IdentifierService (General SQL Parser Java Library Core 4.2.9 API)

Object
- gudusoft.gsqlparser.sqlenv.IdentifierService

```
public class IdentifierService
extends Object
```
标识符服务（Identifier Service）
提供所有标识符规范化和比较的统一入口。

为什么需要它（面向入门者）：
不同数据库对名称的大小写和引号行为差别很大（是否区分大小写、是否折叠为大/小写、是否受 collation 影响）。如果在业务代码里到处手写 toLowerCase/equalsIgnoreCase，很容易出错且难以维护。本服务将“如何折叠与如何比较”的规则集中在一个地方，确保全局一致、可扩展、易测试。

与其它类的关系：
- IdentifierRules：一张“规则卡片”（策略），描述未引号/带引号的折叠（fold）与比较（compare）。
- IdentifierProfile：一个“厂商档案”，为某个数据库厂商打包不同对象组（表/列/函数）的规则，并携带厂商开关（如 MySQL lower_case_table_names、SQL Server collation）。
- IdentifierService：基于 Profile 执行规范化与比较，向外提供统一的 normalize/areEqual/keyForMap 接口（门面）。
设计理念与收益：
- 一致性：所有地方都通过同一入口生成 Map 键与做比较，避免行为分裂。
- 可扩展：新增厂商仅需补一份规则工厂；旧调用不变。
- 可测试：规则与服务可单测覆盖，验证各厂商/对象类型是否符合预期。
- 性能友好：keyForMap(java.lang.String, gudusoft.gsqlparser.sqlenv.ESQLDataObjectType) 先做规范化，Map 查找 O(1)；SQL Server 经 CollatorProvider 做 collation 比较。
使用了哪些设计模式：
- 策略（Strategy）：IdentifierRules 即为可替换的规则策略。
- 工厂（Factory）：IdentifierRules.forOracle()/forPostgreSQL()、IdentifierProfile.forVendor(...) 产出预设策略组合。
- 门面（Facade）：本类用少量方法对外隐藏折叠/比较/引号/Collator 细节。
- 依赖注入（DI）：构造时注入 IdentifierProfile 与可选 CollatorProvider，便于替换与测试。
- 不可变值对象（Immutable）：规则/flags 不可变，线程安全、可缓存。
- Provider：CollatorProvider 解耦 SQL Server 的 collation 依赖。
关键约束（必须遵守）：
- 所有索引键构造必须通过 keyForMap(java.lang.String, gudusoft.gsqlparser.sqlenv.ESQLDataObjectType)
- 所有标识符比较必须通过 areEqual(java.lang.String, java.lang.String, gudusoft.gsqlparser.sqlenv.ESQLDataObjectType)
- 禁止在业务代码中直接调用 String.toUpperCase/toLowerCase/equalsIgnoreCase
使用示例：
```
 IdentifierService service = new IdentifierService(profile, collatorProvider);

 // 规范化标识符（用于索引键）
 String key = service.normalize("MyTable", ESQLDataObjectType.dotTable);
 // Oracle: "MYTABLE", PostgreSQL: "mytable", ClickHouse: "MyTable"

 // 比较两个标识符
 boolean eq = service.areEqual("MyTable", "MYTABLE", ESQLDataObjectType.dotTable);
 // Oracle: true, ClickHouse: false

 // 构造索引键
 String mapKey = service.keyForMap("MyTable", ESQLDataObjectType.dotTable);
 
```
Since:

3.1.0.9

Nested Class Summary

Nested Classes
Modifier and Type	Class and Description
`static class`	`IdentifierService.MixedKindProfileException` Typed rejection for mixed-kind custom profiles (one branch collation-based, one fold-based — the cell shape `areEqual(java.lang.String, java.lang.String, gudusoft.gsqlparser.sqlenv.ESQLDataObjectType)`'s mixed-payload-kind guard never key-matches).

Constructor Summary

Constructors
Constructor and Description

IdentifierService(IdentifierProfile profile, CollatorProvider collatorProvider)
构造标识符服务

Constructors
Constructor and Description
`IdentifierService(IdentifierProfile profile, CollatorProvider collatorProvider)` 构造标识符服务

Method Summary

All Methods Static Methods Instance Methods Concrete Methods
Modifier and Type	Method and Description
`ApproximateIdentifierKey`	`approximateKeyV1(ESQLDataObjectType objectType, String name, IdentifierInputForm form)` Mint the best-effort key that serves EVERY rule cell, including the collation-based and mixed-kind cells `persistentKeyV1(gudusoft.gsqlparser.sqlenv.ESQLDataObjectType, java.lang.String, gudusoft.gsqlparser.sqlenv.IdentifierInputForm)` rejects: quote-strip (lenient — a malformed quoted spelling keeps its identity as written, mirroring `areEqual(java.lang.String, java.lang.String, gudusoft.gsqlparser.sqlenv.ESQLDataObjectType)`'s totality) plus pinned simple case-fold collapse.
`boolean`	`areEqual(String ident1, String ident2, ESQLDataObjectType objectType)` 比较两个标识符是否相等（P0d.1：canonical-key equality）。
`static boolean`	`areEqualStatic(EDbVendor dbVendor, ESQLDataObjectType objectType, String ident1, String ident2)` High-performance static identifier comparison method with caching
`void`	`assertSingleSegmentOrThrow(String identifier, ESQLDataObjectType objectType)` 断言标识符为单段，否则抛出异常
`String`	`buildCompositeKey(List<String> segments, ESQLDataObjectType objectType)` 构造复合键（使用长度前缀编码避免冲突）
`String`	`buildCompositeKey(String qualifiedName, ESQLDataObjectType objectType)` 构造复合键（从完整限定名）
`CanonKey`	`canonKey(String identifier, ESQLDataObjectType objectType)` Build the canonical key of a single identifier segment (unification plan stage P0c).
`static CanonKey`	`canonKeyStatic(EDbVendor dbVendor, ESQLDataObjectType objectType, String identifier)` Static `canonKey(java.lang.String, gudusoft.gsqlparser.sqlenv.ESQLDataObjectType)` accessor over the cached default-flags service per vendor (same cache `areEqualStatic(gudusoft.gsqlparser.EDbVendor, gudusoft.gsqlparser.sqlenv.ESQLDataObjectType, java.lang.String, java.lang.String)` uses, so key equality mirrors that relation).
`boolean`	`canUseCompositeKey(String qualifiedName)` 判断是否可以使用复合键快速路径
`String`	`describePolicyV1(ESQLDataObjectType objectType)` Human-readable rendering of the V1 policies for one object type — a diagnostic, explicitly NOT the identity (`policyId` is the hash of the binary descriptor), so this formatting can improve without rotating any stored key.
`List<String>`	`expandVendorSpecific(List<String> segments, EDbVendor vendor)` 厂商级预处理（展开特殊语法）
`List<String>`	`expandVendorSpecific(List<String> segments, EDbVendor vendor, String defaultSchema)` 厂商级预处理（展开特殊语法）
`CollatorProvider`	`getCollatorProvider()` 获取 Collator 提供者
`IdentifierProfile`	`getProfile()` 获取标识符配置档案
`String`	`keyForMap(String identifier, ESQLDataObjectType objectType)` 为 Map 索引构造键（单个标识符段）
`List<String>`	`keysForHierarchy(String qualifiedName, List<ESQLDataObjectType> partTypes)` 生成层级索引所需的段级键列表
`String`	`normalize(String identifier, ESQLDataObjectType objectType)` 规范化标识符（去引号 + 大小写折叠）
`String`	`normalizeQualifiedName(String qualifiedName, ESQLDataObjectType objectType)` 规范化完整限定名（等价于 SQLUtil.getIdentifierNormalName）
`String`	`normalizeSegment(String segment, ESQLDataObjectType objectType)` 规范化单个段（去引号 + 大小写折叠）
`static String`	`normalizeStatic(EDbVendor dbVendor, ESQLDataObjectType objectType, String identifier)` High-performance static normalize method with caching
`List<String>`	`parseQualifiedName(String qualifiedName)` 解析完整限定名为段列表
`PersistentIdentifierKey`	`persistentKeyV1(ESQLDataObjectType objectType, String name, IdentifierInputForm form)` Mint the persistable canonical key of one identifier segment.

Methods inherited from class Object
clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

- Constructor Detail
  - IdentifierService
```
public IdentifierService(IdentifierProfile profile,
                         CollatorProvider collatorProvider)
```
    构造标识符服务
    
    Parameters:
    
    profile - 厂商标识符配置档案
    
    collatorProvider - Collator 提供者（SQL Server 专用，可为 null）
- Method Detail
  - normalizeStatic
```
public static String normalizeStatic(EDbVendor dbVendor,
                                     ESQLDataObjectType objectType,
                                     String identifier)
```
    High-performance static normalize method with caching
    This method provides a convenient static interface for identifier normalization while leveraging a pre-populated cache of IdentifierService instances.
    
    Performance characteristics:
    - O(1) cache lookup using EnumMap
    - No object creation - services are pre-created and reused
    - Thread-safe - all cached instances are immutable
    - Handles all vendor-specific rules (collation, per-object-type rules, etc.)
    Usage example:
```
 String normalized = IdentifierService.normalizeStatic(
     EDbVendor.dbvoracle,
     ESQLDataObjectType.dotTable,
     "MyTable"
 );
 // Result: "MYTABLE"
 
```
    Parameters:
    
    dbVendor - database vendor
    
    objectType - object type (table, column, schema, etc.)
    
    identifier - identifier to normalize (may be quoted)
    
    Returns:
    
    normalized identifier (unquoted and case-folded)
  - areEqualStatic
```
public static boolean areEqualStatic(EDbVendor dbVendor,
                                     ESQLDataObjectType objectType,
                                     String ident1,
                                     String ident2)
```
    High-performance static identifier comparison method with caching
    This method provides a convenient static interface for identifier comparison while leveraging a pre-populated cache of IdentifierService instances.
    
    Performance characteristics:
    - O(1) cache lookup using EnumMap
    - No object creation - services are pre-created and reused
    - Thread-safe - all cached instances are immutable
    - Handles all vendor-specific rules (case-sensitive, case-insensitive, collation-based)
    Usage example:
```
 boolean equal = IdentifierService.areEqualStatic(
     EDbVendor.dbvoracle,
     ESQLDataObjectType.dotTable,
     "MyTable",
     "MYTABLE"
 );
 // Result: true (Oracle is case-insensitive for unquoted identifiers)
 
```
    Parameters:
    
    dbVendor - database vendor
    
    objectType - object type (table, column, schema, etc.)
    
    ident1 - first identifier to compare (may be quoted)
    
    ident2 - second identifier to compare (may be quoted)
    
    Returns:
    
    true if identifiers are equal according to vendor rules
  - canonKeyStatic
```
public static CanonKey canonKeyStatic(EDbVendor dbVendor,
                                      ESQLDataObjectType objectType,
                                      String identifier)
```
    Static canonKey(java.lang.String, gudusoft.gsqlparser.sqlenv.ESQLDataObjectType) accessor over the cached default-flags service per vendor (same cache areEqualStatic(gudusoft.gsqlparser.EDbVendor, gudusoft.gsqlparser.sqlenv.ESQLDataObjectType, java.lang.String, java.lang.String) uses, so key equality mirrors that relation).
  - normalize
```
public String normalize(String identifier,
                        ESQLDataObjectType objectType)
```
    规范化标识符（去引号 + 大小写折叠）
    用于构造索引键，确保相同语义的标识符生成相同的键。
    
    Parameters:
    
    identifier - 原始标识符（可能带引号）
    
    objectType - 对象类型
    
    Returns:
    
    规范化后的标识符
  - areEqual
```
public boolean areEqual(String ident1,
                        String ident2,
                        ESQLDataObjectType objectType)
```
    比较两个标识符是否相等（P0d.1：canonical-key equality）。
    Defined as canonKey(ident1).equals(canonKey(ident2)), which makes this an equivalence relation by construction. The previous rule selection ("quoted compare if EITHER operand is quoted" over quote-stripped-but-unfolded text) was an OR-relation and non-transitive on quoted-case-preserving dialects: Oracle foo ~ "foo" and foo ~ "FOO" but "foo" ≁ "FOO". Under the canonical relation each operand folds by its own quote state (unquoted foo → FOO matches quoted "FOO", not quoted "foo"), matching how the names are actually stored.
    
    Parameters:
    
    ident1 - 标识符 1
    
    ident2 - 标识符 2
    
    objectType - 对象类型
    
    Returns:
    
    true 如果相等
  - keyForMap
```
public String keyForMap(String identifier,
                        ESQLDataObjectType objectType)
```
    为 Map 索引构造键（单个标识符段）
    注意：此方法仅用于分层索引的单段键，不用于复合键。
    对于 COLLATION_BASED（SQL Server），不做 fold，返回原始标识符（后续通过桶+Collator 比较）。
    
    Parameters:
    
    identifier - 标识符
    
    objectType - 对象类型
    
    Returns:
    
    索引键
  - canonKey
```
public CanonKey canonKey(String identifier,
                         ESQLDataObjectType objectType)
```
    Build the canonical key of a single identifier segment (unification plan stage P0c).
    Fold-based rules use the keyForMap(java.lang.String, gudusoft.gsqlparser.sqlenv.ESQLDataObjectType) text as the payload; collation-based rules (SQL Server / Azure) use CollationKey bytes from the SAME collator areEqual(java.lang.String, java.lang.String, gudusoft.gsqlparser.sqlenv.ESQLDataObjectType) compares with, so key equality mirrors that collator. When a collation-based profile has no CollatorProvider, areEqual falls back to comparing caseCollapse(java.lang.String) payloads (pinned Unicode simple case folding since U5); the key mirrors that exactly by carrying the collapsed decoded text.
    NOTE: since P0d.1, areEqual(java.lang.String, java.lang.String, gudusoft.gsqlparser.sqlenv.ESQLDataObjectType) IS canonical-key equality. The legacy TSQLEnv.compareIdentifier façade migrates to this relation in P0d.2; until then the P0c equivalence proof measures that remaining engine's distance from it.
    
    Parameters:
    
    identifier - single identifier segment (may be quoted); null returns null
    
    objectType - kind of database object the name refers to
    
    Returns:
    
    the canonical key, or null for a null identifier
  - persistentKeyV1
```
public PersistentIdentifierKey persistentKeyV1(ESQLDataObjectType objectType,
                                               String name,
                                               IdentifierInputForm form)
```
    Mint the persistable canonical key of one identifier segment.
    Contract (no carve-out). For two IdentifierInputForm.SQL_LEXICAL inputs a, b under the same service and object type: persistentKeyV1(type, a, SQL_LEXICAL).equals(persistentKeyV1(type, b, SQL_LEXICAL)) ⟺ areEqual(a, b, type). IdentifierInputForm.CATALOG_STORED inputs join the same equivalence domain: a stored name keys equal to the lexical spelling that produced it (Oracle stored MixedCase ≡ lexical "MixedCase"; stored FOO ≡ lexical foo; MySQL mode-1 stored straße ≡ bare lexical straße). In cells whose unquoted branch folds while the quoted branch compares insensitively without folding (MySQL table names, Presto family), the unquoted fold and the quoted collapse are DIFFERENT maps wherever full folding and simple folding disagree (final sigma: ΟΣ lowers to ος while the collapse maps ς to σ) — that asymmetry is areEqual(java.lang.String, java.lang.String, gudusoft.gsqlparser.sqlenv.ESQLDataObjectType)'s own U5-certified branch semantics, and stored names side with the fold-image (the form the catalog actually holds), not with the quoted collapse. Keys are stable across JVMs and JDK versions — folding is pinned (GspCaseData), never JVM case tables — and the policyId rotates exactly when the resolved equivalence behavior changes.
    Payload semantics. SQL_LEXICAL: strict codec decode for recognized quoted spellings (IdentifierCodec.decodeLexical(gudusoft.gsqlparser.sqlenv.IdentifierProfile, gudusoft.gsqlparser.sqlenv.ESQLDataObjectType, java.lang.String) — malformed input throws IdentifierCodec.MalformedIdentifierException; this API never guesses identity), then the quote-state branch's fold, then collapse when the branch compares INSENSITIVE without folding. CATALOG_STORED: no recognition, no decode. SENSITIVE stored-name domains keep the stored text; INSENSITIVE domains take the cell's fold when any branch folds (idempotent on real catalog data — stored names ARE fold-images) and the case-collapse only in never-folding cells. (Refines the plan's "INSENSITIVE cells collapse / SENSITIVE cells keep the stored text": in folding cells the collapse is the wrong canonicalizer for stored names — see persistentPayloadV1 — and for folding quoted branches — Snowflake QUOTED_IDENTIFIERS_IGNORE_CASE — the fold keeps cross-form equality intact.)
    Rejections — before descriptor construction AND before any decode/payload work. Cells with both branches resolving to COLLATION_BASED throw UnsupportedOperationException naming the collation (collator weights are not portably reproducible); mixed-kind custom profiles (one branch collation-based, one fold-based — the areEqual(java.lang.String, java.lang.String, gudusoft.gsqlparser.sqlenv.ESQLDataObjectType) mixed-payload-kind guard case) throw the dedicated subtype IdentifierService.MixedKindProfileException — the biconditional's scope is fold/fold cells. approximateKeyV1(gudusoft.gsqlparser.sqlenv.ESQLDataObjectType, java.lang.String, gudusoft.gsqlparser.sqlenv.IdentifierInputForm) serves both.
    Null name returns null. The empty string returns a defined key (empty payload, real policyId): degenerate but stable.
    
    Parameters:
    
    objectType - kind of database object the name refers to
    
    name - single identifier segment
    
    form - how the string reached the caller — lexical SQL text or a catalog-stored name; the forms are not interchangeable
    
    Returns:
    
    the persistent key, or null for a null name
  - approximateKeyV1
```
public ApproximateIdentifierKey approximateKeyV1(ESQLDataObjectType objectType,
                                                 String name,
                                                 IdentifierInputForm form)
```
    Mint the best-effort key that serves EVERY rule cell, including the collation-based and mixed-kind cells persistentKeyV1(gudusoft.gsqlparser.sqlenv.ESQLDataObjectType, java.lang.String, gudusoft.gsqlparser.sqlenv.IdentifierInputForm) rejects: quote-strip (lenient — a malformed quoted spelling keeps its identity as written, mirroring areEqual(java.lang.String, java.lang.String, gudusoft.gsqlparser.sqlenv.ESQLDataObjectType)'s totality) plus pinned simple case-fold collapse. CATALOG_STORED input is collapsed as-is.
    Claims NO areEqual biconditional — see ApproximateIdentifierKey. The distinct type and the distinct ak1- policy namespace guarantee these keys can never be stored or matched where exact keys are expected.
    
    Parameters:
    
    objectType - kind of database object the name refers to
    
    name - single identifier segment
    
    form - lexical SQL text or catalog-stored name
    
    Returns:
    
    the approximate key, or null for a null name
  - describePolicyV1
```
public String describePolicyV1(ESQLDataObjectType objectType)
```
    Human-readable rendering of the V1 policies for one object type — a diagnostic, explicitly NOT the identity (policyId is the hash of the binary descriptor), so this formatting can improve without rotating any stored key.
    
    Parameters:
    
    objectType - kind of database object
    
    Returns:
    
    one-line description of the persistent and approximate policies
  - canUseCompositeKey
```
public boolean canUseCompositeKey(String qualifiedName)
```
    判断是否可以使用复合键快速路径
    条件（必须全部满足）：
    1. 所有对象组都是 SENSITIVE
    2. 输入没有引号字符
    3. 没有 COLLATION_BASED 类型
    Parameters:
    
    qualifiedName - 完整限定名（如 "db.schema.table"）
    
    Returns:
    
    true 如果可以使用复合键
  - buildCompositeKey
```
public String buildCompositeKey(List<String> segments,
                                ESQLDataObjectType objectType)
```
    构造复合键（使用长度前缀编码避免冲突）
    格式: len1#segment1|len2#segment2|len3#segment3|objectType
    例如: "3#db1|6#schema|5#table|dotTable"
    优势：避免分隔符冲突（标识符可能包含 '.' 或 '|'）
    
    Parameters:
    
    segments - 标识符段列表
    
    objectType - 对象类型
    
    Returns:
    
    复合键
  - buildCompositeKey
```
public String buildCompositeKey(String qualifiedName,
                                ESQLDataObjectType objectType)
```
    构造复合键（从完整限定名）
    
    Parameters:
    
    qualifiedName - 完整限定名（如 "db.schema.table"）
    
    objectType - 对象类型
    
    Returns:
    
    复合键
  - parseQualifiedName
```
public List<String> parseQualifiedName(String qualifiedName)
```
    解析完整限定名为段列表
    包装 SQLUtil#parseNames(String, EDbVendor) 以支持厂商特定解析：
    - MSSQL: 保留 ".." 以便后续展开
    - BigQuery: 反引号内的点号仍视为层级分隔
    - 其它: 按 '.' 分段，处理引号包裹的段
    Parameters:
    
    qualifiedName - 完整限定名（如 "db.schema.table" 或 "db..table"）
    
    Returns:
    
    段列表
  - expandVendorSpecific
```
public List<String> expandVendorSpecific(List<String> segments,
                                         EDbVendor vendor)
```
    厂商级预处理（展开特殊语法）
    处理厂商特定语法：
    - MSSQL/Azure SQL: 将 "db..table" 展开为 "db.<global or dbo>.table"
    Parameters:
    
    segments - 原始段列表
    
    vendor - 数据库厂商
    
    Returns:
    
    展开后的段列表
  - expandVendorSpecific
```
public List<String> expandVendorSpecific(List<String> segments,
                                         EDbVendor vendor,
                                         String defaultSchema)
```
    厂商级预处理（展开特殊语法）
    处理厂商特定语法：
    - MSSQL/Azure SQL: 将 "db..table" 展开为 "db.<defaultSchema or dbo>.table"
    Parameters:
    
    segments - 原始段列表
    
    vendor - 数据库厂商
    
    defaultSchema - 默认 schema，如果为 null 则使用 "dbo"
    
    Returns:
    
    展开后的段列表
  - normalizeSegment
```
public String normalizeSegment(String segment,
                               ESQLDataObjectType objectType)
```
    规范化单个段（去引号 + 大小写折叠）
    与 normalize(String, ESQLDataObjectType) 类似，但语义明确为"单段"处理。
    
    Parameters:
    
    segment - 单个段标识符
    
    objectType - 对象类型
    
    Returns:
    
    规范化后的段
  - normalizeQualifiedName
```
public String normalizeQualifiedName(String qualifiedName,
                                     ESQLDataObjectType objectType)
```
    规范化完整限定名（等价于 SQLUtil.getIdentifierNormalName）
    处理流程：
    1. 解析为段列表：parseQualifiedName(String)
    2. 展开厂商特定语法：expandVendorSpecific(List, EDbVendor)
    3. 根据 supportCatalog/supportSchema 决定各段类型
    4. 逐段规范化：normalizeSegment(String, ESQLDataObjectType)
    5. 重新拼接为 "catalog.schema.table.column" 格式
    Parameters:
    
    qualifiedName - 完整限定名
    
    objectType - 最终对象类型（dotTable/dotColumn/dotSchema 等）
    
    Returns:
    
    规范化后的完整限定名
  - keysForHierarchy
```
public List<String> keysForHierarchy(String qualifiedName,
                                     List<ESQLDataObjectType> partTypes)
```
    生成层级索引所需的段级键列表
    用于分层索引（hierarchical index），确保各层键由 keyForMap(java.lang.String, gudusoft.gsqlparser.sqlenv.ESQLDataObjectType) 统一生成。
    注意：调用前应先调用 parseQualifiedName(String) 和 expandVendorSpecific(List, EDbVendor) 完成解析与展开。
    
    Parameters:
    
    qualifiedName - 完整限定名
    
    partTypes - 各段对应的对象类型列表（长度应与段数一致）
    
    Returns:
    
    段级键列表
    
    Throws:
    
    IllegalArgumentException - 如果段数与类型数不匹配
  - assertSingleSegmentOrThrow
```
public void assertSingleSegmentOrThrow(String identifier,
                                       ESQLDataObjectType objectType)
```
    断言标识符为单段，否则抛出异常
    用于在 keyForMap(java.lang.String, gudusoft.gsqlparser.sqlenv.ESQLDataObjectType) 入口处确保输入为单段标识符（不包含 '.' 分隔符）。
    如果 TBaseType.ALLOW_MULTI_SEGMENT_IN_KEY 为 true（兼容模式），则仅记录警告日志而不抛异常。
    
    Parameters:
    
    identifier - 待检查的标识符
    
    Throws:
    
    IllegalArgumentException - 如果标识符包含多段且未启用兼容模式
  - getProfile
```
public IdentifierProfile getProfile()
```
    获取标识符配置档案
    
    Returns:
    
    配置档案
  - getCollatorProvider
```
public CollatorProvider getCollatorProvider()
```
    获取 Collator 提供者
    
    Returns:
    
    Collator 提供者（可能为 null）

Class IdentifierService

Nested Class Summary

Constructor Summary

Method Summary

Methods inherited from class Object

Constructor Detail

IdentifierService

Method Detail

normalizeStatic

areEqualStatic

canonKeyStatic

normalize

areEqual

keyForMap

canonKey

persistentKeyV1

approximateKeyV1

describePolicyV1

canUseCompositeKey

buildCompositeKey

buildCompositeKey

parseQualifiedName

expandVendorSpecific

expandVendorSpecific

normalizeSegment

normalizeQualifiedName

keysForHierarchy

assertSingleSegmentOrThrow

getProfile

getCollatorProvider