Adding initial rev of my svn-fast-export app, in Python and C.

Makefile

@@ -0,0 +1,5 @@
+SVN = /usr/local/svn
+CFLAGS = -I/usr/include/apr-1.0 -I${SVN}/include/subversion-1 -pipe -g3 -std=c99
+LDFLAGS = -L${SVN}/lib -lsvn_repos-1
+
+svn-fast-export: svn-fast-export.c

svn-fast-export.c

@@ -0,0 +1,169 @@
+/*
+ * svn-fast-export.c
+ * ----------
+ *  Walk through each revision of a local Subversion repository and export it
+ *  in a stream that git-fast-import can consume.
+ *
+ * Author: Chris Lee <clee@kde.org>
+ * License: MIT <http://www.opensource.org/licenses/mit-license.php>
+ */
+
+#include <unistd.h>
+#include <string.h>
+#include <stdio.h>
+
+#ifndef PATH_MAX
+#define PATH_MAX 4096
+#endif
+
+#include <apr_general.h>
+#include <apr_lib.h>
+#include <apr_getopt.h>
+
+#include <svn_types.h>
+#include <svn_pools.h>
+#include <svn_repos.h>
+
+#undef SVN_ERR
+#define SVN_ERR(expr) SVN_INT_ERR(expr)
+#define apr_sane_push(arr, contents) *(char **)apr_array_push(arr) = contents
+
+#define TRUNK "/trunk/"
+
+int dump_blob(svn_fs_root_t *root, char *full_path, apr_pool_t *pool)
+{
+    svn_filesize_t stream_length;
+    svn_stream_t *stream;
+    apr_size_t len;
+    char buf[8];
+
+    SVN_ERR(svn_fs_file_length(&stream_length, root, full_path, pool));
+    SVN_ERR(svn_fs_file_contents(&stream, root, full_path, pool));
+
+    fprintf(stdout, "data %li\n", stream_length);
+
+    do {
+        len = sizeof(buf);
+        SVN_ERR(svn_stream_read(stream, buf, &len));
+        fprintf (stdout, "%s\0", buf);
+    } while (len);
+
+    fprintf(stdout, "\n");
+
+    return 0;
+}
+
+int export_revision(svn_revnum_t rev, svn_repos_t *repo, svn_fs_t *fs, apr_pool_t *pool)
+{
+    apr_array_header_t *file_changes;
+    apr_hash_t *changes, *props;
+    apr_hash_index_t *i;
+    apr_pool_t *revpool;
+
+    svn_fs_path_change_t *change;
+    svn_fs_root_t *root_obj;
+    svn_boolean_t is_dir;
+
+    char *path, *file_change;
+    unsigned int mark;
+    const void *key;
+    void *val;
+ 
+    fprintf(stderr, "Committing revision %li... ", rev);
+
+    SVN_ERR(svn_fs_revision_root(&root_obj, fs, rev, pool));
+    SVN_ERR(svn_fs_paths_changed(&changes, root_obj, pool));
+    SVN_ERR(svn_fs_revision_proplist(&props, fs, rev, pool));
+
+    revpool = svn_pool_create(pool);
+
+    file_changes = apr_array_make(pool, apr_hash_count(changes), sizeof(char *));
+    mark = 1;
+    for (i = apr_hash_first(pool, changes); i; i = apr_hash_next(i)) {
+        svn_pool_clear(revpool);
+        apr_hash_this(i, &key, NULL, &val);
+        path = (char *)key;
+        change = (svn_fs_path_change_t *)val;
+
+        SVN_ERR(svn_fs_is_dir(&is_dir, root_obj, path, revpool));
+
+        if (is_dir || strncmp(TRUNK, path, strlen(TRUNK))) {
+            continue;
+        }
+
+        if (change->change_kind == svn_fs_path_change_delete) {
+            fprintf(stderr, "       delete\n");
+            *(char **)apr_array_push(file_changes) = ((char *)svn_string_createf(pool, "D %s", path + strlen(TRUNK))->data);
+            fprintf(stderr, "       done\n");
+        } else {
+            *(char **)apr_array_push(file_changes) = (char *)svn_string_createf(pool, "M 644 :%u %s", mark, path + strlen(TRUNK))->data;
+            fprintf(stdout, "blob\nmark :%u\n", mark++);
+            // dump_blob(root_obj, (char *)path, revpool);
+        }
+    }
+
+    if (file_changes->nelts == 0) {
+        fprintf(stderr, "skipping.\n");
+        svn_pool_destroy(revpool);
+        return 0;
+    }
+
+    fprintf(stdout, "commit refs/heads/master\n");
+    fprintf(stdout, apr_array_pstrcat(pool, file_changes, '\n'));
+    fprintf(stdout, "\n\n");
+
+    svn_pool_destroy(revpool);
+
+    fprintf(stderr, "done!\n");
+
+    return 0;
+}
+
+int crawl_revisions(char *repos_path)
+{
+    apr_pool_t *pool, *subpool;
+    svn_repos_t *repos;
+    svn_revnum_t youngest_rev, min_rev, max_rev, rev;
+    svn_fs_t *fs;
+
+    pool = svn_pool_create(NULL);
+
+    SVN_ERR(svn_repos_open(&repos, repos_path, pool));
+
+    fs = svn_repos_fs(repos);
+
+    SVN_ERR(svn_fs_initialize(pool));
+    SVN_ERR(svn_fs_youngest_rev(&youngest_rev, fs, pool));
+
+    min_rev = 1;
+    max_rev = youngest_rev;
+
+    subpool = svn_pool_create(pool);
+    for (rev = min_rev; rev <= max_rev; rev++) {
+        svn_pool_clear(subpool);
+        export_revision(rev, repos, fs, subpool);
+    }
+
+    svn_pool_destroy(pool);
+
+    return 0;
+}
+
+int main(int argc, char *argv[])
+{
+    if (argc != 2) {
+        fprintf(stderr, "usage: %s REPOS_PATH\n", argv[0]);
+        return -1;
+    }
+
+    if (apr_initialize() != APR_SUCCESS) {
+        fprintf(stderr, "You lose at apr_initialize().\n");
+        return -1;
+    }
+
+    crawl_revisions(argv[1]);
+
+    apr_terminate();
+
+    return 0;
+}

svn-fast-export.py

@@ -0,0 +1,167 @@
+#!/usr/bin/python
+#
+# svn-fast-export.py
+# ----------
+#  Walk through each revision of a local Subversion repository and export it
+#  in a stream that git-fast-import can consume.
+#
+# Author: Chris Lee <clee@kde.org>
+# License: MIT <http://www.opensource.org/licenses/mit-license.php>
+
+trunk_path = '/trunk/'
+branches_path = '/branches/'
+tags_path = '/tags/'
+
+first_rev = 1
+final_rev = 0
+
+import gc, sys, os.path
+from optparse import OptionParser
+from time import sleep, mktime, localtime, strftime, strptime
+from svn.fs import svn_fs_dir_entries, svn_fs_file_length, svn_fs_file_contents, svn_fs_is_dir, svn_fs_revision_root, svn_fs_youngest_rev, svn_fs_revision_proplist, svn_fs_revision_prop, svn_fs_paths_changed
+from svn.core import svn_pool_create, svn_pool_clear, svn_pool_destroy, svn_stream_read, svn_stream_close, run_app
+from svn.repos import svn_repos_open, svn_repos_fs
+
+ct_short = ['M', 'A', 'D', 'R', 'X']
+
+def dump_file_blob(root, full_path, pool):
+    # Use an iteration subpool.
+    subpool = svn_pool_create(pool)
+
+    # Clear the iteration subpool.
+    svn_pool_clear(subpool)
+
+    # print full_path
+    stream_length = svn_fs_file_length(root, full_path, subpool)
+    stream = svn_fs_file_contents(root, full_path, subpool)
+    contents = svn_stream_read(stream, int(stream_length))
+    sys.stdout.write("data %s\n" % stream_length)
+    sys.stdout.write(contents)
+    sys.stdout.write("\n")
+
+    # Destroy the iteration subpool.
+    svn_pool_destroy(subpool)
+
+
+def export_revision(rev, repo, fs, pool):
+    sys.stderr.write("Committing revision %s... " % rev)
+
+    revpool = svn_pool_create(pool)
+    svn_pool_clear(revpool)
+
+    # Open a root object representing the youngest (HEAD) revision.
+    root = svn_fs_revision_root(fs, rev, revpool)
+
+    # And the list of what changed in this revision.
+    changes = svn_fs_paths_changed(root, revpool)
+
+    i = 1
+    marks = {}
+    file_changes = []
+
+    for path, change_type in changes.iteritems():
+        c_t = ct_short[change_type.change_kind]
+        if svn_fs_is_dir(root, path, revpool):
+            continue
+
+        if not path.startswith(trunk_path):
+            # We don't handle branches. Or tags. Yet.
+            pass
+        else:
+            if c_t == 'D':
+                file_changes.append("D %s" % path.replace(trunk_path, ''))
+            else:
+                sys.stdout.write("blob\nmark :%s\n" % i)
+                marks[i] = path.replace(trunk_path, '')
+                # dump_file_blob(root_obj, path, revpool)
+                file_changes.append("M 644 :%s %s" % (i, marks[i]))
+                i += 1
+
+    # Get the commit author and message
+    props = svn_fs_revision_proplist(fs, rev, revpool)
+
+    # Do the recursive crawl.
+    if props.has_key('svn:author'):
+        author = "%s <%s@localhost>" % (props['svn:author'], props['svn:author'])
+    else:
+        author = 'nobody <nobody@localhost>'
+
+    if len(file_changes) == 0:
+        svn_pool_destroy(revpool)
+        sys.stderr.write("skipping.\n")
+        return
+
+    svndate = props['svn:date'][0:-8]
+    commit_time = mktime(strptime(svndate, '%Y-%m-%dT%H:%M:%S'))
+    sys.stdout.write("commit refs/heads/master\n")
+    #sys.stdout.write("committer %s %s -0000\n" % (author, int(commit_time)))
+    #sys.stdout.write("data %s\n" % len(props['svn:log']))
+    #sys.stdout.write(props['svn:log'])
+    #sys.stdout.write("\n")
+    sys.stdout.write('\n'.join(file_changes))
+    sys.stdout.write("\n\n")
+
+    svn_pool_destroy(revpool)
+    sys.stderr.write("done!\n")
+
+    #if rev % 1000 == 0:
+    #    sys.stderr.write("gc: %s objects\n" % len(gc.get_objects()))
+    #    sleep(5)
+
+
+def crawl_revisions(pool, repos_path):
+    """Open the repository at REPOS_PATH, and recursively crawl all its
+    revisions."""
+    global final_rev
+
+    # Open the repository at REPOS_PATH, and get a reference to its
+    # versioning filesystem.
+    repos_obj = svn_repos_open(repos_path, pool)
+    fs_obj = svn_repos_fs(repos_obj)
+
+    # Query the current youngest revision.
+    youngest_rev = svn_fs_youngest_rev(fs_obj, pool)
+
+
+    first_rev = 1
+    if final_rev == 0:
+        final_rev = youngest_rev
+    for rev in xrange(first_rev, final_rev + 1):
+        export_revision(rev, repos_obj, fs_obj, pool)
+
+
+if __name__ == '__main__':
+    usage = '%prog [options] REPOS_PATH'
+    parser = OptionParser()
+    parser.set_usage(usage)
+    parser.add_option('-f', '--final-rev', help='Final revision to import', 
+                      dest='final_rev', metavar='FINAL_REV', type='int')
+    parser.add_option('-t', '--trunk-path', help='Path in repo to /trunk',
+                      dest='trunk_path', metavar='TRUNK_PATH')
+    parser.add_option('-b', '--branches-path', help='Path in repo to /branches',
+                      dest='branches_path', metavar='BRANCHES_PATH')
+    parser.add_option('-T', '--tags-path', help='Path in repo to /tags',
+                      dest='tags_path', metavar='TAGS_PATH')
+    (options, args) = parser.parse_args()
+
+    if options.trunk_path != None:
+        trunk_path = options.trunk_path
+    if options.branches_path != None:
+        branches_path = options.branches_path
+    if options.tags_path != None:
+        tags_path = options.tags_path
+    if options.final_rev != None:
+        final_rev = options.final_rev
+
+    if len(args) != 1:
+        parser.print_help()
+        sys.exit(2)
+
+    # Canonicalize (enough for Subversion, at least) the repository path.
+    repos_path = os.path.normpath(args[0])
+    if repos_path == '.': 
+        repos_path = ''
+
+    # Call the app-wrapper, which takes care of APR initialization/shutdown
+    # and the creation and cleanup of our top-level memory pool.
+    run_app(crawl_revisions, repos_path)