Git diffs (#25)

* git diffs

* cmdline arg for git diff

* git diff validation

* change structure

* add git diff to model

* nits

* revert webhook

---------

Co-authored-by: Kartik Sarangmath <kartiksarangmath@Kartiks-MacBook-Air.local>

shared/alembic/versions/4e77ec2a7faa_add_git_diff_field_to_agent_instances.py

@@ -0,0 +1,31 @@
+"""Add git_diff field to agent_instances
+
+Revision ID: 4e77ec2a7faa
+Revises: 84cd4a8c9a18
+Create Date: 2025-07-21 12:44:28.861822
+
+"""
+
+from typing import Sequence, Union
+
+from alembic import op
+import sqlalchemy as sa
+
+
+# revision identifiers, used by Alembic.
+revision: str = "4e77ec2a7faa"
+down_revision: Union[str, None] = "84cd4a8c9a18"
+branch_labels: Union[str, Sequence[str], None] = None
+depends_on: Union[str, Sequence[str], None] = None
+
+
+def upgrade() -> None:
+    # ### commands auto generated by Alembic - please adjust! ###
+    op.add_column("agent_instances", sa.Column("git_diff", sa.Text(), nullable=True))
+    # ### end Alembic commands ###
+
+
+def downgrade() -> None:
+    # ### commands auto generated by Alembic - please adjust! ###
+    op.drop_column("agent_instances", "git_diff")
+    # ### end Alembic commands ###

shared/database/models.py

@@ -9,9 +9,11 @@ from sqlalchemy.orm import (
     Mapped,  # type: ignore[attr-defined]
     mapped_column,  # type: ignore[attr-defined]
     relationship,
+    validates,
 )
 
 from .enums import AgentStatus
+from .utils import is_valid_git_diff
 
 if TYPE_CHECKING:
     from .subscription_models import (
@@ -119,6 +121,7 @@ class AgentInstance(Base):
     status: Mapped[AgentStatus] = mapped_column(default=AgentStatus.ACTIVE)
     started_at: Mapped[datetime] = mapped_column(default=lambda: datetime.now(UTC))
     ended_at: Mapped[datetime | None] = mapped_column(default=None)
+    git_diff: Mapped[str | None] = mapped_column(Text, default=None)
 
     # Relationships
     user_agent: Mapped["UserAgent"] = relationship(
@@ -137,6 +140,20 @@ class AgentInstance(Base):
         order_by="AgentUserFeedback.created_at",
     )
 
+    @validates("git_diff")
+    def validate_git_diff(self, key, value):
+        """Validate git diff at the database level.
+
+        Raises ValueError if the git diff is invalid.
+        """
+        if value is None:
+            return value
+
+        if not is_valid_git_diff(value):
+            raise ValueError("Invalid git diff format. Must be a valid unified diff.")
+
+        return value
+
 
 class AgentStep(Base):
     __tablename__ = "agent_steps"

shared/database/utils.py

@@ -0,0 +1,88 @@
+"""Database utility functions."""
+
+import re
+from typing import Optional
+
+
+def is_valid_git_diff(diff: Optional[str]) -> bool:
+    """Validate if a string is a valid git diff.
+
+    Checks for:
+    - Basic git diff format markers
+    - Proper structure
+    - Not just random text
+
+    Args:
+        diff: The string to validate as a git diff
+
+    Returns:
+        True if valid git diff format, False otherwise
+    """
+    if not diff or not isinstance(diff, str):
+        return False
+
+    # Check for essential git diff patterns
+    has_diff_header = re.search(r"^diff --git", diff, re.MULTILINE) is not None
+    has_index_line = (
+        re.search(r"^index [a-f0-9]+\.\.[a-f0-9]+", diff, re.MULTILINE) is not None
+    )
+    has_file_markers = (
+        re.search(r"^--- ", diff, re.MULTILINE) is not None
+        and re.search(r"^\+\+\+ ", diff, re.MULTILINE) is not None
+    )
+    has_hunk_header = re.search(r"^@@[ \-\+,0-9]+@@", diff, re.MULTILINE) is not None
+
+    # For new files (untracked), we might not have index lines
+    has_new_file = re.search(r"^new file mode", diff, re.MULTILINE) is not None
+
+    # A valid diff should have:
+    # 1. diff --git header
+    # 2. Either (index line) OR (new file mode)
+    # 3. File markers (--- and +++)
+    # 4. At least one hunk header (@@)
+
+    is_valid = (
+        has_diff_header
+        and (has_index_line or has_new_file)
+        and has_file_markers
+        and has_hunk_header
+    )
+
+    # Additional check: should have some actual diff content (lines starting with +, -, or space)
+    has_diff_content = re.search(r"^[ \+\-]", diff, re.MULTILINE) is not None
+
+    return is_valid and has_diff_content
+
+
+def sanitize_git_diff(diff: Optional[str]) -> Optional[str]:
+    """Sanitize and validate a git diff for storage.
+
+    Args:
+        diff: The git diff string to sanitize (None means no update needed)
+
+    Returns:
+        - Original diff string if valid
+        - Empty string if diff is empty (clears the git diff)
+        - None if diff is invalid or not provided
+    """
+    if diff is None:
+        return None
+
+    # Strip excessive whitespace
+    diff = diff.strip()
+
+    # If empty after stripping, return empty string (valid case)
+    if not diff:
+        return ""
+
+    # Check if it's a valid git diff
+    if not is_valid_git_diff(diff):
+        return None
+
+    # Limit size to prevent abuse (1MB)
+    max_size = 1024 * 1024  # 1MB
+    if len(diff) > max_size:
+        # Truncate and add marker
+        diff = diff[: max_size - 100] + "\n\n... [TRUNCATED - DIFF TOO LARGE] ..."
+
+    return diff