vcs-svn/fast_export.c

   1 /*
   2  * Licensed under a two-clause BSD-style license.
   3  * See LICENSE for details.
   4  */
   5
   6 #include "git-compat-util.h"
   7 #include "strbuf.h"
   8 #include "quote.h"
   9 #include "fast_export.h"
  10 #include "repo_tree.h"
  11 #include "strbuf.h"
  12 #include "svndiff.h"
  13 #include "sliding_window.h"
  14 #include "line_buffer.h"
  15
  16 #define MAX_GITSVN_LINE_LEN 4096
  17
  18 static uint32_t first_commit_done;
  19 static struct line_buffer postimage = LINE_BUFFER_INIT;
  20 static struct line_buffer report_buffer = LINE_BUFFER_INIT;
  21
  22 /* NEEDSWORK: move to fast_export_init() */
  23 static int init_postimage(void)
  24 {
  25         static int postimage_initialized;
  26         if (postimage_initialized)
  27                 return 0;
  28         postimage_initialized = 1;
  29         return buffer_tmpfile_init(&postimage);
  30 }
  31
  32 void fast_export_init(int fd)
  33 {
  34         first_commit_done = 0;
  35         if (buffer_fdinit(&report_buffer, fd))
  36                 die_errno("cannot read from file descriptor %d", fd);
  37 }
  38
  39 void fast_export_deinit(void)
  40 {
  41         if (buffer_deinit(&report_buffer))
  42                 die_errno("error closing fast-import feedback stream");
  43 }
  44
  45 void fast_export_delete(const char *path)
  46 {
  47         putchar('D');
  48         putchar(' ');
  49         quote_c_style(path, NULL, stdout, 0);
  50         putchar('\n');
  51 }
  52
  53 static void fast_export_truncate(const char *path, uint32_t mode)
  54 {
  55         fast_export_modify(path, mode, "inline");
  56         printf("data 0\n\n");
  57 }
  58
  59 void fast_export_modify(const char *path, uint32_t mode, const char *dataref)
  60 {
  61         /* Mode must be 100644, 100755, 120000, or 160000. */
  62         if (!dataref) {
  63                 fast_export_truncate(path, mode);
  64                 return;
  65         }
  66         printf("M %06"PRIo32" %s ", mode, dataref);
  67         quote_c_style(path, NULL, stdout, 0);
  68         putchar('\n');
  69 }
  70
  71 static char gitsvnline[MAX_GITSVN_LINE_LEN];
  72 void fast_export_begin_commit(uint32_t revision, const char *author,
  73                         const struct strbuf *log,
  74                         const char *uuid, const char *url,
  75                         unsigned long timestamp)
  76 {
  77         static const struct strbuf empty = STRBUF_INIT;
  78         if (!log)
  79                 log = &empty;
  80         if (*uuid && *url) {
  81                 snprintf(gitsvnline, MAX_GITSVN_LINE_LEN,
  82                                 "\n\ngit-svn-id: %s@%"PRIu32" %s\n",
  83                                  url, revision, uuid);
  84         } else {
  85                 *gitsvnline = '\0';
  86         }
  87         printf("commit refs/heads/master\n");
  88         printf("mark :%"PRIu32"\n", revision);
  89         printf("committer %s <%s@%s> %ld +0000\n",
  90                    *author ? author : "nobody",
  91                    *author ? author : "nobody",
  92                    *uuid ? uuid : "local", timestamp);
  93         printf("data %"PRIuMAX"\n",
  94                 (uintmax_t) (log->len + strlen(gitsvnline)));
  95         fwrite(log->buf, log->len, 1, stdout);
  96         printf("%s\n", gitsvnline);
  97         if (!first_commit_done) {
  98                 if (revision > 1)
  99                         printf("from :%"PRIu32"\n", revision - 1);
 100                 first_commit_done = 1;
 101         }
 102 }
 103
 104 void fast_export_end_commit(uint32_t revision)
 105 {
 106         printf("progress Imported commit %"PRIu32".\n\n", revision);
 107 }
 108
 109 static void ls_from_rev(uint32_t rev, const char *path)
 110 {
 111         /* ls :5 path/to/old/file */
 112         printf("ls :%"PRIu32" ", rev);
 113         quote_c_style(path, NULL, stdout, 0);
 114         putchar('\n');
 115         fflush(stdout);
 116 }
 117
 118 static void ls_from_active_commit(const char *path)
 119 {
 120         /* ls "path/to/file" */
 121         printf("ls \"");
 122         quote_c_style(path, NULL, stdout, 1);
 123         printf("\"\n");
 124         fflush(stdout);
 125 }
 126
 127 static const char *get_response_line(void)
 128 {
 129         const char *line = buffer_read_line(&report_buffer);
 130         if (line)
 131                 return line;
 132         if (buffer_ferror(&report_buffer))
 133                 die_errno("error reading from fast-import");
 134         die("unexpected end of fast-import feedback");
 135 }
 136
 137 static void die_short_read(struct line_buffer *input)
 138 {
 139         if (buffer_ferror(input))
 140                 die_errno("error reading dump file");
 141         die("invalid dump: unexpected end of file");
 142 }
 143
 144 static int ends_with(const char *s, size_t len, const char *suffix)
 145 {
 146         const size_t suffixlen = strlen(suffix);
 147         if (len < suffixlen)
 148                 return 0;
 149         return !memcmp(s + len - suffixlen, suffix, suffixlen);
 150 }
 151
 152 static int parse_cat_response_line(const char *header, off_t *len)
 153 {
 154         size_t headerlen = strlen(header);
 155         uintmax_t n;
 156         const char *type;
 157         const char *end;
 158
 159         if (ends_with(header, headerlen, " missing"))
 160                 return error("cat-blob reports missing blob: %s", header);
 161         type = strstr(header, " blob ");
 162         if (!type)
 163                 return error("cat-blob header has wrong object type: %s", header);
 164         n = strtoumax(type + strlen(" blob "), (char **) &end, 10);
 165         if (end == type + strlen(" blob "))
 166                 return error("cat-blob header does not contain length: %s", header);
 167         if (memchr(type + strlen(" blob "), '-', end - type - strlen(" blob ")))
 168                 return error("cat-blob header contains negative length: %s", header);
 169         if (n == UINTMAX_MAX || n > maximum_signed_value_of_type(off_t))
 170                 return error("blob too large for current definition of off_t");
 171         *len = n;
 172         if (*end)
 173                 return error("cat-blob header contains garbage after length: %s", header);
 174         return 0;
 175 }
 176
 177 static void check_preimage_overflow(off_t a, off_t b)
 178 {
 179         if (signed_add_overflows(a, b))
 180                 die("blob too large for current definition of off_t");
 181 }
 182
 183 static long apply_delta(off_t len, struct line_buffer *input,
 184                         const char *old_data, uint32_t old_mode)
 185 {
 186         long ret;
 187         struct sliding_view preimage = SLIDING_VIEW_INIT(&report_buffer, 0);
 188         FILE *out;
 189
 190         if (init_postimage() || !(out = buffer_tmpfile_rewind(&postimage)))
 191                 die("cannot open temporary file for blob retrieval");
 192         if (old_data) {
 193                 const char *response;
 194                 printf("cat-blob %s\n", old_data);
 195                 fflush(stdout);
 196                 response = get_response_line();
 197                 if (parse_cat_response_line(response, &preimage.max_off))
 198                         die("invalid cat-blob response: %s", response);
 199                 check_preimage_overflow(preimage.max_off, 1);
 200         }
 201         if (old_mode == REPO_MODE_LNK) {
 202                 strbuf_addstr(&preimage.buf, "link ");
 203                 check_preimage_overflow(preimage.max_off, strlen("link "));
 204                 preimage.max_off += strlen("link ");
 205                 check_preimage_overflow(preimage.max_off, 1);
 206         }
 207         if (svndiff0_apply(input, len, &preimage, out))
 208                 die("cannot apply delta");
 209         if (old_data) {
 210                 /* Read the remainder of preimage and trailing newline. */
 211                 assert(!signed_add_overflows(preimage.max_off, 1));
 212                 preimage.max_off++;     /* room for newline */
 213                 if (move_window(&preimage, preimage.max_off - 1, 1))
 214                         die("cannot seek to end of input");
 215                 if (preimage.buf.buf[0] != '\n')
 216                         die("missing newline after cat-blob response");
 217         }
 218         ret = buffer_tmpfile_prepare_to_read(&postimage);
 219         if (ret < 0)
 220                 die("cannot read temporary file for blob retrieval");
 221         strbuf_release(&preimage.buf);
 222         return ret;
 223 }
 224
 225 void fast_export_data(uint32_t mode, off_t len, struct line_buffer *input)
 226 {
 227         assert(len >= 0);
 228         if (mode == REPO_MODE_LNK) {
 229                 /* svn symlink blobs start with "link " */
 230                 if (len < 5)
 231                         die("invalid dump: symlink too short for \"link\" prefix");
 232                 len -= 5;
 233                 if (buffer_skip_bytes(input, 5) != 5)
 234                         die_short_read(input);
 235         }
 236         printf("data %"PRIuMAX"\n", (uintmax_t) len);
 237         if (buffer_copy_bytes(input, len) != len)
 238                 die_short_read(input);
 239         fputc('\n', stdout);
 240 }
 241
 242 static int parse_ls_response(const char *response, uint32_t *mode,
 243                                         struct strbuf *dataref)
 244 {
 245         const char *tab;
 246         const char *response_end;
 247
 248         assert(response);
 249         response_end = response + strlen(response);
 250
 251         if (*response == 'm') { /* Missing. */
 252                 errno = ENOENT;
 253                 return -1;
 254         }
 255
 256         /* Mode. */
 257         if (response_end - response < (signed) strlen("100644") ||
 258             response[strlen("100644")] != ' ')
 259                 die("invalid ls response: missing mode: %s", response);
 260         *mode = 0;
 261         for (; *response != ' '; response++) {
 262                 char ch = *response;
 263                 if (ch < '0' || ch > '7')
 264                         die("invalid ls response: mode is not octal: %s", response);
 265                 *mode *= 8;
 266                 *mode += ch - '0';
 267         }
 268
 269         /* ' blob ' or ' tree ' */
 270         if (response_end - response < (signed) strlen(" blob ") ||
 271             (response[1] != 'b' && response[1] != 't'))
 272                 die("unexpected ls response: not a tree or blob: %s", response);
 273         response += strlen(" blob ");
 274
 275         /* Dataref. */
 276         tab = memchr(response, '\t', response_end - response);
 277         if (!tab)
 278                 die("invalid ls response: missing tab: %s", response);
 279         strbuf_add(dataref, response, tab - response);
 280         return 0;
 281 }
 282
 283 int fast_export_ls_rev(uint32_t rev, const char *path,
 284                                 uint32_t *mode, struct strbuf *dataref)
 285 {
 286         ls_from_rev(rev, path);
 287         return parse_ls_response(get_response_line(), mode, dataref);
 288 }
 289
 290 int fast_export_ls(const char *path, uint32_t *mode, struct strbuf *dataref)
 291 {
 292         ls_from_active_commit(path);
 293         return parse_ls_response(get_response_line(), mode, dataref);
 294 }
 295
 296 void fast_export_blob_delta(uint32_t mode,
 297                                 uint32_t old_mode, const char *old_data,
 298                                 off_t len, struct line_buffer *input)
 299 {
 300         long postimage_len;
 301
 302         assert(len >= 0);
 303         postimage_len = apply_delta(len, input, old_data, old_mode);
 304         if (mode == REPO_MODE_LNK) {
 305                 buffer_skip_bytes(&postimage, strlen("link "));
 306                 postimage_len -= strlen("link ");
 307         }
 308         printf("data %ld\n", postimage_len);
 309         buffer_copy_bytes(&postimage, postimage_len);
 310         fputc('\n', stdout);
 311 }