src/svn.cpp

   1 /*
   2  *  Copyright (C) 2007  Thiago Macieira <thiago@kde.org>
   3  *
   4  *  This program is free software: you can redistribute it and/or modify
   5  *  it under the terms of the GNU General Public License as published by
   6  *  the Free Software Foundation, either version 2 of the License, or
   7  *  (at your option) any later version.
   8  *
   9  *  This program is distributed in the hope that it will be useful,
  10  *  but WITHOUT ANY WARRANTY; without even the implied warranty of
  11  *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  12  *  GNU General Public License for more details.
  13  *
  14  *  You should have received a copy of the GNU General Public License
  15  *  along with this program.  If not, see <http://www.gnu.org/licenses/>.
  16  */
  17
  18 /*
  19  * Based on svn-fast-export by Chris Lee <clee@kde.org>
  20  * License: MIT <http://www.opensource.org/licenses/mit-license.php>
  21  * URL: git://repo.or.cz/fast-import.git http://repo.or.cz/w/fast-export.git
  22  */
  23
  24 #define _XOPEN_SOURCE
  25 #define _LARGEFILE_SUPPORT
  26 #define _LARGEFILE64_SUPPORT
  27
  28 #include "svn.h"
  29
  30 #include <unistd.h>
  31 #include <string.h>
  32 #include <stdio.h>
  33 #include <time.h>
  34 #include <unistd.h>
  35
  36 #include <apr_lib.h>
  37 #include <apr_getopt.h>
  38 #include <apr_general.h>
  39
  40 #include <svn_fs.h>
  41 #include <svn_pools.h>
  42 #include <svn_repos.h>
  43 #include <svn_types.h>
  44
  45 #include <QFile>
  46 #include <QDebug>
  47
  48 #include "repository.h"
  49
  50 #undef SVN_ERR
  51 #define SVN_ERR(expr) SVN_INT_ERR(expr)
  52
  53 typedef QList<Rules::Match> MatchRuleList;
  54 typedef QHash<QString, Repository *> RepositoryHash;
  55 typedef QHash<QByteArray, QByteArray> IdentityHash;
  56
  57 class AprAutoPool
  58 {
  59     apr_pool_t *pool;
  60     AprAutoPool(const AprAutoPool &);
  61     AprAutoPool &operator=(const AprAutoPool &);
  62 public:
  63     inline AprAutoPool(apr_pool_t *parent = NULL)
  64         { pool = svn_pool_create(parent); }
  65     inline ~AprAutoPool()
  66         { svn_pool_destroy(pool); }
  67
  68     inline void clear() { svn_pool_clear(pool); }
  69     inline apr_pool_t *data() const { return pool; }
  70     inline operator apr_pool_t *() const { return pool; }
  71 };
  72
  73 class SvnPrivate
  74 {
  75 public:
  76     MatchRuleList matchRules;
  77     RepositoryHash repositories;
  78     IdentityHash identities;
  79
  80     SvnPrivate(const QString &pathToRepository);
  81     ~SvnPrivate();
  82     int youngestRevision();
  83     int exportRevision(int revnum);
  84
  85     int openRepository(const QString &pathToRepository);
  86
  87 private:
  88     AprAutoPool global_pool;
  89     svn_fs_t *fs;
  90     svn_revnum_t youngest_rev;
  91 };
  92
  93 void Svn::initialize()
  94 {
  95     // initialize APR or exit
  96     if (apr_initialize() != APR_SUCCESS) {
  97         fprintf(stderr, "You lose at apr_initialize().\n");
  98         exit(1);
  99     }
 100
 101     // static destructor
 102     static struct Destructor { ~Destructor() { apr_terminate(); } } destructor;
 103 }
 104
 105 Svn::Svn(const QString &pathToRepository)
 106     : d(new SvnPrivate(pathToRepository))
 107 {
 108 }
 109
 110 Svn::~Svn()
 111 {
 112     delete d;
 113 }
 114
 115 void Svn::setMatchRules(const MatchRuleList &matchRules)
 116 {
 117     d->matchRules = matchRules;
 118 }
 119
 120 void Svn::setRepositories(const RepositoryHash &repositories)
 121 {
 122     d->repositories = repositories;
 123 }
 124
 125 int Svn::youngestRevision()
 126 {
 127     return d->youngestRevision();
 128 }
 129
 130 bool Svn::exportRevision(int revnum)
 131 {
 132     return d->exportRevision(revnum) == EXIT_SUCCESS;
 133 }
 134
 135 SvnPrivate::SvnPrivate(const QString &pathToRepository)
 136     : global_pool(NULL)
 137 {
 138     openRepository(pathToRepository);
 139
 140     // get the youngest revision
 141     svn_fs_youngest_rev(&youngest_rev, fs, global_pool);
 142 }
 143
 144 SvnPrivate::~SvnPrivate()
 145 {
 146     svn_pool_destroy(global_pool);
 147 }
 148
 149 int SvnPrivate::youngestRevision()
 150 {
 151     return youngest_rev;
 152 }
 153
 154 int SvnPrivate::openRepository(const QString &pathToRepository)
 155 {
 156     svn_repos_t *repos;
 157     SVN_ERR(svn_repos_open(&repos, QFile::encodeName(pathToRepository), global_pool));
 158     fs = svn_repos_fs(repos);
 159
 160     return EXIT_SUCCESS;
 161 }
 162
 163 static MatchRuleList::ConstIterator
 164 findMatchRule(const MatchRuleList &matchRules, int revnum, const QString &current)
 165 {
 166     MatchRuleList::ConstIterator it = matchRules.constBegin(),
 167                                 end = matchRules.constEnd();
 168     for ( ; it != end; ++it) {
 169         if (it->minRevision > revnum)
 170             continue;
 171         if (it->maxRevision != -1 && it->maxRevision < revnum)
 172             continue;
 173         if (it->rx.indexIn(current) == 0)
 174             return it;
 175     }
 176
 177     // no match
 178     return end;
 179 }
 180
 181 static void splitPathName(const Rules::Match &rule, const QString &pathName, QString *svnprefix_p,
 182                           QString *repository_p, QString *branch_p, QString *path_p)
 183 {
 184     QString svnprefix = pathName;
 185     svnprefix.truncate(rule.rx.matchedLength());
 186     if (svnprefix_p)
 187         *svnprefix_p = svnprefix;
 188
 189     if (repository_p) {
 190         *repository_p = svnprefix;
 191         repository_p->replace(rule.rx, rule.repository);
 192     }
 193
 194     if (branch_p) {
 195         *branch_p = svnprefix;
 196         branch_p->replace(rule.rx, rule.branch);
 197     }
 198
 199     if (path_p)
 200         *path_p = pathName.mid(svnprefix.length());
 201 }
 202
 203 static int pathMode(svn_fs_root_t *fs_root, const char *pathname, apr_pool_t *pool)
 204 {
 205     svn_string_t *propvalue;
 206     SVN_ERR(svn_fs_node_prop(&propvalue, fs_root, pathname, "svn:executable", pool));
 207     int mode = 0100644;
 208     if (propvalue)
 209         mode = 0100755;
 210
 211     // maybe it's a symlink?
 212     SVN_ERR(svn_fs_node_prop(&propvalue, fs_root, pathname, "svn:special", pool));
 213     if (propvalue && strcmp(propvalue->data, "symlink") == 0)
 214         mode = 0120000;
 215
 216     return mode;
 217 }
 218
 219 svn_error_t *QIODevice_write(void *baton, const char *data, apr_size_t *len)
 220 {
 221     QIODevice *device = reinterpret_cast<QIODevice *>(baton);
 222     device->write(data, *len);
 223
 224     if (device->bytesToWrite() > 16384)
 225         device->waitForBytesWritten(0);
 226     return SVN_NO_ERROR;
 227 }
 228
 229 static svn_stream_t *streamForDevice(QIODevice *device, apr_pool_t *pool)
 230 {
 231     svn_stream_t *stream = svn_stream_create(device, pool);
 232     svn_stream_set_write(stream, QIODevice_write);
 233
 234     return stream;
 235 }
 236
 237 static int dumpBlob(Repository::Transaction *txn, svn_fs_root_t *fs_root,
 238                     const char *pathname, const QString &finalPathName, apr_pool_t *pool)
 239 {
 240     // what type is it?
 241     int mode = pathMode(fs_root, pathname, pool);
 242
 243     svn_filesize_t stream_length;
 244
 245     SVN_ERR(svn_fs_file_length(&stream_length, fs_root, pathname, pool));
 246     QIODevice *io = txn->addFile(finalPathName, mode, stream_length);
 247
 248 #ifndef DRY_RUN
 249     // open the file
 250     svn_stream_t *in_stream, *out_stream;
 251     SVN_ERR(svn_fs_file_contents(&in_stream, fs_root, pathname, pool));
 252
 253     // open a generic svn_stream_t for the QIODevice
 254     out_stream = streamForDevice(io, pool);
 255     SVN_ERR(svn_stream_copy(in_stream, out_stream, pool));
 256
 257     // print an ending newline
 258     io->putChar('\n');
 259 #endif
 260
 261     return EXIT_SUCCESS;
 262 }
 263
 264 static int recursiveDumpDir(Repository::Transaction *txn, svn_fs_root_t *fs_root,
 265                             const QByteArray &pathname, const QString &finalPathName,
 266                             apr_pool_t *pool)
 267 {
 268     // get the dir listing
 269     apr_hash_t *entries;
 270     SVN_ERR(svn_fs_dir_entries(&entries, fs_root, pathname, pool));
 271     AprAutoPool dirpool(pool);
 272
 273     for (apr_hash_index_t *i = apr_hash_first(pool, entries); i; i = apr_hash_next(i)) {
 274         dirpool.clear();
 275         const void *vkey;
 276         void *value;
 277         apr_hash_this(i, &vkey, NULL, &value);
 278
 279         svn_fs_dirent_t *dirent = reinterpret_cast<svn_fs_dirent_t *>(value);
 280         QByteArray entryName = pathname + '/' + dirent->name;
 281         QString entryFinalName;
 282         if (finalPathName.isEmpty())
 283             entryFinalName = dirent->name;
 284         else
 285             entryFinalName = finalPathName + '/' + dirent->name;
 286
 287         if (dirent->kind == svn_node_dir) {
 288             if (recursiveDumpDir(txn, fs_root, entryName, entryFinalName, dirpool) == EXIT_FAILURE)
 289                 return EXIT_FAILURE;
 290         } else if (dirent->kind == svn_node_file) {
 291             printf("+");
 292             fflush(stdout);
 293             if (dumpBlob(txn, fs_root, entryName, entryFinalName, dirpool) == EXIT_FAILURE)
 294                 return EXIT_FAILURE;
 295         }
 296     }
 297 }
 298
 299 static bool wasDir(svn_fs_t *fs, int revnum, const char *pathname, apr_pool_t *pool)
 300 {
 301     AprAutoPool subpool(pool);
 302     svn_fs_root_t *fs_root;
 303     if (svn_fs_revision_root(&fs_root, fs, revnum, subpool) != SVN_NO_ERROR)
 304         return false;
 305
 306     svn_boolean_t is_dir;
 307     if (svn_fs_is_dir(&is_dir, fs_root, pathname, subpool) != SVN_NO_ERROR)
 308         return false;
 309
 310     return is_dir;
 311 }
 312
 313 time_t get_epoch(char *svn_date)
 314 {
 315     struct tm tm;
 316     memset(&tm, 0, sizeof tm);
 317     QByteArray date(svn_date, strlen(svn_date) - 8);
 318     strptime(date, "%Y-%m-%dT%H:%M:%S", &tm);
 319     return mktime(&tm);
 320 }
 321
 322 int SvnPrivate::exportRevision(int revnum)
 323 {
 324     AprAutoPool pool(global_pool.data());
 325     QHash<QString, Repository::Transaction *> transactions;
 326
 327     // open this revision:
 328     printf("Exporting revision %d ", revnum);
 329     fflush(stdout);
 330     svn_fs_root_t *fs_root;
 331     SVN_ERR(svn_fs_revision_root(&fs_root, fs, revnum, pool));
 332
 333     // find out what was changed in this revision:
 334     apr_hash_t *changes;
 335     SVN_ERR(svn_fs_paths_changed(&changes, fs_root, pool));
 336     AprAutoPool revpool(pool.data());
 337     for (apr_hash_index_t *i = apr_hash_first(pool, changes); i; i = apr_hash_next(i)) {
 338         revpool.clear();
 339
 340         const void *vkey;
 341         void *value;
 342         apr_hash_this(i, &vkey, NULL, &value);
 343         const char *key = reinterpret_cast<const char *>(vkey);
 344         QString current = QString::fromUtf8(key);
 345         svn_fs_path_change_t *change = reinterpret_cast<svn_fs_path_change_t *>(value);
 346
 347         // was this copied from somewhere?
 348         svn_revnum_t rev_from;
 349         const char *path_from;
 350         SVN_ERR(svn_fs_copied_from(&rev_from, &path_from, fs_root, key, revpool));
 351
 352         // is this a directory?
 353         svn_boolean_t is_dir;
 354         SVN_ERR(svn_fs_is_dir(&is_dir, fs_root, key, revpool));
 355         if (is_dir) {
 356             if (path_from == NULL) {
 357                 // no, it's a new directory being added
 358                 // Git doesn't handle directories, so we don't either
 359                 //qDebug() << "   mkdir ignored:" << key;
 360                 continue;
 361             }
 362
 363             current += '/';
 364             qDebug() << "   " << key << "was copied from" << path_from;
 365         }
 366
 367         // find the first rule that matches this pathname
 368         MatchRuleList::ConstIterator match = findMatchRule(matchRules, revnum, current);
 369         if (match != matchRules.constEnd()) {
 370             const Rules::Match &rule = *match;
 371             if (rule.repository.isEmpty()) {
 372                 // ignore rule
 373                 qDebug() << "   " << qPrintable(current) << "rev" << revnum
 374                          << "-> ignored (rule line" << rule.lineNumber << ")";
 375                 continue;
 376             } else {
 377                 QString svnprefix, repository, branch, path;
 378                 splitPathName(rule, current, &svnprefix, &repository, &branch, &path);
 379
 380                 if (path.isEmpty() && path_from != NULL) {
 381                     QString previous = QString::fromUtf8(path_from) + '/';
 382                     MatchRuleList::ConstIterator prevmatch =
 383                         findMatchRule(matchRules, rev_from, previous);
 384                     if (prevmatch != matchRules.constEnd()) {
 385                         QString prevsvnprefix, prevrepository, prevbranch, prevpath;
 386                         splitPathName(*prevmatch, previous, &prevsvnprefix, &prevrepository,
 387                                       &prevbranch, &prevpath);
 388
 389                         if (!prevpath.isEmpty()) {
 390                             qDebug() << qPrintable(current) << "is a partial branch of repository"
 391                                      << qPrintable(prevrepository) << "branch"
 392                                      << qPrintable(prevbranch) << "subdir"
 393                                      << qPrintable(prevpath);
 394                         } else if (prevrepository != repository) {
 395                             qWarning() << qPrintable(current) << "rev" << revnum
 396                                        << "is a cross-repository copy (from repository"
 397                                        << qPrintable(prevrepository) << "branch"
 398                                        << qPrintable(prevbranch) << "path"
 399                                        << qPrintable(prevpath) << "rev" << rev_from << ")";
 400                         } else if (prevbranch == branch) {
 401                             // same branch and same repository
 402                             qDebug() << qPrintable(current) << "rev" << revnum
 403                                      << "is an SVN rename from"
 404                                      << qPrintable(previous) << "rev" << rev_from;
 405                             continue;
 406                         } else {
 407                             // same repository but not same branch
 408                             // this means this is a plain branch
 409                             qDebug() << qPrintable(repository) << ": branch"
 410                                      << qPrintable(branch) << "is branching from"
 411                                      << qPrintable(prevbranch);
 412
 413                             Repository *repo = repositories.value(repository, 0);
 414                             if (!repo) {
 415                                 qCritical() << "Rule" << rule.rx.pattern() << "line" << rule.lineNumber
 416                                             << "references unknown repository" << repository;
 417                                 return EXIT_FAILURE;
 418                             }
 419
 420                             repo->createBranch(branch, revnum, prevbranch, rev_from);
 421                         }
 422                     }
 423                 }
 424
 425                 printf(".");
 426                 fflush(stdout);
 427 //                qDebug() << "   " << qPrintable(current) << "rev" << revnum << "->"
 428 //                         << qPrintable(repository) << qPrintable(branch) << qPrintable(path);
 429
 430                 Repository::Transaction *txn = transactions.value(repository, 0);
 431                 if (!txn) {
 432                     Repository *repo = repositories.value(repository, 0);
 433                     if (!repo) {
 434                         qCritical() << "Rule" << rule.rx.pattern() << "line" << rule.lineNumber
 435                                     << "references unknown repository" << repository;
 436                         return EXIT_FAILURE;
 437                     }
 438
 439                     txn = repo->newTransaction(branch, svnprefix, revnum);
 440                     if (!txn)
 441                         return EXIT_FAILURE;
 442
 443                     transactions.insert(repository, txn);
 444                 }
 445
 446                 if (change->change_kind == svn_fs_path_change_delete)
 447                     txn->deleteFile(path);
 448                 else if (!is_dir)
 449                     dumpBlob(txn, fs_root, key, path, revpool);
 450                 else
 451                     recursiveDumpDir(txn, fs_root, key, path, revpool);
 452
 453                 continue;
 454             }
 455         }
 456
 457         if (is_dir) {
 458             qDebug() << current << "is a new directory; ignoring";
 459         } else if (wasDir(fs, revnum - 1, key, pool)) {
 460             qDebug() << current << "was a directory; ignoring";
 461         } else if (change->change_kind == svn_fs_path_change_delete) {
 462             qDebug() << current << "is being deleted but I don't know anything about it; ignoring";
 463         } else {
 464             qCritical() << current << "did not match any rules; cannot continue";
 465             return EXIT_FAILURE;
 466         }
 467     }
 468     revpool.clear();
 469
 470     if (transactions.isEmpty()) {
 471         printf("nothing to do\n");
 472         return EXIT_SUCCESS;    // no changes?
 473     }
 474
 475     // now create the commit
 476     apr_hash_t *revprops;
 477     SVN_ERR(svn_fs_revision_proplist(&revprops, fs, revnum, pool));
 478     svn_string_t *svnauthor = (svn_string_t*)apr_hash_get(revprops, "svn:author", APR_HASH_KEY_STRING);
 479     svn_string_t *svndate = (svn_string_t*)apr_hash_get(revprops, "svn:date", APR_HASH_KEY_STRING);
 480     svn_string_t *svnlog = (svn_string_t*)apr_hash_get(revprops, "svn:log", APR_HASH_KEY_STRING);
 481
 482     QByteArray log = (char *)svnlog->data;
 483     QByteArray authorident = svnauthor ? identities.value((char *)svnauthor->data) : QByteArray();
 484     time_t epoch = get_epoch((char*)svndate->data);
 485     if (authorident.isEmpty()) {
 486         if (!svnauthor || svn_string_isempty(svnauthor))
 487             authorident = "nobody <nobody@localhost>";
 488         else
 489             authorident = svnauthor->data + QByteArray(" <") +
 490                           svnauthor->data + QByteArray("@localhost>");
 491     }
 492
 493     foreach (Repository::Transaction *txn, transactions) {
 494         txn->setAuthor(authorident);
 495         txn->setDateTime(epoch);
 496         txn->setLog(log);
 497
 498         txn->commit();
 499         delete txn;
 500     }
 501
 502     printf("done\n");
 503     return EXIT_SUCCESS;
 504 }