Add huggingface inference as text completion source

src/additional-headers.js

@@ -147,6 +147,19 @@ function getKoboldCppHeaders(directories) {
     }) : {};
 }
 
+/**
+ * Gets the headers for the HuggingFace API.
+ * @param {import('./users').UserDirectoryList} directories
+ * @returns {object} Headers for the request
+ */
+function getHuggingFaceHeaders(directories) {
+    const apiKey = readSecret(directories, SECRET_KEYS.HUGGINGFACE);
+
+    return apiKey ? ({
+        'Authorization': `Bearer ${apiKey}`,
+    }) : {};
+}
+
 function getOverrideHeaders(urlHost) {
     const requestOverrides = getConfigValue('requestOverrides', []);
     const overrideHeaders = requestOverrides?.find((e) => e.hosts?.includes(urlHost))?.headers;
@@ -187,6 +200,7 @@ function setAdditionalHeadersByType(requestHeaders, type, server, directories) {
         [TEXTGEN_TYPES.OPENROUTER]: getOpenRouterHeaders,
         [TEXTGEN_TYPES.KOBOLDCPP]: getKoboldCppHeaders,
         [TEXTGEN_TYPES.LLAMACPP]: getLlamaCppHeaders,
+        [TEXTGEN_TYPES.HUGGINGFACE]: getHuggingFaceHeaders,
     };
 
     const getHeaders = headerGetters[type];

src/constants.js

@@ -216,6 +216,7 @@ const TEXTGEN_TYPES = {
     INFERMATICAI: 'infermaticai',
     DREAMGEN: 'dreamgen',
     OPENROUTER: 'openrouter',
+    HUGGINGFACE: 'huggingface',
 };
 
 const INFERMATICAI_KEYS = [

src/endpoints/backends/text-completions.js

@@ -95,13 +95,14 @@ router.post('/status', jsonParser, async function (request, response) {
 
         setAdditionalHeaders(request, args, baseUrl);
 
+        const apiType = request.body.api_type;
         let url = baseUrl;
         let result = '';
 
         if (request.body.legacy_api) {
             url += '/v1/model';
         } else {
-            switch (request.body.api_type) {
+            switch (apiType) {
                 case TEXTGEN_TYPES.OOBA:
                 case TEXTGEN_TYPES.VLLM:
                 case TEXTGEN_TYPES.APHRODITE:
@@ -126,6 +127,9 @@ router.post('/status', jsonParser, async function (request, response) {
                 case TEXTGEN_TYPES.OLLAMA:
                     url += '/api/tags';
                     break;
+                case TEXTGEN_TYPES.HUGGINGFACE:
+                    url += '/info';
+                    break;
             }
         }
 
@@ -144,14 +148,18 @@ router.post('/status', jsonParser, async function (request, response) {
         }
 
         // Rewrap to OAI-like response
-        if (request.body.api_type === TEXTGEN_TYPES.TOGETHERAI && Array.isArray(data)) {
+        if (apiType === TEXTGEN_TYPES.TOGETHERAI && Array.isArray(data)) {
             data = { data: data.map(x => ({ id: x.name, ...x })) };
         }
 
-        if (request.body.api_type === TEXTGEN_TYPES.OLLAMA && Array.isArray(data.models)) {
+        if (apiType === TEXTGEN_TYPES.OLLAMA && Array.isArray(data.models)) {
             data = { data: data.models.map(x => ({ id: x.name, ...x })) };
         }
 
+        if (apiType === TEXTGEN_TYPES.HUGGINGFACE) {
+            data = { data: [] };
+        }
+
         if (!Array.isArray(data.data)) {
             console.log('Models response is not an array.');
             return response.status(400);
@@ -163,7 +171,7 @@ router.post('/status', jsonParser, async function (request, response) {
         // Set result to the first model ID
         result = modelIds[0] || 'Valid';
 
-        if (request.body.api_type === TEXTGEN_TYPES.OOBA) {
+        if (apiType === TEXTGEN_TYPES.OOBA) {
             try {
                 const modelInfoUrl = baseUrl + '/v1/internal/model/info';
                 const modelInfoReply = await fetch(modelInfoUrl, args);
@@ -178,7 +186,7 @@ router.post('/status', jsonParser, async function (request, response) {
             } catch (error) {
                 console.error(`Failed to get Ooba model info: ${error}`);
             }
-        } else if (request.body.api_type === TEXTGEN_TYPES.TABBY) {
+        } else if (apiType === TEXTGEN_TYPES.TABBY) {
             try {
                 const modelInfoUrl = baseUrl + '/v1/model';
                 const modelInfoReply = await fetch(modelInfoUrl, args);
@@ -241,6 +249,7 @@ router.post('/generate', jsonParser, async function (request, response) {
                 case TEXTGEN_TYPES.KOBOLDCPP:
                 case TEXTGEN_TYPES.TOGETHERAI:
                 case TEXTGEN_TYPES.INFERMATICAI:
+                case TEXTGEN_TYPES.HUGGINGFACE:
                     url += '/v1/completions';
                     break;
                 case TEXTGEN_TYPES.DREAMGEN:

src/endpoints/secrets.js

@@ -41,6 +41,7 @@ const SECRET_KEYS = {
     GROQ: 'api_key_groq',
     AZURE_TTS: 'api_key_azure_tts',
     ZEROONEAI: 'api_key_01ai',
+    HUGGINGFACE: 'api_key_huggingface',
 };
 
 // These are the keys that are safe to expose, even if allowKeysExposure is false