claudetool/patch.go

package claudetool

import (
	"bytes"
	"context"
	"encoding/json"
	"errors"
	"fmt"
	"go/parser"
	"go/token"
	"log/slog"
	"os"
	"path/filepath"
	"strings"

	"sketch.dev/claudetool/editbuf"
	"sketch.dev/claudetool/patchkit"
	"sketch.dev/llm"
)

// Patch is a tool for precise text modifications in files.
var Patch = &llm.Tool{
	Name:        PatchName,
	Description: strings.TrimSpace(PatchDescription),
	InputSchema: llm.MustSchema(PatchInputSchema),
	Run:         PatchRun,
}

const (
	PatchName        = "patch"
	PatchDescription = `
File modification tool for precise text edits.

Operations:
- replace: Substitute text with new content
- append_eof: Append new text at the end of the file
- prepend_bof: Insert new text at the beginning of the file
- overwrite: Replace the entire file with new content (automatically creates the file)

Usage notes:
- All inputs are interpreted literally (no automatic newline or whitespace handling)
- For replace operations, oldText must appear EXACTLY ONCE in the file
`

	// If you modify this, update the termui template for prettier rendering.
	PatchInputSchema = `
{
  "type": "object",
  "required": ["path", "patches"],
  "properties": {
    "path": {
      "type": "string",
      "description": "Absolute path to the file to patch"
    },
    "patches": {
      "type": "array",
      "description": "List of patch requests to apply",
      "items": {
        "type": "object",
        "required": ["operation", "newText"],
        "properties": {
          "operation": {
            "type": "string",
            "enum": ["replace", "append_eof", "prepend_bof", "overwrite"],
            "description": "Type of operation to perform"
          },
          "oldText": {
            "type": "string",
            "description": "Text to locate for the operation (must be unique in file, required for replace)"
          },
          "newText": {
            "type": "string",
            "description": "The new text to use (empty for deletions)"
          }
        }
      }
    }
  }
}
`
)

// TODO: maybe rename PatchRequest to PatchOperation or PatchSpec or PatchPart or just Patch?

type patchInput struct {
	Path    string         `json:"path"`
	Patches []patchRequest `json:"patches"`
}

type patchRequest struct {
	Operation string `json:"operation"`
	OldText   string `json:"oldText,omitempty"`
	NewText   string `json:"newText,omitempty"`
}

// PatchRun is the entry point for the user_patch tool.
func PatchRun(ctx context.Context, m json.RawMessage) ([]llm.Content, error) {
	var input patchInput
	if err := json.Unmarshal(m, &input); err != nil {
		return nil, fmt.Errorf("failed to unmarshal user_patch input: %w", err)
	}

	// Validate the input
	if !filepath.IsAbs(input.Path) {
		return nil, fmt.Errorf("path %q is not absolute", input.Path)
	}
	if len(input.Patches) == 0 {
		return nil, fmt.Errorf("no patches provided")
	}
	// TODO: check whether the file is autogenerated, and if so, require a "force" flag to modify it.

	orig, err := os.ReadFile(input.Path)
	// If the file doesn't exist, we can still apply patches
	// that don't require finding existing text.
	switch {
	case errors.Is(err, os.ErrNotExist):
		for _, patch := range input.Patches {
			switch patch.Operation {
			case "prepend_bof", "append_eof", "overwrite":
			default:
				return nil, fmt.Errorf("file %q does not exist", input.Path)
			}
		}
	case err != nil:
		return nil, fmt.Errorf("failed to read file %q: %w", input.Path, err)
	}

	likelyGoFile := strings.HasSuffix(input.Path, ".go")

	autogenerated := likelyGoFile && IsAutogeneratedGoFile(orig)
	parsed := likelyGoFile && parseGo(orig) != nil

	origStr := string(orig)
	// Process the patches "simultaneously", minimizing them along the way.
	// Claude generates patches that interact with each other.
	buf := editbuf.NewBuffer(orig)

	// TODO: is it better to apply the patches that apply cleanly and report on the failures?
	// or instead have it be all-or-nothing?
	// For now, it is all-or-nothing.
	// TODO: when the model gets into a "cannot apply patch" cycle of doom, how do we get it unstuck?
	// Also: how do we detect that it's in a cycle?
	var patchErr error
	for i, patch := range input.Patches {
		switch patch.Operation {
		case "prepend_bof":
			buf.Insert(0, patch.NewText)
		case "append_eof":
			buf.Insert(len(orig), patch.NewText)
		case "overwrite":
			buf.Replace(0, len(orig), patch.NewText)
		case "replace":
			if patch.OldText == "" {
				return nil, fmt.Errorf("patch %d: oldText cannot be empty for %s operation", i, patch.Operation)
			}

			// Attempt to apply the patch.
			spec, count := patchkit.Unique(origStr, patch.OldText, patch.NewText)
			switch count {
			case 0:
				// no matches, maybe recoverable, continued below
			case 1:
				// exact match, apply
				slog.DebugContext(ctx, "patch_applied", "method", "unique")
				spec.ApplyToEditBuf(buf)
				continue
			case 2:
				// multiple matches
				patchErr = errors.Join(patchErr, fmt.Errorf("old text not unique:\n%s", patch.OldText))
			default:
				// TODO: return an error instead of using agentPatch
				slog.ErrorContext(ctx, "unique returned unexpected count", "count", count)
				patchErr = errors.Join(patchErr, fmt.Errorf("internal error"))
				continue
			}

			// The following recovery mechanisms are heuristic.
			// They aren't perfect, but they appear safe,
			// and the cases they cover appear with some regularity.

			// Try adjusting the whitespace prefix.
			spec, ok := patchkit.UniqueDedent(origStr, patch.OldText, patch.NewText)
			if ok {
				slog.DebugContext(ctx, "patch_applied", "method", "unique_dedent")
				spec.ApplyToEditBuf(buf)
				continue
			}

			// Try ignoring leading/trailing whitespace in a semantically safe way.
			spec, ok = patchkit.UniqueInValidGo(origStr, patch.OldText, patch.NewText)
			if ok {
				slog.DebugContext(ctx, "patch_applied", "method", "unique_in_valid_go")
				spec.ApplyToEditBuf(buf)
				continue
			}

			// Try ignoring semantically insignificant whitespace.
			spec, ok = patchkit.UniqueGoTokens(origStr, patch.OldText, patch.NewText)
			if ok {
				slog.DebugContext(ctx, "patch_applied", "method", "unique_go_tokens")
				spec.ApplyToEditBuf(buf)
				continue
			}

			// Try trimming the first line of the patch, if we can do so safely.
			spec, ok = patchkit.UniqueTrim(origStr, patch.OldText, patch.NewText)
			if ok {
				slog.DebugContext(ctx, "patch_applied", "method", "unique_trim")
				spec.ApplyToEditBuf(buf)
				continue
			}

			// No dice.
			patchErr = errors.Join(patchErr, fmt.Errorf("old text not found:\n%s", patch.OldText))
			continue
		default:
			return nil, fmt.Errorf("unrecognized operation %q", patch.Operation)
		}
	}

	if patchErr != nil {
		return nil, patchErr
	}

	patched, err := buf.Bytes()
	if err != nil {
		return nil, err
	}
	if err := os.MkdirAll(filepath.Dir(input.Path), 0o700); err != nil {
		return nil, fmt.Errorf("failed to create directory %q: %w", filepath.Dir(input.Path), err)
	}
	if err := os.WriteFile(input.Path, patched, 0o600); err != nil {
		return nil, fmt.Errorf("failed to write patched contents to file %q: %w", input.Path, err)
	}

	response := new(strings.Builder)
	fmt.Fprintf(response, "- Applied all patches\n")

	if parsed {
		parseErr := parseGo(patched)
		if parseErr != nil {
			return nil, fmt.Errorf("after applying all patches, the file no longer parses:\n%w", parseErr)
		}
	}

	if autogenerated {
		fmt.Fprintf(response, "- WARNING: %q appears to be autogenerated. Patches were applied anyway.\n", input.Path)
	}

	// TODO: maybe report the patch result to the model, i.e. some/all of the new code after the patches and formatting.
	return llm.TextContent(response.String()), nil
}

func parseGo(buf []byte) error {
	fset := token.NewFileSet()
	_, err := parser.ParseFile(fset, "", buf, parser.SkipObjectResolution)
	return err
}

// IsAutogeneratedGoFile reports whether a Go file has markers indicating it was autogenerated.
func IsAutogeneratedGoFile(buf []byte) bool {
	for _, sig := range autogeneratedSignals {
		if bytes.Contains(buf, []byte(sig)) {
			return true
		}
	}

	// https://pkg.go.dev/cmd/go#hdr-Generate_Go_files_by_processing_source
	// "This line must appear before the first non-comment, non-blank text in the file."
	// Approximate that by looking for it at the top of the file, before the last of the imports.
	// (Sometimes people put it after the package declaration, because of course they do.)
	// At least in the imports region we know it's not part of their actual code;
	// we don't want to ignore the generator (which also includes these strings!),
	// just the generated code.
	fset := token.NewFileSet()
	f, err := parser.ParseFile(fset, "x.go", buf, parser.ImportsOnly|parser.ParseComments)
	if err == nil {
		for _, cg := range f.Comments {
			t := strings.ToLower(cg.Text())
			for _, sig := range autogeneratedHeaderSignals {
				if strings.Contains(t, sig) {
					return true
				}
			}
		}
	}

	return false
}

// autogeneratedSignals are signals that a file is autogenerated, when present anywhere in the file.
var autogeneratedSignals = [][]byte{
	[]byte("\nfunc bindataRead("), // pre-embed bindata packed file
}

// autogeneratedHeaderSignals are signals that a file is autogenerated, when present at the top of the file.
var autogeneratedHeaderSignals = []string{
	// canonical would be `(?m)^// Code generated .* DO NOT EDIT\.$`
	// but people screw it up, a lot, so be more lenient
	strings.ToLower("generate"),
	strings.ToLower("DO NOT EDIT"),
	strings.ToLower("export by"),
}