fix(console): preserve literal SQL in GitHub plan diffs

Signed-off-by: Vamsi-klu <nrvamsi13@gmail.com>

Author: Vamsi-klu

sqlmesh/core/console.py

@@ -6,6 +6,7 @@
 import unittest
 import uuid
 import logging
+import re
 import textwrap
 from humanize import metric, naturalsize
 from itertools import zip_longest
@@ -3354,10 +3355,13 @@ def show_environment_difference_summary(
 
         if context_diff.has_environment_statements_changes and not no_diff:
             self._print("\nEnvironment statements:\n")
-            for _, diff in context_diff.environment_statements_diff(
+            for language, diff in context_diff.environment_statements_diff(
                 include_python_env=not context_diff.is_new_environment
             ):
-                self._print(diff)
+                if language == "sql":
+                    self._print_diff(diff)
+                else:
+                    self._print(diff)
 
     def show_model_difference_summary(
         self,
@@ -3474,12 +3478,8 @@ def _print_modified_models(
                     diff_text = context_diff.text_diff(snapshot.name)
                     # sometimes there is no text_diff, like on a seed model where the data has been updated
                     if diff_text:
-                        diff_text = f"\n```diff\n{diff_text}\n```"
                         # these are part of a Markdown list, so indent them by 2 spaces to relate them to the current list item
-                        diff_text_indented = "\n".join(
-                            [f"  {line}" for line in diff_text.splitlines()]
-                        )
-                        self._print(diff_text_indented)
+                        self._print_diff(diff_text, indent="  ")
                     else:
                         if indirectly_modified_children:
                             self._print("\n")
@@ -3573,7 +3573,7 @@ def _show_categorized_snapshots(self, plan: Plan, default_catalog: t.Optional[st
             else:
                 continue
 
-            self._print(f"```diff\n{context_diff.text_diff(snapshot.name)}\n```\n")
+            self._print_diff(context_diff.text_diff(snapshot.name))
             self._print("```\n")
             self._print(tree)
             self._print("\n```")
@@ -3697,6 +3697,15 @@ def _print(self, value: t.Any, **kwargs: t.Any) -> None:
             self.console.print(value, **kwargs)
         self._captured_outputs.append(capture.get())
 
+    def _print_diff(self, diff: str, indent: str = "") -> None:
+        longest_backticks = max((len(run) for run in re.findall(r"`+", diff)), default=0)
+        fence = "`" * max(3, longest_backticks + 1)
+        indented_diff = diff.replace("\n", f"\n{indent}")
+        block = f"{indent}\n{indent}{fence}diff\n{indent}{indented_diff}\n{indent}{fence}\n"
+        # Capture the Markdown source, since even literal Rich output expands tabs.
+        self._captured_outputs.append(block)
+        self.console.print(block, markup=False, emoji=False, soft_wrap=True, end="")
+
 
 class DatabricksMagicConsole(CaptureTerminalConsole):
     """

tests/core/test_console.py

@@ -1,4 +1,229 @@
+import pytest
+
 from sqlmesh.core.console import MarkdownConsole
+from sqlmesh.core.context_diff import ContextDiff
+from sqlmesh.core.dialect import parse
+from sqlmesh.core.environment import EnvironmentNamingInfo, EnvironmentStatements
+from sqlmesh.core.model import load_sql_based_model
+from sqlmesh.core.snapshot import SnapshotChangeCategory
+from sqlmesh.utils.metaprogramming import Executable
+
+
+@pytest.fixture
+def markdown_context_diff(make_snapshot):
+    def make_diff(old_query, new_query, dialect="tsql"):
+        old, new = [
+            make_snapshot(
+                load_sql_based_model(parse(f"MODEL (name test.model, dialect {dialect}); {query}"))
+            )
+            for query in (old_query, new_query)
+        ]
+        new.categorize_as(SnapshotChangeCategory.NON_BREAKING)
+        return ContextDiff(
+            environment="prod",
+            is_new_environment=False,
+            is_unfinalized_environment=False,
+            normalize_environment_name=True,
+            create_from="prod",
+            create_from_env_exists=True,
+            added=set(),
+            removed_snapshots={},
+            modified_snapshots={new.name: (new, old)},
+            snapshots={new.snapshot_id: new},
+            new_snapshots={new.snapshot_id: new},
+            previous_plan_id=None,
+            previously_promoted_snapshot_ids=set(),
+            previous_finalized_snapshots=None,
+            previous_gateway_managed_virtual_layer=False,
+            gateway_managed_virtual_layer=False,
+            environment_statements=[],
+        )
+
+    return make_diff
+
+
+@pytest.mark.parametrize(
+    "dialect,old_column,new_column,table",
+    [
+        ("tsql", "[old_col]", "[new_col]", "[schema].[table]"),
+        ("tsql", "[old column]", "[new]]column]", "[my schema].[my]]table]"),
+        ("duckdb", "old_col", "new_col", "schema_name.table_name"),
+    ],
+)
+def test_markdown_console_model_diff_identifiers(
+    markdown_context_diff, dialect, old_column, new_column, table
+):
+    context_diff = markdown_context_diff(
+        f"SELECT {old_column} FROM {table}",
+        f"SELECT {new_column} FROM {table}",
+        dialect,
+    )
+    snapshot = next(iter(context_diff.snapshots.values()))
+    diff = context_diff.text_diff(snapshot.name)
+    assert old_column in diff and new_column in diff and table in diff
+
+    console = MarkdownConsole()
+    console.show_model_difference_summary(
+        context_diff, EnvironmentNamingInfo(), None, no_diff=False
+    )
+
+    output = console.consume_captured_output()
+    assert (
+        "  ```diff\n" + "\n".join(f"  {line}" for line in diff.split("\n")) + "\n  ```\n" in output
+    )
+    assert "**Directly Modified:**" in output
+    assert "(Non-breaking)" in output
+    assert console.consume_captured_output() == ""
+
+
+@pytest.mark.parametrize("categorized", [False, True])
+@pytest.mark.parametrize("backticks", ["", "```", "`````"])
+def test_markdown_console_literal_model_diffs(
+    markdown_context_diff, mocker, capsys, categorized, backticks
+):
+    literal = f"[bold] [/invalid] :smile: \\path\t{'long text ' * 20}\n{backticks}\nend"
+    context_diff = markdown_context_diff("SELECT 1 AS [old]", f"SELECT '{literal}' AS [new]")
+    snapshot = next(iter(context_diff.snapshots.values()))
+    diff = context_diff.text_diff(snapshot.name)
+    assert all(line in diff for line in literal.split("\n"))
+    console = MarkdownConsole(width=30)
+    console.log_success("before diff")
+    if categorized:
+        plan = mocker.Mock(
+            context_diff=context_diff,
+            categorized=[snapshot],
+            indirectly_modified={},
+            environment_naming_info=EnvironmentNamingInfo(),
+        )
+        console._show_categorized_snapshots(plan, None)
+    else:
+        console.show_model_difference_summary(
+            context_diff, EnvironmentNamingInfo(), None, no_diff=False
+        )
+    console.log_success("after diff")
+
+    output = console.consume_captured_output()
+    fence = "`" * max(3, len(backticks) + 1)
+    indent = "" if categorized else "  "
+    block = (
+        f"{indent}{fence}diff\n"
+        + "\n".join(f"{indent}{line}" for line in diff.split("\n"))
+        + f"\n{indent}{fence}\n"
+    )
+    assert block in output
+    assert output.count(block) == 1
+    assert output.index("before diff") < output.index(block) < output.index("after diff")
+    assert console.consume_captured_output() == ""
+    visible = capsys.readouterr().out
+    assert visible.count(f"{fence}diff") == 1
+    assert "[bold] [/invalid] :smile:" in visible
+    if categorized:
+        assert "Directly Modified:" in output
+        assert "[bold]" not in output.split(block, 1)[1]
+
+
+@pytest.mark.parametrize("no_diff,empty_diff", [(True, False), (False, True)])
+def test_markdown_console_omitted_model_diffs(markdown_context_diff, mocker, no_diff, empty_diff):
+    context_diff = markdown_context_diff("SELECT 1 AS [old]", "SELECT 2 AS [new]")
+    text_diff = mocker.patch.object(ContextDiff, "text_diff", return_value="")
+    console = MarkdownConsole()
+    console.show_model_difference_summary(
+        context_diff, EnvironmentNamingInfo(), None, no_diff=no_diff
+    )
+    output = console.consume_captured_output()
+    assert "```" not in output
+    assert "**Directly Modified:**" in output
+    if empty_diff:
+        text_diff.assert_called_once()
+    else:
+        text_diff.assert_not_called()
+
+
+@pytest.mark.parametrize("no_diff", [False, True])
+def test_markdown_console_environment_sql_diffs(markdown_context_diff, no_diff):
+    context_diff = markdown_context_diff("SELECT 1", "SELECT 2")
+    context_diff.previous_environment_statements = [
+        EnvironmentStatements(
+            before_all=["SELECT [old] FROM [schema].[table]"],
+            after_all=["SELECT [old]"],
+            python_env={"version": Executable.value(1)},
+        )
+    ]
+    context_diff.environment_statements = [
+        EnvironmentStatements(
+            before_all=["SELECT [new], ':smile: [/invalid]\\path\t' FROM [schema].[table]"],
+            after_all=[f"SELECT '```\n{'long text ' * 20}\n```' AS [new]"],
+            python_env={"version": Executable.value(2)},
+        )
+    ]
+    console = MarkdownConsole(width=35)
+    console.show_environment_difference_summary(context_diff, no_diff=no_diff)
+    output = console.consume_captured_output()
+    assert "**Summary of differences from" in output and "`prod`:**" in output
+    if no_diff:
+        assert "Environment statements:" not in output
+        assert "```" not in output
+        return
+
+    assert "Environment statements:" in output
+    for language, diff in context_diff.environment_statements_diff(include_python_env=True):
+        if language == "sql":
+            fence = "````" if "```" in diff else "```"
+            assert f"{fence}diff\n{diff}\n{fence}\n" in output
+        else:
+            assert diff in output
+    assert output.index("before_all:") < output.index("after_all:") < output.index("dependencies:")
+
+
+@pytest.mark.parametrize("indirect_only", [False, True])
+def test_markdown_console_model_change_categories(
+    markdown_context_diff, make_snapshot, indirect_only
+):
+    context_diff = markdown_context_diff("SELECT 1 AS [old]", "SELECT 2 AS [new]")
+    current, previous = next(iter(context_diff.modified_snapshots.values()))
+    child_model = load_sql_based_model(
+        parse("MODEL (name test.child, dialect tsql); SELECT * FROM test.model")
+    )
+    old_child, new_child = [
+        make_snapshot(child_model, nodes={parent.name: parent.model})
+        for parent in (previous, current)
+    ]
+    new_child.categorize_as(SnapshotChangeCategory.INDIRECT_NON_BREAKING)
+    old_metadata, new_metadata = [
+        make_snapshot(
+            load_sql_based_model(
+                parse(f"MODEL (name test.metadata, owner {owner}); SELECT 1 AS value")
+            )
+        )
+        for owner in ("old_owner", "new_owner")
+    ]
+    context_diff.modified_snapshots.update(
+        {
+            new_child.name: (new_child, old_child),
+            new_metadata.name: (new_metadata, old_metadata),
+        }
+    )
+    context_diff.snapshots.update(
+        {new_child.snapshot_id: new_child, new_metadata.snapshot_id: new_metadata}
+    )
+    assert context_diff.indirectly_modified(new_child.name)
+    assert context_diff.metadata_updated(new_metadata.name)
+    if indirect_only:
+        del context_diff.modified_snapshots[current.name]
+
+    console = MarkdownConsole()
+    console.show_model_difference_summary(
+        context_diff, EnvironmentNamingInfo(), None, no_diff=False
+    )
+    output = console.consume_captured_output()
+    assert "**Indirectly Modified:**\n- `test.child` (Indirect Non-breaking)" in output
+    assert "**Metadata Updated:**\n- `test.metadata`" in output
+    if indirect_only:
+        assert "**Directly Modified:**" not in output
+        assert "```" not in output
+    else:
+        assert "  Indirectly Modified Children:\n    - `test.child`" in output
+        assert output.index("  ```diff") < output.index("  Indirectly Modified Children:")
 
 
 def test_markdown_console_warning_block():

tests/integrations/github/cicd/test_github_controller.py

@@ -12,7 +12,9 @@
 from sqlmesh.core import constants as c
 from sqlmesh.core.config import CategorizerConfig
 from sqlmesh.core.dialect import parse_one
+from sqlmesh.core.environment import EnvironmentNamingInfo, EnvironmentStatements
 from sqlmesh.core.model import SqlModel
+from sqlmesh.core.snapshot import SnapshotChangeCategory
 from sqlmesh.core.user import User, UserRole
 from sqlmesh.core.plan.definition import Plan
 from sqlmesh.core.linter.rule import RuleViolation
@@ -800,6 +802,60 @@ def test_get_plan_summary_includes_warnings_and_errors(
     ) in summary
 
 
+def test_get_plan_summary_preserves_literal_sql(
+    github_client, make_controller, make_snapshot, mocker
+):
+    controller = make_controller(
+        "tests/fixtures/github/pull_request_synchronized.json", github_client
+    )
+    context_diff = controller.prod_plan.context_diff
+    old, new = [
+        make_snapshot(
+            SqlModel(
+                name="test.model",
+                dialect="tsql",
+                query=parse_one(f"SELECT [{column}] FROM [schema].[table]", dialect="tsql"),
+            )
+        )
+        for column in ("old_col", "new_col")
+    ]
+    new.categorize_as(SnapshotChangeCategory.NON_BREAKING)
+    context_diff.is_new_environment = False
+    context_diff.added = set()
+    context_diff.snapshots = {new.snapshot_id: new}
+    context_diff.modified_snapshots = {new.name: (new, old)}
+    context_diff.environment_statements = [
+        EnvironmentStatements(
+            before_all=["SELECT [before] FROM [schema].[table]"],
+            after_all=["SELECT ':smile: [/invalid] ```' AS [after]"],
+            python_env={},
+        )
+    ]
+    plan = mocker.Mock(
+        spec=Plan,
+        context_diff=context_diff,
+        restatements={},
+        missing_intervals=[],
+        user_provided_flags={},
+        environment_naming_info=EnvironmentNamingInfo(),
+    )
+    controller._console.log_success("stale output")
+    github_client.reset_mock()
+
+    summary = controller.get_plan_summary(plan)
+
+    diff = context_diff.text_diff(new.name)
+    assert (
+        "  ```diff\n" + "\n".join(f"  {line}" for line in diff.split("\n")) + "\n  ```\n" in summary
+    )
+    for _, diff in context_diff.environment_statements_diff():
+        fence = "````" if "```" in diff else "```"
+        assert f"{fence}diff\n{diff}\n{fence}\n" in summary
+    assert "stale output" not in summary
+    assert controller._console.consume_captured_output() == ""
+    assert not github_client.mock_calls
+
+
 def test_get_pr_environment_summary_includes_warnings_and_errors(
     github_client, make_controller: t.Callable[..., GithubController]
 ):