Impl. link extracting

If absolute paths are enabled, and a link points to something outside of
the archive, a flag is set on the symlink entry to prefer absolute
paths.

file_format.md

@@ -55,6 +55,8 @@ Following the file-count bytes, the following bytes are added for each file:
     2. The second byte.
         1. The first bit is "other write permission".
         2. The second bit is "other execute permission".
+        3. The third bit is UNSET if relative links are preferred, and is SET
+           if absolute links are preferred.
     3. The third byte.
         1. Currently unused.
     4. The fourth byte.

src/archiver.c

@@ -37,7 +37,6 @@
 #include <unistd.h>
 #endif
 
-#include "data_structures/hash_map.h"
 #include "helpers.h"
 
 #define TEMP_FILENAME_CMP "simple_archiver_compressed_%u.tmp"
@@ -104,6 +103,43 @@ void cleanup_temp_filename_delete(void ***ptrs_array) {
 #endif
 }
 
+char *filename_to_absolute_path(const char *filename) {
+  __attribute__((cleanup(free_malloced_memory))) void *path =
+      malloc(strlen(filename) + 1);
+  strncpy(path, filename, strlen(filename) + 1);
+
+  char *path_dir = dirname(path);
+  if (!path_dir) {
+    return NULL;
+  }
+
+  __attribute__((cleanup(free_malloced_memory))) void *dir_realpath =
+      realpath(path_dir, NULL);
+  if (!dir_realpath) {
+    return NULL;
+  }
+
+  // Recreate "path" since it may have been modified by dirname().
+  free_malloced_memory(&path);
+  path = malloc(strlen(filename) + 1);
+  strncpy(path, filename, strlen(filename) + 1);
+
+  char *filename_basename = basename(path);
+  if (!filename_basename) {
+    return NULL;
+  }
+
+  // Get combined full path to file.
+  char *fullpath =
+      malloc(strlen(dir_realpath) + 1 + strlen(filename_basename) + 1);
+  strncpy(fullpath, dir_realpath, strlen(dir_realpath) + 1);
+  fullpath[strlen(dir_realpath)] = '/';
+  strncpy(fullpath + strlen(dir_realpath) + 1, filename_basename,
+          strlen(filename_basename) + 1);
+
+  return fullpath;
+}
+
 int write_files_fn(void *data, void *ud) {
   const SDArchiverFileInfo *file_info = data;
   SDArchiverState *state = ud;
@@ -668,89 +704,90 @@ int write_files_fn(void *data, void *ud) {
     // Unsupported platform. Just set the permission bits for user.
     ((unsigned char *)temp_to_write->buf)[0] |= 0xE;
 #endif
-    simple_archiver_list_add(to_write, temp_to_write, free_internal_to_write);
 
+    // Need to get abs_path for checking/setting a flag before storing flags.
     // Get absolute path.
     __attribute__((cleanup(free_malloced_memory))) void *abs_path =
         realpath(file_info->filename, NULL);
+    __attribute__((cleanup(free_malloced_memory))) void *rel_path = NULL;
+    if (abs_path) {
+      // Get relative path.
+      // First get absolute path of link.
+      __attribute__((cleanup(free_malloced_memory))) void *link_abs_path =
+          filename_to_absolute_path(file_info->filename);
+      if (!link_abs_path) {
+        fprintf(stderr, "WARNING: Failed to get absolute path of link!\n");
+      } else {
+        // fprintf(stderr, "DEBUG: abs_path: %s\nDEBUG: link_abs_path: %s\n",
+        //                 (char*)abs_path, (char*)link_abs_path);
+
+        // Compare paths to get relative path.
+        // Get first non-common char.
+        unsigned int idx;
+        unsigned int last_slash;
+        for (idx = 0, last_slash = 0;
+             idx < strlen(abs_path) && idx < strlen(link_abs_path); ++idx) {
+          if (((const char *)abs_path)[idx] !=
+              ((const char *)link_abs_path)[idx]) {
+            break;
+          } else if (((const char *)abs_path)[idx] == '/') {
+            last_slash = idx + 1;
+          }
+        }
+        // Get substrings of both paths.
+        char *link_substr = (char *)link_abs_path + last_slash;
+        char *dest_substr = (char *)abs_path + last_slash;
+        rel_path = malloc(strlen(dest_substr) + 1);
+        strncpy(rel_path, dest_substr, strlen(dest_substr) + 1);
+        // fprintf(stderr, "DEBUG: link_substr: %s\nDEBUG: dest_substr: %s\n",
+        //         link_substr, dest_substr);
+
+        // Generate the relative path.
+        int has_slash = 0;
+        idx = 0;
+        do {
+          for (; link_substr[idx] != '/' && link_substr[idx] != 0; ++idx);
+          if (link_substr[idx] == 0) {
+            has_slash = 0;
+          } else {
+            has_slash = 1;
+            char *new_rel_path = malloc(strlen(rel_path) + 1 + 3);
+            new_rel_path[0] = '.';
+            new_rel_path[1] = '.';
+            new_rel_path[2] = '/';
+            strncpy(new_rel_path + 3, rel_path, strlen(rel_path) + 1);
+            free(rel_path);
+            rel_path = new_rel_path;
+            ++idx;
+          }
+        } while (has_slash);
+      }
+    }
+
+    // Check if absolute path refers to one of the filenames.
+    if (abs_path && (state->parsed->flags & 0x20) == 0 &&
+        !simple_archiver_hash_map_get(state->map, abs_path,
+                                      strlen(abs_path) + 1)) {
+      // Is not a filename being archived, set preference to absolute path.
+      fprintf(stderr,
+              "NOTICE: abs_path exists, \"--no-abs-symlink\" not specified, "
+              "and link refers to file NOT in archive; preferring abs_path.\n");
+      ((unsigned char *)temp_to_write->buf)[1] |= 0x4;
+    }
+
+    // Store the 4 byte bit-flags for file.
+    simple_archiver_list_add(to_write, temp_to_write, free_internal_to_write);
+
+    // Store the absolute and relative paths.
     if (!abs_path) {
       fprintf(stderr,
               "WARNING: Failed to get absolute path of link destination!\n");
-    }
+      temp_to_write = malloc(sizeof(SDArchiverInternalToWrite));
-
+      temp_to_write->buf = malloc(2);
-    // Get relative path.
+      temp_to_write->size = 2;
-    __attribute__((cleanup(free_malloced_memory))) void *rel_path = NULL;
+      memset(temp_to_write->buf, 0, 2);
-    // First get absolute path of link.
+      simple_archiver_list_add(to_write, temp_to_write, free_internal_to_write);
-    // Get abs path to dirname of link.
+    } else if ((state->parsed->flags & 0x20) == 0) {
-    unsigned int link_dir_path_len = strlen(file_info->filename) + 1;
-    __attribute__((cleanup(free_malloced_memory))) void *link_dir_path =
-        malloc(link_dir_path_len);
-    strncpy(link_dir_path, file_info->filename, link_dir_path_len);
-    char *link_dirname = dirname(link_dir_path);
-    __attribute__((cleanup(free_malloced_memory))) void *link_dir_abs_path =
-        realpath(link_dirname, NULL);
-    if (!link_dir_abs_path) {
-      fprintf(stderr,
-              "WARNING: Failed to get absolute path of link directory!\n");
-    } else {
-      // Get basename of link to append.
-      __attribute__((cleanup(free_malloced_memory))) void *link_filename =
-          malloc(strlen(file_info->filename) + 1);
-      strncpy(link_filename, file_info->filename,
-              strlen(file_info->filename) + 1);
-      char *link_basename = basename(link_filename);
-      // Set up full path to link.
-      unsigned int link_path_len = strlen(link_dir_abs_path);
-      __attribute__((cleanup(free_malloced_memory))) void *combined_path =
-          malloc(link_path_len + 1 + strlen(link_basename) + 1);
-      strncpy(combined_path, link_dir_abs_path, link_path_len + 1);
-      ((char *)combined_path)[link_path_len] = '/';
-      strncpy((char *)combined_path + link_path_len + 1, link_basename,
-              strlen(link_basename) + 1);
-      // fprintf(stderr, "DEBUG: abs_path: %s\nDEBUG: combined_path: %s\n",
-      //                 (char*)abs_path, (char*)combined_path);
-      // Compare paths to get relative path.
-      // Get first non-common char.
-      unsigned int idx;
-      unsigned int last_slash;
-      for (idx = 0, last_slash = 0;
-           idx < strlen(abs_path) && idx < strlen(combined_path); ++idx) {
-        if (((const char *)abs_path)[idx] !=
-            ((const char *)combined_path)[idx]) {
-          break;
-        } else if (((const char *)abs_path)[idx] == '/') {
-          last_slash = idx + 1;
-        }
-      }
-      // Get substrings of both paths.
-      char *link_substr = (char *)combined_path + last_slash;
-      char *dest_substr = (char *)abs_path + last_slash;
-      rel_path = malloc(strlen(dest_substr) + 1);
-      strncpy(rel_path, dest_substr, strlen(dest_substr) + 1);
-      // fprintf(stderr, "DEBUG: link_substr: %s\nDEBUG: dest_substr: %s\n",
-      //         link_substr, dest_substr);
-      // Generate the relative path.
-      int has_slash = 0;
-      idx = 0;
-      do {
-        for (; link_substr[idx] != '/' && link_substr[idx] != 0; ++idx);
-        if (link_substr[idx] == 0) {
-          has_slash = 0;
-        } else {
-          has_slash = 1;
-          char *new_rel_path = malloc(strlen(rel_path) + 1 + 3);
-          new_rel_path[0] = '.';
-          new_rel_path[1] = '.';
-          new_rel_path[2] = '/';
-          strncpy(new_rel_path + 3, rel_path, strlen(rel_path) + 1);
-          free(rel_path);
-          rel_path = new_rel_path;
-          ++idx;
-        }
-      } while (has_slash);
-    }
-
-    if (abs_path) {
       // Write absolute path length.
       u16 = strlen(abs_path);
       simple_archiver_helper_16_bit_be(&u16);
@@ -769,6 +806,9 @@ int write_files_fn(void *data, void *ud) {
       strncpy(temp_to_write->buf, abs_path, u16 + 1);
       simple_archiver_list_add(to_write, temp_to_write, free_internal_to_write);
     } else {
+      fprintf(stderr,
+              "NOTICE: Not saving absolute path since \"--no-abs-symlink\" "
+              "was specified.\n");
       temp_to_write = malloc(sizeof(SDArchiverInternalToWrite));
       temp_to_write->buf = malloc(2);
       temp_to_write->size = 2;
@@ -803,7 +843,9 @@ int write_files_fn(void *data, void *ud) {
 
     // Write all previously set data.
     fprintf(stderr, "Writing symlink info: %s\n", file_info->filename);
-    fprintf(stderr, "  abs path: %s\n", (char *)abs_path);
+    if ((state->parsed->flags & 0x20) == 0) {
+      fprintf(stderr, "  abs path: %s\n", (char *)abs_path);
+    }
     fprintf(stderr, "  rel path: %s\n", (char *)rel_path);
     simple_archiver_list_get(to_write, write_list_datas_fn, state->out_f);
     simple_archiver_list_free(&to_write);
@@ -816,6 +858,54 @@ int write_files_fn(void *data, void *ud) {
 void cleanup_nop_fn(__attribute__((unused)) void *unused) {}
 void cleanup_free_fn(void *data) { free(data); }
 
+int filenames_to_abs_map_fn(void *data, void *ud) {
+  SDArchiverFileInfo *file_info = data;
+  SDArchiverHashMap **abs_filenames = ud;
+
+  // Get combined full path to file.
+  char *fullpath = filename_to_absolute_path(file_info->filename);
+  if (!fullpath) {
+    return 1;
+  }
+
+  simple_archiver_hash_map_insert(abs_filenames, fullpath, fullpath,
+                                  strlen(fullpath) + 1, cleanup_nop_fn, NULL);
+
+  // Try putting all parent dirs up to current working directory.
+  // First get absolute path to current working directory.
+  __attribute__((cleanup(free_malloced_memory))) void *cwd_dirname =
+      realpath(".", NULL);
+  if (!cwd_dirname) {
+    return 1;
+  }
+
+  // Use copy of fullpath to avoid clobbering it.
+  __attribute__((cleanup(free_malloced_memory))) void *fullpath_copy =
+      malloc(strlen(fullpath) + 1);
+  strncpy(fullpath_copy, fullpath, strlen(fullpath) + 1);
+
+  // Get dirnames.
+  char *prev = fullpath_copy;
+  char *fullpath_dirname;
+  while (1) {
+    fullpath_dirname = dirname(prev);
+    if (!fullpath_dirname || strlen(fullpath_dirname) <= strlen(cwd_dirname)) {
+      break;
+    } else {
+      // Make and store copy of fullpath_dirname.
+      char *fullpath_dirname_copy = malloc(strlen(fullpath_dirname) + 1);
+      strncpy(fullpath_dirname_copy, fullpath_dirname,
+              strlen(fullpath_dirname) + 1);
+      simple_archiver_hash_map_insert(
+          abs_filenames, fullpath_dirname_copy, fullpath_dirname_copy,
+          strlen(fullpath_dirname_copy) + 1, cleanup_nop_fn, NULL);
+    }
+    prev = fullpath_dirname;
+  }
+
+  return 0;
+}
+
 char *simple_archiver_error_to_string(enum SDArchiverStateReturns error) {
   switch (error) {
     case SDAS_SUCCESS:
@@ -834,6 +924,10 @@ char *simple_archiver_error_to_string(enum SDArchiverStateReturns error) {
       return "Invalid file";
     case SDAS_INTERNAL_ERROR:
       return "Internal error";
+    case SDAS_FAILED_TO_CREATE_MAP:
+      return "Failed to create set of filenames (internal error)";
+    case SDAS_FAILED_TO_EXTRACT_SYMLINK:
+      return "Failed to extract symlink (internal error)";
     default:
       return "Unknown error";
   }
@@ -848,6 +942,7 @@ SDArchiverState *simple_archiver_init_state(const SDArchiverParsed *parsed) {
   state->flags = 0;
   state->parsed = parsed;
   state->out_f = NULL;
+  state->map = NULL;
 
   return state;
 }
@@ -861,6 +956,14 @@ void simple_archiver_free_state(SDArchiverState **state) {
 
 int simple_archiver_write_all(FILE *out_f, SDArchiverState *state,
                               const SDArchiverLinkedList *filenames) {
+  // First create a "set" of absolute paths to given filenames.
+  __attribute__((cleanup(simple_archiver_hash_map_free)))
+  SDArchiverHashMap *abs_filenames = simple_archiver_hash_map_init();
+  if (simple_archiver_list_get(filenames, filenames_to_abs_map_fn,
+                               &abs_filenames)) {
+    return SDAS_FAILED_TO_CREATE_MAP;
+  }
+
   if (fwrite("SIMPLE_ARCHIVE_VER", 1, 18, out_f) != 18) {
     return SDAS_FAILED_TO_WRITE;
   }
@@ -944,6 +1047,7 @@ int simple_archiver_write_all(FILE *out_f, SDArchiverState *state,
   state->count = 0;
   state->max = filenames->count;
   state->out_f = out_f;
+  state->map = abs_filenames;
   fprintf(stderr, "Begin archiving...\n");
   fprintf(stderr, "[%10u/%10u]\n", state->count, state->max);
   if (simple_archiver_list_get(filenames, write_files_fn, state)) {
@@ -1511,6 +1615,14 @@ int simple_archiver_parse_archive_info(FILE *in_f, int do_extract,
       }
     } else {
       // Is a symbolic link.
+
+      int abs_preferred = (buf[1] & 0x4) != 0 ? 1 : 0;
+      fprintf(stderr, "  Absolute path is %s\n",
+              (abs_preferred ? "preferred" : "NOT preferred"));
+
+      __attribute__((cleanup(free_malloced_memory))) void *abs_path = NULL;
+      __attribute__((cleanup(free_malloced_memory))) void *rel_path = NULL;
+
       if (fread(&u16, 2, 1, in_f) != 1) {
         return SDAS_INVALID_FILE;
       }
@@ -1523,15 +1635,15 @@ int simple_archiver_parse_archive_info(FILE *in_f, int do_extract,
         }
         buf[1023] = 0;
         fprintf(stderr, "  Link absolute path: %s\n", buf);
+        abs_path = malloc((size_t)u16 + 1);
+        strncpy(abs_path, (char *)buf, (size_t)u16 + 1);
       } else {
-        __attribute__((cleanup(free_malloced_memory))) void *heap_buf =
+        abs_path = malloc(u16 + 1);
-            malloc(u16 + 1);
+        if (fread(abs_path, 1, u16 + 1, in_f) != (size_t)u16 + 1) {
-        unsigned char *uc_heap_buf = heap_buf;
-        if (fread(uc_heap_buf, 1, u16 + 1, in_f) != (size_t)u16 + 1) {
           return SDAS_INVALID_FILE;
         }
-        uc_heap_buf[u16 - 1] = 0;
+        ((char *)abs_path)[u16 - 1] = 0;
-        fprintf(stderr, "  Link absolute path: %s\n", uc_heap_buf);
+        fprintf(stderr, "  Link absolute path: %s\n", (char *)abs_path);
       }
 
       if (fread(&u16, 2, 1, in_f) != 1) {
@@ -1546,15 +1658,62 @@ int simple_archiver_parse_archive_info(FILE *in_f, int do_extract,
         }
         buf[1023] = 0;
         fprintf(stderr, "  Link relative path: %s\n", buf);
+        rel_path = malloc((size_t)u16 + 1);
+        strncpy(rel_path, (char *)buf, (size_t)u16 + 1);
       } else {
-        __attribute__((cleanup(free_malloced_memory))) void *heap_buf =
+        rel_path = malloc(u16 + 1);
-            malloc(u16 + 1);
+        if (fread(rel_path, 1, u16 + 1, in_f) != (size_t)u16 + 1) {
-        unsigned char *uc_heap_buf = heap_buf;
-        if (fread(uc_heap_buf, 1, u16 + 1, in_f) != (size_t)u16 + 1) {
           return SDAS_INVALID_FILE;
         }
-        uc_heap_buf[u16 - 1] = 0;
+        ((char *)rel_path)[u16 - 1] = 0;
-        fprintf(stderr, "  Link relative path: %s\n", uc_heap_buf);
+        fprintf(stderr, "  Link relative path: %s\n", (char *)rel_path);
+      }
+
+      if (do_extract) {
+        simple_archiver_helper_make_dirs((const char *)out_f_name);
+        if (abs_path && rel_path) {
+          if (abs_preferred) {
+#if SIMPLE_ARCHIVER_PLATFORM == SIMPLE_ARCHIVER_PLATFORM_COSMOPOLITAN || \
+    SIMPLE_ARCHIVER_PLATFORM == SIMPLE_ARCHIVER_PLATFORM_MAC ||          \
+    SIMPLE_ARCHIVER_PLATFORM == SIMPLE_ARCHIVER_PLATFORM_LINUX
+            int ret = symlink(abs_path, out_f_name);
+            if (ret == -1) {
+              return SDAS_FAILED_TO_EXTRACT_SYMLINK;
+            }
+#endif
+          } else {
+#if SIMPLE_ARCHIVER_PLATFORM == SIMPLE_ARCHIVER_PLATFORM_COSMOPOLITAN || \
+    SIMPLE_ARCHIVER_PLATFORM == SIMPLE_ARCHIVER_PLATFORM_MAC ||          \
+    SIMPLE_ARCHIVER_PLATFORM == SIMPLE_ARCHIVER_PLATFORM_LINUX
+            int ret = symlink(rel_path, out_f_name);
+            if (ret == -1) {
+              return SDAS_FAILED_TO_EXTRACT_SYMLINK;
+            }
+#endif
+          }
+        } else if (abs_path) {
+#if SIMPLE_ARCHIVER_PLATFORM == SIMPLE_ARCHIVER_PLATFORM_COSMOPOLITAN || \
+    SIMPLE_ARCHIVER_PLATFORM == SIMPLE_ARCHIVER_PLATFORM_MAC ||          \
+    SIMPLE_ARCHIVER_PLATFORM == SIMPLE_ARCHIVER_PLATFORM_LINUX
+          int ret = symlink(abs_path, out_f_name);
+          if (ret == -1) {
+            return SDAS_FAILED_TO_EXTRACT_SYMLINK;
+          }
+#endif
+        } else if (rel_path) {
+#if SIMPLE_ARCHIVER_PLATFORM == SIMPLE_ARCHIVER_PLATFORM_COSMOPOLITAN || \
+    SIMPLE_ARCHIVER_PLATFORM == SIMPLE_ARCHIVER_PLATFORM_MAC ||          \
+    SIMPLE_ARCHIVER_PLATFORM == SIMPLE_ARCHIVER_PLATFORM_LINUX
+          int ret = symlink(rel_path, out_f_name);
+          if (ret == -1) {
+            return SDAS_FAILED_TO_EXTRACT_SYMLINK;
+          }
+#endif
+        } else {
+          fprintf(
+              stderr,
+              "WARNING: Symlink entry in archive has no paths to link to!\n");
+        }
       }
     }
   }

src/archiver.h

@@ -21,6 +21,7 @@
 
 #include <stdio.h>
 
+#include "data_structures/hash_map.h"
 #include "data_structures/linked_list.h"
 #include "parser.h"
 
@@ -30,6 +31,7 @@ typedef struct SDArchiverState {
   unsigned int flags;
   const SDArchiverParsed *parsed;
   FILE *out_f;
+  SDArchiverHashMap *map;
   unsigned int count;
   unsigned int max;
 } SDArchiverState;
@@ -42,7 +44,9 @@ enum SDArchiverStateReturns {
   SDAS_NO_DECOMPRESSOR,
   SDAS_INVALID_PARSED_STATE,
   SDAS_INVALID_FILE,
-  SDAS_INTERNAL_ERROR
+  SDAS_INTERNAL_ERROR,
+  SDAS_FAILED_TO_CREATE_MAP,
+  SDAS_FAILED_TO_EXTRACT_SYMLINK
 };
 
 /// Returned pointer must not be freed.

src/parser.c

@@ -151,6 +151,8 @@ void simple_archiver_print_usage(void) {
           "file's stored decompressor\n");
   fprintf(stderr, "--overwrite-create : allows overwriting an archive file\n");
   fprintf(stderr, "--overwrite-extract : allows overwriting when extracting\n");
+  fprintf(stderr,
+          "--no-abs-symlink : do not store absolute paths for symlinks\n");
   fprintf(stderr,
           "-- : specifies remaining arguments are files to archive/extract\n");
   fprintf(
@@ -243,6 +245,8 @@ int simple_archiver_parse_args(int argc, const char **argv,
         out->flags |= 0x4;
       } else if (strcmp(argv[0], "--overwrite-extract") == 0) {
         out->flags |= 0x8;
+      } else if (strcmp(argv[0], "--no-abs-symlink") == 0) {
+        out->flags |= 0x20;
       } else if (argv[0][0] == '-' && argv[0][1] == '-' && argv[0][2] == 0) {
         is_remaining_args = 1;
       } else if (argv[0][0] != '-') {

src/parser.h

@@ -30,6 +30,7 @@ typedef struct SDArchiverParsed {
   /// 0b xxxx x1xx - Allow create archive overwrite.
   /// 0b xxxx 1xxx - Allow extract overwrite.
   /// 0b xxx1 xxxx - Create archive to stdout or read archive from stdin.
+  /// 0b xx1x xxxx - Do not save absolute paths for symlinks.
   unsigned int flags;
   /// Null-terminated string.
   char *filename;