Self-evolving Context Database for AI Agents. Unify Agent Memory, Knowledge RAG and Skills.
Python agent-contextagent-memoryagent-pluginsagentic-ragcontext-databasedsh-pluginknowledge-basememorymemory-managementself-evolvingvector-database
fix(tags): remove charset and length restrictions (#4761)
agent230abf054d31
diff --git a/docs/en/api/06-retrieval.md b/docs/en/api/06-retrieval.mdindex f39aed1bf..bd0b01ebf 100644--- a/docs/en/api/06-retrieval.md+++ b/docs/en/api/06-retrieval.md@@ -58,7 +58,7 @@ The `find()` method performs pure vector similarity search for simple query scen| image_url | str | No | None | Image query as a `data:image/...;base64,...`, `http(s)://`, or `viking://` URI. Requires a multimodal embedding model || target_uri | str \| List[str] | No | "" | Limit search to specific URI prefix || context_type | str \| List[str] | No | None | Limit results to one or more `ContextType` values: `memory`, `resource`, or `skill` |-| tags | List[str] | No | None | Explicit retrieval tags in strict `k=v` form (lowercase letters/digits/`_`/`-`/`.`, starting with a letter or digit, key<=64, value<=128). Multiple tags are combined with AND; a result must contain every requested tag |+| tags | List[str] | No | None | Explicit retrieval tags in strict `k=v` form. Multiple tags are combined with AND; a result must contain every requested tag || node_limit | int | No | None | Maximum number of results || score_threshold | float | No | None | Minimum relevance score threshold || filter | Dict | No | None | Metadata filter |@@ -176,7 +176,7 @@ curl -X POST http://localhost:1933/api/v1/search/find \}'```-Tags must use strict `k=v` strings: both key and value are non-empty, contain exactly one `=`, are made only of lowercase letters, digits, `_`, `-`, `.` and start with a letter or digit (the server trims whitespace and lowercases). The key is capped at 64 and the value at 128 characters. When multiple tags are provided, `find()` requires all of them; the example above only returns contexts whose explicit retrieval tags contain both `env=prod` and `team=search`.+Tags must use strict `k=v` strings. When multiple tags are provided, `find()` requires all of them; the example above only returns contexts whose explicit retrieval tags contain both `env=prod` and `team=search`.**Python SDK**@@ -396,7 +396,7 @@ The `search()` method adds session context understanding and intent analysis cap| session | Session | No | None | Session for context-aware search (SDK) || session_id | str | No | None | Session ID for context-aware search (HTTP) || context_type | str \| List[str] | No | None | Limit results to one or more `ContextType` values: `memory`, `resource`, or `skill` |-| tags | List[str] | No | None | Explicit retrieval tags in strict `k=v` form (lowercase letters/digits/`_`/`-`/`.`, starting with a letter or digit, key<=64, value<=128). Multiple tags are combined with AND; a result must contain every requested tag |+| tags | List[str] | No | None | Explicit retrieval tags in strict `k=v` form. Multiple tags are combined with AND; a result must contain every requested tag || node_limit | int | No | None | Maximum number of results || score_threshold | float | No | None | Minimum relevance score threshold || filter | Dict | No | None | Metadata filter |diff --git a/docs/en/api/12-content.md b/docs/en/api/12-content.mdindex df49478ca..c2ea88e34 100644--- a/docs/en/api/12-content.md+++ b/docs/en/api/12-content.md@@ -491,7 +491,7 @@ Content-Disposition: attachment; filename*=UTF-8''logo.png### set_tags()-Set explicit `k=v` tags used by retrieval filters. Both key and value are non-empty, contain exactly one `=`, are made only of lowercase letters, digits, `_`, `-`, `.` and start with a letter or digit (the server trims whitespace and lowercases); the key is capped at 64 and the value at 128 characters, and invalid tags are rejected. `replace` replaces existing tags, while `append` adds tags. When the target is a directory, `recursive=true` applies the update to files below it.+Set explicit `k=v` tags used by retrieval filters. `replace` replaces existing tags, while `append` adds tags. When the target is a directory, `recursive=true` applies the update to files below it.**Python SDK**diff --git a/docs/zh/api/06-retrieval.md b/docs/zh/api/06-retrieval.mdindex 59c1c1274..169aa01b9 100644--- a/docs/zh/api/06-retrieval.md+++ b/docs/zh/api/06-retrieval.md@@ -58,7 +58,7 @@ OpenViking 提供多种检索方法,包括简单的向量相似度搜索、带| image_url | str | 否 | None | 图片查询,支持 `data:image/...;base64,...`、`http(s)://` 或 `viking://` URI;需要 multimodal embedding 模型 || target_uri | str \| List[str] | 否 | "" | 限制搜索范围到指定的 URI 前缀 || context_type | str \| List[str] | 否 | None | 限定一个或多个 `ContextType` 取值:`memory`、`resource` 或 `skill` |-| tags | List[str] | 否 | None | 显式检索标签,必须是严格的 `k=v` 格式(小写字母/数字/`_`/`-`/`.`,以字母或数字开头,key≤64、value≤128)。多个 tags 之间是 AND 关系,结果必须同时包含所有请求的标签 |+| tags | List[str] | 否 | None | 显式检索标签,必须是严格的 `k=v` 格式。多个 tags 之间是 AND 关系,结果必须同时包含所有请求的标签 || limit | int | 否 | 10 | 最大返回结果数 || node_limit | int | 否 | None | 可选 HTTP 别名;如果提供,会覆盖 limit || score_threshold | float | 否 | None | 最低相关性分数阈值 |@@ -177,7 +177,7 @@ curl -X POST http://localhost:1933/api/v1/search/find \}'```-Tags 必须使用严格的 `k=v` 字符串:key 和 value 都非空,只能有一个 `=`,仅由小写字母、数字、`_`、`-`、`.` 组成且以字母或数字开头(服务端会去空白并转小写),key 最长 64、value 最长 128 字符。传入多个 tags 时,`find()` 会要求全部命中;上面的例子只返回显式检索标签同时包含 `env=prod` 和 `team=search` 的上下文。+Tags 必须使用严格的 `k=v` 字符串。传入多个 tags 时,`find()` 会要求全部命中;上面的例子只返回显式检索标签同时包含 `env=prod` 和 `team=search` 的上下文。**Python SDK**@@ -397,7 +397,7 @@ openviking find "红色海报风格" --image ./poster.png --uri "viking://resour| session | Session | 否 | None | 用于上下文感知搜索的会话(SDK)|| session_id | str | 否 | None | 用于上下文感知搜索的会话 ID(HTTP)|| context_type | str \| List[str] | 否 | None | 限定一个或多个 `ContextType` 取值:`memory`、`resource` 或 `skill` |-| tags | List[str] | 否 | None | 显式检索标签,必须是严格的 `k=v` 格式(小写字母/数字/`_`/`-`/`.`,以字母或数字开头,key≤64、value≤128)。多个 tags 之间是 AND 关系,结果必须同时包含所有请求的标签 |+| tags | List[str] | 否 | None | 显式检索标签,必须是严格的 `k=v` 格式。多个 tags 之间是 AND 关系,结果必须同时包含所有请求的标签 || limit | int | 否 | 10 | 最大返回结果数 || node_limit | int | 否 | None | 可选 HTTP 别名;如果提供,会覆盖 limit || score_threshold | float | 否 | None | 最低相关性分数阈值 |diff --git a/docs/zh/api/12-content.md b/docs/zh/api/12-content.mdindex 6f877ddf3..fe2febb93 100644--- a/docs/zh/api/12-content.md+++ b/docs/zh/api/12-content.md@@ -491,7 +491,7 @@ Content-Disposition: attachment; filename*=UTF-8''logo.png### set_tags()-设置用于检索过滤的显式 `k=v` 标签。key 和 value 都非空、只能有一个 `=`,仅由小写字母、数字、`_`、`-`、`.` 组成且以字母或数字开头(服务端会去空白并转小写),key 最长 64、value 最长 128 字符;不合规的标签会被拒绝。`replace` 替换已有标签,`append` 追加标签;对目录设置 `recursive=true` 时会更新目录下的文件。+设置用于检索过滤的显式 `k=v` 标签。`replace` 替换已有标签,`append` 追加标签;对目录设置 `recursive=true` 时会更新目录下的文件。**Python SDK**diff --git a/openviking/utils/tags.py b/openviking/utils/tags.pyindex a6faa5e08..c9c581ae6 100644--- a/openviking/utils/tags.py+++ b/openviking/utils/tags.py@@ -5,7 +5,6 @@from __future__ import annotationsimport logging-import refrom collections import OrderedDictfrom typing import Any, Iterable@@ -13,25 +12,12 @@ from openviking_cli.exceptions import InvalidArgumentErrorlogger = logging.getLogger(__name__)-# Explicit search tags are meant to be small, enumerable business dimensions-# (env, team, source, ...). Constrain both sides of ``k=v`` so tags stay stable-# identifiers rather than free-form text: bounded length and a predictable-# character set after lower-casing (no internal spaces, no ``=``).-MAX_TAG_KEY_LENGTH = 64-MAX_TAG_VALUE_LENGTH = 128-MAX_TAG_LENGTH = MAX_TAG_KEY_LENGTH + 1 + MAX_TAG_VALUE_LENGTH-_TAG_TOKEN_RE = re.compile(r"^[a-z0-9][a-z0-9_.-]*$")-def normalize_search_tag(tag: str) -> str:"""Validate and normalize a single k=v search tag."""value = str(tag).strip().lower()if not value:raise InvalidArgumentError("search tag must be a non-empty k=v string")- if len(value) > MAX_TAG_LENGTH:- raise InvalidArgumentError(- f"invalid search tag '{tag}': exceeds max length {MAX_TAG_LENGTH}"- )if value.count("=") != 1:raise InvalidArgumentError(f"invalid search tag '{tag}': expected strict k=v format")@@ -40,19 +26,6 @@ def normalize_search_tag(tag: str) -> str:raise InvalidArgumentError(f"invalid search tag '{tag}': key and value must both be non-empty")- if len(key) > MAX_TAG_KEY_LENGTH:- raise InvalidArgumentError(- f"invalid search tag '{tag}': key exceeds max length {MAX_TAG_KEY_LENGTH}"- )- if len(raw_value) > MAX_TAG_VALUE_LENGTH:- raise InvalidArgumentError(- f"invalid search tag '{tag}': value exceeds max length {MAX_TAG_VALUE_LENGTH}"- )- if not _TAG_TOKEN_RE.match(key) or not _TAG_TOKEN_RE.match(raw_value):- raise InvalidArgumentError(- f"invalid search tag '{tag}': key and value may only contain lowercase "- "letters, digits, '_', '-', '.', and must start with a letter or digit"- )return f"{key}={raw_value}"diff --git a/tests/service/test_agent_evolution_service.py b/tests/service/test_agent_evolution_service.pyindex 5cc4b04c0..fd4495473 100644--- a/tests/service/test_agent_evolution_service.py+++ b/tests/service/test_agent_evolution_service.py@@ -22,8 +22,9 @@ def _ctx() -> RequestContext:@pytest.mark.asyncio-async def test_list_trajectories_by_experience_uses_exact_scalar_filter_and_pagination():- experience_uri = "viking://user/alice/memories/experiences/exchange.md"[email protected]("filename", ["exchange.md", "vikingdb_fe_repo_workflows.md"])+async def test_list_trajectories_by_experience_uses_exact_scalar_filter_and_pagination(filename):+ experience_uri = f"viking://user/alice/memories/experiences/{filename}"viking_fs = Mock()viking_fs.stat = AsyncMock(return_value={"isDir": False})vikingdb = Mock()@@ -189,8 +190,9 @@ async def test_list_trajectories_by_experience_accepts_1000_and_paginates_all_re@pytest.mark.asyncio-async def test_get_experience_outcome_distribution_counts_two_tags_on_same_list_field():- experience_uri = "viking://user/alice/memories/experiences/exchange.md"[email protected]("filename", ["exchange.md", "vikingdb_fe_repo_workflows.md"])+async def test_get_experience_outcome_distribution_counts_two_tags_on_same_list_field(filename):+ experience_uri = f"viking://user/alice/memories/experiences/{filename}"viking_fs = Mock()viking_fs.stat = AsyncMock(return_value={"isDir": False})vikingdb = Mock()diff --git a/tests/unit/session/memory/test_experience_lineage.py b/tests/unit/session/memory/test_experience_lineage.pyindex 876bb93dd..f5317438a 100644--- a/tests/unit/session/memory/test_experience_lineage.py+++ b/tests/unit/session/memory/test_experience_lineage.py@@ -15,6 +15,7 @@ from openviking.session.memory.experience_lineage import (from openviking.session.train.components.trajectory_analyzer import (_trajectory_search_tags_by_uri,)+from openviking.utils.tags import build_search_tags_filter, merge_search_tags, normalize_search_tagsfrom openviking_cli.session.user_id import UserIdentifier@@ -180,13 +181,27 @@ def test_collect_read_experience_uris_ignores_removed_dedicated_tool():assert collect_read_experience_uris(messages, ctx=_ctx()) == []-def test_experience_source_tag_uses_experience_uri_as_key():- uri = "viking://user/alice/memories/experiences/无订单号换货处理.md"-+ "uri",+ [+ "viking://user/alice/memories/experiences/cfg_streaming.md",+ "viking://user/alice/memories/experiences/无订单号换货处理.md",+ "viking://user/alice/memories/experiences/vikingdb_fe_repo_workflows.md",+ "viking://user/alice/memories/experiences/" + "nested/" * 40 + "workflow.md",+ ],+)+def test_experience_source_tag_uses_experience_uri_as_key(uri):tag = experience_source_tag(uri)assert tag == f"{uri}=1"assert tag.count("=") == 1+ assert normalize_search_tags([tag], discard_invalid=True) == [tag]+ assert merge_search_tags([tag], ["env=stg"]) == [tag, "env=stg"]+ assert build_search_tags_filter([tag]) == {+ "op": "must",+ "field": "search_tags",+ "conds": [tag],+ }def test_experience_source_tag_preserves_case_and_escapes_equals_without_collisions():@@ -201,6 +216,7 @@ def test_experience_source_tag_preserves_case_and_escapes_equals_without_collisiassert uppercase_tag != lowercase_tagassert uppercase_tag.count("=") == 1assert lowercase_tag.count("=") == 1+ assert merge_search_tags([uppercase_tag], [lowercase_tag]) == [uppercase_tag, lowercase_tag]def test_source_experiences_create_transient_tags_for_every_generated_trajectory():diff --git a/tests/unit/test_search_tags_filter.py b/tests/unit/test_search_tags_filter.pyindex a7b584746..a7021e119 100644--- a/tests/unit/test_search_tags_filter.py+++ b/tests/unit/test_search_tags_filter.py@@ -8,8 +8,6 @@ import pytestfrom openviking.server.routers.search import _resolve_search_filterfrom openviking.utils import tags as tags_modulefrom openviking.utils.tags import (- MAX_TAG_KEY_LENGTH,- MAX_TAG_VALUE_LENGTH,build_search_tags_filter,merge_search_tags,normalize_search_tag,@@ -48,33 +46,31 @@ def test_search_tag_allows_dot_dash_underscore():"tag",["team=search platform", # internal space- "team=with/slash", # unsupported char+ "team=with/slash","team=值", # non-ascii- "-team=search", # must start with letter/digit+ "-team=search","team=-search",- "te=am=search", # more than one '='+ "team=search,platform",+ "team,owner=search",+ "viking://user/alice/memories/experiences/workflow.md=1",+ "viking://user/%41lice/memories/experiences/%45xchange%3d%46low.md=1",],)-def test_search_tag_rejects_disallowed_characters(tag):- with pytest.raises(InvalidArgumentError):- normalize_search_tag(tag)---def test_search_tag_rejects_over_length_key():- over_key = "k" * (MAX_TAG_KEY_LENGTH + 1)- with pytest.raises(InvalidArgumentError):- normalize_search_tag(f"{over_key}=v")+def test_search_tag_accepts_special_characters(tag):+ assert normalize_search_tag(tag) == tag+ assert normalize_search_tags([tag], discard_invalid=True) == [tag]-def test_search_tag_rejects_over_length_value():- over_value = "v" * (MAX_TAG_VALUE_LENGTH + 1)[email protected]("tag", ["", "team", "=search", "team=", "te=am=search"])+def test_search_tag_still_requires_non_empty_kv_format(tag):with pytest.raises(InvalidArgumentError):- normalize_search_tag(f"team={over_value}")+ normalize_search_tag(tag)-def test_search_tag_accepts_boundary_lengths():- key = "k" * MAX_TAG_KEY_LENGTH- value = "v" * MAX_TAG_VALUE_LENGTH[email protected]("key_length,value_length", [(64, 128), (65, 1), (4, 129), (256, 512)])+def test_search_tag_accepts_long_keys_and_values(key_length, value_length):+ key = "k" * key_length+ value = "v" * value_lengthassert normalize_search_tag(f"{key}={value}") == f"{key}={value}"