diff --git versioncontrol_git.log.inc versioncontrol_git.log.inc
index 378f833..3044a46 100644
--- versioncontrol_git.log.inc
+++ versioncontrol_git.log.inc
@@ -36,6 +36,273 @@ function _versioncontrol_git_log_update_repository(VersioncontrolGitRepository &
$repository->updateLock();
$repository->update();
+$autoinc_base = variable_get('versioncontrol_git_autoincrement_base', 4);
+$autoinc_step = variable_get('versioncontrol_git_autoincrement_step', 5);
+
+ // Is this a fresh import?
+ $commits_in_db = $repository->loadCommits(array(), array(), array('may cache' => FALSE));
+ if (empty($commits_in_db)) {
+ $repo_id = $repository->repo_id;
+
+ $file_names = _version_control_fast_export();
+
+ $marks = _versioncontrol_git_log_process_marks($file_names[2]);
+ $logs = _versioncontrol_git_log_process_logfile($file_names[1], $marks);
+ $numlines = _versioncontrol_git_log_process_numstat($file_names[0]);
+
+ db_query("LOCK TABLES {versioncontrol_operations} WRITE");
+ $values = '';
+ $start = db_result(db_query('SELECT MAX(vc_op_id) FROM {versioncontrol_operations}'));
+ if (!is_numeric($start) || $start == 0) {
+ $start = ($autoinc_base - $autoinc_step);
+ }
+ foreach ($logs as $log) {
+ $start += $autoinc_step;
+ //$values = array(
+ // 'type' => 1,
+ // 'repo_id' => $repo_id,
+ // 'date' => $log->date,
+ // 'author' => $log->author,
+ // 'author_uid' => 0,
+ // 'committer' => $log->committer,
+ // 'committer_uid' => 0,
+ // 'revision' => $log->revision,
+ // 'message' => $log->msg,
+ //);
+ $log->vc_op_id = $start;
+ if (empty($log->author)) {
+ $author = '';
+ }
+ else {
+ $author = db_escape_string($log->author);
+ }
+ if (empty($log->committer)) {
+ $committer = '';
+ }
+ else {
+ $committer = db_escape_string($log->committer);
+ }
+ if (is_array($log->msg) || is_array($log->author) || is_array($log->committer)) {
+ dpm($log);
+ }
+ $msg = db_escape_string($log->msg);
+ if (empty($values)) {
+ $values = "(1, $repo_id, $log->date, '$author', 0, '$committer', 0, '$log->revision', '$msg')";
+ }
+ else {
+ $values .= ", (1, $repo_id, $log->date, '$author', 0, '$committer', 0, '$log->revision', '$msg')";
+ }
+ }
+ if ($values) {
+ db_query("INSERT INTO {versioncontrol_operations} (type, repo_id, date, author, author_uid, committer, committer_uid, revision, message) VALUES ". $values);
+ }
+ $end = db_result(db_query('SELECT MAX(vc_op_id) FROM {versioncontrol_operations} WHERE repo_id = '. $repo_id));
+ db_query("UNLOCK TABLES");
+ if ($start != $end) {
+ watchdog('vc_git', "Sequencing problem during inital import of {versioncontrol_operations} for repo id #$repo_id. Expected $start, got $end. Recovering.");
+ $result = db_query("SELECT vc_op_id, revision FROM {versioncontrol_operations} WHERE repo_id = $repo_id ORDER BY vc_op_id ASC");
+ foreach ($logs as $log) {
+ $r = db_fetch_array($result);
+ if ($r['revision'] == $log->revision) {
+ $log->vc_op_id = $r['vc_op_id'];
+ }
+ else {
+ $replacements = array('@op1' => $log->vc_op_id, '@op2' => $r['vc_op_id'], '@rev1' => $log->revision, '@rev2' => $r['revision']);
+ watchdog('vc_git', "Mismatch when correcting {versioncontrol_operations} autoincrement mapping. Was @op1, changing to @op2 for real revision @rev1 and database revision @rev2.", $replacements, WATCHDOG_INFO);
+ }
+ }
+ }
+
+ db_query("LOCK TABLES {versioncontrol_item_revisions} WRITE");
+ $values = '';
+ $start = db_result(db_query('SELECT MAX(item_revision_id) FROM {versioncontrol_item_revisions}')); //Make this happen in the same transaction.
+ if (!is_numeric($start) || $start == 0) {
+ $start = $autoinc_base - $autoinc_step;
+ }
+ foreach ($logs as $log) {
+ foreach($log->files as $hash => $file) {
+ $source_id = 0;
+ $start += $autoinc_step;
+ $log->files[$hash]['rev_id'] = $start;
+ $log->name_block->updateAncestor($log->revision, $file['name'], $start);
+ $type = $file['type'] == VERSIONCONTROL_ACTION_DELETED ? VERSIONCONTROL_ITEM_FILE_DELETED : VERSIONCONTROL_ITEM_FILE;
+ //define('VERSIONCONTROL_ITEM_DIRECTORY', 2); Git doesn't track directories so these are never used.
+ //define('VERSIONCONTROL_ITEM_DIRECTORY_DELETED', 4);
+ if (isset($numlines[$log->revision][$file['name']][0])) {
+ $lines_added = $numlines[$log->revision][$file['name']][0];
+ $lines_removed = $numlines[$log->revision][$file['name']][1];
+ }
+ else {
+ $lines_added = 0;
+ $lines_removed = 0;
+ }
+ //$values = array(
+ // 'vc_op_id' => $log->vc_op_id,
+ // 'repo_id' => $repo_id,
+ // 'path' => $file['name'],
+ // 'revision' => $log->revision,
+ // 'type' => $type,
+ // 'source_item_revision_id' => $log->name_block->getAncestor( $file['name'], $log->revision),
+ // 'action' => $file['type'],
+ // 'line_changes_added' => $lines_added,
+ // 'line_changes_removed' => $lines_removed,
+ //);
+ //$query->values($values);
+
+ $path = db_escape_string($file['name']);
+ $source_id = $log->name_block->getAncestor($file['name'], $log->revision);
+ $action = $file['type'];
+ if (empty($values)) {
+ $values = "($log->vc_op_id, $repo_id, '$path', '$log->revision', $type, $source_id, $action, $lines_added, $lines_removed)";
+ }
+ else {
+ $values .= ", ($log->vc_op_id, $repo_id, '$path', '$log->revision', $type, $source_id, $action, $lines_added, $lines_removed)";
+ }
+ }
+ }
+ if ($values) {
+ db_query("INSERT INTO {versioncontrol_item_revisions} (vc_op_id, repo_id, path, revision, type, source_item_revision_id, action, line_changes_added, line_changes_removed) VALUES " . $values);
+ }
+ $end = db_result(db_query('SELECT MAX(item_revision_id) FROM {versioncontrol_item_revisions} WHERE repo_id = '. $repo_id));
+ db_query("UNLOCK TABLES");
+ if ($start != $end) {
+ watchdog('vc_git', "Sequencing problem during inital import of {versioncontrol_item_revisions} for repo id #$repo_id. Expected $start, got $end. Recovering.");
+ $logs[0]->name_block->history = array();
+ $result = db_query("SELECT item_revision_id, path FROM {versioncontrol_item_revisions} WHERE repo_id = $repo_id ORDER BY item_revision_id ASC");
+ foreach ($logs as $log) {
+ foreach($log->files as $hash => $file) {
+ $r = db_fetch_array($result);
+ if ($r['path'] == $file['name']) {
+ $log->files[$hash]['rev_id'] = $r['item_revision_id'];
+ $log->name_block->updateAncestor($log->revision, $file['name'], $r['item_revision_id']);
+ }
+ else {
+ watchdog('vc_git', "Mismatch when correcting {versioncontrol_item_revisions} autoincrement mapping.", array(), WATCHDOG_INFO);
+ }
+ }
+ }
+ }
+
+ $query = db_insert('versioncontrol_git_item_revisions')->fields(array('item_revision_id', 'blob_hash'));
+ $values = array();
+ foreach ($logs as $log) {
+ foreach(array_keys($log->files) as $hash) {
+ $values = array(
+ 'item_revision_id' => $log->files[$hash]['rev_id'],
+ 'blob_hash' => $hash,
+ );
+ $query->values($values);
+ }
+ }
+ if ($values) {
+ $query->execute();
+ }
+
+
+ $query = db_insert('versioncontrol_git_operations')->fields(array('vc_op_id', 'author_name', 'committer_name', 'parent_commit', 'merge'));
+ $values = array();
+ foreach ($logs as $log) {
+ $merge = empty($log->merge) ? 0 : 1; // We don't reference the additional merge parents by commit.
+ $values = array(
+ 'vc_op_id' => $log->vc_op_id,
+ 'author_name' => $log->author,
+ 'committer_name' => $log->committer,
+ 'parent_commit' => $log->from,
+ 'merge' => $merge,
+ );
+ $query->values($values);
+ }
+ if ($values) {
+ $query->execute();
+ }
+
+ db_query('LOCK TABLES {versioncontrol_labels} WRITE');
+ $values = '';
+ $start = db_result(db_query('SELECT MAX(label_id) FROM {versioncontrol_labels}'));
+ if (!is_numeric($start) || $start == 0) {
+ $start = $autoinc_base - $autoinc_step;
+ }
+ $log = $logs[0];
+ $names = $log->name_block;
+ foreach (array_keys($names->names) as $name) {
+ $start += $autoinc_step;
+ $names->names[$name]['label_id'] = $start; //Save the new label_id in the shared nameBlock for later.
+ //$values = array(
+ // 'repo_id' => $repo_id,
+ // 'name' => $name,
+ // 'type' => $log->type,
+ // 'data' => '',
+ //);
+ //$query->values($values);
+ $name = db_escape_string($name);
+ if (empty($values)) {
+ $values = "($repo_id, '$name', $log->type, '')";
+ }
+ else {
+ $values .= ", ($repo_id, '$name', $log->type, '')";
+ }
+ }
+ if ($values) {
+ db_query("INSERT INTO {versioncontrol_labels} (repo_id, name, type, data) VALUES " . $values);
+ }
+ $end = db_result(db_query('SELECT MAX(label_id) FROM {versioncontrol_labels} WHERE repo_id = '. $repo_id));
+ db_query("UNLOCK TABLES");
+ if ($start != $end) {
+ watchdog('vc_git', "Sequencing problem during inital import of {versioncontrol_labels} for repo id #$repo_id. Expected $start, got $end. Recovering.");
+ $result = db_query("SELECT label_id, name FROM {versioncontrol_labels} WHERE repo_id = $repo_id ORDER BY label_id ASC");
+ $log = $logs[0];
+ $names = $log->name_block;
+ foreach (array_keys($names->names) as $name) {
+ $r = db_fetch_array($result);
+ if ($r['name'] == $name) {
+ $names->names[$name]['label_id'] = $r['label_id'];
+ }
+ else {
+ watchdog('vc_git', "Mismatch when correcting {versioncontrol_labels} autoincrement mapping.", array(), WATCHDOG_INFO);
+ }
+ }
+ }
+
+ $log = $logs[0];
+ $names = $log->name_block;
+ foreach (array_keys($names->names) as $name) {
+ if ($names->names[$name]['type'] == VERSIONCONTROL_LABEL_BRANCH) {
+ $label_obj = (object) array('name' => $name);
+ $revs = _versioncontrol_git_log_get_commits_in_label($label_obj);
+ foreach ($revs as $rev) {
+ $label_map[$rev][] = $name;
+ }
+ }
+ }
+
+ $query = db_insert('versioncontrol_operation_labels')->fields(array('vc_op_id', 'label_id', 'action'));
+ $values = '';
+ foreach ($logs as $log) {
+ if (empty($label_map[$log->revision])) {
+ continue;
+ }
+ // If the same commit appears in 10 branches we should have 10 rows.
+ foreach ($label_map[$log->revision] as $name) {
+ $values = array(
+ 'vc_op_id' => $log->vc_op_id,
+ 'label_id' => $log->name_block->names[$name]['label_id'],
+ 'action' => 2, // It's always two on the live db. Column will be removed.
+ );
+ $query->values($values);
+ }
+ }
+ unset($label_map);
+ if ($values) {
+ $query->execute();
+ }
+
+ $repository->updated = time();
+ $repository->updateLock(0);
+ $repository->update();
+
+ return TRUE;
+ }
+
// 1. Process branches
// Fetch branches from the repo and load them from the db.
@@ -46,11 +313,9 @@ function _versioncontrol_git_log_update_repository(VersioncontrolGitRepository &
$branches_in_db_by_name[$branch->name] = $branch;
}
unset($branches_in_db);
-
// Determine whether we've got branch changes to make.
$branches_new = array_diff_key($branches_in_repo, $branches_in_db_by_name);
$branches_deleted = array_diff_key($branches_in_db_by_name, $branches_in_repo);
-
// Insert new branches in the repository. Later all commits in these new
// branches will be updated.
// Unfortunately we can't say anything about the branch author at this time.
@@ -67,14 +332,11 @@ function _versioncontrol_git_log_update_repository(VersioncontrolGitRepository &
foreach ($branches_in_db as $branch) {
$branches_in_db_by_name[$branch->name] = $branch;
}
-
// Deleted branches are removed, commits in them are not!
foreach($branches_deleted as $branch) {
$branch->delete();
}
-
// 2. Process commits
-
// Fetch commits from the repo and load them from the db.
$commits_in_repo_hashes = $repository->fetchCommits();
$commits_in_db = $repository->loadCommits(array(), array(), array('may cache' => FALSE));
@@ -84,16 +346,13 @@ function _versioncontrol_git_log_update_repository(VersioncontrolGitRepository &
}
unset($commits_in_db);
$commits_new = array_diff($commits_in_repo_hashes, $commits_in_db_hashes);
-
// Insert new commits in the database.
foreach ($commits_new as $hash) {
$command = "show --numstat --summary --pretty=format:\"%H%n%P%n%an%n%ae%n%cn%n%ce%n%ct%n%s%n%b%nENDOFOUTPUTGITMESSAGEHERE\" " . escapeshellarg($hash);
$output = _versioncontrol_git_log_exec($command);
_versioncontrol_git_log_parse_and_insert_commit($repository, $output, $branches_in_db_by_name);
}
-
// 3. Process tags
-
// Insert new tags in the database.
$tags_in_repo = $repository->fetchTags();
$tags_in_db = $repository->loadTags();
@@ -115,9 +374,23 @@ function _versioncontrol_git_log_update_repository(VersioncontrolGitRepository &
$repository->update();
return TRUE;
}
-
/// All general functions
-
+function _version_control_fast_export() {
+ $git_bin = variable_get('versioncontrol_git_binary_path', 'git');
+ if ($errors = _versioncontrol_git_binary_check_path($git_bin)) {
+ watchdog('vc_git', '!errors', array('!errors' => implode('
', $errors)), WATCHDOG_ERROR);
+ return array();
+ }
+ $tmp_mark = tempnam(file_directory_temp(), 'vclpm');
+ $tmp_log = tempnam(file_directory_temp(), 'vclpl');
+ $cmd = "fast-export --no-data --tag-of-filtered-object=drop --signed-tags=verbatim --export-marks=$tmp_mark --all";
+ exec(escapeshellcmd("$git_bin $cmd") ." > $tmp_log", $logs);
+ $tmp_stat = tempnam(file_directory_temp(), 'vclps');
+ $cmd = "log --numstat --pretty=format:%H";
+ exec(escapeshellcmd("$git_bin $cmd") ." > $tmp_stat", $logs);
+
+ return array($tmp_stat, $tmp_log, $tmp_mark);
+}
/**
* Execute a Git command using the root context and the command to be executed.
* @param string $command Command to execute.
@@ -127,7 +400,7 @@ function _versioncontrol_git_log_exec($command) {
$logs = array();
$git_bin = variable_get('versioncontrol_git_binary_path', 'git');
if ($errors = _versioncontrol_git_binary_check_path($git_bin)) {
- watchdog('versioncontrol_git', '!errors', array('!errors' => implode('
', $errors)), WATCHDOG_ERROR);
+ watchdog('vc_git', '!errors', array('!errors' => implode('
', $errors)), WATCHDOG_ERROR);
return array();
}
exec(escapeshellcmd("$git_bin $command"), $logs);
@@ -135,7 +408,6 @@ function _versioncontrol_git_log_exec($command) {
reset($logs); // Reset the array pointer, so that we can use next().
return $logs;
}
-
/// All commit related function
/**
@@ -519,7 +791,7 @@ function _versioncontrol_git_log_process_tags($repository, $tags_new) {
}
}
else {
- watchdog('versioncontrol_git', 'Serious problem in tag parsing, please check that you\'re using a supported version of git!');
+ watchdog('vc_git', 'Serious problem in tag parsing, please check that you\'re using a supported version of git!');
}
$tag_data = array(
@@ -532,3 +804,217 @@ function _versioncontrol_git_log_process_tags($repository, $tags_new) {
$tag->insert();
}
}
+
+function _versioncontrol_git_log_process_marks($mark_file) {
+ $lines = file($mark_file);
+ $marks = array();
+ foreach ($lines as $line) {
+ $e = explode(' ', $line);
+ $marks[intval(ltrim($e[0], ':'))] = trim($e[1]);
+ }
+ unlink($mark_file);
+ return $marks;
+}
+
+
+function _versioncontrol_git_log_process_numstat($stats_file) {
+ $lines = file($stats_file);
+ $stats = array();
+ $hash = GIT_NULL_REV;
+ $mark = array();
+ foreach ($lines as $line) {
+ if (empty($line)) {
+ continue;
+ }
+ if (preg_match('/^([a-f,0-9]{40})$/', $line, $mark)) {
+ $hash = $mark[1];
+ continue;
+ }
+ if (preg_match('/^(\d+)\t(\d+)\t(.+)$/', $line, $mark)) {
+ $name = '/'. $mark[3];
+ $stats[$hash][$name] = array($mark[1], $mark[2]);
+ }
+ }
+ unlink($stats_file);
+ return $stats;
+}
+
+function _versioncontrol_git_log_process_logfile($log_file, &$marks) {
+ $fp = fopen($log_file, 'rb');
+ $name_block = new nameBlock();
+ $log = array();
+ $data_length = 0;
+ while (($line = fgets($fp)) !== FALSE) {
+ $mark = array();
+ if (empty($line)) {
+ continue;
+ }
+ do {
+ $repeat = FALSE;
+ // Add in tagger.
+ if ($data_length) {
+ if ($data_length >= strlen($line)) {
+ $data_length -= $working->addMsg($line);
+ }
+ else {
+ $data_length = 0;
+ $len = $working->addMsg($line);
+ $line = substr($line, $len);
+ $repeat = TRUE;
+ }
+ }
+ elseif (preg_match('/^commit\b/', $line)) {
+ if (isset($working)) {
+ $working->updateCommit();
+ $log[] = $working;
+ }
+ $working = new logBlock();
+ $working->name_block = $name_block;
+ $working->setCommit($line);
+ }
+ elseif (preg_match('/(.) (\d+) ([a-f,0-9]{40}) (.+)$/', $line, $mark)) {
+ $working->addFile($mark[1], $mark[2], $mark[3], $mark[4]);
+ }
+ elseif (preg_match('/^data (\d+)$/', $line, $mark)) {
+ $data_length = $mark[1];
+ }
+ elseif (preg_match('/^mark :(\d+)$/', $line, $mark)) {
+ $working->revision = $marks[$mark[1]];
+ }
+ elseif (preg_match('/^from :(\d+)$/', $line, $mark)) {
+ $working->from = $marks[$mark[1]];
+ }
+ elseif (preg_match('/^author[^<]*<([^>]+)> (\d+)/', $line, $mark)) {
+ $working->author = $mark[1];
+ $working->date = $mark[2];
+ }
+ elseif (preg_match('/^committer[^<]*<([^>]+)> (\d+)/', $line, $mark)) {
+ $working->committer = $mark[1];
+ }
+ elseif (preg_match('/^merge :(\d+)$/', $line, $mark)) {
+ $working->merge[] = $marks[$mark[1]];
+ }
+ } while ($repeat);
+ }
+ if (empty($working)) {
+ watchdog('vc_git', "Can't process empty repository.", array(), WATCHDOG_INFO);
+ return FALSE;
+ }
+ $working->updateCommit();
+ $log[] = $working;
+ fclose($fp);
+ unlink($log_file);
+ return $log;
+}
+
+class logBlock {
+ public $files = array();
+ public $commit = ''; //This actually holds branch/tag names.
+ public $type = 0;
+ public $revision = '';
+ public $date = 0;
+ public $author = '';
+ public $committer = '';
+ public $msg = '';
+ public $from = GIT_NULL_REV;
+ public $vc_op_id;
+ public $name_block;
+ public $merge = array();
+
+ function addMsg($line) {
+ $this->msg .= $line;
+ return strlen($line);
+ }
+
+ function setCommit($commit) {
+ $parts = explode(' ', $commit);
+ $ref = trim($parts[1]);
+ $path = explode('/', $ref);
+ $this->commit = array_pop($path);
+ // We've popped the name off so if any remaining piece says tag we have a tag.
+ if (in_array('tags', $path)) {
+ $this->type = VERSIONCONTROL_LABEL_TAG;
+ // Annotated tags are style of tag: and have a tagger.
+ }
+ else {
+ $this->type = VERSIONCONTROL_LABEL_BRANCH;
+ }
+ }
+
+ function updateCommit() {
+ $this->name_block->add($this->commit, $this->type);
+ }
+
+ function addfile($type, $mode, $hash, $name) {
+ $name = '/'. $name; // Remove this later.
+ switch ($type) {
+ case 'M':
+ $type = VERSIONCONTROL_ACTION_MODIFIED;
+ break;
+ case 'A':
+ $type = VERSIONCONTROL_ACTION_ADDED;
+ break;
+ case 'D':
+ $type = VERSIONCONTROL_ACTION_DELETED;
+ break;
+ case 'R':
+ $type = VERSIONCONTROL_ACTION_MOVED;
+ break;
+ case 'C':
+ $type = VERSIONCONTROL_ACTION_COPIED;
+ break;
+ case '?':
+ $type = VERSIONCONTROL_ACTION_MERGED;
+ break;
+ case '?':
+ $type = VERSIONCONTROL_ACTION_REPLACED;
+ break;
+ default:
+ $type = VERSIONCONTROL_ACTION_OTHER;
+ }
+ $this->files[$hash] = array('name' => $name, 'type' => $type, 'mode' => $mode);
+ $this->name_block->addAncestor($name, $this->revision);
+ }
+}
+
+class nameBlock {
+ public $names = array();
+ public $ancestor = array();
+ public $history = array();
+
+ function add($name, $type) {
+ if (!array_key_exists($name, $this->names)) {
+ $this->names[$name]['type'] = $type; // Will be updated with a label_id
+ $this->names[$name]['label_id'] = 0; // Will be updated with a label_id
+ }
+ }
+
+ function addAncestor($name, $hash) {
+ if (array_key_exists($name, $this->ancestor)) {
+ if (array_key_exists($hash, $this->ancestor[$name])) {
+ return;
+ }
+ $keys = array_keys($this->ancestor[$name]);
+ $last = end($keys);
+ }
+ else {
+ $this->ancestor[$name][$hash] = 0;
+ return;
+ }
+ $this->ancestor[$name][$hash] = $last;
+ }
+
+ function updateAncestor($hash, $name, $id) {
+ $this->history[$name][$hash] = $id;
+ }
+
+ function getAncestor($name, $rev) {
+ if (array_key_exists($name, $this->history)) {
+ $last_rev = $this->ancestor[$name][$rev];
+ if ($last_rev && array_key_exists($last_rev, $this->history[$name])) {
+ return $this->history[$name][$last_rev];
+ }
+ }
+ return 0;
+ }
+}
diff --git versioncontrol_git.module versioncontrol_git.module
index 442ea1e..f0d3365 100644
--- versioncontrol_git.module
+++ versioncontrol_git.module
@@ -95,7 +95,7 @@ function versioncontrol_git_views_api() {
function _versioncontrol_git_get_binary_path($error = TRUE) {
$git_bin = variable_get('versioncontrol_git_binary_path', 'git');
if ($errors = _versioncontrol_git_binary_check_path($git_bin)) {
- watchdog('versioncontrol_git', '!errors', array('!errors' => implode('
', $errors)), WATCHDOG_ERROR);
+ watchdog('vc_git', '!errors', array('!errors' => implode('
', $errors)), WATCHDOG_ERROR);
if ($error) {
throw new Exception('No git binary could be found at the specified location, "' . $git_bin . '"', E_ERROR);
}