opencode-codebase-index 0.6.0 → 0.6.1

package/dist/index.cjs

@@ -763,6 +763,27 @@ var DEFAULT_PROVIDER_MODELS = {
   "ollama": "nomic-embed-text"
 };
 
+// src/config/env-substitution.ts
+var ENV_REFERENCE_PATTERN = /^\{env:([A-Z_][A-Z0-9_]*)\}$/;
+var ENV_REFERENCE_LIKE_PATTERN = /\{env:[^}]+\}/;
+function substituteEnvString(value, keyPath) {
+  const match = value.match(ENV_REFERENCE_PATTERN);
+  if (!match) {
+    if (ENV_REFERENCE_LIKE_PATTERN.test(value)) {
+      throw new Error(
+        `Invalid environment variable reference at '${keyPath}'. Expected the entire string to match '{env:VAR_NAME}' with VAR_NAME matching [A-Z_][A-Z0-9_]*.`
+      );
+    }
+    return value;
+  }
+  const variableName = match[1];
+  const envValue = process.env[variableName];
+  if (envValue === void 0) {
+    throw new Error(`Missing environment variable '${variableName}' referenced by config at '${keyPath}'.`);
+  }
+  return envValue;
+}
+
 // src/config/schema.ts
 function getDefaultIndexingConfig() {
   return {
@@ -820,11 +841,27 @@ function isValidScope(value) {
 function isStringArray(value) {
   return Array.isArray(value) && value.every((item) => typeof item === "string");
 }
+function getResolvedString(value, keyPath) {
+  if (typeof value !== "string") {
+    return void 0;
+  }
+  return substituteEnvString(value, keyPath);
+}
+function getResolvedStringArray(value, keyPath) {
+  if (!isStringArray(value)) {
+    return void 0;
+  }
+  return value.map((item, index) => substituteEnvString(item, `${keyPath}[${index}]`));
+}
 function isValidLogLevel(value) {
   return typeof value === "string" && VALID_LOG_LEVELS.includes(value);
 }
 function parseConfig(raw) {
   const input = raw && typeof raw === "object" ? raw : {};
+  const embeddingProviderValue = getResolvedString(input.embeddingProvider, "$root.embeddingProvider");
+  const scopeValue = getResolvedString(input.scope, "$root.scope");
+  const includeValue = getResolvedStringArray(input.include, "$root.include");
+  const excludeValue = getResolvedStringArray(input.exclude, "$root.exclude");
   const defaultIndexing = getDefaultIndexingConfig();
   const defaultSearch = getDefaultSearchConfig();
   const defaultDebug = getDefaultDebugConfig();
@@ -867,19 +904,23 @@ function parseConfig(raw) {
   let embeddingProvider;
   let embeddingModel = void 0;
   let customProvider = void 0;
-  if (input.embeddingProvider === "custom") {
+  if (embeddingProviderValue === "custom") {
     embeddingProvider = "custom";
     const rawCustom = input.customProvider && typeof input.customProvider === "object" ? input.customProvider : null;
-    if (rawCustom && typeof rawCustom.baseUrl === "string" && rawCustom.baseUrl.trim().length > 0 && typeof rawCustom.model === "string" && rawCustom.model.trim().length > 0 && typeof rawCustom.dimensions === "number" && Number.isInteger(rawCustom.dimensions) && rawCustom.dimensions > 0) {
+    const baseUrlValue = getResolvedString(rawCustom?.baseUrl, "$root.customProvider.baseUrl");
+    const modelValue = getResolvedString(rawCustom?.model, "$root.customProvider.model");
+    const apiKeyValue = getResolvedString(rawCustom?.apiKey, "$root.customProvider.apiKey");
+    if (rawCustom && typeof baseUrlValue === "string" && baseUrlValue.trim().length > 0 && typeof modelValue === "string" && modelValue.trim().length > 0 && typeof rawCustom.dimensions === "number" && Number.isInteger(rawCustom.dimensions) && rawCustom.dimensions > 0) {
       customProvider = {
-        baseUrl: rawCustom.baseUrl.trim().replace(/\/+$/, ""),
-        model: rawCustom.model,
+        baseUrl: baseUrlValue.trim().replace(/\/+$/, ""),
+        model: modelValue,
         dimensions: rawCustom.dimensions,
-        apiKey: typeof rawCustom.apiKey === "string" ? rawCustom.apiKey : void 0,
+        apiKey: apiKeyValue,
         maxTokens: typeof rawCustom.maxTokens === "number" ? rawCustom.maxTokens : void 0,
         timeoutMs: typeof rawCustom.timeoutMs === "number" ? Math.max(1e3, rawCustom.timeoutMs) : void 0,
         concurrency: typeof rawCustom.concurrency === "number" ? Math.max(1, Math.floor(rawCustom.concurrency)) : void 0,
-        requestIntervalMs: typeof rawCustom.requestIntervalMs === "number" ? Math.max(0, Math.floor(rawCustom.requestIntervalMs)) : void 0
+        requestIntervalMs: typeof rawCustom.requestIntervalMs === "number" ? Math.max(0, Math.floor(rawCustom.requestIntervalMs)) : void 0,
+        maxBatchSize: typeof rawCustom.maxBatchSize === "number" ? Math.max(1, Math.floor(rawCustom.maxBatchSize)) : typeof rawCustom.max_batch_size === "number" ? Math.max(1, Math.floor(rawCustom.max_batch_size)) : void 0
       };
       if (!/\/v\d+\/?$/.test(customProvider.baseUrl)) {
         console.warn(
@@ -891,10 +932,16 @@ function parseConfig(raw) {
         "embeddingProvider is 'custom' but customProvider config is missing or invalid. Required fields: baseUrl (string), model (string), dimensions (positive integer)."
       );
     }
-  } else if (isValidProvider(input.embeddingProvider)) {
-    embeddingProvider = input.embeddingProvider;
-    if (input.embeddingModel) {
-      embeddingModel = isValidModel(input.embeddingModel, embeddingProvider) ? input.embeddingModel : DEFAULT_PROVIDER_MODELS[embeddingProvider];
+  } else if (isValidProvider(embeddingProviderValue)) {
+    embeddingProvider = embeddingProviderValue;
+    const rawEmbeddingModel = input.embeddingModel;
+    if (typeof rawEmbeddingModel === "string") {
+      const embeddingModelValue = substituteEnvString(rawEmbeddingModel, "$root.embeddingModel");
+      if (embeddingModelValue) {
+        embeddingModel = isValidModel(embeddingModelValue, embeddingProvider) ? embeddingModelValue : DEFAULT_PROVIDER_MODELS[embeddingProvider];
+      }
+    } else if (rawEmbeddingModel) {
+      embeddingModel = DEFAULT_PROVIDER_MODELS[embeddingProvider];
     }
   } else {
     embeddingProvider = "auto";
@@ -903,9 +950,9 @@ function parseConfig(raw) {
     embeddingProvider,
     embeddingModel,
     customProvider,
-    scope: isValidScope(input.scope) ? input.scope : "project",
-    include: isStringArray(input.include) ? input.include : DEFAULT_INCLUDE,
-    exclude: isStringArray(input.exclude) ? input.exclude : DEFAULT_EXCLUDE,
+    scope: isValidScope(scopeValue) ? scopeValue : "project",
+    include: includeValue ?? DEFAULT_INCLUDE,
+    exclude: excludeValue ?? DEFAULT_EXCLUDE,
     indexing,
     search,
     debug
@@ -2089,7 +2136,8 @@ function createCustomProviderInfo(config) {
       dimensions: config.dimensions,
       maxTokens: config.maxTokens ?? 8192,
       costPer1MTokens: 0,
-      timeoutMs: config.timeoutMs ?? 3e4
+      timeoutMs: config.timeoutMs ?? 3e4,
+      maxBatchSize: config.maxBatchSize
     }
   };
 }
@@ -2352,21 +2400,24 @@ var CustomEmbeddingProvider = class {
     this.credentials = credentials;
     this.modelInfo = modelInfo;
   }
-  async embedQuery(query) {
-    const result = await this.embedBatch([query]);
-    return {
-      embedding: result.embeddings[0],
-      tokensUsed: result.totalTokensUsed
-    };
-  }
-  async embedDocument(document) {
-    const result = await this.embedBatch([document]);
-    return {
-      embedding: result.embeddings[0],
-      tokensUsed: result.totalTokensUsed
-    };
+  splitIntoRequestBatches(texts) {
+    const maxBatchSize = this.modelInfo.maxBatchSize;
+    if (!maxBatchSize || texts.length <= maxBatchSize) {
+      return [texts];
+    }
+    const batches = [];
+    for (let i = 0; i < texts.length; i += maxBatchSize) {
+      batches.push(texts.slice(i, i + maxBatchSize));
+    }
+    return batches;
   }
-  async embedBatch(texts) {
+  async embedRequest(texts) {
+    if (texts.length === 0) {
+      return {
+        embeddings: [],
+        totalTokensUsed: 0
+      };
+    }
     const headers = {
       "Content-Type": "application/json"
     };
@@ -2427,6 +2478,34 @@ var CustomEmbeddingProvider = class {
     }
     throw new Error("Custom embedding API returned unexpected response format. Expected OpenAI-compatible format with data[].embedding.");
   }
+  async embedQuery(query) {
+    const result = await this.embedBatch([query]);
+    return {
+      embedding: result.embeddings[0],
+      tokensUsed: result.totalTokensUsed
+    };
+  }
+  async embedDocument(document) {
+    const result = await this.embedBatch([document]);
+    return {
+      embedding: result.embeddings[0],
+      tokensUsed: result.totalTokensUsed
+    };
+  }
+  async embedBatch(texts) {
+    const requestBatches = this.splitIntoRequestBatches(texts);
+    const embeddings = [];
+    let totalTokensUsed = 0;
+    for (const batch of requestBatches) {
+      const result = await this.embedRequest(batch);
+      embeddings.push(...result.embeddings);
+      totalTokensUsed += result.totalTokensUsed;
+    }
+    return {
+      embeddings,
+      totalTokensUsed
+    };
+  }
   getModelInfo() {
     return this.modelInfo;
   }