Implement standard library glob function (#511)

* Reinstate command to remove /tmp directory.

* Add glob functions to standard library.

* Whitespace changes.

* Reinstate removed blank lines.

src/std/fs.ab

@@ -1,3 +1,5 @@
+import { join, len, replace_regex, split } from "std/text"
+
 /// Checks if a directory exists.
 pub fun dir_exist(path) {
     $[ -d "{path}" ]$ failed {
@@ -76,3 +78,33 @@ pub fun change_owner(user: Text, path: Text): Bool {
 
     return false
 }
+
+/// Escapes all characters in the passed-in glob except "*", "?" and "/",
+/// to prevent injection attacks.
+fun escape_non_glob_chars(path: Text): Text {
+    return replace_regex(path, "\([^*?/]\)", "\\\\\1")
+}
+
+/// Finds all files or directories matching multiple file globs. When
+/// we have union types, this functionality can be merged into the main
+/// `glob` function.
+pub fun glob_multiple(paths: [Text]): [Text]? {
+    let combined = ""
+    if len(paths) == 1 {
+        combined = escape_non_glob_chars(paths[0])
+    } else {
+        let items = [Text]
+        loop item in paths {
+            item = escape_non_glob_chars(item)
+            items += [item]
+        }
+        combined = join(items, " ")
+    }
+    let files = $eval "for file in {combined}; do [ -e \\\"\\\$file\\\" ] && echo \\\"\\\$file\\\"; done"$?
+    return split(files, "\n")
+}
+
+/// Finds all files or directories matching a file glob.
+pub fun glob(path: Text): [Text]? {
+    return glob_multiple([path])?
+}

src/tests/stdlib/create_symbolic_link.ab

@@ -7,5 +7,5 @@ main {
     } else {
         echo "failed"
     }
-    trust $rm {tmpdir}$
+    trust $rm -fr {tmpdir}$
 } 

src/tests/stdlib/glob_absolute_missing_file.ab

@@ -0,0 +1,26 @@
+import * from "std/fs"
+
+// Output
+// FAILED
+
+main {
+    let tmpdir = trust $ mktemp -d $
+    trust {
+        $ touch {tmpdir}/1st\ file\ with\ spaces.txt $
+        $ touch {tmpdir}/2nd\ file\ with\ spaces.txt $
+        $ touch {tmpdir}/file.txt $
+        $ touch {tmpdir}/file1.txt $
+        $ touch {tmpdir}/file2.txt $
+        $ touch {tmpdir}/file99.txt $
+        $ touch {tmpdir}/other.csv $
+    }
+
+    let files = glob("{tmpdir}/missing*") failed {
+        echo "FAILED"
+    }
+    loop file in files {
+        echo "[{file}]"
+    }
+
+    trust $ rm -rf {tmpdir} $
+}

src/tests/stdlib/glob_absolute_multiple_globs.ab

@@ -0,0 +1,47 @@
+import * from "std/array"
+import * from "std/fs"
+import * from "std/text"
+
+fun compare(actual: [Text], expected: [Text]): Bool {
+    if len(actual) != len(expected) {
+        return false
+    }
+    loop file in expected {
+        if not includes(actual, file) {
+            return false
+        }
+    }
+    return true
+}
+
+main {
+    let tmpdir = trust $ mktemp -d $
+    trust {
+        $ touch {tmpdir}/1st\ file\ with\ spaces.txt $
+        $ touch {tmpdir}/2nd\ file\ with\ spaces.txt $
+        $ touch {tmpdir}/file.txt $
+        $ touch {tmpdir}/file1.txt $
+        $ touch {tmpdir}/file2.txt $
+        $ touch {tmpdir}/file99.txt $
+        $ touch {tmpdir}/other.csv $
+    }
+
+    let expected = [
+        "{tmpdir}/file.txt",
+        "{tmpdir}/file1.txt",
+        "{tmpdir}/file2.txt",
+        "{tmpdir}/file99.txt",
+        "{tmpdir}/other.csv",
+    ]
+    let actual = glob_multiple(["{tmpdir}/missing*", "{tmpdir}/file*.txt", "{tmpdir}/other*.csv"]) failed {
+        echo "FAILED"
+    }
+    if compare(actual, expected) {
+        echo "Succeded"
+    } else {
+        echo "Expected: {expected}"
+        echo "Actual: {actual}"
+    }
+
+    trust $ rm -rf {tmpdir} $
+}

src/tests/stdlib/glob_absolute_wild_char.ab

@@ -0,0 +1,44 @@
+import * from "std/array"
+import * from "std/fs"
+import * from "std/text"
+
+fun compare(actual: [Text], expected: [Text]): Bool {
+    if len(actual) != len(expected) {
+        return false
+    }
+    loop file in expected {
+        if not includes(actual, file) {
+            return false
+        }
+    }
+    return true
+}
+
+main {
+    let tmpdir = trust $ mktemp -d $
+    trust {
+        $ touch {tmpdir}/1st\ file\ with\ spaces.txt $
+        $ touch {tmpdir}/2nd\ file\ with\ spaces.txt $
+        $ touch {tmpdir}/file.txt $
+        $ touch {tmpdir}/file1.txt $
+        $ touch {tmpdir}/file2.txt $
+        $ touch {tmpdir}/file99.txt $
+        $ touch {tmpdir}/other.csv $
+    }
+
+    let expected = [
+        "{tmpdir}/file1.txt",
+        "{tmpdir}/file2.txt",
+    ]
+    let actual = glob("{tmpdir}/file?.txt") failed {
+        echo "FAILED"
+    }
+    if compare(actual, expected) {
+        echo "Succeded"
+    } else {
+        echo "Expected: {expected}"
+        echo "Actual: {actual}"
+    }
+
+    trust $ rm -rf {tmpdir} $
+}

src/tests/stdlib/glob_absolute_wild_star.ab

@@ -0,0 +1,46 @@
+import * from "std/array"
+import * from "std/fs"
+import * from "std/text"
+
+fun compare(actual: [Text], expected: [Text]): Bool {
+    if len(actual) != len(expected) {
+        return false
+    }
+    loop file in expected {
+        if not includes(actual, file) {
+            return false
+        }
+    }
+    return true
+}
+
+main {
+    let tmpdir = trust $ mktemp -d $
+    trust {
+        $ touch {tmpdir}/1st\ file\ with\ spaces.txt $
+        $ touch {tmpdir}/2nd\ file\ with\ spaces.txt $
+        $ touch {tmpdir}/file.txt $
+        $ touch {tmpdir}/file1.txt $
+        $ touch {tmpdir}/file2.txt $
+        $ touch {tmpdir}/file99.txt $
+        $ touch {tmpdir}/other.csv $
+    }
+
+    let expected = [
+        "{tmpdir}/file.txt",
+        "{tmpdir}/file1.txt",
+        "{tmpdir}/file2.txt",
+        "{tmpdir}/file99.txt",
+    ]
+    let actual = glob("{tmpdir}/file*.txt") failed {
+        echo "FAILED"
+    }
+    if compare(actual, expected) {
+        echo "Succeded"
+    } else {
+        echo "Expected: {expected}"
+        echo "Actual: {actual}"
+    }
+
+    trust $ rm -rf {tmpdir} $
+}

src/tests/stdlib/glob_absolute_with_spaces.ab

@@ -0,0 +1,44 @@
+import * from "std/array"
+import * from "std/fs"
+import * from "std/text"
+
+fun compare(actual: [Text], expected: [Text]): Bool {
+    if len(actual) != len(expected) {
+        return false
+    }
+    loop file in expected {
+        if not includes(actual, file) {
+            return false
+        }
+    }
+    return true
+}
+
+main {
+    let tmpdir = trust $ mktemp -d $
+    trust {
+        $ touch {tmpdir}/1st\ file\ with\ spaces.txt $
+        $ touch {tmpdir}/2nd\ file\ with\ spaces.txt $
+        $ touch {tmpdir}/file.txt $
+        $ touch {tmpdir}/file1.txt $
+        $ touch {tmpdir}/file2.txt $
+        $ touch {tmpdir}/file99.txt $
+        $ touch {tmpdir}/other.csv $
+    }
+
+    let expected = [
+        "{tmpdir}/1st file with spaces.txt",
+        "{tmpdir}/2nd file with spaces.txt",
+    ]
+    let actual = glob("{tmpdir}/*with spaces*") failed {
+        echo "FAILED"
+    }
+    if compare(actual, expected) {
+        echo "Succeded"
+    } else {
+        echo "Expected: {expected}"
+        echo "Actual: {actual}"
+    }
+
+    trust $ rm -rf {tmpdir} $
+}

src/tests/stdlib/glob_injection_attack.ab

@@ -0,0 +1,25 @@
+import * from "std/fs"
+
+// Output
+// [xxx; do echo HACKED; done; for file in]
+
+main {
+    let tmpdir = trust $ mktemp -d $
+    trust {
+        $ touch "{tmpdir}/xxx; do echo HACKED; done; for file in" $
+    }
+    cd tmpdir
+
+    // The glob function escapes all characters in the passed-in glob
+    // apart from "*", "?" and "/", to prevent injection attacks.  If we
+    // didn't do this, the following code would output "[HACKED]" instead
+    // of the filename.
+    let files = glob("xxx; do echo HACKED; done; for file in") failed {
+        echo "FAILED"
+    }
+    loop file in files {
+        echo "[{file}]"
+    }
+
+    trust $ rm -rf {tmpdir} $
+}

src/tests/stdlib/glob_relative_missing_file.ab

@@ -0,0 +1,27 @@
+import * from "std/fs"
+
+// Output
+// FAILED
+
+main {
+    let tmpdir = trust $ mktemp -d $
+    trust {
+        $ touch {tmpdir}/1st\ file\ with\ spaces.txt $
+        $ touch {tmpdir}/2nd\ file\ with\ spaces.txt $
+        $ touch {tmpdir}/file.txt $
+        $ touch {tmpdir}/file1.txt $
+        $ touch {tmpdir}/file2.txt $
+        $ touch {tmpdir}/file99.txt $
+        $ touch {tmpdir}/other.csv $
+    }
+    cd tmpdir
+
+    let files = glob("missing*") failed {
+        echo "FAILED"
+    }
+    loop file in files {
+        echo "[{file}]"
+    }
+
+    trust $ rm -rf {tmpdir} $
+}

src/tests/stdlib/glob_relative_multiple_globs.ab

@@ -0,0 +1,48 @@
+import * from "std/array"
+import * from "std/fs"
+import * from "std/text"
+
+fun compare(actual: [Text], expected: [Text]): Bool {
+    if len(actual) != len(expected) {
+        return false
+    }
+    loop file in expected {
+        if not includes(actual, file) {
+            return false
+        }
+    }
+    return true
+}
+
+main {
+    let tmpdir = trust $ mktemp -d $
+    trust {
+        $ touch {tmpdir}/1st\ file\ with\ spaces.txt $
+        $ touch {tmpdir}/2nd\ file\ with\ spaces.txt $
+        $ touch {tmpdir}/file.txt $
+        $ touch {tmpdir}/file1.txt $
+        $ touch {tmpdir}/file2.txt $
+        $ touch {tmpdir}/file99.txt $
+        $ touch {tmpdir}/other.csv $
+    }
+    cd tmpdir
+
+    let expected = [
+        "file.txt",
+        "file1.txt",
+        "file2.txt",
+        "file99.txt",
+        "other.csv",
+    ]
+    let actual = glob_multiple(["missing*", "file*.txt", "other*.csv"]) failed {
+        echo "FAILED"
+    }
+    if compare(actual, expected) {
+        echo "Succeded"
+    } else {
+        echo "Expected: {expected}"
+        echo "Actual: {actual}"
+    }
+
+    trust $ rm -rf {tmpdir} $
+}