git.oblomov.eu Git - git/blob - list-objects-filter.c

   1 #include "cache.h"
   2 #include "dir.h"
   3 #include "tag.h"
   4 #include "commit.h"
   5 #include "tree.h"
   6 #include "blob.h"
   7 #include "diff.h"
   8 #include "tree-walk.h"
   9 #include "revision.h"
  10 #include "list-objects.h"
  11 #include "list-objects-filter.h"
  12 #include "list-objects-filter-options.h"
  13 #include "oidset.h"
  14 #include "object-store.h"
  15
  16 /* Remember to update object flag allocation in object.h */
  17 /*
  18  * FILTER_SHOWN_BUT_REVISIT -- we set this bit on tree objects
  19  * that have been shown, but should be revisited if they appear
  20  * in the traversal (until we mark it SEEN).  This is a way to
  21  * let us silently de-dup calls to show() in the caller.  This
  22  * is subtly different from the "revision.h:SHOWN" and the
  23  * "sha1-name.c:ONELINE_SEEN" bits.  And also different from
  24  * the non-de-dup usage in pack-bitmap.c
  25  */
  26 #define FILTER_SHOWN_BUT_REVISIT (1<<21)
  27
  28 /*
  29  * A filter for list-objects to omit ALL blobs from the traversal.
  30  * And to OPTIONALLY collect a list of the omitted OIDs.
  31  */
  32 struct filter_blobs_none_data {
  33         struct oidset *omits;
  34 };
  35
  36 static enum list_objects_filter_result filter_blobs_none(
  37         enum list_objects_filter_situation filter_situation,
  38         struct object *obj,
  39         const char *pathname,
  40         const char *filename,
  41         void *filter_data_)
  42 {
  43         struct filter_blobs_none_data *filter_data = filter_data_;
  44
  45         switch (filter_situation) {
  46         default:
  47                 BUG("unknown filter_situation: %d", filter_situation);
  48
  49         case LOFS_BEGIN_TREE:
  50                 assert(obj->type == OBJ_TREE);
  51                 /* always include all tree objects */
  52                 return LOFR_MARK_SEEN | LOFR_DO_SHOW;
  53
  54         case LOFS_END_TREE:
  55                 assert(obj->type == OBJ_TREE);
  56                 return LOFR_ZERO;
  57
  58         case LOFS_BLOB:
  59                 assert(obj->type == OBJ_BLOB);
  60                 assert((obj->flags & SEEN) == 0);
  61
  62                 if (filter_data->omits)
  63                         oidset_insert(filter_data->omits, &obj->oid);
  64                 return LOFR_MARK_SEEN; /* but not LOFR_DO_SHOW (hard omit) */
  65         }
  66 }
  67
  68 static void *filter_blobs_none__init(
  69         struct oidset *omitted,
  70         struct list_objects_filter_options *filter_options,
  71         filter_object_fn *filter_fn,
  72         filter_free_fn *filter_free_fn)
  73 {
  74         struct filter_blobs_none_data *d = xcalloc(1, sizeof(*d));
  75         d->omits = omitted;
  76
  77         *filter_fn = filter_blobs_none;
  78         *filter_free_fn = free;
  79         return d;
  80 }
  81
  82 /*
  83  * A filter for list-objects to omit large blobs.
  84  * And to OPTIONALLY collect a list of the omitted OIDs.
  85  */
  86 struct filter_blobs_limit_data {
  87         struct oidset *omits;
  88         unsigned long max_bytes;
  89 };
  90
  91 static enum list_objects_filter_result filter_blobs_limit(
  92         enum list_objects_filter_situation filter_situation,
  93         struct object *obj,
  94         const char *pathname,
  95         const char *filename,
  96         void *filter_data_)
  97 {
  98         struct filter_blobs_limit_data *filter_data = filter_data_;
  99         unsigned long object_length;
 100         enum object_type t;
 101
 102         switch (filter_situation) {
 103         default:
 104                 BUG("unknown filter_situation: %d", filter_situation);
 105
 106         case LOFS_BEGIN_TREE:
 107                 assert(obj->type == OBJ_TREE);
 108                 /* always include all tree objects */
 109                 return LOFR_MARK_SEEN | LOFR_DO_SHOW;
 110
 111         case LOFS_END_TREE:
 112                 assert(obj->type == OBJ_TREE);
 113                 return LOFR_ZERO;
 114
 115         case LOFS_BLOB:
 116                 assert(obj->type == OBJ_BLOB);
 117                 assert((obj->flags & SEEN) == 0);
 118
 119                 t = oid_object_info(the_repository, &obj->oid, &object_length);
 120                 if (t != OBJ_BLOB) { /* probably OBJ_NONE */
 121                         /*
 122                          * We DO NOT have the blob locally, so we cannot
 123                          * apply the size filter criteria.  Be conservative
 124                          * and force show it (and let the caller deal with
 125                          * the ambiguity).
 126                          */
 127                         goto include_it;
 128                 }
 129
 130                 if (object_length < filter_data->max_bytes)
 131                         goto include_it;
 132
 133                 if (filter_data->omits)
 134                         oidset_insert(filter_data->omits, &obj->oid);
 135                 return LOFR_MARK_SEEN; /* but not LOFR_DO_SHOW (hard omit) */
 136         }
 137
 138 include_it:
 139         if (filter_data->omits)
 140                 oidset_remove(filter_data->omits, &obj->oid);
 141         return LOFR_MARK_SEEN | LOFR_DO_SHOW;
 142 }
 143
 144 static void *filter_blobs_limit__init(
 145         struct oidset *omitted,
 146         struct list_objects_filter_options *filter_options,
 147         filter_object_fn *filter_fn,
 148         filter_free_fn *filter_free_fn)
 149 {
 150         struct filter_blobs_limit_data *d = xcalloc(1, sizeof(*d));
 151         d->omits = omitted;
 152         d->max_bytes = filter_options->blob_limit_value;
 153
 154         *filter_fn = filter_blobs_limit;
 155         *filter_free_fn = free;
 156         return d;
 157 }
 158
 159 /*
 160  * A filter driven by a sparse-checkout specification to only
 161  * include blobs that a sparse checkout would populate.
 162  *
 163  * The sparse-checkout spec can be loaded from a blob with the
 164  * given OID or from a local pathname.  We allow an OID because
 165  * the repo may be bare or we may be doing the filtering on the
 166  * server.
 167  */
 168 struct frame {
 169         /*
 170          * defval is the usual default include/exclude value that
 171          * should be inherited as we recurse into directories based
 172          * upon pattern matching of the directory itself or of a
 173          * containing directory.
 174          */
 175         int defval;
 176
 177         /*
 178          * 1 if the directory (recursively) contains any provisionally
 179          * omitted objects.
 180          *
 181          * 0 if everything (recursively) contained in this directory
 182          * has been explicitly included (SHOWN) in the result and
 183          * the directory may be short-cut later in the traversal.
 184          */
 185         unsigned child_prov_omit : 1;
 186 };
 187
 188 struct filter_sparse_data {
 189         struct oidset *omits;
 190         struct exclude_list el;
 191
 192         size_t nr, alloc;
 193         struct frame *array_frame;
 194 };
 195
 196 static enum list_objects_filter_result filter_sparse(
 197         enum list_objects_filter_situation filter_situation,
 198         struct object *obj,
 199         const char *pathname,
 200         const char *filename,
 201         void *filter_data_)
 202 {
 203         struct filter_sparse_data *filter_data = filter_data_;
 204         int val, dtype;
 205         struct frame *frame;
 206
 207         switch (filter_situation) {
 208         default:
 209                 BUG("unknown filter_situation: %d", filter_situation);
 210
 211         case LOFS_BEGIN_TREE:
 212                 assert(obj->type == OBJ_TREE);
 213                 dtype = DT_DIR;
 214                 val = is_excluded_from_list(pathname, strlen(pathname),
 215                                             filename, &dtype, &filter_data->el,
 216                                             &the_index);
 217                 if (val < 0)
 218                         val = filter_data->array_frame[filter_data->nr].defval;
 219
 220                 ALLOC_GROW(filter_data->array_frame, filter_data->nr + 1,
 221                            filter_data->alloc);
 222                 filter_data->nr++;
 223                 filter_data->array_frame[filter_data->nr].defval = val;
 224                 filter_data->array_frame[filter_data->nr].child_prov_omit = 0;
 225
 226                 /*
 227                  * A directory with this tree OID may appear in multiple
 228                  * places in the tree. (Think of a directory move or copy,
 229                  * with no other changes, so the OID is the same, but the
 230                  * full pathnames of objects within this directory are new
 231                  * and may match is_excluded() patterns differently.)
 232                  * So we cannot mark this directory as SEEN (yet), since
 233                  * that will prevent process_tree() from revisiting this
 234                  * tree object with other pathname prefixes.
 235                  *
 236                  * Only _DO_SHOW the tree object the first time we visit
 237                  * this tree object.
 238                  *
 239                  * We always show all tree objects.  A future optimization
 240                  * may want to attempt to narrow this.
 241                  */
 242                 if (obj->flags & FILTER_SHOWN_BUT_REVISIT)
 243                         return LOFR_ZERO;
 244                 obj->flags |= FILTER_SHOWN_BUT_REVISIT;
 245                 return LOFR_DO_SHOW;
 246
 247         case LOFS_END_TREE:
 248                 assert(obj->type == OBJ_TREE);
 249                 assert(filter_data->nr > 0);
 250
 251                 frame = &filter_data->array_frame[filter_data->nr];
 252                 filter_data->nr--;
 253
 254                 /*
 255                  * Tell our parent directory if any of our children were
 256                  * provisionally omitted.
 257                  */
 258                 filter_data->array_frame[filter_data->nr].child_prov_omit |=
 259                         frame->child_prov_omit;
 260
 261                 /*
 262                  * If there are NO provisionally omitted child objects (ALL child
 263                  * objects in this folder were INCLUDED), then we can mark the
 264                  * folder as SEEN (so we will not have to revisit it again).
 265                  */
 266                 if (!frame->child_prov_omit)
 267                         return LOFR_MARK_SEEN;
 268                 return LOFR_ZERO;
 269
 270         case LOFS_BLOB:
 271                 assert(obj->type == OBJ_BLOB);
 272                 assert((obj->flags & SEEN) == 0);
 273
 274                 frame = &filter_data->array_frame[filter_data->nr];
 275
 276                 dtype = DT_REG;
 277                 val = is_excluded_from_list(pathname, strlen(pathname),
 278                                             filename, &dtype, &filter_data->el,
 279                                             &the_index);
 280                 if (val < 0)
 281                         val = frame->defval;
 282                 if (val > 0) {
 283                         if (filter_data->omits)
 284                                 oidset_remove(filter_data->omits, &obj->oid);
 285                         return LOFR_MARK_SEEN | LOFR_DO_SHOW;
 286                 }
 287
 288                 /*
 289                  * Provisionally omit it.  We've already established that
 290                  * this pathname is not in the sparse-checkout specification
 291                  * with the CURRENT pathname, so we *WANT* to omit this blob.
 292                  *
 293                  * However, a pathname elsewhere in the tree may also
 294                  * reference this same blob, so we cannot reject it yet.
 295                  * Leave the LOFR_ bits unset so that if the blob appears
 296                  * again in the traversal, we will be asked again.
 297                  */
 298                 if (filter_data->omits)
 299                         oidset_insert(filter_data->omits, &obj->oid);
 300
 301                 /*
 302                  * Remember that at least 1 blob in this tree was
 303                  * provisionally omitted.  This prevents us from short
 304                  * cutting the tree in future iterations.
 305                  */
 306                 frame->child_prov_omit = 1;
 307                 return LOFR_ZERO;
 308         }
 309 }
 310
 311
 312 static void filter_sparse_free(void *filter_data)
 313 {
 314         struct filter_sparse_data *d = filter_data;
 315         /* TODO free contents of 'd' */
 316         free(d);
 317 }
 318
 319 static void *filter_sparse_oid__init(
 320         struct oidset *omitted,
 321         struct list_objects_filter_options *filter_options,
 322         filter_object_fn *filter_fn,
 323         filter_free_fn *filter_free_fn)
 324 {
 325         struct filter_sparse_data *d = xcalloc(1, sizeof(*d));
 326         d->omits = omitted;
 327         if (add_excludes_from_blob_to_list(filter_options->sparse_oid_value,
 328                                            NULL, 0, &d->el) < 0)
 329                 die("could not load filter specification");
 330
 331         ALLOC_GROW(d->array_frame, d->nr + 1, d->alloc);
 332         d->array_frame[d->nr].defval = 0; /* default to include */
 333         d->array_frame[d->nr].child_prov_omit = 0;
 334
 335         *filter_fn = filter_sparse;
 336         *filter_free_fn = filter_sparse_free;
 337         return d;
 338 }
 339
 340 static void *filter_sparse_path__init(
 341         struct oidset *omitted,
 342         struct list_objects_filter_options *filter_options,
 343         filter_object_fn *filter_fn,
 344         filter_free_fn *filter_free_fn)
 345 {
 346         struct filter_sparse_data *d = xcalloc(1, sizeof(*d));
 347         d->omits = omitted;
 348         if (add_excludes_from_file_to_list(filter_options->sparse_path_value,
 349                                            NULL, 0, &d->el, NULL) < 0)
 350                 die("could not load filter specification");
 351
 352         ALLOC_GROW(d->array_frame, d->nr + 1, d->alloc);
 353         d->array_frame[d->nr].defval = 0; /* default to include */
 354         d->array_frame[d->nr].child_prov_omit = 0;
 355
 356         *filter_fn = filter_sparse;
 357         *filter_free_fn = filter_sparse_free;
 358         return d;
 359 }
 360
 361 typedef void *(*filter_init_fn)(
 362         struct oidset *omitted,
 363         struct list_objects_filter_options *filter_options,
 364         filter_object_fn *filter_fn,
 365         filter_free_fn *filter_free_fn);
 366
 367 /*
 368  * Must match "enum list_objects_filter_choice".
 369  */
 370 static filter_init_fn s_filters[] = {
 371         NULL,
 372         filter_blobs_none__init,
 373         filter_blobs_limit__init,
 374         filter_sparse_oid__init,
 375         filter_sparse_path__init,
 376 };
 377
 378 void *list_objects_filter__init(
 379         struct oidset *omitted,
 380         struct list_objects_filter_options *filter_options,
 381         filter_object_fn *filter_fn,
 382         filter_free_fn *filter_free_fn)
 383 {
 384         filter_init_fn init_fn;
 385
 386         assert((sizeof(s_filters) / sizeof(s_filters[0])) == LOFC__COUNT);
 387
 388         if (filter_options->choice >= LOFC__COUNT)
 389                 BUG("invalid list-objects filter choice: %d",
 390                     filter_options->choice);
 391
 392         init_fn = s_filters[filter_options->choice];
 393         if (init_fn)
 394                 return init_fn(omitted, filter_options,
 395                                filter_fn, filter_free_fn);
 396         *filter_fn = NULL;
 397         *filter_free_fn = NULL;
 398         return NULL;
 399 }