gptdiff 0.1.22__tar.gz → 0.1.27__tar.gz

{gptdiff-0.1.22 → gptdiff-0.1.27}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.2
 Name: gptdiff
-Version: 0.1.22
+Version: 0.1.27
 Summary: A tool to generate and apply git diffs using LLMs
 Author: 255labs
 Classifier: License :: OSI Approved :: MIT License
@@ -82,13 +82,15 @@ done
 
 *Requires reasoning model*
 
-### Why GPTDiff?
+## Why Choose GPTDiff?
 
-- **Understands Your Code** - Describe changes in plain English
-- **Safe Modifications** - Keeps existing code working
-- **Auto-Fix** - `--apply` fixes mistakes in generated changes
-- **Works Instantly** - No complex setup needed
-- **Whole Project View** - Handles multiple files together
+- **Describe changes in plain English**  
+- **AI gets your whole project**  
+- **Auto-fixes conflicts**  
+- **Keeps code functional**  
+- **Fast setup, no fuss**  
+- **You approve every change**  
+- **Costs are upfront**
 
 ## Core Capabilities
 

{gptdiff-0.1.22 → gptdiff-0.1.27}/README.md

@@ -55,13 +55,15 @@ done
 
 *Requires reasoning model*
 
-### Why GPTDiff?
-
-- **Understands Your Code** - Describe changes in plain English
-- **Safe Modifications** - Keeps existing code working
-- **Auto-Fix** - `--apply` fixes mistakes in generated changes
-- **Works Instantly** - No complex setup needed
-- **Whole Project View** - Handles multiple files together
+## Why Choose GPTDiff?
+
+- **Describe changes in plain English**  
+- **AI gets your whole project**  
+- **Auto-fixes conflicts**  
+- **Keeps code functional**  
+- **Fast setup, no fuss**  
+- **You approve every change**  
+- **Costs are upfront**
 
 ## Core Capabilities
 

gptdiff-0.1.27/gptdiff/applydiff.py

@@ -0,0 +1,265 @@
+"""
+Module: applydiff
+
+Contains the function to apply unified git diffs to files on disk.
+"""
+
+from pathlib import Path
+import re
+import hashlib
+
+def apply_diff(project_dir, diff_text):
+    """
+    Applies a unified diff (as generated by git diff) to the files in project_dir
+    using pure Python (without calling the external 'patch' command).
+
+    Handles file modifications, new file creation, and file deletions.
+
+    Returns:
+        True if at least one file was modified (or deleted/created) as a result of the patch,
+        False otherwise.
+    """
+    from pathlib import Path
+    import re, hashlib
+
+    def file_hash(filepath):
+        h = hashlib.sha256()
+        with open(filepath, "rb") as f:
+            h.update(f.read())
+        return h.hexdigest()
+
+    def apply_patch_to_file(file_path, patch):
+        """
+        Applies a unified diff patch (for a single file) to file_path.
+
+        Returns True if the patch was applied successfully, False otherwise.
+        """
+        # Read the original file lines; if the file doesn't exist, treat it as empty.
+        if file_path.exists():
+            original_lines = file_path.read_text(encoding="utf8").splitlines(keepends=True)
+        else:
+            original_lines = []
+        new_lines = []
+        current_index = 0
+
+        patch_lines = patch.splitlines()
+        # Regex for a hunk header, e.g., @@ -3,7 +3,6 @@
+        hunk_header_re = re.compile(r"^@@(?: -(\d+)(?:,(\d+))?)?(?: \+(\d+)(?:,(\d+))?)? @@")
+        i = 0
+        while i < len(patch_lines):
+            line = patch_lines[i]
+            if line.lstrip().startswith("@@"):
+                if line.strip() == "@@":
+                    # Handle minimal hunk header without line numbers.
+                    orig_start = 1
+                else:
+                    m = hunk_header_re.match(line.strip())
+                    if not m:
+                        print("Invalid hunk header:", line)
+                        return False
+                    orig_start = int(m.group(1)) if m.group(1) is not None else 1
+                hunk_start_index = orig_start - 1  # diff headers are 1-indexed
+                if hunk_start_index > len(original_lines):
+                    print("Hunk start index beyond file length")
+                    return False
+                new_lines.extend(original_lines[current_index:hunk_start_index])
+                current_index = hunk_start_index
+                i += 1
+                # Process the hunk lines until the next hunk header.
+                while i < len(patch_lines) and not patch_lines[i].startswith("@@"):
+                    pline = patch_lines[i]
+                    if pline.startswith(" "):
+                        # Context line must match exactly.
+                        expected = pline[1:]
+                        if current_index >= len(original_lines):
+                            print("Context line expected but file ended")
+                            return False
+                        orig_line = original_lines[current_index].rstrip("\n")
+                        if orig_line != expected:
+                            print("Context line mismatch. Expected:", expected, "Got:", orig_line)
+                            return False
+                        new_lines.append(original_lines[current_index])
+                        current_index += 1
+                    elif pline.startswith("-"):
+                        # Removal line: verify and skip from original.
+                        expected = pline[1:]
+                        if current_index >= len(original_lines):
+                            print("Removal line expected but file ended")
+                            return False
+                        orig_line = original_lines[current_index].rstrip("\n")
+                        if orig_line != expected:
+                            print("Removal line mismatch. Expected:", expected, "Got:", orig_line)
+                            return False
+                        current_index += 1
+                    elif pline.startswith("+"):
+                        # Addition line: add to new_lines.
+                        new_lines.append(pline[1:] + "\n")
+                    else:
+                        print("Unexpected line in hunk:", pline)
+                        return False
+                    i += 1
+            else:
+                # Skip non-hunk header lines.
+                i += 1
+
+        # Append any remaining lines from the original file.
+        new_lines.extend(original_lines[current_index:])
+        # Ensure parent directories exist before writing the file.
+        file_path.parent.mkdir(parents=True, exist_ok=True)
+        # Write the new content back to the file.
+        file_path.write_text("".join(new_lines), encoding="utf8")
+        return True
+
+    # Parse the diff into per-file patches.
+    file_patches = parse_diff_per_file(diff_text)
+    if not file_patches:
+        print("No file patches found in diff.")
+        return False
+
+    # Record original file hashes.
+    original_hashes = {}
+    for file_path, _ in file_patches:
+        target_file = Path(project_dir) / file_path
+        if target_file.exists():
+            original_hashes[file_path] = file_hash(target_file)
+        else:
+            original_hashes[file_path] = None
+
+    any_change = False
+    # Process each file patch.
+    for file_path, patch in file_patches:
+        target_file = Path(project_dir) / file_path
+        if "+++ /dev/null" in patch:
+            # Deletion patch: delete the file if it exists.
+            if target_file.exists():
+                target_file.unlink()
+                if not target_file.exists():
+                    any_change = True
+                else:
+                    print(f"Failed to delete file: {target_file}")
+                    return False
+        else:
+            # Modification or new file creation.
+            success = apply_patch_to_file(target_file, patch)
+            if not success:
+                print(f"Failed to apply patch to file: {target_file}")
+                return False
+
+    # Verify that at least one file was changed by comparing hashes.
+    for file_path, patch in file_patches:
+        target_file = Path(project_dir) / file_path
+        if "+++ /dev/null" in patch:
+            if not target_file.exists():
+                any_change = True
+            else:
+                print(f"Expected deletion but file still exists: {target_file}")
+                return False
+        else:
+            old_hash = original_hashes.get(file_path)
+            if target_file.exists():
+                new_hash = file_hash(target_file)
+                if old_hash != new_hash:
+                    any_change = True
+                else:
+                    print(f"No change detected in file: {target_file}")
+            else:
+                print(f"Expected modification or creation but file is missing: {target_file}")
+                return False
+
+    if not any_change:
+        print("Patch applied but no file modifications detected.")
+        return False
+    return True
+
+def parse_diff_per_file(diff_text):
+    """Parse unified diff text into individual file patches.
+
+    Splits a multi-file diff into per-file entries for processing. Handles:
+    - File creations (+++ /dev/null)
+    - File deletions (--- /dev/null)
+    - Standard modifications
+
+    Args:
+        diff_text: Unified diff string as generated by `git diff`
+
+    Returns:
+        List of tuples (file_path, patch) where:
+        - file_path: Relative path to modified file
+        - patch: Full diff fragment for this file
+
+    Note:
+        Uses 'b/' prefix detection from git diffs to determine target paths
+        This doesn't work all the time and needs to be revised with stronger models
+    """
+    header_re = re.compile(r'^(?:diff --git\s+)?(a/[^ ]+)\s+(b/[^ ]+)\s*$', re.MULTILINE)
+    lines = diff_text.splitlines()
+
+    # Check if any header line exists.
+    if not any(header_re.match(line) for line in lines):
+        # Fallback strategy: detect file headers starting with '--- a/' or '-- a/'
+        diffs = []
+        current_lines = []
+        current_file = None
+        deletion_mode = False
+        header_line_re = re.compile(r'^-{2,3}\s+a/(.+)$')
+
+        for line in lines:
+            if header_line_re.match(line):
+                if current_file is not None and current_lines:
+                    if deletion_mode and not any(l.startswith("+++ /dev/null") for l in current_lines):
+                        current_lines.append("+++ /dev/null")
+                    diffs.append((current_file, "\n".join(current_lines)))
+                current_lines = [line]
+                deletion_mode = False
+                file_from = header_line_re.match(line).group(1).strip()
+                current_file = file_from
+            else:
+                current_lines.append(line)
+                if "deleted file mode" in line:
+                    deletion_mode = True
+                if line.startswith("+++ "):
+                    parts = line.split()
+                    if len(parts) >= 2:
+                        file_to = parts[1].strip()
+                        if file_to != "/dev/null":
+                            current_file = file_to[2:] if (file_to.startswith("a/") or file_to.startswith("b/")) else file_to
+        if current_file is not None and current_lines:
+            if deletion_mode and not any(l.startswith("+++ ") for l in current_lines):
+                current_lines.append("+++ /dev/null")
+            diffs.append((current_file, "\n".join(current_lines)))
+        return diffs
+    else:
+        # Use header-based strategy.
+        diffs = []
+        current_lines = []
+        current_file = None
+        deletion_mode = False
+        for line in lines:
+            m = header_re.match(line)
+            if m:
+                if current_file is not None and current_lines:
+                    if deletion_mode and not any(l.startswith("+++ ") for l in current_lines):
+                        current_lines.append("+++ /dev/null")
+                    diffs.append((current_file, "\n".join(current_lines)))
+                current_lines = [line]
+                deletion_mode = False
+                file_from = m.group(1)  # e.g. "a/index.html"
+                file_to = m.group(2)    # e.g. "b/index.html"
+                current_file = file_to[2:] if file_to.startswith("b/") else file_to
+            else:
+                current_lines.append(line)
+                if "deleted file mode" in line:
+                    deletion_mode = True
+                if line.startswith("+++ "):
+                    parts = line.split()
+                    if len(parts) >= 2:
+                        file_to = parts[1].strip()
+                        if file_to != "/dev/null":
+                            current_file = file_to[2:] if (file_to.startswith("a/") or file_to.startswith("b/")) else file_to
+        if current_file is not None and current_lines:
+            if deletion_mode and not any(l.startswith("+++ ") for l in current_lines):
+                current_lines.append("+++ /dev/null")
+            diffs.append((current_file, "\n".join(current_lines)))
+        return diffs
+
+