]> andersk Git - svn-all-fast-export.git/blobdiff - src/svn.cpp
Avoid ending slashes and mid double-slashes
[svn-all-fast-export.git] / src / svn.cpp
index 9e917abe1499fc62de369d9c06b039953b3b7cde..de3a9779ae0563674ce4df7ba199210988795533 100644 (file)
@@ -57,12 +57,15 @@ typedef QHash<QByteArray, QByteArray> IdentityHash;
 class AprAutoPool
 {
     apr_pool_t *pool;
+    AprAutoPool(const AprAutoPool &);
+    AprAutoPool &operator=(const AprAutoPool &);
 public:
     inline AprAutoPool(apr_pool_t *parent = NULL)
         { pool = svn_pool_create(parent); }
     inline ~AprAutoPool()
         { svn_pool_destroy(pool); }
 
+    inline void clear() { svn_pool_clear(pool); }
     inline apr_pool_t *data() const { return pool; }
     inline operator apr_pool_t *() const { return pool; }
 };
@@ -157,6 +160,51 @@ int SvnPrivate::openRepository(const QString &pathToRepository)
     return EXIT_SUCCESS;
 }
 
+enum RuleType { AnyRule = 0, NoIgnoreRule = 0x01 };
+
+static MatchRuleList::ConstIterator
+findMatchRule(const MatchRuleList &matchRules, int revnum, const QString &current,
+              int ruleMask = AnyRule)
+{
+    MatchRuleList::ConstIterator it = matchRules.constBegin(),
+                                end = matchRules.constEnd();
+    for ( ; it != end; ++it) {
+        if (it->minRevision > revnum)
+            continue;
+        if (it->maxRevision != -1 && it->maxRevision < revnum)
+            continue;
+        if (it->action == Rules::Match::Ignore && ruleMask & NoIgnoreRule)
+            continue;
+        if (it->rx.indexIn(current) == 0)
+            return it;
+    }
+
+    // no match
+    return end;
+}
+
+static void splitPathName(const Rules::Match &rule, const QString &pathName, QString *svnprefix_p,
+                          QString *repository_p, QString *branch_p, QString *path_p)
+{
+    QString svnprefix = pathName;
+    svnprefix.truncate(rule.rx.matchedLength());
+    if (svnprefix_p)
+        *svnprefix_p = svnprefix;
+
+    if (repository_p) {
+        *repository_p = svnprefix;
+        repository_p->replace(rule.rx, rule.repository);
+    }
+
+    if (branch_p) {
+        *branch_p = svnprefix;
+        branch_p->replace(rule.rx, rule.branch);
+    }
+
+    if (path_p)
+        *path_p = pathName.mid(svnprefix.length());
+}
+
 static int pathMode(svn_fs_root_t *fs_root, const char *pathname, apr_pool_t *pool)
 {
     svn_string_t *propvalue;
@@ -167,7 +215,7 @@ static int pathMode(svn_fs_root_t *fs_root, const char *pathname, apr_pool_t *po
 
     // maybe it's a symlink?
     SVN_ERR(svn_fs_node_prop(&propvalue, fs_root, pathname, "svn:special", pool));
-    if (strcmp(propvalue->data, "symlink") == 0)
+    if (propvalue && strcmp(propvalue->data, "symlink") == 0)
         mode = 0120000;
 
     return mode;
@@ -177,6 +225,9 @@ svn_error_t *QIODevice_write(void *baton, const char *data, apr_size_t *len)
 {
     QIODevice *device = reinterpret_cast<QIODevice *>(baton);
     device->write(data, *len);
+
+    if (device->bytesToWrite() > 16384)
+        device->waitForBytesWritten(0);
     return SVN_NO_ERROR;
 }
 
@@ -189,27 +240,81 @@ static svn_stream_t *streamForDevice(QIODevice *device, apr_pool_t *pool)
 }
 
 static int dumpBlob(Repository::Transaction *txn, svn_fs_root_t *fs_root,
-                    const char *pathname, apr_pool_t *pool)
+                    const char *pathname, const QString &finalPathName, apr_pool_t *pool)
 {
+    AprAutoPool dumppool(pool);
     // what type is it?
-    int mode = pathMode(fs_root, pathname, pool);
+    int mode = pathMode(fs_root, pathname, dumppool);
 
-    svn_stream_t *in_stream, *out_stream;
     svn_filesize_t stream_length;
 
-    SVN_ERR(svn_fs_file_length(&stream_length, fs_root, pathname, pool));
-    QIODevice *io = txn->addFile(pathname, mode, stream_length);
+    SVN_ERR(svn_fs_file_length(&stream_length, fs_root, pathname, dumppool));
+    QIODevice *io = txn->addFile(finalPathName, mode, stream_length);
 
+#ifndef DRY_RUN
     // open the file
-    SVN_ERR(svn_fs_file_contents(&in_stream, fs_root, pathname, pool));
+    svn_stream_t *in_stream, *out_stream;
+    SVN_ERR(svn_fs_file_contents(&in_stream, fs_root, pathname, dumppool));
 
     // open a generic svn_stream_t for the QIODevice
-    out_stream = streamForDevice(io, pool);
-    SVN_ERR(svn_stream_copy(in_stream, out_stream, pool));
+    out_stream = streamForDevice(io, dumppool);
+    SVN_ERR(svn_stream_copy(in_stream, out_stream, dumppool));
+
+    // print an ending newline
+    io->putChar('\n');
+#endif
 
     return EXIT_SUCCESS;
 }
 
+static int recursiveDumpDir(Repository::Transaction *txn, svn_fs_root_t *fs_root,
+                            const QByteArray &pathname, const QString &finalPathName,
+                            apr_pool_t *pool)
+{
+    // get the dir listing
+    apr_hash_t *entries;
+    SVN_ERR(svn_fs_dir_entries(&entries, fs_root, pathname, pool));
+    AprAutoPool dirpool(pool);
+
+    for (apr_hash_index_t *i = apr_hash_first(pool, entries); i; i = apr_hash_next(i)) {
+        dirpool.clear();
+        const void *vkey;
+        void *value;
+        apr_hash_this(i, &vkey, NULL, &value);
+
+        svn_fs_dirent_t *dirent = reinterpret_cast<svn_fs_dirent_t *>(value);
+        QByteArray entryName = pathname + '/' + dirent->name;
+        QString entryFinalName = finalPathName + dirent->name;
+
+        if (dirent->kind == svn_node_dir) {
+            entryFinalName += '/';
+            if (recursiveDumpDir(txn, fs_root, entryName, entryFinalName, dirpool) == EXIT_FAILURE)
+                return EXIT_FAILURE;
+        } else if (dirent->kind == svn_node_file) {
+            printf("+");
+            fflush(stdout);
+            if (dumpBlob(txn, fs_root, entryName, entryFinalName, dirpool) == EXIT_FAILURE)
+                return EXIT_FAILURE;
+        }
+    }
+
+    return EXIT_SUCCESS;
+}
+
+static bool wasDir(svn_fs_t *fs, int revnum, const char *pathname, apr_pool_t *pool)
+{
+    AprAutoPool subpool(pool);
+    svn_fs_root_t *fs_root;
+    if (svn_fs_revision_root(&fs_root, fs, revnum, subpool) != SVN_NO_ERROR)
+        return false;
+
+    svn_boolean_t is_dir;
+    if (svn_fs_is_dir(&is_dir, fs_root, pathname, subpool) != SVN_NO_ERROR)
+        return false;
+
+    return is_dir;
+}
+
 time_t get_epoch(char *svn_date)
 {
     struct tm tm;
@@ -219,92 +324,92 @@ time_t get_epoch(char *svn_date)
     return mktime(&tm);
 }
 
+class SvnRevision
+{
+public:
+    AprAutoPool pool;
+    QHash<QString, Repository::Transaction *> transactions;
+    MatchRuleList matchRules;
+    RepositoryHash repositories;
+    IdentityHash identities;
+
+    svn_fs_t *fs;
+    svn_fs_root_t *fs_root;
+    int revnum;
+
+    SvnRevision(int revision, svn_fs_t *f, apr_pool_t *parent_pool)
+        : pool(parent_pool), fs(f), fs_root(0), revnum(revision)
+    {
+    }
+
+    int open()
+    {
+        SVN_ERR(svn_fs_revision_root(&fs_root, fs, revnum, pool));
+        return EXIT_SUCCESS;
+    }
+
+    int prepareTransactions();
+    int commit();
+
+    int exportEntry(const char *path, const svn_fs_path_change_t *change, apr_hash_t *changes);
+    int exportInternal(const char *path, const svn_fs_path_change_t *change,
+                       const char *path_from, svn_revnum_t rev_from,
+                       const QString &current, const Rules::Match &rule);
+    int recurse(const char *path, const svn_fs_path_change_t *change,
+                const char *path_from, svn_revnum_t rev_from,
+                apr_hash_t *changes, apr_pool_t *pool);
+};
+
 int SvnPrivate::exportRevision(int revnum)
 {
-    AprAutoPool pool(global_pool);
+    SvnRevision rev(revnum, fs, global_pool);
+    rev.matchRules = matchRules;
+    rev.repositories = repositories;
+    rev.identities = identities;
 
     // open this revision:
-    svn_fs_root_t *fs_root;
-    SVN_ERR(svn_fs_revision_root(&fs_root, fs, revnum, pool));
-    qDebug() << "Exporting revision" << revnum;
+    printf("Exporting revision %d ", revnum);
+    fflush(stdout);
+
+    if (rev.open() == EXIT_FAILURE)
+        return EXIT_FAILURE;
+
+    if (rev.prepareTransactions() == EXIT_FAILURE)
+        return EXIT_FAILURE;
 
+    if (rev.transactions.isEmpty()) {
+        printf(" nothing to do\n");
+        return EXIT_SUCCESS;    // no changes?
+    }
+
+    if (rev.commit() == EXIT_FAILURE)
+        return EXIT_FAILURE;
+
+    printf(" done\n");
+    return EXIT_SUCCESS;
+}
+
+int SvnRevision::prepareTransactions()
+{
     // find out what was changed in this revision:
-    QHash<QString, Repository::Transaction *> transactions;
     apr_hash_t *changes;
     SVN_ERR(svn_fs_paths_changed(&changes, fs_root, pool));
-    AprAutoPool revpool(pool);
     for (apr_hash_index_t *i = apr_hash_first(pool, changes); i; i = apr_hash_next(i)) {
-        svn_pool_clear(revpool);
-
         const void *vkey;
         void *value;
         apr_hash_this(i, &vkey, NULL, &value);
         const char *key = reinterpret_cast<const char *>(vkey);
+        svn_fs_path_change_t *change = reinterpret_cast<svn_fs_path_change_t *>(value);
 
-        // is this a directory?
-        svn_boolean_t is_dir;
-        SVN_ERR(svn_fs_is_dir(&is_dir, fs_root, key, revpool));
-        if (is_dir)
-            continue;           // Git doesn't handle directories, so we don't either
-
-        QString current = QString::fromUtf8(key);
-
-        // find the first rule that matches this pathname
-        bool foundMatch = false;
-        foreach (Rules::Match rule, matchRules)
-            if (rule.rx.exactMatch(current)) {
-                foundMatch = true;
-                QString repository = current;
-                QString branch = current;
-                QString path = current;
-
-                // do the replacement
-                repository.replace(rule.rx, rule.repository);
-                branch.replace(rule.rx, rule.branch);
-                path.replace(rule.rx, rule.path);
-
-                qDebug() << "..." << current << "->"
-                         << repository << branch << path;
-
-                Repository::Transaction *txn = transactions.value(repository, 0);
-                if (!txn) {
-                    Repository *repo = repositories.value(repository, 0);
-                    if (!repo) {
-                        qCritical() << "Rule" << rule.rx.pattern()
-                                    << "references unknown repository" << repository;
-                        return EXIT_FAILURE;
-                    }
-
-                    QString svnprefix = current;
-                    if (current.endsWith(path))
-                        current.chop(path.length());
-
-                    txn = repo->newTransaction(branch, svnprefix, revnum);
-                    if (!txn)
-                        return EXIT_FAILURE;
-
-                    transactions.insert(repository, txn);
-                }
-
-                svn_fs_path_change_t *change = reinterpret_cast<svn_fs_path_change_t *>(value);
-                if (change->change_kind == svn_fs_path_change_delete)
-                    txn->deleteFile(path);
-                else
-                    dumpBlob(txn, fs_root, key, revpool);
-
-                break;
-            }
-
-        if (!foundMatch) {
-            qCritical() << current << "did not match any rules; cannot continue";
+        if (exportEntry(key, change, changes) == EXIT_FAILURE)
             return EXIT_FAILURE;
-        }
     }
-    svn_pool_clear(revpool);
 
-    if (transactions.isEmpty())
-        return true;            // no changes?
+    return EXIT_SUCCESS;
+}
 
+int SvnRevision::commit()
+{
     // now create the commit
     apr_hash_t *revprops;
     SVN_ERR(svn_fs_revision_proplist(&revprops, fs, revnum, pool));
@@ -313,7 +418,7 @@ int SvnPrivate::exportRevision(int revnum)
     svn_string_t *svnlog = (svn_string_t*)apr_hash_get(revprops, "svn:log", APR_HASH_KEY_STRING);
 
     QByteArray log = (char *)svnlog->data;
-    QByteArray authorident = identities.value((char *)svnauthor->data);
+    QByteArray authorident = svnauthor ? identities.value((char *)svnauthor->data) : QByteArray();
     time_t epoch = get_epoch((char*)svndate->data);
     if (authorident.isEmpty()) {
         if (!svnauthor || svn_string_isempty(svnauthor))
@@ -334,3 +439,192 @@ int SvnPrivate::exportRevision(int revnum)
 
     return EXIT_SUCCESS;
 }
+
+int SvnRevision::exportEntry(const char *key, const svn_fs_path_change_t *change,
+                             apr_hash_t *changes)
+{
+    AprAutoPool revpool(pool.data());
+    QString current = QString::fromUtf8(key);
+
+    // was this copied from somewhere?
+    svn_revnum_t rev_from;
+    const char *path_from;
+    SVN_ERR(svn_fs_copied_from(&rev_from, &path_from, fs_root, key, revpool));
+
+    // is this a directory?
+    svn_boolean_t is_dir;
+    SVN_ERR(svn_fs_is_dir(&is_dir, fs_root, key, revpool));
+    if (is_dir) {
+        if (path_from == NULL) {
+            // no, it's a new directory being added
+            // Git doesn't handle directories, so we don't either
+            //qDebug() << "   mkdir ignored:" << key;
+            return EXIT_SUCCESS;
+        }
+
+        current += '/';
+        qDebug() << "   " << key << "was copied from" << path_from;
+    }
+
+    // find the first rule that matches this pathname
+    MatchRuleList::ConstIterator match = findMatchRule(matchRules, revnum, current);
+    if (match != matchRules.constEnd()) {
+        const Rules::Match &rule = *match;
+        return exportInternal(key, change, path_from, rev_from, current, rule);
+    }
+
+    if (is_dir && path_from != NULL) {
+        qDebug() << current << "is a copy-with-history, auto-recursing";
+        return recurse(key, change, path_from, rev_from, changes, revpool);
+    } else if (wasDir(fs, revnum - 1, key, revpool)) {
+        qDebug() << current << "was a directory; ignoring";
+    } else if (change->change_kind == svn_fs_path_change_delete) {
+        qDebug() << current << "is being deleted but I don't know anything about it; ignoring";
+    } else {
+        qCritical() << current << "did not match any rules; cannot continue";
+        return EXIT_FAILURE;
+    }
+
+    return EXIT_SUCCESS;
+}
+
+int SvnRevision::exportInternal(const char *key, const svn_fs_path_change_t *change,
+                                const char *path_from, svn_revnum_t rev_from,
+                                const QString &current, const Rules::Match &rule)
+{
+    if (rule.action == Rules::Match::Ignore) {
+        // ignore rule
+        qDebug() << "   " << qPrintable(current) << "rev" << revnum
+                 << "-> ignored (rule" << rule << ")";
+        return EXIT_SUCCESS;
+    }
+
+    QString svnprefix, repository, branch, path;
+    splitPathName(rule, current, &svnprefix, &repository, &branch, &path);
+
+    printf(".");
+    fflush(stdout);
+//                qDebug() << "   " << qPrintable(current) << "rev" << revnum << "->"
+//                         << qPrintable(repository) << qPrintable(branch) << qPrintable(path);
+
+    if (path.isEmpty() && path_from != NULL) {
+        QString previous = QString::fromUtf8(path_from) + '/';
+        MatchRuleList::ConstIterator prevmatch =
+            findMatchRule(matchRules, rev_from, previous, NoIgnoreRule);
+        if (prevmatch != matchRules.constEnd()) {
+            QString prevsvnprefix, prevrepository, prevbranch, prevpath;
+            splitPathName(*prevmatch, previous, &prevsvnprefix, &prevrepository,
+                          &prevbranch, &prevpath);
+
+            if (!prevpath.isEmpty()) {
+                qDebug() << qPrintable(current) << "is a partial branch of repository"
+                         << qPrintable(prevrepository) << "branch"
+                         << qPrintable(prevbranch) << "subdir"
+                         << qPrintable(prevpath);
+            } else if (prevrepository != repository) {
+                qWarning() << qPrintable(current) << "rev" << revnum
+                           << "is a cross-repository copy (from repository"
+                           << qPrintable(prevrepository) << "branch"
+                           << qPrintable(prevbranch) << "path"
+                           << qPrintable(prevpath) << "rev" << rev_from << ")";
+            } else if (prevbranch == branch) {
+                // same branch and same repository
+                qDebug() << qPrintable(current) << "rev" << revnum
+                         << "is an SVN rename from"
+                         << qPrintable(previous) << "rev" << rev_from;
+                return EXIT_SUCCESS;
+            } else {
+                // same repository but not same branch
+                // this means this is a plain branch
+                qDebug() << qPrintable(repository) << ": branch"
+                         << qPrintable(branch) << "is branching from"
+                         << qPrintable(prevbranch);
+
+                Repository *repo = repositories.value(repository, 0);
+                if (!repo) {
+                    qCritical() << "Rule" << rule
+                                << "references unknown repository" << repository;
+                    return EXIT_FAILURE;
+                }
+
+                repo->createBranch(branch, revnum, prevbranch, rev_from);
+                return EXIT_SUCCESS;
+            }
+        }
+    }
+
+    Repository::Transaction *txn = transactions.value(repository, 0);
+    if (!txn) {
+        Repository *repo = repositories.value(repository, 0);
+        if (!repo) {
+            qCritical() << "Rule" << rule
+                        << "references unknown repository" << repository;
+            return EXIT_FAILURE;
+        }
+
+        txn = repo->newTransaction(branch, svnprefix, revnum);
+        if (!txn)
+            return EXIT_FAILURE;
+
+        transactions.insert(repository, txn);
+    }
+
+    if (change->change_kind == svn_fs_path_change_delete) {
+        txn->deleteFile(path);
+    } else if (!current.endsWith('/')) {
+        dumpBlob(txn, fs_root, key, path, pool);
+    } else {
+        QString pathNoSlash = path;
+        pathNoSlash.chop(1);
+        txn->deleteFile(pathNoSlash);
+        recursiveDumpDir(txn, fs_root, key, path, pool);
+    }
+
+    return EXIT_SUCCESS;
+}
+
+int SvnRevision::recurse(const char *path, const svn_fs_path_change_t *change,
+                         const char *path_from, svn_revnum_t rev_from,
+                         apr_hash_t *changes, apr_pool_t *pool)
+{
+    // get the dir listing
+    apr_hash_t *entries;
+    SVN_ERR(svn_fs_dir_entries(&entries, fs_root, path, pool));
+
+    AprAutoPool dirpool(pool);
+    for (apr_hash_index_t *i = apr_hash_first(pool, entries); i; i = apr_hash_next(i)) {
+        dirpool.clear();
+        const void *vkey;
+        void *value;
+        apr_hash_this(i, &vkey, NULL, &value);
+
+        svn_fs_dirent_t *dirent = reinterpret_cast<svn_fs_dirent_t *>(value);
+        QByteArray entry = path + QByteArray("/") + dirent->name;
+        QByteArray entryFrom;
+        if (path_from)
+            entryFrom = path_from + QByteArray("/") + dirent->name;
+
+        // check if this entry is in the changelist for this revision already
+        if (apr_hash_get(changes, entry.constData(), APR_HASH_KEY_STRING)) {
+            qDebug() << entry << "rev" << revnum
+                     << "is in the change-list, deferring to that one";
+            continue;
+        }
+
+        QString current = QString::fromUtf8(entry);
+        if (dirent->kind == svn_node_dir)
+            current += '/';
+
+        // find the first rule that matches this pathname
+        MatchRuleList::ConstIterator match = findMatchRule(matchRules, revnum, current);
+        if (match != matchRules.constEnd()) {
+            if (exportInternal(entry, change, entryFrom.isNull() ? 0 : entryFrom.constData(),
+                               rev_from, current, *match) == EXIT_FAILURE)
+                return EXIT_FAILURE;
+        } else {
+            qCritical() << current << "rev" << revnum
+                        << "did not match any rules; cannot continue";
+            return EXIT_FAILURE;
+        }
+    }
+}
This page took 0.059326 seconds and 4 git commands to generate.