Add support for patterns on %include #25140

Also adds generic tor_glob function to expand globs.

1 files changed, 296 insertions, 0 deletions

diff --git a/src/lib/fs/path.c b/src/lib/fs/path.c
index 0d57be4b06..4532bfb7d1 100644
--- a/src/lib/fs/path.c
+++ b/src/lib/fs/path.c
@@ -13,18 +13,44 @@
 #include "lib/malloc/malloc.h"
 #include "lib/log/log.h"
 #include "lib/log/util_bug.h"
+#include "lib/container/smartlist.h"
+#include "lib/sandbox/sandbox.h"
 #include "lib/string/printf.h"
 #include "lib/string/util_string.h"
 #include "lib/string/compat_ctype.h"
+#include "lib/string/compat_string.h"
+#include "lib/fs/files.h"
+#include "lib/fs/dir.h"
 #include "lib/fs/userdb.h"
 
+#ifdef HAVE_SYS_TYPES_H
+#include <sys/types.h>
+#endif
+#ifdef HAVE_SYS_STAT_H
+#include <sys/stat.h>
+#endif
 #ifdef HAVE_UNISTD_H
 #include <unistd.h>
 #endif
 
+#ifdef _WIN32
+#include <windows.h>
+#include <shlwapi.h>
+#else /* !(defined(_WIN32)) */
+#include <dirent.h>
+#include <glob.h>
+#endif /* defined(_WIN32) */
+
 #include <errno.h>
 #include <string.h>
 
+#ifdef _WIN32
+#define IS_GLOB_CHAR(s,i) (((s)[(i)]) == '*' || ((s)[(i)]) == '?')
+#else
+#define IS_GLOB_CHAR(s,i) ((((s)[(i)]) == '*' || ((s)[(i)]) == '?') &&\
+                           ((i) == 0 || (s)[(i)-1] != '\\')) /* check escape */
+#endif
+
 /** Removes enclosing quotes from <b>path</b> and unescapes quotes between the
  * enclosing quotes. Backslashes are not unescaped. Return the unquoted
  * <b>path</b> on success or 0 if <b>path</b> is not quoted correctly. */
@@ -294,3 +320,273 @@ make_path_absolute(const char *fname)
   return absfname;
 #endif /* defined(_WIN32) */
 }
+
+/** Expands globs in <b>pattern</b> for the path fragment between
+ * <b>prev_sep</b> and <b>next_sep</b>. Returns NULL on failure. */
+typedef struct smartlist_t * unglob_fn(const char *pattern, int prev_sep,
+                                       int next_sep);
+
+/** Adds <b>path</b> to <b>result</b> if it exists and is a file type we can
+ * handle. Returns false if <b>path</b> is a file type we cannot handle,
+ * returns true otherwise. Used on tor_glob for WIN32. */
+static bool
+add_non_glob_path(const char *path, struct smartlist_t *result)
+{
+  file_status_t file_type = file_status(path);
+  if (file_type == FN_ERROR) {
+    return false;
+  } else if (file_type != FN_NOENT) {
+    char *to_add = tor_strdup(path);
+    clean_fname_for_stat(to_add);
+    smartlist_add(result, to_add);
+  }
+  /* If WIN32 tor_glob is called with a non-existing path, we want it to
+   * return an empty list instead of error to match the regular version */
+  return true;
+}
+
+/** Auxiliary function used by get_glob_opened_files and WIN32 tor_glob.
+ * Returns a list of paths obtained from <b>pattern</b> using <b>unglob</b> to
+ * expand each path fragment. If <b>final</b> is true, the paths are the result
+ * of the glob expansion of <b>pattern</b> (implements tor_glob). Otherwise,
+ * the paths are the paths opened by glob while expanding <b>pattern</b>
+ * (implements get_glb_opened_files). Returns NULL on failure. */
+static struct smartlist_t *
+get_glob_paths(const char *pattern, unglob_fn unglob, bool final)
+{
+  smartlist_t *result = smartlist_new();
+  int i, prev_sep = -1, next_sep = -1;
+  bool is_glob = false, error_found = false, is_sep = false, is_last = false;
+
+  // find first path fragment with globs
+  for (i = 0; pattern[i]; i++) {
+    is_glob = is_glob || IS_GLOB_CHAR(pattern, i);
+    is_last = !pattern[i+1];
+    is_sep = pattern[i] == *PATH_SEPARATOR || pattern[i] == '/';
+    if (is_sep || is_last) {
+      prev_sep = next_sep;
+      next_sep = i; // next_sep+1 is start of next fragment or end of string
+      if (is_glob) {
+        break;
+      }
+    }
+  }
+
+  if (!is_glob) { // pattern fully expanded or no glob in pattern
+    if (final && !add_non_glob_path(pattern, result)) {
+      error_found = true;
+      goto end;
+    }
+    return result;
+  }
+
+  if (!final) {
+    // add path before the glob to result
+    int len = prev_sep < 1 ? prev_sep + 1 : prev_sep; // handle /*
+    char *path_until_glob = tor_strndup(pattern, len);
+    smartlist_add(result, path_until_glob);
+  }
+
+  smartlist_t *unglobbed_paths = unglob(pattern, prev_sep, next_sep);
+  if (!unglobbed_paths) {
+    error_found = true;
+  } else {
+    // for each path for current fragment, add the rest of the pattern
+    // and call recursively to get all expanded paths
+    SMARTLIST_FOREACH_BEGIN(unglobbed_paths, char *, current_path) {
+      char *next_path;
+      tor_asprintf(&next_path, "%s"PATH_SEPARATOR"%s", current_path,
+                   &pattern[next_sep+1]);
+      smartlist_t *opened_next = get_glob_paths(next_path, unglob, final);
+      tor_free(next_path);
+      if (!opened_next) {
+        error_found = true;
+        break;
+      }
+      smartlist_add_all(result, opened_next);
+      smartlist_free(opened_next);
+    } SMARTLIST_FOREACH_END(current_path);
+    SMARTLIST_FOREACH(unglobbed_paths, char *, p, tor_free(p));
+    smartlist_free(unglobbed_paths);
+  }
+
+end:
+  if (error_found) {
+    SMARTLIST_FOREACH(result, char *, p, tor_free(p));
+    smartlist_free(result);
+    result = NULL;
+  }
+  return result;
+}
+
+#ifdef _WIN32
+/** Expands globs in <b>pattern</b> for the path fragment between
+ * <b>prev_sep</b> and <b>next_sep</b> using the WIN32 API. Returns NULL on
+ * failure. Used by the WIN32 implementation of tor_glob. */
+static struct smartlist_t *
+unglob_win32(const char *pattern, int prev_sep, int next_sep)
+{
+  smartlist_t *result = smartlist_new();
+  int len = prev_sep < 1 ? prev_sep + 1 : prev_sep; // handle /*
+  char *path_until_glob = tor_strndup(pattern, len);
+
+  if (!is_file(file_status(path_until_glob))) {
+    smartlist_t *filenames = tor_listdir(path_until_glob);
+    if (!filenames) {
+      smartlist_free(result);
+      result = NULL;
+    } else {
+      SMARTLIST_FOREACH_BEGIN(filenames, char *, filename) {
+        TCHAR tpattern[MAX_PATH] = {0};
+        TCHAR tfile[MAX_PATH] = {0};
+        char *full_path;
+        tor_asprintf(&full_path, "%s"PATH_SEPARATOR"%s",
+                     path_until_glob, filename);
+        char *path_curr_glob = tor_strndup(pattern, next_sep + 1);
+        // *\ must return only dirs, remove \ from the pattern so it matches
+        if (is_dir(file_status(full_path))) {
+          clean_fname_for_stat(path_curr_glob);
+        }
+#ifdef UNICODE
+        mbstowcs(tpattern, path_curr_glob, MAX_PATH);
+        mbstowcs(tfile, full_path, MAX_PATH);
+#else
+        strlcpy(tpattern, path_curr_glob, MAX_PATH);
+        strlcpy(tfile, full_path, MAX_PATH);
+#endif
+        if (PathMatchSpec(tfile, tpattern)) {
+          smartlist_add(result, full_path);
+        } else {
+          tor_free(full_path);
+        }
+        tor_free(path_curr_glob);
+      } SMARTLIST_FOREACH_END(filename);
+      SMARTLIST_FOREACH(filenames, char *, p, tor_free(p));
+      smartlist_free(filenames);
+    }
+  }
+  tor_free(path_until_glob);
+  return result;
+}
+#else /* !defined(_WIN32) */
+/** Same as opendir but calls sandbox_intern_string before */
+static DIR *
+prot_opendir(const char *name)
+{
+  return opendir(sandbox_intern_string(name));
+}
+
+/** Same as stat but calls sandbox_intern_string before */
+static int
+prot_stat(const char *pathname, struct stat *buf)
+{
+  return stat(sandbox_intern_string(pathname), buf);
+}
+
+/** Same as lstat but calls sandbox_intern_string before */
+static int
+prot_lstat(const char *pathname, struct stat *buf)
+{
+  return lstat(sandbox_intern_string(pathname), buf);
+}
+#endif /* defined(_WIN32) */
+
+/** Return a new list containing the paths that match the pattern
+ * <b>pattern</b>. Return NULL on error.
+ */
+struct smartlist_t *
+tor_glob(const char *pattern)
+{
+  smartlist_t *result;
+#ifdef _WIN32
+  // PathMatchSpec does not support forward slashes, change them to backslashes
+  char *pattern_normalized = tor_strdup(pattern);
+  tor_strreplacechar(pattern_normalized, '/', *PATH_SEPARATOR);
+  result = get_glob_paths(pattern_normalized, unglob_win32, true);
+  tor_free(pattern_normalized);
+#else /* !(defined(_WIN32)) */
+  glob_t matches;
+  int flags = GLOB_ERR | GLOB_NOSORT;
+#ifdef GLOB_ALTDIRFUNC
+  /* use functions that call sandbox_intern_string */
+  flags |= GLOB_ALTDIRFUNC;
+  typedef void *(*gl_opendir)(const char * name);
+  typedef struct dirent *(*gl_readdir)(void *);
+  typedef void (*gl_closedir)(void *);
+  matches.gl_opendir = (gl_opendir) &prot_opendir;
+  matches.gl_readdir = (gl_readdir) &readdir;
+  matches.gl_closedir = (gl_closedir) &closedir;
+  matches.gl_stat = &prot_stat;
+  matches.gl_lstat = &prot_lstat;
+#endif /* defined(GLOB_ALTDIRFUNC) */
+  int ret = glob(pattern, flags, NULL, &matches);
+  if (ret == GLOB_NOMATCH) {
+    return smartlist_new();
+  } else if (ret != 0) {
+    return NULL;
+  }
+
+  result = smartlist_new();
+  size_t i;
+  for (i = 0; i < matches.gl_pathc; i++) {
+    char *match = tor_strdup(matches.gl_pathv[i]);
+    size_t len = strlen(match);
+    if (len > 0 && match[len-1] == *PATH_SEPARATOR) {
+      match[len-1] = '\0';
+    }
+    smartlist_add(result, match);
+  }
+  globfree(&matches);
+#endif /* defined(_WIN32) */
+  return result;
+}
+
+/** Returns true if <b>s</b> contains characters that can be globbed.
+ * Returns false otherwise. */
+bool
+has_glob(const char *s)
+{
+  int i;
+  for (i = 0; s[i]; i++) {
+    if (IS_GLOB_CHAR(s, i)) {
+      return true;
+    }
+  }
+  return false;
+}
+
+/** Expands globs in <b>pattern</b> for the path fragment between
+ * <b>prev_sep</b> and <b>next_sep</b> using tor_glob. Returns NULL on
+ * failure. Used by get_glob_opened_files. */
+static struct smartlist_t *
+unglob_opened_files(const char *pattern, int prev_sep, int next_sep)
+{
+  (void)prev_sep;
+  smartlist_t *result = smartlist_new();
+  // if the following fragments have no globs, we're done
+  if (has_glob(&pattern[next_sep+1])) {
+    // if there is a glob after next_sep, we know it is a separator and not the
+    // last char and glob_path will have the path without the separator
+    char *glob_path = tor_strndup(pattern, next_sep);
+    smartlist_t *child_paths = tor_glob(glob_path);
+    tor_free(glob_path);
+    if (!child_paths) {
+      smartlist_free(result);
+      result = NULL;
+    } else {
+      smartlist_add_all(result, child_paths);
+      smartlist_free(child_paths);
+    }
+  }
+  return result;
+}
+
+/** Returns a list of files that are opened by the tor_glob function when
+ * called with <b>pattern</b>. Returns NULL on error. The purpose of this
+ * function is to create a list of files to be added to the sandbox white list
+ * before the sandbox is enabled. */
+struct smartlist_t *
+get_glob_opened_files(const char *pattern)
+{
+  return get_glob_paths(pattern, unglob_opened_files, false);
+}