Enhance diff command with normalized output and compatibility handling

CHANGELOG.md

@@ -15,14 +15,34 @@ and this project adheres to [Semantic Versioning](https://semver.org/).
 - Treat equivalent role-membership statements written as `EXEC sp_addrolemember ...` or `ALTER ROLE ... ADD MEMBER ...` as compatible during comparison.
 - Treat legacy Service Broker message-type validation synonyms and equivalent contract/service body formatting and item ordering as compatible during comparison.
 - Treat equivalent `TableData` scripts as compatible during comparison when the normalized `INSERT` statements differ only by row ordering within the same contiguous data block.
+- Treat equivalent `TableData` scripts as compatible during comparison when single-row `INSERT` column lists and corresponding value tuples are reordered consistently.
 - Treat equivalent `Table` scripts as compatible during comparison when post-create statement packages differ only by ordering after the base `CREATE TABLE` block.
+- Treat equivalent legacy `Table` statement formatting as compatible during comparison when normalized table definitions, post-create table statements, and persisted option values are otherwise identical.
+- Treat equivalent legacy `UserDefinedType` `CREATE TYPE` formatting as compatible during comparison when the normalized type definition is otherwise identical.
 - Treat omitted `TEXTIMAGE_ON` on `Table` scripts as compatible during comparison only when DB metadata shows the table LOB data space matches the current default data space.
 - Treat equivalent extended-property blocks as compatible during comparison when the normalized `sp_addextendedproperty` statements differ only by ordering, argument spacing, or named-vs-positional argument forms within the same contiguous block.
+- Treat equivalent extended-property blocks as compatible during comparison when the normalized `sp_addextendedproperty` statements differ only by ordering, argument spacing, named-vs-positional forms, or top-level `N'...'` string literal prefixes.
+- Treat leading SSMS-generated banner comments on programmable objects as compatible during comparison.
 - Treat redundant empty or otherwise no-op `GO` batches as compatible during comparison.
+- Treat an omitted terminal `GO` after the final batch as compatible during comparison with an explicit final `GO`.
+- Keep `diff` output readable by rendering compatible `Table` and `UserDefinedType` changes from readable script text instead of opaque comparison-normalized text.
+- Keep normal `diff` output readable for non-table objects such as users and roles by rendering permission changes from readable script text instead of lowercase comparison-normalized tokens.
+- Keep readable `diff` output for `Table` and table-valued `UserDefinedType` bodies at per-entry granularity instead of collapsing the entire body into one changed line.
+- Align readable `Table` and table-valued `UserDefinedType` diffs by individual body entries so a single changed column or inline constraint does not mark the entire body as changed.
+- Exclude SSMS database-diagram support stored procedures from discovery and scripting even when SQL Server does not mark them as system-shipped.
+- Exclude SSMS database-diagram support table and function objects from discovery and scripting even when SQL Server does not mark them as system-shipped.
+- Script `TYPE::` permissions for scalar and table-valued `UserDefinedType` objects.
+- Script database-level permissions granted directly to `Role` and `User` principals, and emit `CREATE USER ... WITHOUT LOGIN` when no server-login metadata is available.
+- Treat equivalent contiguous permission statement ordering as compatible during comparison.
+- Treat legacy standalone table-level inline `PRIMARY KEY` and `UNIQUE` constraints as compatible during comparison with canonical post-create key-constraint statements.
+- Treat legacy CLR table-valued function return-column collation clauses as compatible during comparison when SQL Server ignores them in the effective return shape.
 - Treat legacy explicit `NULL` tokens on CLR table-valued function return columns as compatible during comparison and preserve them during compatibility reconciliation when the rest of the definition matches.
+- Treat equivalent legacy `Assembly` scripts as compatible during comparison when they differ only by banner comments, wrapped or case-varied hex payload formatting, `PERMISSION_SET` spacing, or quoted versus bracketed `ADD FILE` names.
+- Rewrite programmable-object declaration lines to the current metadata name when SQL Server stores stale module text after an object rename.
+- Fix table-trigger scripting after the declaration rewrite change by resolving trigger schema without referencing a non-existent `sys.triggers.schema_id` column.
 - Trailing semicolon differences on `INSERT` statement lines in data scripts are now suppressed during comparison normalization; scripts emitted with and without statement terminators compare as compatible (#47).
 - Legacy `TableData` scripts now compare as compatible when they differ from canonical output only by `SET IDENTITY_INSERT` semicolons or top-level `N'...'` string literal prefixes, including inside multi-line `INSERT ... VALUES (...)` statements.
-- Whitespace-only separator lines now compare as compatible with empty blank lines during `status` and `diff`.
+- Empty separator lines are now ignored during `status` and `diff`, and whitespace-only separator lines compare as compatible after normalization.
 - Preserve reference banner-comment formatting and module-declaration identifier quoting during programmable-object compatibility reconciliation.
 - Preserve compatible computed-column arithmetic grouping parentheses during table compatibility reconciliation.
 
@@ -31,6 +51,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/).
 - Discover and script SQL CLR scalar functions as `Function` objects.
 - Discover and script SQL CLR table-valued functions as `Function` objects.
 - Discover and script SQL CLR stored procedures as `StoredProcedure` objects.
+- Discover and script SQL CLR aggregates as `Aggregate` objects in `Functions/`.
 - Discover and script built-in `dbo` as a `Schema` object when it has explicit schema permissions or schema-level extended properties, without emitting `CREATE SCHEMA`.
 - `sqlct init` now prompts interactively for connection details (server, database, auth, credentials, trust-server-certificate) when run without flags in a new project directory (#36).
 - Connection flags (`--server`, `--database`, `--auth`, `--user`, `--password`, `--trust-server-certificate`) for non-interactive/scripted `init` use (#36).
@@ -39,11 +60,13 @@ and this project adheres to [Semantic Versioning](https://semver.org/).
 - Next-steps suggestions are printed after `sqlct init` to guide users toward `pull`, `status`, and `diff` (#36).
 - Add `--object <selector>` to `sqlct pull` for exact-match filtering using the same selector forms as `diff --object` (#35).
 - Add `--filter <pattern>` to `sqlct pull` for regex-based filtering; multiple patterns may be provided and matching is case-insensitive (#35).
+- Add `--normalized-diff` to `sqlct diff` to render comparison-normalized diff text for debugging while preserving readable diff output by default.
 - Add `--filter <pattern>` to `sqlct diff` for regex-based filtering; without `--object` filters the output to matching objects, with `--object` additionally constrains the single-object result (#35).
 - SQL Authentication support: set `database.auth` to `"sql"` and supply `database.user` (and optionally `database.password`) in `sqlct.config.json` to connect using SQL Server Authentication (#30).
 - Support active object type `Assembly`, with deterministic scripting to `Assemblies/*.sql` for user-defined SQL Server assemblies.
 - Support additional active object types: `TableType`, `XmlSchemaCollection`, `MessageType`, `Contract`, `Queue`, `Service`, `Route`, `EventNotification`, `ServiceBinding`, `FullTextCatalog`, `FullTextStoplist`, and `SearchPropertyList`.
 - Script standalone user-created table statistics as deterministic post-create table statements, including filtered, effective sampling, persisted-sample, incremental, and auto-drop metadata when available.
+- Script persisted key and index storage options such as fill factor, pad index, duplicate-key handling, and row/page locking when those options differ from defaults.
 - Add `--object <pattern>` to `sqlct data track` and `sqlct data untrack` as a flag alias for the positional pattern argument.
 - Add `--filter <regex>` to `sqlct data track` and `sqlct data untrack` for regex-based table matching; matched case-insensitively against the full `schema.table` display name. Exactly one of the positional pattern, `--object`, or `--filter` must be provided; combining any two returns exit code 2.
 - `sqlct diff` now uses a chunked diff format: only changed segments and configurable surrounding context lines are shown instead of the full file. Use `--context <N>` to control the number of context lines (default: 3) (#39).

README.md

@@ -23,7 +23,7 @@ validate configuration, and generate reproducible scripts for Git and CI/CD.
 - `sqlct data untrack [<pattern>] [--object <pattern>] [--filter <regex>] [--project-dir <path>]`
 - `sqlct data list [--project-dir <path>]`
 - `sqlct status [--project-dir <path>] [--target <db|folder>]`
-- `sqlct diff [--project-dir <path>] [--target <db|folder>] [--object <selector>] [--filter <pattern>...] [--context <N>]`
+- `sqlct diff [--project-dir <path>] [--target <db|folder>] [--object <selector>] [--filter <pattern>...] [--context <N>] [--normalized-diff]`
 - `sqlct pull [--project-dir <path>] [--object <selector>] [--filter <pattern>...]`
 
 Current runtime scope for `status`, `diff`, and `pull` covers:
@@ -32,6 +32,7 @@ Current runtime scope for `status`, `diff`, and `pull` covers:
 - `View`
 - `StoredProcedure`
 - `Function`
+- `Aggregate`
 - `Sequence`
 - `Schema`
 - `Role`
@@ -56,10 +57,14 @@ Schema scripting covers user-defined schemas and also emits built-in `dbo` when
 
 Stored procedure scripting covers T-SQL procedures and SQL CLR stored procedures (`sys.objects.type = 'P'` and `PC`).
 
+Role and user scripting includes database-level permissions granted directly to those principals, and loginless users are scripted as `CREATE USER ... WITHOUT LOGIN`.
+
 Table scripting also includes standalone user-created table statistics (`CREATE STATISTICS`) as post-create table statements. Current statistics option coverage includes effective sampling (`FULLSCAN` or `SAMPLE <n> PERCENT`), `PERSIST_SAMPLE_PERCENT = ON`, `NORECOMPUTE`, `INCREMENTAL=ON`, and `AUTO_DROP = ON|OFF` when the source server exposes the required metadata. `MAXDOP`, `STATS_STREAM`, `ROWCOUNT`, and `PAGECOUNT` remain deferred.
 
 Function scripting covers T-SQL scalar/table functions and SQL CLR scalar/table-valued functions (`sys.objects.type = 'FS'` and `FT`), including `EXTERNAL NAME` assembly bindings.
 
+Aggregate scripting covers SQL CLR aggregates (`sys.objects.type = 'AF'`), including `CREATE AGGREGATE ... RETURNS ... EXTERNAL NAME` bindings.
+
 When `data.trackedTables` is configured, `status`, `diff`, and `pull` also process `TableData` artifacts for those explicit tracked tables.
 
 `--object` selectors support: