codevault 1.4.1 → 1.6.0
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/.env.example +79 -44
- package/README.md +115 -343
- package/package.json +4 -4
- package/dist/chunking/file-grouper.d.ts +0 -39
- package/dist/chunking/file-grouper.d.ts.map +0 -1
- package/dist/chunking/file-grouper.js +0 -181
- package/dist/chunking/file-grouper.js.map +0 -1
- package/dist/chunking/semantic-chunker.d.ts +0 -37
- package/dist/chunking/semantic-chunker.d.ts.map +0 -1
- package/dist/chunking/semantic-chunker.js +0 -172
- package/dist/chunking/semantic-chunker.js.map +0 -1
- package/dist/chunking/token-counter.d.ts +0 -28
- package/dist/chunking/token-counter.d.ts.map +0 -1
- package/dist/chunking/token-counter.js +0 -207
- package/dist/chunking/token-counter.js.map +0 -1
- package/dist/cli/commands/ask-cmd.d.ts +0 -3
- package/dist/cli/commands/ask-cmd.d.ts.map +0 -1
- package/dist/cli/commands/ask-cmd.js +0 -130
- package/dist/cli/commands/ask-cmd.js.map +0 -1
- package/dist/cli/commands/config-cmd.d.ts +0 -3
- package/dist/cli/commands/config-cmd.d.ts.map +0 -1
- package/dist/cli/commands/config-cmd.js +0 -245
- package/dist/cli/commands/config-cmd.js.map +0 -1
- package/dist/cli/commands/context.d.ts +0 -3
- package/dist/cli/commands/context.d.ts.map +0 -1
- package/dist/cli/commands/context.js +0 -98
- package/dist/cli/commands/context.js.map +0 -1
- package/dist/cli/commands/interactive-config.d.ts +0 -2
- package/dist/cli/commands/interactive-config.d.ts.map +0 -1
- package/dist/cli/commands/interactive-config.js +0 -304
- package/dist/cli/commands/interactive-config.js.map +0 -1
- package/dist/cli.d.ts +0 -3
- package/dist/cli.d.ts.map +0 -1
- package/dist/cli.js +0 -398
- package/dist/cli.js.map +0 -1
- package/dist/codemap/io.d.ts +0 -5
- package/dist/codemap/io.d.ts.map +0 -1
- package/dist/codemap/io.js +0 -30
- package/dist/codemap/io.js.map +0 -1
- package/dist/config/apply-env.d.ts +0 -15
- package/dist/config/apply-env.d.ts.map +0 -1
- package/dist/config/apply-env.js +0 -107
- package/dist/config/apply-env.js.map +0 -1
- package/dist/config/loader.d.ts +0 -57
- package/dist/config/loader.d.ts.map +0 -1
- package/dist/config/loader.js +0 -304
- package/dist/config/loader.js.map +0 -1
- package/dist/config/types.d.ts +0 -48
- package/dist/config/types.d.ts.map +0 -1
- package/dist/config/types.js +0 -2
- package/dist/config/types.js.map +0 -1
- package/dist/context/packs.d.ts +0 -33
- package/dist/context/packs.d.ts.map +0 -1
- package/dist/context/packs.js +0 -180
- package/dist/context/packs.js.map +0 -1
- package/dist/core/batch-indexer.d.ts +0 -44
- package/dist/core/batch-indexer.d.ts.map +0 -1
- package/dist/core/batch-indexer.js +0 -161
- package/dist/core/batch-indexer.js.map +0 -1
- package/dist/core/indexer.d.ts +0 -3
- package/dist/core/indexer.d.ts.map +0 -1
- package/dist/core/indexer.js +0 -624
- package/dist/core/indexer.js.map +0 -1
- package/dist/core/metadata.d.ts +0 -19
- package/dist/core/metadata.d.ts.map +0 -1
- package/dist/core/metadata.js +0 -161
- package/dist/core/metadata.js.map +0 -1
- package/dist/core/search.d.ts +0 -7
- package/dist/core/search.d.ts.map +0 -1
- package/dist/core/search.js +0 -542
- package/dist/core/search.js.map +0 -1
- package/dist/core/symbol-extractor.d.ts +0 -3
- package/dist/core/symbol-extractor.d.ts.map +0 -1
- package/dist/core/symbol-extractor.js +0 -78
- package/dist/core/symbol-extractor.js.map +0 -1
- package/dist/core/types.d.ts +0 -104
- package/dist/core/types.d.ts.map +0 -1
- package/dist/core/types.js +0 -2
- package/dist/core/types.js.map +0 -1
- package/dist/database/db.d.ts +0 -63
- package/dist/database/db.d.ts.map +0 -1
- package/dist/database/db.js +0 -205
- package/dist/database/db.js.map +0 -1
- package/dist/indexer/merkle.d.ts +0 -13
- package/dist/indexer/merkle.d.ts.map +0 -1
- package/dist/indexer/merkle.js +0 -86
- package/dist/indexer/merkle.js.map +0 -1
- package/dist/indexer/update.d.ts +0 -19
- package/dist/indexer/update.d.ts.map +0 -1
- package/dist/indexer/update.js +0 -40
- package/dist/indexer/update.js.map +0 -1
- package/dist/indexer/watch.d.ts +0 -21
- package/dist/indexer/watch.d.ts.map +0 -1
- package/dist/indexer/watch.js +0 -222
- package/dist/indexer/watch.js.map +0 -1
- package/dist/languages/rules.d.ts +0 -11
- package/dist/languages/rules.d.ts.map +0 -1
- package/dist/languages/rules.js +0 -371
- package/dist/languages/rules.js.map +0 -1
- package/dist/languages/tree-sitter-loader.d.ts +0 -27
- package/dist/languages/tree-sitter-loader.d.ts.map +0 -1
- package/dist/languages/tree-sitter-loader.js +0 -76
- package/dist/languages/tree-sitter-loader.js.map +0 -1
- package/dist/mcp/tools/ask-codebase.d.ts +0 -85
- package/dist/mcp/tools/ask-codebase.d.ts.map +0 -1
- package/dist/mcp/tools/ask-codebase.js +0 -125
- package/dist/mcp/tools/ask-codebase.js.map +0 -1
- package/dist/mcp/tools/use-context-pack.d.ts +0 -57
- package/dist/mcp/tools/use-context-pack.d.ts.map +0 -1
- package/dist/mcp/tools/use-context-pack.js +0 -91
- package/dist/mcp/tools/use-context-pack.js.map +0 -1
- package/dist/mcp-server.d.ts +0 -3
- package/dist/mcp-server.d.ts.map +0 -1
- package/dist/mcp-server.js +0 -518
- package/dist/mcp-server.js.map +0 -1
- package/dist/providers/base.d.ts +0 -39
- package/dist/providers/base.d.ts.map +0 -1
- package/dist/providers/base.js +0 -199
- package/dist/providers/base.js.map +0 -1
- package/dist/providers/chat-llm.d.ts +0 -41
- package/dist/providers/chat-llm.d.ts.map +0 -1
- package/dist/providers/chat-llm.js +0 -155
- package/dist/providers/chat-llm.js.map +0 -1
- package/dist/providers/index.d.ts +0 -6
- package/dist/providers/index.d.ts.map +0 -1
- package/dist/providers/index.js +0 -23
- package/dist/providers/index.js.map +0 -1
- package/dist/providers/ollama.d.ts +0 -13
- package/dist/providers/ollama.d.ts.map +0 -1
- package/dist/providers/ollama.js +0 -50
- package/dist/providers/ollama.js.map +0 -1
- package/dist/providers/openai.d.ts +0 -14
- package/dist/providers/openai.d.ts.map +0 -1
- package/dist/providers/openai.js +0 -122
- package/dist/providers/openai.js.map +0 -1
- package/dist/providers/token-counter.d.ts +0 -2
- package/dist/providers/token-counter.d.ts.map +0 -1
- package/dist/providers/token-counter.js +0 -18
- package/dist/providers/token-counter.js.map +0 -1
- package/dist/ranking/api-reranker.d.ts +0 -18
- package/dist/ranking/api-reranker.d.ts.map +0 -1
- package/dist/ranking/api-reranker.js +0 -134
- package/dist/ranking/api-reranker.js.map +0 -1
- package/dist/ranking/symbol-boost.d.ts +0 -15
- package/dist/ranking/symbol-boost.d.ts.map +0 -1
- package/dist/ranking/symbol-boost.js +0 -154
- package/dist/ranking/symbol-boost.js.map +0 -1
- package/dist/search/bm25.d.ts +0 -17
- package/dist/search/bm25.d.ts.map +0 -1
- package/dist/search/bm25.js +0 -56
- package/dist/search/bm25.js.map +0 -1
- package/dist/search/hybrid.d.ts +0 -21
- package/dist/search/hybrid.d.ts.map +0 -1
- package/dist/search/hybrid.js +0 -50
- package/dist/search/hybrid.js.map +0 -1
- package/dist/search/scope.d.ts +0 -5
- package/dist/search/scope.d.ts.map +0 -1
- package/dist/search/scope.js +0 -107
- package/dist/search/scope.js.map +0 -1
- package/dist/storage/encrypted-chunks.d.ts +0 -40
- package/dist/storage/encrypted-chunks.d.ts.map +0 -1
- package/dist/storage/encrypted-chunks.js +0 -237
- package/dist/storage/encrypted-chunks.js.map +0 -1
- package/dist/symbols/extract.d.ts +0 -15
- package/dist/symbols/extract.d.ts.map +0 -1
- package/dist/symbols/extract.js +0 -187
- package/dist/symbols/extract.js.map +0 -1
- package/dist/symbols/graph.d.ts +0 -3
- package/dist/symbols/graph.d.ts.map +0 -1
- package/dist/symbols/graph.js +0 -89
- package/dist/symbols/graph.js.map +0 -1
- package/dist/synthesis/markdown-formatter.d.ts +0 -13
- package/dist/synthesis/markdown-formatter.d.ts.map +0 -1
- package/dist/synthesis/markdown-formatter.js +0 -104
- package/dist/synthesis/markdown-formatter.js.map +0 -1
- package/dist/synthesis/prompt-builder.d.ts +0 -21
- package/dist/synthesis/prompt-builder.d.ts.map +0 -1
- package/dist/synthesis/prompt-builder.js +0 -129
- package/dist/synthesis/prompt-builder.js.map +0 -1
- package/dist/synthesis/synthesizer.d.ts +0 -30
- package/dist/synthesis/synthesizer.d.ts.map +0 -1
- package/dist/synthesis/synthesizer.js +0 -210
- package/dist/synthesis/synthesizer.js.map +0 -1
- package/dist/types/ast.d.ts +0 -3
- package/dist/types/ast.d.ts.map +0 -1
- package/dist/types/ast.js +0 -2
- package/dist/types/ast.js.map +0 -1
- package/dist/types/codemap.d.ts +0 -58
- package/dist/types/codemap.d.ts.map +0 -1
- package/dist/types/codemap.js +0 -224
- package/dist/types/codemap.js.map +0 -1
- package/dist/types/context-pack.d.ts +0 -47
- package/dist/types/context-pack.d.ts.map +0 -1
- package/dist/types/context-pack.js +0 -44
- package/dist/types/context-pack.js.map +0 -1
- package/dist/types/search.d.ts +0 -15
- package/dist/types/search.d.ts.map +0 -1
- package/dist/types/search.js +0 -11
- package/dist/types/search.js.map +0 -1
- package/dist/utils/cli-ui.d.ts +0 -44
- package/dist/utils/cli-ui.d.ts.map +0 -1
- package/dist/utils/cli-ui.js +0 -139
- package/dist/utils/cli-ui.js.map +0 -1
- package/dist/utils/indexer-with-progress.d.ts +0 -10
- package/dist/utils/indexer-with-progress.d.ts.map +0 -1
- package/dist/utils/indexer-with-progress.js +0 -58
- package/dist/utils/indexer-with-progress.js.map +0 -1
- package/dist/utils/rate-limiter.d.ts +0 -34
- package/dist/utils/rate-limiter.d.ts.map +0 -1
- package/dist/utils/rate-limiter.js +0 -179
- package/dist/utils/rate-limiter.js.map +0 -1
|
@@ -1,181 +0,0 @@
|
|
|
1
|
-
import { batchAnalyzeCodeSize } from './token-counter.js';
|
|
2
|
-
function getSizeLimits(profile) {
|
|
3
|
-
if (profile.useTokens && profile.tokenCounter) {
|
|
4
|
-
return {
|
|
5
|
-
optimal: profile.optimalTokens,
|
|
6
|
-
min: profile.minChunkTokens,
|
|
7
|
-
max: profile.maxChunkTokens,
|
|
8
|
-
overlap: profile.overlapTokens,
|
|
9
|
-
unit: 'tokens'
|
|
10
|
-
};
|
|
11
|
-
}
|
|
12
|
-
return {
|
|
13
|
-
optimal: profile.optimalChars,
|
|
14
|
-
min: profile.minChunkChars,
|
|
15
|
-
max: profile.maxChunkChars,
|
|
16
|
-
overlap: profile.overlapChars,
|
|
17
|
-
unit: 'characters'
|
|
18
|
-
};
|
|
19
|
-
}
|
|
20
|
-
async function batchAnalyzeNodesInternal(nodes, source, profile) {
|
|
21
|
-
const codes = nodes.map(node => source.slice(node.startIndex, node.endIndex));
|
|
22
|
-
const limits = getSizeLimits(profile);
|
|
23
|
-
if (profile.useTokens && profile.tokenCounter) {
|
|
24
|
-
const analyses = await batchAnalyzeCodeSize(codes, limits, profile.tokenCounter, false);
|
|
25
|
-
return nodes.map((node, i) => ({
|
|
26
|
-
node,
|
|
27
|
-
size: analyses[i].size,
|
|
28
|
-
code: codes[i]
|
|
29
|
-
}));
|
|
30
|
-
}
|
|
31
|
-
return nodes.map((node, i) => ({
|
|
32
|
-
node,
|
|
33
|
-
size: codes[i].length,
|
|
34
|
-
code: codes[i]
|
|
35
|
-
}));
|
|
36
|
-
}
|
|
37
|
-
function isContainerNode(node, rule) {
|
|
38
|
-
const containerTypes = [
|
|
39
|
-
'class_declaration',
|
|
40
|
-
'class_definition',
|
|
41
|
-
'interface_declaration',
|
|
42
|
-
'module_declaration',
|
|
43
|
-
'namespace_declaration',
|
|
44
|
-
'trait_declaration',
|
|
45
|
-
'enum_declaration'
|
|
46
|
-
];
|
|
47
|
-
return containerTypes.includes(node.type);
|
|
48
|
-
}
|
|
49
|
-
function identifySemanticGroups(nodes, source, nodeAnalyses, rule) {
|
|
50
|
-
const groups = [];
|
|
51
|
-
let currentGroup = {
|
|
52
|
-
type: 'file_section',
|
|
53
|
-
nodes: [],
|
|
54
|
-
analyses: [],
|
|
55
|
-
parentNode: null
|
|
56
|
-
};
|
|
57
|
-
for (let i = 0; i < nodes.length; i++) {
|
|
58
|
-
const node = nodes[i];
|
|
59
|
-
const analysis = nodeAnalyses[i];
|
|
60
|
-
if (isContainerNode(node, rule)) {
|
|
61
|
-
if (currentGroup.nodes.length > 0) {
|
|
62
|
-
groups.push(currentGroup);
|
|
63
|
-
}
|
|
64
|
-
currentGroup = {
|
|
65
|
-
type: 'container',
|
|
66
|
-
containerType: node.type,
|
|
67
|
-
nodes: [node],
|
|
68
|
-
analyses: [analysis],
|
|
69
|
-
parentNode: node
|
|
70
|
-
};
|
|
71
|
-
groups.push(currentGroup);
|
|
72
|
-
currentGroup = {
|
|
73
|
-
type: 'file_section',
|
|
74
|
-
nodes: [],
|
|
75
|
-
analyses: [],
|
|
76
|
-
parentNode: null
|
|
77
|
-
};
|
|
78
|
-
}
|
|
79
|
-
else {
|
|
80
|
-
currentGroup.nodes.push(node);
|
|
81
|
-
currentGroup.analyses.push(analysis);
|
|
82
|
-
}
|
|
83
|
-
}
|
|
84
|
-
if (currentGroup.nodes.length > 0) {
|
|
85
|
-
groups.push(currentGroup);
|
|
86
|
-
}
|
|
87
|
-
return groups;
|
|
88
|
-
}
|
|
89
|
-
async function combineGroupsToOptimalSize(semanticGroups, source, profile, limits) {
|
|
90
|
-
const optimalGroups = [];
|
|
91
|
-
let currentCombinedGroup = {
|
|
92
|
-
nodes: [],
|
|
93
|
-
totalSize: 0,
|
|
94
|
-
groupInfo: []
|
|
95
|
-
};
|
|
96
|
-
for (const group of semanticGroups) {
|
|
97
|
-
const groupTotalSize = group.analyses.reduce((sum, a) => sum + a.size, 0);
|
|
98
|
-
if (groupTotalSize > limits.optimal) {
|
|
99
|
-
if (currentCombinedGroup.nodes.length > 0) {
|
|
100
|
-
optimalGroups.push(currentCombinedGroup);
|
|
101
|
-
}
|
|
102
|
-
optimalGroups.push({
|
|
103
|
-
nodes: group.nodes,
|
|
104
|
-
totalSize: groupTotalSize,
|
|
105
|
-
groupInfo: [group]
|
|
106
|
-
});
|
|
107
|
-
currentCombinedGroup = {
|
|
108
|
-
nodes: [],
|
|
109
|
-
totalSize: 0,
|
|
110
|
-
groupInfo: []
|
|
111
|
-
};
|
|
112
|
-
continue;
|
|
113
|
-
}
|
|
114
|
-
if (currentCombinedGroup.totalSize + groupTotalSize > limits.max) {
|
|
115
|
-
if (currentCombinedGroup.nodes.length > 0) {
|
|
116
|
-
optimalGroups.push(currentCombinedGroup);
|
|
117
|
-
}
|
|
118
|
-
currentCombinedGroup = {
|
|
119
|
-
nodes: group.nodes,
|
|
120
|
-
totalSize: groupTotalSize,
|
|
121
|
-
groupInfo: [group]
|
|
122
|
-
};
|
|
123
|
-
continue;
|
|
124
|
-
}
|
|
125
|
-
currentCombinedGroup.nodes.push(...group.nodes);
|
|
126
|
-
currentCombinedGroup.totalSize += groupTotalSize;
|
|
127
|
-
currentCombinedGroup.groupInfo.push(group);
|
|
128
|
-
if (currentCombinedGroup.totalSize >= limits.optimal * 0.9) {
|
|
129
|
-
optimalGroups.push(currentCombinedGroup);
|
|
130
|
-
currentCombinedGroup = {
|
|
131
|
-
nodes: [],
|
|
132
|
-
totalSize: 0,
|
|
133
|
-
groupInfo: []
|
|
134
|
-
};
|
|
135
|
-
}
|
|
136
|
-
}
|
|
137
|
-
if (currentCombinedGroup.nodes.length > 0) {
|
|
138
|
-
optimalGroups.push(currentCombinedGroup);
|
|
139
|
-
}
|
|
140
|
-
return optimalGroups;
|
|
141
|
-
}
|
|
142
|
-
export async function groupNodesForChunking(nodes, source, profile, rule) {
|
|
143
|
-
if (!nodes || nodes.length === 0)
|
|
144
|
-
return [];
|
|
145
|
-
const limits = getSizeLimits(profile);
|
|
146
|
-
if (nodes.length <= 10) {
|
|
147
|
-
return nodes.map(node => ({
|
|
148
|
-
nodes: [node],
|
|
149
|
-
totalSize: 0,
|
|
150
|
-
groupInfo: []
|
|
151
|
-
}));
|
|
152
|
-
}
|
|
153
|
-
const nodeAnalyses = await batchAnalyzeNodesInternal(nodes, source, profile);
|
|
154
|
-
const semanticGroups = identifySemanticGroups(nodes, source, nodeAnalyses, rule);
|
|
155
|
-
const optimalGroups = await combineGroupsToOptimalSize(semanticGroups, source, profile, limits);
|
|
156
|
-
return optimalGroups;
|
|
157
|
-
}
|
|
158
|
-
export function createCombinedChunk(nodeGroup, source, filerel) {
|
|
159
|
-
if (!nodeGroup.nodes || nodeGroup.nodes.length === 0) {
|
|
160
|
-
return null;
|
|
161
|
-
}
|
|
162
|
-
const codes = nodeGroup.nodes.map(node => source.slice(node.startIndex, node.endIndex));
|
|
163
|
-
const combinedCode = codes.join('\n\n');
|
|
164
|
-
const firstNode = nodeGroup.nodes[0];
|
|
165
|
-
const lastNode = nodeGroup.nodes[nodeGroup.nodes.length - 1];
|
|
166
|
-
return {
|
|
167
|
-
code: combinedCode,
|
|
168
|
-
node: {
|
|
169
|
-
...firstNode,
|
|
170
|
-
type: `${firstNode.type}_group_${nodeGroup.nodes.length}`,
|
|
171
|
-
endIndex: lastNode.endIndex
|
|
172
|
-
},
|
|
173
|
-
metadata: {
|
|
174
|
-
isGroup: true,
|
|
175
|
-
nodeCount: nodeGroup.nodes.length,
|
|
176
|
-
totalSize: nodeGroup.totalSize,
|
|
177
|
-
groupTypes: nodeGroup.groupInfo?.map(g => g.type) || ['combined']
|
|
178
|
-
}
|
|
179
|
-
};
|
|
180
|
-
}
|
|
181
|
-
//# sourceMappingURL=file-grouper.js.map
|
|
@@ -1 +0,0 @@
|
|
|
1
|
-
{"version":3,"file":"file-grouper.js","sourceRoot":"","sources":["../../src/chunking/file-grouper.ts"],"names":[],"mappings":"AAAA,OAAO,EAAmB,oBAAoB,EAAyB,MAAM,oBAAoB,CAAC;AAqClG,SAAS,aAAa,CAAC,OAAqB;IAC1C,IAAI,OAAO,CAAC,SAAS,IAAI,OAAO,CAAC,YAAY,EAAE,CAAC;QAC9C,OAAO;YACL,OAAO,EAAE,OAAO,CAAC,aAAa;YAC9B,GAAG,EAAE,OAAO,CAAC,cAAc;YAC3B,GAAG,EAAE,OAAO,CAAC,cAAc;YAC3B,OAAO,EAAE,OAAO,CAAC,aAAa;YAC9B,IAAI,EAAE,QAAQ;SACf,CAAC;IACJ,CAAC;IACD,OAAO;QACL,OAAO,EAAE,OAAO,CAAC,YAAY;QAC7B,GAAG,EAAE,OAAO,CAAC,aAAa;QAC1B,GAAG,EAAE,OAAO,CAAC,aAAa;QAC1B,OAAO,EAAE,OAAO,CAAC,YAAY;QAC7B,IAAI,EAAE,YAAY;KACnB,CAAC;AACJ,CAAC;AAED,KAAK,UAAU,yBAAyB,CAAC,KAAuB,EAAE,MAAc,EAAE,OAAqB;IACrG,MAAM,KAAK,GAAG,KAAK,CAAC,GAAG,CAAC,IAAI,CAAC,EAAE,CAAC,MAAM,CAAC,KAAK,CAAC,IAAI,CAAC,UAAU,EAAE,IAAI,CAAC,QAAQ,CAAC,CAAC,CAAC;IAC9E,MAAM,MAAM,GAAG,aAAa,CAAC,OAAO,CAAC,CAAC;IAEtC,IAAI,OAAO,CAAC,SAAS,IAAI,OAAO,CAAC,YAAY,EAAE,CAAC;QAC9C,MAAM,QAAQ,GAAG,MAAM,oBAAoB,CAAC,KAAK,EAAE,MAAM,EAAE,OAAO,CAAC,YAAY,EAAE,KAAK,CAAC,CAAC;QACxF,OAAO,KAAK,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC;YAC7B,IAAI;YACJ,IAAI,EAAE,QAAQ,CAAC,CAAC,CAAC,CAAC,IAAI;YACtB,IAAI,EAAE,KAAK,CAAC,CAAC,CAAC;SACf,CAAC,CAAC,CAAC;IACN,CAAC;IAED,OAAO,KAAK,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC;QAC7B,IAAI;QACJ,IAAI,EAAE,KAAK,CAAC,CAAC,CAAC,CAAC,MAAM;QACrB,IAAI,EAAE,KAAK,CAAC,CAAC,CAAC;KACf,CAAC,CAAC,CAAC;AACN,CAAC;AAED,SAAS,eAAe,CAAC,IAAoB,EAAE,IAAkB;IAC/D,MAAM,cAAc,GAAG;QACrB,mBAAmB;QACnB,kBAAkB;QAClB,uBAAuB;QACvB,oBAAoB;QACpB,uBAAuB;QACvB,mBAAmB;QACnB,kBAAkB;KACnB,CAAC;IAEF,OAAO,cAAc,CAAC,QAAQ,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;AAC5C,CAAC;AAED,SAAS,sBAAsB,CAAC,KAAuB,EAAE,MAAc,EAAE,YAA4B,EAAE,IAAkB;IACvH,MAAM,MAAM,GAAoB,EAAE,CAAC;IACnC,IAAI,YAAY,GAAkB;QAChC,IAAI,EAAE,cAAc;QACpB,KAAK,EAAE,EAAE;QACT,QAAQ,EAAE,EAAE;QACZ,UAAU,EAAE,IAAI;KACjB,CAAC;IAEF,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,KAAK,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC;QACtC,MAAM,IAAI,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC;QACtB,MAAM,QAAQ,GAAG,YAAY,CAAC,CAAC,CAAC,CAAC;QAEjC,IAAI,eAAe,CAAC,IAAI,EAAE,IAAI,CAAC,EAAE,CAAC;YAChC,IAAI,YAAY,CAAC,KAAK,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;gBAClC,MAAM,CAAC,IAAI,CAAC,YAAY,CAAC,CAAC;YAC5B,CAAC;YAED,YAAY,GAAG;gBACb,IAAI,EAAE,WAAW;gBACjB,aAAa,EAAE,IAAI,CAAC,IAAI;gBACxB,KAAK,EAAE,CAAC,IAAI,CAAC;gBACb,QAAQ,EAAE,CAAC,QAAQ,CAAC;gBACpB,UAAU,EAAE,IAAI;aACjB,CAAC;YAEF,MAAM,CAAC,IAAI,CAAC,YAAY,CAAC,CAAC;YAE1B,YAAY,GAAG;gBACb,IAAI,EAAE,cAAc;gBACpB,KAAK,EAAE,EAAE;gBACT,QAAQ,EAAE,EAAE;gBACZ,UAAU,EAAE,IAAI;aACjB,CAAC;QACJ,CAAC;aAAM,CAAC;YACN,YAAY,CAAC,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;YAC9B,YAAY,CAAC,QAAQ,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC;QACvC,CAAC;IACH,CAAC;IAED,IAAI,YAAY,CAAC,KAAK,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;QAClC,MAAM,CAAC,IAAI,CAAC,YAAY,CAAC,CAAC;IAC5B,CAAC;IAED,OAAO,MAAM,CAAC;AAChB,CAAC;AAED,KAAK,UAAU,0BAA0B,CACvC,cAA+B,EAC/B,MAAc,EACd,OAAqB,EACrB,MAAkB;IAElB,MAAM,aAAa,GAAgB,EAAE,CAAC;IACtC,IAAI,oBAAoB,GAAc;QACpC,KAAK,EAAE,EAAE;QACT,SAAS,EAAE,CAAC;QACZ,SAAS,EAAE,EAAE;KACd,CAAC;IAEF,KAAK,MAAM,KAAK,IAAI,cAAc,EAAE,CAAC;QACnC,MAAM,cAAc,GAAG,KAAK,CAAC,QAAQ,CAAC,MAAM,CAAC,CAAC,GAAG,EAAE,CAAC,EAAE,EAAE,CAAC,GAAG,GAAG,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC,CAAC;QAE1E,IAAI,cAAc,GAAG,MAAM,CAAC,OAAO,EAAE,CAAC;YACpC,IAAI,oBAAoB,CAAC,KAAK,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;gBAC1C,aAAa,CAAC,IAAI,CAAC,oBAAoB,CAAC,CAAC;YAC3C,CAAC;YAED,aAAa,CAAC,IAAI,CAAC;gBACjB,KAAK,EAAE,KAAK,CAAC,KAAK;gBAClB,SAAS,EAAE,cAAc;gBACzB,SAAS,EAAE,CAAC,KAAK,CAAC;aACnB,CAAC,CAAC;YAEH,oBAAoB,GAAG;gBACrB,KAAK,EAAE,EAAE;gBACT,SAAS,EAAE,CAAC;gBACZ,SAAS,EAAE,EAAE;aACd,CAAC;YACF,SAAS;QACX,CAAC;QAED,IAAI,oBAAoB,CAAC,SAAS,GAAG,cAAc,GAAG,MAAM,CAAC,GAAG,EAAE,CAAC;YACjE,IAAI,oBAAoB,CAAC,KAAK,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;gBAC1C,aAAa,CAAC,IAAI,CAAC,oBAAoB,CAAC,CAAC;YAC3C,CAAC;YAED,oBAAoB,GAAG;gBACrB,KAAK,EAAE,KAAK,CAAC,KAAK;gBAClB,SAAS,EAAE,cAAc;gBACzB,SAAS,EAAE,CAAC,KAAK,CAAC;aACnB,CAAC;YACF,SAAS;QACX,CAAC;QAED,oBAAoB,CAAC,KAAK,CAAC,IAAI,CAAC,GAAG,KAAK,CAAC,KAAK,CAAC,CAAC;QAChD,oBAAoB,CAAC,SAAS,IAAI,cAAc,CAAC;QACjD,oBAAoB,CAAC,SAAS,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC;QAE3C,IAAI,oBAAoB,CAAC,SAAS,IAAI,MAAM,CAAC,OAAO,GAAG,GAAG,EAAE,CAAC;YAC3D,aAAa,CAAC,IAAI,CAAC,oBAAoB,CAAC,CAAC;YACzC,oBAAoB,GAAG;gBACrB,KAAK,EAAE,EAAE;gBACT,SAAS,EAAE,CAAC;gBACZ,SAAS,EAAE,EAAE;aACd,CAAC;QACJ,CAAC;IACH,CAAC;IAED,IAAI,oBAAoB,CAAC,KAAK,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;QAC1C,aAAa,CAAC,IAAI,CAAC,oBAAoB,CAAC,CAAC;IAC3C,CAAC;IAED,OAAO,aAAa,CAAC;AACvB,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,qBAAqB,CACzC,KAAuB,EACvB,MAAc,EACd,OAAqB,EACrB,IAAkB;IAElB,IAAI,CAAC,KAAK,IAAI,KAAK,CAAC,MAAM,KAAK,CAAC;QAAE,OAAO,EAAE,CAAC;IAE5C,MAAM,MAAM,GAAG,aAAa,CAAC,OAAO,CAAC,CAAC;IAEtC,IAAI,KAAK,CAAC,MAAM,IAAI,EAAE,EAAE,CAAC;QACvB,OAAO,KAAK,CAAC,GAAG,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC;YACxB,KAAK,EAAE,CAAC,IAAI,CAAC;YACb,SAAS,EAAE,CAAC;YACZ,SAAS,EAAE,EAAE;SACd,CAAC,CAAC,CAAC;IACN,CAAC;IAED,MAAM,YAAY,GAAG,MAAM,yBAAyB,CAAC,KAAK,EAAE,MAAM,EAAE,OAAO,CAAC,CAAC;IAC7E,MAAM,cAAc,GAAG,sBAAsB,CAAC,KAAK,EAAE,MAAM,EAAE,YAAY,EAAE,IAAI,CAAC,CAAC;IACjF,MAAM,aAAa,GAAG,MAAM,0BAA0B,CAAC,cAAc,EAAE,MAAM,EAAE,OAAO,EAAE,MAAM,CAAC,CAAC;IAEhG,OAAO,aAAa,CAAC;AACvB,CAAC;AAaD,MAAM,UAAU,mBAAmB,CAAC,SAAoB,EAAE,MAAc,EAAE,OAAe;IACvF,IAAI,CAAC,SAAS,CAAC,KAAK,IAAI,SAAS,CAAC,KAAK,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;QACrD,OAAO,IAAI,CAAC;IACd,CAAC;IAED,MAAM,KAAK,GAAG,SAAS,CAAC,KAAK,CAAC,GAAG,CAAC,IAAI,CAAC,EAAE,CACvC,MAAM,CAAC,KAAK,CAAC,IAAI,CAAC,UAAU,EAAE,IAAI,CAAC,QAAQ,CAAC,CAC7C,CAAC;IAEF,MAAM,YAAY,GAAG,KAAK,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;IAExC,MAAM,SAAS,GAAG,SAAS,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC;IACrC,MAAM,QAAQ,GAAG,SAAS,CAAC,KAAK,CAAC,SAAS,CAAC,KAAK,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC;IAE7D,OAAO;QACL,IAAI,EAAE,YAAY;QAClB,IAAI,EAAE;YACJ,GAAG,SAAS;YACZ,IAAI,EAAE,GAAG,SAAS,CAAC,IAAI,UAAU,SAAS,CAAC,KAAK,CAAC,MAAM,EAAE;YACzD,QAAQ,EAAE,QAAQ,CAAC,QAAQ;SAC5B;QACD,QAAQ,EAAE;YACR,OAAO,EAAE,IAAI;YACb,SAAS,EAAE,SAAS,CAAC,KAAK,CAAC,MAAM;YACjC,SAAS,EAAE,SAAS,CAAC,SAAS;YAC9B,UAAU,EAAE,SAAS,CAAC,SAAS,EAAE,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,UAAU,CAAC;SAClE;KACF,CAAC;AACJ,CAAC"}
|
|
@@ -1,37 +0,0 @@
|
|
|
1
|
-
import type { ModelProfile } from '../providers/base.js';
|
|
2
|
-
import type { TreeSitterNode } from '../types/ast.js';
|
|
3
|
-
interface LanguageRule {
|
|
4
|
-
subdivisionTypes?: Record<string, string[]>;
|
|
5
|
-
[key: string]: any;
|
|
6
|
-
}
|
|
7
|
-
export declare function findSemanticSubdivisions(node: TreeSitterNode, rule: LanguageRule): TreeSitterNode[];
|
|
8
|
-
export declare function findLastCompleteBoundary(code: string, maxSize: number): number;
|
|
9
|
-
export declare function extractSignature(node: TreeSitterNode, source: string): string;
|
|
10
|
-
export declare function extractLinesBeforeNode(node: TreeSitterNode, source: string, numLines: number): string;
|
|
11
|
-
export declare function extractParentContext(node: TreeSitterNode, source: string): {
|
|
12
|
-
signature: string;
|
|
13
|
-
startLine: number;
|
|
14
|
-
endLine: number;
|
|
15
|
-
};
|
|
16
|
-
export declare function getLineNumber(byteOffset: number, source: string): number;
|
|
17
|
-
export interface NodeAnalysis {
|
|
18
|
-
isSingleChunk: boolean;
|
|
19
|
-
needsSubdivision: boolean;
|
|
20
|
-
subdivisionCandidates: TreeSitterNode[];
|
|
21
|
-
size: number;
|
|
22
|
-
unit: string;
|
|
23
|
-
method: string;
|
|
24
|
-
estimatedSubchunks: number;
|
|
25
|
-
}
|
|
26
|
-
export declare function analyzeNodeForChunking(node: TreeSitterNode, source: string, rule: LanguageRule, profile: ModelProfile): Promise<NodeAnalysis>;
|
|
27
|
-
export declare function batchAnalyzeNodes(nodes: TreeSitterNode[], source: string, rule: LanguageRule, profile: ModelProfile, isSubdivision?: boolean): Promise<Array<NodeAnalysis & {
|
|
28
|
-
node: TreeSitterNode;
|
|
29
|
-
}>>;
|
|
30
|
-
export interface StatementChunk {
|
|
31
|
-
code: string;
|
|
32
|
-
size: number;
|
|
33
|
-
unit: string;
|
|
34
|
-
}
|
|
35
|
-
export declare function yieldStatementChunks(node: TreeSitterNode, source: string, maxSize: number, overlapSize: number, profile: ModelProfile): Promise<StatementChunk[]>;
|
|
36
|
-
export {};
|
|
37
|
-
//# sourceMappingURL=semantic-chunker.d.ts.map
|
|
@@ -1 +0,0 @@
|
|
|
1
|
-
{"version":3,"file":"semantic-chunker.d.ts","sourceRoot":"","sources":["../../src/chunking/semantic-chunker.ts"],"names":[],"mappings":"AACA,OAAO,KAAK,EAAE,YAAY,EAAE,MAAM,sBAAsB,CAAC;AACzD,OAAO,KAAK,EAAE,cAAc,EAAE,MAAM,iBAAiB,CAAC;AAEtD,UAAU,YAAY;IACpB,gBAAgB,CAAC,EAAE,MAAM,CAAC,MAAM,EAAE,MAAM,EAAE,CAAC,CAAC;IAC5C,CAAC,GAAG,EAAE,MAAM,GAAG,GAAG,CAAC;CACpB;AAUD,wBAAgB,wBAAwB,CAAC,IAAI,EAAE,cAAc,EAAE,IAAI,EAAE,YAAY,GAAG,cAAc,EAAE,CAwBnG;AAED,wBAAgB,wBAAwB,CAAC,IAAI,EAAE,MAAM,EAAE,OAAO,EAAE,MAAM,GAAG,MAAM,CAgB9E;AAED,wBAAgB,gBAAgB,CAAC,IAAI,EAAE,cAAc,EAAE,MAAM,EAAE,MAAM,GAAG,MAAM,CAO7E;AAED,wBAAgB,sBAAsB,CAAC,IAAI,EAAE,cAAc,EAAE,MAAM,EAAE,MAAM,EAAE,QAAQ,EAAE,MAAM,GAAG,MAAM,CAIrG;AAED,wBAAgB,oBAAoB,CAAC,IAAI,EAAE,cAAc,EAAE,MAAM,EAAE,MAAM,GAAG;IAC1E,SAAS,EAAE,MAAM,CAAC;IAClB,SAAS,EAAE,MAAM,CAAC;IAClB,OAAO,EAAE,MAAM,CAAC;CACjB,CAMA;AAED,wBAAgB,aAAa,CAAC,UAAU,EAAE,MAAM,EAAE,MAAM,EAAE,MAAM,GAAG,MAAM,CAGxE;AAqBD,MAAM,WAAW,YAAY;IAC3B,aAAa,EAAE,OAAO,CAAC;IACvB,gBAAgB,EAAE,OAAO,CAAC;IAC1B,qBAAqB,EAAE,cAAc,EAAE,CAAC;IACxC,IAAI,EAAE,MAAM,CAAC;IACb,IAAI,EAAE,MAAM,CAAC;IACb,MAAM,EAAE,MAAM,CAAC;IACf,kBAAkB,EAAE,MAAM,CAAC;CAC5B;AAED,wBAAsB,sBAAsB,CAC1C,IAAI,EAAE,cAAc,EACpB,MAAM,EAAE,MAAM,EACd,IAAI,EAAE,YAAY,EAClB,OAAO,EAAE,YAAY,GACpB,OAAO,CAAC,YAAY,CAAC,CA2BvB;AAED,wBAAsB,iBAAiB,CACrC,KAAK,EAAE,cAAc,EAAE,EACvB,MAAM,EAAE,MAAM,EACd,IAAI,EAAE,YAAY,EAClB,OAAO,EAAE,YAAY,EACrB,aAAa,UAAQ,GACpB,OAAO,CAAC,KAAK,CAAC,YAAY,GAAG;IAAE,IAAI,EAAE,cAAc,CAAA;CAAE,CAAC,CAAC,CAgCzD;AAED,MAAM,WAAW,cAAc;IAC7B,IAAI,EAAE,MAAM,CAAC;IACb,IAAI,EAAE,MAAM,CAAC;IACb,IAAI,EAAE,MAAM,CAAC;CACd;AAED,wBAAsB,oBAAoB,CACxC,IAAI,EAAE,cAAc,EACpB,MAAM,EAAE,MAAM,EACd,OAAO,EAAE,MAAM,EACf,WAAW,EAAE,MAAM,EACnB,OAAO,EAAE,YAAY,GACpB,OAAO,CAAC,cAAc,EAAE,CAAC,CAwC3B"}
|
|
@@ -1,172 +0,0 @@
|
|
|
1
|
-
import { analyzeCodeSize, batchAnalyzeCodeSize } from './token-counter.js';
|
|
2
|
-
export function findSemanticSubdivisions(node, rule) {
|
|
3
|
-
if (!node || !rule)
|
|
4
|
-
return [];
|
|
5
|
-
const subdivisionTypes = rule.subdivisionTypes?.[node.type] || [];
|
|
6
|
-
if (subdivisionTypes.length === 0)
|
|
7
|
-
return [];
|
|
8
|
-
const candidates = [];
|
|
9
|
-
function walk(n, depth = 0) {
|
|
10
|
-
if (depth > 0 && subdivisionTypes.includes(n.type)) {
|
|
11
|
-
candidates.push(n);
|
|
12
|
-
return;
|
|
13
|
-
}
|
|
14
|
-
for (let i = 0; i < n.childCount; i++) {
|
|
15
|
-
const child = n.child(i);
|
|
16
|
-
if (child) {
|
|
17
|
-
walk(child, depth + 1);
|
|
18
|
-
}
|
|
19
|
-
}
|
|
20
|
-
}
|
|
21
|
-
walk(node);
|
|
22
|
-
return candidates;
|
|
23
|
-
}
|
|
24
|
-
export function findLastCompleteBoundary(code, maxSize) {
|
|
25
|
-
const boundaries = [
|
|
26
|
-
{ pattern: /\n\s*}\s*$/gm, priority: 1 },
|
|
27
|
-
{ pattern: /;\s*$/gm, priority: 2 },
|
|
28
|
-
{ pattern: /\n\s*$/gm, priority: 3 }
|
|
29
|
-
];
|
|
30
|
-
for (const boundary of boundaries) {
|
|
31
|
-
const matches = [...code.substring(0, maxSize).matchAll(boundary.pattern)];
|
|
32
|
-
if (matches.length > 0) {
|
|
33
|
-
const lastMatch = matches[matches.length - 1];
|
|
34
|
-
return lastMatch.index + lastMatch[0].length;
|
|
35
|
-
}
|
|
36
|
-
}
|
|
37
|
-
return maxSize;
|
|
38
|
-
}
|
|
39
|
-
export function extractSignature(node, source) {
|
|
40
|
-
const code = source.slice(node.startIndex, node.endIndex);
|
|
41
|
-
const firstBrace = code.indexOf('{');
|
|
42
|
-
if (firstBrace !== -1) {
|
|
43
|
-
return code.substring(0, firstBrace).trim() + ' {';
|
|
44
|
-
}
|
|
45
|
-
return code.split('\n')[0];
|
|
46
|
-
}
|
|
47
|
-
export function extractLinesBeforeNode(node, source, numLines) {
|
|
48
|
-
const beforeCode = source.substring(0, node.startIndex);
|
|
49
|
-
const lines = beforeCode.split('\n');
|
|
50
|
-
return lines.slice(-numLines).join('\n');
|
|
51
|
-
}
|
|
52
|
-
export function extractParentContext(node, source) {
|
|
53
|
-
return {
|
|
54
|
-
signature: extractSignature(node, source),
|
|
55
|
-
startLine: getLineNumber(node.startIndex, source),
|
|
56
|
-
endLine: getLineNumber(node.endIndex, source)
|
|
57
|
-
};
|
|
58
|
-
}
|
|
59
|
-
export function getLineNumber(byteOffset, source) {
|
|
60
|
-
const before = source.substring(0, byteOffset);
|
|
61
|
-
return before.split('\n').length;
|
|
62
|
-
}
|
|
63
|
-
function getSizeLimits(profile) {
|
|
64
|
-
if (profile.useTokens && profile.tokenCounter) {
|
|
65
|
-
return {
|
|
66
|
-
optimal: profile.optimalTokens,
|
|
67
|
-
min: profile.minChunkTokens,
|
|
68
|
-
max: profile.maxChunkTokens,
|
|
69
|
-
overlap: profile.overlapTokens,
|
|
70
|
-
unit: 'tokens'
|
|
71
|
-
};
|
|
72
|
-
}
|
|
73
|
-
return {
|
|
74
|
-
optimal: profile.optimalChars,
|
|
75
|
-
min: profile.minChunkChars,
|
|
76
|
-
max: profile.maxChunkChars,
|
|
77
|
-
overlap: profile.overlapChars,
|
|
78
|
-
unit: 'characters'
|
|
79
|
-
};
|
|
80
|
-
}
|
|
81
|
-
export async function analyzeNodeForChunking(node, source, rule, profile) {
|
|
82
|
-
const code = source.slice(node.startIndex, node.endIndex);
|
|
83
|
-
const limits = getSizeLimits(profile);
|
|
84
|
-
let actualSize;
|
|
85
|
-
let method;
|
|
86
|
-
if (profile.useTokens && profile.tokenCounter) {
|
|
87
|
-
const analysis = await analyzeCodeSize(code, limits, profile.tokenCounter);
|
|
88
|
-
actualSize = analysis.size;
|
|
89
|
-
method = analysis.method;
|
|
90
|
-
}
|
|
91
|
-
else {
|
|
92
|
-
actualSize = code.length;
|
|
93
|
-
method = 'chars';
|
|
94
|
-
}
|
|
95
|
-
const subdivisionThreshold = limits.max;
|
|
96
|
-
return {
|
|
97
|
-
isSingleChunk: actualSize <= subdivisionThreshold,
|
|
98
|
-
needsSubdivision: actualSize > subdivisionThreshold,
|
|
99
|
-
subdivisionCandidates: findSemanticSubdivisions(node, rule),
|
|
100
|
-
size: actualSize,
|
|
101
|
-
unit: limits.unit,
|
|
102
|
-
method,
|
|
103
|
-
estimatedSubchunks: Math.ceil(actualSize / limits.optimal)
|
|
104
|
-
};
|
|
105
|
-
}
|
|
106
|
-
export async function batchAnalyzeNodes(nodes, source, rule, profile, isSubdivision = false) {
|
|
107
|
-
const codes = nodes.map(node => source.slice(node.startIndex, node.endIndex));
|
|
108
|
-
const limits = getSizeLimits(profile);
|
|
109
|
-
let analyses;
|
|
110
|
-
if (profile.useTokens && profile.tokenCounter) {
|
|
111
|
-
analyses = await batchAnalyzeCodeSize(codes, limits, profile.tokenCounter, isSubdivision);
|
|
112
|
-
}
|
|
113
|
-
else {
|
|
114
|
-
analyses = codes.map(code => ({
|
|
115
|
-
size: code.length,
|
|
116
|
-
decision: code.length < limits.min ? 'too_small'
|
|
117
|
-
: code.length > limits.max ? 'too_large'
|
|
118
|
-
: code.length <= limits.optimal ? 'optimal'
|
|
119
|
-
: 'needs_tokenization',
|
|
120
|
-
method: 'chars'
|
|
121
|
-
}));
|
|
122
|
-
}
|
|
123
|
-
return nodes.map((node, i) => {
|
|
124
|
-
const analysis = analyses[i];
|
|
125
|
-
const subdivisionThreshold = limits.max;
|
|
126
|
-
return {
|
|
127
|
-
node,
|
|
128
|
-
isSingleChunk: analysis.size <= subdivisionThreshold,
|
|
129
|
-
needsSubdivision: analysis.size > subdivisionThreshold,
|
|
130
|
-
subdivisionCandidates: findSemanticSubdivisions(node, rule),
|
|
131
|
-
size: analysis.size,
|
|
132
|
-
unit: limits.unit,
|
|
133
|
-
method: analysis.method,
|
|
134
|
-
estimatedSubchunks: Math.ceil(analysis.size / limits.optimal)
|
|
135
|
-
};
|
|
136
|
-
});
|
|
137
|
-
}
|
|
138
|
-
export async function yieldStatementChunks(node, source, maxSize, overlapSize, profile) {
|
|
139
|
-
const code = source.slice(node.startIndex, node.endIndex);
|
|
140
|
-
const lines = code.split('\n');
|
|
141
|
-
const chunks = [];
|
|
142
|
-
let currentChunk = [];
|
|
143
|
-
let currentSize = 0;
|
|
144
|
-
for (const line of lines) {
|
|
145
|
-
const lineSize = profile.useTokens && profile.tokenCounter
|
|
146
|
-
? await profile.tokenCounter(line)
|
|
147
|
-
: line.length;
|
|
148
|
-
if (currentSize + lineSize > maxSize && currentChunk.length > 0) {
|
|
149
|
-
chunks.push({
|
|
150
|
-
code: currentChunk.join('\n'),
|
|
151
|
-
size: currentSize,
|
|
152
|
-
unit: profile.useTokens ? 'tokens' : 'characters'
|
|
153
|
-
});
|
|
154
|
-
const overlapLines = Math.floor(currentChunk.length * 0.2);
|
|
155
|
-
currentChunk = currentChunk.slice(-overlapLines);
|
|
156
|
-
currentSize = profile.useTokens && profile.tokenCounter
|
|
157
|
-
? await profile.tokenCounter(currentChunk.join('\n'))
|
|
158
|
-
: currentChunk.join('\n').length;
|
|
159
|
-
}
|
|
160
|
-
currentChunk.push(line);
|
|
161
|
-
currentSize += lineSize;
|
|
162
|
-
}
|
|
163
|
-
if (currentChunk.length > 0) {
|
|
164
|
-
chunks.push({
|
|
165
|
-
code: currentChunk.join('\n'),
|
|
166
|
-
size: currentSize,
|
|
167
|
-
unit: profile.useTokens ? 'tokens' : 'characters'
|
|
168
|
-
});
|
|
169
|
-
}
|
|
170
|
-
return chunks;
|
|
171
|
-
}
|
|
172
|
-
//# sourceMappingURL=semantic-chunker.js.map
|
|
@@ -1 +0,0 @@
|
|
|
1
|
-
{"version":3,"file":"semantic-chunker.js","sourceRoot":"","sources":["../../src/chunking/semantic-chunker.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,eAAe,EAAE,oBAAoB,EAAyB,MAAM,oBAAoB,CAAC;AAiBlG,MAAM,UAAU,wBAAwB,CAAC,IAAoB,EAAE,IAAkB;IAC/E,IAAI,CAAC,IAAI,IAAI,CAAC,IAAI;QAAE,OAAO,EAAE,CAAC;IAE9B,MAAM,gBAAgB,GAAG,IAAI,CAAC,gBAAgB,EAAE,CAAC,IAAI,CAAC,IAAI,CAAC,IAAI,EAAE,CAAC;IAClE,IAAI,gBAAgB,CAAC,MAAM,KAAK,CAAC;QAAE,OAAO,EAAE,CAAC;IAE7C,MAAM,UAAU,GAAqB,EAAE,CAAC;IAExC,SAAS,IAAI,CAAC,CAAiB,EAAE,KAAK,GAAG,CAAC;QACxC,IAAI,KAAK,GAAG,CAAC,IAAI,gBAAgB,CAAC,QAAQ,CAAC,CAAC,CAAC,IAAI,CAAC,EAAE,CAAC;YACnD,UAAU,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC;YACnB,OAAO;QACT,CAAC;QAED,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,CAAC,CAAC,UAAU,EAAE,CAAC,EAAE,EAAE,CAAC;YACtC,MAAM,KAAK,GAAG,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC;YACzB,IAAI,KAAK,EAAE,CAAC;gBACV,IAAI,CAAC,KAAK,EAAE,KAAK,GAAG,CAAC,CAAC,CAAC;YACzB,CAAC;QACH,CAAC;IACH,CAAC;IAED,IAAI,CAAC,IAAI,CAAC,CAAC;IACX,OAAO,UAAU,CAAC;AACpB,CAAC;AAED,MAAM,UAAU,wBAAwB,CAAC,IAAY,EAAE,OAAe;IACpE,MAAM,UAAU,GAAG;QACjB,EAAE,OAAO,EAAE,cAAc,EAAE,QAAQ,EAAE,CAAC,EAAE;QACxC,EAAE,OAAO,EAAE,SAAS,EAAE,QAAQ,EAAE,CAAC,EAAE;QACnC,EAAE,OAAO,EAAE,UAAU,EAAE,QAAQ,EAAE,CAAC,EAAE;KACrC,CAAC;IAEF,KAAK,MAAM,QAAQ,IAAI,UAAU,EAAE,CAAC;QAClC,MAAM,OAAO,GAAG,CAAC,GAAG,IAAI,CAAC,SAAS,CAAC,CAAC,EAAE,OAAO,CAAC,CAAC,QAAQ,CAAC,QAAQ,CAAC,OAAO,CAAC,CAAC,CAAC;QAC3E,IAAI,OAAO,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;YACvB,MAAM,SAAS,GAAG,OAAO,CAAC,OAAO,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC;YAC9C,OAAO,SAAS,CAAC,KAAM,GAAG,SAAS,CAAC,CAAC,CAAC,CAAC,MAAM,CAAC;QAChD,CAAC;IACH,CAAC;IAED,OAAO,OAAO,CAAC;AACjB,CAAC;AAED,MAAM,UAAU,gBAAgB,CAAC,IAAoB,EAAE,MAAc;IACnE,MAAM,IAAI,GAAG,MAAM,CAAC,KAAK,CAAC,IAAI,CAAC,UAAU,EAAE,IAAI,CAAC,QAAQ,CAAC,CAAC;IAC1D,MAAM,UAAU,GAAG,IAAI,CAAC,OAAO,CAAC,GAAG,CAAC,CAAC;IACrC,IAAI,UAAU,KAAK,CAAC,CAAC,EAAE,CAAC;QACtB,OAAO,IAAI,CAAC,SAAS,CAAC,CAAC,EAAE,UAAU,CAAC,CAAC,IAAI,EAAE,GAAG,IAAI,CAAC;IACrD,CAAC;IACD,OAAO,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC;AAC7B,CAAC;AAED,MAAM,UAAU,sBAAsB,CAAC,IAAoB,EAAE,MAAc,EAAE,QAAgB;IAC3F,MAAM,UAAU,GAAG,MAAM,CAAC,SAAS,CAAC,CAAC,EAAE,IAAI,CAAC,UAAU,CAAC,CAAC;IACxD,MAAM,KAAK,GAAG,UAAU,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC;IACrC,OAAO,KAAK,CAAC,KAAK,CAAC,CAAC,QAAQ,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;AAC3C,CAAC;AAED,MAAM,UAAU,oBAAoB,CAAC,IAAoB,EAAE,MAAc;IAKvE,OAAO;QACL,SAAS,EAAE,gBAAgB,CAAC,IAAI,EAAE,MAAM,CAAC;QACzC,SAAS,EAAE,aAAa,CAAC,IAAI,CAAC,UAAU,EAAE,MAAM,CAAC;QACjD,OAAO,EAAE,aAAa,CAAC,IAAI,CAAC,QAAQ,EAAE,MAAM,CAAC;KAC9C,CAAC;AACJ,CAAC;AAED,MAAM,UAAU,aAAa,CAAC,UAAkB,EAAE,MAAc;IAC9D,MAAM,MAAM,GAAG,MAAM,CAAC,SAAS,CAAC,CAAC,EAAE,UAAU,CAAC,CAAC;IAC/C,OAAO,MAAM,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC,MAAM,CAAC;AACnC,CAAC;AAED,SAAS,aAAa,CAAC,OAAqB;IAC1C,IAAI,OAAO,CAAC,SAAS,IAAI,OAAO,CAAC,YAAY,EAAE,CAAC;QAC9C,OAAO;YACL,OAAO,EAAE,OAAO,CAAC,aAAa;YAC9B,GAAG,EAAE,OAAO,CAAC,cAAc;YAC3B,GAAG,EAAE,OAAO,CAAC,cAAc;YAC3B,OAAO,EAAE,OAAO,CAAC,aAAa;YAC9B,IAAI,EAAE,QAAQ;SACf,CAAC;IACJ,CAAC;IACD,OAAO;QACL,OAAO,EAAE,OAAO,CAAC,YAAY;QAC7B,GAAG,EAAE,OAAO,CAAC,aAAa;QAC1B,GAAG,EAAE,OAAO,CAAC,aAAa;QAC1B,OAAO,EAAE,OAAO,CAAC,YAAY;QAC7B,IAAI,EAAE,YAAY;KACnB,CAAC;AACJ,CAAC;AAYD,MAAM,CAAC,KAAK,UAAU,sBAAsB,CAC1C,IAAoB,EACpB,MAAc,EACd,IAAkB,EAClB,OAAqB;IAErB,MAAM,IAAI,GAAG,MAAM,CAAC,KAAK,CAAC,IAAI,CAAC,UAAU,EAAE,IAAI,CAAC,QAAQ,CAAC,CAAC;IAC1D,MAAM,MAAM,GAAG,aAAa,CAAC,OAAO,CAAC,CAAC;IAEtC,IAAI,UAAkB,CAAC;IACvB,IAAI,MAAc,CAAC;IAEnB,IAAI,OAAO,CAAC,SAAS,IAAI,OAAO,CAAC,YAAY,EAAE,CAAC;QAC9C,MAAM,QAAQ,GAAG,MAAM,eAAe,CAAC,IAAI,EAAE,MAAM,EAAE,OAAO,CAAC,YAAY,CAAC,CAAC;QAC3E,UAAU,GAAG,QAAQ,CAAC,IAAI,CAAC;QAC3B,MAAM,GAAG,QAAQ,CAAC,MAAM,CAAC;IAC3B,CAAC;SAAM,CAAC;QACN,UAAU,GAAG,IAAI,CAAC,MAAM,CAAC;QACzB,MAAM,GAAG,OAAO,CAAC;IACnB,CAAC;IAED,MAAM,oBAAoB,GAAG,MAAM,CAAC,GAAG,CAAC;IAExC,OAAO;QACL,aAAa,EAAE,UAAU,IAAI,oBAAoB;QACjD,gBAAgB,EAAE,UAAU,GAAG,oBAAoB;QACnD,qBAAqB,EAAE,wBAAwB,CAAC,IAAI,EAAE,IAAI,CAAC;QAC3D,IAAI,EAAE,UAAU;QAChB,IAAI,EAAE,MAAM,CAAC,IAAI;QACjB,MAAM;QACN,kBAAkB,EAAE,IAAI,CAAC,IAAI,CAAC,UAAU,GAAG,MAAM,CAAC,OAAO,CAAC;KAC3D,CAAC;AACJ,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,iBAAiB,CACrC,KAAuB,EACvB,MAAc,EACd,IAAkB,EAClB,OAAqB,EACrB,aAAa,GAAG,KAAK;IAErB,MAAM,KAAK,GAAG,KAAK,CAAC,GAAG,CAAC,IAAI,CAAC,EAAE,CAAC,MAAM,CAAC,KAAK,CAAC,IAAI,CAAC,UAAU,EAAE,IAAI,CAAC,QAAQ,CAAC,CAAC,CAAC;IAC9E,MAAM,MAAM,GAAG,aAAa,CAAC,OAAO,CAAC,CAAC;IAEtC,IAAI,QAA4B,CAAC;IACjC,IAAI,OAAO,CAAC,SAAS,IAAI,OAAO,CAAC,YAAY,EAAE,CAAC;QAC9C,QAAQ,GAAG,MAAM,oBAAoB,CAAC,KAAK,EAAE,MAAM,EAAE,OAAO,CAAC,YAAY,EAAE,aAAa,CAAC,CAAC;IAC5F,CAAC;SAAM,CAAC;QACN,QAAQ,GAAG,KAAK,CAAC,GAAG,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC;YAC5B,IAAI,EAAE,IAAI,CAAC,MAAM;YACjB,QAAQ,EAAE,IAAI,CAAC,MAAM,GAAG,MAAM,CAAC,GAAG,CAAC,CAAC,CAAC,WAAoB;gBACnD,CAAC,CAAC,IAAI,CAAC,MAAM,GAAG,MAAM,CAAC,GAAG,CAAC,CAAC,CAAC,WAAoB;oBACjD,CAAC,CAAC,IAAI,CAAC,MAAM,IAAI,MAAM,CAAC,OAAO,CAAC,CAAC,CAAC,SAAkB;wBACpD,CAAC,CAAC,oBAA6B;YACrC,MAAM,EAAE,OAAO;SAChB,CAAC,CAAC,CAAC;IACN,CAAC;IAED,OAAO,KAAK,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,CAAC,EAAE,EAAE;QAC3B,MAAM,QAAQ,GAAG,QAAQ,CAAC,CAAC,CAAC,CAAC;QAC7B,MAAM,oBAAoB,GAAG,MAAM,CAAC,GAAG,CAAC;QACxC,OAAO;YACL,IAAI;YACJ,aAAa,EAAE,QAAQ,CAAC,IAAI,IAAI,oBAAoB;YACpD,gBAAgB,EAAE,QAAQ,CAAC,IAAI,GAAG,oBAAoB;YACtD,qBAAqB,EAAE,wBAAwB,CAAC,IAAI,EAAE,IAAI,CAAC;YAC3D,IAAI,EAAE,QAAQ,CAAC,IAAI;YACnB,IAAI,EAAE,MAAM,CAAC,IAAI;YACjB,MAAM,EAAE,QAAQ,CAAC,MAAM;YACvB,kBAAkB,EAAE,IAAI,CAAC,IAAI,CAAC,QAAQ,CAAC,IAAI,GAAG,MAAM,CAAC,OAAO,CAAC;SAC9D,CAAC;IACJ,CAAC,CAAC,CAAC;AACL,CAAC;AAQD,MAAM,CAAC,KAAK,UAAU,oBAAoB,CACxC,IAAoB,EACpB,MAAc,EACd,OAAe,EACf,WAAmB,EACnB,OAAqB;IAErB,MAAM,IAAI,GAAG,MAAM,CAAC,KAAK,CAAC,IAAI,CAAC,UAAU,EAAE,IAAI,CAAC,QAAQ,CAAC,CAAC;IAC1D,MAAM,KAAK,GAAG,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC;IAE/B,MAAM,MAAM,GAAqB,EAAE,CAAC;IACpC,IAAI,YAAY,GAAa,EAAE,CAAC;IAChC,IAAI,WAAW,GAAG,CAAC,CAAC;IAEpB,KAAK,MAAM,IAAI,IAAI,KAAK,EAAE,CAAC;QACzB,MAAM,QAAQ,GAAG,OAAO,CAAC,SAAS,IAAI,OAAO,CAAC,YAAY;YACxD,CAAC,CAAC,MAAM,OAAO,CAAC,YAAY,CAAC,IAAI,CAAC;YAClC,CAAC,CAAC,IAAI,CAAC,MAAM,CAAC;QAEhB,IAAI,WAAW,GAAG,QAAQ,GAAG,OAAO,IAAI,YAAY,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;YAChE,MAAM,CAAC,IAAI,CAAC;gBACV,IAAI,EAAE,YAAY,CAAC,IAAI,CAAC,IAAI,CAAC;gBAC7B,IAAI,EAAE,WAAW;gBACjB,IAAI,EAAE,OAAO,CAAC,SAAS,CAAC,CAAC,CAAC,QAAQ,CAAC,CAAC,CAAC,YAAY;aAClD,CAAC,CAAC;YAEH,MAAM,YAAY,GAAG,IAAI,CAAC,KAAK,CAAC,YAAY,CAAC,MAAM,GAAG,GAAG,CAAC,CAAC;YAC3D,YAAY,GAAG,YAAY,CAAC,KAAK,CAAC,CAAC,YAAY,CAAC,CAAC;YACjD,WAAW,GAAG,OAAO,CAAC,SAAS,IAAI,OAAO,CAAC,YAAY;gBACrD,CAAC,CAAC,MAAM,OAAO,CAAC,YAAY,CAAC,YAAY,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;gBACrD,CAAC,CAAC,YAAY,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,MAAM,CAAC;QACrC,CAAC;QAED,YAAY,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;QACxB,WAAW,IAAI,QAAQ,CAAC;IAC1B,CAAC;IAED,IAAI,YAAY,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;QAC5B,MAAM,CAAC,IAAI,CAAC;YACV,IAAI,EAAE,YAAY,CAAC,IAAI,CAAC,IAAI,CAAC;YAC7B,IAAI,EAAE,WAAW;YACjB,IAAI,EAAE,OAAO,CAAC,SAAS,CAAC,CAAC,CAAC,QAAQ,CAAC,CAAC,CAAC,YAAY;SAClD,CAAC,CAAC;IACL,CAAC;IAED,OAAO,MAAM,CAAC;AAChB,CAAC"}
|
|
@@ -1,28 +0,0 @@
|
|
|
1
|
-
interface TokenCountStats {
|
|
2
|
-
totalRequests: number;
|
|
3
|
-
cacheHits: number;
|
|
4
|
-
charFilterSkips: number;
|
|
5
|
-
actualTokenizations: number;
|
|
6
|
-
batchTokenizations: number;
|
|
7
|
-
cacheHitRate: string;
|
|
8
|
-
charFilterRate: string;
|
|
9
|
-
tokenizationRate: string;
|
|
10
|
-
}
|
|
11
|
-
interface SizeLimits {
|
|
12
|
-
min: number;
|
|
13
|
-
max: number;
|
|
14
|
-
optimal: number;
|
|
15
|
-
}
|
|
16
|
-
type Decision = 'too_small' | 'too_large' | 'needs_tokenization' | 'optimal';
|
|
17
|
-
export interface CodeSizeAnalysis {
|
|
18
|
-
size: number;
|
|
19
|
-
decision: Decision;
|
|
20
|
-
method: string;
|
|
21
|
-
}
|
|
22
|
-
export declare function analyzeCodeSize(code: string, limits: SizeLimits, tokenCounter: (text: string) => number | Promise<number>, allowEstimateForSkip?: boolean): Promise<CodeSizeAnalysis>;
|
|
23
|
-
export declare function batchAnalyzeCodeSize(codeSnippets: string[], limits: SizeLimits, tokenCounter: (text: string) => number | Promise<number>, allowEstimateForSkip?: boolean): Promise<CodeSizeAnalysis[]>;
|
|
24
|
-
export declare function getTokenCountStats(): TokenCountStats;
|
|
25
|
-
export declare function resetTokenCountStats(): void;
|
|
26
|
-
export declare function clearTokenCache(): void;
|
|
27
|
-
export {};
|
|
28
|
-
//# sourceMappingURL=token-counter.d.ts.map
|
|
@@ -1 +0,0 @@
|
|
|
1
|
-
{"version":3,"file":"token-counter.d.ts","sourceRoot":"","sources":["../../src/chunking/token-counter.ts"],"names":[],"mappings":"AAqCA,UAAU,eAAe;IACvB,aAAa,EAAE,MAAM,CAAC;IACtB,SAAS,EAAE,MAAM,CAAC;IAClB,eAAe,EAAE,MAAM,CAAC;IACxB,mBAAmB,EAAE,MAAM,CAAC;IAC5B,kBAAkB,EAAE,MAAM,CAAC;IAC3B,YAAY,EAAE,MAAM,CAAC;IACrB,cAAc,EAAE,MAAM,CAAC;IACvB,gBAAgB,EAAE,MAAM,CAAC;CAC1B;AAeD,UAAU,UAAU;IAClB,GAAG,EAAE,MAAM,CAAC;IACZ,GAAG,EAAE,MAAM,CAAC;IACZ,OAAO,EAAE,MAAM,CAAC;CACjB;AAED,KAAK,QAAQ,GAAG,WAAW,GAAG,WAAW,GAAG,oBAAoB,GAAG,SAAS,CAAC;AAwF7E,MAAM,WAAW,gBAAgB;IAC/B,IAAI,EAAE,MAAM,CAAC;IACb,QAAQ,EAAE,QAAQ,CAAC;IACnB,MAAM,EAAE,MAAM,CAAC;CAChB;AAED,wBAAsB,eAAe,CACnC,IAAI,EAAE,MAAM,EACZ,MAAM,EAAE,UAAU,EAClB,YAAY,EAAE,CAAC,IAAI,EAAE,MAAM,KAAK,MAAM,GAAG,OAAO,CAAC,MAAM,CAAC,EACxD,oBAAoB,UAAQ,GAC3B,OAAO,CAAC,gBAAgB,CAAC,CAgC3B;AAED,wBAAsB,oBAAoB,CACxC,YAAY,EAAE,MAAM,EAAE,EACtB,MAAM,EAAE,UAAU,EAClB,YAAY,EAAE,CAAC,IAAI,EAAE,MAAM,KAAK,MAAM,GAAG,OAAO,CAAC,MAAM,CAAC,EACxD,oBAAoB,UAAQ,GAC3B,OAAO,CAAC,gBAAgB,EAAE,CAAC,CAiD7B;AAED,wBAAgB,kBAAkB,IAAI,eAAe,CAapD;AAED,wBAAgB,oBAAoB,IAAI,IAAI,CAM3C;AAED,wBAAgB,eAAe,IAAI,IAAI,CAEtC"}
|
|
@@ -1,207 +0,0 @@
|
|
|
1
|
-
class LRUCache {
|
|
2
|
-
maxSize;
|
|
3
|
-
cache = new Map();
|
|
4
|
-
constructor(maxSize = 1000) {
|
|
5
|
-
this.maxSize = maxSize;
|
|
6
|
-
}
|
|
7
|
-
get(key) {
|
|
8
|
-
if (!this.cache.has(key))
|
|
9
|
-
return undefined;
|
|
10
|
-
const value = this.cache.get(key);
|
|
11
|
-
this.cache.delete(key);
|
|
12
|
-
this.cache.set(key, value);
|
|
13
|
-
return value;
|
|
14
|
-
}
|
|
15
|
-
set(key, value) {
|
|
16
|
-
if (this.cache.has(key)) {
|
|
17
|
-
this.cache.delete(key);
|
|
18
|
-
}
|
|
19
|
-
this.cache.set(key, value);
|
|
20
|
-
if (this.cache.size > this.maxSize) {
|
|
21
|
-
const firstKey = this.cache.keys().next().value;
|
|
22
|
-
if (firstKey !== undefined) {
|
|
23
|
-
this.cache.delete(firstKey);
|
|
24
|
-
}
|
|
25
|
-
}
|
|
26
|
-
}
|
|
27
|
-
clear() {
|
|
28
|
-
this.cache.clear();
|
|
29
|
-
}
|
|
30
|
-
}
|
|
31
|
-
const tokenCountCache = new LRUCache(1000);
|
|
32
|
-
const stats = {
|
|
33
|
-
totalRequests: 0,
|
|
34
|
-
cacheHits: 0,
|
|
35
|
-
charFilterSkips: 0,
|
|
36
|
-
actualTokenizations: 0,
|
|
37
|
-
batchTokenizations: 0
|
|
38
|
-
};
|
|
39
|
-
function estimateTokensFromChars(charCount) {
|
|
40
|
-
return Math.ceil(charCount / 4);
|
|
41
|
-
}
|
|
42
|
-
function preFilterByChars(code, limits) {
|
|
43
|
-
const charCount = code.length;
|
|
44
|
-
const estimatedTokens = estimateTokensFromChars(charCount);
|
|
45
|
-
const minEstimate = limits.min * 0.8;
|
|
46
|
-
const maxEstimate = limits.max * 1.2;
|
|
47
|
-
const optimalLow = limits.optimal * 0.8;
|
|
48
|
-
const optimalHigh = limits.optimal * 1.2;
|
|
49
|
-
if (estimatedTokens < minEstimate) {
|
|
50
|
-
return { decision: 'too_small', estimate: estimatedTokens };
|
|
51
|
-
}
|
|
52
|
-
if (estimatedTokens > maxEstimate) {
|
|
53
|
-
return { decision: 'too_large', estimate: estimatedTokens };
|
|
54
|
-
}
|
|
55
|
-
if (estimatedTokens >= optimalLow && estimatedTokens <= optimalHigh) {
|
|
56
|
-
return { decision: 'optimal', estimate: estimatedTokens };
|
|
57
|
-
}
|
|
58
|
-
return { decision: 'needs_tokenization', estimate: estimatedTokens };
|
|
59
|
-
}
|
|
60
|
-
async function countTokensWithCache(code, tokenCounter) {
|
|
61
|
-
stats.totalRequests++;
|
|
62
|
-
const cached = tokenCountCache.get(code);
|
|
63
|
-
if (cached !== undefined) {
|
|
64
|
-
stats.cacheHits++;
|
|
65
|
-
return cached;
|
|
66
|
-
}
|
|
67
|
-
stats.actualTokenizations++;
|
|
68
|
-
const result = tokenCounter(code);
|
|
69
|
-
const count = result instanceof Promise ? await result : result;
|
|
70
|
-
tokenCountCache.set(code, count);
|
|
71
|
-
return count;
|
|
72
|
-
}
|
|
73
|
-
async function batchCountTokens(codeSnippets, tokenCounter) {
|
|
74
|
-
stats.batchTokenizations++;
|
|
75
|
-
const results = [];
|
|
76
|
-
const uncached = [];
|
|
77
|
-
const uncachedIndices = [];
|
|
78
|
-
for (let i = 0; i < codeSnippets.length; i++) {
|
|
79
|
-
const code = codeSnippets[i];
|
|
80
|
-
const cached = tokenCountCache.get(code);
|
|
81
|
-
if (cached !== undefined) {
|
|
82
|
-
stats.cacheHits++;
|
|
83
|
-
results[i] = cached;
|
|
84
|
-
}
|
|
85
|
-
else {
|
|
86
|
-
uncached.push(code);
|
|
87
|
-
uncachedIndices.push(i);
|
|
88
|
-
}
|
|
89
|
-
}
|
|
90
|
-
if (uncached.length > 0) {
|
|
91
|
-
stats.actualTokenizations += uncached.length;
|
|
92
|
-
const counts = await Promise.all(uncached.map(async (code) => {
|
|
93
|
-
const result = tokenCounter(code);
|
|
94
|
-
return result instanceof Promise ? await result : result;
|
|
95
|
-
}));
|
|
96
|
-
for (let i = 0; i < counts.length; i++) {
|
|
97
|
-
const code = uncached[i];
|
|
98
|
-
const count = counts[i];
|
|
99
|
-
tokenCountCache.set(code, count);
|
|
100
|
-
results[uncachedIndices[i]] = count;
|
|
101
|
-
}
|
|
102
|
-
}
|
|
103
|
-
return results;
|
|
104
|
-
}
|
|
105
|
-
export async function analyzeCodeSize(code, limits, tokenCounter, allowEstimateForSkip = false) {
|
|
106
|
-
stats.totalRequests++;
|
|
107
|
-
const preFilter = preFilterByChars(code, limits);
|
|
108
|
-
if (allowEstimateForSkip && preFilter.decision === 'too_large') {
|
|
109
|
-
stats.charFilterSkips++;
|
|
110
|
-
return {
|
|
111
|
-
size: preFilter.estimate,
|
|
112
|
-
decision: preFilter.decision,
|
|
113
|
-
method: 'char_estimate'
|
|
114
|
-
};
|
|
115
|
-
}
|
|
116
|
-
const actualSize = await countTokensWithCache(code, tokenCounter);
|
|
117
|
-
let decision;
|
|
118
|
-
if (actualSize < limits.min) {
|
|
119
|
-
decision = 'too_small';
|
|
120
|
-
}
|
|
121
|
-
else if (actualSize > limits.max) {
|
|
122
|
-
decision = 'too_large';
|
|
123
|
-
}
|
|
124
|
-
else if (actualSize <= limits.optimal) {
|
|
125
|
-
decision = 'optimal';
|
|
126
|
-
}
|
|
127
|
-
else {
|
|
128
|
-
decision = 'needs_tokenization';
|
|
129
|
-
}
|
|
130
|
-
return {
|
|
131
|
-
size: actualSize,
|
|
132
|
-
decision,
|
|
133
|
-
method: 'tokenized'
|
|
134
|
-
};
|
|
135
|
-
}
|
|
136
|
-
export async function batchAnalyzeCodeSize(codeSnippets, limits, tokenCounter, allowEstimateForSkip = false) {
|
|
137
|
-
const results = [];
|
|
138
|
-
const needsTokenization = [];
|
|
139
|
-
const needsTokenizationIndices = [];
|
|
140
|
-
for (let i = 0; i < codeSnippets.length; i++) {
|
|
141
|
-
const code = codeSnippets[i];
|
|
142
|
-
const preFilter = preFilterByChars(code, limits);
|
|
143
|
-
if (allowEstimateForSkip && preFilter.decision === 'too_large') {
|
|
144
|
-
stats.charFilterSkips++;
|
|
145
|
-
results[i] = {
|
|
146
|
-
size: preFilter.estimate,
|
|
147
|
-
decision: preFilter.decision,
|
|
148
|
-
method: 'char_estimate'
|
|
149
|
-
};
|
|
150
|
-
}
|
|
151
|
-
else {
|
|
152
|
-
needsTokenization.push(code);
|
|
153
|
-
needsTokenizationIndices.push(i);
|
|
154
|
-
}
|
|
155
|
-
}
|
|
156
|
-
if (needsTokenization.length > 0) {
|
|
157
|
-
const tokenCounts = await batchCountTokens(needsTokenization, tokenCounter);
|
|
158
|
-
for (let i = 0; i < tokenCounts.length; i++) {
|
|
159
|
-
const actualSize = tokenCounts[i];
|
|
160
|
-
const idx = needsTokenizationIndices[i];
|
|
161
|
-
let decision;
|
|
162
|
-
if (actualSize < limits.min) {
|
|
163
|
-
decision = 'too_small';
|
|
164
|
-
}
|
|
165
|
-
else if (actualSize > limits.max) {
|
|
166
|
-
decision = 'too_large';
|
|
167
|
-
}
|
|
168
|
-
else if (actualSize <= limits.optimal) {
|
|
169
|
-
decision = 'optimal';
|
|
170
|
-
}
|
|
171
|
-
else {
|
|
172
|
-
decision = 'needs_tokenization';
|
|
173
|
-
}
|
|
174
|
-
results[idx] = {
|
|
175
|
-
size: actualSize,
|
|
176
|
-
decision,
|
|
177
|
-
method: 'tokenized'
|
|
178
|
-
};
|
|
179
|
-
}
|
|
180
|
-
}
|
|
181
|
-
return results;
|
|
182
|
-
}
|
|
183
|
-
export function getTokenCountStats() {
|
|
184
|
-
return {
|
|
185
|
-
...stats,
|
|
186
|
-
cacheHitRate: stats.totalRequests > 0
|
|
187
|
-
? (stats.cacheHits / stats.totalRequests * 100).toFixed(1) + '%'
|
|
188
|
-
: '0%',
|
|
189
|
-
charFilterRate: stats.totalRequests > 0
|
|
190
|
-
? (stats.charFilterSkips / stats.totalRequests * 100).toFixed(1) + '%'
|
|
191
|
-
: '0%',
|
|
192
|
-
tokenizationRate: stats.totalRequests > 0
|
|
193
|
-
? (stats.actualTokenizations / stats.totalRequests * 100).toFixed(1) + '%'
|
|
194
|
-
: '0%'
|
|
195
|
-
};
|
|
196
|
-
}
|
|
197
|
-
export function resetTokenCountStats() {
|
|
198
|
-
stats.totalRequests = 0;
|
|
199
|
-
stats.cacheHits = 0;
|
|
200
|
-
stats.charFilterSkips = 0;
|
|
201
|
-
stats.actualTokenizations = 0;
|
|
202
|
-
stats.batchTokenizations = 0;
|
|
203
|
-
}
|
|
204
|
-
export function clearTokenCache() {
|
|
205
|
-
tokenCountCache.clear();
|
|
206
|
-
}
|
|
207
|
-
//# sourceMappingURL=token-counter.js.map
|