@tensorcost/otel-genai-cost 0.0.0-stage → 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 +202 -0
- package/README.md +70 -2
- package/dist/index.d.ts +3 -0
- package/dist/index.js +3 -0
- package/dist/map-span.d.ts +80 -0
- package/dist/map-span.js +140 -0
- package/dist/provider-name.d.ts +8 -0
- package/dist/provider-name.js +27 -0
- package/package.json +46 -4
package/LICENSE
ADDED
|
@@ -0,0 +1,202 @@
|
|
|
1
|
+
|
|
2
|
+
Apache License
|
|
3
|
+
Version 2.0, January 2004
|
|
4
|
+
http://www.apache.org/licenses/
|
|
5
|
+
|
|
6
|
+
TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
|
|
7
|
+
|
|
8
|
+
1. Definitions.
|
|
9
|
+
|
|
10
|
+
"License" shall mean the terms and conditions for use, reproduction,
|
|
11
|
+
and distribution as defined by Sections 1 through 9 of this document.
|
|
12
|
+
|
|
13
|
+
"Licensor" shall mean the copyright owner or entity authorized by
|
|
14
|
+
the copyright owner that is granting the License.
|
|
15
|
+
|
|
16
|
+
"Legal Entity" shall mean the union of the acting entity and all
|
|
17
|
+
other entities that control, are controlled by, or are under common
|
|
18
|
+
control with that entity. For the purposes of this definition,
|
|
19
|
+
"control" means (i) the power, direct or indirect, to cause the
|
|
20
|
+
direction or management of such entity, whether by contract or
|
|
21
|
+
otherwise, or (ii) ownership of fifty percent (50%) or more of the
|
|
22
|
+
outstanding shares, or (iii) beneficial ownership of such entity.
|
|
23
|
+
|
|
24
|
+
"You" (or "Your") shall mean an individual or Legal Entity
|
|
25
|
+
exercising permissions granted by this License.
|
|
26
|
+
|
|
27
|
+
"Source" form shall mean the preferred form for making modifications,
|
|
28
|
+
including but not limited to software source code, documentation
|
|
29
|
+
source, and configuration files.
|
|
30
|
+
|
|
31
|
+
"Object" form shall mean any form resulting from mechanical
|
|
32
|
+
transformation or translation of a Source form, including but
|
|
33
|
+
not limited to compiled object code, generated documentation,
|
|
34
|
+
and conversions to other media types.
|
|
35
|
+
|
|
36
|
+
"Work" shall mean the work of authorship, whether in Source or
|
|
37
|
+
Object form, made available under the License, as indicated by a
|
|
38
|
+
copyright notice that is included in or attached to the work
|
|
39
|
+
(an example is provided in the Appendix below).
|
|
40
|
+
|
|
41
|
+
"Derivative Works" shall mean any work, whether in Source or Object
|
|
42
|
+
form, that is based on (or derived from) the Work and for which the
|
|
43
|
+
editorial revisions, annotations, elaborations, or other modifications
|
|
44
|
+
represent, as a whole, an original work of authorship. For the purposes
|
|
45
|
+
of this License, Derivative Works shall not include works that remain
|
|
46
|
+
separable from, or merely link (or bind by name) to the interfaces of,
|
|
47
|
+
the Work and Derivative Works thereof.
|
|
48
|
+
|
|
49
|
+
"Contribution" shall mean any work of authorship, including
|
|
50
|
+
the original version of the Work and any modifications or additions
|
|
51
|
+
to that Work or Derivative Works thereof, that is intentionally
|
|
52
|
+
submitted to Licensor for inclusion in the Work by the copyright owner
|
|
53
|
+
or by an individual or Legal Entity authorized to submit on behalf of
|
|
54
|
+
the copyright owner. For the purposes of this definition, "submitted"
|
|
55
|
+
means any form of electronic, verbal, or written communication sent
|
|
56
|
+
to the Licensor or its representatives, including but not limited to
|
|
57
|
+
communication on electronic mailing lists, source code control systems,
|
|
58
|
+
and issue tracking systems that are managed by, or on behalf of, the
|
|
59
|
+
Licensor for the purpose of discussing and improving the Work, but
|
|
60
|
+
excluding communication that is conspicuously marked or otherwise
|
|
61
|
+
designated in writing by the copyright owner as "Not a Contribution."
|
|
62
|
+
|
|
63
|
+
"Contributor" shall mean Licensor and any individual or Legal Entity
|
|
64
|
+
on behalf of whom a Contribution has been received by Licensor and
|
|
65
|
+
subsequently incorporated within the Work.
|
|
66
|
+
|
|
67
|
+
2. Grant of Copyright License. Subject to the terms and conditions of
|
|
68
|
+
this License, each Contributor hereby grants to You a perpetual,
|
|
69
|
+
worldwide, non-exclusive, no-charge, royalty-free, irrevocable
|
|
70
|
+
copyright license to reproduce, prepare Derivative Works of,
|
|
71
|
+
publicly display, publicly perform, sublicense, and distribute the
|
|
72
|
+
Work and such Derivative Works in Source or Object form.
|
|
73
|
+
|
|
74
|
+
3. Grant of Patent License. Subject to the terms and conditions of
|
|
75
|
+
this License, each Contributor hereby grants to You a perpetual,
|
|
76
|
+
worldwide, non-exclusive, no-charge, royalty-free, irrevocable
|
|
77
|
+
(except as stated in this section) patent license to make, have made,
|
|
78
|
+
use, offer to sell, sell, import, and otherwise transfer the Work,
|
|
79
|
+
where such license applies only to those patent claims licensable
|
|
80
|
+
by such Contributor that are necessarily infringed by their
|
|
81
|
+
Contribution(s) alone or by combination of their Contribution(s)
|
|
82
|
+
with the Work to which such Contribution(s) was submitted. If You
|
|
83
|
+
institute patent litigation against any entity (including a
|
|
84
|
+
cross-claim or counterclaim in a lawsuit) alleging that the Work
|
|
85
|
+
or a Contribution incorporated within the Work constitutes direct
|
|
86
|
+
or contributory patent infringement, then any patent licenses
|
|
87
|
+
granted to You under this License for that Work shall terminate
|
|
88
|
+
as of the date such litigation is filed.
|
|
89
|
+
|
|
90
|
+
4. Redistribution. You may reproduce and distribute copies of the
|
|
91
|
+
Work or Derivative Works thereof in any medium, with or without
|
|
92
|
+
modifications, and in Source or Object form, provided that You
|
|
93
|
+
meet the following conditions:
|
|
94
|
+
|
|
95
|
+
(a) You must give any other recipients of the Work or
|
|
96
|
+
Derivative Works a copy of this License; and
|
|
97
|
+
|
|
98
|
+
(b) You must cause any modified files to carry prominent notices
|
|
99
|
+
stating that You changed the files; and
|
|
100
|
+
|
|
101
|
+
(c) You must retain, in the Source form of any Derivative Works
|
|
102
|
+
that You distribute, all copyright, patent, trademark, and
|
|
103
|
+
attribution notices from the Source form of the Work,
|
|
104
|
+
excluding those notices that do not pertain to any part of
|
|
105
|
+
the Derivative Works; and
|
|
106
|
+
|
|
107
|
+
(d) If the Work includes a "NOTICE" text file as part of its
|
|
108
|
+
distribution, then any Derivative Works that You distribute must
|
|
109
|
+
include a readable copy of the attribution notices contained
|
|
110
|
+
within such NOTICE file, excluding those notices that do not
|
|
111
|
+
pertain to any part of the Derivative Works, in at least one
|
|
112
|
+
of the following places: within a NOTICE text file distributed
|
|
113
|
+
as part of the Derivative Works; within the Source form or
|
|
114
|
+
documentation, if provided along with the Derivative Works; or,
|
|
115
|
+
within a display generated by the Derivative Works, if and
|
|
116
|
+
wherever such third-party notices normally appear. The contents
|
|
117
|
+
of the NOTICE file are for informational purposes only and
|
|
118
|
+
do not modify the License. You may add Your own attribution
|
|
119
|
+
notices within Derivative Works that You distribute, alongside
|
|
120
|
+
or as an addendum to the NOTICE text from the Work, provided
|
|
121
|
+
that such additional attribution notices cannot be construed
|
|
122
|
+
as modifying the License.
|
|
123
|
+
|
|
124
|
+
You may add Your own copyright statement to Your modifications and
|
|
125
|
+
may provide additional or different license terms and conditions
|
|
126
|
+
for use, reproduction, or distribution of Your modifications, or
|
|
127
|
+
for any such Derivative Works as a whole, provided Your use,
|
|
128
|
+
reproduction, and distribution of the Work otherwise complies with
|
|
129
|
+
the conditions stated in this License.
|
|
130
|
+
|
|
131
|
+
5. Submission of Contributions. Unless You explicitly state otherwise,
|
|
132
|
+
any Contribution intentionally submitted for inclusion in the Work
|
|
133
|
+
by You to the Licensor shall be under the terms and conditions of
|
|
134
|
+
this License, without any additional terms or conditions.
|
|
135
|
+
Notwithstanding the above, nothing herein shall supersede or modify
|
|
136
|
+
the terms of any separate license agreement you may have executed
|
|
137
|
+
with Licensor regarding such Contributions.
|
|
138
|
+
|
|
139
|
+
6. Trademarks. This License does not grant permission to use the trade
|
|
140
|
+
names, trademarks, service marks, or product names of the Licensor,
|
|
141
|
+
except as required for reasonable and customary use in describing the
|
|
142
|
+
origin of the Work and reproducing the content of the NOTICE file.
|
|
143
|
+
|
|
144
|
+
7. Disclaimer of Warranty. Unless required by applicable law or
|
|
145
|
+
agreed to in writing, Licensor provides the Work (and each
|
|
146
|
+
Contributor provides its Contributions) on an "AS IS" BASIS,
|
|
147
|
+
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
|
|
148
|
+
implied, including, without limitation, any warranties or conditions
|
|
149
|
+
of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
|
|
150
|
+
PARTICULAR PURPOSE. You are solely responsible for determining the
|
|
151
|
+
appropriateness of using or redistributing the Work and assume any
|
|
152
|
+
risks associated with Your exercise of permissions under this License.
|
|
153
|
+
|
|
154
|
+
8. Limitation of Liability. In no event and under no legal theory,
|
|
155
|
+
whether in tort (including negligence), contract, or otherwise,
|
|
156
|
+
unless required by applicable law (such as deliberate and grossly
|
|
157
|
+
negligent acts) or agreed to in writing, shall any Contributor be
|
|
158
|
+
liable to You for damages, including any direct, indirect, special,
|
|
159
|
+
incidental, or consequential damages of any character arising as a
|
|
160
|
+
result of this License or out of the use or inability to use the
|
|
161
|
+
Work (including but not limited to damages for loss of goodwill,
|
|
162
|
+
work stoppage, computer failure or malfunction, or any and all
|
|
163
|
+
other commercial damages or losses), even if such Contributor
|
|
164
|
+
has been advised of the possibility of such damages.
|
|
165
|
+
|
|
166
|
+
9. Accepting Warranty or Additional Liability. While redistributing
|
|
167
|
+
the Work or Derivative Works thereof, You may choose to offer,
|
|
168
|
+
and charge a fee for, acceptance of support, warranty, indemnity,
|
|
169
|
+
or other liability obligations and/or rights consistent with this
|
|
170
|
+
License. However, in accepting such obligations, You may act only
|
|
171
|
+
on Your own behalf and on Your sole responsibility, not on behalf
|
|
172
|
+
of any other Contributor, and only if You agree to indemnify,
|
|
173
|
+
defend, and hold each Contributor harmless for any liability
|
|
174
|
+
incurred by, or claims asserted against, such Contributor by reason
|
|
175
|
+
of your accepting any such warranty or additional liability.
|
|
176
|
+
|
|
177
|
+
END OF TERMS AND CONDITIONS
|
|
178
|
+
|
|
179
|
+
APPENDIX: How to apply the Apache License to your work.
|
|
180
|
+
|
|
181
|
+
To apply the Apache License to your work, attach the following
|
|
182
|
+
boilerplate notice, with the fields enclosed by brackets "[]"
|
|
183
|
+
replaced with your own identifying information. (Don't include
|
|
184
|
+
the brackets!) The text should be enclosed in the appropriate
|
|
185
|
+
comment syntax for the file format. We also recommend that a
|
|
186
|
+
file or class name and description of purpose be included on the
|
|
187
|
+
same "printed page" as the copyright notice for easier
|
|
188
|
+
identification within third-party archives.
|
|
189
|
+
|
|
190
|
+
Copyright [yyyy] [name of copyright owner]
|
|
191
|
+
|
|
192
|
+
Licensed under the Apache License, Version 2.0 (the "License");
|
|
193
|
+
you may not use this file except in compliance with the License.
|
|
194
|
+
You may obtain a copy of the License at
|
|
195
|
+
|
|
196
|
+
http://www.apache.org/licenses/LICENSE-2.0
|
|
197
|
+
|
|
198
|
+
Unless required by applicable law or agreed to in writing, software
|
|
199
|
+
distributed under the License is distributed on an "AS IS" BASIS,
|
|
200
|
+
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
|
201
|
+
See the License for the specific language governing permissions and
|
|
202
|
+
limitations under the License.
|
package/README.md
CHANGED
|
@@ -1,3 +1,71 @@
|
|
|
1
|
-
#
|
|
1
|
+
# @tensorcost/otel-genai-cost
|
|
2
2
|
|
|
3
|
-
|
|
3
|
+
Reads an OpenTelemetry GenAI span and returns one call shape: provider, model, input and output tokens, cache read and write tokens, the response id and who made the call. It is the same code TensorCost's inference proxy runs on spans from Kong, LiteLLM and plain OpenTelemetry exporters, published so the mapping can be read, checked and corrected in the open.
|
|
4
|
+
|
|
5
|
+
It has no runtime dependencies and does not read the network. Input is an OTLP/JSON span, output is a plain object.
|
|
6
|
+
|
|
7
|
+
```ts
|
|
8
|
+
import { mapGenAiSpan, storedInputTokens } from "@tensorcost/otel-genai-cost";
|
|
9
|
+
|
|
10
|
+
const call = mapGenAiSpan(span); // null when the span is not an LLM call
|
|
11
|
+
call?.provider; // "anthropic" | "openai" | "bedrock" | "azure_openai" | "vertex" | null
|
|
12
|
+
call?.inputTokens; // already adjusted by storedInputTokens
|
|
13
|
+
```
|
|
14
|
+
|
|
15
|
+
Exports: `mapGenAiSpan`, `detectEmitter`, `storedInputTokens`, `spanCorrelationId`, `normalizeProviderName` and the types `GenAiCall`, `GatewayCaller`, `GatewayEmitter`, `GenAiProviderId`, `OtlpSpanLike`, `OtlpAttributeLike`.
|
|
16
|
+
|
|
17
|
+
## Attributes read
|
|
18
|
+
|
|
19
|
+
Where a row lists several keys, they are tried in order and the first one present wins, except for the provider, which is described below the table.
|
|
20
|
+
|
|
21
|
+
| Field | Attribute keys |
|
|
22
|
+
| --- | --- |
|
|
23
|
+
| provider | `gen_ai.provider.name`, then `gen_ai.system`, then `llm.provider` (first one that names a known provider) |
|
|
24
|
+
| model | `gen_ai.request.model`, `gen_ai.response.model`, `llm.model_name` |
|
|
25
|
+
| input tokens | `gen_ai.usage.input_tokens`, `llm.token_count.prompt` |
|
|
26
|
+
| output tokens | `gen_ai.usage.output_tokens`, `llm.token_count.completion` |
|
|
27
|
+
| cache read tokens | `gen_ai.usage.cache_read.input_tokens`, `llm.token_count.prompt_details.cache_read` |
|
|
28
|
+
| cache write tokens | `gen_ai.usage.cache_creation.input_tokens`, `llm.token_count.prompt_details.cache_write` |
|
|
29
|
+
| response id | `gen_ai.response.id` |
|
|
30
|
+
| max tokens | `gen_ai.request.max_tokens` |
|
|
31
|
+
| status | `gen_ai.response.status`, then `error.type`; any value other than `success` is an error |
|
|
32
|
+
| caller (consumer) | `kong.auth.consumer.name` |
|
|
33
|
+
| caller (workspace) | `kong.workspace.name` |
|
|
34
|
+
| caller (user) | `user.email` |
|
|
35
|
+
|
|
36
|
+
The operation is the span name, or `gen_ai.operation.name` when the span has none. Missing token and cache counts are 0.
|
|
37
|
+
|
|
38
|
+
Provider names are matched case-insensitively against a fixed list: `anthropic`; `openai`; `aws.bedrock`, `bedrock`, `amazon_bedrock` (all `bedrock`); `azure.ai.openai`, `azure.ai.inference`, `azure`, `azure_openai` (all `azure_openai`); `gcp.vertex_ai`, `gcp.gemini`, `gcp.gen_ai`, `vertex_ai`, `vertex` (all `vertex`). The raw value is kept in `providerRaw`.
|
|
39
|
+
|
|
40
|
+
## Unknown providers and non-LLM spans
|
|
41
|
+
|
|
42
|
+
The three provider keys are tried in order and the first value that matches the list gives `provider`. `providerRaw` is the first provider value present on the span, matched or not, so `gen_ai.provider.name=foo` with `gen_ai.system=anthropic` gives `provider: "anthropic"` and `providerRaw: "foo"`. If no key matches, `provider` is `null`. It is never replaced with a default vendor, because a guessed vendor prices the call from the wrong rate card. Check `providerRaw` to see what the gateway sent.
|
|
43
|
+
|
|
44
|
+
A span that is not an LLM call gives `null` from `mapGenAiSpan`. That is any span with no model, and any span with a model but no token usage and no response id, which is a wrapper around the real call and would count the spend twice if kept.
|
|
45
|
+
|
|
46
|
+
## Token convention
|
|
47
|
+
|
|
48
|
+
The stored input token count includes cache reads and cache writes. A call that sent 300 uncached input tokens, read 800 tokens from the cache and wrote 100 to it is stored as 1200 input tokens, with `cacheReadTokens` 800 and `cacheWriteTokens` 100 kept alongside.
|
|
49
|
+
|
|
50
|
+
Emitters do not agree on whether the number on the span already includes the cache, so `storedInputTokens(emitter, provider, reportedInput, cacheRead, cacheWrite)` adds the cache back where it is missing. `mapGenAiSpan` calls it for you.
|
|
51
|
+
|
|
52
|
+
| Emitter | Reported input | Stored input |
|
|
53
|
+
| --- | --- | --- |
|
|
54
|
+
| `otel` (OpenTelemetry semconv) | includes cache | as reported |
|
|
55
|
+
| `litellm` | includes cache | as reported |
|
|
56
|
+
| `kong`, provider `anthropic` | excludes cache (unverified) | reported + cache read + cache write |
|
|
57
|
+
| `kong`, any other provider | includes cache | as reported |
|
|
58
|
+
|
|
59
|
+
The Kong and Anthropic row is unverified. It follows Anthropic's own usage block, which reports the uncached remainder, and has not been checked against a live Kong gateway. If you run one and see otherwise, open an issue.
|
|
60
|
+
|
|
61
|
+
`detectEmitter` returns `kong` when any attribute starts with `kong.` and `otel` otherwise. LiteLLM spans carry no marker, so pass `"litellm"` to `mapGenAiSpan` yourself.
|
|
62
|
+
|
|
63
|
+
`spanCorrelationId(traceId, spanId)` returns a UUID-shaped id from both ids. A trace id alone is shared by every call in an agent run.
|
|
64
|
+
|
|
65
|
+
## Reporting a semconv rename
|
|
66
|
+
|
|
67
|
+
The OpenTelemetry GenAI conventions are still changing, and attribute names move between releases. If a gateway or SDK sends a name this package does not read, or a rename changes what a field means, open an issue on https://github.com/vaadhlabs/tensorcost-integrations. Include the attribute names, the semconv or gateway version, and one span as OTLP/JSON with prompts and any personal data removed. A pull request that adds the key to the table above and a test is welcome.
|
|
68
|
+
|
|
69
|
+
## License
|
|
70
|
+
|
|
71
|
+
Apache-2.0.
|
package/dist/index.d.ts
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
1
|
+
export { normalizeProviderName, type GenAiProviderId } from "./provider-name.js";
|
|
2
|
+
export { detectEmitter, mapGenAiSpan, spanCorrelationId, storedInputTokens, type GatewayCaller, type GatewayEmitter, type GenAiCall, type OtlpAttributeLike, type OtlpSpanLike, } from "./map-span.js";
|
|
3
|
+
//# sourceMappingURL=index.d.ts.map
|
package/dist/index.js
ADDED
|
@@ -0,0 +1,80 @@
|
|
|
1
|
+
import { type GenAiProviderId } from "./provider-name.js";
|
|
2
|
+
export interface OtlpAttributeLike {
|
|
3
|
+
key: string;
|
|
4
|
+
value?: {
|
|
5
|
+
stringValue?: string;
|
|
6
|
+
intValue?: string | number;
|
|
7
|
+
doubleValue?: number;
|
|
8
|
+
boolValue?: boolean;
|
|
9
|
+
};
|
|
10
|
+
}
|
|
11
|
+
export interface OtlpSpanLike {
|
|
12
|
+
traceId?: string;
|
|
13
|
+
spanId?: string;
|
|
14
|
+
name?: string;
|
|
15
|
+
startTimeUnixNano?: string;
|
|
16
|
+
endTimeUnixNano?: string;
|
|
17
|
+
attributes?: OtlpAttributeLike[];
|
|
18
|
+
}
|
|
19
|
+
export type GatewayEmitter = "kong" | "litellm" | "otel";
|
|
20
|
+
export interface GatewayCaller {
|
|
21
|
+
kind: string;
|
|
22
|
+
id: string;
|
|
23
|
+
}
|
|
24
|
+
/**
|
|
25
|
+
* One LLM call as a gateway reported it. inputTokens is what the span said;
|
|
26
|
+
* run it through storedInputTokens before storing, because emitters disagree
|
|
27
|
+
* on whether it already contains the cache tokens.
|
|
28
|
+
*/
|
|
29
|
+
export interface GenAiCall {
|
|
30
|
+
provider: GenAiProviderId | null;
|
|
31
|
+
providerRaw: string | null;
|
|
32
|
+
model: string;
|
|
33
|
+
operation: string | null;
|
|
34
|
+
inputTokens: number;
|
|
35
|
+
outputTokens: number;
|
|
36
|
+
cacheReadTokens: number;
|
|
37
|
+
cacheWriteTokens: number;
|
|
38
|
+
responseId: string | null;
|
|
39
|
+
traceId: string | null;
|
|
40
|
+
spanId: string | null;
|
|
41
|
+
maxTokens: number | null;
|
|
42
|
+
callers: GatewayCaller[];
|
|
43
|
+
startedAt: string;
|
|
44
|
+
endedAt: string | null;
|
|
45
|
+
status: "success" | "error";
|
|
46
|
+
errorType: string | null;
|
|
47
|
+
}
|
|
48
|
+
/**
|
|
49
|
+
* Kong's AI plugins stamp kong.* attributes on the span; plain OTel and
|
|
50
|
+
* LiteLLM spans don't. LiteLLM is never detected here (its spans carry no
|
|
51
|
+
* reliable marker), so pass it explicitly.
|
|
52
|
+
*/
|
|
53
|
+
export declare function detectEmitter(span: OtlpSpanLike): GatewayEmitter;
|
|
54
|
+
/**
|
|
55
|
+
* Reads one span. Null means the span is not an LLM call: it has no model, or
|
|
56
|
+
* it has a model but neither usage nor a response id (a wrapper span around
|
|
57
|
+
* the real call, which would double count if it were kept).
|
|
58
|
+
*
|
|
59
|
+
* inputTokens is the stored value: for an emitter whose input excludes the
|
|
60
|
+
* cache (Kong with Anthropic) the cache reads and writes are added back, see
|
|
61
|
+
* storedInputTokens. The emitter is detected from kong.* attributes unless
|
|
62
|
+
* the caller passes it.
|
|
63
|
+
*/
|
|
64
|
+
export declare function mapGenAiSpan(span: OtlpSpanLike, emitter?: GatewayEmitter): GenAiCall | null;
|
|
65
|
+
/**
|
|
66
|
+
* A UUID-shaped id from the trace id and span id together. The trace id alone
|
|
67
|
+
* is shared by every call in an agent run, so five calls would collapse into
|
|
68
|
+
* one row.
|
|
69
|
+
*/
|
|
70
|
+
export declare function spanCorrelationId(traceId: string, spanId: string): string;
|
|
71
|
+
/**
|
|
72
|
+
* The input_tokens value to store, where stored input counts everything the
|
|
73
|
+
* model read, cache reads and writes included. Whether the reported number
|
|
74
|
+
* already includes the cache depends on who emitted the span: OTel semconv
|
|
75
|
+
* and LiteLLM include it; Kong follows each provider's own usage block, so an
|
|
76
|
+
* Anthropic span reports the uncached remainder only. The Kong rows are
|
|
77
|
+
* unverified against a live gateway.
|
|
78
|
+
*/
|
|
79
|
+
export declare function storedInputTokens(emitter: GatewayEmitter, provider: GenAiProviderId | null, reportedInput: number, cacheRead: number, cacheWrite: number): number;
|
|
80
|
+
//# sourceMappingURL=map-span.d.ts.map
|
package/dist/map-span.js
ADDED
|
@@ -0,0 +1,140 @@
|
|
|
1
|
+
import { createHash } from "node:crypto";
|
|
2
|
+
import { normalizeProviderName } from "./provider-name.js";
|
|
3
|
+
const PROVIDER_KEYS = ["gen_ai.provider.name", "gen_ai.system", "llm.provider"];
|
|
4
|
+
const MODEL_KEYS = ["gen_ai.request.model", "gen_ai.response.model", "llm.model_name"];
|
|
5
|
+
const INPUT_KEYS = ["gen_ai.usage.input_tokens", "llm.token_count.prompt"];
|
|
6
|
+
const OUTPUT_KEYS = ["gen_ai.usage.output_tokens", "llm.token_count.completion"];
|
|
7
|
+
const CACHE_READ_KEYS = [
|
|
8
|
+
"gen_ai.usage.cache_read.input_tokens",
|
|
9
|
+
"llm.token_count.prompt_details.cache_read",
|
|
10
|
+
];
|
|
11
|
+
const CACHE_WRITE_KEYS = [
|
|
12
|
+
"gen_ai.usage.cache_creation.input_tokens",
|
|
13
|
+
"llm.token_count.prompt_details.cache_write",
|
|
14
|
+
];
|
|
15
|
+
const CALLER_KEYS = [
|
|
16
|
+
{ key: "kong.auth.consumer.name", kind: "consumer" },
|
|
17
|
+
{ key: "kong.workspace.name", kind: "workspace" },
|
|
18
|
+
{ key: "user.email", kind: "user" },
|
|
19
|
+
];
|
|
20
|
+
function attrString(attrs, keys) {
|
|
21
|
+
for (const key of keys) {
|
|
22
|
+
const v = attrs.find((a) => a.key === key)?.value;
|
|
23
|
+
if (!v)
|
|
24
|
+
continue;
|
|
25
|
+
if (v.stringValue)
|
|
26
|
+
return v.stringValue;
|
|
27
|
+
if (v.intValue != null)
|
|
28
|
+
return String(v.intValue);
|
|
29
|
+
if (v.doubleValue != null)
|
|
30
|
+
return String(v.doubleValue);
|
|
31
|
+
}
|
|
32
|
+
return undefined;
|
|
33
|
+
}
|
|
34
|
+
function attrNumber(attrs, keys) {
|
|
35
|
+
const raw = attrString(attrs, keys);
|
|
36
|
+
if (raw == null)
|
|
37
|
+
return undefined;
|
|
38
|
+
const n = Number(raw);
|
|
39
|
+
return Number.isFinite(n) ? n : undefined;
|
|
40
|
+
}
|
|
41
|
+
function nanoToIso(nano) {
|
|
42
|
+
if (!nano)
|
|
43
|
+
return null;
|
|
44
|
+
const ms = Math.floor(Number(nano) / 1_000_000);
|
|
45
|
+
if (!Number.isFinite(ms))
|
|
46
|
+
return null;
|
|
47
|
+
return new Date(ms).toISOString();
|
|
48
|
+
}
|
|
49
|
+
/**
|
|
50
|
+
* Kong's AI plugins stamp kong.* attributes on the span; plain OTel and
|
|
51
|
+
* LiteLLM spans don't. LiteLLM is never detected here (its spans carry no
|
|
52
|
+
* reliable marker), so pass it explicitly.
|
|
53
|
+
*/
|
|
54
|
+
export function detectEmitter(span) {
|
|
55
|
+
const kong = (span.attributes ?? []).some((a) => a.key.startsWith("kong."));
|
|
56
|
+
return kong ? "kong" : "otel";
|
|
57
|
+
}
|
|
58
|
+
/**
|
|
59
|
+
* Reads one span. Null means the span is not an LLM call: it has no model, or
|
|
60
|
+
* it has a model but neither usage nor a response id (a wrapper span around
|
|
61
|
+
* the real call, which would double count if it were kept).
|
|
62
|
+
*
|
|
63
|
+
* inputTokens is the stored value: for an emitter whose input excludes the
|
|
64
|
+
* cache (Kong with Anthropic) the cache reads and writes are added back, see
|
|
65
|
+
* storedInputTokens. The emitter is detected from kong.* attributes unless
|
|
66
|
+
* the caller passes it.
|
|
67
|
+
*/
|
|
68
|
+
export function mapGenAiSpan(span, emitter) {
|
|
69
|
+
const attrs = span.attributes ?? [];
|
|
70
|
+
const model = attrString(attrs, MODEL_KEYS);
|
|
71
|
+
if (!model)
|
|
72
|
+
return null;
|
|
73
|
+
const input = attrNumber(attrs, INPUT_KEYS);
|
|
74
|
+
const output = attrNumber(attrs, OUTPUT_KEYS);
|
|
75
|
+
const responseId = attrString(attrs, ["gen_ai.response.id"]) ?? null;
|
|
76
|
+
if (input === undefined && output === undefined && responseId === null)
|
|
77
|
+
return null;
|
|
78
|
+
let providerRaw = null;
|
|
79
|
+
let provider = null;
|
|
80
|
+
for (const key of PROVIDER_KEYS) {
|
|
81
|
+
const raw = attrString(attrs, [key]);
|
|
82
|
+
if (!raw)
|
|
83
|
+
continue;
|
|
84
|
+
providerRaw ??= raw;
|
|
85
|
+
provider = normalizeProviderName(raw);
|
|
86
|
+
if (provider)
|
|
87
|
+
break;
|
|
88
|
+
}
|
|
89
|
+
const callers = [];
|
|
90
|
+
for (const { key, kind } of CALLER_KEYS) {
|
|
91
|
+
const id = attrString(attrs, [key]);
|
|
92
|
+
if (id)
|
|
93
|
+
callers.push({ kind, id });
|
|
94
|
+
}
|
|
95
|
+
const cacheRead = attrNumber(attrs, CACHE_READ_KEYS) ?? 0;
|
|
96
|
+
const cacheWrite = attrNumber(attrs, CACHE_WRITE_KEYS) ?? 0;
|
|
97
|
+
const statusRaw = attrString(attrs, ["gen_ai.response.status", "error.type"]);
|
|
98
|
+
const failed = statusRaw !== undefined && statusRaw !== "success";
|
|
99
|
+
return {
|
|
100
|
+
provider,
|
|
101
|
+
providerRaw,
|
|
102
|
+
model,
|
|
103
|
+
operation: span.name ?? attrString(attrs, ["gen_ai.operation.name"]) ?? null,
|
|
104
|
+
inputTokens: storedInputTokens(emitter ?? detectEmitter(span), provider, input ?? 0, cacheRead, cacheWrite),
|
|
105
|
+
outputTokens: output ?? 0,
|
|
106
|
+
cacheReadTokens: cacheRead,
|
|
107
|
+
cacheWriteTokens: cacheWrite,
|
|
108
|
+
responseId,
|
|
109
|
+
traceId: span.traceId ?? null,
|
|
110
|
+
spanId: span.spanId ?? null,
|
|
111
|
+
maxTokens: attrNumber(attrs, ["gen_ai.request.max_tokens"]) ?? null,
|
|
112
|
+
callers,
|
|
113
|
+
startedAt: nanoToIso(span.startTimeUnixNano) ?? new Date().toISOString(),
|
|
114
|
+
endedAt: nanoToIso(span.endTimeUnixNano),
|
|
115
|
+
status: failed ? "error" : "success",
|
|
116
|
+
errorType: failed ? (statusRaw ?? null) : null,
|
|
117
|
+
};
|
|
118
|
+
}
|
|
119
|
+
/**
|
|
120
|
+
* A UUID-shaped id from the trace id and span id together. The trace id alone
|
|
121
|
+
* is shared by every call in an agent run, so five calls would collapse into
|
|
122
|
+
* one row.
|
|
123
|
+
*/
|
|
124
|
+
export function spanCorrelationId(traceId, spanId) {
|
|
125
|
+
const h = createHash("sha256").update(`${traceId}:${spanId}`).digest("hex");
|
|
126
|
+
return `${h.slice(0, 8)}-${h.slice(8, 12)}-${h.slice(12, 16)}-${h.slice(16, 20)}-${h.slice(20, 32)}`;
|
|
127
|
+
}
|
|
128
|
+
/**
|
|
129
|
+
* The input_tokens value to store, where stored input counts everything the
|
|
130
|
+
* model read, cache reads and writes included. Whether the reported number
|
|
131
|
+
* already includes the cache depends on who emitted the span: OTel semconv
|
|
132
|
+
* and LiteLLM include it; Kong follows each provider's own usage block, so an
|
|
133
|
+
* Anthropic span reports the uncached remainder only. The Kong rows are
|
|
134
|
+
* unverified against a live gateway.
|
|
135
|
+
*/
|
|
136
|
+
export function storedInputTokens(emitter, provider, reportedInput, cacheRead, cacheWrite) {
|
|
137
|
+
const excludesCache = emitter === "kong" && provider === "anthropic";
|
|
138
|
+
return excludesCache ? reportedInput + cacheRead + cacheWrite : reportedInput;
|
|
139
|
+
}
|
|
140
|
+
//# sourceMappingURL=map-span.js.map
|
|
@@ -0,0 +1,8 @@
|
|
|
1
|
+
export type GenAiProviderId = "anthropic" | "openai" | "bedrock" | "azure_openai" | "vertex";
|
|
2
|
+
/**
|
|
3
|
+
* Maps the provider name a gateway puts on a span (the OTel GenAI names plus
|
|
4
|
+
* the spellings Kong and LiteLLM use) to the id TensorCost stores. Anything
|
|
5
|
+
* else is null: guessing a provider writes spend under the wrong price list.
|
|
6
|
+
*/
|
|
7
|
+
export declare function normalizeProviderName(raw: string | null | undefined): GenAiProviderId | null;
|
|
8
|
+
//# sourceMappingURL=provider-name.d.ts.map
|
|
@@ -0,0 +1,27 @@
|
|
|
1
|
+
const PROVIDER_NAMES = {
|
|
2
|
+
anthropic: "anthropic",
|
|
3
|
+
openai: "openai",
|
|
4
|
+
"aws.bedrock": "bedrock",
|
|
5
|
+
bedrock: "bedrock",
|
|
6
|
+
amazon_bedrock: "bedrock",
|
|
7
|
+
"azure.ai.openai": "azure_openai",
|
|
8
|
+
"azure.ai.inference": "azure_openai",
|
|
9
|
+
azure: "azure_openai",
|
|
10
|
+
azure_openai: "azure_openai",
|
|
11
|
+
"gcp.vertex_ai": "vertex",
|
|
12
|
+
"gcp.gemini": "vertex",
|
|
13
|
+
"gcp.gen_ai": "vertex",
|
|
14
|
+
vertex_ai: "vertex",
|
|
15
|
+
vertex: "vertex",
|
|
16
|
+
};
|
|
17
|
+
/**
|
|
18
|
+
* Maps the provider name a gateway puts on a span (the OTel GenAI names plus
|
|
19
|
+
* the spellings Kong and LiteLLM use) to the id TensorCost stores. Anything
|
|
20
|
+
* else is null: guessing a provider writes spend under the wrong price list.
|
|
21
|
+
*/
|
|
22
|
+
export function normalizeProviderName(raw) {
|
|
23
|
+
if (!raw)
|
|
24
|
+
return null;
|
|
25
|
+
return PROVIDER_NAMES[raw.trim().toLowerCase()] ?? null;
|
|
26
|
+
}
|
|
27
|
+
//# sourceMappingURL=provider-name.js.map
|
package/package.json
CHANGED
|
@@ -1,6 +1,48 @@
|
|
|
1
1
|
{
|
|
2
2
|
"name": "@tensorcost/otel-genai-cost",
|
|
3
|
-
"version": "0.
|
|
4
|
-
"
|
|
5
|
-
"description": "
|
|
6
|
-
|
|
3
|
+
"version": "0.1.0",
|
|
4
|
+
"license": "Apache-2.0",
|
|
5
|
+
"description": "Reads OpenTelemetry GenAI spans (Kong, LiteLLM, plain OTel) into one call shape with provider, cache tokens and caller identity.",
|
|
6
|
+
"repository": {
|
|
7
|
+
"type": "git",
|
|
8
|
+
"url": "git+https://github.com/vaadhlabs/tensorcost-integrations.git",
|
|
9
|
+
"directory": "packages/otel-genai-cost"
|
|
10
|
+
},
|
|
11
|
+
"publishConfig": {
|
|
12
|
+
"access": "public"
|
|
13
|
+
},
|
|
14
|
+
"keywords": [
|
|
15
|
+
"opentelemetry",
|
|
16
|
+
"genai",
|
|
17
|
+
"llm",
|
|
18
|
+
"cost",
|
|
19
|
+
"tokens"
|
|
20
|
+
],
|
|
21
|
+
"type": "module",
|
|
22
|
+
"main": "./dist/index.js",
|
|
23
|
+
"types": "./dist/index.d.ts",
|
|
24
|
+
"exports": {
|
|
25
|
+
".": {
|
|
26
|
+
"types": "./dist/index.d.ts",
|
|
27
|
+
"import": "./dist/index.js"
|
|
28
|
+
}
|
|
29
|
+
},
|
|
30
|
+
"scripts": {
|
|
31
|
+
"build": "tsc -b",
|
|
32
|
+
"dev": "tsc -b --watch",
|
|
33
|
+
"typecheck": "tsc --noEmit",
|
|
34
|
+
"test": "vitest run",
|
|
35
|
+
"lint": "echo 'no lint configured'"
|
|
36
|
+
},
|
|
37
|
+
"devDependencies": {
|
|
38
|
+
"@types/node": "^22.0.0",
|
|
39
|
+
"typescript": "^5.6.0",
|
|
40
|
+
"vitest": "^3.2.6"
|
|
41
|
+
},
|
|
42
|
+
"files": [
|
|
43
|
+
"dist",
|
|
44
|
+
"!dist/**/*.map",
|
|
45
|
+
"README.md",
|
|
46
|
+
"LICENSE"
|
|
47
|
+
]
|
|
48
|
+
}
|