From 4e3926d43b956ff69c8e6ef69ab43b398ee47730 Mon Sep 17 00:00:00 2001
From: Amy Boyd <amy@amyboyd.co.uk>
Date: Wed, 22 Jun 2016 23:09:59 +0100
Subject: [PATCH] Add source file finder.

---
 editorconfig/source_file_finder.go      | 102 ++++++++++++++++++++++++
 editorconfig/source_file_finder_test.go |  41 ++++++++++
 2 files changed, 143 insertions(+)
 create mode 100644 editorconfig/source_file_finder.go
 create mode 100644 editorconfig/source_file_finder_test.go

diff --git a/editorconfig/source_file_finder.go b/editorconfig/source_file_finder.go
new file mode 100644
index 0000000..090d192
--- /dev/null
+++ b/editorconfig/source_file_finder.go
@@ -0,0 +1,102 @@
+package editorconfig
+
+import (
+	"bytes"
+	"github.com/codegangsta/cli"
+	"os"
+	"path/filepath"
+	"regexp"
+	"strings"
+)
+
+var containsDotDirectoryRegex = regexp.MustCompile(`(^|/)\.[^/]+/`)
+
+var fileExtensions = GetSourceFileExtensions()
+
+var fileExtensionsRegex = CreateSourceFileExtensionRegex(fileExtensions)
+
+func GetSourceFileExtensions() []string {
+	var file, err = Asset("data/file-extensions.txt")
+	if err != nil {
+		ExitBecauseOfInternalError("Could not open list of file extensions")
+	}
+
+	reader := bytes.NewBuffer(file)
+	exts := []string{}
+
+	for {
+		ext, _ := reader.ReadBytes(' ')
+		if len(ext) == 0 {
+			break
+		}
+		ext = ext[0 : len(ext)-1]
+		exts = append(exts, string(ext))
+
+		// Skip the rest of the line (a description of what the file extension means).
+		reader.ReadBytes('\n')
+	}
+
+	if len(exts) == 0 {
+		ExitBecauseOfInternalError("No file extensions found")
+	}
+
+	return exts
+}
+
+func CreateSourceFileExtensionRegex(fileExtensions []string) *regexp.Regexp {
+	for i, e := range fileExtensions {
+		fileExtensions[i] = regexp.QuoteMeta(e)
+	}
+
+	r, err := regexp.Compile("\\.(" + strings.Join(fileExtensions, "|") + ")$")
+	if err != nil {
+		ExitBecauseOfInternalError("Could not compile regex: " + err.Error())
+	}
+
+	return r
+}
+
+func FindSourceFiles(searchPaths []string) ([]string, error) {
+	files := []string{}
+
+	for _, searchPath := range searchPaths {
+		if searchPath == "/" {
+			errMessage := "The path / was given.\n" +
+				"Exiting because this could be a mistake," +
+				" e.g. an environment variable infront of / might be unintentionally empty.\n" +
+				"Here is an example of why we won't run on /:\n" +
+				"http://www.theregister.co.uk/2015/01/17/scary_code_of_the_week_steam_cleans_linux_pcs/"
+			return nil, cli.NewExitError(errMessage, 2)
+		}
+
+		_ = filepath.Walk(searchPath, func(path string, fileInfo os.FileInfo, err error) error {
+			// Don't add paths that don't exist.
+			if err != nil {
+				return nil
+			}
+
+			// Only add files, not directories.
+			if fileInfo.IsDir() {
+				return nil
+			}
+
+			// Don't run on files inside dot directories. These directories are likely to
+			// be source control directories like ".git" or ".svn", or temporary directories
+			// like ".vagrant" or ".idea".
+			if containsDotDirectoryRegex.MatchString(path) {
+				return nil
+			}
+
+			// Only keep files with extensions that we know are textual.
+			if !fileExtensionsRegex.MatchString(path) {
+				return nil
+			}
+
+			files = append(files, path)
+
+			return nil
+		})
+	}
+
+	return files, nil
+}
diff --git a/editorconfig/source_file_finder_test.go b/editorconfig/source_file_finder_test.go
new file mode 100644
index 0000000..ba725ed
--- /dev/null
+++ b/editorconfig/source_file_finder_test.go
@@ -0,0 +1,41 @@
+package editorconfig
+
+import (
+	"strconv"
+	"testing"
+)
+
+func TestGetSourceFileExtensions(t *testing.T) {
+	result := GetSourceFileExtensions()
+
+	ExpectExtension := func(ext string) {
+		if !ContainsString(result, ext) {
+			t.Error("Result does not contain extension '" + ext + "'")
+		}
+	}
+
+	ExpectExtension("go")
+	ExpectExtension("java")
+	ExpectExtension("php")
+}
+
+func TestFindSourceFiles(t *testing.T) {
+	result, _ := FindSourceFiles([]string{"tests/a/b/c"})
+
+	if len(result) != 6 {
+		t.Error("Result should have 6 files, but has " + strconv.Itoa(len(result)))
+	}
+
+	ExpectPath := func(path string) {
+		if !ContainsString(result, path) {
+			t.Error("Result does not contain path '" + path + "'")
+		}
+	}
+
+	ExpectPath("tests/a/b/c/.editorconfig")
+	ExpectPath("tests/a/b/c/d/file.go")
+	ExpectPath("tests/a/b/c/d/file.java")
+	ExpectPath("tests/a/b/c/d/file.php")
+	ExpectPath("tests/a/b/c/d/keep-trailing-spaces.txt")
+	ExpectPath("tests/a/b/c/file.java")
+}