c65gm/internal/preproc/preproc.go

package preproc

import (
	"bufio"
	"errors"
	"fmt"
	"os"
	"path/filepath"
	"strings"
)

// Line represents one post-processed source line and its provenance.
type Line struct {
	Text     string   // post-preprocessor line text (after define replacement)
	Filename string   // file the line came from (after resolving includes)
	LineNo   int      // 1-based line number in Filename
	Tokens   []string // whitespace-split tokens from Text (space or tab; consecutive collapsed)
}

// HaltError is returned when a `#HALT` directive is encountered.
type HaltError struct{}

func (HaltError) Error() string { return "preprocessor HALT" }

// PreProcess processes the given root source file and returns the flattened, define-expanded
// lines. Directive lines are not emitted.
func PreProcess(rootFilename string) ([]Line, error) {
	pp := newPreproc()
	return pp.run(rootFilename)
}

// -------------------- internal --------------------

type preproc struct {
	defs   *DefineList       // from definelist.go
	pragma map[string]string // #PRAGMA NAME VALUE (stored, not interpreted here)
	cond   []bool            // conditional stack; a line is active if all are true
}

func newPreproc() *preproc {
	return &preproc{
		defs:   NewDefineList(),
		pragma: make(map[string]string),
		cond:   []bool{},
	}
}

func (p *preproc) run(root string) ([]Line, error) {
	var out []Line

	type frame struct {
		path string
		f    *os.File
		s    *bufio.Scanner
		line int
		dir  string
	}

	newFrame := func(path string) (*frame, error) {
		f, err := os.Open(path)
		if err != nil {
			return nil, err
		}
		sc := bufio.NewScanner(f)
		// Allow long lines (Pascal source and macro expansion can be large)
		sc.Buffer(make([]byte, 0, 64*1024), 10*1024*1024)
		return &frame{path: path, f: f, s: sc, line: 0, dir: filepath.Dir(path)}, nil
	}

	var stack []*frame

	absRoot, _ := filepath.Abs(root)
	fr, err := newFrame(absRoot)
	if err != nil {
		return nil, err
	}
	stack = append(stack, fr)

	for len(stack) > 0 {
		fr := stack[len(stack)-1]
		if !fr.s.Scan() {
			if err := fr.s.Err(); err != nil {
				_ = fr.f.Close()
				return nil, err
			}
			_ = fr.f.Close()
			stack = stack[:len(stack)-1]
			continue
		}
		fr.line++
		raw := fr.s.Text()

		trim := strings.TrimSpace(raw)
		isDirective := strings.HasPrefix(trim, "#")
		active := p.isActive()

		if isDirective {
			parts := strings.Fields(trim)
			if len(parts) == 0 {
				continue
			}
			switch strings.ToUpper(parts[0]) {
			case "#DEFINE":
				if active && len(parts) >= 2 {
					name := parts[1]
					val := ""
					if len(parts) > 2 {
						val = strings.Join(parts[2:], " ")
						val = p.defs.ReplaceDefines(val) // allow nested defines in values
					}
					p.defs.Add(name, val)
				}
				continue
			case "#UNDEF":
				if active && len(parts) >= 2 {
					p.defs.Delete(parts[1])
				}
				continue
			case "#IFDEF":
				if len(parts) != 2 {
					return nil, fmt.Errorf("#IFDEF requires exactly one argument at %s:%d", fr.path, fr.line)
				}
				p.cond = append(p.cond, p.defs.Defined(parts[1]))
				continue

			case "#IFNDEF":
				if len(parts) != 2 {
					return nil, fmt.Errorf("#IFNDEF requires exactly one argument at %s:%d", fr.path, fr.line)
				}
				p.cond = append(p.cond, !p.defs.Defined(parts[1]))
				continue
			case "#IFEND":
				if len(p.cond) > 0 {
					p.cond = p.cond[:len(p.cond)-1]
				}
				continue
			case "#INCLUDE":
				if active {
					if len(parts) < 2 {
						return nil, fmt.Errorf("#INCLUDE without path at %s:%d", fr.path, fr.line)
					}
					incPathRaw := parts[1]
					nextPath, err := resolveInclude(incPathRaw, fr.dir)
					if err != nil {
						return nil, fmt.Errorf("%s at %s:%d", err, fr.path, fr.line)
					}
					next, err := newFrame(nextPath)
					if err != nil {
						return nil, fmt.Errorf("include open failed %q: %w", nextPath, err)
					}
					stack = append(stack, next)
				}
				continue
			case "#PRINT":
				if active {
					msg := strings.TrimSpace(strings.TrimPrefix(trim, "#PRINT"))
					msg = p.defs.ReplaceDefines(msg)
					fmt.Println(msg)
				}
				continue
			case "#HALT":
				if active {
					return nil, HaltError{}
				}
				continue
			case "#PRAGMA":
				if active && len(parts) >= 2 {
					name := strings.ToUpper(parts[1])
					val := ""
					if len(parts) > 2 {
						val = strings.Join(parts[2:], " ")
						val = p.defs.ReplaceDefines(val)
					}
					p.pragma[name] = val
				}
				continue
			default:
				// Unknown directive: drop (do not emit)
				continue
			}
		}

		if !active {
			continue
		}

		// Non-directive: expand defines and emit.
		text := p.defs.ReplaceDefines(raw)
		out = append(out, Line{
			Text:     text,
			Filename: fr.path,
			LineNo:   fr.line,
			Tokens:   strings.Fields(text),
		})
	}

	// Check any scanner errors on remaining frames (if any)
	for _, fr := range stack {
		if err := fr.s.Err(); err != nil {
			return nil, err
		}
	}
	return out, nil
}

func (p *preproc) isActive() bool {
	for _, v := range p.cond {
		if !v {
			return false
		}
	}
	return true
}

func resolveInclude(spec string, curDir string) (string, error) {
	if len(spec) == 0 {
		return "", errors.New("empty include path")
	}
	// <file> -> library include via C65LIBPATH
	if strings.HasPrefix(spec, "<") && strings.HasSuffix(spec, ">") {
		base := strings.TrimSpace(spec[1 : len(spec)-1])
		lib := os.Getenv("C65LIBPATH")
		if lib == "" {
			return "", errors.New("C65LIBPATH not set for angle-bracket include")
		}
		path := filepath.Join(lib, base)
		if _, err := os.Stat(path); err != nil {
			return "", fmt.Errorf("library include not found: %s", path)
		}
		return path, nil
	}
	// quoted or bare -> relative to current file dir
	base := strings.Trim(spec, `"'`)
	path := filepath.Join(curDir, base)
	if _, err := os.Stat(path); err != nil {
		return "", fmt.Errorf("include not found: %s", path)
	}
	return path, nil
}

// fieldsCollapsed splits on spaces and tabs, ignoring empties (i.e., collapses runs).
/*
func fieldsCollapsed(s string) []string {
	out := make([]string, 0, 8)
	field := strings.Builder{}
	flush := func() {
		if field.Len() > 0 {
			out = append(out, field.String())
			field.Reset()
		}
	}
	for _, r := range s {
		if r == ' ' || r == '\t' {
			flush()
		} else {
			field.WriteRune(r)
		}
	}
	flush()
	return out
}
*/