScript(schema) {
   return formatType(schema, 0, new WeakSet);
 }
@@ -926,7 +931,7 @@ ${lines.join(`
 ${indent}}`;
 }
 function formatKey(key) {
-  return /^[A-Za-z_$][A-Za-z0-9_$]*$/.test(key) ? key : JSON.stringify(key);
+  return RE_SIMPLE_IDENTIFIER.test(key) ? key : JSON.stringify(key);
 }
 function requiresParentheses(typeText) {
   return typeText.includes(" | ") || typeText.includes(" & ");
@@ -972,7 +977,7 @@ function readSchemaDescription(schema) {
   return;
 }
 function sanitizeDescription(value) {
-  return value.replace(/\s+/g, " ").trim();
+  return value.replace(RE_WHITESPACE, " ").trim();
 }
 
 // src/format.ts
@@ -997,6 +1002,8 @@ function resolveSchemaInstruction(instruction) {
 }
 // src/think.ts
 var THINK_TAG_NAME = "think";
+var RE_IDENTIFIER_CHAR = /[a-zA-Z0-9:_-]/;
+var RE_NON_LINE_BREAK = /[^\r\n]/g;
 function sanitizeThink(input) {
   const thinkBlocks = [];
   const diagnostics = {
@@ -1095,6 +1102,9 @@ function parseThinkTagAt(input, index) {
   if (input[cursor] === "/") {
     closing = true;
     cursor += 1;
+    while (cursor < input.length && isWhitespace(input[cursor])) {
+      cursor += 1;
+    }
   }
   if (!matchesIgnoreCase(input, cursor, THINK_TAG_NAME)) {
     return null;
@@ -1151,7 +1161,7 @@ function matchesIgnoreCase(input, index, expected) {
   return input.slice(index, index + expected.length).toLowerCase() === expected;
 }
 function isIdentifierChar(char) {
-  return /[a-zA-Z0-9:_-]/.test(char);
+  return RE_IDENTIFIER_CHAR.test(char);
 }
 function countHiddenChars(value) {
   let count = 0;
@@ -1165,9 +1175,10 @@ function countHiddenChars(value) {
   return count;
 }
 function maskKeepingLineBreaks(value) {
-  return value.replace(/[^\r\n]/g, " ");
+  return value.replace(RE_NON_LINE_BREAK, " ");
 }
 // src/providers/stream-utils.ts
+var RE_LINE_ENDING = /\r?\n/;
 async function consumeSSE(response, onEvent) {
   if (!response.body) {
     return;
@@ -1190,7 +1201,7 @@ async function consumeSSE(response, onEvent) {
       buffer = buffer.slice(boundary + (buffer.startsWith(`\r
 \r
 `, boundary) ? 4 : 2));
-      const dataLines = rawEvent.split(/\r?\n/).filter((line) => line.startsWith("data:")).map((line) => line.slice(5).trim());
+      const dataLines = rawEvent.split(RE_LINE_ENDING).filter((line) => line.startsWith("data:")).map((line) => line.slice(5).trim());
       if (dataLines.length === 0) {
         continue;
       }
@@ -1200,7 +1211,7 @@ async function consumeSSE(response, onEvent) {
   }
   const remainder = buffer.trim();
   if (remainder.length > 0) {
-    const dataLines = remainder.split(/\r?\n/).filter((line) => line.startsWith("data:")).map((line) => line.slice(5).trim());
+    const dataLines = remainder.split(RE_LINE_ENDING).filter((line) => line.startsWith("data:")).map((line) => line.slice(5).trim());
     if (dataLines.length > 0) {
       onEvent(dataLines.join(`
 `));
@@ -1490,13 +1501,18 @@ function describeTool(clientId, tool, hasCollision) {
   }
   return;
 }
+var RE_NON_ALPHANUMERIC = /[^A-Za-z0-9_]/g;
+var RE_MULTIPLE_UNDERSCORES = /_+/g;
+var RE_LEADING_UNDERSCORES = /^_+/;
+var RE_TRAILING_UNDERSCORES = /_+$/;
+var RE_STARTS_WITH_DIGIT = /^[0-9]/;
 function sanitizeToolName(input) {
-  const sanitized = input.replace(/[^A-Za-z0-9_]/g, "_").replace(/_+/g, "_");
-  const trimmed = sanitized.replace(/^_+/, "").replace(/_+$/, "");
+  const sanitized = input.replace(RE_NON_ALPHANUMERIC, "_").replace(RE_MULTIPLE_UNDERSCORES, "_");
+  const trimmed = sanitized.replace(RE_LEADING_UNDERSCORES, "").replace(RE_TRAILING_UNDERSCORES, "");
   if (!trimmed) {
     return "tool";
   }
-  if (/^[0-9]/.test(trimmed)) {
+  if (RE_STARTS_WITH_DIGIT.test(trimmed)) {
     return `tool_${trimmed}`;
   }
   return trimmed;
@@ -1687,8 +1703,6 @@ async function completeWithChatCompletionsPassThrough(options, fetcher, path, re
   };
 }
 async function completeWithChatCompletionsWithMCP(options, fetcher, path, request) {
-  const mcpToolset = await resolveMCPToolset(request.mcpClients);
-  const transportTools = toProviderFunctionTools(mcpToolset);
   const maxToolRounds = normalizeMaxToolRounds(request.maxToolRounds ?? options.defaultMaxToolRounds);
   let messages = buildMessages(request);
   let aggregatedUsage;
@@ -1697,6 +1711,8 @@ async function completeWithChatCompletionsWithMCP(options, fetcher, path, reques
   const toolCalls = [];
   const toolExecutions = [];
   for (let round = 1;round <= maxToolRounds + 1; round += 1) {
+    const mcpToolset = await resolveMCPToolset(request.mcpClients);
+    const transportTools = toProviderFunctionTools(mcpToolset);
     const response = await fetcher(buildURL(options.baseURL, path), {
       method: "POST",
       headers: buildHeaders(options),
@@ -1792,8 +1808,6 @@ async function completeWithResponsesAPIPassThrough(options, fetcher, path, reque
   };
 }
 async function completeWithResponsesAPIWithMCP(options, fetcher, path, request) {
-  const mcpToolset = await resolveMCPToolset(request.mcpClients);
-  const transportTools = toResponsesTools(toProviderFunctionTools(mcpToolset));
   const maxToolRounds = normalizeMaxToolRounds(request.maxToolRounds ?? options.defaultMaxToolRounds);
   let input = buildResponsesInput(request);
   let previousResponseId = pickString(isRecord2(request.body) ? request.body.previous_response_id : undefined);
@@ -1803,6 +1817,8 @@ async function completeWithResponsesAPIWithMCP(options, fetcher, path, request)
   const executedToolCalls = [];
   const toolExecutions = [];
   for (let round = 1;round <= maxToolRounds + 1; round += 1) {
+    const mcpToolset = await resolveMCPToolset(request.mcpClients);
+    const transportTools = toResponsesTools(toProviderFunctionTools(mcpToolset));
     const response = await fetcher(buildURL(options.baseURL, path), {
       method: "POST",
       headers: buildHeaders(options),
@@ -2242,8 +2258,6 @@ async function completePassThrough(options, fetcher, path, request) {
   };
 }
 async function completeWithMCPToolLoop(options, fetcher, path, request) {
-  const mcpToolset = await resolveMCPToolset(request.mcpClients);
-  const tools = toAnthropicTools(toProviderFunctionTools(mcpToolset));
   const maxToolRounds = normalizeMaxToolRounds(request.maxToolRounds ?? options.defaultMaxToolRounds);
   let messages = [{ role: "user", content: request.prompt }];
   let aggregatedUsage;
@@ -2252,6 +2266,8 @@ async function completeWithMCPToolLoop(options, fetcher, path, request) {
   const toolCalls = [];
   const toolExecutions = [];
   for (let round = 1;round <= maxToolRounds + 1; round += 1) {
+    const mcpToolset = await resolveMCPToolset(request.mcpClients);
+    const tools = toAnthropicTools(toProviderFunctionTools(mcpToolset));
     const response = await fetcher(buildURL(options.baseURL, path), {
       method: "POST",
       headers: buildHeaders2(options),
@@ -3061,6 +3077,9 @@ var DEFAULT_SELF_HEAL_PROTOCOL = "extrait.self-heal.v2";
 var DEFAULT_SELF_HEAL_MAX_CONTEXT_CHARS = 12000;
 var DEFAULT_SELF_HEAL_STOP_ON_NO_PROGRESS = true;
 var DEFAULT_SELF_HEAL_MAX_ERRORS = 8;
+var RE_SIMPLE_IDENTIFIER2 = /^[A-Za-z_$][A-Za-z0-9_$]*$/;
+var RE_ESCAPE_QUOTE = /"/g;
+var RE_WHITESPACE2 = /\s+/g;
 var DEFAULT_SELF_HEAL_MAX_DIAGNOSTICS = 8;
 var structuredOutdent = createOutdent({
   trimLeadingNewline: true,
@@ -3405,11 +3424,11 @@ function formatIssuePath(path) {
       out += `[${segment}]`;
       continue;
     }
-    if (/^[A-Za-z_$][A-Za-z0-9_$]*$/.test(segment)) {
+    if (RE_SIMPLE_IDENTIFIER2.test(segment)) {
       out += `.${segment}`;
       continue;
     }
-    out += `["${segment.replace(/"/g, "\\\"")}"]`;
+    out += `["${segment.replace(RE_ESCAPE_QUOTE, "\\\"")}"]`;
   }
   return out;
 }
@@ -3452,7 +3471,7 @@ function buildSelfHealFailureFingerprint(attempt) {
   return [issues, errors, source].join("::");
 }
 function normalizeWhitespace(value) {
-  return value.replace(/\s+/g, " ").trim();
+  return value.replace(RE_WHITESPACE2, " ").trim();
 }
 function normalizeStreamConfig(option) {
   if (typeof option === "boolean") {

package/dist/index.js

@@ -219,6 +219,9 @@ function isOnlyWhitespace(value) {
 }
 
 // src/extract.ts
+var RE_EMPTY_OBJECT = /^\{\s*\}$/;
+var RE_EMPTY_ARRAY = /^\[\s*\]$/;
+var RE_BOUNDARY_CHAR = /[\s,.;:!?`"'()\[\]{}<>]/;
 var DEFAULT_EXTRACTION_HEURISTICS = {
   firstPassMin: 12,
   firstPassCap: 30,
@@ -443,7 +446,7 @@ function jsonShapeScore(content, acceptArrays) {
   const commaCount = countChar(trimmed, ",");
   const quoteCount = countChar(trimmed, '"');
   if (root === "{") {
-    if (/^\{\s*\}$/.test(trimmed)) {
+    if (RE_EMPTY_OBJECT.test(trimmed)) {
       score += 12;
     } else if (colonCount > 0) {
       score += 22;
@@ -454,7 +457,7 @@ function jsonShapeScore(content, acceptArrays) {
       score += quoteCount % 2 === 0 ? 8 : -8;
     }
   } else {
-    score += /^\[\s*\]$/.test(trimmed) ? 8 : 4;
+    score += RE_EMPTY_ARRAY.test(trimmed) ? 8 : 4;
     if (colonCount > 0) {
       score += 4;
     }
@@ -487,7 +490,7 @@ function isBoundary(char) {
   if (!char) {
     return true;
   }
-  return /[\s,.;:!?`"'()[\]{}<>]/.test(char);
+  return RE_BOUNDARY_CHAR.test(char);
 }
 function lengthScore(length) {
   return Math.min(120, Math.floor(Math.sqrt(length) * 6));
@@ -676,6 +679,8 @@ function clamp(value, min, max) {
   return Math.max(min, Math.min(max, Math.floor(value)));
 }
 // src/schema.ts
+var RE_SIMPLE_IDENTIFIER = /^[A-Za-z_$][A-Za-z0-9_$]*$/;
+var RE_WHITESPACE = /\s+/g;
 function formatZodSchemaLikeTypeScript(schema) {
   return formatType(schema, 0, new WeakSet);
 }
@@ -847,7 +852,7 @@ ${lines.join(`
 ${indent}}`;
 }
 function formatKey(key) {
-  return /^[A-Za-z_$][A-Za-z0-9_$]*$/.test(key) ? key : JSON.stringify(key);
+  return RE_SIMPLE_IDENTIFIER.test(key) ? key : JSON.stringify(key);
 }
 function requiresParentheses(typeText) {
   return typeText.includes(" | ") || typeText.includes(" & ");
@@ -893,7 +898,7 @@ function readSchemaDescription(schema) {
   return;
 }
 function sanitizeDescription(value) {
-  return value.replace(/\s+/g, " ").trim();
+  return value.replace(RE_WHITESPACE, " ").trim();
 }
 
 // src/format.ts
@@ -918,6 +923,8 @@ function resolveSchemaInstruction(instruction) {
 }
 // src/think.ts
 var THINK_TAG_NAME = "think";
+var RE_IDENTIFIER_CHAR = /[a-zA-Z0-9:_-]/;
+var RE_NON_LINE_BREAK = /[^\r\n]/g;
 function sanitizeThink(input) {
   const thinkBlocks = [];
   const diagnostics = {
@@ -1016,6 +1023,9 @@ function parseThinkTagAt(input, index) {
   if (input[cursor] === "/") {
     closing = true;
     cursor += 1;
+    while (cursor < input.length && isWhitespace(input[cursor])) {
+      cursor += 1;
+    }
   }
   if (!matchesIgnoreCase(input, cursor, THINK_TAG_NAME)) {
     return null;
@@ -1072,7 +1082,7 @@ function matchesIgnoreCase(input, index, expected) {
   return input.slice(index, index + expected.length).toLowerCase() === expected;
 }
 function isIdentifierChar(char) {
-  return /[a-zA-Z0-9:_-]/.test(char);
+  return RE_IDENTIFIER_CHAR.test(char);
 }
 function countHiddenChars(value) {
   let count = 0;
@@ -1086,9 +1096,10 @@ function countHiddenChars(value) {
   return count;
 }
 function maskKeepingLineBreaks(value) {
-  return value.replace(/[^\r\n]/g, " ");
+  return value.replace(RE_NON_LINE_BREAK, " ");
 }
 // src/providers/stream-utils.ts
+var RE_LINE_ENDING = /\r?\n/;
 async function consumeSSE(response, onEvent) {
   if (!response.body) {
     return;
@@ -1111,7 +1122,7 @@ async function consumeSSE(response, onEvent) {
       buffer = buffer.slice(boundary + (buffer.startsWith(`\r
 \r
 `, boundary) ? 4 : 2));
-      const dataLines = rawEvent.split(/\r?\n/).filter((line) => line.startsWith("data:")).map((line) => line.slice(5).trim());
+      const dataLines = rawEvent.split(RE_LINE_ENDING).filter((line) => line.startsWith("data:")).map((line) => line.slice(5).trim());
       if (dataLines.length === 0) {
         continue;
       }
@@ -1121,7 +1132,7 @@ async function consumeSSE(response, onEvent) {
   }
   const remainder = buffer.trim();
   if (remainder.length > 0) {
-    const dataLines = remainder.split(/\r?\n/).filter((line) => line.startsWith("data:")).map((line) => line.slice(5).trim());
+    const dataLines = remainder.split(RE_LINE_ENDING).filter((line) => line.startsWith("data:")).map((line) => line.slice(5).trim());
     if (dataLines.length > 0) {
       onEvent(dataLines.join(`
 `));
@@ -1411,13 +1422,18 @@ function describeTool(clientId, tool, hasCollision) {
   }
   return;
 }
+var RE_NON_ALPHANUMERIC = /[^A-Za-z0-9_]/g;
+var RE_MULTIPLE_UNDERSCORES = /_+/g;
+var RE_LEADING_UNDERSCORES = /^_+/;
+var RE_TRAILING_UNDERSCORES = /_+$/;
+var RE_STARTS_WITH_DIGIT = /^[0-9]/;
 function sanitizeToolName(input) {
-  const sanitized = input.replace(/[^A-Za-z0-9_]/g, "_").replace(/_+/g, "_");
-  const trimmed = sanitized.replace(/^_+/, "").replace(/_+$/, "");
+  const sanitized = input.replace(RE_NON_ALPHANUMERIC, "_").replace(RE_MULTIPLE_UNDERSCORES, "_");
+  const trimmed = sanitized.replace(RE_LEADING_UNDERSCORES, "").replace(RE_TRAILING_UNDERSCORES, "");
   if (!trimmed) {
     return "tool";
   }
-  if (/^[0-9]/.test(trimmed)) {
+  if (RE_STARTS_WITH_DIGIT.test(trimmed)) {
     return `tool_${trimmed}`;
   }
   return trimmed;
@@ -1608,8 +1624,6 @@ async function completeWithChatCompletionsPassThrough(options, fetcher, path, re
   };
 }
 async function completeWithChatCompletionsWithMCP(options, fetcher, path, request) {
-  const mcpToolset = await resolveMCPToolset(request.mcpClients);
-  const transportTools = toProviderFunctionTools(mcpToolset);
   const maxToolRounds = normalizeMaxToolRounds(request.maxToolRounds ?? options.defaultMaxToolRounds);
   let messages = buildMessages(request);
   let aggregatedUsage;
@@ -1618,6 +1632,8 @@ async function completeWithChatCompletionsWithMCP(options, fetcher, path, reques
   const toolCalls = [];
   const toolExecutions = [];
   for (let round = 1;round <= maxToolRounds + 1; round += 1) {
+    const mcpToolset = await resolveMCPToolset(request.mcpClients);
+    const transportTools = toProviderFunctionTools(mcpToolset);
     const response = await fetcher(buildURL(options.baseURL, path), {
       method: "POST",
       headers: buildHeaders(options),
@@ -1713,8 +1729,6 @@ async function completeWithResponsesAPIPassThrough(options, fetcher, path, reque
   };
 }
 async function completeWithResponsesAPIWithMCP(options, fetcher, path, request) {
-  const mcpToolset = await resolveMCPToolset(request.mcpClients);
-  const transportTools = toResponsesTools(toProviderFunctionTools(mcpToolset));
   const maxToolRounds = normalizeMaxToolRounds(request.maxToolRounds ?? options.defaultMaxToolRounds);
   let input = buildResponsesInput(request);
   let previousResponseId = pickString(isRecord2(request.body) ? request.body.previous_response_id : undefined);
@@ -1724,6 +1738,8 @@ async function completeWithResponsesAPIWithMCP(options, fetcher, path, request)
   const executedToolCalls = [];
   const toolExecutions = [];
   for (let round = 1;round <= maxToolRounds + 1; round += 1) {
+    const mcpToolset = await resolveMCPToolset(request.mcpClients);
+    const transportTools = toResponsesTools(toProviderFunctionTools(mcpToolset));
     const response = await fetcher(buildURL(options.baseURL, path), {
       method: "POST",
       headers: buildHeaders(options),
@@ -2163,8 +2179,6 @@ async function completePassThrough(options, fetcher, path, request) {
   };
 }
 async function completeWithMCPToolLoop(options, fetcher, path, request) {
-  const mcpToolset = await resolveMCPToolset(request.mcpClients);
-  const tools = toAnthropicTools(toProviderFunctionTools(mcpToolset));
   const maxToolRounds = normalizeMaxToolRounds(request.maxToolRounds ?? options.defaultMaxToolRounds);
   let messages = [{ role: "user", content: request.prompt }];
   let aggregatedUsage;
@@ -2173,6 +2187,8 @@ async function completeWithMCPToolLoop(options, fetcher, path, request) {
   const toolCalls = [];
   const toolExecutions = [];
   for (let round = 1;round <= maxToolRounds + 1; round += 1) {
+    const mcpToolset = await resolveMCPToolset(request.mcpClients);
+    const tools = toAnthropicTools(toProviderFunctionTools(mcpToolset));
     const response = await fetcher(buildURL(options.baseURL, path), {
       method: "POST",
       headers: buildHeaders2(options),
@@ -2982,6 +2998,9 @@ var DEFAULT_SELF_HEAL_PROTOCOL = "extrait.self-heal.v2";
 var DEFAULT_SELF_HEAL_MAX_CONTEXT_CHARS = 12000;
 var DEFAULT_SELF_HEAL_STOP_ON_NO_PROGRESS = true;
 var DEFAULT_SELF_HEAL_MAX_ERRORS = 8;
+var RE_SIMPLE_IDENTIFIER2 = /^[A-Za-z_$][A-Za-z0-9_$]*$/;
+var RE_ESCAPE_QUOTE = /"/g;
+var RE_WHITESPACE2 = /\s+/g;
 var DEFAULT_SELF_HEAL_MAX_DIAGNOSTICS = 8;
 var structuredOutdent = createOutdent({
   trimLeadingNewline: true,
@@ -3326,11 +3345,11 @@ function formatIssuePath(path) {
       out += `[${segment}]`;
       continue;
     }
-    if (/^[A-Za-z_$][A-Za-z0-9_$]*$/.test(segment)) {
+    if (RE_SIMPLE_IDENTIFIER2.test(segment)) {
       out += `.${segment}`;
       continue;
     }
-    out += `["${segment.replace(/"/g, "\\\"")}"]`;
+    out += `["${segment.replace(RE_ESCAPE_QUOTE, "\\\"")}"]`;
   }
   return out;
 }
@@ -3373,7 +3392,7 @@ function buildSelfHealFailureFingerprint(attempt) {
   return [issues, errors, source].join("::");
 }
 function normalizeWhitespace(value) {
-  return value.replace(/\s+/g, " ").trim();
+  return value.replace(RE_WHITESPACE2, " ").trim();
 }
 function normalizeStreamConfig(option) {
   if (typeof option === "boolean") {

package/dist/providers/mcp-runtime.d.ts

@@ -36,3 +36,4 @@ export declare function normalizeMaxToolRounds(value: number | undefined): numbe
 export declare function parseToolArguments(value: unknown): unknown;
 export declare function stringifyToolOutput(value: unknown): string;
 export declare function formatToolExecutionDebugLine(execution: LLMToolExecution): string;
+export declare function sanitizeToolName(input: string): string;

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "extrait",
-  "version": "0.1.2",
+  "version": "0.2.0",
   "license": "MIT",
   "repository": {
     "type": "git",