searchsocket 0.4.0 → 0.5.0
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/README.md +49 -31
- package/dist/cli.js +634 -1326
- package/dist/client.cjs +41 -117
- package/dist/client.d.cts +3 -17
- package/dist/client.d.ts +3 -17
- package/dist/client.js +41 -117
- package/dist/index.cjs +608 -1398
- package/dist/index.d.cts +73 -35
- package/dist/index.d.ts +73 -35
- package/dist/index.js +605 -1392
- package/dist/plugin-B_npJSux.d.cts +36 -0
- package/dist/plugin-M-aW0ev6.d.ts +36 -0
- package/dist/scroll.cjs +185 -0
- package/dist/scroll.d.cts +42 -0
- package/dist/scroll.d.ts +42 -0
- package/dist/scroll.js +183 -0
- package/dist/sveltekit.cjs +781 -1278
- package/dist/sveltekit.d.cts +3 -43
- package/dist/sveltekit.d.ts +3 -43
- package/dist/sveltekit.js +779 -1276
- package/dist/{types-z2dw3H6E.d.cts → types-Dk43uz25.d.cts} +46 -141
- package/dist/{types-z2dw3H6E.d.ts → types-Dk43uz25.d.ts} +46 -141
- package/package.json +10 -3
package/dist/index.js
CHANGED
|
@@ -3,12 +3,12 @@ import path from 'path';
|
|
|
3
3
|
import { createJiti } from 'jiti';
|
|
4
4
|
import { z } from 'zod';
|
|
5
5
|
import { execSync, spawn } from 'child_process';
|
|
6
|
-
import pLimit2 from 'p-limit';
|
|
7
6
|
import { createHash } from 'crypto';
|
|
8
7
|
import { load } from 'cheerio';
|
|
9
8
|
import matter from 'gray-matter';
|
|
10
|
-
import fs4 from 'fs/promises';
|
|
11
9
|
import fg from 'fast-glob';
|
|
10
|
+
import pLimit from 'p-limit';
|
|
11
|
+
import fs3 from 'fs/promises';
|
|
12
12
|
import net from 'net';
|
|
13
13
|
import { gunzipSync } from 'zlib';
|
|
14
14
|
import { McpServer } from '@modelcontextprotocol/sdk/server/mcp.js';
|
|
@@ -2759,12 +2759,12 @@ var require_ChildNode = __commonJS({
|
|
|
2759
2759
|
"node_modules/.pnpm/@mixmark-io+domino@2.2.0/node_modules/@mixmark-io/domino/lib/ChildNode.js"(exports$1, module) {
|
|
2760
2760
|
var Node2 = require_Node();
|
|
2761
2761
|
var LinkedList = require_LinkedList();
|
|
2762
|
-
var createDocumentFragmentFromArguments = function(
|
|
2763
|
-
var docFrag =
|
|
2762
|
+
var createDocumentFragmentFromArguments = function(document2, args) {
|
|
2763
|
+
var docFrag = document2.createDocumentFragment();
|
|
2764
2764
|
for (var i = 0; i < args.length; i++) {
|
|
2765
2765
|
var argItem = args[i];
|
|
2766
2766
|
var isNode = argItem instanceof Node2;
|
|
2767
|
-
docFrag.appendChild(isNode ? argItem :
|
|
2767
|
+
docFrag.appendChild(isNode ? argItem : document2.createTextNode(String(argItem)));
|
|
2768
2768
|
}
|
|
2769
2769
|
return docFrag;
|
|
2770
2770
|
};
|
|
@@ -2922,7 +2922,7 @@ var require_NamedNodeMap = __commonJS({
|
|
|
2922
2922
|
// node_modules/.pnpm/@mixmark-io+domino@2.2.0/node_modules/@mixmark-io/domino/lib/Element.js
|
|
2923
2923
|
var require_Element = __commonJS({
|
|
2924
2924
|
"node_modules/.pnpm/@mixmark-io+domino@2.2.0/node_modules/@mixmark-io/domino/lib/Element.js"(exports$1, module) {
|
|
2925
|
-
module.exports =
|
|
2925
|
+
module.exports = Element2;
|
|
2926
2926
|
var xml = require_xmlnames();
|
|
2927
2927
|
var utils = require_utils();
|
|
2928
2928
|
var NAMESPACE = utils.NAMESPACE;
|
|
@@ -2939,7 +2939,7 @@ var require_Element = __commonJS({
|
|
|
2939
2939
|
var NonDocumentTypeChildNode = require_NonDocumentTypeChildNode();
|
|
2940
2940
|
var NamedNodeMap = require_NamedNodeMap();
|
|
2941
2941
|
var uppercaseCache = /* @__PURE__ */ Object.create(null);
|
|
2942
|
-
function
|
|
2942
|
+
function Element2(doc, localName, namespaceURI, prefix) {
|
|
2943
2943
|
ContainerNode.call(this);
|
|
2944
2944
|
this.nodeType = Node2.ELEMENT_NODE;
|
|
2945
2945
|
this.ownerDocument = doc;
|
|
@@ -2959,7 +2959,7 @@ var require_Element = __commonJS({
|
|
|
2959
2959
|
recursiveGetText(node.childNodes[i], a);
|
|
2960
2960
|
}
|
|
2961
2961
|
}
|
|
2962
|
-
|
|
2962
|
+
Element2.prototype = Object.create(ContainerNode.prototype, {
|
|
2963
2963
|
isHTML: { get: function isHTML() {
|
|
2964
2964
|
return this.namespaceURI === NAMESPACE.HTML && this.ownerDocument.isHTML;
|
|
2965
2965
|
} },
|
|
@@ -3029,7 +3029,7 @@ var require_Element = __commonJS({
|
|
|
3029
3029
|
return NodeUtils.serializeOne(this, { nodeType: 0 });
|
|
3030
3030
|
},
|
|
3031
3031
|
set: function(v) {
|
|
3032
|
-
var
|
|
3032
|
+
var document2 = this.ownerDocument;
|
|
3033
3033
|
var parent = this.parentNode;
|
|
3034
3034
|
if (parent === null) {
|
|
3035
3035
|
return;
|
|
@@ -3040,8 +3040,8 @@ var require_Element = __commonJS({
|
|
|
3040
3040
|
if (parent.nodeType === Node2.DOCUMENT_FRAGMENT_NODE) {
|
|
3041
3041
|
parent = parent.ownerDocument.createElement("body");
|
|
3042
3042
|
}
|
|
3043
|
-
var parser =
|
|
3044
|
-
|
|
3043
|
+
var parser = document2.implementation.mozHTMLParser(
|
|
3044
|
+
document2._address,
|
|
3045
3045
|
parent
|
|
3046
3046
|
);
|
|
3047
3047
|
parser.parse(v === null ? "" : String(v), true);
|
|
@@ -3100,7 +3100,7 @@ var require_Element = __commonJS({
|
|
|
3100
3100
|
default:
|
|
3101
3101
|
utils.SyntaxError();
|
|
3102
3102
|
}
|
|
3103
|
-
if (!(context instanceof
|
|
3103
|
+
if (!(context instanceof Element2) || context.ownerDocument.isHTML && context.localName === "html" && context.namespaceURI === NAMESPACE.HTML) {
|
|
3104
3104
|
context = context.ownerDocument.createElementNS(NAMESPACE.HTML, "body");
|
|
3105
3105
|
}
|
|
3106
3106
|
var parser = this.ownerDocument.implementation.mozHTMLParser(
|
|
@@ -3708,10 +3708,10 @@ var require_Element = __commonJS({
|
|
|
3708
3708
|
return nodes.item ? nodes : new NodeList(nodes);
|
|
3709
3709
|
} }
|
|
3710
3710
|
});
|
|
3711
|
-
Object.defineProperties(
|
|
3712
|
-
Object.defineProperties(
|
|
3711
|
+
Object.defineProperties(Element2.prototype, ChildNode);
|
|
3712
|
+
Object.defineProperties(Element2.prototype, NonDocumentTypeChildNode);
|
|
3713
3713
|
attributes.registerChangeHandler(
|
|
3714
|
-
|
|
3714
|
+
Element2,
|
|
3715
3715
|
"id",
|
|
3716
3716
|
function(element, lname, oldval, newval) {
|
|
3717
3717
|
if (element.rooted) {
|
|
@@ -3725,7 +3725,7 @@ var require_Element = __commonJS({
|
|
|
3725
3725
|
}
|
|
3726
3726
|
);
|
|
3727
3727
|
attributes.registerChangeHandler(
|
|
3728
|
-
|
|
3728
|
+
Element2,
|
|
3729
3729
|
"class",
|
|
3730
3730
|
function(element, lname, oldval, newval) {
|
|
3731
3731
|
if (element._classList) {
|
|
@@ -3824,7 +3824,7 @@ var require_Element = __commonJS({
|
|
|
3824
3824
|
}
|
|
3825
3825
|
}
|
|
3826
3826
|
});
|
|
3827
|
-
|
|
3827
|
+
Element2._Attr = Attr;
|
|
3828
3828
|
function AttributesArray(elt) {
|
|
3829
3829
|
NamedNodeMap.call(this, elt);
|
|
3830
3830
|
for (var name in elt._attrsByQName) {
|
|
@@ -4226,7 +4226,7 @@ var require_DocumentFragment = __commonJS({
|
|
|
4226
4226
|
var Node2 = require_Node();
|
|
4227
4227
|
var NodeList = require_NodeList();
|
|
4228
4228
|
var ContainerNode = require_ContainerNode();
|
|
4229
|
-
var
|
|
4229
|
+
var Element2 = require_Element();
|
|
4230
4230
|
var select = require_select();
|
|
4231
4231
|
var utils = require_utils();
|
|
4232
4232
|
function DocumentFragment(doc) {
|
|
@@ -4244,9 +4244,9 @@ var require_DocumentFragment = __commonJS({
|
|
|
4244
4244
|
}
|
|
4245
4245
|
},
|
|
4246
4246
|
// Copy the text content getter/setter from Element
|
|
4247
|
-
textContent: Object.getOwnPropertyDescriptor(
|
|
4247
|
+
textContent: Object.getOwnPropertyDescriptor(Element2.prototype, "textContent"),
|
|
4248
4248
|
// Copy the text content getter/setter from Element
|
|
4249
|
-
innerText: Object.getOwnPropertyDescriptor(
|
|
4249
|
+
innerText: Object.getOwnPropertyDescriptor(Element2.prototype, "innerText"),
|
|
4250
4250
|
querySelector: { value: function(selector) {
|
|
4251
4251
|
var nodes = this.querySelectorAll(selector);
|
|
4252
4252
|
return nodes.length ? nodes[0] : null;
|
|
@@ -4254,8 +4254,8 @@ var require_DocumentFragment = __commonJS({
|
|
|
4254
4254
|
querySelectorAll: { value: function(selector) {
|
|
4255
4255
|
var context = Object.create(this);
|
|
4256
4256
|
context.isHTML = true;
|
|
4257
|
-
context.getElementsByTagName =
|
|
4258
|
-
context.nextElement = Object.getOwnPropertyDescriptor(
|
|
4257
|
+
context.getElementsByTagName = Element2.prototype.getElementsByTagName;
|
|
4258
|
+
context.nextElement = Object.getOwnPropertyDescriptor(Element2.prototype, "firstElementChild").get;
|
|
4259
4259
|
var nodes = select(selector, context);
|
|
4260
4260
|
return nodes.item ? nodes : new NodeList(nodes);
|
|
4261
4261
|
} },
|
|
@@ -4337,7 +4337,7 @@ var require_ProcessingInstruction = __commonJS({
|
|
|
4337
4337
|
// node_modules/.pnpm/@mixmark-io+domino@2.2.0/node_modules/@mixmark-io/domino/lib/NodeFilter.js
|
|
4338
4338
|
var require_NodeFilter = __commonJS({
|
|
4339
4339
|
"node_modules/.pnpm/@mixmark-io+domino@2.2.0/node_modules/@mixmark-io/domino/lib/NodeFilter.js"(exports$1, module) {
|
|
4340
|
-
var
|
|
4340
|
+
var NodeFilter2 = {
|
|
4341
4341
|
// Constants for acceptNode()
|
|
4342
4342
|
FILTER_ACCEPT: 1,
|
|
4343
4343
|
FILTER_REJECT: 2,
|
|
@@ -4362,7 +4362,7 @@ var require_NodeFilter = __commonJS({
|
|
|
4362
4362
|
SHOW_NOTATION: 2048
|
|
4363
4363
|
// historical
|
|
4364
4364
|
};
|
|
4365
|
-
module.exports =
|
|
4365
|
+
module.exports = NodeFilter2.constructor = NodeFilter2.prototype = NodeFilter2;
|
|
4366
4366
|
}
|
|
4367
4367
|
});
|
|
4368
4368
|
|
|
@@ -4437,7 +4437,7 @@ var require_TreeWalker = __commonJS({
|
|
|
4437
4437
|
"node_modules/.pnpm/@mixmark-io+domino@2.2.0/node_modules/@mixmark-io/domino/lib/TreeWalker.js"(exports$1, module) {
|
|
4438
4438
|
module.exports = TreeWalker;
|
|
4439
4439
|
var Node2 = require_Node();
|
|
4440
|
-
var
|
|
4440
|
+
var NodeFilter2 = require_NodeFilter();
|
|
4441
4441
|
var NodeTraversal = require_NodeTraversal();
|
|
4442
4442
|
var utils = require_utils();
|
|
4443
4443
|
var mapChild = {
|
|
@@ -4457,11 +4457,11 @@ var require_TreeWalker = __commonJS({
|
|
|
4457
4457
|
node = tw._currentNode[mapChild[type]];
|
|
4458
4458
|
while (node !== null) {
|
|
4459
4459
|
result = tw._internalFilter(node);
|
|
4460
|
-
if (result ===
|
|
4460
|
+
if (result === NodeFilter2.FILTER_ACCEPT) {
|
|
4461
4461
|
tw._currentNode = node;
|
|
4462
4462
|
return node;
|
|
4463
4463
|
}
|
|
4464
|
-
if (result ===
|
|
4464
|
+
if (result === NodeFilter2.FILTER_SKIP) {
|
|
4465
4465
|
child = node[mapChild[type]];
|
|
4466
4466
|
if (child !== null) {
|
|
4467
4467
|
node = child;
|
|
@@ -4495,12 +4495,12 @@ var require_TreeWalker = __commonJS({
|
|
|
4495
4495
|
while (sibling !== null) {
|
|
4496
4496
|
node = sibling;
|
|
4497
4497
|
result = tw._internalFilter(node);
|
|
4498
|
-
if (result ===
|
|
4498
|
+
if (result === NodeFilter2.FILTER_ACCEPT) {
|
|
4499
4499
|
tw._currentNode = node;
|
|
4500
4500
|
return node;
|
|
4501
4501
|
}
|
|
4502
4502
|
sibling = node[mapChild[type]];
|
|
4503
|
-
if (result ===
|
|
4503
|
+
if (result === NodeFilter2.FILTER_REJECT || sibling === null) {
|
|
4504
4504
|
sibling = node[mapSibling[type]];
|
|
4505
4505
|
}
|
|
4506
4506
|
}
|
|
@@ -4508,7 +4508,7 @@ var require_TreeWalker = __commonJS({
|
|
|
4508
4508
|
if (node === null || node === tw.root) {
|
|
4509
4509
|
return null;
|
|
4510
4510
|
}
|
|
4511
|
-
if (tw._internalFilter(node) ===
|
|
4511
|
+
if (tw._internalFilter(node) === NodeFilter2.FILTER_ACCEPT) {
|
|
4512
4512
|
return null;
|
|
4513
4513
|
}
|
|
4514
4514
|
}
|
|
@@ -4556,11 +4556,11 @@ var require_TreeWalker = __commonJS({
|
|
|
4556
4556
|
utils.InvalidStateError();
|
|
4557
4557
|
}
|
|
4558
4558
|
if (!(1 << node.nodeType - 1 & this._whatToShow)) {
|
|
4559
|
-
return
|
|
4559
|
+
return NodeFilter2.FILTER_SKIP;
|
|
4560
4560
|
}
|
|
4561
4561
|
filter = this._filter;
|
|
4562
4562
|
if (filter === null) {
|
|
4563
|
-
result =
|
|
4563
|
+
result = NodeFilter2.FILTER_ACCEPT;
|
|
4564
4564
|
} else {
|
|
4565
4565
|
this._active = true;
|
|
4566
4566
|
try {
|
|
@@ -4589,7 +4589,7 @@ var require_TreeWalker = __commonJS({
|
|
|
4589
4589
|
if (node === null) {
|
|
4590
4590
|
return null;
|
|
4591
4591
|
}
|
|
4592
|
-
if (this._internalFilter(node) ===
|
|
4592
|
+
if (this._internalFilter(node) === NodeFilter2.FILTER_ACCEPT) {
|
|
4593
4593
|
this._currentNode = node;
|
|
4594
4594
|
return node;
|
|
4595
4595
|
}
|
|
@@ -4642,17 +4642,17 @@ var require_TreeWalker = __commonJS({
|
|
|
4642
4642
|
for (previousSibling = node.previousSibling; previousSibling; previousSibling = node.previousSibling) {
|
|
4643
4643
|
node = previousSibling;
|
|
4644
4644
|
result = this._internalFilter(node);
|
|
4645
|
-
if (result ===
|
|
4645
|
+
if (result === NodeFilter2.FILTER_REJECT) {
|
|
4646
4646
|
continue;
|
|
4647
4647
|
}
|
|
4648
4648
|
for (lastChild = node.lastChild; lastChild; lastChild = node.lastChild) {
|
|
4649
4649
|
node = lastChild;
|
|
4650
4650
|
result = this._internalFilter(node);
|
|
4651
|
-
if (result ===
|
|
4651
|
+
if (result === NodeFilter2.FILTER_REJECT) {
|
|
4652
4652
|
break;
|
|
4653
4653
|
}
|
|
4654
4654
|
}
|
|
4655
|
-
if (result ===
|
|
4655
|
+
if (result === NodeFilter2.FILTER_ACCEPT) {
|
|
4656
4656
|
this._currentNode = node;
|
|
4657
4657
|
return node;
|
|
4658
4658
|
}
|
|
@@ -4661,7 +4661,7 @@ var require_TreeWalker = __commonJS({
|
|
|
4661
4661
|
return null;
|
|
4662
4662
|
}
|
|
4663
4663
|
node = node.parentNode;
|
|
4664
|
-
if (this._internalFilter(node) ===
|
|
4664
|
+
if (this._internalFilter(node) === NodeFilter2.FILTER_ACCEPT) {
|
|
4665
4665
|
this._currentNode = node;
|
|
4666
4666
|
return node;
|
|
4667
4667
|
}
|
|
@@ -4678,26 +4678,26 @@ var require_TreeWalker = __commonJS({
|
|
|
4678
4678
|
nextNode: { value: function nextNode() {
|
|
4679
4679
|
var node, result, firstChild, nextSibling;
|
|
4680
4680
|
node = this._currentNode;
|
|
4681
|
-
result =
|
|
4681
|
+
result = NodeFilter2.FILTER_ACCEPT;
|
|
4682
4682
|
CHILDREN:
|
|
4683
4683
|
while (true) {
|
|
4684
4684
|
for (firstChild = node.firstChild; firstChild; firstChild = node.firstChild) {
|
|
4685
4685
|
node = firstChild;
|
|
4686
4686
|
result = this._internalFilter(node);
|
|
4687
|
-
if (result ===
|
|
4687
|
+
if (result === NodeFilter2.FILTER_ACCEPT) {
|
|
4688
4688
|
this._currentNode = node;
|
|
4689
4689
|
return node;
|
|
4690
|
-
} else if (result ===
|
|
4690
|
+
} else if (result === NodeFilter2.FILTER_REJECT) {
|
|
4691
4691
|
break;
|
|
4692
4692
|
}
|
|
4693
4693
|
}
|
|
4694
4694
|
for (nextSibling = NodeTraversal.nextSkippingChildren(node, this.root); nextSibling; nextSibling = NodeTraversal.nextSkippingChildren(node, this.root)) {
|
|
4695
4695
|
node = nextSibling;
|
|
4696
4696
|
result = this._internalFilter(node);
|
|
4697
|
-
if (result ===
|
|
4697
|
+
if (result === NodeFilter2.FILTER_ACCEPT) {
|
|
4698
4698
|
this._currentNode = node;
|
|
4699
4699
|
return node;
|
|
4700
|
-
} else if (result ===
|
|
4700
|
+
} else if (result === NodeFilter2.FILTER_SKIP) {
|
|
4701
4701
|
continue CHILDREN;
|
|
4702
4702
|
}
|
|
4703
4703
|
}
|
|
@@ -4716,7 +4716,7 @@ var require_TreeWalker = __commonJS({
|
|
|
4716
4716
|
var require_NodeIterator = __commonJS({
|
|
4717
4717
|
"node_modules/.pnpm/@mixmark-io+domino@2.2.0/node_modules/@mixmark-io/domino/lib/NodeIterator.js"(exports$1, module) {
|
|
4718
4718
|
module.exports = NodeIterator;
|
|
4719
|
-
var
|
|
4719
|
+
var NodeFilter2 = require_NodeFilter();
|
|
4720
4720
|
var NodeTraversal = require_NodeTraversal();
|
|
4721
4721
|
var utils = require_utils();
|
|
4722
4722
|
function move(node, stayWithin, directionIsNext) {
|
|
@@ -4751,7 +4751,7 @@ var require_NodeIterator = __commonJS({
|
|
|
4751
4751
|
}
|
|
4752
4752
|
}
|
|
4753
4753
|
var result = ni._internalFilter(node);
|
|
4754
|
-
if (result ===
|
|
4754
|
+
if (result === NodeFilter2.FILTER_ACCEPT) {
|
|
4755
4755
|
break;
|
|
4756
4756
|
}
|
|
4757
4757
|
}
|
|
@@ -4799,11 +4799,11 @@ var require_NodeIterator = __commonJS({
|
|
|
4799
4799
|
utils.InvalidStateError();
|
|
4800
4800
|
}
|
|
4801
4801
|
if (!(1 << node.nodeType - 1 & this._whatToShow)) {
|
|
4802
|
-
return
|
|
4802
|
+
return NodeFilter2.FILTER_SKIP;
|
|
4803
4803
|
}
|
|
4804
4804
|
filter = this._filter;
|
|
4805
4805
|
if (filter === null) {
|
|
4806
|
-
result =
|
|
4806
|
+
result = NodeFilter2.FILTER_ACCEPT;
|
|
4807
4807
|
} else {
|
|
4808
4808
|
this._active = true;
|
|
4809
4809
|
try {
|
|
@@ -5013,32 +5013,32 @@ var require_URL = __commonJS({
|
|
|
5013
5013
|
else
|
|
5014
5014
|
return basepath.substring(0, lastslash + 1) + refpath;
|
|
5015
5015
|
}
|
|
5016
|
-
function remove_dot_segments(
|
|
5017
|
-
if (!
|
|
5016
|
+
function remove_dot_segments(path13) {
|
|
5017
|
+
if (!path13) return path13;
|
|
5018
5018
|
var output = "";
|
|
5019
|
-
while (
|
|
5020
|
-
if (
|
|
5021
|
-
|
|
5019
|
+
while (path13.length > 0) {
|
|
5020
|
+
if (path13 === "." || path13 === "..") {
|
|
5021
|
+
path13 = "";
|
|
5022
5022
|
break;
|
|
5023
5023
|
}
|
|
5024
|
-
var twochars =
|
|
5025
|
-
var threechars =
|
|
5026
|
-
var fourchars =
|
|
5024
|
+
var twochars = path13.substring(0, 2);
|
|
5025
|
+
var threechars = path13.substring(0, 3);
|
|
5026
|
+
var fourchars = path13.substring(0, 4);
|
|
5027
5027
|
if (threechars === "../") {
|
|
5028
|
-
|
|
5028
|
+
path13 = path13.substring(3);
|
|
5029
5029
|
} else if (twochars === "./") {
|
|
5030
|
-
|
|
5030
|
+
path13 = path13.substring(2);
|
|
5031
5031
|
} else if (threechars === "/./") {
|
|
5032
|
-
|
|
5033
|
-
} else if (twochars === "/." &&
|
|
5034
|
-
|
|
5035
|
-
} else if (fourchars === "/../" || threechars === "/.." &&
|
|
5036
|
-
|
|
5032
|
+
path13 = "/" + path13.substring(3);
|
|
5033
|
+
} else if (twochars === "/." && path13.length === 2) {
|
|
5034
|
+
path13 = "/";
|
|
5035
|
+
} else if (fourchars === "/../" || threechars === "/.." && path13.length === 3) {
|
|
5036
|
+
path13 = "/" + path13.substring(4);
|
|
5037
5037
|
output = output.replace(/\/?[^\/]*$/, "");
|
|
5038
5038
|
} else {
|
|
5039
|
-
var segment =
|
|
5039
|
+
var segment = path13.match(/(\/?([^\/]*))/)[0];
|
|
5040
5040
|
output += segment;
|
|
5041
|
-
|
|
5041
|
+
path13 = path13.substring(segment.length);
|
|
5042
5042
|
}
|
|
5043
5043
|
}
|
|
5044
5044
|
return output;
|
|
@@ -5603,9 +5603,9 @@ var require_defineElement = __commonJS({
|
|
|
5603
5603
|
});
|
|
5604
5604
|
return c;
|
|
5605
5605
|
};
|
|
5606
|
-
function EventHandlerBuilder(body,
|
|
5606
|
+
function EventHandlerBuilder(body, document2, form, element) {
|
|
5607
5607
|
this.body = body;
|
|
5608
|
-
this.document =
|
|
5608
|
+
this.document = document2;
|
|
5609
5609
|
this.form = form;
|
|
5610
5610
|
this.element = element;
|
|
5611
5611
|
}
|
|
@@ -5639,7 +5639,7 @@ var require_defineElement = __commonJS({
|
|
|
5639
5639
|
var require_htmlelts = __commonJS({
|
|
5640
5640
|
"node_modules/.pnpm/@mixmark-io+domino@2.2.0/node_modules/@mixmark-io/domino/lib/htmlelts.js"(exports$1) {
|
|
5641
5641
|
var Node2 = require_Node();
|
|
5642
|
-
var
|
|
5642
|
+
var Element2 = require_Element();
|
|
5643
5643
|
var CSSStyleDeclaration = require_CSSStyleDeclaration();
|
|
5644
5644
|
var utils = require_utils();
|
|
5645
5645
|
var URLUtils = require_URLUtils();
|
|
@@ -5707,10 +5707,10 @@ var require_htmlelts = __commonJS({
|
|
|
5707
5707
|
this._form = null;
|
|
5708
5708
|
};
|
|
5709
5709
|
var HTMLElement = exports$1.HTMLElement = define({
|
|
5710
|
-
superclass:
|
|
5710
|
+
superclass: Element2,
|
|
5711
5711
|
name: "HTMLElement",
|
|
5712
5712
|
ctor: function HTMLElement2(doc, localName, prefix) {
|
|
5713
|
-
|
|
5713
|
+
Element2.call(this, doc, localName, utils.NAMESPACE.HTML, prefix);
|
|
5714
5714
|
},
|
|
5715
5715
|
props: {
|
|
5716
5716
|
dangerouslySetInnerHTML: {
|
|
@@ -7192,7 +7192,7 @@ var require_htmlelts = __commonJS({
|
|
|
7192
7192
|
// node_modules/.pnpm/@mixmark-io+domino@2.2.0/node_modules/@mixmark-io/domino/lib/svg.js
|
|
7193
7193
|
var require_svg = __commonJS({
|
|
7194
7194
|
"node_modules/.pnpm/@mixmark-io+domino@2.2.0/node_modules/@mixmark-io/domino/lib/svg.js"(exports$1) {
|
|
7195
|
-
var
|
|
7195
|
+
var Element2 = require_Element();
|
|
7196
7196
|
var defineElement = require_defineElement();
|
|
7197
7197
|
var utils = require_utils();
|
|
7198
7198
|
var CSSStyleDeclaration = require_CSSStyleDeclaration();
|
|
@@ -7206,10 +7206,10 @@ var require_svg = __commonJS({
|
|
|
7206
7206
|
return defineElement(spec, SVGElement, svgElements, svgNameToImpl);
|
|
7207
7207
|
}
|
|
7208
7208
|
var SVGElement = define({
|
|
7209
|
-
superclass:
|
|
7209
|
+
superclass: Element2,
|
|
7210
7210
|
name: "SVGElement",
|
|
7211
7211
|
ctor: function SVGElement2(doc, localName, prefix) {
|
|
7212
|
-
|
|
7212
|
+
Element2.call(this, doc, localName, utils.NAMESPACE.SVG, prefix);
|
|
7213
7213
|
},
|
|
7214
7214
|
props: {
|
|
7215
7215
|
style: { get: function() {
|
|
@@ -7344,7 +7344,7 @@ var require_Document = __commonJS({
|
|
|
7344
7344
|
var Node2 = require_Node();
|
|
7345
7345
|
var NodeList = require_NodeList();
|
|
7346
7346
|
var ContainerNode = require_ContainerNode();
|
|
7347
|
-
var
|
|
7347
|
+
var Element2 = require_Element();
|
|
7348
7348
|
var Text = require_Text();
|
|
7349
7349
|
var Comment = require_Comment();
|
|
7350
7350
|
var Event = require_Event();
|
|
@@ -7353,7 +7353,7 @@ var require_Document = __commonJS({
|
|
|
7353
7353
|
var DOMImplementation = require_DOMImplementation();
|
|
7354
7354
|
var TreeWalker = require_TreeWalker();
|
|
7355
7355
|
var NodeIterator = require_NodeIterator();
|
|
7356
|
-
var
|
|
7356
|
+
var NodeFilter2 = require_NodeFilter();
|
|
7357
7357
|
var URL2 = require_URL();
|
|
7358
7358
|
var select = require_select();
|
|
7359
7359
|
var events = require_events();
|
|
@@ -7492,13 +7492,13 @@ var require_Document = __commonJS({
|
|
|
7492
7492
|
if (this.isHTML) {
|
|
7493
7493
|
localName = utils.toASCIILowerCase(localName);
|
|
7494
7494
|
}
|
|
7495
|
-
return new
|
|
7495
|
+
return new Element2._Attr(null, localName, null, null, "");
|
|
7496
7496
|
} },
|
|
7497
7497
|
createAttributeNS: { value: function(namespace, qualifiedName) {
|
|
7498
7498
|
namespace = namespace === null || namespace === void 0 || namespace === "" ? null : String(namespace);
|
|
7499
7499
|
qualifiedName = String(qualifiedName);
|
|
7500
7500
|
var ve = validateAndExtract(namespace, qualifiedName);
|
|
7501
|
-
return new
|
|
7501
|
+
return new Element2._Attr(null, ve.localName, ve.prefix, ve.namespace, "");
|
|
7502
7502
|
} },
|
|
7503
7503
|
createElement: { value: function(localName) {
|
|
7504
7504
|
localName = String(localName);
|
|
@@ -7510,7 +7510,7 @@ var require_Document = __commonJS({
|
|
|
7510
7510
|
} else if (this.contentType === "application/xhtml+xml") {
|
|
7511
7511
|
return html.createElement(this, localName, null);
|
|
7512
7512
|
} else {
|
|
7513
|
-
return new
|
|
7513
|
+
return new Element2(this, localName, null, null);
|
|
7514
7514
|
}
|
|
7515
7515
|
}, writable: isApiWritable },
|
|
7516
7516
|
createElementNS: { value: function(namespace, qualifiedName) {
|
|
@@ -7527,7 +7527,7 @@ var require_Document = __commonJS({
|
|
|
7527
7527
|
} else if (namespace === NAMESPACE.SVG) {
|
|
7528
7528
|
return svg.createElement(this, localName, prefix);
|
|
7529
7529
|
}
|
|
7530
|
-
return new
|
|
7530
|
+
return new Element2(this, localName, namespace, prefix);
|
|
7531
7531
|
} },
|
|
7532
7532
|
createEvent: { value: function createEvent(interfaceName) {
|
|
7533
7533
|
interfaceName = interfaceName.toLowerCase();
|
|
@@ -7549,7 +7549,7 @@ var require_Document = __commonJS({
|
|
|
7549
7549
|
if (!(root3 instanceof Node2)) {
|
|
7550
7550
|
throw new TypeError("root not a node");
|
|
7551
7551
|
}
|
|
7552
|
-
whatToShow = whatToShow === void 0 ?
|
|
7552
|
+
whatToShow = whatToShow === void 0 ? NodeFilter2.SHOW_ALL : +whatToShow;
|
|
7553
7553
|
filter = filter === void 0 ? null : filter;
|
|
7554
7554
|
return new TreeWalker(root3, whatToShow, filter);
|
|
7555
7555
|
} },
|
|
@@ -7561,7 +7561,7 @@ var require_Document = __commonJS({
|
|
|
7561
7561
|
if (!(root3 instanceof Node2)) {
|
|
7562
7562
|
throw new TypeError("root not a node");
|
|
7563
7563
|
}
|
|
7564
|
-
whatToShow = whatToShow === void 0 ?
|
|
7564
|
+
whatToShow = whatToShow === void 0 ? NodeFilter2.SHOW_ALL : +whatToShow;
|
|
7565
7565
|
filter = filter === void 0 ? null : filter;
|
|
7566
7566
|
return new NodeIterator(root3, whatToShow, filter);
|
|
7567
7567
|
} },
|
|
@@ -7622,10 +7622,10 @@ var require_Document = __commonJS({
|
|
|
7622
7622
|
return this.byId[id] instanceof MultiId;
|
|
7623
7623
|
} },
|
|
7624
7624
|
// Just copy this method from the Element prototype
|
|
7625
|
-
getElementsByName: { value:
|
|
7626
|
-
getElementsByTagName: { value:
|
|
7627
|
-
getElementsByTagNameNS: { value:
|
|
7628
|
-
getElementsByClassName: { value:
|
|
7625
|
+
getElementsByName: { value: Element2.prototype.getElementsByName },
|
|
7626
|
+
getElementsByTagName: { value: Element2.prototype.getElementsByTagName },
|
|
7627
|
+
getElementsByTagNameNS: { value: Element2.prototype.getElementsByTagNameNS },
|
|
7628
|
+
getElementsByClassName: { value: Element2.prototype.getElementsByClassName },
|
|
7629
7629
|
adoptNode: { value: function adoptNode(node) {
|
|
7630
7630
|
if (node.nodeType === Node2.DOCUMENT_NODE) utils.NotSupportedError();
|
|
7631
7631
|
if (node.nodeType === Node2.ATTRIBUTE_NODE) {
|
|
@@ -16451,8 +16451,8 @@ var require_Window = __commonJS({
|
|
|
16451
16451
|
var Location = require_Location();
|
|
16452
16452
|
var utils = require_utils();
|
|
16453
16453
|
module.exports = Window;
|
|
16454
|
-
function Window(
|
|
16455
|
-
this.document =
|
|
16454
|
+
function Window(document2) {
|
|
16455
|
+
this.document = document2 || new DOMImplementation(null).createHTMLDocument("");
|
|
16456
16456
|
this.document._scripting_enabled = true;
|
|
16457
16457
|
this.document.defaultView = this;
|
|
16458
16458
|
this.location = new Location(this, this.document._address || "about:blank");
|
|
@@ -16582,11 +16582,11 @@ var require_lib = __commonJS({
|
|
|
16582
16582
|
};
|
|
16583
16583
|
};
|
|
16584
16584
|
exports$1.createWindow = function(html, address) {
|
|
16585
|
-
var
|
|
16585
|
+
var document2 = exports$1.createDocument(html);
|
|
16586
16586
|
if (address !== void 0) {
|
|
16587
|
-
|
|
16587
|
+
document2._address = address;
|
|
16588
16588
|
}
|
|
16589
|
-
return new impl.Window(
|
|
16589
|
+
return new impl.Window(document2);
|
|
16590
16590
|
};
|
|
16591
16591
|
exports$1.impl = impl;
|
|
16592
16592
|
}
|
|
@@ -16651,29 +16651,18 @@ var searchSocketConfigSchema = z.object({
|
|
|
16651
16651
|
prependTitle: z.boolean().optional(),
|
|
16652
16652
|
pageSummaryChunk: z.boolean().optional()
|
|
16653
16653
|
}).optional(),
|
|
16654
|
-
|
|
16655
|
-
|
|
16656
|
-
|
|
16657
|
-
|
|
16658
|
-
|
|
16659
|
-
batchSize: z.number().int().positive().optional(),
|
|
16660
|
-
concurrency: z.number().int().positive().optional(),
|
|
16661
|
-
pricePer1kTokens: z.number().positive().optional()
|
|
16654
|
+
upstash: z.object({
|
|
16655
|
+
url: z.string().url().optional(),
|
|
16656
|
+
token: z.string().min(1).optional(),
|
|
16657
|
+
urlEnv: z.string().min(1).optional(),
|
|
16658
|
+
tokenEnv: z.string().min(1).optional()
|
|
16662
16659
|
}).optional(),
|
|
16663
|
-
|
|
16664
|
-
|
|
16665
|
-
|
|
16666
|
-
|
|
16667
|
-
|
|
16668
|
-
|
|
16669
|
-
authTokenEnv: z.string().optional(),
|
|
16670
|
-
localPath: z.string().optional()
|
|
16671
|
-
}).optional()
|
|
16672
|
-
}).optional(),
|
|
16673
|
-
rerank: z.object({
|
|
16674
|
-
enabled: z.boolean().optional(),
|
|
16675
|
-
topN: z.number().int().positive().optional(),
|
|
16676
|
-
model: z.string().optional()
|
|
16660
|
+
search: z.object({
|
|
16661
|
+
semanticWeight: z.number().min(0).max(1).optional(),
|
|
16662
|
+
inputEnrichment: z.boolean().optional(),
|
|
16663
|
+
reranking: z.boolean().optional(),
|
|
16664
|
+
dualSearch: z.boolean().optional(),
|
|
16665
|
+
pageSearchWeight: z.number().min(0).max(1).optional()
|
|
16677
16666
|
}).optional(),
|
|
16678
16667
|
ranking: z.object({
|
|
16679
16668
|
enableIncomingLinkBoost: z.boolean().optional(),
|
|
@@ -16683,11 +16672,12 @@ var searchSocketConfigSchema = z.object({
|
|
|
16683
16672
|
aggregationDecay: z.number().min(0).max(1).optional(),
|
|
16684
16673
|
minChunkScoreRatio: z.number().min(0).max(1).optional(),
|
|
16685
16674
|
minScore: z.number().min(0).max(1).optional(),
|
|
16675
|
+
scoreGapThreshold: z.number().min(0).max(1).optional(),
|
|
16686
16676
|
weights: z.object({
|
|
16687
16677
|
incomingLinks: z.number().optional(),
|
|
16688
16678
|
depth: z.number().optional(),
|
|
16689
|
-
|
|
16690
|
-
|
|
16679
|
+
aggregation: z.number().optional(),
|
|
16680
|
+
titleMatch: z.number().optional()
|
|
16691
16681
|
}).optional()
|
|
16692
16682
|
}).optional(),
|
|
16693
16683
|
api: z.object({
|
|
@@ -16709,8 +16699,7 @@ var searchSocketConfigSchema = z.object({
|
|
|
16709
16699
|
}).optional()
|
|
16710
16700
|
}).optional(),
|
|
16711
16701
|
state: z.object({
|
|
16712
|
-
dir: z.string().optional()
|
|
16713
|
-
writeMirror: z.boolean().optional()
|
|
16702
|
+
dir: z.string().optional()
|
|
16714
16703
|
}).optional()
|
|
16715
16704
|
});
|
|
16716
16705
|
|
|
@@ -16764,24 +16753,16 @@ function createDefaultConfig(projectId) {
|
|
|
16764
16753
|
prependTitle: true,
|
|
16765
16754
|
pageSummaryChunk: true
|
|
16766
16755
|
},
|
|
16767
|
-
|
|
16768
|
-
|
|
16769
|
-
|
|
16770
|
-
apiKeyEnv: "JINA_API_KEY",
|
|
16771
|
-
batchSize: 64,
|
|
16772
|
-
concurrency: 4
|
|
16773
|
-
},
|
|
16774
|
-
vector: {
|
|
16775
|
-
turso: {
|
|
16776
|
-
urlEnv: "TURSO_DATABASE_URL",
|
|
16777
|
-
authTokenEnv: "TURSO_AUTH_TOKEN",
|
|
16778
|
-
localPath: ".searchsocket/vectors.db"
|
|
16779
|
-
}
|
|
16756
|
+
upstash: {
|
|
16757
|
+
urlEnv: "UPSTASH_SEARCH_REST_URL",
|
|
16758
|
+
tokenEnv: "UPSTASH_SEARCH_REST_TOKEN"
|
|
16780
16759
|
},
|
|
16781
|
-
|
|
16782
|
-
|
|
16783
|
-
|
|
16784
|
-
|
|
16760
|
+
search: {
|
|
16761
|
+
semanticWeight: 0.75,
|
|
16762
|
+
inputEnrichment: true,
|
|
16763
|
+
reranking: true,
|
|
16764
|
+
dualSearch: true,
|
|
16765
|
+
pageSearchWeight: 0.3
|
|
16785
16766
|
},
|
|
16786
16767
|
ranking: {
|
|
16787
16768
|
enableIncomingLinkBoost: true,
|
|
@@ -16790,12 +16771,13 @@ function createDefaultConfig(projectId) {
|
|
|
16790
16771
|
aggregationCap: 5,
|
|
16791
16772
|
aggregationDecay: 0.5,
|
|
16792
16773
|
minChunkScoreRatio: 0.5,
|
|
16793
|
-
minScore: 0,
|
|
16774
|
+
minScore: 0.3,
|
|
16775
|
+
scoreGapThreshold: 0.4,
|
|
16794
16776
|
weights: {
|
|
16795
16777
|
incomingLinks: 0.05,
|
|
16796
16778
|
depth: 0.03,
|
|
16797
|
-
|
|
16798
|
-
|
|
16779
|
+
aggregation: 0.1,
|
|
16780
|
+
titleMatch: 0.15
|
|
16799
16781
|
}
|
|
16800
16782
|
},
|
|
16801
16783
|
api: {
|
|
@@ -16813,8 +16795,7 @@ function createDefaultConfig(projectId) {
|
|
|
16813
16795
|
}
|
|
16814
16796
|
},
|
|
16815
16797
|
state: {
|
|
16816
|
-
dir: ".searchsocket"
|
|
16817
|
-
writeMirror: false
|
|
16798
|
+
dir: ".searchsocket"
|
|
16818
16799
|
}
|
|
16819
16800
|
};
|
|
16820
16801
|
}
|
|
@@ -16938,21 +16919,13 @@ ${issues}`
|
|
|
16938
16919
|
...defaults.chunking,
|
|
16939
16920
|
...parsed.chunking
|
|
16940
16921
|
},
|
|
16941
|
-
|
|
16942
|
-
...defaults.
|
|
16943
|
-
...parsed.
|
|
16922
|
+
upstash: {
|
|
16923
|
+
...defaults.upstash,
|
|
16924
|
+
...parsed.upstash
|
|
16944
16925
|
},
|
|
16945
|
-
|
|
16946
|
-
...defaults.
|
|
16947
|
-
...parsed.
|
|
16948
|
-
turso: {
|
|
16949
|
-
...defaults.vector.turso,
|
|
16950
|
-
...parsed.vector?.turso
|
|
16951
|
-
}
|
|
16952
|
-
},
|
|
16953
|
-
rerank: {
|
|
16954
|
-
...defaults.rerank,
|
|
16955
|
-
...parsed.rerank
|
|
16926
|
+
search: {
|
|
16927
|
+
...defaults.search,
|
|
16928
|
+
...parsed.search
|
|
16956
16929
|
},
|
|
16957
16930
|
ranking: {
|
|
16958
16931
|
...defaults.ranking,
|
|
@@ -17131,660 +17104,245 @@ function resolveScope(config, override) {
|
|
|
17131
17104
|
scopeId: `${config.project.id}:${scopeName}`
|
|
17132
17105
|
};
|
|
17133
17106
|
}
|
|
17134
|
-
function sleep(ms) {
|
|
17135
|
-
return new Promise((resolve) => {
|
|
17136
|
-
setTimeout(resolve, ms);
|
|
17137
|
-
});
|
|
17138
|
-
}
|
|
17139
|
-
var JinaEmbeddingsProvider = class {
|
|
17140
|
-
apiKey;
|
|
17141
|
-
batchSize;
|
|
17142
|
-
concurrency;
|
|
17143
|
-
defaultTask;
|
|
17144
|
-
constructor(options) {
|
|
17145
|
-
if (!Number.isInteger(options.batchSize) || options.batchSize <= 0) {
|
|
17146
|
-
throw new Error(`Invalid batchSize: ${options.batchSize}. batchSize must be a positive integer.`);
|
|
17147
|
-
}
|
|
17148
|
-
if (!Number.isInteger(options.concurrency) || options.concurrency <= 0) {
|
|
17149
|
-
throw new Error(`Invalid concurrency: ${options.concurrency}. concurrency must be a positive integer.`);
|
|
17150
|
-
}
|
|
17151
|
-
this.apiKey = options.apiKey;
|
|
17152
|
-
this.batchSize = options.batchSize;
|
|
17153
|
-
this.concurrency = options.concurrency;
|
|
17154
|
-
this.defaultTask = options.task ?? "retrieval.passage";
|
|
17155
|
-
}
|
|
17156
|
-
estimateTokens(text) {
|
|
17157
|
-
const normalized = text.trim();
|
|
17158
|
-
if (!normalized) {
|
|
17159
|
-
return 0;
|
|
17160
|
-
}
|
|
17161
|
-
const wordCount = normalized.match(/[A-Za-z0-9_]+/g)?.length ?? 0;
|
|
17162
|
-
const punctuationCount = normalized.match(/[^\s\w]/g)?.length ?? 0;
|
|
17163
|
-
const cjkCount = normalized.match(/[\u3400-\u9fff]/g)?.length ?? 0;
|
|
17164
|
-
const charEstimate = Math.ceil(normalized.length / 4);
|
|
17165
|
-
const lexicalEstimate = Math.ceil(wordCount * 1.25 + punctuationCount * 0.45 + cjkCount * 1.6);
|
|
17166
|
-
return Math.max(1, Math.max(charEstimate, lexicalEstimate));
|
|
17167
|
-
}
|
|
17168
|
-
async embedTexts(texts, modelId, task) {
|
|
17169
|
-
if (texts.length === 0) {
|
|
17170
|
-
return [];
|
|
17171
|
-
}
|
|
17172
|
-
const batches = [];
|
|
17173
|
-
for (let i = 0; i < texts.length; i += this.batchSize) {
|
|
17174
|
-
batches.push({
|
|
17175
|
-
index: i,
|
|
17176
|
-
values: texts.slice(i, i + this.batchSize)
|
|
17177
|
-
});
|
|
17178
|
-
}
|
|
17179
|
-
const outputs = new Array(batches.length);
|
|
17180
|
-
const limit = pLimit2(this.concurrency);
|
|
17181
|
-
await Promise.all(
|
|
17182
|
-
batches.map(
|
|
17183
|
-
(batch, position) => limit(async () => {
|
|
17184
|
-
outputs[position] = await this.embedWithRetry(batch.values, modelId, task ?? this.defaultTask);
|
|
17185
|
-
})
|
|
17186
|
-
)
|
|
17187
|
-
);
|
|
17188
|
-
return outputs.flat();
|
|
17189
|
-
}
|
|
17190
|
-
async embedWithRetry(texts, modelId, task) {
|
|
17191
|
-
const maxAttempts = 5;
|
|
17192
|
-
let attempt = 0;
|
|
17193
|
-
while (attempt < maxAttempts) {
|
|
17194
|
-
attempt += 1;
|
|
17195
|
-
let response;
|
|
17196
|
-
try {
|
|
17197
|
-
response = await fetch("https://api.jina.ai/v1/embeddings", {
|
|
17198
|
-
method: "POST",
|
|
17199
|
-
headers: {
|
|
17200
|
-
"content-type": "application/json",
|
|
17201
|
-
authorization: `Bearer ${this.apiKey}`
|
|
17202
|
-
},
|
|
17203
|
-
body: JSON.stringify({
|
|
17204
|
-
model: modelId,
|
|
17205
|
-
input: texts,
|
|
17206
|
-
task
|
|
17207
|
-
})
|
|
17208
|
-
});
|
|
17209
|
-
} catch (error) {
|
|
17210
|
-
if (attempt >= maxAttempts) {
|
|
17211
|
-
throw error;
|
|
17212
|
-
}
|
|
17213
|
-
await sleep(Math.min(2 ** attempt * 300, 5e3));
|
|
17214
|
-
continue;
|
|
17215
|
-
}
|
|
17216
|
-
if (!response.ok) {
|
|
17217
|
-
const retryable = response.status === 429 || response.status >= 500;
|
|
17218
|
-
if (!retryable || attempt >= maxAttempts) {
|
|
17219
|
-
const errorBody = await response.text();
|
|
17220
|
-
throw new Error(`Jina embeddings failed (${response.status}): ${errorBody}`);
|
|
17221
|
-
}
|
|
17222
|
-
await sleep(Math.min(2 ** attempt * 300, 5e3));
|
|
17223
|
-
continue;
|
|
17224
|
-
}
|
|
17225
|
-
const payload = await response.json();
|
|
17226
|
-
if (!payload.data || !Array.isArray(payload.data)) {
|
|
17227
|
-
throw new Error("Invalid Jina embeddings response format");
|
|
17228
|
-
}
|
|
17229
|
-
return payload.data.map((entry) => entry.embedding);
|
|
17230
|
-
}
|
|
17231
|
-
throw new Error("Unreachable retry state");
|
|
17232
|
-
}
|
|
17233
|
-
};
|
|
17234
|
-
|
|
17235
|
-
// src/embeddings/factory.ts
|
|
17236
|
-
function createEmbeddingsProvider(config) {
|
|
17237
|
-
if (config.embeddings.provider !== "jina") {
|
|
17238
|
-
throw new SearchSocketError(
|
|
17239
|
-
"CONFIG_MISSING",
|
|
17240
|
-
`Unsupported embeddings provider ${config.embeddings.provider}`
|
|
17241
|
-
);
|
|
17242
|
-
}
|
|
17243
|
-
const apiKey = config.embeddings.apiKey ?? process.env[config.embeddings.apiKeyEnv];
|
|
17244
|
-
if (!apiKey) {
|
|
17245
|
-
throw new SearchSocketError(
|
|
17246
|
-
"CONFIG_MISSING",
|
|
17247
|
-
`Missing embeddings API key: provide embeddings.apiKey or set env var ${config.embeddings.apiKeyEnv}`
|
|
17248
|
-
);
|
|
17249
|
-
}
|
|
17250
|
-
return new JinaEmbeddingsProvider({
|
|
17251
|
-
apiKey,
|
|
17252
|
-
batchSize: config.embeddings.batchSize,
|
|
17253
|
-
concurrency: config.embeddings.concurrency
|
|
17254
|
-
});
|
|
17255
|
-
}
|
|
17256
|
-
|
|
17257
|
-
// src/rerank/jina.ts
|
|
17258
|
-
function sleep2(ms) {
|
|
17259
|
-
return new Promise((resolve) => {
|
|
17260
|
-
setTimeout(resolve, ms);
|
|
17261
|
-
});
|
|
17262
|
-
}
|
|
17263
|
-
var JinaReranker = class {
|
|
17264
|
-
apiKey;
|
|
17265
|
-
model;
|
|
17266
|
-
maxRetries;
|
|
17267
|
-
constructor(options) {
|
|
17268
|
-
this.apiKey = options.apiKey;
|
|
17269
|
-
this.model = options.model;
|
|
17270
|
-
this.maxRetries = options.maxRetries ?? 2;
|
|
17271
|
-
}
|
|
17272
|
-
async rerank(query, candidates, topN) {
|
|
17273
|
-
if (candidates.length === 0) {
|
|
17274
|
-
return [];
|
|
17275
|
-
}
|
|
17276
|
-
const body = {
|
|
17277
|
-
model: this.model,
|
|
17278
|
-
query,
|
|
17279
|
-
documents: candidates.map((candidate) => candidate.text),
|
|
17280
|
-
top_n: topN ?? candidates.length,
|
|
17281
|
-
return_documents: false
|
|
17282
|
-
};
|
|
17283
|
-
let attempt = 0;
|
|
17284
|
-
while (attempt <= this.maxRetries) {
|
|
17285
|
-
attempt += 1;
|
|
17286
|
-
let response;
|
|
17287
|
-
try {
|
|
17288
|
-
response = await fetch("https://api.jina.ai/v1/rerank", {
|
|
17289
|
-
method: "POST",
|
|
17290
|
-
headers: {
|
|
17291
|
-
"content-type": "application/json",
|
|
17292
|
-
authorization: `Bearer ${this.apiKey}`
|
|
17293
|
-
},
|
|
17294
|
-
body: JSON.stringify(body)
|
|
17295
|
-
});
|
|
17296
|
-
} catch (error) {
|
|
17297
|
-
if (attempt <= this.maxRetries) {
|
|
17298
|
-
await sleep2(Math.min(300 * 2 ** attempt, 4e3));
|
|
17299
|
-
continue;
|
|
17300
|
-
}
|
|
17301
|
-
throw error;
|
|
17302
|
-
}
|
|
17303
|
-
if (!response.ok) {
|
|
17304
|
-
const retryable = response.status === 429 || response.status >= 500;
|
|
17305
|
-
if (retryable && attempt <= this.maxRetries) {
|
|
17306
|
-
await sleep2(Math.min(300 * 2 ** attempt, 4e3));
|
|
17307
|
-
continue;
|
|
17308
|
-
}
|
|
17309
|
-
const errorBody = await response.text();
|
|
17310
|
-
throw new Error(`Jina rerank failed (${response.status}): ${errorBody}`);
|
|
17311
|
-
}
|
|
17312
|
-
const payload = await response.json();
|
|
17313
|
-
const rawResults = payload.results ?? payload.data ?? [];
|
|
17314
|
-
if (!Array.isArray(rawResults)) {
|
|
17315
|
-
throw new Error("Invalid Jina rerank response format");
|
|
17316
|
-
}
|
|
17317
|
-
return rawResults.flatMap((item) => {
|
|
17318
|
-
const index = item.index;
|
|
17319
|
-
if (typeof index !== "number" || index < 0 || index >= candidates.length) {
|
|
17320
|
-
return [];
|
|
17321
|
-
}
|
|
17322
|
-
const candidate = candidates[index];
|
|
17323
|
-
if (!candidate) {
|
|
17324
|
-
return [];
|
|
17325
|
-
}
|
|
17326
|
-
const score = typeof item.relevance_score === "number" ? item.relevance_score : item.score ?? 0;
|
|
17327
|
-
return [
|
|
17328
|
-
{
|
|
17329
|
-
id: candidate.id,
|
|
17330
|
-
score
|
|
17331
|
-
}
|
|
17332
|
-
];
|
|
17333
|
-
}).sort((a, b) => b.score - a.score);
|
|
17334
|
-
}
|
|
17335
|
-
throw new Error("Jina rerank request failed after retries");
|
|
17336
|
-
}
|
|
17337
|
-
};
|
|
17338
|
-
|
|
17339
|
-
// src/rerank/factory.ts
|
|
17340
|
-
function createReranker(config) {
|
|
17341
|
-
if (!config.rerank.enabled) {
|
|
17342
|
-
return null;
|
|
17343
|
-
}
|
|
17344
|
-
const apiKey = config.embeddings.apiKey ?? process.env[config.embeddings.apiKeyEnv];
|
|
17345
|
-
if (!apiKey) {
|
|
17346
|
-
return null;
|
|
17347
|
-
}
|
|
17348
|
-
return new JinaReranker({
|
|
17349
|
-
apiKey,
|
|
17350
|
-
model: config.rerank.model
|
|
17351
|
-
});
|
|
17352
|
-
}
|
|
17353
17107
|
function ensureStateDirs(cwd, stateDir, scope) {
|
|
17354
17108
|
const statePath = path.resolve(cwd, stateDir);
|
|
17355
|
-
|
|
17356
|
-
|
|
17357
|
-
return { statePath, pagesPath };
|
|
17109
|
+
fs.mkdirSync(statePath, { recursive: true });
|
|
17110
|
+
return { statePath };
|
|
17358
17111
|
}
|
|
17359
17112
|
|
|
17360
|
-
// src/vector/
|
|
17361
|
-
|
|
17113
|
+
// src/vector/upstash.ts
|
|
17114
|
+
function chunkIndexName(scope) {
|
|
17115
|
+
return `${scope.projectId}--${scope.scopeName}`;
|
|
17116
|
+
}
|
|
17117
|
+
function pageIndexName(scope) {
|
|
17118
|
+
return `${scope.projectId}--${scope.scopeName}--pages`;
|
|
17119
|
+
}
|
|
17120
|
+
var UpstashSearchStore = class {
|
|
17362
17121
|
client;
|
|
17363
|
-
dimension;
|
|
17364
|
-
chunksReady = false;
|
|
17365
|
-
registryReady = false;
|
|
17366
|
-
pagesReady = false;
|
|
17367
17122
|
constructor(opts) {
|
|
17368
17123
|
this.client = opts.client;
|
|
17369
|
-
this.dimension = opts.dimension;
|
|
17370
17124
|
}
|
|
17371
|
-
|
|
17372
|
-
|
|
17373
|
-
await this.client.execute(`
|
|
17374
|
-
CREATE TABLE IF NOT EXISTS registry (
|
|
17375
|
-
scope_key TEXT PRIMARY KEY,
|
|
17376
|
-
project_id TEXT NOT NULL,
|
|
17377
|
-
scope_name TEXT NOT NULL,
|
|
17378
|
-
model_id TEXT NOT NULL,
|
|
17379
|
-
last_indexed_at TEXT NOT NULL,
|
|
17380
|
-
vector_count INTEGER,
|
|
17381
|
-
last_estimate_tokens INTEGER,
|
|
17382
|
-
last_estimate_cost_usd REAL,
|
|
17383
|
-
last_estimate_changed_chunks INTEGER
|
|
17384
|
-
)
|
|
17385
|
-
`);
|
|
17386
|
-
const estimateCols = [
|
|
17387
|
-
{ name: "last_estimate_tokens", def: "INTEGER" },
|
|
17388
|
-
{ name: "last_estimate_cost_usd", def: "REAL" },
|
|
17389
|
-
{ name: "last_estimate_changed_chunks", def: "INTEGER" }
|
|
17390
|
-
];
|
|
17391
|
-
for (const col of estimateCols) {
|
|
17392
|
-
try {
|
|
17393
|
-
await this.client.execute(`ALTER TABLE registry ADD COLUMN ${col.name} ${col.def}`);
|
|
17394
|
-
} catch (error) {
|
|
17395
|
-
if (error instanceof Error && !error.message.includes("duplicate column")) {
|
|
17396
|
-
throw error;
|
|
17397
|
-
}
|
|
17398
|
-
}
|
|
17399
|
-
}
|
|
17400
|
-
this.registryReady = true;
|
|
17401
|
-
}
|
|
17402
|
-
async ensureChunks(dim) {
|
|
17403
|
-
if (this.chunksReady) return;
|
|
17404
|
-
const exists = await this.chunksTableExists();
|
|
17405
|
-
if (exists) {
|
|
17406
|
-
const currentDim = await this.getChunksDimension();
|
|
17407
|
-
if (currentDim !== null && currentDim !== dim) {
|
|
17408
|
-
await this.client.batch([
|
|
17409
|
-
"DROP INDEX IF EXISTS idx",
|
|
17410
|
-
"DROP TABLE IF EXISTS chunks"
|
|
17411
|
-
]);
|
|
17412
|
-
}
|
|
17413
|
-
}
|
|
17414
|
-
await this.client.batch([
|
|
17415
|
-
`CREATE TABLE IF NOT EXISTS chunks (
|
|
17416
|
-
id TEXT PRIMARY KEY,
|
|
17417
|
-
project_id TEXT NOT NULL,
|
|
17418
|
-
scope_name TEXT NOT NULL,
|
|
17419
|
-
url TEXT NOT NULL,
|
|
17420
|
-
path TEXT NOT NULL,
|
|
17421
|
-
title TEXT NOT NULL,
|
|
17422
|
-
section_title TEXT NOT NULL DEFAULT '',
|
|
17423
|
-
heading_path TEXT NOT NULL DEFAULT '[]',
|
|
17424
|
-
snippet TEXT NOT NULL DEFAULT '',
|
|
17425
|
-
chunk_text TEXT NOT NULL DEFAULT '',
|
|
17426
|
-
ordinal INTEGER NOT NULL DEFAULT 0,
|
|
17427
|
-
content_hash TEXT NOT NULL DEFAULT '',
|
|
17428
|
-
model_id TEXT NOT NULL DEFAULT '',
|
|
17429
|
-
depth INTEGER NOT NULL DEFAULT 0,
|
|
17430
|
-
incoming_links INTEGER NOT NULL DEFAULT 0,
|
|
17431
|
-
route_file TEXT NOT NULL DEFAULT '',
|
|
17432
|
-
tags TEXT NOT NULL DEFAULT '[]',
|
|
17433
|
-
description TEXT NOT NULL DEFAULT '',
|
|
17434
|
-
keywords TEXT NOT NULL DEFAULT '[]',
|
|
17435
|
-
embedding F32_BLOB(${dim})
|
|
17436
|
-
)`,
|
|
17437
|
-
`CREATE INDEX IF NOT EXISTS idx ON chunks (libsql_vector_idx(embedding, 'metric=cosine'))`
|
|
17438
|
-
]);
|
|
17439
|
-
this.chunksReady = true;
|
|
17125
|
+
chunkIndex(scope) {
|
|
17126
|
+
return this.client.index(chunkIndexName(scope));
|
|
17440
17127
|
}
|
|
17441
|
-
|
|
17442
|
-
|
|
17443
|
-
await this.client.execute(`
|
|
17444
|
-
CREATE TABLE IF NOT EXISTS pages (
|
|
17445
|
-
project_id TEXT NOT NULL,
|
|
17446
|
-
scope_name TEXT NOT NULL,
|
|
17447
|
-
url TEXT NOT NULL,
|
|
17448
|
-
title TEXT NOT NULL,
|
|
17449
|
-
markdown TEXT NOT NULL,
|
|
17450
|
-
route_file TEXT NOT NULL DEFAULT '',
|
|
17451
|
-
route_resolution TEXT NOT NULL DEFAULT 'exact',
|
|
17452
|
-
incoming_links INTEGER NOT NULL DEFAULT 0,
|
|
17453
|
-
outgoing_links INTEGER NOT NULL DEFAULT 0,
|
|
17454
|
-
depth INTEGER NOT NULL DEFAULT 0,
|
|
17455
|
-
tags TEXT NOT NULL DEFAULT '[]',
|
|
17456
|
-
indexed_at TEXT NOT NULL,
|
|
17457
|
-
PRIMARY KEY (project_id, scope_name, url)
|
|
17458
|
-
)
|
|
17459
|
-
`);
|
|
17460
|
-
this.pagesReady = true;
|
|
17128
|
+
pageIndex(scope) {
|
|
17129
|
+
return this.client.index(pageIndexName(scope));
|
|
17461
17130
|
}
|
|
17462
|
-
async
|
|
17463
|
-
|
|
17464
|
-
|
|
17465
|
-
return true;
|
|
17466
|
-
} catch (error) {
|
|
17467
|
-
if (error instanceof Error && error.message.includes("no such table")) {
|
|
17468
|
-
return false;
|
|
17469
|
-
}
|
|
17470
|
-
throw error;
|
|
17471
|
-
}
|
|
17472
|
-
}
|
|
17473
|
-
/**
|
|
17474
|
-
* Read the current F32_BLOB dimension from the chunks table schema.
|
|
17475
|
-
* Returns null if the table doesn't exist or the dimension can't be parsed.
|
|
17476
|
-
*/
|
|
17477
|
-
async getChunksDimension() {
|
|
17478
|
-
try {
|
|
17479
|
-
const rs = await this.client.execute(
|
|
17480
|
-
"SELECT sql FROM sqlite_master WHERE type='table' AND name='chunks'"
|
|
17481
|
-
);
|
|
17482
|
-
if (rs.rows.length === 0) return null;
|
|
17483
|
-
const sql = rs.rows[0].sql;
|
|
17484
|
-
const match = sql.match(/F32_BLOB\((\d+)\)/i);
|
|
17485
|
-
return match ? parseInt(match[1], 10) : null;
|
|
17486
|
-
} catch {
|
|
17487
|
-
return null;
|
|
17488
|
-
}
|
|
17489
|
-
}
|
|
17490
|
-
/**
|
|
17491
|
-
* Drop all SearchSocket tables (chunks, registry, pages) and their indexes.
|
|
17492
|
-
* Used by `clean --remote` for a full reset.
|
|
17493
|
-
*/
|
|
17494
|
-
async dropAllTables() {
|
|
17495
|
-
await this.client.batch([
|
|
17496
|
-
"DROP INDEX IF EXISTS idx",
|
|
17497
|
-
"DROP TABLE IF EXISTS chunks",
|
|
17498
|
-
"DROP TABLE IF EXISTS registry",
|
|
17499
|
-
"DROP TABLE IF EXISTS pages"
|
|
17500
|
-
]);
|
|
17501
|
-
this.chunksReady = false;
|
|
17502
|
-
this.registryReady = false;
|
|
17503
|
-
this.pagesReady = false;
|
|
17504
|
-
}
|
|
17505
|
-
async upsert(records, _scope) {
|
|
17506
|
-
if (records.length === 0) return;
|
|
17507
|
-
const dim = this.dimension ?? records[0].vector.length;
|
|
17508
|
-
await this.ensureChunks(dim);
|
|
17131
|
+
async upsertChunks(chunks, scope) {
|
|
17132
|
+
if (chunks.length === 0) return;
|
|
17133
|
+
const index = this.chunkIndex(scope);
|
|
17509
17134
|
const BATCH_SIZE = 100;
|
|
17510
|
-
for (let i = 0; i <
|
|
17511
|
-
const batch =
|
|
17512
|
-
|
|
17513
|
-
sql: `INSERT OR REPLACE INTO chunks
|
|
17514
|
-
(id, project_id, scope_name, url, path, title, section_title,
|
|
17515
|
-
heading_path, snippet, chunk_text, ordinal, content_hash, model_id, depth,
|
|
17516
|
-
incoming_links, route_file, tags, description, keywords, embedding)
|
|
17517
|
-
VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, vector(?))`,
|
|
17518
|
-
args: [
|
|
17519
|
-
r.id,
|
|
17520
|
-
r.metadata.projectId,
|
|
17521
|
-
r.metadata.scopeName,
|
|
17522
|
-
r.metadata.url,
|
|
17523
|
-
r.metadata.path,
|
|
17524
|
-
r.metadata.title,
|
|
17525
|
-
r.metadata.sectionTitle,
|
|
17526
|
-
JSON.stringify(r.metadata.headingPath),
|
|
17527
|
-
r.metadata.snippet,
|
|
17528
|
-
r.metadata.chunkText,
|
|
17529
|
-
r.metadata.ordinal,
|
|
17530
|
-
r.metadata.contentHash,
|
|
17531
|
-
r.metadata.modelId,
|
|
17532
|
-
r.metadata.depth,
|
|
17533
|
-
r.metadata.incomingLinks,
|
|
17534
|
-
r.metadata.routeFile,
|
|
17535
|
-
JSON.stringify(r.metadata.tags),
|
|
17536
|
-
r.metadata.description ?? "",
|
|
17537
|
-
JSON.stringify(r.metadata.keywords ?? []),
|
|
17538
|
-
JSON.stringify(r.vector)
|
|
17539
|
-
]
|
|
17540
|
-
}));
|
|
17541
|
-
await this.client.batch(stmts);
|
|
17135
|
+
for (let i = 0; i < chunks.length; i += BATCH_SIZE) {
|
|
17136
|
+
const batch = chunks.slice(i, i + BATCH_SIZE);
|
|
17137
|
+
await index.upsert(batch);
|
|
17542
17138
|
}
|
|
17543
17139
|
}
|
|
17544
|
-
async query
|
|
17545
|
-
const
|
|
17546
|
-
await
|
|
17547
|
-
|
|
17548
|
-
|
|
17549
|
-
|
|
17550
|
-
|
|
17551
|
-
|
|
17552
|
-
|
|
17553
|
-
c.description, c.keywords,
|
|
17554
|
-
vector_distance_cos(c.embedding, vector(?)) AS distance
|
|
17555
|
-
FROM vector_top_k('idx', vector(?), ?) AS v
|
|
17556
|
-
JOIN chunks AS c ON c.rowid = v.id`,
|
|
17557
|
-
args: [queryJson, queryJson, opts.topK]
|
|
17140
|
+
async search(query, opts, scope) {
|
|
17141
|
+
const index = this.chunkIndex(scope);
|
|
17142
|
+
const results = await index.search({
|
|
17143
|
+
query,
|
|
17144
|
+
limit: opts.limit,
|
|
17145
|
+
semanticWeight: opts.semanticWeight,
|
|
17146
|
+
inputEnrichment: opts.inputEnrichment,
|
|
17147
|
+
reranking: opts.reranking,
|
|
17148
|
+
filter: opts.filter
|
|
17558
17149
|
});
|
|
17559
|
-
|
|
17560
|
-
|
|
17561
|
-
|
|
17562
|
-
|
|
17563
|
-
|
|
17564
|
-
|
|
17150
|
+
return results.map((doc) => ({
|
|
17151
|
+
id: doc.id,
|
|
17152
|
+
score: doc.score,
|
|
17153
|
+
metadata: {
|
|
17154
|
+
projectId: doc.metadata?.projectId ?? "",
|
|
17155
|
+
scopeName: doc.metadata?.scopeName ?? "",
|
|
17156
|
+
url: doc.content.url,
|
|
17157
|
+
path: doc.metadata?.path ?? "",
|
|
17158
|
+
title: doc.content.title,
|
|
17159
|
+
sectionTitle: doc.content.sectionTitle,
|
|
17160
|
+
headingPath: doc.content.headingPath ? doc.content.headingPath.split(" > ").filter(Boolean) : [],
|
|
17161
|
+
snippet: doc.metadata?.snippet ?? "",
|
|
17162
|
+
chunkText: doc.content.text,
|
|
17163
|
+
ordinal: doc.metadata?.ordinal ?? 0,
|
|
17164
|
+
contentHash: doc.metadata?.contentHash ?? "",
|
|
17165
|
+
depth: doc.metadata?.depth ?? 0,
|
|
17166
|
+
incomingLinks: doc.metadata?.incomingLinks ?? 0,
|
|
17167
|
+
routeFile: doc.metadata?.routeFile ?? "",
|
|
17168
|
+
tags: doc.content.tags ? doc.content.tags.split(",").filter(Boolean) : [],
|
|
17169
|
+
description: doc.metadata?.description || void 0,
|
|
17170
|
+
keywords: doc.metadata?.keywords ? doc.metadata.keywords.split(",").filter(Boolean) : void 0
|
|
17565
17171
|
}
|
|
17566
|
-
|
|
17567
|
-
|
|
17568
|
-
|
|
17569
|
-
|
|
17570
|
-
|
|
17571
|
-
|
|
17572
|
-
|
|
17573
|
-
|
|
17574
|
-
|
|
17575
|
-
|
|
17576
|
-
|
|
17577
|
-
|
|
17578
|
-
|
|
17579
|
-
continue;
|
|
17580
|
-
}
|
|
17581
|
-
}
|
|
17582
|
-
const distance = row.distance;
|
|
17583
|
-
const score = 1 - distance;
|
|
17584
|
-
const description = row.description || void 0;
|
|
17585
|
-
const keywords = (() => {
|
|
17586
|
-
const raw = row.keywords || "[]";
|
|
17587
|
-
const parsed = JSON.parse(raw);
|
|
17588
|
-
return parsed.length > 0 ? parsed : void 0;
|
|
17589
|
-
})();
|
|
17590
|
-
hits.push({
|
|
17591
|
-
id: row.id,
|
|
17592
|
-
score,
|
|
17593
|
-
metadata: {
|
|
17594
|
-
projectId,
|
|
17595
|
-
scopeName,
|
|
17596
|
-
url: row.url,
|
|
17597
|
-
path: rowPath,
|
|
17598
|
-
title: row.title,
|
|
17599
|
-
sectionTitle: row.section_title,
|
|
17600
|
-
headingPath: JSON.parse(row.heading_path || "[]"),
|
|
17601
|
-
snippet: row.snippet,
|
|
17602
|
-
chunkText: row.chunk_text || "",
|
|
17603
|
-
ordinal: row.ordinal || 0,
|
|
17604
|
-
contentHash: row.content_hash,
|
|
17605
|
-
modelId: row.model_id,
|
|
17606
|
-
depth: row.depth,
|
|
17607
|
-
incomingLinks: row.incoming_links,
|
|
17608
|
-
routeFile: row.route_file,
|
|
17609
|
-
tags,
|
|
17610
|
-
description,
|
|
17611
|
-
keywords
|
|
17612
|
-
}
|
|
17172
|
+
}));
|
|
17173
|
+
}
|
|
17174
|
+
async searchPages(query, opts, scope) {
|
|
17175
|
+
const index = this.pageIndex(scope);
|
|
17176
|
+
let results;
|
|
17177
|
+
try {
|
|
17178
|
+
results = await index.search({
|
|
17179
|
+
query,
|
|
17180
|
+
limit: opts.limit,
|
|
17181
|
+
semanticWeight: opts.semanticWeight,
|
|
17182
|
+
inputEnrichment: opts.inputEnrichment,
|
|
17183
|
+
reranking: true,
|
|
17184
|
+
filter: opts.filter
|
|
17613
17185
|
});
|
|
17186
|
+
} catch {
|
|
17187
|
+
return [];
|
|
17614
17188
|
}
|
|
17615
|
-
|
|
17616
|
-
|
|
17189
|
+
return results.map((doc) => ({
|
|
17190
|
+
id: doc.id,
|
|
17191
|
+
score: doc.score,
|
|
17192
|
+
title: doc.content.title,
|
|
17193
|
+
url: doc.content.url,
|
|
17194
|
+
description: doc.content.description ?? "",
|
|
17195
|
+
tags: doc.content.tags ? doc.content.tags.split(",").filter(Boolean) : [],
|
|
17196
|
+
depth: doc.metadata?.depth ?? 0,
|
|
17197
|
+
incomingLinks: doc.metadata?.incomingLinks ?? 0,
|
|
17198
|
+
routeFile: doc.metadata?.routeFile ?? ""
|
|
17199
|
+
}));
|
|
17617
17200
|
}
|
|
17618
17201
|
async deleteByIds(ids, scope) {
|
|
17619
17202
|
if (ids.length === 0) return;
|
|
17203
|
+
const index = this.chunkIndex(scope);
|
|
17620
17204
|
const BATCH_SIZE = 500;
|
|
17621
17205
|
for (let i = 0; i < ids.length; i += BATCH_SIZE) {
|
|
17622
17206
|
const batch = ids.slice(i, i + BATCH_SIZE);
|
|
17623
|
-
|
|
17624
|
-
await this.client.execute({
|
|
17625
|
-
sql: `DELETE FROM chunks WHERE project_id = ? AND scope_name = ? AND id IN (${placeholders})`,
|
|
17626
|
-
args: [scope.projectId, scope.scopeName, ...batch]
|
|
17627
|
-
});
|
|
17207
|
+
await index.delete(batch);
|
|
17628
17208
|
}
|
|
17629
17209
|
}
|
|
17630
17210
|
async deleteScope(scope) {
|
|
17631
|
-
await this.ensureRegistry();
|
|
17632
17211
|
try {
|
|
17633
|
-
|
|
17634
|
-
|
|
17635
|
-
|
|
17636
|
-
});
|
|
17637
|
-
} catch (error) {
|
|
17638
|
-
if (error instanceof Error && !error.message.includes("no such table")) {
|
|
17639
|
-
throw error;
|
|
17640
|
-
}
|
|
17212
|
+
const chunkIdx = this.chunkIndex(scope);
|
|
17213
|
+
await chunkIdx.deleteIndex();
|
|
17214
|
+
} catch {
|
|
17641
17215
|
}
|
|
17642
17216
|
try {
|
|
17643
|
-
|
|
17644
|
-
|
|
17645
|
-
|
|
17646
|
-
});
|
|
17647
|
-
} catch (error) {
|
|
17648
|
-
if (error instanceof Error && !error.message.includes("no such table")) {
|
|
17649
|
-
throw error;
|
|
17650
|
-
}
|
|
17217
|
+
const pageIdx = this.pageIndex(scope);
|
|
17218
|
+
await pageIdx.deleteIndex();
|
|
17219
|
+
} catch {
|
|
17651
17220
|
}
|
|
17652
|
-
await this.client.execute({
|
|
17653
|
-
sql: `DELETE FROM registry WHERE project_id = ? AND scope_name = ?`,
|
|
17654
|
-
args: [scope.projectId, scope.scopeName]
|
|
17655
|
-
});
|
|
17656
|
-
}
|
|
17657
|
-
async listScopes(scopeProjectId) {
|
|
17658
|
-
await this.ensureRegistry();
|
|
17659
|
-
const rs = await this.client.execute({
|
|
17660
|
-
sql: `SELECT project_id, scope_name, model_id, last_indexed_at, vector_count,
|
|
17661
|
-
last_estimate_tokens, last_estimate_cost_usd, last_estimate_changed_chunks
|
|
17662
|
-
FROM registry WHERE project_id = ?`,
|
|
17663
|
-
args: [scopeProjectId]
|
|
17664
|
-
});
|
|
17665
|
-
return rs.rows.map((row) => ({
|
|
17666
|
-
projectId: row.project_id,
|
|
17667
|
-
scopeName: row.scope_name,
|
|
17668
|
-
modelId: row.model_id,
|
|
17669
|
-
lastIndexedAt: row.last_indexed_at,
|
|
17670
|
-
vectorCount: row.vector_count,
|
|
17671
|
-
lastEstimateTokens: row.last_estimate_tokens,
|
|
17672
|
-
lastEstimateCostUSD: row.last_estimate_cost_usd,
|
|
17673
|
-
lastEstimateChangedChunks: row.last_estimate_changed_chunks
|
|
17674
|
-
}));
|
|
17675
17221
|
}
|
|
17676
|
-
async
|
|
17677
|
-
await this.
|
|
17678
|
-
const
|
|
17679
|
-
|
|
17680
|
-
|
|
17681
|
-
|
|
17682
|
-
|
|
17683
|
-
|
|
17684
|
-
|
|
17685
|
-
|
|
17686
|
-
|
|
17687
|
-
|
|
17688
|
-
|
|
17689
|
-
|
|
17690
|
-
|
|
17691
|
-
|
|
17692
|
-
|
|
17693
|
-
|
|
17694
|
-
|
|
17695
|
-
|
|
17222
|
+
async listScopes(projectId) {
|
|
17223
|
+
const allIndexes = await this.client.listIndexes();
|
|
17224
|
+
const prefix = `${projectId}--`;
|
|
17225
|
+
const scopeNames = /* @__PURE__ */ new Set();
|
|
17226
|
+
for (const name of allIndexes) {
|
|
17227
|
+
if (name.startsWith(prefix) && !name.endsWith("--pages")) {
|
|
17228
|
+
const scopeName = name.slice(prefix.length);
|
|
17229
|
+
scopeNames.add(scopeName);
|
|
17230
|
+
}
|
|
17231
|
+
}
|
|
17232
|
+
const scopes = [];
|
|
17233
|
+
for (const scopeName of scopeNames) {
|
|
17234
|
+
const scope = {
|
|
17235
|
+
projectId,
|
|
17236
|
+
scopeName,
|
|
17237
|
+
scopeId: `${projectId}:${scopeName}`
|
|
17238
|
+
};
|
|
17239
|
+
try {
|
|
17240
|
+
const info = await this.chunkIndex(scope).info();
|
|
17241
|
+
scopes.push({
|
|
17242
|
+
projectId,
|
|
17243
|
+
scopeName,
|
|
17244
|
+
lastIndexedAt: (/* @__PURE__ */ new Date()).toISOString(),
|
|
17245
|
+
documentCount: info.documentCount
|
|
17246
|
+
});
|
|
17247
|
+
} catch {
|
|
17248
|
+
scopes.push({
|
|
17249
|
+
projectId,
|
|
17250
|
+
scopeName,
|
|
17251
|
+
lastIndexedAt: "unknown",
|
|
17252
|
+
documentCount: 0
|
|
17253
|
+
});
|
|
17254
|
+
}
|
|
17255
|
+
}
|
|
17256
|
+
return scopes;
|
|
17696
17257
|
}
|
|
17697
17258
|
async getContentHashes(scope) {
|
|
17698
|
-
const exists = await this.chunksTableExists();
|
|
17699
|
-
if (!exists) return /* @__PURE__ */ new Map();
|
|
17700
|
-
const rs = await this.client.execute({
|
|
17701
|
-
sql: `SELECT id, content_hash FROM chunks WHERE project_id = ? AND scope_name = ?`,
|
|
17702
|
-
args: [scope.projectId, scope.scopeName]
|
|
17703
|
-
});
|
|
17704
17259
|
const map = /* @__PURE__ */ new Map();
|
|
17705
|
-
|
|
17706
|
-
|
|
17260
|
+
const index = this.chunkIndex(scope);
|
|
17261
|
+
let cursor = "0";
|
|
17262
|
+
try {
|
|
17263
|
+
for (; ; ) {
|
|
17264
|
+
const result = await index.range({ cursor, limit: 100 });
|
|
17265
|
+
for (const doc of result.documents) {
|
|
17266
|
+
if (doc.metadata?.contentHash) {
|
|
17267
|
+
map.set(doc.id, doc.metadata.contentHash);
|
|
17268
|
+
}
|
|
17269
|
+
}
|
|
17270
|
+
if (!result.nextCursor || result.nextCursor === "0") break;
|
|
17271
|
+
cursor = result.nextCursor;
|
|
17272
|
+
}
|
|
17273
|
+
} catch {
|
|
17707
17274
|
}
|
|
17708
17275
|
return map;
|
|
17709
17276
|
}
|
|
17710
17277
|
async upsertPages(pages, scope) {
|
|
17711
17278
|
if (pages.length === 0) return;
|
|
17712
|
-
|
|
17713
|
-
|
|
17714
|
-
if (page.projectId !== scope.projectId || page.scopeName !== scope.scopeName) {
|
|
17715
|
-
throw new Error(
|
|
17716
|
-
`Page scope mismatch: page has ${page.projectId}:${page.scopeName} but scope is ${scope.projectId}:${scope.scopeName}`
|
|
17717
|
-
);
|
|
17718
|
-
}
|
|
17719
|
-
}
|
|
17720
|
-
const BATCH_SIZE = 100;
|
|
17279
|
+
const index = this.pageIndex(scope);
|
|
17280
|
+
const BATCH_SIZE = 50;
|
|
17721
17281
|
for (let i = 0; i < pages.length; i += BATCH_SIZE) {
|
|
17722
17282
|
const batch = pages.slice(i, i + BATCH_SIZE);
|
|
17723
|
-
const
|
|
17724
|
-
|
|
17725
|
-
|
|
17726
|
-
|
|
17727
|
-
|
|
17728
|
-
|
|
17729
|
-
p.
|
|
17730
|
-
p.
|
|
17731
|
-
p.
|
|
17732
|
-
p.
|
|
17733
|
-
|
|
17734
|
-
|
|
17735
|
-
p.
|
|
17736
|
-
p.
|
|
17737
|
-
p.
|
|
17738
|
-
p.
|
|
17739
|
-
|
|
17740
|
-
p.
|
|
17741
|
-
|
|
17283
|
+
const docs = batch.map((p) => ({
|
|
17284
|
+
id: p.url,
|
|
17285
|
+
content: {
|
|
17286
|
+
title: p.title,
|
|
17287
|
+
url: p.url,
|
|
17288
|
+
type: "page",
|
|
17289
|
+
description: p.description ?? "",
|
|
17290
|
+
keywords: (p.keywords ?? []).join(","),
|
|
17291
|
+
summary: p.summary ?? "",
|
|
17292
|
+
tags: p.tags.join(",")
|
|
17293
|
+
},
|
|
17294
|
+
metadata: {
|
|
17295
|
+
markdown: p.markdown,
|
|
17296
|
+
projectId: p.projectId,
|
|
17297
|
+
scopeName: p.scopeName,
|
|
17298
|
+
routeFile: p.routeFile,
|
|
17299
|
+
routeResolution: p.routeResolution,
|
|
17300
|
+
incomingLinks: p.incomingLinks,
|
|
17301
|
+
outgoingLinks: p.outgoingLinks,
|
|
17302
|
+
depth: p.depth,
|
|
17303
|
+
indexedAt: p.indexedAt
|
|
17304
|
+
}
|
|
17742
17305
|
}));
|
|
17743
|
-
await
|
|
17306
|
+
await index.upsert(docs);
|
|
17744
17307
|
}
|
|
17745
17308
|
}
|
|
17746
17309
|
async getPage(url, scope) {
|
|
17747
|
-
|
|
17748
|
-
|
|
17749
|
-
|
|
17750
|
-
|
|
17751
|
-
|
|
17752
|
-
|
|
17753
|
-
|
|
17754
|
-
|
|
17755
|
-
|
|
17756
|
-
|
|
17757
|
-
|
|
17758
|
-
|
|
17759
|
-
|
|
17760
|
-
|
|
17761
|
-
|
|
17762
|
-
|
|
17763
|
-
|
|
17764
|
-
|
|
17765
|
-
|
|
17766
|
-
|
|
17767
|
-
|
|
17310
|
+
const index = this.pageIndex(scope);
|
|
17311
|
+
try {
|
|
17312
|
+
const results = await index.fetch([url]);
|
|
17313
|
+
const doc = results[0];
|
|
17314
|
+
if (!doc) return null;
|
|
17315
|
+
return {
|
|
17316
|
+
url: doc.content.url,
|
|
17317
|
+
title: doc.content.title,
|
|
17318
|
+
markdown: doc.metadata.markdown,
|
|
17319
|
+
projectId: doc.metadata.projectId,
|
|
17320
|
+
scopeName: doc.metadata.scopeName,
|
|
17321
|
+
routeFile: doc.metadata.routeFile,
|
|
17322
|
+
routeResolution: doc.metadata.routeResolution,
|
|
17323
|
+
incomingLinks: doc.metadata.incomingLinks,
|
|
17324
|
+
outgoingLinks: doc.metadata.outgoingLinks,
|
|
17325
|
+
depth: doc.metadata.depth,
|
|
17326
|
+
tags: doc.content.tags ? doc.content.tags.split(",").filter(Boolean) : [],
|
|
17327
|
+
indexedAt: doc.metadata.indexedAt,
|
|
17328
|
+
summary: doc.content.summary || void 0,
|
|
17329
|
+
description: doc.content.description || void 0,
|
|
17330
|
+
keywords: doc.content.keywords ? doc.content.keywords.split(",").filter(Boolean) : void 0
|
|
17331
|
+
};
|
|
17332
|
+
} catch {
|
|
17333
|
+
return null;
|
|
17334
|
+
}
|
|
17768
17335
|
}
|
|
17769
17336
|
async deletePages(scope) {
|
|
17770
|
-
|
|
17771
|
-
|
|
17772
|
-
|
|
17773
|
-
|
|
17774
|
-
}
|
|
17775
|
-
}
|
|
17776
|
-
async getScopeModelId(scope) {
|
|
17777
|
-
await this.ensureRegistry();
|
|
17778
|
-
const rs = await this.client.execute({
|
|
17779
|
-
sql: `SELECT model_id FROM registry WHERE project_id = ? AND scope_name = ?`,
|
|
17780
|
-
args: [scope.projectId, scope.scopeName]
|
|
17781
|
-
});
|
|
17782
|
-
if (rs.rows.length === 0) return null;
|
|
17783
|
-
return rs.rows[0].model_id;
|
|
17337
|
+
try {
|
|
17338
|
+
const index = this.pageIndex(scope);
|
|
17339
|
+
await index.reset();
|
|
17340
|
+
} catch {
|
|
17341
|
+
}
|
|
17784
17342
|
}
|
|
17785
17343
|
async health() {
|
|
17786
17344
|
try {
|
|
17787
|
-
await this.client.
|
|
17345
|
+
await this.client.info();
|
|
17788
17346
|
return { ok: true };
|
|
17789
17347
|
} catch (error) {
|
|
17790
17348
|
return {
|
|
@@ -17793,40 +17351,34 @@ var TursoVectorStore = class {
|
|
|
17793
17351
|
};
|
|
17794
17352
|
}
|
|
17795
17353
|
}
|
|
17354
|
+
async dropAllIndexes(projectId) {
|
|
17355
|
+
const allIndexes = await this.client.listIndexes();
|
|
17356
|
+
const prefix = `${projectId}--`;
|
|
17357
|
+
for (const name of allIndexes) {
|
|
17358
|
+
if (name.startsWith(prefix)) {
|
|
17359
|
+
try {
|
|
17360
|
+
const index = this.client.index(name);
|
|
17361
|
+
await index.deleteIndex();
|
|
17362
|
+
} catch {
|
|
17363
|
+
}
|
|
17364
|
+
}
|
|
17365
|
+
}
|
|
17366
|
+
}
|
|
17796
17367
|
};
|
|
17797
17368
|
|
|
17798
17369
|
// src/vector/factory.ts
|
|
17799
|
-
async function
|
|
17800
|
-
const
|
|
17801
|
-
const
|
|
17802
|
-
if (
|
|
17803
|
-
const { createClient: createClient2 } = await import('@libsql/client/http');
|
|
17804
|
-
const authToken = turso.authToken ?? process.env[turso.authTokenEnv];
|
|
17805
|
-
const client2 = createClient2({
|
|
17806
|
-
url: remoteUrl,
|
|
17807
|
-
authToken
|
|
17808
|
-
});
|
|
17809
|
-
return new TursoVectorStore({
|
|
17810
|
-
client: client2,
|
|
17811
|
-
dimension: config.vector.dimension
|
|
17812
|
-
});
|
|
17813
|
-
}
|
|
17814
|
-
if (isServerless()) {
|
|
17370
|
+
async function createUpstashStore(config) {
|
|
17371
|
+
const url = config.upstash.url ?? process.env[config.upstash.urlEnv];
|
|
17372
|
+
const token = config.upstash.token ?? process.env[config.upstash.tokenEnv];
|
|
17373
|
+
if (!url || !token) {
|
|
17815
17374
|
throw new SearchSocketError(
|
|
17816
17375
|
"VECTOR_BACKEND_UNAVAILABLE",
|
|
17817
|
-
`
|
|
17376
|
+
`Missing Upstash Search credentials. Set ${config.upstash.urlEnv} and ${config.upstash.tokenEnv} environment variables, or pass upstash.url and upstash.token in your config.`
|
|
17818
17377
|
);
|
|
17819
17378
|
}
|
|
17820
|
-
const {
|
|
17821
|
-
const
|
|
17822
|
-
|
|
17823
|
-
const client = createClient({
|
|
17824
|
-
url: `file:${localPath}`
|
|
17825
|
-
});
|
|
17826
|
-
return new TursoVectorStore({
|
|
17827
|
-
client,
|
|
17828
|
-
dimension: config.vector.dimension
|
|
17829
|
-
});
|
|
17379
|
+
const { Search } = await import('@upstash/search');
|
|
17380
|
+
const client = new Search({ url, token });
|
|
17381
|
+
return new UpstashSearchStore({ client });
|
|
17830
17382
|
}
|
|
17831
17383
|
function sha1(input) {
|
|
17832
17384
|
return createHash("sha1").update(input).digest("hex");
|
|
@@ -17845,13 +17397,6 @@ function normalizeUrlPath(rawPath) {
|
|
|
17845
17397
|
}
|
|
17846
17398
|
return out;
|
|
17847
17399
|
}
|
|
17848
|
-
function urlPathToMirrorRelative(urlPath) {
|
|
17849
|
-
const normalized = normalizeUrlPath(urlPath);
|
|
17850
|
-
if (normalized === "/") {
|
|
17851
|
-
return "index.md";
|
|
17852
|
-
}
|
|
17853
|
-
return `${normalized.slice(1)}.md`;
|
|
17854
|
-
}
|
|
17855
17400
|
function staticHtmlFileToUrl(filePath, rootDir) {
|
|
17856
17401
|
const relative = path.relative(rootDir, filePath).replace(/\\/g, "/");
|
|
17857
17402
|
if (relative === "index.html") {
|
|
@@ -18126,7 +17671,7 @@ function buildEmbeddingText(chunk, prependTitle) {
|
|
|
18126
17671
|
|
|
18127
17672
|
${chunk.chunkText}`;
|
|
18128
17673
|
}
|
|
18129
|
-
function
|
|
17674
|
+
function chunkPage(page, config, scope) {
|
|
18130
17675
|
const sections = parseHeadingSections(page.markdown, config.chunking.headingPathDepth);
|
|
18131
17676
|
const rawChunks = sections.flatMap((section) => splitSection(section, config.chunking));
|
|
18132
17677
|
const chunks = [];
|
|
@@ -19157,53 +18702,6 @@ function extractFromMarkdown(url, markdown, title) {
|
|
|
19157
18702
|
weight: mdWeight
|
|
19158
18703
|
};
|
|
19159
18704
|
}
|
|
19160
|
-
function yamlString(value) {
|
|
19161
|
-
return JSON.stringify(value);
|
|
19162
|
-
}
|
|
19163
|
-
function yamlArray(values) {
|
|
19164
|
-
return `[${values.map((v) => JSON.stringify(v)).join(", ")}]`;
|
|
19165
|
-
}
|
|
19166
|
-
function buildMirrorMarkdown(page) {
|
|
19167
|
-
const frontmatterLines = [
|
|
19168
|
-
"---",
|
|
19169
|
-
`url: ${yamlString(page.url)}`,
|
|
19170
|
-
`title: ${yamlString(page.title)}`,
|
|
19171
|
-
`scope: ${yamlString(page.scope)}`,
|
|
19172
|
-
`routeFile: ${yamlString(page.routeFile)}`,
|
|
19173
|
-
`routeResolution: ${yamlString(page.routeResolution)}`,
|
|
19174
|
-
`generatedAt: ${yamlString(page.generatedAt)}`,
|
|
19175
|
-
`incomingLinks: ${page.incomingLinks}`,
|
|
19176
|
-
`outgoingLinks: ${page.outgoingLinks}`,
|
|
19177
|
-
`depth: ${page.depth}`,
|
|
19178
|
-
`tags: ${yamlArray(page.tags)}`,
|
|
19179
|
-
"---",
|
|
19180
|
-
""
|
|
19181
|
-
];
|
|
19182
|
-
return `${frontmatterLines.join("\n")}${normalizeMarkdown(page.markdown)}`;
|
|
19183
|
-
}
|
|
19184
|
-
function stripGeneratedAt(content) {
|
|
19185
|
-
return content.replace(/^generatedAt: .*$/m, "");
|
|
19186
|
-
}
|
|
19187
|
-
async function writeMirrorPage(statePath, scope, page) {
|
|
19188
|
-
const relative = urlPathToMirrorRelative(page.url);
|
|
19189
|
-
const outputPath = path.join(statePath, "pages", scope.scopeName, relative);
|
|
19190
|
-
await fs4.mkdir(path.dirname(outputPath), { recursive: true });
|
|
19191
|
-
const newContent = buildMirrorMarkdown(page);
|
|
19192
|
-
try {
|
|
19193
|
-
const existing = await fs4.readFile(outputPath, "utf8");
|
|
19194
|
-
if (stripGeneratedAt(existing) === stripGeneratedAt(newContent)) {
|
|
19195
|
-
return outputPath;
|
|
19196
|
-
}
|
|
19197
|
-
} catch {
|
|
19198
|
-
}
|
|
19199
|
-
await fs4.writeFile(outputPath, newContent, "utf8");
|
|
19200
|
-
return outputPath;
|
|
19201
|
-
}
|
|
19202
|
-
async function cleanMirrorForScope(statePath, scope) {
|
|
19203
|
-
const target = path.join(statePath, "pages", scope.scopeName);
|
|
19204
|
-
await fs4.rm(target, { recursive: true, force: true });
|
|
19205
|
-
await fs4.mkdir(target, { recursive: true });
|
|
19206
|
-
}
|
|
19207
18705
|
function segmentToRegex(segment) {
|
|
19208
18706
|
if (segment.startsWith("(") && segment.endsWith(")")) {
|
|
19209
18707
|
return { regex: "", score: 0 };
|
|
@@ -19396,7 +18894,7 @@ async function parseManifest(cwd, outputDir) {
|
|
|
19396
18894
|
const manifestPath = path.resolve(cwd, outputDir, "server", "manifest-full.js");
|
|
19397
18895
|
let content;
|
|
19398
18896
|
try {
|
|
19399
|
-
content = await
|
|
18897
|
+
content = await fs3.readFile(manifestPath, "utf8");
|
|
19400
18898
|
} catch {
|
|
19401
18899
|
throw new SearchSocketError(
|
|
19402
18900
|
"BUILD_MANIFEST_NOT_FOUND",
|
|
@@ -19569,7 +19067,7 @@ async function discoverPages(server, buildConfig, pipelineMaxPages) {
|
|
|
19569
19067
|
const visited = /* @__PURE__ */ new Set();
|
|
19570
19068
|
const pages = [];
|
|
19571
19069
|
const queue = [];
|
|
19572
|
-
const limit =
|
|
19070
|
+
const limit = pLimit(8);
|
|
19573
19071
|
for (const seed of seedUrls) {
|
|
19574
19072
|
const normalized = normalizeUrlPath(seed);
|
|
19575
19073
|
if (!visited.has(normalized) && !isExcluded(normalized, exclude)) {
|
|
@@ -19651,7 +19149,7 @@ async function loadBuildPages(cwd, config, maxPages) {
|
|
|
19651
19149
|
const selected = typeof maxCount === "number" ? expanded.slice(0, maxCount) : expanded;
|
|
19652
19150
|
const server = await startPreviewServer(cwd, { previewTimeout: buildConfig.previewTimeout }, logger);
|
|
19653
19151
|
try {
|
|
19654
|
-
const concurrencyLimit =
|
|
19152
|
+
const concurrencyLimit = pLimit(8);
|
|
19655
19153
|
const results = await Promise.allSettled(
|
|
19656
19154
|
selected.map(
|
|
19657
19155
|
(route) => concurrencyLimit(async () => {
|
|
@@ -19725,7 +19223,7 @@ async function loadContentFilesPages(cwd, config, maxPages) {
|
|
|
19725
19223
|
const selected = typeof limit === "number" ? files.slice(0, limit) : files;
|
|
19726
19224
|
const pages = [];
|
|
19727
19225
|
for (const filePath of selected) {
|
|
19728
|
-
const raw = await
|
|
19226
|
+
const raw = await fs3.readFile(filePath, "utf8");
|
|
19729
19227
|
const markdown = filePath.endsWith(".md") ? raw : normalizeSvelteToMarkdown(raw);
|
|
19730
19228
|
pages.push({
|
|
19731
19229
|
url: filePathToUrl(filePath, baseDir),
|
|
@@ -19820,7 +19318,7 @@ async function loadCrawledPages(config, maxPages) {
|
|
|
19820
19318
|
const routes = await resolveRoutes(config);
|
|
19821
19319
|
const maxCount = typeof maxPages === "number" ? Math.max(0, Math.floor(maxPages)) : void 0;
|
|
19822
19320
|
const selected = typeof maxCount === "number" ? routes.slice(0, maxCount) : routes;
|
|
19823
|
-
const concurrencyLimit =
|
|
19321
|
+
const concurrencyLimit = pLimit(8);
|
|
19824
19322
|
const results = await Promise.allSettled(
|
|
19825
19323
|
selected.map(
|
|
19826
19324
|
(route) => concurrencyLimit(async () => {
|
|
@@ -19861,7 +19359,7 @@ async function loadStaticOutputPages(cwd, config, maxPages) {
|
|
|
19861
19359
|
const selected = typeof limit === "number" ? htmlFiles.slice(0, limit) : htmlFiles;
|
|
19862
19360
|
const pages = [];
|
|
19863
19361
|
for (const filePath of selected) {
|
|
19864
|
-
const html = await
|
|
19362
|
+
const html = await fs3.readFile(filePath, "utf8");
|
|
19865
19363
|
pages.push({
|
|
19866
19364
|
url: staticHtmlFileToUrl(filePath, outputDir),
|
|
19867
19365
|
html,
|
|
@@ -19924,7 +19422,7 @@ function isBlockedByRobots(urlPath, rules3) {
|
|
|
19924
19422
|
}
|
|
19925
19423
|
async function loadRobotsTxtFromDir(dir) {
|
|
19926
19424
|
try {
|
|
19927
|
-
const content = await
|
|
19425
|
+
const content = await fs3.readFile(path.join(dir, "robots.txt"), "utf8");
|
|
19928
19426
|
return parseRobotsTxt(content);
|
|
19929
19427
|
} catch {
|
|
19930
19428
|
return null;
|
|
@@ -19949,7 +19447,12 @@ function nonNegativeOrZero(value) {
|
|
|
19949
19447
|
}
|
|
19950
19448
|
return Math.max(0, value);
|
|
19951
19449
|
}
|
|
19952
|
-
function
|
|
19450
|
+
function normalizeForTitleMatch(text) {
|
|
19451
|
+
return text.toLowerCase().replace(/[^a-z0-9\s]/g, "").replace(/\s+/g, " ").trim();
|
|
19452
|
+
}
|
|
19453
|
+
function rankHits(hits, config, query) {
|
|
19454
|
+
const normalizedQuery = query ? normalizeForTitleMatch(query) : "";
|
|
19455
|
+
const titleMatchWeight = config.ranking.weights.titleMatch;
|
|
19953
19456
|
return hits.map((hit) => {
|
|
19954
19457
|
let score = Number.isFinite(hit.score) ? hit.score : Number.NEGATIVE_INFINITY;
|
|
19955
19458
|
if (config.ranking.enableIncomingLinkBoost) {
|
|
@@ -19960,6 +19463,12 @@ function rankHits(hits, config) {
|
|
|
19960
19463
|
const depthBoost = 1 / (1 + nonNegativeOrZero(hit.metadata.depth));
|
|
19961
19464
|
score += depthBoost * config.ranking.weights.depth;
|
|
19962
19465
|
}
|
|
19466
|
+
if (normalizedQuery && titleMatchWeight > 0) {
|
|
19467
|
+
const normalizedTitle = normalizeForTitleMatch(hit.metadata.title);
|
|
19468
|
+
if (normalizedQuery.length > 0 && normalizedTitle.length > 0 && (normalizedTitle.includes(normalizedQuery) || normalizedQuery.includes(normalizedTitle))) {
|
|
19469
|
+
score += titleMatchWeight;
|
|
19470
|
+
}
|
|
19471
|
+
}
|
|
19963
19472
|
return {
|
|
19964
19473
|
hit,
|
|
19965
19474
|
finalScore: Number.isFinite(score) ? score : Number.NEGATIVE_INFINITY
|
|
@@ -19969,6 +19478,30 @@ function rankHits(hits, config) {
|
|
|
19969
19478
|
return Number.isNaN(delta) ? 0 : delta;
|
|
19970
19479
|
});
|
|
19971
19480
|
}
|
|
19481
|
+
function trimByScoreGap(results, config) {
|
|
19482
|
+
if (results.length === 0) return results;
|
|
19483
|
+
const threshold = config.ranking.scoreGapThreshold;
|
|
19484
|
+
const minScore = config.ranking.minScore;
|
|
19485
|
+
if (minScore > 0 && results.length > 0) {
|
|
19486
|
+
const sortedScores = results.map((r) => r.pageScore).sort((a, b) => a - b);
|
|
19487
|
+
const mid = Math.floor(sortedScores.length / 2);
|
|
19488
|
+
const median = sortedScores.length % 2 === 0 ? (sortedScores[mid - 1] + sortedScores[mid]) / 2 : sortedScores[mid];
|
|
19489
|
+
if (median < minScore) return [];
|
|
19490
|
+
}
|
|
19491
|
+
if (threshold > 0 && results.length > 1) {
|
|
19492
|
+
for (let i = 1; i < results.length; i++) {
|
|
19493
|
+
const prev = results[i - 1].pageScore;
|
|
19494
|
+
const current = results[i].pageScore;
|
|
19495
|
+
if (prev > 0) {
|
|
19496
|
+
const gap = (prev - current) / prev;
|
|
19497
|
+
if (gap >= threshold) {
|
|
19498
|
+
return results.slice(0, i);
|
|
19499
|
+
}
|
|
19500
|
+
}
|
|
19501
|
+
}
|
|
19502
|
+
}
|
|
19503
|
+
return results;
|
|
19504
|
+
}
|
|
19972
19505
|
function findPageWeight(url, pageWeights) {
|
|
19973
19506
|
let bestPattern = "";
|
|
19974
19507
|
let bestWeight = 1;
|
|
@@ -20023,6 +19556,61 @@ function aggregateByPage(ranked, config) {
|
|
|
20023
19556
|
return Number.isNaN(delta) ? 0 : delta;
|
|
20024
19557
|
});
|
|
20025
19558
|
}
|
|
19559
|
+
function mergePageAndChunkResults(pageHits, rankedChunks, config) {
|
|
19560
|
+
if (pageHits.length === 0) return rankedChunks;
|
|
19561
|
+
const w = config.search.pageSearchWeight;
|
|
19562
|
+
const pageScoreMap = /* @__PURE__ */ new Map();
|
|
19563
|
+
for (const ph of pageHits) {
|
|
19564
|
+
pageScoreMap.set(ph.url, ph);
|
|
19565
|
+
}
|
|
19566
|
+
const pagesWithChunks = /* @__PURE__ */ new Set();
|
|
19567
|
+
const merged = rankedChunks.map((ranked) => {
|
|
19568
|
+
const url = ranked.hit.metadata.url;
|
|
19569
|
+
const pageHit = pageScoreMap.get(url);
|
|
19570
|
+
if (pageHit) {
|
|
19571
|
+
pagesWithChunks.add(url);
|
|
19572
|
+
const blended = (1 - w) * ranked.finalScore + w * pageHit.score;
|
|
19573
|
+
return {
|
|
19574
|
+
hit: ranked.hit,
|
|
19575
|
+
finalScore: Number.isFinite(blended) ? blended : ranked.finalScore
|
|
19576
|
+
};
|
|
19577
|
+
}
|
|
19578
|
+
return ranked;
|
|
19579
|
+
});
|
|
19580
|
+
for (const [url, pageHit] of pageScoreMap) {
|
|
19581
|
+
if (pagesWithChunks.has(url)) continue;
|
|
19582
|
+
const syntheticScore = pageHit.score * w;
|
|
19583
|
+
const syntheticHit = {
|
|
19584
|
+
id: `page:${url}`,
|
|
19585
|
+
score: pageHit.score,
|
|
19586
|
+
metadata: {
|
|
19587
|
+
projectId: "",
|
|
19588
|
+
scopeName: "",
|
|
19589
|
+
url: pageHit.url,
|
|
19590
|
+
path: pageHit.url,
|
|
19591
|
+
title: pageHit.title,
|
|
19592
|
+
sectionTitle: "",
|
|
19593
|
+
headingPath: [],
|
|
19594
|
+
snippet: pageHit.description || pageHit.title,
|
|
19595
|
+
chunkText: pageHit.description || pageHit.title,
|
|
19596
|
+
ordinal: 0,
|
|
19597
|
+
contentHash: "",
|
|
19598
|
+
depth: pageHit.depth,
|
|
19599
|
+
incomingLinks: pageHit.incomingLinks,
|
|
19600
|
+
routeFile: pageHit.routeFile,
|
|
19601
|
+
tags: pageHit.tags
|
|
19602
|
+
}
|
|
19603
|
+
};
|
|
19604
|
+
merged.push({
|
|
19605
|
+
hit: syntheticHit,
|
|
19606
|
+
finalScore: Number.isFinite(syntheticScore) ? syntheticScore : 0
|
|
19607
|
+
});
|
|
19608
|
+
}
|
|
19609
|
+
return merged.sort((a, b) => {
|
|
19610
|
+
const delta = b.finalScore - a.finalScore;
|
|
19611
|
+
return Number.isNaN(delta) ? 0 : delta;
|
|
19612
|
+
});
|
|
19613
|
+
}
|
|
20026
19614
|
|
|
20027
19615
|
// src/utils/time.ts
|
|
20028
19616
|
function nowIso() {
|
|
@@ -20033,34 +19621,41 @@ function hrTimeMs(start) {
|
|
|
20033
19621
|
}
|
|
20034
19622
|
|
|
20035
19623
|
// src/indexing/pipeline.ts
|
|
20036
|
-
|
|
20037
|
-
|
|
20038
|
-
|
|
20039
|
-
|
|
20040
|
-
|
|
19624
|
+
function buildPageSummary(page, maxChars = 3500) {
|
|
19625
|
+
const parts = [page.title];
|
|
19626
|
+
if (page.description) {
|
|
19627
|
+
parts.push(page.description);
|
|
19628
|
+
}
|
|
19629
|
+
if (page.keywords && page.keywords.length > 0) {
|
|
19630
|
+
parts.push(page.keywords.join(", "));
|
|
19631
|
+
}
|
|
19632
|
+
const plainBody = page.markdown.replace(/```[\s\S]*?```/g, " ").replace(/`([^`]+)`/g, "$1").replace(/!?\[([^\]]*)\]\([^)]*\)/g, "$1").replace(/^#{1,6}\s+/gm, "").replace(/[>*_|~\-]/g, " ").replace(/\s+/g, " ").trim();
|
|
19633
|
+
if (plainBody) {
|
|
19634
|
+
parts.push(plainBody);
|
|
19635
|
+
}
|
|
19636
|
+
const joined = parts.join("\n\n");
|
|
19637
|
+
if (joined.length <= maxChars) return joined;
|
|
19638
|
+
return joined.slice(0, maxChars).trim();
|
|
19639
|
+
}
|
|
20041
19640
|
var IndexPipeline = class _IndexPipeline {
|
|
20042
19641
|
cwd;
|
|
20043
19642
|
config;
|
|
20044
|
-
|
|
20045
|
-
vectorStore;
|
|
19643
|
+
store;
|
|
20046
19644
|
logger;
|
|
20047
19645
|
constructor(options) {
|
|
20048
19646
|
this.cwd = options.cwd;
|
|
20049
19647
|
this.config = options.config;
|
|
20050
|
-
this.
|
|
20051
|
-
this.vectorStore = options.vectorStore;
|
|
19648
|
+
this.store = options.store;
|
|
20052
19649
|
this.logger = options.logger;
|
|
20053
19650
|
}
|
|
20054
19651
|
static async create(options = {}) {
|
|
20055
19652
|
const cwd = path.resolve(options.cwd ?? process.cwd());
|
|
20056
19653
|
const config = options.config ?? await loadConfig({ cwd, configPath: options.configPath });
|
|
20057
|
-
const
|
|
20058
|
-
const vectorStore = options.vectorStore ?? await createVectorStore(config, cwd);
|
|
19654
|
+
const store = options.store ?? await createUpstashStore(config);
|
|
20059
19655
|
return new _IndexPipeline({
|
|
20060
19656
|
cwd,
|
|
20061
19657
|
config,
|
|
20062
|
-
|
|
20063
|
-
vectorStore,
|
|
19658
|
+
store,
|
|
20064
19659
|
logger: options.logger ?? new Logger()
|
|
20065
19660
|
});
|
|
20066
19661
|
}
|
|
@@ -20080,25 +19675,17 @@ var IndexPipeline = class _IndexPipeline {
|
|
|
20080
19675
|
stageTimingsMs[name] = Math.round(hrTimeMs(start));
|
|
20081
19676
|
};
|
|
20082
19677
|
const scope = resolveScope(this.config, options.scopeOverride);
|
|
20083
|
-
|
|
19678
|
+
ensureStateDirs(this.cwd, this.config.state.dir);
|
|
20084
19679
|
const sourceMode = options.sourceOverride ?? this.config.source.mode;
|
|
20085
|
-
this.logger.info(`Indexing scope "${scope.scopeName}" (source: ${sourceMode},
|
|
19680
|
+
this.logger.info(`Indexing scope "${scope.scopeName}" (source: ${sourceMode}, backend: upstash-search)`);
|
|
20086
19681
|
if (options.force) {
|
|
20087
19682
|
this.logger.info("Force mode enabled \u2014 full rebuild");
|
|
20088
|
-
await cleanMirrorForScope(statePath, scope);
|
|
20089
19683
|
}
|
|
20090
19684
|
if (options.dryRun) {
|
|
20091
19685
|
this.logger.info("Dry run \u2014 no writes will be performed");
|
|
20092
19686
|
}
|
|
20093
19687
|
const manifestStart = stageStart();
|
|
20094
|
-
const existingHashes = await this.
|
|
20095
|
-
const existingModelId = await this.vectorStore.getScopeModelId(scope);
|
|
20096
|
-
if (existingModelId && existingModelId !== this.config.embeddings.model && !options.force) {
|
|
20097
|
-
throw new SearchSocketError(
|
|
20098
|
-
"EMBEDDING_MODEL_MISMATCH",
|
|
20099
|
-
`Scope ${scope.scopeName} uses model ${existingModelId}. Re-run with --force to migrate.`
|
|
20100
|
-
);
|
|
20101
|
-
}
|
|
19688
|
+
const existingHashes = options.force ? /* @__PURE__ */ new Map() : await this.store.getContentHashes(scope);
|
|
20102
19689
|
stageEnd("manifest", manifestStart);
|
|
20103
19690
|
this.logger.debug(`Manifest: ${existingHashes.size} existing chunk hashes loaded`);
|
|
20104
19691
|
const sourceStart = stageStart();
|
|
@@ -20227,9 +19814,9 @@ var IndexPipeline = class _IndexPipeline {
|
|
|
20227
19814
|
}
|
|
20228
19815
|
stageEnd("links", linkStart);
|
|
20229
19816
|
this.logger.debug(`Link analysis: computed incoming links for ${incomingLinkCount.size} pages (${stageTimingsMs["links"]}ms)`);
|
|
20230
|
-
const
|
|
20231
|
-
this.logger.info("
|
|
20232
|
-
const
|
|
19817
|
+
const pagesStart = stageStart();
|
|
19818
|
+
this.logger.info("Building indexed pages...");
|
|
19819
|
+
const pages = [];
|
|
20233
19820
|
let routeExact = 0;
|
|
20234
19821
|
let routeBestEffort = 0;
|
|
20235
19822
|
const precomputedRoutes = /* @__PURE__ */ new Map();
|
|
@@ -20258,7 +19845,7 @@ var IndexPipeline = class _IndexPipeline {
|
|
|
20258
19845
|
} else {
|
|
20259
19846
|
routeExact += 1;
|
|
20260
19847
|
}
|
|
20261
|
-
const
|
|
19848
|
+
const indexedPage = {
|
|
20262
19849
|
url: page.url,
|
|
20263
19850
|
title: page.title,
|
|
20264
19851
|
scope: scope.scopeName,
|
|
@@ -20273,35 +19860,38 @@ var IndexPipeline = class _IndexPipeline {
|
|
|
20273
19860
|
description: page.description,
|
|
20274
19861
|
keywords: page.keywords
|
|
20275
19862
|
};
|
|
20276
|
-
|
|
20277
|
-
|
|
20278
|
-
await writeMirrorPage(statePath, scope, mirror);
|
|
20279
|
-
}
|
|
20280
|
-
this.logger.event("markdown_written", { url: page.url });
|
|
19863
|
+
pages.push(indexedPage);
|
|
19864
|
+
this.logger.event("page_indexed", { url: page.url });
|
|
20281
19865
|
}
|
|
20282
19866
|
if (!options.dryRun) {
|
|
20283
|
-
const pageRecords =
|
|
20284
|
-
|
|
20285
|
-
|
|
20286
|
-
|
|
20287
|
-
|
|
20288
|
-
|
|
20289
|
-
|
|
20290
|
-
|
|
20291
|
-
|
|
20292
|
-
|
|
20293
|
-
|
|
20294
|
-
|
|
20295
|
-
|
|
20296
|
-
|
|
20297
|
-
|
|
20298
|
-
|
|
19867
|
+
const pageRecords = pages.map((p) => {
|
|
19868
|
+
const summary = buildPageSummary(p);
|
|
19869
|
+
return {
|
|
19870
|
+
url: p.url,
|
|
19871
|
+
title: p.title,
|
|
19872
|
+
markdown: p.markdown,
|
|
19873
|
+
projectId: scope.projectId,
|
|
19874
|
+
scopeName: scope.scopeName,
|
|
19875
|
+
routeFile: p.routeFile,
|
|
19876
|
+
routeResolution: p.routeResolution,
|
|
19877
|
+
incomingLinks: p.incomingLinks,
|
|
19878
|
+
outgoingLinks: p.outgoingLinks,
|
|
19879
|
+
depth: p.depth,
|
|
19880
|
+
tags: p.tags,
|
|
19881
|
+
indexedAt: p.generatedAt,
|
|
19882
|
+
summary,
|
|
19883
|
+
description: p.description,
|
|
19884
|
+
keywords: p.keywords
|
|
19885
|
+
};
|
|
19886
|
+
});
|
|
19887
|
+
await this.store.deletePages(scope);
|
|
19888
|
+
await this.store.upsertPages(pageRecords, scope);
|
|
20299
19889
|
}
|
|
20300
|
-
stageEnd("
|
|
20301
|
-
this.logger.info(`
|
|
19890
|
+
stageEnd("pages", pagesStart);
|
|
19891
|
+
this.logger.info(`Indexed ${pages.length} page${pages.length === 1 ? "" : "s"} (${routeExact} exact, ${routeBestEffort} best-effort) (${stageTimingsMs["pages"]}ms)`);
|
|
20302
19892
|
const chunkStart = stageStart();
|
|
20303
19893
|
this.logger.info("Chunking pages...");
|
|
20304
|
-
let chunks =
|
|
19894
|
+
let chunks = pages.flatMap((page) => chunkPage(page, this.config, scope));
|
|
20305
19895
|
const maxChunks = typeof options.maxChunks === "number" ? Math.max(0, Math.floor(options.maxChunks)) : void 0;
|
|
20306
19896
|
if (typeof maxChunks === "number") {
|
|
20307
19897
|
chunks = chunks.slice(0, maxChunks);
|
|
@@ -20333,125 +19923,59 @@ var IndexPipeline = class _IndexPipeline {
|
|
|
20333
19923
|
});
|
|
20334
19924
|
const deletes = [...existingHashes.keys()].filter((chunkKey) => !currentChunkMap.has(chunkKey));
|
|
20335
19925
|
this.logger.info(`Changes detected: ${changedChunks.length} changed, ${deletes.length} deleted, ${chunks.length - changedChunks.length} unchanged`);
|
|
20336
|
-
const
|
|
20337
|
-
|
|
20338
|
-
for (const chunk of changedChunks) {
|
|
20339
|
-
chunkTokenEstimates.set(chunk.chunkKey, this.embeddings.estimateTokens(buildEmbeddingText(chunk, this.config.chunking.prependTitle)));
|
|
20340
|
-
}
|
|
20341
|
-
const estimatedTokens = changedChunks.reduce(
|
|
20342
|
-
(sum, chunk) => sum + (chunkTokenEstimates.get(chunk.chunkKey) ?? 0),
|
|
20343
|
-
0
|
|
20344
|
-
);
|
|
20345
|
-
const pricePer1k = this.config.embeddings.pricePer1kTokens ?? EMBEDDING_PRICE_PER_1K_TOKENS_USD[this.config.embeddings.model] ?? DEFAULT_EMBEDDING_PRICE_PER_1K;
|
|
20346
|
-
const estimatedCostUSD = estimatedTokens / 1e3 * pricePer1k;
|
|
20347
|
-
let newEmbeddings = 0;
|
|
20348
|
-
const vectorsByChunk = /* @__PURE__ */ new Map();
|
|
19926
|
+
const upsertStart = stageStart();
|
|
19927
|
+
let documentsUpserted = 0;
|
|
20349
19928
|
if (!options.dryRun && changedChunks.length > 0) {
|
|
20350
|
-
this.logger.info(`
|
|
20351
|
-
const
|
|
20352
|
-
|
|
20353
|
-
|
|
20354
|
-
|
|
20355
|
-
|
|
20356
|
-
|
|
20357
|
-
|
|
20358
|
-
|
|
20359
|
-
|
|
20360
|
-
);
|
|
20361
|
-
|
|
20362
|
-
for (let i = 0; i < changedChunks.length; i += 1) {
|
|
20363
|
-
const chunk = changedChunks[i];
|
|
20364
|
-
const embedding = embeddings[i];
|
|
20365
|
-
if (!chunk || !embedding || embedding.length === 0 || embedding.some((value) => !Number.isFinite(value))) {
|
|
20366
|
-
throw new SearchSocketError(
|
|
20367
|
-
"VECTOR_BACKEND_UNAVAILABLE",
|
|
20368
|
-
`Embedding provider returned an invalid vector for chunk index ${i}.`
|
|
20369
|
-
);
|
|
20370
|
-
}
|
|
20371
|
-
vectorsByChunk.set(chunk.chunkKey, embedding);
|
|
20372
|
-
newEmbeddings += 1;
|
|
20373
|
-
this.logger.event("embedded_new", { chunkKey: chunk.chunkKey });
|
|
20374
|
-
}
|
|
20375
|
-
}
|
|
20376
|
-
stageEnd("embedding", embedStart);
|
|
20377
|
-
if (changedChunks.length > 0) {
|
|
20378
|
-
this.logger.info(`Embedded ${newEmbeddings} chunk${newEmbeddings === 1 ? "" : "s"} (${stageTimingsMs["embedding"]}ms)`);
|
|
20379
|
-
} else {
|
|
20380
|
-
this.logger.info("No chunks to embed \u2014 all up to date");
|
|
20381
|
-
}
|
|
20382
|
-
const syncStart = stageStart();
|
|
20383
|
-
if (!options.dryRun) {
|
|
20384
|
-
this.logger.info("Syncing vectors...");
|
|
20385
|
-
const upserts = [];
|
|
20386
|
-
for (const chunk of changedChunks) {
|
|
20387
|
-
const vector = vectorsByChunk.get(chunk.chunkKey);
|
|
20388
|
-
if (!vector) {
|
|
20389
|
-
continue;
|
|
20390
|
-
}
|
|
20391
|
-
upserts.push({
|
|
19929
|
+
this.logger.info(`Upserting ${changedChunks.length} chunk${changedChunks.length === 1 ? "" : "s"} to Upstash Search...`);
|
|
19930
|
+
const UPSTASH_CONTENT_LIMIT = 4096;
|
|
19931
|
+
const docs = changedChunks.map((chunk) => {
|
|
19932
|
+
const title = chunk.title;
|
|
19933
|
+
const sectionTitle = chunk.sectionTitle ?? "";
|
|
19934
|
+
const url = chunk.url;
|
|
19935
|
+
const tags = chunk.tags.join(",");
|
|
19936
|
+
const headingPath = chunk.headingPath.join(" > ");
|
|
19937
|
+
const otherFieldsLen = title.length + sectionTitle.length + url.length + tags.length + headingPath.length;
|
|
19938
|
+
const textBudget = Math.max(500, UPSTASH_CONTENT_LIMIT - otherFieldsLen - 50);
|
|
19939
|
+
const text = buildEmbeddingText(chunk, this.config.chunking.prependTitle).slice(0, textBudget);
|
|
19940
|
+
return {
|
|
20392
19941
|
id: chunk.chunkKey,
|
|
20393
|
-
|
|
19942
|
+
content: { title, sectionTitle, text, url, tags, headingPath },
|
|
20394
19943
|
metadata: {
|
|
20395
19944
|
projectId: scope.projectId,
|
|
20396
19945
|
scopeName: scope.scopeName,
|
|
20397
|
-
url: chunk.url,
|
|
20398
19946
|
path: chunk.path,
|
|
20399
|
-
title: chunk.title,
|
|
20400
|
-
sectionTitle: chunk.sectionTitle ?? "",
|
|
20401
|
-
headingPath: chunk.headingPath,
|
|
20402
19947
|
snippet: chunk.snippet,
|
|
20403
|
-
chunkText: chunk.chunkText.slice(0, 4e3),
|
|
20404
19948
|
ordinal: chunk.ordinal,
|
|
20405
19949
|
contentHash: chunk.contentHash,
|
|
20406
|
-
modelId: this.config.embeddings.model,
|
|
20407
19950
|
depth: chunk.depth,
|
|
20408
19951
|
incomingLinks: chunk.incomingLinks,
|
|
20409
19952
|
routeFile: chunk.routeFile,
|
|
20410
|
-
|
|
20411
|
-
|
|
20412
|
-
keywords: chunk.keywords
|
|
19953
|
+
description: chunk.description ?? "",
|
|
19954
|
+
keywords: (chunk.keywords ?? []).join(",")
|
|
20413
19955
|
}
|
|
20414
|
-
}
|
|
20415
|
-
}
|
|
20416
|
-
if (upserts.length > 0) {
|
|
20417
|
-
await this.vectorStore.upsert(upserts, scope);
|
|
20418
|
-
this.logger.event("upserted", { count: upserts.length });
|
|
20419
|
-
}
|
|
20420
|
-
if (deletes.length > 0) {
|
|
20421
|
-
await this.vectorStore.deleteByIds(deletes, scope);
|
|
20422
|
-
this.logger.event("deleted", { count: deletes.length });
|
|
20423
|
-
}
|
|
20424
|
-
}
|
|
20425
|
-
stageEnd("sync", syncStart);
|
|
20426
|
-
this.logger.debug(`Sync complete (${stageTimingsMs["sync"]}ms)`);
|
|
20427
|
-
const finalizeStart = stageStart();
|
|
20428
|
-
if (!options.dryRun) {
|
|
20429
|
-
const scopeInfo = {
|
|
20430
|
-
projectId: scope.projectId,
|
|
20431
|
-
scopeName: scope.scopeName,
|
|
20432
|
-
modelId: this.config.embeddings.model,
|
|
20433
|
-
lastIndexedAt: nowIso(),
|
|
20434
|
-
vectorCount: chunks.length,
|
|
20435
|
-
lastEstimateTokens: estimatedTokens,
|
|
20436
|
-
lastEstimateCostUSD: Number(estimatedCostUSD.toFixed(8)),
|
|
20437
|
-
lastEstimateChangedChunks: changedChunks.length
|
|
20438
|
-
};
|
|
20439
|
-
await this.vectorStore.recordScope(scopeInfo);
|
|
20440
|
-
this.logger.event("registry_updated", {
|
|
20441
|
-
scope: scope.scopeName,
|
|
20442
|
-
vectorCount: chunks.length
|
|
19956
|
+
};
|
|
20443
19957
|
});
|
|
19958
|
+
await this.store.upsertChunks(docs, scope);
|
|
19959
|
+
documentsUpserted = docs.length;
|
|
19960
|
+
this.logger.event("upserted", { count: docs.length });
|
|
19961
|
+
}
|
|
19962
|
+
if (!options.dryRun && deletes.length > 0) {
|
|
19963
|
+
await this.store.deleteByIds(deletes, scope);
|
|
19964
|
+
this.logger.event("deleted", { count: deletes.length });
|
|
19965
|
+
}
|
|
19966
|
+
stageEnd("upsert", upsertStart);
|
|
19967
|
+
if (changedChunks.length > 0) {
|
|
19968
|
+
this.logger.info(`Upserted ${documentsUpserted} document${documentsUpserted === 1 ? "" : "s"} (${stageTimingsMs["upsert"]}ms)`);
|
|
19969
|
+
} else {
|
|
19970
|
+
this.logger.info("No chunks to upsert \u2014 all up to date");
|
|
20444
19971
|
}
|
|
20445
|
-
stageEnd("finalize", finalizeStart);
|
|
20446
19972
|
this.logger.info("Done.");
|
|
20447
19973
|
return {
|
|
20448
|
-
pagesProcessed:
|
|
19974
|
+
pagesProcessed: pages.length,
|
|
20449
19975
|
chunksTotal: chunks.length,
|
|
20450
19976
|
chunksChanged: changedChunks.length,
|
|
20451
|
-
|
|
19977
|
+
documentsUpserted,
|
|
20452
19978
|
deletes: deletes.length,
|
|
20453
|
-
estimatedTokens,
|
|
20454
|
-
estimatedCostUSD: Number(estimatedCostUSD.toFixed(8)),
|
|
20455
19979
|
routeExact,
|
|
20456
19980
|
routeBestEffort,
|
|
20457
19981
|
stageTimingsMs
|
|
@@ -20464,35 +19988,25 @@ var requestSchema = z.object({
|
|
|
20464
19988
|
scope: z.string().optional(),
|
|
20465
19989
|
pathPrefix: z.string().optional(),
|
|
20466
19990
|
tags: z.array(z.string()).optional(),
|
|
20467
|
-
|
|
20468
|
-
groupBy: z.enum(["page", "chunk"]).optional(),
|
|
20469
|
-
stream: z.boolean().optional()
|
|
19991
|
+
groupBy: z.enum(["page", "chunk"]).optional()
|
|
20470
19992
|
});
|
|
20471
19993
|
var SearchEngine = class _SearchEngine {
|
|
20472
19994
|
cwd;
|
|
20473
19995
|
config;
|
|
20474
|
-
|
|
20475
|
-
vectorStore;
|
|
20476
|
-
reranker;
|
|
19996
|
+
store;
|
|
20477
19997
|
constructor(options) {
|
|
20478
19998
|
this.cwd = options.cwd;
|
|
20479
19999
|
this.config = options.config;
|
|
20480
|
-
this.
|
|
20481
|
-
this.vectorStore = options.vectorStore;
|
|
20482
|
-
this.reranker = options.reranker;
|
|
20000
|
+
this.store = options.store;
|
|
20483
20001
|
}
|
|
20484
20002
|
static async create(options = {}) {
|
|
20485
20003
|
const cwd = path.resolve(options.cwd ?? process.cwd());
|
|
20486
20004
|
const config = options.config ?? await loadConfig({ cwd, configPath: options.configPath });
|
|
20487
|
-
const
|
|
20488
|
-
const vectorStore = options.vectorStore ?? await createVectorStore(config, cwd);
|
|
20489
|
-
const reranker = options.reranker === void 0 ? createReranker(config) : options.reranker;
|
|
20005
|
+
const store = options.store ?? await createUpstashStore(config);
|
|
20490
20006
|
return new _SearchEngine({
|
|
20491
20007
|
cwd,
|
|
20492
20008
|
config,
|
|
20493
|
-
|
|
20494
|
-
vectorStore,
|
|
20495
|
-
reranker
|
|
20009
|
+
store
|
|
20496
20010
|
});
|
|
20497
20011
|
}
|
|
20498
20012
|
getConfig() {
|
|
@@ -20506,142 +20020,90 @@ var SearchEngine = class _SearchEngine {
|
|
|
20506
20020
|
const input = parsed.data;
|
|
20507
20021
|
const totalStart = process.hrtime.bigint();
|
|
20508
20022
|
const resolvedScope = resolveScope(this.config, input.scope);
|
|
20509
|
-
await this.assertModelCompatibility(resolvedScope);
|
|
20510
20023
|
const topK = input.topK ?? 10;
|
|
20511
|
-
const wantsRerank = Boolean(input.rerank);
|
|
20512
20024
|
const groupByPage = (input.groupBy ?? "page") === "page";
|
|
20513
20025
|
const candidateK = groupByPage ? Math.max(topK * 10, 50) : Math.max(50, topK);
|
|
20514
|
-
const
|
|
20515
|
-
|
|
20516
|
-
|
|
20517
|
-
|
|
20518
|
-
|
|
20519
|
-
|
|
20520
|
-
|
|
20521
|
-
|
|
20522
|
-
|
|
20523
|
-
|
|
20524
|
-
|
|
20525
|
-
|
|
20526
|
-
|
|
20527
|
-
|
|
20528
|
-
|
|
20529
|
-
|
|
20530
|
-
|
|
20531
|
-
|
|
20532
|
-
|
|
20533
|
-
|
|
20534
|
-
|
|
20535
|
-
|
|
20536
|
-
|
|
20537
|
-
|
|
20538
|
-
|
|
20539
|
-
|
|
20540
|
-
|
|
20026
|
+
const filterParts = [];
|
|
20027
|
+
if (input.pathPrefix) {
|
|
20028
|
+
const prefix = input.pathPrefix.startsWith("/") ? input.pathPrefix : `/${input.pathPrefix}`;
|
|
20029
|
+
filterParts.push(`url GLOB '${prefix}*'`);
|
|
20030
|
+
}
|
|
20031
|
+
if (input.tags && input.tags.length > 0) {
|
|
20032
|
+
for (const tag of input.tags) {
|
|
20033
|
+
filterParts.push(`tags GLOB '*${tag}*'`);
|
|
20034
|
+
}
|
|
20035
|
+
}
|
|
20036
|
+
const filter = filterParts.length > 0 ? filterParts.join(" AND ") : void 0;
|
|
20037
|
+
const useDualSearch = this.config.search.dualSearch && groupByPage;
|
|
20038
|
+
const searchStart = process.hrtime.bigint();
|
|
20039
|
+
let ranked;
|
|
20040
|
+
if (useDualSearch) {
|
|
20041
|
+
const chunkLimit = Math.max(topK * 10, 100);
|
|
20042
|
+
const pageLimit = 20;
|
|
20043
|
+
const [pageHits, chunkHits] = await Promise.all([
|
|
20044
|
+
this.store.searchPages(
|
|
20045
|
+
input.q,
|
|
20046
|
+
{
|
|
20047
|
+
limit: pageLimit,
|
|
20048
|
+
semanticWeight: this.config.search.semanticWeight,
|
|
20049
|
+
inputEnrichment: this.config.search.inputEnrichment,
|
|
20050
|
+
filter
|
|
20051
|
+
},
|
|
20052
|
+
resolvedScope
|
|
20053
|
+
),
|
|
20054
|
+
this.store.search(
|
|
20055
|
+
input.q,
|
|
20056
|
+
{
|
|
20057
|
+
limit: chunkLimit,
|
|
20058
|
+
semanticWeight: this.config.search.semanticWeight,
|
|
20059
|
+
inputEnrichment: this.config.search.inputEnrichment,
|
|
20060
|
+
reranking: false,
|
|
20061
|
+
filter
|
|
20062
|
+
},
|
|
20063
|
+
resolvedScope
|
|
20064
|
+
)
|
|
20065
|
+
]);
|
|
20066
|
+
const rankedChunks = rankHits(chunkHits, this.config, input.q);
|
|
20067
|
+
ranked = mergePageAndChunkResults(pageHits, rankedChunks, this.config);
|
|
20068
|
+
} else {
|
|
20069
|
+
const hits = await this.store.search(
|
|
20070
|
+
input.q,
|
|
20071
|
+
{
|
|
20072
|
+
limit: candidateK,
|
|
20073
|
+
semanticWeight: this.config.search.semanticWeight,
|
|
20074
|
+
inputEnrichment: this.config.search.inputEnrichment,
|
|
20075
|
+
reranking: this.config.search.reranking,
|
|
20076
|
+
filter
|
|
20077
|
+
},
|
|
20078
|
+
resolvedScope
|
|
20079
|
+
);
|
|
20080
|
+
ranked = rankHits(hits, this.config, input.q);
|
|
20541
20081
|
}
|
|
20542
|
-
const
|
|
20082
|
+
const searchMs = hrTimeMs(searchStart);
|
|
20083
|
+
const results = this.buildResults(ranked, topK, groupByPage, input.q);
|
|
20543
20084
|
return {
|
|
20544
20085
|
q: input.q,
|
|
20545
20086
|
scope: resolvedScope.scopeName,
|
|
20546
20087
|
results,
|
|
20547
20088
|
meta: {
|
|
20548
20089
|
timingsMs: {
|
|
20549
|
-
|
|
20550
|
-
vector: Math.round(vectorMs),
|
|
20551
|
-
rerank: Math.round(rerankMs),
|
|
20090
|
+
search: Math.round(searchMs),
|
|
20552
20091
|
total: Math.round(hrTimeMs(totalStart))
|
|
20553
|
-
},
|
|
20554
|
-
usedRerank,
|
|
20555
|
-
modelId: this.config.embeddings.model
|
|
20556
|
-
}
|
|
20557
|
-
};
|
|
20558
|
-
}
|
|
20559
|
-
async *searchStreaming(request) {
|
|
20560
|
-
const parsed = requestSchema.safeParse(request);
|
|
20561
|
-
if (!parsed.success) {
|
|
20562
|
-
throw new SearchSocketError("INVALID_REQUEST", parsed.error.issues[0]?.message ?? "Invalid request", 400);
|
|
20563
|
-
}
|
|
20564
|
-
const input = parsed.data;
|
|
20565
|
-
const wantsRerank = Boolean(input.rerank);
|
|
20566
|
-
if (!wantsRerank) {
|
|
20567
|
-
const response = await this.search(request);
|
|
20568
|
-
yield { phase: "initial", data: response };
|
|
20569
|
-
return;
|
|
20570
|
-
}
|
|
20571
|
-
const totalStart = process.hrtime.bigint();
|
|
20572
|
-
const resolvedScope = resolveScope(this.config, input.scope);
|
|
20573
|
-
await this.assertModelCompatibility(resolvedScope);
|
|
20574
|
-
const topK = input.topK ?? 10;
|
|
20575
|
-
const groupByPage = (input.groupBy ?? "page") === "page";
|
|
20576
|
-
const candidateK = groupByPage ? Math.max(topK * 10, 50) : Math.max(50, topK);
|
|
20577
|
-
const embedStart = process.hrtime.bigint();
|
|
20578
|
-
const queryEmbeddings = await this.embeddings.embedTexts([input.q], this.config.embeddings.model, "retrieval.query");
|
|
20579
|
-
const queryVector = queryEmbeddings[0];
|
|
20580
|
-
if (!queryVector || queryVector.length === 0 || queryVector.some((value) => !Number.isFinite(value))) {
|
|
20581
|
-
throw new SearchSocketError("VECTOR_BACKEND_UNAVAILABLE", "Unable to create query embedding.");
|
|
20582
|
-
}
|
|
20583
|
-
const embedMs = hrTimeMs(embedStart);
|
|
20584
|
-
const vectorStart = process.hrtime.bigint();
|
|
20585
|
-
const hits = await this.vectorStore.query(
|
|
20586
|
-
queryVector,
|
|
20587
|
-
{
|
|
20588
|
-
topK: candidateK,
|
|
20589
|
-
pathPrefix: input.pathPrefix,
|
|
20590
|
-
tags: input.tags
|
|
20591
|
-
},
|
|
20592
|
-
resolvedScope
|
|
20593
|
-
);
|
|
20594
|
-
const vectorMs = hrTimeMs(vectorStart);
|
|
20595
|
-
const ranked = rankHits(hits, this.config);
|
|
20596
|
-
const initialResults = this.buildResults(ranked, topK, groupByPage);
|
|
20597
|
-
yield {
|
|
20598
|
-
phase: "initial",
|
|
20599
|
-
data: {
|
|
20600
|
-
q: input.q,
|
|
20601
|
-
scope: resolvedScope.scopeName,
|
|
20602
|
-
results: initialResults,
|
|
20603
|
-
meta: {
|
|
20604
|
-
timingsMs: {
|
|
20605
|
-
embed: Math.round(embedMs),
|
|
20606
|
-
vector: Math.round(vectorMs),
|
|
20607
|
-
rerank: 0,
|
|
20608
|
-
total: Math.round(hrTimeMs(totalStart))
|
|
20609
|
-
},
|
|
20610
|
-
usedRerank: false,
|
|
20611
|
-
modelId: this.config.embeddings.model
|
|
20612
|
-
}
|
|
20613
|
-
}
|
|
20614
|
-
};
|
|
20615
|
-
const rerankStart = process.hrtime.bigint();
|
|
20616
|
-
const reranked = await this.rerankHits(input.q, ranked, topK);
|
|
20617
|
-
const rerankMs = hrTimeMs(rerankStart);
|
|
20618
|
-
const rerankedResults = this.buildResults(reranked, topK, groupByPage);
|
|
20619
|
-
yield {
|
|
20620
|
-
phase: "reranked",
|
|
20621
|
-
data: {
|
|
20622
|
-
q: input.q,
|
|
20623
|
-
scope: resolvedScope.scopeName,
|
|
20624
|
-
results: rerankedResults,
|
|
20625
|
-
meta: {
|
|
20626
|
-
timingsMs: {
|
|
20627
|
-
embed: Math.round(embedMs),
|
|
20628
|
-
vector: Math.round(vectorMs),
|
|
20629
|
-
rerank: Math.round(rerankMs),
|
|
20630
|
-
total: Math.round(hrTimeMs(totalStart))
|
|
20631
|
-
},
|
|
20632
|
-
usedRerank: true,
|
|
20633
|
-
modelId: this.config.embeddings.model
|
|
20634
20092
|
}
|
|
20635
20093
|
}
|
|
20636
20094
|
};
|
|
20637
20095
|
}
|
|
20638
|
-
|
|
20639
|
-
const
|
|
20096
|
+
ensureSnippet(hit) {
|
|
20097
|
+
const snippet = hit.hit.metadata.snippet;
|
|
20098
|
+
if (snippet && snippet.length >= 30) return snippet;
|
|
20099
|
+
const chunkText = hit.hit.metadata.chunkText;
|
|
20100
|
+
if (chunkText) return toSnippet(chunkText);
|
|
20101
|
+
return snippet || "";
|
|
20102
|
+
}
|
|
20103
|
+
buildResults(ordered, topK, groupByPage, _query) {
|
|
20640
20104
|
if (groupByPage) {
|
|
20641
20105
|
let pages = aggregateByPage(ordered, this.config);
|
|
20642
|
-
|
|
20643
|
-
pages = pages.filter((p) => p.pageScore >= minScore);
|
|
20644
|
-
}
|
|
20106
|
+
pages = trimByScoreGap(pages, this.config);
|
|
20645
20107
|
const minRatio = this.config.ranking.minChunkScoreRatio;
|
|
20646
20108
|
return pages.slice(0, topK).map((page) => {
|
|
20647
20109
|
const bestScore = page.bestChunk.finalScore;
|
|
@@ -20651,12 +20113,12 @@ var SearchEngine = class _SearchEngine {
|
|
|
20651
20113
|
url: page.url,
|
|
20652
20114
|
title: page.title,
|
|
20653
20115
|
sectionTitle: page.bestChunk.hit.metadata.sectionTitle || void 0,
|
|
20654
|
-
snippet: page.bestChunk
|
|
20116
|
+
snippet: this.ensureSnippet(page.bestChunk),
|
|
20655
20117
|
score: Number(page.pageScore.toFixed(6)),
|
|
20656
20118
|
routeFile: page.routeFile,
|
|
20657
20119
|
chunks: meaningful.length > 1 ? meaningful.map((c) => ({
|
|
20658
20120
|
sectionTitle: c.hit.metadata.sectionTitle || void 0,
|
|
20659
|
-
snippet: c
|
|
20121
|
+
snippet: this.ensureSnippet(c),
|
|
20660
20122
|
headingPath: c.hit.metadata.headingPath,
|
|
20661
20123
|
score: Number(c.finalScore.toFixed(6))
|
|
20662
20124
|
})) : void 0
|
|
@@ -20664,6 +20126,7 @@ var SearchEngine = class _SearchEngine {
|
|
|
20664
20126
|
});
|
|
20665
20127
|
} else {
|
|
20666
20128
|
let filtered = ordered;
|
|
20129
|
+
const minScore = this.config.ranking.minScore;
|
|
20667
20130
|
if (minScore > 0) {
|
|
20668
20131
|
filtered = ordered.filter((entry) => entry.finalScore >= minScore);
|
|
20669
20132
|
}
|
|
@@ -20671,7 +20134,7 @@ var SearchEngine = class _SearchEngine {
|
|
|
20671
20134
|
url: hit.metadata.url,
|
|
20672
20135
|
title: hit.metadata.title,
|
|
20673
20136
|
sectionTitle: hit.metadata.sectionTitle || void 0,
|
|
20674
|
-
snippet: hit
|
|
20137
|
+
snippet: this.ensureSnippet({ hit, finalScore }),
|
|
20675
20138
|
score: Number(finalScore.toFixed(6)),
|
|
20676
20139
|
routeFile: hit.metadata.routeFile
|
|
20677
20140
|
}));
|
|
@@ -20680,7 +20143,7 @@ var SearchEngine = class _SearchEngine {
|
|
|
20680
20143
|
async getPage(pathOrUrl, scope) {
|
|
20681
20144
|
const resolvedScope = resolveScope(this.config, scope);
|
|
20682
20145
|
const urlPath = this.resolveInputPath(pathOrUrl);
|
|
20683
|
-
const page = await this.
|
|
20146
|
+
const page = await this.store.getPage(urlPath, resolvedScope);
|
|
20684
20147
|
if (!page) {
|
|
20685
20148
|
throw new SearchSocketError("INVALID_REQUEST", `Indexed page not found for ${urlPath}`, 404);
|
|
20686
20149
|
}
|
|
@@ -20701,7 +20164,7 @@ var SearchEngine = class _SearchEngine {
|
|
|
20701
20164
|
};
|
|
20702
20165
|
}
|
|
20703
20166
|
async health() {
|
|
20704
|
-
return this.
|
|
20167
|
+
return this.store.health();
|
|
20705
20168
|
}
|
|
20706
20169
|
resolveInputPath(pathOrUrl) {
|
|
20707
20170
|
try {
|
|
@@ -20713,92 +20176,8 @@ var SearchEngine = class _SearchEngine {
|
|
|
20713
20176
|
const withoutQueryOrHash = pathOrUrl.split(/[?#]/)[0] ?? pathOrUrl;
|
|
20714
20177
|
return normalizeUrlPath(withoutQueryOrHash);
|
|
20715
20178
|
}
|
|
20716
|
-
async assertModelCompatibility(scope) {
|
|
20717
|
-
const modelId = await this.vectorStore.getScopeModelId(scope);
|
|
20718
|
-
if (modelId && modelId !== this.config.embeddings.model) {
|
|
20719
|
-
throw new SearchSocketError(
|
|
20720
|
-
"EMBEDDING_MODEL_MISMATCH",
|
|
20721
|
-
`Scope ${scope.scopeName} was indexed with ${modelId}. Current config uses ${this.config.embeddings.model}. Re-index with --force.`
|
|
20722
|
-
);
|
|
20723
|
-
}
|
|
20724
|
-
}
|
|
20725
|
-
async rerankHits(query, ranked, topK) {
|
|
20726
|
-
if (!this.config.rerank.enabled) {
|
|
20727
|
-
throw new SearchSocketError(
|
|
20728
|
-
"INVALID_REQUEST",
|
|
20729
|
-
"rerank=true requested but rerank.enabled is not set to true.",
|
|
20730
|
-
400
|
|
20731
|
-
);
|
|
20732
|
-
}
|
|
20733
|
-
if (!this.reranker) {
|
|
20734
|
-
throw new SearchSocketError(
|
|
20735
|
-
"CONFIG_MISSING",
|
|
20736
|
-
`rerank=true requested but ${this.config.embeddings.apiKeyEnv} is not set.`,
|
|
20737
|
-
400
|
|
20738
|
-
);
|
|
20739
|
-
}
|
|
20740
|
-
const pageGroups = /* @__PURE__ */ new Map();
|
|
20741
|
-
for (const entry of ranked) {
|
|
20742
|
-
const url = entry.hit.metadata.url;
|
|
20743
|
-
const group = pageGroups.get(url);
|
|
20744
|
-
if (group) group.push(entry);
|
|
20745
|
-
else pageGroups.set(url, [entry]);
|
|
20746
|
-
}
|
|
20747
|
-
const MAX_CHUNKS_PER_PAGE = 5;
|
|
20748
|
-
const MIN_CHUNKS_PER_PAGE = 1;
|
|
20749
|
-
const MIN_CHUNK_SCORE_RATIO = 0.5;
|
|
20750
|
-
const MAX_DOC_CHARS = 2e3;
|
|
20751
|
-
const pageCandidates = [];
|
|
20752
|
-
for (const [url, chunks] of pageGroups) {
|
|
20753
|
-
const byScore = [...chunks].sort((a, b) => b.finalScore - a.finalScore);
|
|
20754
|
-
const bestScore = byScore[0].finalScore;
|
|
20755
|
-
const scoreFloor = Number.isFinite(bestScore) ? bestScore * MIN_CHUNK_SCORE_RATIO : Number.NEGATIVE_INFINITY;
|
|
20756
|
-
const selected = byScore.filter(
|
|
20757
|
-
(c, i) => i < MIN_CHUNKS_PER_PAGE || c.finalScore >= scoreFloor
|
|
20758
|
-
).slice(0, MAX_CHUNKS_PER_PAGE);
|
|
20759
|
-
selected.sort((a, b) => (a.hit.metadata.ordinal ?? 0) - (b.hit.metadata.ordinal ?? 0));
|
|
20760
|
-
const first = selected[0].hit.metadata;
|
|
20761
|
-
const parts = [first.title];
|
|
20762
|
-
if (first.description) {
|
|
20763
|
-
parts.push(first.description);
|
|
20764
|
-
}
|
|
20765
|
-
if (first.keywords && first.keywords.length > 0) {
|
|
20766
|
-
parts.push(first.keywords.join(", "));
|
|
20767
|
-
}
|
|
20768
|
-
const body = selected.map((c) => c.hit.metadata.chunkText || c.hit.metadata.snippet).join("\n\n");
|
|
20769
|
-
parts.push(body);
|
|
20770
|
-
let text = parts.join("\n\n");
|
|
20771
|
-
if (text.length > MAX_DOC_CHARS) {
|
|
20772
|
-
text = text.slice(0, MAX_DOC_CHARS);
|
|
20773
|
-
}
|
|
20774
|
-
pageCandidates.push({ id: url, text });
|
|
20775
|
-
}
|
|
20776
|
-
const maxCandidates = Math.max(topK, this.config.rerank.topN);
|
|
20777
|
-
const cappedCandidates = pageCandidates.slice(0, maxCandidates);
|
|
20778
|
-
const reranked = await this.reranker.rerank(
|
|
20779
|
-
query,
|
|
20780
|
-
cappedCandidates,
|
|
20781
|
-
maxCandidates
|
|
20782
|
-
);
|
|
20783
|
-
const scoreByUrl = new Map(reranked.map((e) => [e.id, e.score]));
|
|
20784
|
-
return ranked.map((entry) => {
|
|
20785
|
-
const pageScore = scoreByUrl.get(entry.hit.metadata.url);
|
|
20786
|
-
const base = Number.isFinite(entry.finalScore) ? entry.finalScore : Number.NEGATIVE_INFINITY;
|
|
20787
|
-
if (pageScore === void 0 || !Number.isFinite(pageScore)) {
|
|
20788
|
-
return { ...entry, finalScore: base };
|
|
20789
|
-
}
|
|
20790
|
-
const combined = pageScore * this.config.ranking.weights.rerank + base * 1e-3;
|
|
20791
|
-
return {
|
|
20792
|
-
...entry,
|
|
20793
|
-
finalScore: Number.isFinite(combined) ? combined : base
|
|
20794
|
-
};
|
|
20795
|
-
}).sort((a, b) => {
|
|
20796
|
-
const delta = b.finalScore - a.finalScore;
|
|
20797
|
-
return Number.isNaN(delta) ? 0 : delta;
|
|
20798
|
-
});
|
|
20799
|
-
}
|
|
20800
20179
|
};
|
|
20801
|
-
function createServer(engine
|
|
20180
|
+
function createServer(engine) {
|
|
20802
20181
|
const server = new McpServer({
|
|
20803
20182
|
name: "searchsocket-mcp",
|
|
20804
20183
|
version: "0.1.0"
|
|
@@ -20806,15 +20185,14 @@ function createServer(engine, config) {
|
|
|
20806
20185
|
server.registerTool(
|
|
20807
20186
|
"search",
|
|
20808
20187
|
{
|
|
20809
|
-
description: "Semantic site search. Returns url/title/snippet/score/routeFile for each match. Supports optional scope, pathPrefix, tags, topK, and
|
|
20188
|
+
description: "Semantic site search powered by Upstash Search. Returns url/title/snippet/score/routeFile for each match. Supports optional scope, pathPrefix, tags, topK, and groupBy.",
|
|
20810
20189
|
inputSchema: {
|
|
20811
20190
|
query: z.string().min(1),
|
|
20812
20191
|
scope: z.string().optional(),
|
|
20813
20192
|
topK: z.number().int().positive().max(100).optional(),
|
|
20814
20193
|
pathPrefix: z.string().optional(),
|
|
20815
20194
|
tags: z.array(z.string()).optional(),
|
|
20816
|
-
groupBy: z.enum(["page", "chunk"]).optional()
|
|
20817
|
-
rerank: z.boolean().optional().describe("Enable reranking for better relevance (uses Jina Reranker). Defaults to true when rerank is enabled in config.")
|
|
20195
|
+
groupBy: z.enum(["page", "chunk"]).optional()
|
|
20818
20196
|
}
|
|
20819
20197
|
},
|
|
20820
20198
|
async (input) => {
|
|
@@ -20824,8 +20202,7 @@ function createServer(engine, config) {
|
|
|
20824
20202
|
scope: input.scope,
|
|
20825
20203
|
pathPrefix: input.pathPrefix,
|
|
20826
20204
|
tags: input.tags,
|
|
20827
|
-
groupBy: input.groupBy
|
|
20828
|
-
rerank: input.rerank ?? config.rerank.enabled
|
|
20205
|
+
groupBy: input.groupBy
|
|
20829
20206
|
});
|
|
20830
20207
|
return {
|
|
20831
20208
|
content: [
|
|
@@ -20951,10 +20328,10 @@ async function runMcpServer(options = {}) {
|
|
|
20951
20328
|
config
|
|
20952
20329
|
});
|
|
20953
20330
|
if (resolvedTransport === "http") {
|
|
20954
|
-
await startHttpServer(() => createServer(engine
|
|
20331
|
+
await startHttpServer(() => createServer(engine), config, options);
|
|
20955
20332
|
return;
|
|
20956
20333
|
}
|
|
20957
|
-
const server = createServer(engine
|
|
20334
|
+
const server = createServer(engine);
|
|
20958
20335
|
const stdioTransport = new StdioServerTransport();
|
|
20959
20336
|
await server.connect(stdioTransport);
|
|
20960
20337
|
}
|
|
@@ -21111,42 +20488,6 @@ function searchsocketHandle(options = {}) {
|
|
|
21111
20488
|
}
|
|
21112
20489
|
const engine = await getEngine();
|
|
21113
20490
|
const searchRequest = body;
|
|
21114
|
-
if (searchRequest.stream && searchRequest.rerank) {
|
|
21115
|
-
const encoder = new TextEncoder();
|
|
21116
|
-
const stream = new ReadableStream({
|
|
21117
|
-
async start(controller) {
|
|
21118
|
-
try {
|
|
21119
|
-
for await (const event2 of engine.searchStreaming(searchRequest)) {
|
|
21120
|
-
const line = JSON.stringify(event2) + "\n";
|
|
21121
|
-
controller.enqueue(encoder.encode(line));
|
|
21122
|
-
}
|
|
21123
|
-
} catch (streamError) {
|
|
21124
|
-
const errorEvent = {
|
|
21125
|
-
phase: "error",
|
|
21126
|
-
data: {
|
|
21127
|
-
error: {
|
|
21128
|
-
code: streamError instanceof SearchSocketError ? streamError.code : "INTERNAL_ERROR",
|
|
21129
|
-
message: streamError instanceof Error ? streamError.message : "Unknown error"
|
|
21130
|
-
}
|
|
21131
|
-
}
|
|
21132
|
-
};
|
|
21133
|
-
controller.enqueue(encoder.encode(JSON.stringify(errorEvent) + "\n"));
|
|
21134
|
-
} finally {
|
|
21135
|
-
controller.close();
|
|
21136
|
-
}
|
|
21137
|
-
}
|
|
21138
|
-
});
|
|
21139
|
-
return withCors(
|
|
21140
|
-
new Response(stream, {
|
|
21141
|
-
status: 200,
|
|
21142
|
-
headers: {
|
|
21143
|
-
"content-type": "application/x-ndjson"
|
|
21144
|
-
}
|
|
21145
|
-
}),
|
|
21146
|
-
event.request,
|
|
21147
|
-
config
|
|
21148
|
-
);
|
|
21149
|
-
}
|
|
21150
20491
|
const result = await engine.search(searchRequest);
|
|
21151
20492
|
return withCors(
|
|
21152
20493
|
new Response(JSON.stringify(result), {
|
|
@@ -21232,13 +20573,6 @@ function searchsocketVitePlugin(options = {}) {
|
|
|
21232
20573
|
let running = false;
|
|
21233
20574
|
return {
|
|
21234
20575
|
name: "searchsocket:auto-index",
|
|
21235
|
-
config() {
|
|
21236
|
-
return {
|
|
21237
|
-
ssr: {
|
|
21238
|
-
external: ["@libsql/client", "libsql"]
|
|
21239
|
-
}
|
|
21240
|
-
};
|
|
21241
|
-
},
|
|
21242
20576
|
async closeBundle() {
|
|
21243
20577
|
if (executed || running) {
|
|
21244
20578
|
return;
|
|
@@ -21266,9 +20600,8 @@ function searchsocketVitePlugin(options = {}) {
|
|
|
21266
20600
|
verbose: options.verbose
|
|
21267
20601
|
});
|
|
21268
20602
|
logger3.info(
|
|
21269
|
-
`[searchsocket] indexed pages=${stats.pagesProcessed} chunks=${stats.chunksTotal} changed=${stats.chunksChanged}
|
|
20603
|
+
`[searchsocket] indexed pages=${stats.pagesProcessed} chunks=${stats.chunksTotal} changed=${stats.chunksChanged} upserted=${stats.documentsUpserted}`
|
|
21270
20604
|
);
|
|
21271
|
-
logger3.info("[searchsocket] markdown mirror written under .searchsocket/pages/<scope> (safe to commit for content workflows).");
|
|
21272
20605
|
executed = true;
|
|
21273
20606
|
} finally {
|
|
21274
20607
|
running = false;
|
|
@@ -21277,60 +20610,6 @@ function searchsocketVitePlugin(options = {}) {
|
|
|
21277
20610
|
};
|
|
21278
20611
|
}
|
|
21279
20612
|
|
|
21280
|
-
// src/merge.ts
|
|
21281
|
-
function mergeSearchResults(initial, reranked, options) {
|
|
21282
|
-
const maxDisplacement = options?.maxDisplacement ?? 3;
|
|
21283
|
-
const initialUrls = initial.results.map((r) => r.url);
|
|
21284
|
-
const rerankedUrls = reranked.results.map((r) => r.url);
|
|
21285
|
-
const initialPos = /* @__PURE__ */ new Map();
|
|
21286
|
-
for (let i = 0; i < initialUrls.length; i++) {
|
|
21287
|
-
initialPos.set(initialUrls[i], i);
|
|
21288
|
-
}
|
|
21289
|
-
const rerankedPos = /* @__PURE__ */ new Map();
|
|
21290
|
-
for (let i = 0; i < rerankedUrls.length; i++) {
|
|
21291
|
-
rerankedPos.set(rerankedUrls[i], i);
|
|
21292
|
-
}
|
|
21293
|
-
const displacements = [];
|
|
21294
|
-
for (const url of initialUrls) {
|
|
21295
|
-
const iPos = initialPos.get(url);
|
|
21296
|
-
const rPos = rerankedPos.get(url);
|
|
21297
|
-
const displacement = rPos !== void 0 ? Math.abs(iPos - rPos) : 0;
|
|
21298
|
-
displacements.push({ url, displacement });
|
|
21299
|
-
}
|
|
21300
|
-
const totalResults = displacements.length;
|
|
21301
|
-
if (totalResults === 0) {
|
|
21302
|
-
return {
|
|
21303
|
-
response: reranked,
|
|
21304
|
-
usedRerankedOrder: true,
|
|
21305
|
-
displacements
|
|
21306
|
-
};
|
|
21307
|
-
}
|
|
21308
|
-
const hasLargeDisplacement = displacements.some((d) => d.displacement > maxDisplacement);
|
|
21309
|
-
if (hasLargeDisplacement) {
|
|
21310
|
-
return {
|
|
21311
|
-
response: reranked,
|
|
21312
|
-
usedRerankedOrder: true,
|
|
21313
|
-
displacements
|
|
21314
|
-
};
|
|
21315
|
-
}
|
|
21316
|
-
const rerankedScoreMap = /* @__PURE__ */ new Map();
|
|
21317
|
-
for (const result of reranked.results) {
|
|
21318
|
-
rerankedScoreMap.set(result.url, result.score);
|
|
21319
|
-
}
|
|
21320
|
-
const mergedResults = initial.results.map((result) => ({
|
|
21321
|
-
...result,
|
|
21322
|
-
score: rerankedScoreMap.get(result.url) ?? result.score
|
|
21323
|
-
}));
|
|
21324
|
-
return {
|
|
21325
|
-
response: {
|
|
21326
|
-
...reranked,
|
|
21327
|
-
results: mergedResults
|
|
21328
|
-
},
|
|
21329
|
-
usedRerankedOrder: false,
|
|
21330
|
-
displacements
|
|
21331
|
-
};
|
|
21332
|
-
}
|
|
21333
|
-
|
|
21334
20613
|
// src/client.ts
|
|
21335
20614
|
function createSearchClient(options = {}) {
|
|
21336
20615
|
const endpoint = options.endpoint ?? "/api/search";
|
|
@@ -21358,72 +20637,6 @@ function createSearchClient(options = {}) {
|
|
|
21358
20637
|
throw new Error(message);
|
|
21359
20638
|
}
|
|
21360
20639
|
return payload;
|
|
21361
|
-
},
|
|
21362
|
-
async streamSearch(request, onPhase) {
|
|
21363
|
-
const response = await fetchImpl(endpoint, {
|
|
21364
|
-
method: "POST",
|
|
21365
|
-
headers: {
|
|
21366
|
-
"content-type": "application/json"
|
|
21367
|
-
},
|
|
21368
|
-
body: JSON.stringify(request)
|
|
21369
|
-
});
|
|
21370
|
-
if (!response.ok) {
|
|
21371
|
-
let payload;
|
|
21372
|
-
try {
|
|
21373
|
-
payload = await response.json();
|
|
21374
|
-
} catch {
|
|
21375
|
-
throw new Error("Search failed");
|
|
21376
|
-
}
|
|
21377
|
-
const message = payload.error?.message ?? "Search failed";
|
|
21378
|
-
throw new Error(message);
|
|
21379
|
-
}
|
|
21380
|
-
const contentType = response.headers.get("content-type") ?? "";
|
|
21381
|
-
if (contentType.includes("application/json")) {
|
|
21382
|
-
const data = await response.json();
|
|
21383
|
-
onPhase({ phase: "initial", data });
|
|
21384
|
-
return data;
|
|
21385
|
-
}
|
|
21386
|
-
if (!response.body) {
|
|
21387
|
-
throw new Error("Response body is not readable");
|
|
21388
|
-
}
|
|
21389
|
-
const reader = response.body.getReader();
|
|
21390
|
-
const decoder = new TextDecoder();
|
|
21391
|
-
let buffer = "";
|
|
21392
|
-
let lastResponse = null;
|
|
21393
|
-
for (; ; ) {
|
|
21394
|
-
const { done, value } = await reader.read();
|
|
21395
|
-
if (done) break;
|
|
21396
|
-
buffer += decoder.decode(value, { stream: true });
|
|
21397
|
-
let newlineIdx;
|
|
21398
|
-
while ((newlineIdx = buffer.indexOf("\n")) !== -1) {
|
|
21399
|
-
const line = buffer.slice(0, newlineIdx).trim();
|
|
21400
|
-
buffer = buffer.slice(newlineIdx + 1);
|
|
21401
|
-
if (line.length === 0) continue;
|
|
21402
|
-
const event = JSON.parse(line);
|
|
21403
|
-
if (event.phase === "error") {
|
|
21404
|
-
const errData = event.data;
|
|
21405
|
-
throw new Error(errData.error.message ?? "Streaming search error");
|
|
21406
|
-
}
|
|
21407
|
-
const searchEvent = event;
|
|
21408
|
-
onPhase(searchEvent);
|
|
21409
|
-
lastResponse = searchEvent.data;
|
|
21410
|
-
}
|
|
21411
|
-
}
|
|
21412
|
-
const remaining = buffer.trim();
|
|
21413
|
-
if (remaining.length > 0) {
|
|
21414
|
-
const event = JSON.parse(remaining);
|
|
21415
|
-
if (event.phase === "error") {
|
|
21416
|
-
const errData = event.data;
|
|
21417
|
-
throw new Error(errData.error.message ?? "Streaming search error");
|
|
21418
|
-
}
|
|
21419
|
-
const searchEvent = event;
|
|
21420
|
-
onPhase(searchEvent);
|
|
21421
|
-
lastResponse = searchEvent.data;
|
|
21422
|
-
}
|
|
21423
|
-
if (!lastResponse) {
|
|
21424
|
-
throw new Error("No search results received");
|
|
21425
|
-
}
|
|
21426
|
-
return lastResponse;
|
|
21427
20640
|
}
|
|
21428
20641
|
};
|
|
21429
20642
|
}
|
|
@@ -21439,6 +20652,6 @@ function createSearchClient(options = {}) {
|
|
|
21439
20652
|
*)
|
|
21440
20653
|
*/
|
|
21441
20654
|
|
|
21442
|
-
export { IndexPipeline,
|
|
20655
|
+
export { IndexPipeline, SearchEngine, UpstashSearchStore, createSearchClient, createUpstashStore, isServerless, loadConfig, mergeConfig, mergeConfigServerless, resolveScope, runMcpServer, searchsocketHandle, searchsocketVitePlugin };
|
|
21443
20656
|
//# sourceMappingURL=index.js.map
|
|
21444
20657
|
//# sourceMappingURL=index.js.map
|