Add new function autoar_extract_start to do the extraction

This function is not tested and does not write files to disks currently.

2 files changed, 230 insertions, 2 deletions

diff --git a/autoarchive/autoar-extract.c b/autoarchive/autoar-extract.c
index ecaf20d8a..1dce79008 100644
--- a/autoarchive/autoar-extract.c
+++ b/autoarchive/autoar-extract.c
@@ -27,10 +27,10 @@
 
 #include "autoar-extract.h"
 
-#include <gio/gio.h>
-#include <string.h>
 #include <archive.h>
 #include <archive_entry.h>
+#include <gio/gio.h>
+#include <string.h>
 
 G_DEFINE_TYPE (AutoarExtract, autoar_extract, G_TYPE_OBJECT)
 
@@ -331,6 +331,49 @@ libarchive_read_read_cb (struct archive *ar_read,
   return read_size;
 }
 
+static char*
+_g_filename_basename_remove_extension (const char *filename)
+{
+  char *dot_location;
+  char *basename;
+
+  if (filename == NULL) {
+    return NULL;
+  }
+
+  /* filename must not be directory, so we do not get a bad basename. */
+  basename = g_path_get_basename (filename);
+
+  dot_location = strrchr (basename, '.');
+  if (dot_location == NULL || dot_location == basename) {
+    return basename;
+  }
+
+  if (dot_location - 4 > basename) {
+    if (strcmp (dot_location - 4, ".tar") == 0) {
+      dot_location -= 4;
+    }
+  }
+
+  *dot_location = '\0';
+
+  return basename;
+}
+
+static gboolean
+archive_extract_do_pattern_check (const char *pathname)
+{
+  return TRUE;
+}
+
+static void
+archive_extract_do_write_entry (AutoarExtract *arextract,
+                                struct archive *a,
+                                struct archive_entry *entry,
+                                GFile *dest)
+{
+}
+
 /* Additional marshaller generated by glib-genmarshal
  * Command: echo "VOID:DOUBLE,DOUBLE" | glib-genmarshal --header */
 
@@ -574,3 +617,186 @@ autoar_extract_new (const char *source,
                        output,
                        NULL);
 }
+
+void
+autoar_extract_start (AutoarExtract* arextract)
+{
+  struct archive *a;
+  struct archive_entry *entry;
+
+  char *pathname_prefix;
+  int pathname_prefix_len;
+
+  gboolean has_top_level_dir;
+  char *top_level_dir_basename;
+  char *top_level_dir_basename_modified;
+  GFile *top_level_parent_dir;
+  GFile *top_level_dir;
+
+  GFile *source;
+
+  int i;
+
+  g_return_if_fail (AUTOAR_IS_EXTRACT (arextract));
+  g_return_if_fail (arextract->priv->source != NULL);
+  g_return_if_fail (arextract->priv->output != NULL);
+
+  a = archive_read_new ();
+  archive_read_support_filter_all (a);
+  archive_read_support_format_all (a);
+
+  /* Reset all counter variables */
+  arextract->priv->size = 0;
+  arextract->priv->completed_size = 0;
+  arextract->priv->files = 0;
+  arextract->priv->completed_files = 0;
+
+  /* Step 1: Scan all file names in the archive
+   * We have to check whether the archive contains a top-level directory
+   * before performing the extraction. We emit the "scanned" signal when
+   * the checking is completed. */
+  archive_read_open (a,
+                     arextract,
+                     libarchive_read_open_cb,
+                     libarchive_read_read_cb,
+                     libarchive_read_close_cb);
+  pathname_prefix = NULL;
+  has_top_level_dir = TRUE;
+  while (archive_read_next_header (a, &entry) == ARCHIVE_OK) {
+    const char *pathname, *dir_sep_location;
+    size_t skip_len, prefix_len;
+
+    pathname = archive_entry_pathname (entry);
+
+    /* TODO: Add file name pattern check here */
+
+    if (pathname_prefix == NULL) {
+      skip_len = strspn (pathname, "./");
+      dir_sep_location = strchr (pathname + skip_len, '/');
+      if (dir_sep_location == NULL) {
+        prefix_len = strlen (pathname);
+      } else {
+        prefix_len = dir_sep_location - pathname;
+      }
+      pathname_prefix = g_strndup (pathname, prefix_len);
+      pathname_prefix_len = prefix_len;
+    } else {
+      if (!g_str_has_prefix (pathname, pathname_prefix)) {
+        has_top_level_dir = FALSE;
+      }
+    }
+    archive_read_data_skip (a);
+    arextract->priv->files++;
+  }
+  g_free (pathname_prefix);
+  archive_read_close (a);
+  if (arextract->priv->error != NULL) {
+    g_signal_emit_by_name (arextract, "error", arextract->priv->error);
+    archive_read_free (a);
+    return;
+  }
+  g_signal_emit_by_name (arextract, "scanned");
+
+  /* Step 2: Create necessary directories */
+  top_level_dir_basename = _g_filename_basename_remove_extension (arextract->priv->source);
+  top_level_parent_dir = g_file_new_for_commandline_arg (arextract->priv->output);
+  top_level_dir = g_file_get_child (top_level_parent_dir, top_level_dir_basename);
+
+  top_level_dir_basename_modified = NULL;
+  for (i=1; g_file_query_exists (top_level_dir, NULL); i++) {
+    g_free (top_level_dir_basename_modified);
+    g_object_unref (top_level_dir);
+    top_level_dir_basename_modified = g_strdup_printf ("%s (%d)",
+                                                       top_level_dir_basename,
+                                                       i);
+    top_level_dir = g_file_get_child (top_level_parent_dir,
+                                      top_level_dir_basename_modified);
+  }
+
+  g_file_make_directory_with_parents (top_level_dir, NULL, &(arextract->priv->error));
+
+  g_free (top_level_dir_basename);
+  g_free (top_level_dir_basename_modified);
+  g_object_unref (top_level_parent_dir);
+
+  if (arextract->priv->error != NULL) {
+    g_signal_emit_by_name (arextract, "error", arextract->priv->error);
+    g_object_unref (top_level_dir);
+    archive_read_free (a);
+    return;
+  }
+
+  /* Step 3: Extract files
+   * We have to re-open the archive to extract files */
+  archive_read_open (a,
+                     arextract,
+                     libarchive_read_open_cb,
+                     libarchive_read_read_cb,
+                     libarchive_read_close_cb);
+  while (archive_read_next_header (a, &entry) == ARCHIVE_OK) {
+    const char *pathname;
+    const char *pathname_skip_prefix;
+    char **pathname_chunks;
+
+    GFile *extracted_filename;
+    GFileInfo *restored_fileinfo;
+
+    pathname = archive_entry_pathname (entry);
+    if (has_top_level_dir)
+      pathname_skip_prefix = pathname + pathname_prefix_len;
+    else
+      pathname_skip_prefix = pathname + strspn (pathname, "./");
+
+    extracted_filename = g_file_get_child (top_level_dir, pathname_skip_prefix);
+
+    /* Extracted file should not be located outside the top level directory. */
+    if (!g_file_has_prefix (extracted_filename, top_level_dir)) {
+      pathname_chunks = g_strsplit (pathname_skip_prefix, "/", G_MAXINT);
+      for (i = 0; pathname_chunks[i] != NULL; i++) {
+        if (strcmp (pathname_chunks[i], "..") == 0) {
+          char *pathname_sanitized;
+
+          *pathname_chunks[i] = '\0';
+          pathname_sanitized = g_strjoinv ("/", pathname_chunks);
+
+          g_object_unref (extracted_filename);
+          extracted_filename = g_file_get_child (top_level_dir, pathname_sanitized);
+
+          g_free (pathname_sanitized);
+
+          if (g_file_has_prefix (extracted_filename, top_level_dir))
+            break;
+        }
+      }
+      g_strfreev (pathname_chunks);
+    }
+
+    /* TODO: Add file name pattern check here */
+    archive_extract_do_pattern_check (pathname_skip_prefix);
+
+    /* TODO: Write entry to disk */
+    archive_extract_do_write_entry (arextract, a, entry, extracted_filename);
+
+    arextract->priv->completed_files++;
+    g_signal_emit_by_name (arextract,
+                           "progress",
+                           ((double)(arextract->priv->size)) / ((double)(arextract->priv->completed_size)),
+                           ((double)(arextract->priv->files)) / ((double)(arextract->priv->completed_files)));
+    g_object_unref (extracted_filename);
+  }
+
+  g_object_unref (top_level_dir);
+  archive_read_close (a);
+  archive_read_free (a);
+  if (arextract->priv->error != NULL) {
+    g_signal_emit_by_name (arextract, "error", arextract->priv->error);
+    return;
+  }
+
+  /* If the extraction is completed successfully, remove the source file.
+   * Errors are not fatal because we have completed our work. */
+  source = g_file_new_for_commandline_arg (arextract->priv->source);
+  g_file_delete (source, NULL, NULL);
+  g_object_unref (source);
+  g_signal_emit_by_name (arextract, "completed");
+}
diff --git a/autoarchive/autoar-extract.h b/autoarchive/autoar-extract.h
index 2d93228f2..66bdbd367 100644
--- a/autoarchive/autoar-extract.h
+++ b/autoarchive/autoar-extract.h
@@ -67,6 +67,8 @@ GType           autoar_extract_get_type            (void) G_GNUC_CONST;
 AutoarExtract  *autoar_extract_new                 (const char *source,
                                                     const char *output);
 
+void            autoar_extract_start               (AutoarExtract* arextract);
+
 char           *autoar_extract_get_source          (AutoarExtract *arextract);
 char           *autoar_extract_get_output          (AutoarExtract *arextract);
 guint64         autoar_extract_get_size            (AutoarExtract *arextract);