portacode 0.3.19.dev4__py3-none-any.whl → 1.4.11.dev1__py3-none-any.whl

portacode/utils/diff_apply.py

@@ -0,0 +1,456 @@
+"""Utilities for parsing and applying unified diff patches."""
+
+from __future__ import annotations
+
+import os
+import re
+from dataclasses import dataclass
+from typing import List, Optional, Tuple
+
+
+class DiffParseError(Exception):
+    """Raised when a diff cannot be parsed."""
+
+
+class DiffApplyError(Exception):
+    """Raised when a diff cannot be applied to a file."""
+
+    def __init__(self, message: str, *, file_path: Optional[str] = None, line_number: Optional[int] = None):
+        super().__init__(message)
+        self.file_path = file_path
+        self.line_number = line_number
+
+
+@dataclass
+class PatchLine:
+    """Represents a single line in a hunk."""
+
+    op: str  # ' ', '+', '-'
+    text: str
+
+
+@dataclass
+class Hunk:
+    """Represents a parsed hunk."""
+
+    old_start: int
+    old_length: int
+    new_start: int
+    new_length: int
+    lines: List[PatchLine]
+
+
+@dataclass
+class FilePatch:
+    """Represents an entire file diff."""
+
+    old_path: Optional[str]
+    new_path: Optional[str]
+    hunks: List[Hunk]
+
+    @property
+    def is_new_file(self) -> bool:
+        return self.old_path in (None, "/dev/null")
+
+    @property
+    def is_delete(self) -> bool:
+        return self.new_path in (None, "/dev/null")
+
+    @property
+    def target_path(self) -> Optional[str]:
+        if self.is_delete and self.old_path not in (None, "/dev/null"):
+            return self.old_path
+        if self.new_path not in (None, "/dev/null"):
+            return self.new_path
+        return self.old_path
+
+
+_HUNK_HEADER_RE = re.compile(
+    r"@@ -(?P<old_start>\d+)(?:,(?P<old_len>\d+))? \+(?P<new_start>\d+)(?:,(?P<new_len>\d+))? @@"
+)
+
+
+def _normalize_diff_path(raw_path: str) -> Optional[str]:
+    """Normalize diff path lines (handles prefixes and tabs)."""
+    path = raw_path.strip()
+    if not path or path == "/dev/null":
+        return None
+    # Drop git prefixes like a/ and b/
+    if path.startswith("a/") or path.startswith("b/"):
+        path = path[2:]
+    # Remove any trailing metadata after tab (e.g., timestamps)
+    if "\t" in path:
+        path = path.split("\t", 1)[0]
+    return path.strip()
+
+
+def parse_unified_diff(diff_text: str) -> List[FilePatch]:
+    """Parse unified diff text into FilePatch objects."""
+    if not diff_text or not diff_text.strip():
+        raise DiffParseError("Diff content is empty")
+
+    lines = diff_text.splitlines(keepends=True)
+    patches: List[FilePatch] = []
+    i = 0
+
+    while i < len(lines):
+        line = lines[i]
+        if line.startswith("diff --git"):
+            i += 1
+            continue
+
+        if not line.startswith("--- "):
+            i += 1
+            continue
+
+        old_path = _normalize_diff_path(line[4:].strip())
+        i += 1
+        if i >= len(lines) or not lines[i].startswith("+++ "):
+            raise DiffParseError("Missing +++ line after --- line")
+        new_path = _normalize_diff_path(lines[i][4:].strip())
+        i += 1
+
+        hunks: List[Hunk] = []
+        while i < len(lines) and lines[i].startswith("@@"):
+            header = lines[i]
+            match = _HUNK_HEADER_RE.match(header)
+            if not match:
+                raise DiffParseError(f"Invalid hunk header: {header.strip()}")
+            old_start = int(match.group("old_start"))
+            old_len = int(match.group("old_len") or "1")
+            new_start = int(match.group("new_start"))
+            new_len = int(match.group("new_len") or "1")
+            i += 1
+
+            hunk_lines: List[PatchLine] = []
+            while i < len(lines):
+                current_line = lines[i]
+                prefix = current_line[:1]
+                # Stop if we encounter the start of the next file diff
+                if current_line.startswith("diff --git ") or current_line.startswith("--- "):
+                    break
+                if prefix in {" ", "+", "-"}:
+                    # Guard against accidental file headers inside a hunk
+                    if prefix == "-" and current_line.startswith("--- "):
+                        break
+                    if prefix == "+" and current_line.startswith("+++ "):
+                        break
+                    hunk_lines.append(PatchLine(prefix, current_line[1:]))
+                    i += 1
+                elif current_line.startswith("\"):
+                    # Skip metadata line but keep processing
+                    i += 1
+                else:
+                    break
+
+            hunks.append(Hunk(old_start, old_len, new_start, new_len, hunk_lines))
+
+        if not hunks:
+            raise DiffParseError("No hunks found for file diff")
+
+        patches.append(FilePatch(old_path, new_path, hunks))
+
+    if not patches:
+        raise DiffParseError("No valid file patches found in diff")
+
+    return patches
+
+
+def _normalize_target_path(path: str, base_path: Optional[str]) -> str:
+    """Compute the absolute path for a diff target."""
+    if os.path.isabs(path):
+        return path
+    base = base_path or os.getcwd()
+    return os.path.abspath(os.path.join(base, path))
+
+
+def _load_file_lines(path: str) -> Tuple[List[str], bool]:
+    """Load file contents as a list of lines with newline characters preserved."""
+    try:
+        with open(path, "r", encoding="utf-8") as f:
+            data = f.read()
+        return data.splitlines(keepends=True), True
+    except FileNotFoundError:
+        return [], False
+
+
+def _collect_original_line_chunks(hunk: Hunk) -> List[List[str]]:
+    """Collect consecutive lines from the original file that can anchor a hunk."""
+    chunks: List[List[str]] = []
+    current: List[str] = []
+
+    for line in hunk.lines:
+        if line.op in {" ", "-"}:
+            current.append(line.text)
+        elif current:
+            chunks.append(list(current))
+            current = []
+    if current:
+        chunks.append(list(current))
+
+    return [chunk for chunk in chunks if chunk]
+
+
+def _find_unique_chunk_position(original_lines: List[str], chunk: List[str]) -> Optional[int]:
+    """Return the sole index where chunk appears, or None if ambiguous."""
+    if not chunk or len(chunk) > len(original_lines):
+        return None
+
+    matches: List[int] = []
+    max_start = len(original_lines) - len(chunk)
+    for idx in range(0, max_start + 1):
+        if original_lines[idx : idx + len(chunk)] == chunk:
+            matches.append(idx)
+            if len(matches) > 1:
+                break
+    if len(matches) == 1:
+        return matches[0]
+    return None
+
+
+def _find_hunk_fallback_index(
+    original_lines: List[str], hunk: Hunk
+) -> Optional[Tuple[int, int]]:
+    """Try to find a unique match for hunk context and return (index, context_length)."""
+    chunks = _collect_original_line_chunks(hunk)
+    if not chunks:
+        return None
+
+    chunks.sort(key=len, reverse=True)
+    for chunk in chunks:
+        pos = _find_unique_chunk_position(original_lines, chunk)
+        if pos is not None:
+            return pos, len(chunk)
+    return None
+
+
+def _apply_hunk_segment(
+    original_lines: List[str],
+    start_index: int,
+    hunk: Hunk,
+    *,
+    file_path: str,
+) -> Tuple[List[str], int]:
+    """Apply a single hunk starting at a specific index."""
+    if start_index > len(original_lines):
+        raise DiffApplyError(
+            "Hunk start position past end of file",
+            file_path=file_path,
+            line_number=hunk.old_start,
+        )
+
+    result: List[str] = []
+    src_index = start_index
+    current_line_number = start_index + 1
+
+    for line in hunk.lines:
+        if line.op == " ":
+            if src_index >= len(original_lines):
+                raise DiffApplyError(
+                    "Unexpected end of file while matching context",
+                    file_path=file_path,
+                    line_number=current_line_number,
+                )
+            if original_lines[src_index] != line.text:
+                raise DiffApplyError(
+                    f"Context mismatch. Expected {original_lines[src_index]!r} but got {line.text!r}",
+                    file_path=file_path,
+                    line_number=current_line_number,
+                )
+            result.append(line.text)
+            src_index += 1
+            current_line_number += 1
+        elif line.op == "-":
+            if src_index >= len(original_lines):
+                raise DiffApplyError(
+                    "Unexpected end of file while removing line",
+                    file_path=file_path,
+                    line_number=current_line_number,
+                )
+            if original_lines[src_index] != line.text:
+                raise DiffApplyError(
+                    f"Deletion mismatch. Expected {original_lines[src_index]!r} but got {line.text!r}",
+                    file_path=file_path,
+                    line_number=current_line_number,
+                )
+            src_index += 1
+            current_line_number += 1
+        elif line.op == "+":
+            result.append(line.text)
+        else:
+            raise DiffApplyError(
+                f"Unsupported hunk operation {line.op!r}",
+                file_path=file_path,
+                line_number=current_line_number,
+            )
+
+    return result, src_index
+
+
+def _apply_hunks(
+    original_lines: List[str],
+    hunks: List[Hunk],
+    *,
+    file_path: str,
+    heuristic_log: Optional[List[str]] = None,
+) -> List[str]:
+    """Apply hunks to the provided original lines."""
+    result: List[str] = []
+    src_index = 0  # zero-based
+
+    for hunk in hunks:
+        desired_index = max(hunk.old_start - 1, 0)
+        fallback = _find_hunk_fallback_index(original_lines, hunk)
+        attempts: List[int] = [desired_index]
+        fallback_context_len = 0
+        fallback_index: Optional[int] = None
+        if fallback:
+            fallback_index, fallback_context_len = fallback
+            if fallback_index != desired_index and fallback_index >= src_index:
+                attempts.append(fallback_index)
+
+        last_error: Optional[DiffApplyError] = None
+        for attempt_index in attempts:
+            if attempt_index < src_index:
+                continue
+
+            prefix_length = len(result)
+            prefix_src = src_index
+            if attempt_index > src_index:
+                result.extend(original_lines[src_index:attempt_index])
+                src_index = attempt_index
+
+            try:
+                applied_lines, new_src_index = _apply_hunk_segment(
+                    original_lines,
+                    attempt_index,
+                    hunk,
+                    file_path=file_path,
+                )
+                result.extend(applied_lines)
+                src_index = new_src_index
+
+                if (
+                    heuristic_log is not None
+                    and fallback_index is not None
+                    and attempt_index == fallback_index
+                ):
+                    heuristic_log.append(
+                        f"Hunk for {file_path} was shifted from line {desired_index + 1} to {fallback_index + 1} "
+                        f"using a unique {fallback_context_len}-line context match."
+                    )
+                break
+            except DiffApplyError as exc:
+                last_error = exc
+                del result[prefix_length:]
+                src_index = prefix_src
+                continue
+        else:
+            if last_error:
+                raise last_error
+            raise DiffApplyError(
+                "Unable to apply hunk",
+                file_path=file_path,
+                line_number=hunk.old_start,
+            )
+
+    # Append remaining content
+    if src_index < len(original_lines):
+        result.extend(original_lines[src_index:])
+
+    return result
+
+
+def apply_file_patch(
+    file_patch: FilePatch,
+    base_path: Optional[str],
+    heuristic_log: Optional[List[str]] = None,
+) -> Tuple[str, str, int]:
+    """Apply a parsed FilePatch to disk.
+
+    Returns:
+        Tuple[target_path, action, bytes_written]
+    """
+    target_rel = file_patch.target_path
+    if not target_rel:
+        raise DiffApplyError("Unable to determine target path for diff")
+
+    target_path = _normalize_target_path(target_rel, base_path)
+    original_lines, file_exists = _load_file_lines(target_path)
+
+    if file_patch.is_new_file and file_exists:
+        # For new files we expect the file to not exist, but if it does we treat it as modification
+        pass
+    elif not file_patch.is_new_file and not file_exists and not file_patch.is_delete:
+        raise DiffApplyError(f"File does not exist: {target_path}", file_path=target_path)
+
+    if file_patch.is_delete:
+        if not file_exists:
+            raise DiffApplyError(f"File does not exist: {target_path}", file_path=target_path)
+        # Applying hunks ensures they match before deletion
+        updated_lines = _apply_hunks(original_lines, file_patch.hunks, file_path=target_path)
+        if os.path.exists(target_path):
+            os.remove(target_path)
+        return target_path, "deleted", 0
+
+    updated_lines = _apply_hunks(
+        original_lines,
+        file_patch.hunks,
+        file_path=target_path,
+        heuristic_log=heuristic_log,
+    )
+
+    dir_name = os.path.dirname(target_path)
+    if dir_name and not os.path.exists(dir_name):
+        os.makedirs(dir_name, exist_ok=True)
+    with open(target_path, "w", encoding="utf-8") as f:
+        f.write("".join(updated_lines))
+
+    bytes_written = sum(len(chunk.encode("utf-8")) for chunk in updated_lines)
+    action = "created" if not file_exists else "modified"
+    return target_path, action, bytes_written
+
+
+def preview_file_patch(
+    file_patch: FilePatch,
+    base_path: Optional[str],
+    heuristic_log: Optional[List[str]] = None,
+) -> Tuple[str, str, List[str], List[str]]:
+    """Compute the before/after contents for a FilePatch without writing to disk.
+
+    Returns:
+        Tuple[target_path, action, original_lines, updated_lines]
+    """
+    target_rel = file_patch.target_path
+    if not target_rel:
+        raise DiffApplyError("Unable to determine target path for diff preview")
+
+    target_path = _normalize_target_path(target_rel, base_path)
+    original_lines, file_exists = _load_file_lines(target_path)
+
+    if file_patch.is_new_file and file_exists:
+        # Treat as modification to allow previewing changes atop an existing file
+        pass
+    elif not file_patch.is_new_file and not file_exists and not file_patch.is_delete:
+        raise DiffApplyError(f"File does not exist: {target_path}", file_path=target_path)
+
+    if file_patch.is_delete:
+        # Validate the hunks but the resulting file will be removed entirely
+        _apply_hunks(
+            original_lines,
+            file_patch.hunks,
+            file_path=target_path,
+            heuristic_log=heuristic_log,
+        )
+        updated_lines = []
+        action = "deleted"
+    else:
+        updated_lines = _apply_hunks(
+            original_lines,
+            file_patch.hunks,
+            file_path=target_path,
+            heuristic_log=heuristic_log,
+        )
+        action = "created" if not file_exists else "modified"
+
+    return target_path, action, original_lines, updated_lines