deer-flow/backend/packages/harness/deerflow/workspace_changes/diff.py

from __future__ import annotations

import difflib

from .types import (
    DiffUnavailableReason,
    FileSnapshot,
    WorkspaceChangeLimits,
    WorkspaceChangeResult,
    WorkspaceChangeStatus,
    WorkspaceChangeSummary,
    WorkspaceFileChange,
    WorkspaceSnapshot,
)


def compare_snapshots(
    before: WorkspaceSnapshot,
    after: WorkspaceSnapshot,
    *,
    limits: WorkspaceChangeLimits | None = None,
) -> WorkspaceChangeResult:
    resolved_limits = limits or WorkspaceChangeLimits()
    changes: list[WorkspaceFileChange] = []
    created = modified = deleted = symlink_created = additions = deletions = 0
    total_diff_bytes = 0
    truncated = before.truncated or after.truncated

    for path in sorted(get_changed_paths(before, after)):
        before_file = before.files.get(path)
        after_file = after.files.get(path)

        status = _status(before_file, after_file)
        if status == "created":
            created += 1
        elif status == "modified":
            modified += 1
        elif status == "symlink_created":
            symlink_created += 1
        else:
            deleted += 1

        diff, line_additions, line_deletions, diff_truncated, reason = _build_diff(
            path,
            before_file,
            after_file,
            remaining_bytes=max(0, resolved_limits.max_total_diff_bytes - total_diff_bytes),
        )
        if diff:
            total_diff_bytes += len(diff.encode("utf-8"))
        if diff_truncated or reason in {"large", "truncated"}:
            truncated = True
        additions += line_additions
        deletions += line_deletions

        if len(changes) < resolved_limits.max_files:
            sample = after_file or before_file
            assert sample is not None
            changes.append(
                WorkspaceFileChange(
                    path=path,
                    root=sample.root,
                    status=status,
                    binary=bool((after_file or before_file).binary if (after_file or before_file) else False),
                    sensitive=bool((after_file or before_file).sensitive if (after_file or before_file) else False),
                    size_before=before_file.size if before_file else None,
                    size_after=after_file.size if after_file else None,
                    sha256_before=before_file.sha256 if before_file else None,
                    sha256_after=after_file.sha256 if after_file else None,
                    diff=diff,
                    diff_truncated=diff_truncated,
                    diff_unavailable_reason=reason,
                    additions=line_additions,
                    deletions=line_deletions,
                    symlink=bool((after_file or before_file).symlink if (after_file or before_file) else False),
                    symlink_target_before=before_file.symlink_target if before_file else None,
                    symlink_target_after=after_file.symlink_target if after_file else None,
                )
            )
        else:
            truncated = True

    return WorkspaceChangeResult(
        summary=WorkspaceChangeSummary(
            created=created,
            modified=modified,
            deleted=deleted,
            symlink_created=symlink_created,
            additions=additions,
            deletions=deletions,
            truncated=truncated,
        ),
        files=changes,
        limits=resolved_limits,
    )


def get_changed_paths(before: WorkspaceSnapshot, after: WorkspaceSnapshot) -> set[str]:
    changed: set[str] = set()
    for path in set(before.files) | set(after.files):
        before_file = before.files.get(path)
        after_file = after.files.get(path)
        # Absence from a partial scan does not prove creation or deletion.
        if (before_file is None and before.truncated) or (after_file is None and after.truncated):
            continue
        if before_file and after_file and _same_file(before_file, after_file):
            continue
        changed.add(path)
    return changed


def get_changed_output_paths(before: WorkspaceSnapshot, after: WorkspaceSnapshot) -> list[str]:
    """Return regular output changes established by the bounded snapshots.

    Paths absent from a truncated before snapshot are omitted, including
    genuinely new files; modifications observed in both snapshots are retained.
    The worker's delivery check relies on this list and is skipped when it is
    empty, so an empty result does not prove that the run produced no outputs.
    """
    paths: list[str] = []
    for path in sorted(get_changed_paths(before, after)):
        snapshot = after.files.get(path)
        if snapshot is not None and snapshot.root == "outputs" and not snapshot.symlink:
            paths.append(path)
    return paths


def _status(
    before_file: FileSnapshot | None,
    after_file: FileSnapshot | None,
) -> WorkspaceChangeStatus:
    # A symlink now occupying a path that was not already a symlink is always
    # surfaced distinctly - whether it is brand new (before_file is None) or it
    # just replaced a regular file (before_file is None => "deleted" would
    # otherwise be reported even though the path is still alive on disk, just
    # as a symlink that may point anywhere on the host).
    before_was_symlink = before_file is not None and before_file.symlink
    after_is_symlink = after_file is not None and after_file.symlink
    if after_is_symlink and not before_was_symlink:
        return "symlink_created"
    if before_file is None:
        return "created"
    if after_file is None:
        return "deleted"
    return "modified"


def _same_file(before_file: FileSnapshot, after_file: FileSnapshot) -> bool:
    if before_file.sha256 is not None and after_file.sha256 is not None:
        return before_file.sha256 == after_file.sha256
    return before_file.size == after_file.size and before_file.mtime_ns == after_file.mtime_ns


def _build_diff(
    path: str,
    before_file: FileSnapshot | None,
    after_file: FileSnapshot | None,
    *,
    remaining_bytes: int,
) -> tuple[str, int, int, bool, DiffUnavailableReason | None]:
    reason = _diff_unavailable_reason(before_file, after_file)
    if reason is not None:
        return "", 0, 0, False, reason

    before_text = _snapshot_text(before_file) if before_file else ""
    after_text = _snapshot_text(after_file) if after_file else ""

    if before_file is not None and before_text is None:
        return "", 0, 0, False, None
    if after_file is not None and after_text is None:
        return "", 0, 0, False, None

    lines = list(
        difflib.unified_diff(
            before_text.splitlines(),
            after_text.splitlines(),
            fromfile=f"a{path}",
            tofile=f"b{path}",
            lineterm="",
        )
    )
    diff = "\n".join(lines)
    additions, deletions = _count_diff_lines(lines)
    if len(diff.encode("utf-8")) > remaining_bytes:
        return "", additions, deletions, True, "truncated"
    return diff, additions, deletions, False, None


def _diff_unavailable_reason(
    before_file: FileSnapshot | None,
    after_file: FileSnapshot | None,
) -> DiffUnavailableReason | None:
    files = [file for file in (before_file, after_file) if file is not None]
    for preferred in ("symlink", "sensitive", "binary", "large"):
        if any(file.content_unavailable_reason == preferred for file in files):
            return preferred  # type: ignore[return-value]
    return None


def _snapshot_text(file: FileSnapshot | None) -> str | None:
    if file is None:
        return ""
    if file.text is not None:
        return file.text
    if file.text_path:
        try:
            # The cache is published byte-exactly (the scanner pins
            # newline="\n"), so read it back without newline translation:
            # universal newlines would silently rewrite a decoded CRLF
            # snapshot into LF text and corrupt the served content.
            with open(file.text_path, encoding="utf-8", newline="") as cached:
                return cached.read()
        except OSError:
            return None
    return None


def _count_diff_lines(lines: list[str]) -> tuple[int, int]:
    additions = 0
    deletions = 0
    # difflib.unified_diff always emits the two file headers ("--- a/…" then
    # "+++ b/…") first; skip them by position. A prefix test can't tell them
    # apart from a hunk-body line whose content starts with "-- "/"++ " (e.g. a
    # deleted SQL comment "-- get users" becomes the diff line "--- get users"),
    # which would then be dropped from the count.
    for line in lines[2:]:
        if line.startswith("+"):
            additions += 1
        elif line.startswith("-"):
            deletions += 1
    return additions, deletions