Implemented proper .d file parser

Author: cmaglie

internal/arduino/builder/internal/utils/ansi_others.go renamed to internal/arduino/builder/cpp/ansi_others.go

@@ -15,7 +15,7 @@
 
 //go:build !windows
 
-package utils
+package cpp
 
 import (
 	"errors"

internal/arduino/builder/internal/utils/ansi_windows.go renamed to internal/arduino/builder/cpp/ansi_windows.go

@@ -13,7 +13,7 @@
 // Arduino software without disclosing the source code of your own applications.
 // To purchase a commercial license, send an email to [email protected].
 
-package utils
+package cpp
 
 import (
 	"golang.org/x/sys/windows"

internal/arduino/builder/cpp/depfile.go

@@ -0,0 +1,140 @@
+// This file is part of arduino-cli.
+//
+// Copyright 2025 ARDUINO SA (http://www.arduino.cc/)
+//
+// This software is released under the GNU General Public License version 3,
+// which covers the main part of arduino-cli.
+// The terms of this license can be found at:
+// https://www.gnu.org/licenses/gpl-3.0.en.html
+//
+// You can be released from the requirements of the above licenses by purchasing
+// a commercial license. Buying such a license is mandatory if you want to
+// modify or otherwise use the software for commercial activities involving the
+// Arduino software without disclosing the source code of your own applications.
+// To purchase a commercial license, send an email to [email protected].
+
+package cpp
+
+import (
+	"errors"
+	"runtime"
+	"strings"
+	"unicode"
+
+	"github.com/arduino/go-paths-helper"
+	"go.bug.st/f"
+)
+
+// Dependencies represents the dependencies of a source file.
+type Dependencies struct {
+	ObjectFile   string
+	Dependencies []string
+}
+
+// ReadDepFile reads a dependency file and returns the dependencies.
+// It may return nil if the dependency file is empty.
+func ReadDepFile(depFilePath *paths.Path) (*Dependencies, error) {
+	depFileData, err := depFilePath.ReadFile()
+	if err != nil {
+		return nil, err
+	}
+
+	if runtime.GOOS == "windows" {
+		// This is required because on Windows we don't know which encoding is used
+		// by gcc to write the dep file (it could be UTF-8 or any of the Windows
+		// ANSI mappings).
+		if decoded, err := convertAnsiBytesToString(depFileData); err == nil {
+			if res, err := readDepFile(decoded); err == nil && res != nil {
+				return res, nil
+			}
+		}
+		// Fallback to UTF-8...
+	}
+
+	return readDepFile(string(depFileData))
+}
+
+func readDepFile(depFile string) (*Dependencies, error) {
+	rows, err := unescapeAndSplit(strings.ReplaceAll(depFile, "\r\n", "\n"))
+	if err != nil {
+		return nil, err
+	}
+	rows = f.Map(rows, strings.TrimSpace)
+	rows = f.Filter(rows, f.NotEquals(""))
+	if len(rows) == 0 {
+		return &Dependencies{}, nil
+	}
+
+	// The first line of the depfile contains the path to the object file to generate.
+	// The second line of the depfile contains the path to the source file.
+	// All subsequent lines contain the header files necessary to compile the object file.
+
+	if !strings.HasSuffix(rows[0], ":") {
+		return nil, errors.New("no colon in first item of depfile")
+	}
+	res := &Dependencies{
+		ObjectFile:   strings.TrimSuffix(rows[0], ":"),
+		Dependencies: rows[1:],
+	}
+	return res, nil
+}
+
+func unescapeAndSplit(s string) ([]string, error) {
+	var res []string
+	backslash := false
+	dollar := false
+	current := strings.Builder{}
+	for _, c := range s {
+		if backslash {
+			switch c {
+			case ' ':
+				current.WriteByte(' ')
+			case 't':
+				current.WriteByte('\t')
+			case '#':
+				current.WriteByte('#')
+			case '\\':
+				current.WriteByte('\\')
+			case '\n':
+				// ignore
+			default:
+				return nil, errors.New("invalid escape sequence: \\" + string(c))
+			}
+			backslash = false
+			continue
+		}
+		if dollar {
+			if c != '$' {
+				return nil, errors.New("invalid dollar sequence: $" + string(c))
+			}
+			current.WriteByte('$')
+			dollar = false
+			continue
+		}
+
+		if c == '\\' {
+			backslash = true
+			continue
+		}
+		if c == '$' {
+			dollar = true
+			continue
+		}
+
+		if unicode.IsSpace(c) {
+			if current.Len() > 0 {
+				res = append(res, current.String())
+				current.Reset()
+			}
+			continue
+		}
+		current.WriteRune(c)
+	}
+	if backslash || dollar {
+		return nil, errors.New("unclosed escape sequence at end of depfile")
+	}
+	if current.Len() > 0 {
+		res = append(res, current.String())
+	}
+	return res, nil
+}

internal/arduino/builder/cpp/depfile_test.go

@@ -0,0 +1,100 @@
+// This file is part of arduino-cli.
+//
+// Copyright 2025 ARDUINO SA (http://www.arduino.cc/)
+//
+// This software is released under the GNU General Public License version 3,
+// which covers the main part of arduino-cli.
+// The terms of this license can be found at:
+// https://www.gnu.org/licenses/gpl-3.0.en.html
+//
+// You can be released from the requirements of the above licenses by purchasing
+// a commercial license. Buying such a license is mandatory if you want to
+// modify or otherwise use the software for commercial activities involving the
+// Arduino software without disclosing the source code of your own applications.
+// To purchase a commercial license, send an email to [email protected].
+
+package cpp
+
+import (
+	"testing"
+
+	"github.com/arduino/go-paths-helper"
+	"github.com/stretchr/testify/require"
+)
+
+func TestDepFileReader(t *testing.T) {
+	t.Run("0", func(t *testing.T) {
+		deps, err := ReadDepFile(paths.New("testdata", "depcheck.0.d"))
+		require.NoError(t, err)
+		require.NotNil(t, deps)
+		require.Len(t, deps.Dependencies, 302)
+		require.Equal(t, "sketch.ino.cpp.o", deps.ObjectFile)
+		require.Equal(t, "/home/megabug/Arduino/sketch/build/sketch/sketch.ino.cpp.merged", deps.Dependencies[0])
+		require.Equal(t, "/home/megabug/.arduino15/packages/arduino/hardware/zephyr/0.10.0-rc.10/variants/b_u585i_iot02a_stm32u585xx/llext-edk/include/zephyr/include/generated/zephyr/autoconf.h", deps.Dependencies[1])
+		require.Equal(t, "/home/megabug/.arduino15/packages/arduino/hardware/zephyr/0.10.0-rc.10/variants/b_u585i_iot02a_stm32u585xx/llext-edk/include/zephyr/include/zephyr/toolchain/zephyr_stdint.h", deps.Dependencies[2])
+		require.Equal(t, "/home/megabug/.arduino15/packages/arduino/hardware/zephyr/0.10.0-rc.10/libraries/Arduino_RPCLite/src/dispatcher.h", deps.Dependencies[301])
+	})
+	t.Run("1", func(t *testing.T) {
+		deps, err := ReadDepFile(paths.New("testdata", "depcheck.1.d"))
+		require.NoError(t, err)
+		require.NotNil(t, deps)
+		require.Equal(t, "sketch.ino.o", deps.ObjectFile)
+		require.Len(t, deps.Dependencies, 302)
+		require.Equal(t, "/home/megabug/Arduino/sketch/build/sketch/sketch.ino.cpp", deps.Dependencies[0])
+		require.Equal(t, "/home/megabug/.arduino15/packages/arduino/hardware/zephyr/0.10.0-rc.10/variants/b_u585i_iot02a_stm32u585xx/llext-edk/include/zephyr/include/generated/zephyr/autoconf.h", deps.Dependencies[1])
+		require.Equal(t, "/home/megabug/.arduino15/packages/arduino/hardware/zephyr/0.10.0-rc.10/variants/b_u585i_iot02a_stm32u585xx/llext-edk/include/zephyr/include/zephyr/toolchain/zephyr_stdint.h", deps.Dependencies[2])
+		require.Equal(t, "/home/megabug/.arduino15/packages/arduino/hardware/zephyr/0.10.0-rc.10/libraries/Arduino_RPCLite/src/dispatcher.h", deps.Dependencies[301])
+	})
+	t.Run("2", func(t *testing.T) {
+		deps, err := ReadDepFile(paths.New("testdata", "depcheck.2.d"))
+		require.NoError(t, err)
+		require.NotNil(t, deps)
+		require.Equal(t, "ske tch.ino.cpp.o", deps.ObjectFile)
+		require.Len(t, deps.Dependencies, 302)
+		require.Equal(t, "/home/megabug/Arduino/ske tch/build/sketch/ske tch.ino.cpp.merged", deps.Dependencies[0])
+		require.Equal(t, "/home/megabug/.arduino15/packages/arduino/hardware/zephyr/0.10.0-rc.10/variants/b_u585i_iot02a_stm32u585xx/llext-edk/include/zephyr/include/generated/zephyr/autoconf.h", deps.Dependencies[1])
+		require.Equal(t, "/home/megabug/.arduino15/packages/arduino/hardware/zephyr/0.10.0-rc.10/variants/b_u585i_iot02a_stm32u585xx/llext-edk/include/zephyr/include/zephyr/toolchain/zephyr_stdint.h", deps.Dependencies[2])
+		require.Equal(t, "/home/megabug/.arduino15/packages/arduino/hardware/zephyr/0.10.0-rc.10/libraries/Arduino_RPCLite/src/dispatcher.h", deps.Dependencies[301])
+	})
+	t.Run("3", func(t *testing.T) {
+		deps, err := ReadDepFile(paths.New("testdata", "depcheck.3.d"))
+		require.NoError(t, err)
+		require.NotNil(t, deps)
+		require.Equal(t, "myfile.o", deps.ObjectFile)
+		require.Len(t, deps.Dependencies, 3)
+		require.Equal(t, "/some/path\twith/tabs and spaces/file.cpp", deps.Dependencies[0])
+		require.Equal(t, "/some/other$/path#/file.h", deps.Dependencies[1])
+		require.Equal(t, "/yet/ano\\ther/path/file.h", deps.Dependencies[2])
+	})
+	t.Run("4", func(t *testing.T) {
+		deps, err := ReadDepFile(paths.New("testdata", "depcheck.4.d"))
+		require.EqualError(t, err, "invalid dollar sequence: $a")
+		require.Nil(t, deps)
+	})
+	t.Run("5", func(t *testing.T) {
+		deps, err := ReadDepFile(paths.New("testdata", "depcheck.5.d"))
+		require.EqualError(t, err, "invalid escape sequence: \\h")
+		require.Nil(t, deps)
+	})
+	t.Run("6", func(t *testing.T) {
+		deps, err := ReadDepFile(paths.New("testdata", "depcheck.6.d"))
+		require.EqualError(t, err, "unclosed escape sequence at end of depfile")
+		require.Nil(t, deps)
+	})
+	t.Run("7", func(t *testing.T) {
+		deps, err := ReadDepFile(paths.New("testdata", "depcheck.7.d"))
+		require.EqualError(t, err, "no colon in first item of depfile")
+		require.Nil(t, deps)
+	})
+	t.Run("8", func(t *testing.T) {
+		deps, err := ReadDepFile(paths.New("testdata", "depcheck.8.d"))
+		require.NoError(t, err)
+		require.Nil(t, deps.Dependencies)
+		require.Empty(t, deps.ObjectFile)
+	})
+	t.Run("9", func(t *testing.T) {
+		deps, err := ReadDepFile(paths.New("testdata", "nonexistent.d"))
+		require.Error(t, err)
+		require.Nil(t, deps)
+	})
+}