Function parse_patch

ggshield/core/scan/commit_utils.py:204–313

Parse a patch generated with `git show` or `git diff` using PATCH_COMMON_ARGS. Returns a list of Scannable. A patch looks like this: ``` commit $SHA Author: $NAME <$EMAIL> Date: $DATE $SUBJECT $BODY1 $BODY2 ... $AFFECTED_FILE_LI

(
    sha: Optional[str], patch: str, exclusion_regexes: Optional[Set[Pattern[str]]]
)

def parse_patch(
    sha: Optional[str], patch: str, exclusion_regexes: Optional[Set[Pattern[str]]]
) -> Iterable[Scannable]:
    """
    Parse a patch generated with `git show` or `git diff` using PATCH_COMMON_ARGS.
    Returns a list of Scannable.

    A patch looks like this:

    ```
    commit $SHA
    Author: $NAME <$EMAIL>
    Date:   $DATE

        $SUBJECT

        $BODY1
        $BODY2
        ...

    $AFFECTED_FILE_LINE\0$DIFF1
    $DIFF2
    ...
    ```

    For a non-merge commit, $DIFFn looks like this:

    ```
    diff --git $A_NAME $B_NAME
    $META_INFO
    $META_INFO
    ...
    --- $A_NAME
    +++ $A_NAME
    @@ $FROM $TO @@
    $ACTUAL_CHANGES
    @@ $FROM $TO @@
    $MORE_CHANGES
    ```

    $A_NAME and $B_NAME may be /dev/null in case of creation or removal. When they are
    not, they start with "a/" and "b/" respectively.

    For a 2-parent merge commit with resolved conflicts, $DIFFn looks like this:

    ```
    diff --cc $NAME
    $META_INFO
    $META_INFO
    ...
    --- $A_NAME
    +++ $B_NAME
    @@@ $FROM1 $FROM2 $TO @@@
    $ACTUAL_CHANGES
    ```

    Note that:
    - The diff line only contains one name, without any "a/" or "b/" prefixes.