@zvec/zvec-grep 0.1.0
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/LICENSE +201 -0
- package/README.md +222 -0
- package/README_CN.md +222 -0
- package/dist/cli/args.d.ts +2 -0
- package/dist/cli/args.js +611 -0
- package/dist/cli/args.js.map +1 -0
- package/dist/cli/commands.d.ts +2 -0
- package/dist/cli/commands.js +270 -0
- package/dist/cli/commands.js.map +1 -0
- package/dist/cli/errors.d.ts +6 -0
- package/dist/cli/errors.js +97 -0
- package/dist/cli/errors.js.map +1 -0
- package/dist/cli/format/context.d.ts +5 -0
- package/dist/cli/format/context.js +490 -0
- package/dist/cli/format/context.js.map +1 -0
- package/dist/cli/format/debug.d.ts +4 -0
- package/dist/cli/format/debug.js +42 -0
- package/dist/cli/format/debug.js.map +1 -0
- package/dist/cli/format/highlight.d.ts +3 -0
- package/dist/cli/format/highlight.js +106 -0
- package/dist/cli/format/highlight.js.map +1 -0
- package/dist/cli/format/progress.d.ts +9 -0
- package/dist/cli/format/progress.js +92 -0
- package/dist/cli/format/progress.js.map +1 -0
- package/dist/cli/format/range.d.ts +3 -0
- package/dist/cli/format/range.js +24 -0
- package/dist/cli/format/range.js.map +1 -0
- package/dist/cli/format/status.d.ts +7 -0
- package/dist/cli/format/status.js +200 -0
- package/dist/cli/format/status.js.map +1 -0
- package/dist/cli/help.d.ts +1 -0
- package/dist/cli/help.js +88 -0
- package/dist/cli/help.js.map +1 -0
- package/dist/cli/index.d.ts +2 -0
- package/dist/cli/index.js +32 -0
- package/dist/cli/index.js.map +1 -0
- package/dist/cli/types.d.ts +56 -0
- package/dist/cli/types.js +10 -0
- package/dist/cli/types.js.map +1 -0
- package/dist/cli/version.d.ts +1 -0
- package/dist/cli/version.js +11 -0
- package/dist/cli/version.js.map +1 -0
- package/dist/engine/collection/index.d.ts +50 -0
- package/dist/engine/collection/index.js +388 -0
- package/dist/engine/collection/index.js.map +1 -0
- package/dist/engine/errors/details.d.ts +6 -0
- package/dist/engine/errors/details.js +27 -0
- package/dist/engine/errors/details.js.map +1 -0
- package/dist/engine/errors/index.d.ts +14 -0
- package/dist/engine/errors/index.js +16 -0
- package/dist/engine/errors/index.js.map +1 -0
- package/dist/engine/models/catalog.d.ts +47 -0
- package/dist/engine/models/catalog.js +68 -0
- package/dist/engine/models/catalog.js.map +1 -0
- package/dist/engine/models/embeddings.d.ts +25 -0
- package/dist/engine/models/embeddings.js +100 -0
- package/dist/engine/models/embeddings.js.map +1 -0
- package/dist/engine/models/factory.d.ts +5 -0
- package/dist/engine/models/factory.js +66 -0
- package/dist/engine/models/factory.js.map +1 -0
- package/dist/engine/models/index.d.ts +6 -0
- package/dist/engine/models/index.js +5 -0
- package/dist/engine/models/index.js.map +1 -0
- package/dist/engine/models/providers/index.d.ts +2 -0
- package/dist/engine/models/providers/index.js +3 -0
- package/dist/engine/models/providers/index.js.map +1 -0
- package/dist/engine/models/providers/llama-cpp/embedding.d.ts +94 -0
- package/dist/engine/models/providers/llama-cpp/embedding.js +459 -0
- package/dist/engine/models/providers/llama-cpp/embedding.js.map +1 -0
- package/dist/engine/models/providers/llama-cpp/index.d.ts +1 -0
- package/dist/engine/models/providers/llama-cpp/index.js +2 -0
- package/dist/engine/models/providers/llama-cpp/index.js.map +1 -0
- package/dist/engine/models/providers/qwen/embedding.d.ts +38 -0
- package/dist/engine/models/providers/qwen/embedding.js +322 -0
- package/dist/engine/models/providers/qwen/embedding.js.map +1 -0
- package/dist/engine/models/providers/qwen/index.d.ts +1 -0
- package/dist/engine/models/providers/qwen/index.js +2 -0
- package/dist/engine/models/providers/qwen/index.js.map +1 -0
- package/dist/engine/models/ranking.d.ts +15 -0
- package/dist/engine/models/ranking.js +2 -0
- package/dist/engine/models/ranking.js.map +1 -0
- package/dist/engine/models/types.d.ts +30 -0
- package/dist/engine/models/types.js +2 -0
- package/dist/engine/models/types.js.map +1 -0
- package/dist/engine/pipeline/indexing/extractor/code/adapter.d.ts +19 -0
- package/dist/engine/pipeline/indexing/extractor/code/adapter.js +24 -0
- package/dist/engine/pipeline/indexing/extractor/code/adapter.js.map +1 -0
- package/dist/engine/pipeline/indexing/extractor/code/extractor.d.ts +14 -0
- package/dist/engine/pipeline/indexing/extractor/code/extractor.js +594 -0
- package/dist/engine/pipeline/indexing/extractor/code/extractor.js.map +1 -0
- package/dist/engine/pipeline/indexing/extractor/code/families/c-family.d.ts +2 -0
- package/dist/engine/pipeline/indexing/extractor/code/families/c-family.js +116 -0
- package/dist/engine/pipeline/indexing/extractor/code/families/c-family.js.map +1 -0
- package/dist/engine/pipeline/indexing/extractor/code/families/js-ts.d.ts +12 -0
- package/dist/engine/pipeline/indexing/extractor/code/families/js-ts.js +112 -0
- package/dist/engine/pipeline/indexing/extractor/code/families/js-ts.js.map +1 -0
- package/dist/engine/pipeline/indexing/extractor/code/families/metadata.d.ts +8 -0
- package/dist/engine/pipeline/indexing/extractor/code/families/metadata.js +83 -0
- package/dist/engine/pipeline/indexing/extractor/code/families/metadata.js.map +1 -0
- package/dist/engine/pipeline/indexing/extractor/code/families/name-field.d.ts +3 -0
- package/dist/engine/pipeline/indexing/extractor/code/families/name-field.js +22 -0
- package/dist/engine/pipeline/indexing/extractor/code/families/name-field.js.map +1 -0
- package/dist/engine/pipeline/indexing/extractor/code/languages/c.d.ts +1 -0
- package/dist/engine/pipeline/indexing/extractor/code/languages/c.js +15 -0
- package/dist/engine/pipeline/indexing/extractor/code/languages/c.js.map +1 -0
- package/dist/engine/pipeline/indexing/extractor/code/languages/cpp.d.ts +1 -0
- package/dist/engine/pipeline/indexing/extractor/code/languages/cpp.js +18 -0
- package/dist/engine/pipeline/indexing/extractor/code/languages/cpp.js.map +1 -0
- package/dist/engine/pipeline/indexing/extractor/code/languages/go.d.ts +2 -0
- package/dist/engine/pipeline/indexing/extractor/code/languages/go.js +72 -0
- package/dist/engine/pipeline/indexing/extractor/code/languages/go.js.map +1 -0
- package/dist/engine/pipeline/indexing/extractor/code/languages/java.d.ts +1 -0
- package/dist/engine/pipeline/indexing/extractor/code/languages/java.js +17 -0
- package/dist/engine/pipeline/indexing/extractor/code/languages/java.js.map +1 -0
- package/dist/engine/pipeline/indexing/extractor/code/languages/javascript.d.ts +2 -0
- package/dist/engine/pipeline/indexing/extractor/code/languages/javascript.js +25 -0
- package/dist/engine/pipeline/indexing/extractor/code/languages/javascript.js.map +1 -0
- package/dist/engine/pipeline/indexing/extractor/code/languages/python.d.ts +2 -0
- package/dist/engine/pipeline/indexing/extractor/code/languages/python.js +53 -0
- package/dist/engine/pipeline/indexing/extractor/code/languages/python.js.map +1 -0
- package/dist/engine/pipeline/indexing/extractor/code/languages/rust.d.ts +2 -0
- package/dist/engine/pipeline/indexing/extractor/code/languages/rust.js +29 -0
- package/dist/engine/pipeline/indexing/extractor/code/languages/rust.js.map +1 -0
- package/dist/engine/pipeline/indexing/extractor/code/languages/typescript.d.ts +2 -0
- package/dist/engine/pipeline/indexing/extractor/code/languages/typescript.js +37 -0
- package/dist/engine/pipeline/indexing/extractor/code/languages/typescript.js.map +1 -0
- package/dist/engine/pipeline/indexing/extractor/index.d.ts +44 -0
- package/dist/engine/pipeline/indexing/extractor/index.js +281 -0
- package/dist/engine/pipeline/indexing/extractor/index.js.map +1 -0
- package/dist/engine/pipeline/indexing/extractor/markdown/extractor.d.ts +13 -0
- package/dist/engine/pipeline/indexing/extractor/markdown/extractor.js +339 -0
- package/dist/engine/pipeline/indexing/extractor/markdown/extractor.js.map +1 -0
- package/dist/engine/pipeline/indexing/extractor/tree-sitter/grammar.d.ts +4 -0
- package/dist/engine/pipeline/indexing/extractor/tree-sitter/grammar.js +49 -0
- package/dist/engine/pipeline/indexing/extractor/tree-sitter/grammar.js.map +1 -0
- package/dist/engine/pipeline/indexing/extractor/tree-sitter/nodes.d.ts +3 -0
- package/dist/engine/pipeline/indexing/extractor/tree-sitter/nodes.js +26 -0
- package/dist/engine/pipeline/indexing/extractor/tree-sitter/nodes.js.map +1 -0
- package/dist/engine/pipeline/indexing/extractor/tree-sitter/parser.d.ts +2 -0
- package/dist/engine/pipeline/indexing/extractor/tree-sitter/parser.js +31 -0
- package/dist/engine/pipeline/indexing/extractor/tree-sitter/parser.js.map +1 -0
- package/dist/engine/pipeline/indexing/ids.d.ts +1 -0
- package/dist/engine/pipeline/indexing/ids.js +5 -0
- package/dist/engine/pipeline/indexing/ids.js.map +1 -0
- package/dist/engine/pipeline/indexing/index.d.ts +12 -0
- package/dist/engine/pipeline/indexing/index.js +538 -0
- package/dist/engine/pipeline/indexing/index.js.map +1 -0
- package/dist/engine/pipeline/indexing/root-paths.d.ts +7 -0
- package/dist/engine/pipeline/indexing/root-paths.js +151 -0
- package/dist/engine/pipeline/indexing/root-paths.js.map +1 -0
- package/dist/engine/pipeline/indexing/scanner/index.d.ts +5 -0
- package/dist/engine/pipeline/indexing/scanner/index.js +504 -0
- package/dist/engine/pipeline/indexing/scanner/index.js.map +1 -0
- package/dist/engine/pipeline/search/index.d.ts +12 -0
- package/dist/engine/pipeline/search/index.js +761 -0
- package/dist/engine/pipeline/search/index.js.map +1 -0
- package/dist/engine/service/index.d.ts +2 -0
- package/dist/engine/service/index.js +2 -0
- package/dist/engine/service/index.js.map +1 -0
- package/dist/engine/service/lexical.d.ts +29 -0
- package/dist/engine/service/lexical.js +482 -0
- package/dist/engine/service/lexical.js.map +1 -0
- package/dist/engine/service/root.d.ts +15 -0
- package/dist/engine/service/root.js +44 -0
- package/dist/engine/service/root.js.map +1 -0
- package/dist/engine/service/types.d.ts +157 -0
- package/dist/engine/service/types.js +2 -0
- package/dist/engine/service/types.js.map +1 -0
- package/dist/engine/service/zvec-grep.d.ts +2 -0
- package/dist/engine/service/zvec-grep.js +935 -0
- package/dist/engine/service/zvec-grep.js.map +1 -0
- package/dist/engine/storage/index.d.ts +42 -0
- package/dist/engine/storage/index.js +2 -0
- package/dist/engine/storage/index.js.map +1 -0
- package/dist/engine/storage/zvec.d.ts +51 -0
- package/dist/engine/storage/zvec.js +706 -0
- package/dist/engine/storage/zvec.js.map +1 -0
- package/dist/engine/types.d.ts +266 -0
- package/dist/engine/types.js +14 -0
- package/dist/engine/types.js.map +1 -0
- package/dist/engine/utils/glob.d.ts +6 -0
- package/dist/engine/utils/glob.js +110 -0
- package/dist/engine/utils/glob.js.map +1 -0
- package/dist/engine/utils/hash.d.ts +2 -0
- package/dist/engine/utils/hash.js +8 -0
- package/dist/engine/utils/hash.js.map +1 -0
- package/dist/engine/utils/json.d.ts +4 -0
- package/dist/engine/utils/json.js +45 -0
- package/dist/engine/utils/json.js.map +1 -0
- package/dist/engine/utils/lock.d.ts +19 -0
- package/dist/engine/utils/lock.js +254 -0
- package/dist/engine/utils/lock.js.map +1 -0
- package/dist/engine/utils/path.d.ts +4 -0
- package/dist/engine/utils/path.js +18 -0
- package/dist/engine/utils/path.js.map +1 -0
- package/dist/index.d.ts +6 -0
- package/dist/index.js +4 -0
- package/dist/index.js.map +1 -0
- package/package.json +47 -0
- package/skills/zvec-grep/SKILL.md +141 -0
- package/skills/zvec-grep/agents/openai.yaml +4 -0
|
@@ -0,0 +1,53 @@
|
|
|
1
|
+
import { extractCommonModifiers, extractGenericSignature, extractPrecedingDoc, } from "../families/metadata.js";
|
|
2
|
+
export const PYTHON_ADAPTER = {
|
|
3
|
+
format: "python",
|
|
4
|
+
entityTypes: new Set([
|
|
5
|
+
"class_definition",
|
|
6
|
+
"decorated_definition",
|
|
7
|
+
"function_definition",
|
|
8
|
+
]),
|
|
9
|
+
scopeTypes: new Set([
|
|
10
|
+
"class_definition",
|
|
11
|
+
"decorated_definition",
|
|
12
|
+
]),
|
|
13
|
+
extractName(node) {
|
|
14
|
+
if (node.type === "decorated_definition") {
|
|
15
|
+
const inner = node.namedChildren.find((child) => child.type === "function_definition" || child.type === "class_definition");
|
|
16
|
+
return inner ? this.extractName(inner) : undefined;
|
|
17
|
+
}
|
|
18
|
+
return node.childForFieldName("name")?.text;
|
|
19
|
+
},
|
|
20
|
+
shouldEnterScope(node) {
|
|
21
|
+
if (node.type !== "decorated_definition") {
|
|
22
|
+
return true;
|
|
23
|
+
}
|
|
24
|
+
return node.namedChildren.some((child) => child.type === "class_definition");
|
|
25
|
+
},
|
|
26
|
+
enterScopeNode(node) {
|
|
27
|
+
if (node.type !== "decorated_definition") {
|
|
28
|
+
return node;
|
|
29
|
+
}
|
|
30
|
+
return node.namedChildren.find((child) => child.type === "class_definition") ?? node;
|
|
31
|
+
},
|
|
32
|
+
extractSignature(node) {
|
|
33
|
+
return extractGenericSignature(innerPythonDefinition(node) ?? node);
|
|
34
|
+
},
|
|
35
|
+
extractDoc: extractPrecedingDoc,
|
|
36
|
+
extractModifiers(node) {
|
|
37
|
+
const modifiers = new Set(extractCommonModifiers(node));
|
|
38
|
+
if (/^\s*async\s+def\b/m.test(node.text)) {
|
|
39
|
+
modifiers.add("async");
|
|
40
|
+
}
|
|
41
|
+
if (/^\s*@staticmethod\b/m.test(node.text)) {
|
|
42
|
+
modifiers.add("static");
|
|
43
|
+
}
|
|
44
|
+
return [...modifiers];
|
|
45
|
+
},
|
|
46
|
+
};
|
|
47
|
+
function innerPythonDefinition(node) {
|
|
48
|
+
if (node.type !== "decorated_definition") {
|
|
49
|
+
return undefined;
|
|
50
|
+
}
|
|
51
|
+
return node.namedChildren.find((child) => child.type === "function_definition" || child.type === "class_definition");
|
|
52
|
+
}
|
|
53
|
+
//# sourceMappingURL=python.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"python.js","sourceRoot":"","sources":["../../../../../../../src/engine/pipeline/indexing/extractor/code/languages/python.ts"],"names":[],"mappings":"AAGA,OAAO,EACL,sBAAsB,EACtB,uBAAuB,EACvB,mBAAmB,GACpB,MAAM,yBAAyB,CAAC;AAGjC,MAAM,CAAC,MAAM,cAAc,GAAoB;IAC7C,MAAM,EAAE,QAAQ;IAChB,WAAW,EAAE,IAAI,GAAG,CAAC;QACnB,kBAAkB;QAClB,sBAAsB;QACtB,qBAAqB;KACtB,CAAC;IACF,UAAU,EAAE,IAAI,GAAG,CAAC;QAClB,kBAAkB;QAClB,sBAAsB;KACvB,CAAC;IACF,WAAW,CAAC,IAAI;QACd,IAAI,IAAI,CAAC,IAAI,KAAK,sBAAsB,EAAE,CAAC;YACzC,MAAM,KAAK,GAAG,IAAI,CAAC,aAAa,CAAC,IAAI,CAAC,CAAC,KAAK,EAAE,EAAE,CAC9C,KAAK,CAAC,IAAI,KAAK,qBAAqB,IAAI,KAAK,CAAC,IAAI,KAAK,kBAAkB,CAC1E,CAAC;YACF,OAAO,KAAK,CAAC,CAAC,CAAC,IAAI,CAAC,WAAW,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,SAAS,CAAC;QACrD,CAAC;QAED,OAAO,IAAI,CAAC,iBAAiB,CAAC,MAAM,CAAC,EAAE,IAAI,CAAC;IAC9C,CAAC;IACD,gBAAgB,CAAC,IAAI;QACnB,IAAI,IAAI,CAAC,IAAI,KAAK,sBAAsB,EAAE,CAAC;YACzC,OAAO,IAAI,CAAC;QACd,CAAC;QAED,OAAO,IAAI,CAAC,aAAa,CAAC,IAAI,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,KAAK,CAAC,IAAI,KAAK,kBAAkB,CAAC,CAAC;IAC/E,CAAC;IACD,cAAc,CAAC,IAAI;QACjB,IAAI,IAAI,CAAC,IAAI,KAAK,sBAAsB,EAAE,CAAC;YACzC,OAAO,IAAI,CAAC;QACd,CAAC;QAED,OAAO,IAAI,CAAC,aAAa,CAAC,IAAI,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,KAAK,CAAC,IAAI,KAAK,kBAAkB,CAAC,IAAI,IAAI,CAAC;IACvF,CAAC;IACD,gBAAgB,CAAC,IAAI;QACnB,OAAO,uBAAuB,CAAC,qBAAqB,CAAC,IAAI,CAAC,IAAI,IAAI,CAAC,CAAC;IACtE,CAAC;IACD,UAAU,EAAE,mBAAmB;IAC/B,gBAAgB,CAAC,IAAI;QACnB,MAAM,SAAS,GAAG,IAAI,GAAG,CAAqB,sBAAsB,CAAC,IAAI,CAAC,CAAC,CAAC;QAE5E,IAAI,oBAAoB,CAAC,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC;YACzC,SAAS,CAAC,GAAG,CAAC,OAAO,CAAC,CAAC;QACzB,CAAC;QAED,IAAI,sBAAsB,CAAC,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC;YAC3C,SAAS,CAAC,GAAG,CAAC,QAAQ,CAAC,CAAC;QAC1B,CAAC;QAED,OAAO,CAAC,GAAG,SAAS,CAAC,CAAC;IACxB,CAAC;CACF,CAAC;AAGF,SAAS,qBAAqB,CAAC,IAAY;IACzC,IAAI,IAAI,CAAC,IAAI,KAAK,sBAAsB,EAAE,CAAC;QACzC,OAAO,SAAS,CAAC;IACnB,CAAC;IAED,OAAO,IAAI,CAAC,aAAa,CAAC,IAAI,CAAC,CAAC,KAAK,EAAE,EAAE,CACvC,KAAK,CAAC,IAAI,KAAK,qBAAqB,IAAI,KAAK,CAAC,IAAI,KAAK,kBAAkB,CAC1E,CAAC;AACJ,CAAC"}
|
|
@@ -0,0 +1,29 @@
|
|
|
1
|
+
import { extractCommonModifiers, extractGenericSignature, extractPrecedingDoc, } from "../families/metadata.js";
|
|
2
|
+
export const RUST_ADAPTER = {
|
|
3
|
+
format: "rust",
|
|
4
|
+
entityTypes: new Set([
|
|
5
|
+
"enum_item",
|
|
6
|
+
"function_item",
|
|
7
|
+
"function_signature_item",
|
|
8
|
+
"impl_item",
|
|
9
|
+
"struct_item",
|
|
10
|
+
"trait_item",
|
|
11
|
+
"type_item",
|
|
12
|
+
"union_item",
|
|
13
|
+
]),
|
|
14
|
+
scopeTypes: new Set([
|
|
15
|
+
"impl_item",
|
|
16
|
+
"mod_item",
|
|
17
|
+
"trait_item",
|
|
18
|
+
]),
|
|
19
|
+
extractName(node) {
|
|
20
|
+
if (node.type === "impl_item") {
|
|
21
|
+
return node.childForFieldName("type")?.text;
|
|
22
|
+
}
|
|
23
|
+
return node.childForFieldName("name")?.text;
|
|
24
|
+
},
|
|
25
|
+
extractSignature: extractGenericSignature,
|
|
26
|
+
extractDoc: extractPrecedingDoc,
|
|
27
|
+
extractModifiers: extractCommonModifiers,
|
|
28
|
+
};
|
|
29
|
+
//# sourceMappingURL=rust.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"rust.js","sourceRoot":"","sources":["../../../../../../../src/engine/pipeline/indexing/extractor/code/languages/rust.ts"],"names":[],"mappings":"AACA,OAAO,EACL,sBAAsB,EACtB,uBAAuB,EACvB,mBAAmB,GACpB,MAAM,yBAAyB,CAAC;AAGjC,MAAM,CAAC,MAAM,YAAY,GAAoB;IAC3C,MAAM,EAAE,MAAM;IACd,WAAW,EAAE,IAAI,GAAG,CAAC;QACnB,WAAW;QACX,eAAe;QACf,yBAAyB;QACzB,WAAW;QACX,aAAa;QACb,YAAY;QACZ,WAAW;QACX,YAAY;KACb,CAAC;IACF,UAAU,EAAE,IAAI,GAAG,CAAC;QAClB,WAAW;QACX,UAAU;QACV,YAAY;KACb,CAAC;IACF,WAAW,CAAC,IAAI;QACd,IAAI,IAAI,CAAC,IAAI,KAAK,WAAW,EAAE,CAAC;YAC9B,OAAO,IAAI,CAAC,iBAAiB,CAAC,MAAM,CAAC,EAAE,IAAI,CAAC;QAC9C,CAAC;QAED,OAAO,IAAI,CAAC,iBAAiB,CAAC,MAAM,CAAC,EAAE,IAAI,CAAC;IAC9C,CAAC;IACD,gBAAgB,EAAE,uBAAuB;IACzC,UAAU,EAAE,mBAAmB;IAC/B,gBAAgB,EAAE,sBAAsB;CACzC,CAAC"}
|
|
@@ -0,0 +1,37 @@
|
|
|
1
|
+
import { classifyJavascriptTypescriptNode, extractJavascriptTypescriptDoc, extractJavascriptTypescriptModifiers, extractJavascriptTypescriptName, extractJavascriptTypescriptSignature, javascriptTypescriptScopeBreadcrumb, resolveJavascriptTypescriptEntities, shouldIndexJavascriptTypescriptEntity, } from "../families/js-ts.js";
|
|
2
|
+
export const TYPESCRIPT_ADAPTER = {
|
|
3
|
+
format: "typescript",
|
|
4
|
+
entityTypes: new Set([
|
|
5
|
+
"abstract_class_declaration",
|
|
6
|
+
"abstract_method_signature",
|
|
7
|
+
"class_declaration",
|
|
8
|
+
"enum_declaration",
|
|
9
|
+
"field_definition",
|
|
10
|
+
"function_declaration",
|
|
11
|
+
"generator_function_declaration",
|
|
12
|
+
"interface_declaration",
|
|
13
|
+
"method_signature",
|
|
14
|
+
"method_definition",
|
|
15
|
+
"pair",
|
|
16
|
+
"public_field_definition",
|
|
17
|
+
"type_alias_declaration",
|
|
18
|
+
"variable_declarator",
|
|
19
|
+
]),
|
|
20
|
+
scopeTypes: new Set([
|
|
21
|
+
"abstract_class_declaration",
|
|
22
|
+
"class_declaration",
|
|
23
|
+
"internal_module",
|
|
24
|
+
"interface_declaration",
|
|
25
|
+
"module_declaration",
|
|
26
|
+
"namespace_declaration",
|
|
27
|
+
]),
|
|
28
|
+
extractName: extractJavascriptTypescriptName,
|
|
29
|
+
shouldIndexEntity: shouldIndexJavascriptTypescriptEntity,
|
|
30
|
+
resolveEntities: resolveJavascriptTypescriptEntities,
|
|
31
|
+
scopeBreadcrumb: javascriptTypescriptScopeBreadcrumb,
|
|
32
|
+
classifyNode: classifyJavascriptTypescriptNode,
|
|
33
|
+
extractSignature: extractJavascriptTypescriptSignature,
|
|
34
|
+
extractDoc: extractJavascriptTypescriptDoc,
|
|
35
|
+
extractModifiers: extractJavascriptTypescriptModifiers,
|
|
36
|
+
};
|
|
37
|
+
//# sourceMappingURL=typescript.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"typescript.js","sourceRoot":"","sources":["../../../../../../../src/engine/pipeline/indexing/extractor/code/languages/typescript.ts"],"names":[],"mappings":"AACA,OAAO,EACL,gCAAgC,EAChC,8BAA8B,EAC9B,oCAAoC,EACpC,+BAA+B,EAC/B,oCAAoC,EACpC,mCAAmC,EACnC,mCAAmC,EACnC,qCAAqC,GACtC,MAAM,sBAAsB,CAAC;AAG9B,MAAM,CAAC,MAAM,kBAAkB,GAAoB;IACjD,MAAM,EAAE,YAAY;IACpB,WAAW,EAAE,IAAI,GAAG,CAAC;QACnB,4BAA4B;QAC5B,2BAA2B;QAC3B,mBAAmB;QACnB,kBAAkB;QAClB,kBAAkB;QAClB,sBAAsB;QACtB,gCAAgC;QAChC,uBAAuB;QACvB,kBAAkB;QAClB,mBAAmB;QACnB,MAAM;QACN,yBAAyB;QACzB,wBAAwB;QACxB,qBAAqB;KACtB,CAAC;IACF,UAAU,EAAE,IAAI,GAAG,CAAC;QAClB,4BAA4B;QAC5B,mBAAmB;QACnB,iBAAiB;QACjB,uBAAuB;QACvB,oBAAoB;QACpB,uBAAuB;KACxB,CAAC;IACF,WAAW,EAAE,+BAA+B;IAC5C,iBAAiB,EAAE,qCAAqC;IACxD,eAAe,EAAE,mCAAmC;IACpD,eAAe,EAAE,mCAAmC;IACpD,YAAY,EAAE,gCAAgC;IAC9C,gBAAgB,EAAE,oCAAoC;IACtD,UAAU,EAAE,8BAA8B;IAC1C,gBAAgB,EAAE,oCAAoC;CACvD,CAAC"}
|
|
@@ -0,0 +1,44 @@
|
|
|
1
|
+
import type { EntityFragment, FileInfo, ImageFormat } from "../../../types.js";
|
|
2
|
+
export type SourceKind = "text" | "image";
|
|
3
|
+
type SourceBase = {
|
|
4
|
+
kind: SourceKind;
|
|
5
|
+
file: FileInfo;
|
|
6
|
+
};
|
|
7
|
+
export type TextSource = SourceBase & {
|
|
8
|
+
kind: "text";
|
|
9
|
+
text: string;
|
|
10
|
+
};
|
|
11
|
+
export type ImageSource = SourceBase & {
|
|
12
|
+
kind: "image";
|
|
13
|
+
data: Uint8Array;
|
|
14
|
+
format: ImageFormat;
|
|
15
|
+
};
|
|
16
|
+
export type Source = TextSource | ImageSource;
|
|
17
|
+
export interface Extractor {
|
|
18
|
+
supports(source: Source): boolean;
|
|
19
|
+
extract(source: Source): Promise<EntityFragment[]>;
|
|
20
|
+
}
|
|
21
|
+
export type TextExtractorOptions = {
|
|
22
|
+
maxChunkChars?: number;
|
|
23
|
+
chunkOverlapChars?: number;
|
|
24
|
+
};
|
|
25
|
+
export declare class TextExtractor implements Extractor {
|
|
26
|
+
private readonly maxChunkChars;
|
|
27
|
+
private readonly chunkOverlapChars;
|
|
28
|
+
constructor(options?: TextExtractorOptions);
|
|
29
|
+
supports(source: Source): boolean;
|
|
30
|
+
extract(source: Source): Promise<EntityFragment[]>;
|
|
31
|
+
}
|
|
32
|
+
export declare class ImageExtractor implements Extractor {
|
|
33
|
+
supports(source: Source): boolean;
|
|
34
|
+
extract(source: Source): Promise<EntityFragment[]>;
|
|
35
|
+
}
|
|
36
|
+
export declare class ExtractorRegistry {
|
|
37
|
+
private readonly extractors;
|
|
38
|
+
constructor(extractors?: readonly Extractor[]);
|
|
39
|
+
resolve(source: Source): Extractor;
|
|
40
|
+
extract(source: Source): Promise<EntityFragment[]>;
|
|
41
|
+
private extractFirst;
|
|
42
|
+
}
|
|
43
|
+
export declare function extractFragments(source: Source): Promise<EntityFragment[]>;
|
|
44
|
+
export {};
|
|
@@ -0,0 +1,281 @@
|
|
|
1
|
+
import { EngineError } from "../../../errors/index.js";
|
|
2
|
+
import { makeEntityId } from "../ids.js";
|
|
3
|
+
import { CodeExtractor } from "./code/extractor.js";
|
|
4
|
+
import { MarkdownExtractor } from "./markdown/extractor.js";
|
|
5
|
+
const DEFAULT_TEXT_CHUNK_CHARS = 3600;
|
|
6
|
+
const DEFAULT_TEXT_CHUNK_OVERLAP_CHARS = 540;
|
|
7
|
+
export class TextExtractor {
|
|
8
|
+
maxChunkChars;
|
|
9
|
+
chunkOverlapChars;
|
|
10
|
+
constructor(options = {}) {
|
|
11
|
+
const maxChunkChars = options.maxChunkChars ?? DEFAULT_TEXT_CHUNK_CHARS;
|
|
12
|
+
const chunkOverlapChars = options.chunkOverlapChars ?? DEFAULT_TEXT_CHUNK_OVERLAP_CHARS;
|
|
13
|
+
if (!Number.isInteger(maxChunkChars) || maxChunkChars <= 0) {
|
|
14
|
+
throw new EngineError("Text extractor requires a positive integer chunk size", {
|
|
15
|
+
code: "ZVEC_GREP.ENGINE.EXTRACTORS.TEXT_INVALID_CHUNK_SIZE",
|
|
16
|
+
context: `maxChunkChars=${maxChunkChars}`,
|
|
17
|
+
});
|
|
18
|
+
}
|
|
19
|
+
if (!Number.isInteger(chunkOverlapChars)
|
|
20
|
+
|| chunkOverlapChars < 0
|
|
21
|
+
|| chunkOverlapChars >= maxChunkChars) {
|
|
22
|
+
throw new EngineError("Text extractor requires overlap to be smaller than chunk size", {
|
|
23
|
+
code: "ZVEC_GREP.ENGINE.EXTRACTORS.TEXT_INVALID_CHUNK_OVERLAP",
|
|
24
|
+
context: `maxChunkChars=${maxChunkChars} chunkOverlapChars=${chunkOverlapChars}`,
|
|
25
|
+
});
|
|
26
|
+
}
|
|
27
|
+
this.maxChunkChars = maxChunkChars;
|
|
28
|
+
this.chunkOverlapChars = chunkOverlapChars;
|
|
29
|
+
}
|
|
30
|
+
supports(source) {
|
|
31
|
+
return source.kind === "text";
|
|
32
|
+
}
|
|
33
|
+
async extract(source) {
|
|
34
|
+
if (source.kind !== "text") {
|
|
35
|
+
throw new EngineError("Text extractor received a non-text source", {
|
|
36
|
+
code: "ZVEC_GREP.ENGINE.EXTRACTORS.TEXT_UNSUPPORTED_SOURCE",
|
|
37
|
+
context: `fileId=${source.file.id} sourceKind=${source.kind}`,
|
|
38
|
+
});
|
|
39
|
+
}
|
|
40
|
+
validateSourceFile(source);
|
|
41
|
+
const chunks = chunkText(source.text, this.maxChunkChars, this.chunkOverlapChars);
|
|
42
|
+
return chunks.map((chunk, index) => ({
|
|
43
|
+
id: makeEntityId(source.file.id, index),
|
|
44
|
+
fileId: source.file.id,
|
|
45
|
+
range: chunk.range,
|
|
46
|
+
content: {
|
|
47
|
+
kind: "text",
|
|
48
|
+
text: chunk.text,
|
|
49
|
+
},
|
|
50
|
+
}));
|
|
51
|
+
}
|
|
52
|
+
}
|
|
53
|
+
export class ImageExtractor {
|
|
54
|
+
supports(source) {
|
|
55
|
+
return source.kind === "image";
|
|
56
|
+
}
|
|
57
|
+
async extract(source) {
|
|
58
|
+
if (source.kind !== "image") {
|
|
59
|
+
throw new EngineError("Image extractor received a non-image source", {
|
|
60
|
+
code: "ZVEC_GREP.ENGINE.EXTRACTORS.IMAGE_UNSUPPORTED_SOURCE",
|
|
61
|
+
context: `fileId=${source.file.id} sourceKind=${source.kind}`,
|
|
62
|
+
});
|
|
63
|
+
}
|
|
64
|
+
validateSourceFile(source);
|
|
65
|
+
if (source.data.byteLength === 0) {
|
|
66
|
+
throw new EngineError("Image extractor requires non-empty image data", {
|
|
67
|
+
code: "ZVEC_GREP.ENGINE.EXTRACTORS.IMAGE_EMPTY_DATA",
|
|
68
|
+
context: `fileId=${source.file.id} format=${source.format}`,
|
|
69
|
+
});
|
|
70
|
+
}
|
|
71
|
+
return [
|
|
72
|
+
{
|
|
73
|
+
id: makeEntityId(source.file.id, 0),
|
|
74
|
+
fileId: source.file.id,
|
|
75
|
+
range: {
|
|
76
|
+
kind: "file",
|
|
77
|
+
},
|
|
78
|
+
content: {
|
|
79
|
+
kind: "image",
|
|
80
|
+
data: source.data,
|
|
81
|
+
format: source.format,
|
|
82
|
+
},
|
|
83
|
+
},
|
|
84
|
+
];
|
|
85
|
+
}
|
|
86
|
+
}
|
|
87
|
+
const DEFAULT_EXTRACTORS = [
|
|
88
|
+
new CodeExtractor(),
|
|
89
|
+
new MarkdownExtractor(),
|
|
90
|
+
new TextExtractor(),
|
|
91
|
+
new ImageExtractor(),
|
|
92
|
+
];
|
|
93
|
+
export class ExtractorRegistry {
|
|
94
|
+
extractors;
|
|
95
|
+
constructor(extractors = DEFAULT_EXTRACTORS) {
|
|
96
|
+
if (extractors.length === 0) {
|
|
97
|
+
throw new EngineError("Extractor registry requires at least one extractor", {
|
|
98
|
+
code: "ZVEC_GREP.ENGINE.EXTRACTORS.EMPTY_REGISTRY",
|
|
99
|
+
});
|
|
100
|
+
}
|
|
101
|
+
this.extractors = extractors;
|
|
102
|
+
}
|
|
103
|
+
resolve(source) {
|
|
104
|
+
for (const extractor of this.extractors) {
|
|
105
|
+
if (extractor.supports(source)) {
|
|
106
|
+
return extractor;
|
|
107
|
+
}
|
|
108
|
+
}
|
|
109
|
+
throw new EngineError("No extractor supports source", {
|
|
110
|
+
code: "ZVEC_GREP.ENGINE.EXTRACTORS.UNSUPPORTED_SOURCE",
|
|
111
|
+
context: `fileId=${source.file.id} sourceKind=${source.kind}`,
|
|
112
|
+
});
|
|
113
|
+
}
|
|
114
|
+
extract(source) {
|
|
115
|
+
return this.extractFirst(source);
|
|
116
|
+
}
|
|
117
|
+
async extractFirst(source) {
|
|
118
|
+
let matched = false;
|
|
119
|
+
let lastResult = [];
|
|
120
|
+
for (const extractor of this.extractors) {
|
|
121
|
+
if (!extractor.supports(source)) {
|
|
122
|
+
continue;
|
|
123
|
+
}
|
|
124
|
+
matched = true;
|
|
125
|
+
lastResult = await extractor.extract(source);
|
|
126
|
+
if (lastResult.length > 0) {
|
|
127
|
+
return lastResult;
|
|
128
|
+
}
|
|
129
|
+
}
|
|
130
|
+
if (matched) {
|
|
131
|
+
return lastResult;
|
|
132
|
+
}
|
|
133
|
+
throw new EngineError("No extractor supports source", {
|
|
134
|
+
code: "ZVEC_GREP.ENGINE.EXTRACTORS.UNSUPPORTED_SOURCE",
|
|
135
|
+
context: `fileId=${source.file.id} sourceKind=${source.kind}`,
|
|
136
|
+
});
|
|
137
|
+
}
|
|
138
|
+
}
|
|
139
|
+
export function extractFragments(source) {
|
|
140
|
+
return new ExtractorRegistry().extract(source);
|
|
141
|
+
}
|
|
142
|
+
function chunkText(text, maxChars, overlapChars) {
|
|
143
|
+
if (text.trim().length === 0) {
|
|
144
|
+
return [];
|
|
145
|
+
}
|
|
146
|
+
const lines = text.split("\n");
|
|
147
|
+
const lineOffsets = computeLineOffsets(lines);
|
|
148
|
+
const chunks = [];
|
|
149
|
+
let startIndex = 0;
|
|
150
|
+
while (startIndex < lines.length) {
|
|
151
|
+
if (lines[startIndex].length + 1 > maxChars) {
|
|
152
|
+
const line = lines[startIndex];
|
|
153
|
+
const lineNumber = startIndex + 1;
|
|
154
|
+
let offset = 0;
|
|
155
|
+
while (offset < line.length) {
|
|
156
|
+
const remaining = line.length - offset;
|
|
157
|
+
const sliceLength = remaining <= maxChars
|
|
158
|
+
? remaining
|
|
159
|
+
: findLineCut(line.slice(offset), maxChars);
|
|
160
|
+
const slice = line.slice(offset, offset + sliceLength);
|
|
161
|
+
if (slice.trim().length > 0) {
|
|
162
|
+
chunks.push({
|
|
163
|
+
text: slice,
|
|
164
|
+
range: {
|
|
165
|
+
kind: "text",
|
|
166
|
+
startLine: lineNumber,
|
|
167
|
+
endLine: lineNumber,
|
|
168
|
+
startOffset: lineOffsets[startIndex] + offset,
|
|
169
|
+
endOffset: lineOffsets[startIndex] + offset + slice.length,
|
|
170
|
+
},
|
|
171
|
+
});
|
|
172
|
+
}
|
|
173
|
+
offset += sliceLength;
|
|
174
|
+
}
|
|
175
|
+
startIndex++;
|
|
176
|
+
continue;
|
|
177
|
+
}
|
|
178
|
+
let usedChars = 0;
|
|
179
|
+
let endIndex = startIndex;
|
|
180
|
+
while (endIndex < lines.length) {
|
|
181
|
+
const lineLength = lines[endIndex].length + 1;
|
|
182
|
+
if (usedChars + lineLength > maxChars && endIndex > startIndex) {
|
|
183
|
+
break;
|
|
184
|
+
}
|
|
185
|
+
usedChars += lineLength;
|
|
186
|
+
endIndex++;
|
|
187
|
+
}
|
|
188
|
+
const chunkLines = lines.slice(startIndex, endIndex);
|
|
189
|
+
const chunk = chunkLines.join("\n");
|
|
190
|
+
if (chunk.trim().length > 0) {
|
|
191
|
+
const endLineIndex = endIndex - 1;
|
|
192
|
+
chunks.push({
|
|
193
|
+
text: chunk,
|
|
194
|
+
range: {
|
|
195
|
+
kind: "text",
|
|
196
|
+
startLine: startIndex + 1,
|
|
197
|
+
endLine: endIndex,
|
|
198
|
+
startOffset: lineOffsets[startIndex],
|
|
199
|
+
endOffset: lineOffsets[endLineIndex] + lines[endLineIndex].length,
|
|
200
|
+
},
|
|
201
|
+
});
|
|
202
|
+
}
|
|
203
|
+
if (endIndex >= lines.length) {
|
|
204
|
+
break;
|
|
205
|
+
}
|
|
206
|
+
startIndex = computeNextStartLine(lines, startIndex, endIndex, overlapChars);
|
|
207
|
+
}
|
|
208
|
+
return chunks;
|
|
209
|
+
}
|
|
210
|
+
function computeLineOffsets(lines) {
|
|
211
|
+
const offsets = [];
|
|
212
|
+
let offset = 0;
|
|
213
|
+
for (const line of lines) {
|
|
214
|
+
offsets.push(offset);
|
|
215
|
+
offset += line.length + 1;
|
|
216
|
+
}
|
|
217
|
+
return offsets;
|
|
218
|
+
}
|
|
219
|
+
function computeNextStartLine(lines, startIndex, endIndex, overlapChars) {
|
|
220
|
+
if (overlapChars === 0) {
|
|
221
|
+
return endIndex;
|
|
222
|
+
}
|
|
223
|
+
let overlapLines = 0;
|
|
224
|
+
let overlapCount = 0;
|
|
225
|
+
for (let index = endIndex - 1; index >= startIndex && overlapCount < overlapChars; index--) {
|
|
226
|
+
overlapCount += lines[index].length + 1;
|
|
227
|
+
overlapLines++;
|
|
228
|
+
}
|
|
229
|
+
const nextStart = endIndex - overlapLines;
|
|
230
|
+
return nextStart > startIndex ? nextStart : endIndex;
|
|
231
|
+
}
|
|
232
|
+
function findLineCut(line, maxChars) {
|
|
233
|
+
if (line.length <= maxChars) {
|
|
234
|
+
return line.length;
|
|
235
|
+
}
|
|
236
|
+
const minPosition = Math.floor(maxChars * 0.7);
|
|
237
|
+
let bestPosition = -1;
|
|
238
|
+
let bestScore = 0;
|
|
239
|
+
for (let index = minPosition; index < maxChars; index++) {
|
|
240
|
+
const character = line[index];
|
|
241
|
+
let score = 0;
|
|
242
|
+
if (character === "." || character === "!" || character === "?") {
|
|
243
|
+
score = 4;
|
|
244
|
+
}
|
|
245
|
+
else if (character === "," || character === ";" || character === ":") {
|
|
246
|
+
score = 3;
|
|
247
|
+
}
|
|
248
|
+
else if (character === " " || character === "\t") {
|
|
249
|
+
score = 2;
|
|
250
|
+
}
|
|
251
|
+
else if (character === "-" || character === "/" || character === "\\") {
|
|
252
|
+
score = 1;
|
|
253
|
+
}
|
|
254
|
+
if (score > 0 && score >= bestScore) {
|
|
255
|
+
bestScore = score;
|
|
256
|
+
bestPosition = index + 1;
|
|
257
|
+
}
|
|
258
|
+
}
|
|
259
|
+
return bestPosition > 0 ? bestPosition : maxChars;
|
|
260
|
+
}
|
|
261
|
+
function validateSourceFile(source) {
|
|
262
|
+
if (source.file.id.trim().length === 0) {
|
|
263
|
+
throw new EngineError("Extractor source requires a non-empty file id", {
|
|
264
|
+
code: "ZVEC_GREP.ENGINE.EXTRACTORS.EMPTY_FILE_ID",
|
|
265
|
+
context: `sourceKind=${source.kind}`,
|
|
266
|
+
});
|
|
267
|
+
}
|
|
268
|
+
if (source.file.absolutePath.trim().length === 0) {
|
|
269
|
+
throw new EngineError("Extractor source requires a non-empty absolute file path", {
|
|
270
|
+
code: "ZVEC_GREP.ENGINE.EXTRACTORS.EMPTY_ABSOLUTE_FILE_PATH",
|
|
271
|
+
context: `fileId=${source.file.id} sourceKind=${source.kind}`,
|
|
272
|
+
});
|
|
273
|
+
}
|
|
274
|
+
if (source.file.relativePath.trim().length === 0) {
|
|
275
|
+
throw new EngineError("Extractor source requires a non-empty relative file path", {
|
|
276
|
+
code: "ZVEC_GREP.ENGINE.EXTRACTORS.EMPTY_RELATIVE_FILE_PATH",
|
|
277
|
+
context: `fileId=${source.file.id} sourceKind=${source.kind}`,
|
|
278
|
+
});
|
|
279
|
+
}
|
|
280
|
+
}
|
|
281
|
+
//# sourceMappingURL=index.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"index.js","sourceRoot":"","sources":["../../../../../src/engine/pipeline/indexing/extractor/index.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,WAAW,EAAE,MAAM,0BAA0B,CAAC;AAOvD,OAAO,EAAE,YAAY,EAAE,MAAM,WAAW,CAAC;AACzC,OAAO,EAAE,aAAa,EAAE,MAAM,qBAAqB,CAAC;AACpD,OAAO,EAAE,iBAAiB,EAAE,MAAM,yBAAyB,CAAC;AA4C5D,MAAM,wBAAwB,GAAG,IAAI,CAAC;AAGtC,MAAM,gCAAgC,GAAG,GAAG,CAAC;AAG7C,MAAM,OAAO,aAAa;IACP,aAAa,CAAS;IACtB,iBAAiB,CAAS;IAG3C,YAAY,UAAgC,EAAE;QAC5C,MAAM,aAAa,GAAG,OAAO,CAAC,aAAa,IAAI,wBAAwB,CAAC;QACxE,MAAM,iBAAiB,GAAG,OAAO,CAAC,iBAAiB,IAAI,gCAAgC,CAAC;QAExF,IAAI,CAAC,MAAM,CAAC,SAAS,CAAC,aAAa,CAAC,IAAI,aAAa,IAAI,CAAC,EAAE,CAAC;YAC3D,MAAM,IAAI,WAAW,CAAC,uDAAuD,EAAE;gBAC7E,IAAI,EAAE,qDAAqD;gBAC3D,OAAO,EAAE,iBAAiB,aAAa,EAAE;aAC1C,CAAC,CAAC;QACL,CAAC;QAED,IACE,CAAC,MAAM,CAAC,SAAS,CAAC,iBAAiB,CAAC;eACjC,iBAAiB,GAAG,CAAC;eACrB,iBAAiB,IAAI,aAAa,EACrC,CAAC;YACD,MAAM,IAAI,WAAW,CAAC,+DAA+D,EAAE;gBACrF,IAAI,EAAE,wDAAwD;gBAC9D,OAAO,EAAE,iBAAiB,aAAa,sBAAsB,iBAAiB,EAAE;aACjF,CAAC,CAAC;QACL,CAAC;QAED,IAAI,CAAC,aAAa,GAAG,aAAa,CAAC;QACnC,IAAI,CAAC,iBAAiB,GAAG,iBAAiB,CAAC;IAC7C,CAAC;IAGD,QAAQ,CAAC,MAAc;QACrB,OAAO,MAAM,CAAC,IAAI,KAAK,MAAM,CAAC;IAChC,CAAC;IAGD,KAAK,CAAC,OAAO,CAAC,MAAc;QAC1B,IAAI,MAAM,CAAC,IAAI,KAAK,MAAM,EAAE,CAAC;YAC3B,MAAM,IAAI,WAAW,CAAC,2CAA2C,EAAE;gBACjE,IAAI,EAAE,qDAAqD;gBAC3D,OAAO,EAAE,UAAU,MAAM,CAAC,IAAI,CAAC,EAAE,eAAe,MAAM,CAAC,IAAI,EAAE;aAC9D,CAAC,CAAC;QACL,CAAC;QAED,kBAAkB,CAAC,MAAM,CAAC,CAAC;QAE3B,MAAM,MAAM,GAAG,SAAS,CAAC,MAAM,CAAC,IAAI,EAAE,IAAI,CAAC,aAAa,EAAE,IAAI,CAAC,iBAAiB,CAAC,CAAC;QAElF,OAAO,MAAM,CAAC,GAAG,CAAC,CAAC,KAAK,EAAE,KAAK,EAAkB,EAAE,CAAC,CAAC;YACnD,EAAE,EAAE,YAAY,CAAC,MAAM,CAAC,IAAI,CAAC,EAAE,EAAE,KAAK,CAAC;YACvC,MAAM,EAAE,MAAM,CAAC,IAAI,CAAC,EAAE;YACtB,KAAK,EAAE,KAAK,CAAC,KAAK;YAClB,OAAO,EAAE;gBACP,IAAI,EAAE,MAAM;gBACZ,IAAI,EAAE,KAAK,CAAC,IAAI;aACjB;SACF,CAAC,CAAC,CAAC;IACN,CAAC;CACF;AAGD,MAAM,OAAO,cAAc;IACzB,QAAQ,CAAC,MAAc;QACrB,OAAO,MAAM,CAAC,IAAI,KAAK,OAAO,CAAC;IACjC,CAAC;IAGD,KAAK,CAAC,OAAO,CAAC,MAAc;QAC1B,IAAI,MAAM,CAAC,IAAI,KAAK,OAAO,EAAE,CAAC;YAC5B,MAAM,IAAI,WAAW,CAAC,6CAA6C,EAAE;gBACnE,IAAI,EAAE,sDAAsD;gBAC5D,OAAO,EAAE,UAAU,MAAM,CAAC,IAAI,CAAC,EAAE,eAAe,MAAM,CAAC,IAAI,EAAE;aAC9D,CAAC,CAAC;QACL,CAAC;QAED,kBAAkB,CAAC,MAAM,CAAC,CAAC;QAE3B,IAAI,MAAM,CAAC,IAAI,CAAC,UAAU,KAAK,CAAC,EAAE,CAAC;YACjC,MAAM,IAAI,WAAW,CAAC,+CAA+C,EAAE;gBACrE,IAAI,EAAE,8CAA8C;gBACpD,OAAO,EAAE,UAAU,MAAM,CAAC,IAAI,CAAC,EAAE,WAAW,MAAM,CAAC,MAAM,EAAE;aAC5D,CAAC,CAAC;QACL,CAAC;QAED,OAAO;YACL;gBACE,EAAE,EAAE,YAAY,CAAC,MAAM,CAAC,IAAI,CAAC,EAAE,EAAE,CAAC,CAAC;gBACnC,MAAM,EAAE,MAAM,CAAC,IAAI,CAAC,EAAE;gBACtB,KAAK,EAAE;oBACL,IAAI,EAAE,MAAM;iBACb;gBACD,OAAO,EAAE;oBACP,IAAI,EAAE,OAAO;oBACb,IAAI,EAAE,MAAM,CAAC,IAAI;oBACjB,MAAM,EAAE,MAAM,CAAC,MAAM;iBACtB;aACF;SACF,CAAC;IACJ,CAAC;CACF;AAGD,MAAM,kBAAkB,GAAyB;IAC/C,IAAI,aAAa,EAAE;IACnB,IAAI,iBAAiB,EAAE;IACvB,IAAI,aAAa,EAAE;IACnB,IAAI,cAAc,EAAE;CACrB,CAAC;AAGF,MAAM,OAAO,iBAAiB;IACX,UAAU,CAAuB;IAGlD,YAAY,aAAmC,kBAAkB;QAC/D,IAAI,UAAU,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;YAC5B,MAAM,IAAI,WAAW,CAAC,oDAAoD,EAAE;gBAC1E,IAAI,EAAE,4CAA4C;aACnD,CAAC,CAAC;QACL,CAAC;QAED,IAAI,CAAC,UAAU,GAAG,UAAU,CAAC;IAC/B,CAAC;IAGD,OAAO,CAAC,MAAc;QACpB,KAAK,MAAM,SAAS,IAAI,IAAI,CAAC,UAAU,EAAE,CAAC;YACxC,IAAI,SAAS,CAAC,QAAQ,CAAC,MAAM,CAAC,EAAE,CAAC;gBAC/B,OAAO,SAAS,CAAC;YACnB,CAAC;QACH,CAAC;QAED,MAAM,IAAI,WAAW,CAAC,8BAA8B,EAAE;YACpD,IAAI,EAAE,gDAAgD;YACtD,OAAO,EAAE,UAAU,MAAM,CAAC,IAAI,CAAC,EAAE,eAAe,MAAM,CAAC,IAAI,EAAE;SAC9D,CAAC,CAAC;IACL,CAAC;IAGD,OAAO,CAAC,MAAc;QACpB,OAAO,IAAI,CAAC,YAAY,CAAC,MAAM,CAAC,CAAC;IACnC,CAAC;IAGO,KAAK,CAAC,YAAY,CAAC,MAAc;QACvC,IAAI,OAAO,GAAG,KAAK,CAAC;QACpB,IAAI,UAAU,GAAqB,EAAE,CAAC;QAEtC,KAAK,MAAM,SAAS,IAAI,IAAI,CAAC,UAAU,EAAE,CAAC;YACxC,IAAI,CAAC,SAAS,CAAC,QAAQ,CAAC,MAAM,CAAC,EAAE,CAAC;gBAChC,SAAS;YACX,CAAC;YAED,OAAO,GAAG,IAAI,CAAC;YACf,UAAU,GAAG,MAAM,SAAS,CAAC,OAAO,CAAC,MAAM,CAAC,CAAC;YAE7C,IAAI,UAAU,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;gBAC1B,OAAO,UAAU,CAAC;YACpB,CAAC;QACH,CAAC;QAED,IAAI,OAAO,EAAE,CAAC;YACZ,OAAO,UAAU,CAAC;QACpB,CAAC;QAED,MAAM,IAAI,WAAW,CAAC,8BAA8B,EAAE;YACpD,IAAI,EAAE,gDAAgD;YACtD,OAAO,EAAE,UAAU,MAAM,CAAC,IAAI,CAAC,EAAE,eAAe,MAAM,CAAC,IAAI,EAAE;SAC9D,CAAC,CAAC;IACL,CAAC;CACF;AAGD,MAAM,UAAU,gBAAgB,CAAC,MAAc;IAC7C,OAAO,IAAI,iBAAiB,EAAE,CAAC,OAAO,CAAC,MAAM,CAAC,CAAC;AACjD,CAAC;AASD,SAAS,SAAS,CAAC,IAAY,EAAE,QAAgB,EAAE,YAAoB;IACrE,IAAI,IAAI,CAAC,IAAI,EAAE,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;QAC7B,OAAO,EAAE,CAAC;IACZ,CAAC;IAED,MAAM,KAAK,GAAG,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC;IAC/B,MAAM,WAAW,GAAG,kBAAkB,CAAC,KAAK,CAAC,CAAC;IAC9C,MAAM,MAAM,GAAgB,EAAE,CAAC;IAE/B,IAAI,UAAU,GAAG,CAAC,CAAC;IAEnB,OAAO,UAAU,GAAG,KAAK,CAAC,MAAM,EAAE,CAAC;QACjC,IAAI,KAAK,CAAC,UAAU,CAAC,CAAC,MAAM,GAAG,CAAC,GAAG,QAAQ,EAAE,CAAC;YAC5C,MAAM,IAAI,GAAG,KAAK,CAAC,UAAU,CAAC,CAAC;YAC/B,MAAM,UAAU,GAAG,UAAU,GAAG,CAAC,CAAC;YAClC,IAAI,MAAM,GAAG,CAAC,CAAC;YAEf,OAAO,MAAM,GAAG,IAAI,CAAC,MAAM,EAAE,CAAC;gBAC5B,MAAM,SAAS,GAAG,IAAI,CAAC,MAAM,GAAG,MAAM,CAAC;gBACvC,MAAM,WAAW,GAAG,SAAS,IAAI,QAAQ;oBACvC,CAAC,CAAC,SAAS;oBACX,CAAC,CAAC,WAAW,CAAC,IAAI,CAAC,KAAK,CAAC,MAAM,CAAC,EAAE,QAAQ,CAAC,CAAC;gBAC9C,MAAM,KAAK,GAAG,IAAI,CAAC,KAAK,CAAC,MAAM,EAAE,MAAM,GAAG,WAAW,CAAC,CAAC;gBAEvD,IAAI,KAAK,CAAC,IAAI,EAAE,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;oBAC5B,MAAM,CAAC,IAAI,CAAC;wBACV,IAAI,EAAE,KAAK;wBACX,KAAK,EAAE;4BACL,IAAI,EAAE,MAAM;4BACZ,SAAS,EAAE,UAAU;4BACrB,OAAO,EAAE,UAAU;4BACnB,WAAW,EAAE,WAAW,CAAC,UAAU,CAAC,GAAG,MAAM;4BAC7C,SAAS,EAAE,WAAW,CAAC,UAAU,CAAC,GAAG,MAAM,GAAG,KAAK,CAAC,MAAM;yBAC3D;qBACF,CAAC,CAAC;gBACL,CAAC;gBAED,MAAM,IAAI,WAAW,CAAC;YACxB,CAAC;YAED,UAAU,EAAE,CAAC;YACb,SAAS;QACX,CAAC;QAED,IAAI,SAAS,GAAG,CAAC,CAAC;QAClB,IAAI,QAAQ,GAAG,UAAU,CAAC;QAE1B,OAAO,QAAQ,GAAG,KAAK,CAAC,MAAM,EAAE,CAAC;YAC/B,MAAM,UAAU,GAAG,KAAK,CAAC,QAAQ,CAAC,CAAC,MAAM,GAAG,CAAC,CAAC;YAE9C,IAAI,SAAS,GAAG,UAAU,GAAG,QAAQ,IAAI,QAAQ,GAAG,UAAU,EAAE,CAAC;gBAC/D,MAAM;YACR,CAAC;YAED,SAAS,IAAI,UAAU,CAAC;YACxB,QAAQ,EAAE,CAAC;QACb,CAAC;QAED,MAAM,UAAU,GAAG,KAAK,CAAC,KAAK,CAAC,UAAU,EAAE,QAAQ,CAAC,CAAC;QACrD,MAAM,KAAK,GAAG,UAAU,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;QAEpC,IAAI,KAAK,CAAC,IAAI,EAAE,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;YAC5B,MAAM,YAAY,GAAG,QAAQ,GAAG,CAAC,CAAC;YAElC,MAAM,CAAC,IAAI,CAAC;gBACV,IAAI,EAAE,KAAK;gBACX,KAAK,EAAE;oBACL,IAAI,EAAE,MAAM;oBACZ,SAAS,EAAE,UAAU,GAAG,CAAC;oBACzB,OAAO,EAAE,QAAQ;oBACjB,WAAW,EAAE,WAAW,CAAC,UAAU,CAAC;oBACpC,SAAS,EAAE,WAAW,CAAC,YAAY,CAAC,GAAG,KAAK,CAAC,YAAY,CAAC,CAAC,MAAM;iBAClE;aACF,CAAC,CAAC;QACL,CAAC;QAED,IAAI,QAAQ,IAAI,KAAK,CAAC,MAAM,EAAE,CAAC;YAC7B,MAAM;QACR,CAAC;QAED,UAAU,GAAG,oBAAoB,CAAC,KAAK,EAAE,UAAU,EAAE,QAAQ,EAAE,YAAY,CAAC,CAAC;IAC/E,CAAC;IAED,OAAO,MAAM,CAAC;AAChB,CAAC;AAGD,SAAS,kBAAkB,CAAC,KAAwB;IAClD,MAAM,OAAO,GAAa,EAAE,CAAC;IAC7B,IAAI,MAAM,GAAG,CAAC,CAAC;IAEf,KAAK,MAAM,IAAI,IAAI,KAAK,EAAE,CAAC;QACzB,OAAO,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;QACrB,MAAM,IAAI,IAAI,CAAC,MAAM,GAAG,CAAC,CAAC;IAC5B,CAAC;IAED,OAAO,OAAO,CAAC;AACjB,CAAC;AAGD,SAAS,oBAAoB,CAC3B,KAAwB,EACxB,UAAkB,EAClB,QAAgB,EAChB,YAAoB;IAEpB,IAAI,YAAY,KAAK,CAAC,EAAE,CAAC;QACvB,OAAO,QAAQ,CAAC;IAClB,CAAC;IAED,IAAI,YAAY,GAAG,CAAC,CAAC;IACrB,IAAI,YAAY,GAAG,CAAC,CAAC;IAErB,KAAK,IAAI,KAAK,GAAG,QAAQ,GAAG,CAAC,EAAE,KAAK,IAAI,UAAU,IAAI,YAAY,GAAG,YAAY,EAAE,KAAK,EAAE,EAAE,CAAC;QAC3F,YAAY,IAAI,KAAK,CAAC,KAAK,CAAC,CAAC,MAAM,GAAG,CAAC,CAAC;QACxC,YAAY,EAAE,CAAC;IACjB,CAAC;IAED,MAAM,SAAS,GAAG,QAAQ,GAAG,YAAY,CAAC;IAE1C,OAAO,SAAS,GAAG,UAAU,CAAC,CAAC,CAAC,SAAS,CAAC,CAAC,CAAC,QAAQ,CAAC;AACvD,CAAC;AAGD,SAAS,WAAW,CAAC,IAAY,EAAE,QAAgB;IACjD,IAAI,IAAI,CAAC,MAAM,IAAI,QAAQ,EAAE,CAAC;QAC5B,OAAO,IAAI,CAAC,MAAM,CAAC;IACrB,CAAC;IAED,MAAM,WAAW,GAAG,IAAI,CAAC,KAAK,CAAC,QAAQ,GAAG,GAAG,CAAC,CAAC;IAC/C,IAAI,YAAY,GAAG,CAAC,CAAC,CAAC;IACtB,IAAI,SAAS,GAAG,CAAC,CAAC;IAElB,KAAK,IAAI,KAAK,GAAG,WAAW,EAAE,KAAK,GAAG,QAAQ,EAAE,KAAK,EAAE,EAAE,CAAC;QACxD,MAAM,SAAS,GAAG,IAAI,CAAC,KAAK,CAAC,CAAC;QAC9B,IAAI,KAAK,GAAG,CAAC,CAAC;QAEd,IAAI,SAAS,KAAK,GAAG,IAAI,SAAS,KAAK,GAAG,IAAI,SAAS,KAAK,GAAG,EAAE,CAAC;YAChE,KAAK,GAAG,CAAC,CAAC;QACZ,CAAC;aAAM,IAAI,SAAS,KAAK,GAAG,IAAI,SAAS,KAAK,GAAG,IAAI,SAAS,KAAK,GAAG,EAAE,CAAC;YACvE,KAAK,GAAG,CAAC,CAAC;QACZ,CAAC;aAAM,IAAI,SAAS,KAAK,GAAG,IAAI,SAAS,KAAK,IAAI,EAAE,CAAC;YACnD,KAAK,GAAG,CAAC,CAAC;QACZ,CAAC;aAAM,IAAI,SAAS,KAAK,GAAG,IAAI,SAAS,KAAK,GAAG,IAAI,SAAS,KAAK,IAAI,EAAE,CAAC;YACxE,KAAK,GAAG,CAAC,CAAC;QACZ,CAAC;QAED,IAAI,KAAK,GAAG,CAAC,IAAI,KAAK,IAAI,SAAS,EAAE,CAAC;YACpC,SAAS,GAAG,KAAK,CAAC;YAClB,YAAY,GAAG,KAAK,GAAG,CAAC,CAAC;QAC3B,CAAC;IACH,CAAC;IAED,OAAO,YAAY,GAAG,CAAC,CAAC,CAAC,CAAC,YAAY,CAAC,CAAC,CAAC,QAAQ,CAAC;AACpD,CAAC;AAGD,SAAS,kBAAkB,CAAC,MAAc;IACxC,IAAI,MAAM,CAAC,IAAI,CAAC,EAAE,CAAC,IAAI,EAAE,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;QACvC,MAAM,IAAI,WAAW,CAAC,+CAA+C,EAAE;YACrE,IAAI,EAAE,2CAA2C;YACjD,OAAO,EAAE,cAAc,MAAM,CAAC,IAAI,EAAE;SACrC,CAAC,CAAC;IACL,CAAC;IAED,IAAI,MAAM,CAAC,IAAI,CAAC,YAAY,CAAC,IAAI,EAAE,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;QACjD,MAAM,IAAI,WAAW,CAAC,0DAA0D,EAAE;YAChF,IAAI,EAAE,sDAAsD;YAC5D,OAAO,EAAE,UAAU,MAAM,CAAC,IAAI,CAAC,EAAE,eAAe,MAAM,CAAC,IAAI,EAAE;SAC9D,CAAC,CAAC;IACL,CAAC;IAED,IAAI,MAAM,CAAC,IAAI,CAAC,YAAY,CAAC,IAAI,EAAE,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;QACjD,MAAM,IAAI,WAAW,CAAC,0DAA0D,EAAE;YAChF,IAAI,EAAE,sDAAsD;YAC5D,OAAO,EAAE,UAAU,MAAM,CAAC,IAAI,CAAC,EAAE,eAAe,MAAM,CAAC,IAAI,EAAE;SAC9D,CAAC,CAAC;IACL,CAAC;AACH,CAAC"}
|
|
@@ -0,0 +1,13 @@
|
|
|
1
|
+
import type { EntityFragment } from "../../../../types.js";
|
|
2
|
+
import type { Extractor, Source } from "../index.js";
|
|
3
|
+
export type MarkdownExtractorOptions = {
|
|
4
|
+
maxChunkChars?: number;
|
|
5
|
+
chunkOverlapChars?: number;
|
|
6
|
+
};
|
|
7
|
+
export declare class MarkdownExtractor implements Extractor {
|
|
8
|
+
private readonly maxChunkChars;
|
|
9
|
+
private readonly chunkOverlapChars;
|
|
10
|
+
constructor(options?: MarkdownExtractorOptions);
|
|
11
|
+
supports(source: Source): boolean;
|
|
12
|
+
extract(source: Source): Promise<EntityFragment[]>;
|
|
13
|
+
}
|