@matheusbsbs / matheus-bagliano / commits / 2a113aee9b

Fork 0

fast-import: Proper notes tree manipulation

This patch teaches 'git fast-import' to automatically organize note objects in a fast-import stream into an appropriate fanout structure. The notes API in notes.h is NOT used to accomplish this, because trying to keep the fast-import and notes data structures in sync would yield a significantly larger patch with higher complexity. Note objects are added with the 'N' command, and accounted for with a per-branch counter, which is used to trigger fanout restructuring when needed. Note that when restructuring the branch tree, _any_ entry whose path consists of 40 hex chars (not including directory separators) will be recognized as a note object. It is therefore not advisable to manipulate note entries with M/D/R/C commands. Since note objects are stored in the same tree structure as other objects, the unloading and reloading of a fast-import branches handle note objects transparently. This patch has been improved by the following contributions: - Shawn O. Pearce: Several style- and logic-related improvements Cc: Shawn O. Pearce <spearce@spearce.org> Signed-off-by: Johan Herland <johan@herland.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>

Johan Herland committed Dec 7, 2009 at 12:27 UTC 2a113aee9ba6ba55fc9a4c714efc958ab0de6510

2 files changed +272 -18

fast-import.c

+129 -5

@@ -245,6 +245,7 @@ struct branch

245 const char *name;

246 struct tree_entry branch_tree;

247 uintmax_t last_commit;

248 + uintmax_t num_notes;

249 unsigned active : 1;

250 unsigned pack_id : PACK_ID_BITS;

251 unsigned char sha1[20];

@@ -693,6 +694,7 @@ static struct branch *new_branch(const char *name)

694 b->table_next_branch = branch_table[hc];

695 b->branch_tree.versions[0].mode = S_IFDIR;

696 b->branch_tree.versions[1].mode = S_IFDIR;

697 + b->num_notes = 0;

698 b->active = 0;

699 b->pack_id = MAX_PACK_ID;

700 branch_table[hc] = b;

@@ -1860,6 +1862,109 @@ static void load_branch(struct branch *b)

1862 }

1863 }

1864

1865 +static unsigned char convert_num_notes_to_fanout(uintmax_t num_notes)

1866 +{

1867 + unsigned char fanout = 0;

1868 + while ((num_notes >>= 8))

1869 + fanout++;

1870 + return fanout;

1871 +}

1872 +

1873 +static void construct_path_with_fanout(const char *hex_sha1,

1874 + unsigned char fanout, char *path)

1875 +{

1876 + unsigned int i = 0, j = 0;

1877 + if (fanout >= 20)

1878 + die("Too large fanout (%u)", fanout);

1879 + while (fanout) {

1880 + path[i++] = hex_sha1[j++];

1881 + path[i++] = hex_sha1[j++];

1882 + path[i++] = '/';

1883 + fanout--;

1884 + }

1885 + memcpy(path + i, hex_sha1 + j, 40 - j);

1886 + path[i + 40 - j] = '\0';

1887 +}

1888 +

1889 +static uintmax_t do_change_note_fanout(

1890 + struct tree_entry *orig_root, struct tree_entry *root,

1891 + char *hex_sha1, unsigned int hex_sha1_len,

1892 + char *fullpath, unsigned int fullpath_len,

1893 + unsigned char fanout)

1894 +{

1895 + struct tree_content *t = root->tree;

1896 + struct tree_entry *e, leaf;

1897 + unsigned int i, tmp_hex_sha1_len, tmp_fullpath_len;

1898 + uintmax_t num_notes = 0;

1899 + unsigned char sha1[20];

1900 + char realpath[60];

1901 +

1902 + for (i = 0; t && i < t->entry_count; i++) {

1903 + e = t->entries[i];

1904 + tmp_hex_sha1_len = hex_sha1_len + e->name->str_len;

1905 + tmp_fullpath_len = fullpath_len;

1906 +

1907 + /*

1908 + * We're interested in EITHER existing note entries (entries

1909 + * with exactly 40 hex chars in path, not including directory

1910 + * separators), OR directory entries that may contain note

1911 + * entries (with < 40 hex chars in path).

1912 + * Also, each path component in a note entry must be a multiple

1913 + * of 2 chars.

1914 + */

1915 + if (!e->versions[1].mode ||

1916 + tmp_hex_sha1_len > 40 ||

1917 + e->name->str_len % 2)

1918 + continue;

1919 +

1920 + /* This _may_ be a note entry, or a subdir containing notes */

1921 + memcpy(hex_sha1 + hex_sha1_len, e->name->str_dat,

1922 + e->name->str_len);

1923 + if (tmp_fullpath_len)

1924 + fullpath[tmp_fullpath_len++] = '/';

1925 + memcpy(fullpath + tmp_fullpath_len, e->name->str_dat,

1926 + e->name->str_len);

1927 + tmp_fullpath_len += e->name->str_len;

1928 + fullpath[tmp_fullpath_len] = '\0';

1929 +

1930 + if (tmp_hex_sha1_len == 40 && !get_sha1_hex(hex_sha1, sha1)) {

1931 + /* This is a note entry */

1932 + construct_path_with_fanout(hex_sha1, fanout, realpath);

1933 + if (!strcmp(fullpath, realpath)) {

1934 + /* Note entry is in correct location */

1935 + num_notes++;

1936 + continue;

1937 + }

1938 +

1939 + /* Rename fullpath to realpath */

1940 + if (!tree_content_remove(orig_root, fullpath, &leaf))

1941 + die("Failed to remove path %s", fullpath);

1942 + tree_content_set(orig_root, realpath,

1943 + leaf.versions[1].sha1,

1944 + leaf.versions[1].mode,

1945 + leaf.tree);

1946 + } else if (S_ISDIR(e->versions[1].mode)) {

1947 + /* This is a subdir that may contain note entries */

1948 + if (!e->tree)

1949 + load_tree(e);

1950 + num_notes += do_change_note_fanout(orig_root, e,

1951 + hex_sha1, tmp_hex_sha1_len,

1952 + fullpath, tmp_fullpath_len, fanout);

1953 + }

1954 +

1955 + /* The above may have reallocated the current tree_content */

1956 + t = root->tree;

1957 + }

1958 + return num_notes;

1959 +}

1960 +

1961 +static uintmax_t change_note_fanout(struct tree_entry *root,

1962 + unsigned char fanout)

1963 +{

1964 + char hex_sha1[40], path[60];

1965 + return do_change_note_fanout(root, root, hex_sha1, 0, path, 0, fanout);

1966 +}

1967 +

1968 static void file_change_m(struct branch *b)

1969 {

1970 const char *p = command_buf.buf + 2;

@@ -2010,14 +2115,16 @@ static void file_change_cr(struct branch *b, int rename)

2115 leaf.tree);

2116 }

2117

2013 -static void note_change_n(struct branch *b)

2118 +static void note_change_n(struct branch *b, unsigned char old_fanout)

2119 {

2120 const char *p = command_buf.buf + 2;

2121 static struct strbuf uq = STRBUF_INIT;

2122 struct object_entry *oe = oe;

2123 struct branch *s;

2124 unsigned char sha1[20], commit_sha1[20];

2125 + char path[60];

2126 uint16_t inline_data = 0;

2127 + unsigned char new_fanout;

2128

2129 /* <dataref> or 'inline' */

2130 if (*p == ':') {

@@ -2071,7 +2178,7 @@ static void note_change_n(struct branch *b)

2178 if (oe->type != OBJ_BLOB)

2179 die("Not a blob (actually a %s): %s",

2180 typename(oe->type), command_buf.buf);

2074 - } else {

2181 + } else if (!is_null_sha1(sha1)) {

2182 enum object_type type = sha1_object_info(sha1, NULL);

2183 if (type < 0)

2184 die("Blob not found: %s", command_buf.buf);

@@ -2080,8 +2187,17 @@ static void note_change_n(struct branch *b)

2187 typename(type), command_buf.buf);

2188 }

2189

2083 - tree_content_set(&b->branch_tree, sha1_to_hex(commit_sha1), sha1,

2084 - S_IFREG | 0644, NULL);

2190 + construct_path_with_fanout(sha1_to_hex(commit_sha1), old_fanout, path);

2191 + if (tree_content_remove(&b->branch_tree, path, NULL))

2192 + b->num_notes--;

2193 +

2194 + if (is_null_sha1(sha1))

2195 + return; /* nothing to insert */

2196 +

2197 + b->num_notes++;

2198 + new_fanout = convert_num_notes_to_fanout(b->num_notes);

2199 + construct_path_with_fanout(sha1_to_hex(commit_sha1), new_fanout, path);

2200 + tree_content_set(&b->branch_tree, path, sha1, S_IFREG | 0644, NULL);

2201 }

2202

2203 static void file_change_deleteall(struct branch *b)

@@ -2090,6 +2206,7 @@ static void file_change_deleteall(struct branch *b)

2206 hashclr(b->branch_tree.versions[0].sha1);

2207 hashclr(b->branch_tree.versions[1].sha1);

2208 load_tree(&b->branch_tree);

2209 + b->num_notes = 0;

2210 }

2211

2212 static void parse_from_commit(struct branch *b, char *buf, unsigned long size)

@@ -2213,6 +2330,7 @@ static void parse_new_commit(void)

2330 char *committer = NULL;

2331 struct hash_list *merge_list = NULL;

2332 unsigned int merge_count;

2333 + unsigned char prev_fanout, new_fanout;

2334

2335 /* Obtain the branch name from the rest of our command */

2336 sp = strchr(command_buf.buf, ' ') + 1;

@@ -2243,6 +2361,8 @@ static void parse_new_commit(void)

2361 load_branch(b);

2362 }

2363

2364 + prev_fanout = convert_num_notes_to_fanout(b->num_notes);

2365 +

2366 /* file_change* */

2367 while (command_buf.len > 0) {

2368 if (!prefixcmp(command_buf.buf, "M "))

@@ -2254,7 +2374,7 @@ static void parse_new_commit(void)

2374 else if (!prefixcmp(command_buf.buf, "C "))

2375 file_change_cr(b, 0);

2376 else if (!prefixcmp(command_buf.buf, "N "))

2257 - note_change_n(b);

2377 + note_change_n(b, prev_fanout);

2378 else if (!strcmp("deleteall", command_buf.buf))

2379 file_change_deleteall(b);

2380 else {

@@ -2265,6 +2385,10 @@ static void parse_new_commit(void)

2385 break;

2386 }

2387

2388 + new_fanout = convert_num_notes_to_fanout(b->num_notes);

2389 + if (new_fanout != prev_fanout)

2390 + b->num_notes = change_note_fanout(&b->branch_tree, new_fanout);

2391 +

2392 /* build the tree and the commit */

2393 store_tree(&b->branch_tree);

2394 hashcpy(b->branch_tree.versions[0].sha1,

t/t9300-fast-import.sh

+143 -13

@@ -1092,9 +1092,12 @@ test_expect_success 'P: fail on blob mark in gitlink' '

1092 ### series Q (notes)

1093 ###

1094

1095 -note1_data="Note for the first commit"

1096 -note2_data="Note for the second commit"

1097 -note3_data="Note for the third commit"

1095 +note1_data="The first note for the first commit"

1096 +note2_data="The first note for the second commit"

1097 +note3_data="The first note for the third commit"

1098 +note1b_data="The second note for the first commit"

1099 +note1c_data="The third note for the first commit"

1100 +note2b_data="The second note for the second commit"

1101

1102 test_tick

1103 cat >input <<INPUT_END

@@ -1169,7 +1172,45 @@ data <<EOF

1172 $note3_data

1173 EOF

1174

1175 +commit refs/notes/foobar

1176 +mark :10

1177 +committer $GIT_COMMITTER_NAME <$GIT_COMMITTER_EMAIL> $GIT_COMMITTER_DATE

1178 +data <<COMMIT

1179 +notes (:10)

1180 +COMMIT

1181 +

1182 +N inline :3

1183 +data <<EOF

1184 +$note1b_data

1185 +EOF

1186 +

1187 +commit refs/notes/foobar2

1188 +mark :11

1189 +committer $GIT_COMMITTER_NAME <$GIT_COMMITTER_EMAIL> $GIT_COMMITTER_DATE

1190 +data <<COMMIT

1191 +notes (:11)

1192 +COMMIT

1193 +

1194 +N inline :3

1195 +data <<EOF

1196 +$note1c_data

1197 +EOF

1198 +

1199 +commit refs/notes/foobar

1200 +mark :12

1201 +committer $GIT_COMMITTER_NAME <$GIT_COMMITTER_EMAIL> $GIT_COMMITTER_DATE

1202 +data <<COMMIT

1203 +notes (:12)

1204 +COMMIT

1205 +

1206 +deleteall

1207 +N inline :5

1208 +data <<EOF

1209 +$note2b_data

1210 +EOF

1211 +

1212 INPUT_END

1213 +

1214 test_expect_success \

1215 'Q: commit notes' \

1216 'git fast-import <input &&

@@ -1224,8 +1265,8 @@ committer $GIT_COMMITTER_NAME <$GIT_COMMITTER_EMAIL> $GIT_COMMITTER_DATE

1265 notes (:9)

1266 EOF

1267 test_expect_success \

1227 - 'Q: verify notes commit' \

1228 - 'git cat-file commit refs/notes/foobar | sed 1d >actual &&

1268 + 'Q: verify first notes commit' \

1269 + 'git cat-file commit refs/notes/foobar~2 | sed 1d >actual &&

1270 test_cmp expect actual'

1271

1272 cat >expect.unsorted <<EOF

@@ -1235,23 +1276,112 @@ cat >expect.unsorted <<EOF

1276 EOF

1277 cat expect.unsorted | sort >expect

1278 test_expect_success \

1238 - 'Q: verify notes tree' \

1239 - 'git cat-file -p refs/notes/foobar^{tree} | sed "s/ [0-9a-f]* / /" >actual &&

1279 + 'Q: verify first notes tree' \

1280 + 'git cat-file -p refs/notes/foobar~2^{tree} | sed "s/ [0-9a-f]* / /" >actual &&

1281 test_cmp expect actual'

1282

1283 echo "$note1_data" >expect

1284 test_expect_success \

1244 - 'Q: verify note for first commit' \

1245 - 'git cat-file blob refs/notes/foobar:$commit1 >actual && test_cmp expect actual'

1285 + 'Q: verify first note for first commit' \

1286 + 'git cat-file blob refs/notes/foobar~2:$commit1 >actual && test_cmp expect actual'

1287

1288 echo "$note2_data" >expect

1289 test_expect_success \

1249 - 'Q: verify note for second commit' \

1250 - 'git cat-file blob refs/notes/foobar:$commit2 >actual && test_cmp expect actual'

1290 + 'Q: verify first note for second commit' \

1291 + 'git cat-file blob refs/notes/foobar~2:$commit2 >actual && test_cmp expect actual'

1292 +

1293 +echo "$note3_data" >expect

1294 +test_expect_success \

1295 + 'Q: verify first note for third commit' \

1296 + 'git cat-file blob refs/notes/foobar~2:$commit3 >actual && test_cmp expect actual'

1297 +

1298 +cat >expect <<EOF

1299 +parent `git rev-parse --verify refs/notes/foobar~2`

1300 +author $GIT_COMMITTER_NAME <$GIT_COMMITTER_EMAIL> $GIT_COMMITTER_DATE

1301 +committer $GIT_COMMITTER_NAME <$GIT_COMMITTER_EMAIL> $GIT_COMMITTER_DATE

1302 +

1303 +notes (:10)

1304 +EOF

1305 +test_expect_success \

1306 + 'Q: verify second notes commit' \

1307 + 'git cat-file commit refs/notes/foobar^ | sed 1d >actual &&

1308 + test_cmp expect actual'

1309 +

1310 +cat >expect.unsorted <<EOF

1311 +100644 blob $commit1

1312 +100644 blob $commit2

1313 +100644 blob $commit3

1314 +EOF

1315 +cat expect.unsorted | sort >expect

1316 +test_expect_success \

1317 + 'Q: verify second notes tree' \

1318 + 'git cat-file -p refs/notes/foobar^^{tree} | sed "s/ [0-9a-f]* / /" >actual &&

1319 + test_cmp expect actual'

1320 +

1321 +echo "$note1b_data" >expect

1322 +test_expect_success \

1323 + 'Q: verify second note for first commit' \

1324 + 'git cat-file blob refs/notes/foobar^:$commit1 >actual && test_cmp expect actual'

1325 +

1326 +echo "$note2_data" >expect

1327 +test_expect_success \

1328 + 'Q: verify first note for second commit' \

1329 + 'git cat-file blob refs/notes/foobar^:$commit2 >actual && test_cmp expect actual'

1330

1331 echo "$note3_data" >expect

1332 test_expect_success \

1254 - 'Q: verify note for third commit' \

1255 - 'git cat-file blob refs/notes/foobar:$commit3 >actual && test_cmp expect actual'

1333 + 'Q: verify first note for third commit' \

1334 + 'git cat-file blob refs/notes/foobar^:$commit3 >actual && test_cmp expect actual'

1335 +

1336 +cat >expect <<EOF

1337 +author $GIT_COMMITTER_NAME <$GIT_COMMITTER_EMAIL> $GIT_COMMITTER_DATE

1338 +committer $GIT_COMMITTER_NAME <$GIT_COMMITTER_EMAIL> $GIT_COMMITTER_DATE

1339 +

1340 +notes (:11)

1341 +EOF

1342 +test_expect_success \

1343 + 'Q: verify third notes commit' \

1344 + 'git cat-file commit refs/notes/foobar2 | sed 1d >actual &&

1345 + test_cmp expect actual'

1346 +

1347 +cat >expect.unsorted <<EOF

1348 +100644 blob $commit1

1349 +EOF

1350 +cat expect.unsorted | sort >expect

1351 +test_expect_success \

1352 + 'Q: verify third notes tree' \

1353 + 'git cat-file -p refs/notes/foobar2^{tree} | sed "s/ [0-9a-f]* / /" >actual &&

1354 + test_cmp expect actual'

1355 +

1356 +echo "$note1c_data" >expect

1357 +test_expect_success \

1358 + 'Q: verify third note for first commit' \

1359 + 'git cat-file blob refs/notes/foobar2:$commit1 >actual && test_cmp expect actual'

1360 +

1361 +cat >expect <<EOF

1362 +parent `git rev-parse --verify refs/notes/foobar^`

1363 +author $GIT_COMMITTER_NAME <$GIT_COMMITTER_EMAIL> $GIT_COMMITTER_DATE

1364 +committer $GIT_COMMITTER_NAME <$GIT_COMMITTER_EMAIL> $GIT_COMMITTER_DATE

1365 +

1366 +notes (:12)

1367 +EOF

1368 +test_expect_success \

1369 + 'Q: verify fourth notes commit' \

1370 + 'git cat-file commit refs/notes/foobar | sed 1d >actual &&

1371 + test_cmp expect actual'

1372 +

1373 +cat >expect.unsorted <<EOF

1374 +100644 blob $commit2

1375 +EOF

1376 +cat expect.unsorted | sort >expect

1377 +test_expect_success \

1378 + 'Q: verify fourth notes tree' \

1379 + 'git cat-file -p refs/notes/foobar^{tree} | sed "s/ [0-9a-f]* / /" >actual &&

1380 + test_cmp expect actual'

1381 +

1382 +echo "$note2b_data" >expect

1383 +test_expect_success \

1384 + 'Q: verify second note for second commit' \

1385 + 'git cat-file blob refs/notes/foobar:$commit2 >actual && test_cmp expect actual'

1386

1387 test_done