Add a warning about tag processing in the standardlayout file
[svn-all-fast-export.git] / src / svn.cpp
blob489ef141991d3fe87d3a2049f48af99a01661cd0
1 /*
2 * Copyright (C) 2007 Thiago Macieira <thiago@kde.org>
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.
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.
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/>.
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
24 #define _XOPEN_SOURCE
25 #define _LARGEFILE_SUPPORT
26 #define _LARGEFILE64_SUPPORT
28 #include "svn.h"
30 #include <unistd.h>
31 #include <string.h>
32 #include <stdio.h>
33 #include <time.h>
34 #include <unistd.h>
36 #include <apr_lib.h>
37 #include <apr_getopt.h>
38 #include <apr_general.h>
40 #include <svn_fs.h>
41 #include <svn_pools.h>
42 #include <svn_repos.h>
43 #include <svn_types.h>
45 #include <QFile>
46 #include <QDebug>
48 #include "repository.h"
50 #undef SVN_ERR
51 #define SVN_ERR(expr) SVN_INT_ERR(expr)
53 typedef QList<Rules::Match> MatchRuleList;
54 typedef QHash<QString, Repository *> RepositoryHash;
55 typedef QHash<QByteArray, QByteArray> IdentityHash;
57 class AprAutoPool
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); }
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; }
73 class SvnPrivate
75 public:
76 MatchRuleList matchRules;
77 RepositoryHash repositories;
78 IdentityHash identities;
80 SvnPrivate(const QString &pathToRepository);
81 ~SvnPrivate();
82 int youngestRevision();
83 int exportRevision(int revnum);
85 int openRepository(const QString &pathToRepository);
87 private:
88 AprAutoPool global_pool;
89 svn_fs_t *fs;
90 svn_revnum_t youngest_rev;
93 void Svn::initialize()
95 // initialize APR or exit
96 if (apr_initialize() != APR_SUCCESS) {
97 fprintf(stderr, "You lose at apr_initialize().\n");
98 exit(1);
101 // static destructor
102 static struct Destructor { ~Destructor() { apr_terminate(); } } destructor;
105 Svn::Svn(const QString &pathToRepository)
106 : d(new SvnPrivate(pathToRepository))
110 Svn::~Svn()
112 delete d;
115 void Svn::setMatchRules(const MatchRuleList &matchRules)
117 d->matchRules = matchRules;
120 void Svn::setRepositories(const RepositoryHash &repositories)
122 d->repositories = repositories;
125 void Svn::setIdentityMap(const IdentityHash &identityMap)
127 d->identities = identityMap;
130 int Svn::youngestRevision()
132 return d->youngestRevision();
135 bool Svn::exportRevision(int revnum)
137 return d->exportRevision(revnum) == EXIT_SUCCESS;
140 SvnPrivate::SvnPrivate(const QString &pathToRepository)
141 : global_pool(NULL)
143 openRepository(pathToRepository);
145 // get the youngest revision
146 svn_fs_youngest_rev(&youngest_rev, fs, global_pool);
149 SvnPrivate::~SvnPrivate()
151 svn_pool_destroy(global_pool);
154 int SvnPrivate::youngestRevision()
156 return youngest_rev;
159 int SvnPrivate::openRepository(const QString &pathToRepository)
161 svn_repos_t *repos;
162 SVN_ERR(svn_repos_open(&repos, QFile::encodeName(pathToRepository), global_pool));
163 fs = svn_repos_fs(repos);
165 return EXIT_SUCCESS;
168 enum RuleType { AnyRule = 0, NoIgnoreRule = 0x01, NoRecurseRule = 0x02 };
170 static MatchRuleList::ConstIterator
171 findMatchRule(const MatchRuleList &matchRules, int revnum, const QString &current,
172 int ruleMask = AnyRule)
174 MatchRuleList::ConstIterator it = matchRules.constBegin(),
175 end = matchRules.constEnd();
176 for ( ; it != end; ++it) {
177 if (it->minRevision > revnum)
178 continue;
179 if (it->maxRevision != -1 && it->maxRevision < revnum)
180 continue;
181 if (it->action == Rules::Match::Ignore && ruleMask & NoIgnoreRule)
182 continue;
183 if (it->action == Rules::Match::Recurse && ruleMask & NoRecurseRule)
184 continue;
185 if (it->rx.indexIn(current) == 0)
186 return it;
189 // no match
190 return end;
193 static void splitPathName(const Rules::Match &rule, const QString &pathName, QString *svnprefix_p,
194 QString *repository_p, QString *branch_p, QString *path_p)
196 QString svnprefix = pathName;
197 svnprefix.truncate(rule.rx.matchedLength());
198 if (svnprefix_p)
199 *svnprefix_p = svnprefix;
201 if (repository_p) {
202 *repository_p = svnprefix;
203 repository_p->replace(rule.rx, rule.repository);
206 if (branch_p) {
207 *branch_p = svnprefix;
208 branch_p->replace(rule.rx, rule.branch);
211 if (path_p)
212 *path_p = pathName.mid(svnprefix.length());
215 static int pathMode(svn_fs_root_t *fs_root, const char *pathname, apr_pool_t *pool)
217 svn_string_t *propvalue;
218 SVN_ERR(svn_fs_node_prop(&propvalue, fs_root, pathname, "svn:executable", pool));
219 int mode = 0100644;
220 if (propvalue)
221 mode = 0100755;
223 return mode;
226 svn_error_t *QIODevice_write(void *baton, const char *data, apr_size_t *len)
228 QIODevice *device = reinterpret_cast<QIODevice *>(baton);
229 device->write(data, *len);
231 while (device->bytesToWrite() > 32*1024) {
232 if (!device->waitForBytesWritten(-1)) {
233 qFatal("Failed to write to process: %s", qPrintable(device->errorString()));
234 return svn_error_createf(APR_EOF, SVN_NO_ERROR, "Failed to write to process: %s",
235 qPrintable(device->errorString()));
238 return SVN_NO_ERROR;
241 static svn_stream_t *streamForDevice(QIODevice *device, apr_pool_t *pool)
243 svn_stream_t *stream = svn_stream_create(device, pool);
244 svn_stream_set_write(stream, QIODevice_write);
246 return stream;
249 static int dumpBlob(Repository::Transaction *txn, svn_fs_root_t *fs_root,
250 const char *pathname, const QString &finalPathName, apr_pool_t *pool)
252 AprAutoPool dumppool(pool);
253 // what type is it?
254 int mode = pathMode(fs_root, pathname, dumppool);
256 svn_filesize_t stream_length;
258 SVN_ERR(svn_fs_file_length(&stream_length, fs_root, pathname, dumppool));
260 #ifndef DRY_RUN
261 // open the file
262 svn_stream_t *in_stream, *out_stream;
263 SVN_ERR(svn_fs_file_contents(&in_stream, fs_root, pathname, dumppool));
264 #endif
266 // maybe it's a symlink?
267 svn_string_t *propvalue;
268 SVN_ERR(svn_fs_node_prop(&propvalue, fs_root, pathname, "svn:special", dumppool));
269 if (propvalue) {
270 apr_size_t len = strlen("link ");
271 #ifndef DRY_RUN
272 QByteArray buf;
273 buf.reserve(len);
274 SVN_ERR(svn_stream_read(in_stream, buf.data(), &len));
275 if (len != strlen("link ") || strncmp(buf, "link ", len) != 0)
276 qFatal("file %s is svn:special but not a symlink", pathname);
277 #endif
278 mode = 0120000;
279 stream_length -= len;
282 QIODevice *io = txn->addFile(finalPathName, mode, stream_length);
284 #ifndef DRY_RUN
285 // open a generic svn_stream_t for the QIODevice
286 out_stream = streamForDevice(io, dumppool);
287 SVN_ERR(svn_stream_copy(in_stream, out_stream, dumppool));
288 svn_stream_close(out_stream);
289 svn_stream_close(in_stream);
291 // print an ending newline
292 io->putChar('\n');
293 #endif
295 return EXIT_SUCCESS;
298 static int recursiveDumpDir(Repository::Transaction *txn, svn_fs_root_t *fs_root,
299 const QByteArray &pathname, const QString &finalPathName,
300 apr_pool_t *pool)
302 // get the dir listing
303 apr_hash_t *entries;
304 SVN_ERR(svn_fs_dir_entries(&entries, fs_root, pathname, pool));
305 AprAutoPool dirpool(pool);
307 for (apr_hash_index_t *i = apr_hash_first(pool, entries); i; i = apr_hash_next(i)) {
308 dirpool.clear();
309 const void *vkey;
310 void *value;
311 apr_hash_this(i, &vkey, NULL, &value);
313 svn_fs_dirent_t *dirent = reinterpret_cast<svn_fs_dirent_t *>(value);
314 QByteArray entryName = pathname + '/' + dirent->name;
315 QString entryFinalName = finalPathName + dirent->name;
317 if (dirent->kind == svn_node_dir) {
318 entryFinalName += '/';
319 if (recursiveDumpDir(txn, fs_root, entryName, entryFinalName, dirpool) == EXIT_FAILURE)
320 return EXIT_FAILURE;
321 } else if (dirent->kind == svn_node_file) {
322 printf("+");
323 fflush(stdout);
324 if (dumpBlob(txn, fs_root, entryName, entryFinalName, dirpool) == EXIT_FAILURE)
325 return EXIT_FAILURE;
329 return EXIT_SUCCESS;
332 static bool wasDir(svn_fs_t *fs, int revnum, const char *pathname, apr_pool_t *pool)
334 AprAutoPool subpool(pool);
335 svn_fs_root_t *fs_root;
336 if (svn_fs_revision_root(&fs_root, fs, revnum, subpool) != SVN_NO_ERROR)
337 return false;
339 svn_boolean_t is_dir;
340 if (svn_fs_is_dir(&is_dir, fs_root, pathname, subpool) != SVN_NO_ERROR)
341 return false;
343 return is_dir;
346 time_t get_epoch(char *svn_date)
348 struct tm tm;
349 memset(&tm, 0, sizeof tm);
350 QByteArray date(svn_date, strlen(svn_date) - 8);
351 strptime(date, "%Y-%m-%dT%H:%M:%S", &tm);
352 return timegm(&tm);
355 class SvnRevision
357 public:
358 AprAutoPool pool;
359 QHash<QString, Repository::Transaction *> transactions;
360 MatchRuleList matchRules;
361 RepositoryHash repositories;
362 IdentityHash identities;
364 svn_fs_t *fs;
365 svn_fs_root_t *fs_root;
366 int revnum;
368 SvnRevision(int revision, svn_fs_t *f, apr_pool_t *parent_pool)
369 : pool(parent_pool), fs(f), fs_root(0), revnum(revision)
373 int open()
375 SVN_ERR(svn_fs_revision_root(&fs_root, fs, revnum, pool));
376 return EXIT_SUCCESS;
379 int prepareTransactions();
380 int commit();
382 int exportEntry(const char *path, const svn_fs_path_change_t *change, apr_hash_t *changes);
383 int exportDispatch(const char *path, const svn_fs_path_change_t *change,
384 const char *path_from, svn_revnum_t rev_from,
385 apr_hash_t *changes, const QString &current, const Rules::Match &rule,
386 apr_pool_t *pool);
387 int exportInternal(const char *path, const svn_fs_path_change_t *change,
388 const char *path_from, svn_revnum_t rev_from,
389 const QString &current, const Rules::Match &rule);
390 int recurse(const char *path, const svn_fs_path_change_t *change,
391 const char *path_from, svn_revnum_t rev_from,
392 apr_hash_t *changes, apr_pool_t *pool);
395 int SvnPrivate::exportRevision(int revnum)
397 SvnRevision rev(revnum, fs, global_pool);
398 rev.matchRules = matchRules;
399 rev.repositories = repositories;
400 rev.identities = identities;
402 // open this revision:
403 printf("Exporting revision %d ", revnum);
404 fflush(stdout);
406 if (rev.open() == EXIT_FAILURE)
407 return EXIT_FAILURE;
409 if (rev.prepareTransactions() == EXIT_FAILURE)
410 return EXIT_FAILURE;
412 if (rev.transactions.isEmpty()) {
413 printf(" nothing to do\n");
414 return EXIT_SUCCESS; // no changes?
417 if (rev.commit() == EXIT_FAILURE)
418 return EXIT_FAILURE;
420 printf(" done\n");
421 return EXIT_SUCCESS;
424 int SvnRevision::prepareTransactions()
426 // find out what was changed in this revision:
427 apr_hash_t *changes;
428 SVN_ERR(svn_fs_paths_changed(&changes, fs_root, pool));
429 for (apr_hash_index_t *i = apr_hash_first(pool, changes); i; i = apr_hash_next(i)) {
430 const void *vkey;
431 void *value;
432 apr_hash_this(i, &vkey, NULL, &value);
433 const char *key = reinterpret_cast<const char *>(vkey);
434 svn_fs_path_change_t *change = reinterpret_cast<svn_fs_path_change_t *>(value);
436 if (exportEntry(key, change, changes) == EXIT_FAILURE)
437 return EXIT_FAILURE;
440 return EXIT_SUCCESS;
443 int SvnRevision::commit()
445 // now create the commit
446 apr_hash_t *revprops;
447 SVN_ERR(svn_fs_revision_proplist(&revprops, fs, revnum, pool));
448 svn_string_t *svnauthor = (svn_string_t*)apr_hash_get(revprops, "svn:author", APR_HASH_KEY_STRING);
449 svn_string_t *svndate = (svn_string_t*)apr_hash_get(revprops, "svn:date", APR_HASH_KEY_STRING);
450 svn_string_t *svnlog = (svn_string_t*)apr_hash_get(revprops, "svn:log", APR_HASH_KEY_STRING);
452 QByteArray log = (char *)svnlog->data;
453 QByteArray authorident = svnauthor ? identities.value((char *)svnauthor->data) : QByteArray();
454 time_t epoch = get_epoch((char*)svndate->data);
455 if (authorident.isEmpty()) {
456 if (!svnauthor || svn_string_isempty(svnauthor))
457 authorident = "nobody <nobody@localhost>";
458 else
459 authorident = svnauthor->data + QByteArray(" <") +
460 svnauthor->data + QByteArray("@localhost>");
463 foreach (Repository::Transaction *txn, transactions) {
464 txn->setAuthor(authorident);
465 txn->setDateTime(epoch);
466 txn->setLog(log);
468 txn->commit();
469 delete txn;
472 return EXIT_SUCCESS;
475 int SvnRevision::exportEntry(const char *key, const svn_fs_path_change_t *change,
476 apr_hash_t *changes)
478 AprAutoPool revpool(pool.data());
479 QString current = QString::fromUtf8(key);
481 // was this copied from somewhere?
482 svn_revnum_t rev_from;
483 const char *path_from;
484 SVN_ERR(svn_fs_copied_from(&rev_from, &path_from, fs_root, key, revpool));
486 // is this a directory?
487 svn_boolean_t is_dir;
488 SVN_ERR(svn_fs_is_dir(&is_dir, fs_root, key, revpool));
489 if (is_dir) {
490 if (path_from == NULL) {
491 // no, it's a new directory being added
492 // Git doesn't handle directories, so we don't either
493 //qDebug() << " mkdir ignored:" << key;
494 return EXIT_SUCCESS;
497 current += '/';
498 qDebug() << " " << key << "was copied from" << path_from << "rev" << rev_from;
501 // find the first rule that matches this pathname
502 MatchRuleList::ConstIterator match = findMatchRule(matchRules, revnum, current);
503 if (match != matchRules.constEnd()) {
504 const Rules::Match &rule = *match;
505 return exportDispatch(key, change, path_from, rev_from, changes, current, rule, revpool);
508 if (is_dir && path_from != NULL) {
509 qDebug() << current << "is a copy-with-history, auto-recursing";
510 return recurse(key, change, path_from, rev_from, changes, revpool);
511 } else if (wasDir(fs, revnum - 1, key, revpool)) {
512 qDebug() << current << "was a directory; ignoring";
513 } else if (change->change_kind == svn_fs_path_change_delete) {
514 qDebug() << current << "is being deleted but I don't know anything about it; ignoring";
515 } else {
516 qCritical() << current << "did not match any rules; cannot continue";
517 return EXIT_FAILURE;
520 return EXIT_SUCCESS;
523 int SvnRevision::exportDispatch(const char *key, const svn_fs_path_change_t *change,
524 const char *path_from, svn_revnum_t rev_from,
525 apr_hash_t *changes, const QString &current,
526 const Rules::Match &rule, apr_pool_t *pool)
528 switch (rule.action) {
529 case Rules::Match::Ignore:
530 // ignore rule
531 //qDebug() << " " << qPrintable(current) << "rev" << revnum
532 // << "-> ignored (rule" << rule << ")";
533 return EXIT_SUCCESS;
535 case Rules::Match::Recurse:
536 return recurse(key, change, path_from, rev_from, changes, pool);
538 case Rules::Match::Export:
539 return exportInternal(key, change, path_from, rev_from, current, rule);
542 // never reached
543 return EXIT_FAILURE;
546 int SvnRevision::exportInternal(const char *key, const svn_fs_path_change_t *change,
547 const char *path_from, svn_revnum_t rev_from,
548 const QString &current, const Rules::Match &rule)
550 QString svnprefix, repository, branch, path;
551 splitPathName(rule, current, &svnprefix, &repository, &branch, &path);
553 printf(".");
554 fflush(stdout);
555 // qDebug() << " " << qPrintable(current) << "rev" << revnum << "->"
556 // << qPrintable(repository) << qPrintable(branch) << qPrintable(path);
558 if (path.isEmpty() && path_from != NULL) {
559 QString previous = QString::fromUtf8(path_from) + '/';
560 MatchRuleList::ConstIterator prevmatch =
561 findMatchRule(matchRules, rev_from, previous, NoIgnoreRule);
562 if (prevmatch != matchRules.constEnd()) {
563 QString prevsvnprefix, prevrepository, prevbranch, prevpath;
564 splitPathName(*prevmatch, previous, &prevsvnprefix, &prevrepository,
565 &prevbranch, &prevpath);
567 if (!prevpath.isEmpty()) {
568 qDebug() << qPrintable(current) << "is a partial branch of repository"
569 << qPrintable(prevrepository) << "branch"
570 << qPrintable(prevbranch) << "subdir"
571 << qPrintable(prevpath);
572 } else if (prevrepository != repository) {
573 qWarning() << qPrintable(current) << "rev" << revnum
574 << "is a cross-repository copy (from repository"
575 << qPrintable(prevrepository) << "branch"
576 << qPrintable(prevbranch) << "path"
577 << qPrintable(prevpath) << "rev" << rev_from << ")";
578 } else if (prevbranch == branch) {
579 // same branch and same repository
580 qDebug() << qPrintable(current) << "rev" << revnum
581 << "is an SVN rename from"
582 << qPrintable(previous) << "rev" << rev_from;
583 return EXIT_SUCCESS;
584 } else {
585 // same repository but not same branch
586 // this means this is a plain branch
587 qDebug() << qPrintable(repository) << ": branch"
588 << qPrintable(branch) << "is branching from"
589 << qPrintable(prevbranch);
591 Repository *repo = repositories.value(repository, 0);
592 if (!repo) {
593 qCritical() << "Rule" << rule
594 << "references unknown repository" << repository;
595 return EXIT_FAILURE;
598 repo->createBranch(branch, revnum, prevbranch, rev_from);
599 return EXIT_SUCCESS;
604 Repository::Transaction *txn = transactions.value(repository + branch, 0);
605 if (!txn) {
606 Repository *repo = repositories.value(repository, 0);
607 if (!repo) {
608 qCritical() << "Rule" << rule
609 << "references unknown repository" << repository;
610 return EXIT_FAILURE;
613 txn = repo->newTransaction(branch, svnprefix, revnum);
614 if (!txn)
615 return EXIT_FAILURE;
617 transactions.insert(repository + branch, txn);
620 if (change->change_kind == svn_fs_path_change_delete) {
621 txn->deleteFile(path);
622 } else if (!current.endsWith('/')) {
623 dumpBlob(txn, fs_root, key, path, pool);
624 } else {
625 QString pathNoSlash = path;
626 pathNoSlash.chop(1);
627 txn->deleteFile(pathNoSlash);
628 recursiveDumpDir(txn, fs_root, key, path, pool);
631 return EXIT_SUCCESS;
634 int SvnRevision::recurse(const char *path, const svn_fs_path_change_t *change,
635 const char *path_from, svn_revnum_t rev_from,
636 apr_hash_t *changes, apr_pool_t *pool)
638 // get the dir listing
639 apr_hash_t *entries;
640 SVN_ERR(svn_fs_dir_entries(&entries, fs_root, path, pool));
642 AprAutoPool dirpool(pool);
643 for (apr_hash_index_t *i = apr_hash_first(pool, entries); i; i = apr_hash_next(i)) {
644 dirpool.clear();
645 const void *vkey;
646 void *value;
647 apr_hash_this(i, &vkey, NULL, &value);
649 svn_fs_dirent_t *dirent = reinterpret_cast<svn_fs_dirent_t *>(value);
650 if (dirent->kind != svn_node_dir)
651 continue; // not a directory, so can't recurse; skip
653 QByteArray entry = path + QByteArray("/") + dirent->name;
654 QByteArray entryFrom;
655 if (path_from)
656 entryFrom = path_from + QByteArray("/") + dirent->name;
658 // check if this entry is in the changelist for this revision already
659 if (apr_hash_get(changes, entry.constData(), APR_HASH_KEY_STRING)) {
660 qDebug() << entry << "rev" << revnum
661 << "is in the change-list, deferring to that one";
662 continue;
665 QString current = QString::fromUtf8(entry);
666 if (dirent->kind == svn_node_dir)
667 current += '/';
669 // find the first rule that matches this pathname
670 MatchRuleList::ConstIterator match = findMatchRule(matchRules, revnum, current);
671 if (match != matchRules.constEnd()) {
672 if (exportDispatch(entry, change, entryFrom.isNull() ? 0 : entryFrom.constData(),
673 rev_from, changes, current, *match, dirpool) == EXIT_FAILURE)
674 return EXIT_FAILURE;
675 } else {
676 qCritical() << current << "rev" << revnum
677 << "did not match any rules; cannot continue";
678 return EXIT_FAILURE;
682 return EXIT_SUCCESS;