|
| 1 | +# SPDX-FileCopyrightText: 2025-present Datadog, Inc. <dev@datadoghq.com> |
| 2 | +# |
| 3 | +# SPDX-License-Identifier: MIT |
| 4 | +from collections.abc import Generator, Iterable |
| 5 | +from dataclasses import dataclass |
| 6 | +from enum import StrEnum |
| 7 | +from functools import cached_property |
| 8 | +from typing import Self |
| 9 | + |
| 10 | +from dda.utils.fs import Path |
| 11 | +from dda.utils.git.sha1hash import SHA1Hash |
| 12 | + |
| 13 | +FILECHANGES_GIT_DIFF_ARGS = ["diff", "-U0", "--no-color", "--no-prefix", "--no-renames"] |
| 14 | + |
| 15 | + |
| 16 | +class ChangeType(StrEnum): |
| 17 | + ADDED = "A" |
| 18 | + MODIFIED = "M" |
| 19 | + DELETED = "D" |
| 20 | + |
| 21 | + @classmethod |
| 22 | + def from_github_status(cls, status: str) -> "ChangeType": |
| 23 | + mapping = { |
| 24 | + "added": cls.ADDED, |
| 25 | + "modified": cls.MODIFIED, |
| 26 | + "deleted": cls.DELETED, |
| 27 | + } |
| 28 | + try: |
| 29 | + return mapping[status] |
| 30 | + except KeyError as e: |
| 31 | + msg = f"Invalid GitHub change type message: {status}" |
| 32 | + raise ValueError(msg) from e |
| 33 | + |
| 34 | + |
| 35 | +@dataclass(frozen=True, order=True) |
| 36 | +class FileChanges: |
| 37 | + """Represents changes to a single file in a git repository.""" |
| 38 | + |
| 39 | + file: Path |
| 40 | + type: ChangeType |
| 41 | + |
| 42 | + patch: str |
| 43 | + """ |
| 44 | + The patch representing the changes to the file, in unified diff format. |
| 45 | + We only keep the hunk lines (starting with @@) and the lines starting with + or - (no extra context lines). |
| 46 | + This is similar to the format used by the `patch` fields in GitHub's API. |
| 47 | +
|
| 48 | + Example: |
| 49 | + ```diff |
| 50 | + @@ -15,2 +15 @@ if TYPE_CHECKING: |
| 51 | + - from dda.utils.git.commit import Commit, CommitDetails |
| 52 | + - from dda.utils.git.sha1hash import SHA1Hash |
| 53 | + + from dda.utils.git.commit import Commit |
| 54 | + ``` |
| 55 | + """ |
| 56 | + |
| 57 | + # TODO: This might be a bit brittle - or we might want to move this to a separate file ? |
| 58 | + @classmethod |
| 59 | + def generate_from_diff_output(cls, diff_output: Iterable[str] | str) -> Generator[Self, None, None]: |
| 60 | + """ |
| 61 | + Generate a list of FileChanges from the output of _some_ git diff commands. |
| 62 | + Not all outputs from `git diff` are supported (ex: renames), you should run: |
| 63 | + ```bash |
| 64 | + git diff -U0 --no-color --no-prefix --no-renames <oldrev> <newrev> |
| 65 | + ``` |
| 66 | + Accepts an Iterable of lines or a single string as argument. |
| 67 | + """ |
| 68 | + if isinstance(diff_output, str): |
| 69 | + diff_output = diff_output.strip().splitlines() |
| 70 | + |
| 71 | + line_iterator = iter(diff_output) |
| 72 | + |
| 73 | + current_file: Path | None = None |
| 74 | + current_type: ChangeType | None = None |
| 75 | + current_patch_lines: list[str] = [] |
| 76 | + iterator_exhausted = False |
| 77 | + |
| 78 | + try: |
| 79 | + line = next(line_iterator) |
| 80 | + while True: |
| 81 | + # Start processing a new file - the line looks like `diff --git a/<path> b/<path>` |
| 82 | + if not line.startswith("diff --git "): |
| 83 | + msg = f"Unexpected line in git diff output: {line}" |
| 84 | + raise ValueError(msg) |
| 85 | + |
| 86 | + # Go forward until we find the 'old file' line (---) |
| 87 | + while not line.startswith("--- "): |
| 88 | + try: |
| 89 | + line = next(line_iterator) |
| 90 | + except StopIteration: |
| 91 | + msg = "Unexpected end of git diff output while looking for --- line" |
| 92 | + raise ValueError(msg) # noqa: B904 |
| 93 | + |
| 94 | + # When we get here, we are on the --- line |
| 95 | + # It should always be followed by a +++ line |
| 96 | + old_file_line = line |
| 97 | + |
| 98 | + try: |
| 99 | + new_file_line = next(line_iterator) |
| 100 | + except StopIteration: |
| 101 | + msg = "Unexpected end of git diff output while looking for +++ line" |
| 102 | + raise ValueError(msg) # noqa: B904 |
| 103 | + if not new_file_line.startswith("+++ "): |
| 104 | + msg = f"Unexpected line in git diff output, expected +++ line: {new_file_line}" |
| 105 | + raise ValueError(msg) |
| 106 | + |
| 107 | + old_file_path = old_file_line[4:].strip() |
| 108 | + new_file_path = new_file_line[4:].strip() |
| 109 | + |
| 110 | + if old_file_path == "/dev/null": |
| 111 | + current_type = ChangeType.ADDED |
| 112 | + current_file = Path(new_file_path) |
| 113 | + elif new_file_path == "/dev/null": |
| 114 | + current_type = ChangeType.DELETED |
| 115 | + current_file = Path(old_file_path) |
| 116 | + elif old_file_path == new_file_path: |
| 117 | + current_type = ChangeType.MODIFIED |
| 118 | + current_file = Path(new_file_path) |
| 119 | + else: |
| 120 | + msg = f"Unexpected file paths in git diff output: {old_file_path} -> {new_file_path} - this indicates a rename which we do not support" |
| 121 | + raise ValueError( |
| 122 | + msg, |
| 123 | + ) |
| 124 | + |
| 125 | + # Now, we should be at the start of the patch hunks (lines starting with @@) |
| 126 | + line = next(line_iterator) |
| 127 | + if not line.startswith("@@ "): |
| 128 | + msg = f"Unexpected line in git diff output, expected hunk start: {line}" |
| 129 | + raise ValueError(msg) |
| 130 | + current_patch_lines.append(line) |
| 131 | + # Collect all hunk lines and lines starting with + or - |
| 132 | + line = next(line_iterator) |
| 133 | + while line.startswith(("+", "-")): |
| 134 | + current_patch_lines.append(line) |
| 135 | + try: |
| 136 | + line = next(line_iterator) |
| 137 | + except StopIteration: |
| 138 | + # Just break out of the loop, we will handle yielding below |
| 139 | + # Set a flag to indicate we reached the end of the iterator |
| 140 | + iterator_exhausted = True |
| 141 | + break |
| 142 | + |
| 143 | + # Yield the file we were building now that we have reached the end of its patch |
| 144 | + yield cls( |
| 145 | + file=current_file, |
| 146 | + type=current_type, |
| 147 | + patch="\n".join(current_patch_lines), |
| 148 | + ) |
| 149 | + current_file = None |
| 150 | + current_type = None |
| 151 | + current_patch_lines = [] |
| 152 | + |
| 153 | + if iterator_exhausted: |
| 154 | + return |
| 155 | + |
| 156 | + except StopIteration: |
| 157 | + msg = "Unexpected end of git diff output while parsing" |
| 158 | + raise ValueError(msg) # noqa: B904 |
| 159 | + |
| 160 | + |
| 161 | +# Easier and safer to subclass UserDict than Dict directly |
| 162 | +class ChangeSet(dict[Path, FileChanges]): |
| 163 | + """ |
| 164 | + Represents a set of changes to files in a git repository. |
| 165 | + This can both be a change between two commits, or the changes in the working directory. |
| 166 | +
|
| 167 | + When considering the changes to the working directory, the untracked files are considered as added files. |
| 168 | + """ |
| 169 | + |
| 170 | + @cached_property |
| 171 | + def added(self) -> set[Path]: |
| 172 | + """List of files that were added.""" |
| 173 | + return {change.file for change in self.values() if change.type == ChangeType.ADDED} |
| 174 | + |
| 175 | + @cached_property |
| 176 | + def modified(self) -> set[Path]: |
| 177 | + """List of files that were modified.""" |
| 178 | + return {change.file for change in self.values() if change.type == ChangeType.MODIFIED} |
| 179 | + |
| 180 | + @cached_property |
| 181 | + def deleted(self) -> set[Path]: |
| 182 | + """List of files that were deleted.""" |
| 183 | + return {change.file for change in self.values() if change.type == ChangeType.DELETED} |
| 184 | + |
| 185 | + @cached_property |
| 186 | + def changed(self) -> set[Path]: |
| 187 | + """List of files that were changed (added, modified, or deleted).""" |
| 188 | + return set(self.keys()) |
| 189 | + |
| 190 | + def add(self, change: FileChanges) -> None: |
| 191 | + """Add a file change to the changeset.""" |
| 192 | + self[change.file] = change |
| 193 | + |
| 194 | + def digest(self) -> SHA1Hash: |
| 195 | + """Compute a hash of the changeset.""" |
| 196 | + from hashlib import sha1 |
| 197 | + |
| 198 | + digester = sha1() # noqa: S324 |
| 199 | + for change in sorted(self.values()): |
| 200 | + digester.update(change.file.as_posix().encode()) |
| 201 | + digester.update(change.type.value.encode()) |
| 202 | + digester.update(change.patch.encode()) |
| 203 | + |
| 204 | + return SHA1Hash(digester.hexdigest()) |
| 205 | + |
| 206 | + @classmethod |
| 207 | + def generate_from_diff_output(cls, diff_output: Iterable[str] | str) -> Self: |
| 208 | + """Generate a changeset from the output of a git diff command.""" |
| 209 | + changeset = cls() |
| 210 | + for change in FileChanges.generate_from_diff_output(diff_output): |
| 211 | + changeset.add(change) |
| 212 | + return changeset |
0 commit comments