zzet--gortex/internal/parser/languages/cmake.go

package languages

import (
	"regexp"
	"strings"

	"github.com/zzet/gortex/internal/graph"
	"github.com/zzet/gortex/internal/parser"
)

// CMake is command-call-structured. `function(NAME ...)` /
// `macro(NAME ...)` introduce callable blocks terminated by
// `endfunction()` / `endmacro()`; `add_library` / `add_executable`
// declare build targets (modelled as function nodes); `set(NAME ...)`
// declares variables; `include(...)` and `add_subdirectory(...)`
// are imports.
var (
	cmakeFunctionRe      = regexp.MustCompile(`(?mi)^\s*function\s*\(\s*([A-Za-z_][\w-]*)`)
	cmakeMacroRe         = regexp.MustCompile(`(?mi)^\s*macro\s*\(\s*([A-Za-z_][\w-]*)`)
	cmakeIncludeRe       = regexp.MustCompile(`(?mi)^\s*include\s*\(\s*([^)\s]+)`)
	cmakeAddSubdirRe     = regexp.MustCompile(`(?mi)^\s*add_subdirectory\s*\(\s*([^)\s]+)`)
	cmakeSetRe           = regexp.MustCompile(`(?mi)^\s*set\s*\(\s*([A-Za-z_][\w]*)`)
	cmakeAddLibraryRe    = regexp.MustCompile(`(?mi)^\s*add_library\s*\(\s*([A-Za-z_][\w-]*)`)
	cmakeAddExecutableRe = regexp.MustCompile(`(?mi)^\s*add_executable\s*\(\s*([A-Za-z_][\w-]*)`)
)

// CMakeExtractor extracts CMake source using regex.
type CMakeExtractor struct{}

func NewCMakeExtractor() *CMakeExtractor { return &CMakeExtractor{} }

func (e *CMakeExtractor) Language() string     { return "cmake" }
func (e *CMakeExtractor) Extensions() []string { return []string{".cmake", "CMakeLists.txt"} }

func (e *CMakeExtractor) Extract(filePath string, src []byte) (*parser.ExtractionResult, error) {
	lines := strings.Split(string(src), "\n")
	result := &parser.ExtractionResult{}

	fileNode := &graph.Node{
		ID: filePath, Kind: graph.KindFile, Name: filePath,
		FilePath: filePath, StartLine: 1, EndLine: len(lines),
		Language: "cmake",
	}
	result.Nodes = append(result.Nodes, fileNode)

	seen := make(map[string]bool)
	add := func(name string, kind graph.NodeKind, start, end int) {
		if name == "" {
			return
		}
		id := filePath + "::" + name
		if seen[id] {
			return
		}
		seen[id] = true
		result.Nodes = append(result.Nodes, &graph.Node{
			ID: id, Kind: kind, Name: name,
			FilePath: filePath, StartLine: start, EndLine: end,
			Language: "cmake",
		})
		result.Edges = append(result.Edges, &graph.Edge{
			From: fileNode.ID, To: id, Kind: graph.EdgeDefines,
			FilePath: filePath, Line: start,
		})
	}

	for _, m := range cmakeFunctionRe.FindAllSubmatchIndex(src, -1) {
		name := string(src[m[2]:m[3]])
		line := lineAt(src, m[0])
		add(name, graph.KindFunction, line, findKeywordBlockEnd(lines, line, "endfunction"))
	}
	for _, m := range cmakeMacroRe.FindAllSubmatchIndex(src, -1) {
		name := string(src[m[2]:m[3]])
		line := lineAt(src, m[0])
		add(name, graph.KindFunction, line, findKeywordBlockEnd(lines, line, "endmacro"))
	}
	for _, m := range cmakeAddLibraryRe.FindAllSubmatchIndex(src, -1) {
		name := string(src[m[2]:m[3]])
		line := lineAt(src, m[0])
		add(name, graph.KindFunction, line, line)
	}
	for _, m := range cmakeAddExecutableRe.FindAllSubmatchIndex(src, -1) {
		name := string(src[m[2]:m[3]])
		line := lineAt(src, m[0])
		add(name, graph.KindFunction, line, line)
	}
	for _, m := range cmakeSetRe.FindAllSubmatchIndex(src, -1) {
		name := string(src[m[2]:m[3]])
		line := lineAt(src, m[0])
		add(name, graph.KindVariable, line, line)
	}

	for _, m := range cmakeIncludeRe.FindAllSubmatchIndex(src, -1) {
		mod := strings.Trim(string(src[m[2]:m[3]]), `"`)
		line := lineAt(src, m[0])
		result.Edges = append(result.Edges, &graph.Edge{
			From: fileNode.ID, To: "unresolved::import::" + mod,
			Kind: graph.EdgeImports, FilePath: filePath, Line: line,
		})
	}
	for _, m := range cmakeAddSubdirRe.FindAllSubmatchIndex(src, -1) {
		mod := strings.Trim(string(src[m[2]:m[3]]), `"`)
		line := lineAt(src, m[0])
		result.Edges = append(result.Edges, &graph.Edge{
			From: fileNode.ID, To: "unresolved::import::" + mod,
			Kind: graph.EdgeImports, FilePath: filePath, Line: line,
		})
	}

	return result, nil
}

var _ parser.Extractor = (*CMakeExtractor)(nil)