diff --git a/includes/classes.inc b/includes/classes.inc index 5134d40..acd5d3f 100644 --- a/includes/classes.inc +++ b/includes/classes.inc @@ -11,8 +11,6 @@ class VersioncontrolGitBackend extends VersioncontrolBackend { // Use the commit hash for to identify the commit instead of an individual // revision for each file. VERSIONCONTROL_CAPABILITY_ATOMIC_COMMITS, - VERSIONCONTROL_CAPABILITY_COMMIT_RESTRICTIONS, - VERSIONCONTROL_CAPABILITY_BRANCH_TAG_RESTRICTIONS ); $this->classes = array( 'repo' => 'VersioncontrolGitRepository', diff --git a/versioncontrol_git.log.inc b/versioncontrol_git.log.inc index b9c6acb..79ab4e3 100755 --- a/versioncontrol_git.log.inc +++ b/versioncontrol_git.log.inc @@ -6,13 +6,16 @@ * account management as a pluggable backend. * * Copyright 2008 by Jimmy Berry ("boombatower", http://drupal.org/user/214218) - * Copyright 2009 by Cornelius Riemenschneider ("CorniI", http://drupal.org/user/136353) + * Copyright 2009-2010 by Cornelius Riemenschneider ("CorniI", http://drupal.org/user/136353) */ include_once(drupal_get_path('module', 'versioncontrol') .'/includes/VersioncontrolBranch.php'); include_once(drupal_get_path('module', 'versioncontrol') .'/includes/VersioncontrolOperation.php'); include_once(drupal_get_path('module', 'versioncontrol') .'/includes/VersioncontrolItem.php'); +//TODO: in the update hooks, forbid including our magic constants in the commit/tag messages! +// Unfortunately, I wasn't able to get random magic strings working :( + /** * Actually update the repository by fetching commits and other stuff * directly from the repository, invoking the git executable. @@ -33,109 +36,78 @@ function _versioncontrol_git_log_update_repository(&$repository) { } $repository->data['versioncontrol_git']['locked'] = 1; $repository->update(); + + $branches_in_repo = _versioncontrol_git_log_get_branches_in_repo(); + $branches_in_db = _versioncontrol_git_log_get_branches_in_db($repository); + $branches_new = array_diff($branches_in_repo, $branches_in_db); + $branches_deleted = array_diff($branches_in_db, $branches_in_repo); - // Get the list of current branches from Git. - // $branch_list is the list of all branches currently in the repo. - $branch_list = _versioncontrol_git_log_get_branches(); - // $branches will be an assoc array $branch_name => $label with $label being - // a VersioncontrolBranch object. - $branches = array(); - // This inserts new branches into the repository as needed. - // For deleting branches, see the @TODO below. - foreach ($branch_list as $branch_name) { - $label = new VersioncontrolBranch($branch_name, + // 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. + // The post-update hook could do this, though. + // We also can't insert a VCOperation which adds the branch, because + // we don't know anything about the branch. This is all stuff a hook could + // figure out. + // Here we will just ensure that each branch is in the database. + foreach($branches_new as $branch_name) { + $label = new VersioncontrolBranch($branch_name, VERSIONCONTROL_ACTION_MODIFIED, NULL, $repository); $label->ensure(); - $branches[$branch_name] = $label; } - //jpetso and me (corni) came to the conclusion that we will not delete branches. - // TODO: revisit this! - - // Record new commits. - $constraints = array( - 'vcs' => array('git'), - 'repo_ids' => array($repository->repo_id), - 'types' => array(VERSIONCONTROL_OPERATION_COMMIT), - 'branches' => array() // used in the loop - ); - $branches_per_commit = array(); - $existing_revs = array(); - // Get the existing revisions from the cache. - $cache_object = cache_get('versioncontrol_git_rev_cache'); - // Check wether the cache object exists or not. - if (is_object($cache_object)) { - $existing_revs = $cache_object->data; + // Later we need a list of $branch => VersioncontrolBranch so construct it here + $branch_labels_in_db = $repository->getLabels(array('type' => VERSIONCONTROL_LABEL_BRANCH)); + $branch_label_list = array(); + foreach($branch_labels_in_db as $label) { + $branch_label_list[$label->name] = $label; } - // Get the list of current branches from Git. - // Generate the range per branch with which git shall be called. - foreach ($branches as $branch_name => $label) { - if (is_object($cache_object)) { - // We get all commits we have in this branch to not process them later. - $constraints['branches'] = array($branch_name); - $latest_commit_date = 0; - $commit_op = VersioncontrolOperationCache::getInstance()->getOperations($constraints); - $latest_commit = FALSE; - foreach ($commit_op as $vc_op_id => $c_op) { - if ($latest_commit_date < $c_op['date']) { - $latest_commit = $c_op['revision']; - $latest_commit_date = $c_op['date']; - $existing_revs[$branch_name][$latest_commit] = TRUE; - } - } - } - // No way to free the damned mysql result!! - unset($commit_op); - - $commits_in_branch = _versioncontrol_git_log_get_commits_in_branch($repository, escapeshellarg($branch_name)); - foreach ($commits_in_branch as $commit) { - if (!isset($existing_revs[$branch_name][$commit])) { - if (!isset($branches_per_commit[$commit]) || !is_array($branches_per_commit[$commit])) { - $branches_per_commit[$commit] = array($label); - } - else { - $branches_per_commit[$commit][] = $label; - } - } - } + + // Deleted branches are removed, commits in them are not! + // TODO: the db_query in here is not nice, that should be part of vcapi! + foreach($branches_deleted as $branch_name) { + _versioncontrol_git_log_remove_branch_from_commits($repository, $branch_name); + db_query('DELETE FROM {versioncontrol_labels} + WHERE repo_id = %d AND type = %d AND label_id = %d', + $repository->repo_id, VERSIONCONTROL_LABEL_BRANCH, + $branch_label_list[$branch_name]->label_id); } - // This uses an extra loop on purpose! - // Process all commits on a per-branch base. - foreach ($branches_per_commit as $revision => $branch) { - // Update commits from Git. - _versioncontrol_git_process_commits($repository, $revision, $branches_per_commit, $existing_revs); + + $commits_in_db = _versioncontrol_git_log_get_commits_in_db($repository); + $commits_in_repo = _versioncontrol_git_log_get_commits_in_repo($repository); + $commits_new = array_diff($commits_in_repo, $commits_in_db); + + // Insert new commits in the database. + foreach($commits_new as $new_commit) { + $new_commit_data = _versioncontrol_git_log_get_raw_commit_data($new_commit); + _versioncontrol_git_log_parse_and_insert_commit($repository, $new_commit_data, $commits_in_db, $branch_label_list); } - // Check tags. - $tags = _versioncontrol_git_log_get_tags(); //Now we have the current list of tags as array of strings. - $constraints = array( - 'vcs' => array('git'), - 'repo_ids' => array($repository->repo_id), - 'types' => array(VERSIONCONTROL_OPERATION_TAG) - ); - $existing_tag_ops = VersioncontrolOperationCache::getInstance()->getOperations($constraints); - $existing_tags = array(); - foreach ($existing_tag_ops as $tag_op) { - if (!in_array($tag_op->labels[0]->name, $existing_tags)) { - $existing_tags[] = $tag_op->labels[0]->name; - } + + // Add a new branch to all commits contained in that branch. + foreach($branches_new as $branch_new) { + $commits = _versioncontrol_git_log_get_commits_in_branch($branch_new); + _versioncontrol_git_log_attach_branch_to_commits($branch_label_list[$branch_new], $commits); } + + // Insert new tags in the database. + $tags_in_repo = _versioncontrol_git_log_get_tags_in_repo(); + $tags_in_db = _versioncontrol_git_log_get_tags_in_db($repository); // Deleting tags is *not* supported. Read the manual if you want to know why... // Check for new tags. - $new_tags = array_diff($tags, $existing_tags); - if (!empty($new_tags)) { - _versioncontrol_git_process_tags($repository, $new_tags); + $tags_new = array_diff($tags_in_repo, $tags_in_db); + if (!empty($tags_new)) { + _versioncontrol_git_log_process_tags($repository, $tags_new); } // Update repository updated field. Displayed on administration interface for documentation purposes. $repository->data['versioncontrol_git']['updated'] = time(); $repository->data['versioncontrol_git']['locked'] = 0; $repository->update(); - - // Write back the cache. - cache_set('versioncontrol_git_rev_cache', $existing_revs); return TRUE; } +/// All general functions + /** * Execute a Git command using the root context and the command to be executed. * @param string $command Command to execute. @@ -150,231 +122,109 @@ function _versioncontrol_git_log_exec($command) { } /** - * Get branches from Git using 'branch -l' command. - * @return array List of branches. - */ -function _versioncontrol_git_log_get_branches() { - $logs = _versioncontrol_git_log_exec('git show-ref --heads'); // Query branches. - $branches = _versioncontrol_git_log_parse_branches($logs); // Parse output. - return $branches; -} - -/** - * Parse the branch list output from Git. + * Returns a list of all existing labels in the database of $type. + * @param VersioncontrolRepository $repository + * @param $type + * @return array An array of branch/tag names. */ -function _versioncontrol_git_log_parse_branches(&$logs) { - $branches = array(); - while (($line = next($logs)) !== FALSE) { - // the output of git show-ref --heads looks like this - // c5bbe46edb29eb6142942ba971a51dc9d5becbc0 refs/heads/master - // to get 'master', we have to skip 52 chars. - $branches[] = substr(trim($line), 52); +function _versioncontrol_git_log_get_labels_in_db($repository, $type) { + $labels_in_db = $repository->getLabels(array('type' => $type)); + $label_names_in_db = array(); + foreach($labels_in_db as $label) { + $label_names_in_db[] = $label->name; } - return $branches; + return $label_names_in_db; } -/** - * Get tags from Git using 'tag -l' command. - */ -function _versioncontrol_git_log_get_tags() { - //TODO: incorporate a --dereference and parse better, saves one git call for tags - $logs = _versioncontrol_git_log_exec('git show-ref --tags'); // Query tags. - $tags = _versioncontrol_git_log_parse_tags($logs); // Parse output. - return $tags; -} +/// All commit related function /** - * Parse the tag list output from Git. + * Return one commit op. + * @param VersioncontrolRepository $repository + * @param string $revision + * @return VersioncontrolOperation */ -function _versioncontrol_git_log_parse_tags(&$logs) { - $tags = array(); - while (($line = next($logs)) !== FALSE) { - // the output of git show-ref --tags looks like this - // 94a5915923d5a9a6af935e4055c95582fd1a1136 refs/tags/DRUPAL-5--1-0 - // to get 'DRUPAL-5--1-0', we have to skip 51 chars. - $tags[] = substr(trim($line), 51); - } - return $tags; +function _versioncontrol_git_log_get_commit($repository, $revision) { + $constraints = array( + 'repo_ids' => array($repository->repo_id), + 'types' => array(VERSIONCONTROL_OPERATION_COMMIT), + 'revisions' => array($revision), + ); + $commit_op = VersioncontrolOperationCache::getInstance()->getOperations($constraints); + return array_pop($commit_op); } /** - * Parses output of git show $tag_name provided by _versioncontrol_git_get_tag_operation() to retrieve an $operation for inserting a tag. - * @param $repository - * @param string $tag_name The name of the parsed tag - * @param $logs The output of git show - * @return array An $operation array which contains the info for the tag. + * Returns a list of all existing commits in the database. + * @param VersioncontrolRepository $repository + * @return array An array of commit revisions. */ -function _versioncontrol_git_log_parse_tag_info($repository, &$logs, $tag_commits) { - $line = next($logs); // Get op type - if ($line === FALSE) { - return FALSE; - } - if ($line == 'commit') { - //let's get the author and the date from the tagged commit, better than nothing. - $tagged_commit = next($logs); // Get the tagged commit - $tag_name = substr(strrchr(next($logs), '/'), 1 ); // Get the name of the tag based on %(refname) - next($logs); // Skip these two lines - next($logs); - // Get the tag/commit message - $message = ''; - $i = 0; - while (($line = next($logs)) !== FALSE) { - if ($line == 'ENDOFGITTAGOUTPUTMESAGEHERE') { - break; - } - if ($i == 1) { - $message .= "\n"; - } - $message .= $line ."\n"; - $i++; - } - $constraints = array( - 'vcs' => array('git'), - 'repo_ids' => array($repository->repo_id), - 'types' => array(VERSIONCONTROL_OPERATION_COMMIT), - 'revisions' => array($tagged_commit) - ); - $op = VersioncontrolOperationCache::getInstance()->getOperations($constraints); - $op = array_pop($op); - //FIXME get author/commiter - $o_operation = new VersioncontrolGitOperation( - VERSIONCONTROL_OPERATION_TAG, - $op->author, - $op->date+1, // We want to be displayed *after* the tagged commit. - $tagged_commit, - $message, - $op->author, - $repository - ); - $o_operation->labels = array( - 0 => new VersioncontrolTag($tag_name, VERSIONCONTROL_ACTION_ADDED, null, $repository) - ); - return $o_operation; - } - $line = next($logs); // Skip op sha1 - $tag_name = substr(strrchr(next($logs), '/'), 1 ); // Get the name of the tag based on %(refname) - $tagger = next($logs); // Get tagger - $date = strtotime(next($logs)); // Get date - // Get the tag message - $message = ''; - $i = 0; - while (($line = next($logs)) !== FALSE) { - if ($line == 'ENDOFGITTAGOUTPUTMESAGEHERE') { - break; - } - if ($i == 1) { - $message .= "\n"; - } - $message .= $line ."\n"; - $i++; - } - $tagged_commit = $tag_commits[$tag_name]; - // By now, we're done with the parsing, construct the op array - $o_operation = new VersioncontrolGitOperation( - VERSIONCONTROL_OPERATION_TAG, - $tagger, - $date, - $tagged_commit, - $message, - $tagger, - $repository +function _versioncontrol_git_log_get_commits_in_db($repository) { + $constraints = array( + 'vcs' => array('git'), + 'repo_ids' => array($repository->repo_id), + 'types' => array(VERSIONCONTROL_OPERATION_COMMIT), ); - $o_operation->labels = array( - 0 => new VersioncontrolTag($tag_name, VERSIONCONTROL_ACTION_ADDED, null, $repository) - ); - return $o_operation; + $commits_in_db_as_op = VersioncontrolOperationCache::getInstance()->getOperations($constraints); + $commits_in_db = array(); + foreach($commits_in_db_as_op as $vc_op_id => $vc_op) { + $commits_in_db[] = $vc_op->revision; + } + return $commits_in_db; } /** - * Invokes 'git-show tag' to get information about a tag. - * It's output is later parsed by _versioncontrol_git_log_parse_tag_info(). - * @param $repository - * @param string $tag The name of the tag. - * @return An $operation array which contains the info for the tag. + * Get the raw git output for $revision. + * @param string $revision + * @return The raw git output which can be fed into the commit parser for a given commit. */ -function _versioncontrol_git_get_tag_operations($repository, $tags) { - $tag_ops = array(); - $tag_string = ''; - if (empty($tags)) { - return array(); - } - foreach ($tags as $tag) { - $tag_string .= escapeshellarg("refs/tags/$tag") .' '; - } - $format = "%(objecttype)\n%(objectname)\n%(refname)\n%(taggername) %(taggeremail)\n%(taggerdate)\n%(contents)\nENDOFGITTAGOUTPUTMESAGEHERE"; - $exec = "git for-each-ref --format=\"$format\" $tag_string"; - $logs_tag_msg = _versioncontrol_git_log_exec($exec); - $exec = "git show-ref -d $tag_string"; - $logs_tag_commits = _versioncontrol_git_log_exec($exec); - $tag_commits = array(); - foreach ($logs_tag_commits as $line) { - if (substr($line, -3, 3) == '^{}') { - $commit = substr($line, 0, 40); - $tag = substr($line, 41); - $tag = substr(substr($line, 41, strlen($tag) -3), 10); - $tag_commits[$tag] = $commit; - } - } - do { - $ret = _versioncontrol_git_log_parse_tag_info($repository, $logs_tag_msg, $tag_commits); - if ($ret !== FALSE) { - $tag_ops[] = $ret; - } - }while ($ret !== FALSE); - return $tag_ops; +function _versioncontrol_git_log_get_raw_commit_data($revision) { + $revision = escapeshellarg($revision); + $command = "git log $revision --numstat --summary --pretty=format:\"%H%n%P%n%aN <%ae>%n%ct%n%s%n%b%nENDOFOUTPUTGITMESSAGEHERE\" -n 1 -c --"; + return _versioncontrol_git_log_exec($command); } /** - * Does all the processing for all new tags. - * @param $repository - * @param array $new_tags An array of strings for all new tags which shall be processed + * Returns an array of all branches a given commit is in. + * @param string $revision + * @param array $branch_label_list + * @return VersioncontrolBranch */ -function _versioncontrol_git_process_tags($repository, $new_tags) { - $tag_ops = _versioncontrol_git_get_tag_operations($repository, $new_tags); - foreach ($tag_ops as $tag_op) { - $op_items = array(); - $tag_op->insert($op_items); - $constraints = array( - 'vcs' => array('git'), - 'repo_ids' => array($repository->repo_id), - 'types' => array(VERSIONCONTROL_OPERATION_COMMIT), - 'revisions' => array($tag_op->revision) - ); - $tag_commits = VersioncontrolOperationCache::getInstance()->getOperations($constraints); - foreach ($tag_commits as $vc_op_id => $tag_commit_op) { - $tag_commit_op->labels[] = new VersioncontrolTag( - $tag_op['labels'][0]['name'], - VERSIONCONTROL_ACTION_MODIFIED, - null, - $repository - ); - $tag_commit_op->updateLabels($tag_commit_op->labels); +function _versioncontrol_git_log_get_branches_of_commit($revision, $branch_label_list) { + $exec = 'git branch --contains ' . escapeshellarg($revision); + $logs = _versioncontrol_git_log_exec($exec); + $branches = array(); + while (($line = next($logs)) !== FALSE) { + $line = trim($line); + if($line[0] == '*') { + $line = substr($line, 2); } + $branches[] = $branch_label_list[$line]; } + return $branches; } /** - * Get all commits from Git using 'git log' command. + * This function returns all commits in the repository * @param $repository - * @param string $range the computed range for the branch we check - * @param array $branches_per_commit An array of all commits we will encounter with a list of branches they are in. + * @return array An array of strings with all commit id's in it */ -function _versioncontrol_git_process_commits($repository, $revision, &$branches_per_commit, &$existing_revs) { - $rev_shell = escapeshellarg($revision); - $command = "git log $rev_shell --numstat --summary --pretty=format:\"%H%n%P%n%aN <%ae>%n%ct%n%s%n%b%nENDOFOUTPUTGITMESSAGEHERE\" -n 1 --"; - $logs = _versioncontrol_git_log_exec($command); - _versioncontrol_git_log_parse_commits($repository, $logs, $branches_per_commit, $existing_revs); // Parse the info from the raw output. +function _versioncontrol_git_log_get_commits_in_repo($repository) { + $logs = _versioncontrol_git_log_exec("git rev-list --all"); + $commits = array(); + while (($line = next($logs)) !== FALSE) { + $commits[] = trim($line); + } + return $commits; } /** - * This function returns all commits in the given range. - * It is used to get all new commits in a branch, which is specified by @p $range - * @param $repository - * @param string $range The range of the commits to retrieve + * This function returns all commits in the given branch. + * @param string $branch * @return array An array of strings with all commit id's in it */ -function _versioncontrol_git_log_get_commits_in_branch($repository, $range) { - $logs = _versioncontrol_git_log_exec("git rev-list $range --reverse --"); // Query tags. +function _versioncontrol_git_log_get_commits_in_branch($branch) { + $logs = _versioncontrol_git_log_exec("git rev-list " . escapeshellarg($branch) . " --"); $commits = array(); while (($line = next($logs)) !== FALSE) { $commits[] = trim($line); @@ -383,219 +233,131 @@ function _versioncontrol_git_log_get_commits_in_branch($repository, $range) { } /** - * A helper function to get the source_items. - * @param $repository - * @param $revision The revision of the parent item. - * @param $filename The filename of the parent item. - * @return array An $item array ready to use for $operation['source_items'] + * Returns the previous commits which touched a given file. + * @param string $path + * @param string $revision_current + * @param int $parent_count + * @return string */ -function _versioncontrol_git_get_source_item_helper($repository, $revision, $filename, $branches) { - $branch_names = array(); - foreach ($branches as $branch) { - $branch_names[] = $branch['name']; - } - $constraints = array( - 'vcs' => array('git'), - 'repo_ids' => array($repository->repo_id), - 'types' => array(VERSIONCONTROL_OPERATION_COMMIT), - 'paths' => array($filename), - 'branches' => $branch_names - ); - $commit_op = versioncontrol_get_operations($constraints); - ksort($commit_op); - $commit_op = array_pop($commit_op); - $op_items = versioncontrol_get_operation_items($commit_op); - $type = $op_items[$filename]['type'] ? $op_items[$filename]['type'] : VERSIONCONTROL_ITEM_FILE; - // ['action'] not needed for source items :) - return array( - 'path' => $filename, - 'type' => $type, - 'revision' => $op_items[$filename]['revision'], - ); +function _versioncontrol_git_get_prev_commits_touching_file($path, $revision_current, $parent_count) { + // We need one more revision to fulfill git rev-list's requirements. + $parent_count += 1; + $exec = 'git rev-list -n ' . $parent_count . ' ' . escapeshellarg($revision_current) . ' -- ' . escapeshellarg($path); + return _versioncontrol_git_log_exec($exec); } /** - * A function to a source_item for a specific file. - * @param $repository as we get it from the API - * @param string $filename the revision of the current item we shall get it's source from - * @return array $source_items array for use in an $operation. + * A function to fill in the source_item for a specific VersioncontrolItem. + * @param VersioncontrolItem $item + * @param array $parents The parent commit(s) + * @return none */ -function _versioncontrol_git_get_source_item($repository, $filename, $parents, $branches, $commit_rev) { - $ret = array(); - if (count($parents) == 1) { - $filenameg = substr($filename, 1); - $filenameg = escapeshellarg($filenameg); - $commit_rev = escapeshellarg($commit_rev); - $exec = "git rev-list -n 1 ". $commit_rev ."^ -- $filenameg"; - $logs = _versioncontrol_git_log_exec($exec); // Query tags. - $revision = next($logs); - $ret = array( - new VersioncontrolGitItem( - VERSIONCONTROL_ITEM_FILE, - $filename, - $revision, - null, - $repository - ) - ); - } - else { - foreach ($parents as $rev) { - $ret[] = _versioncontrol_git_get_source_item_helper($repository, $rev, $filename, $branches); - } - } - return $ret; -} +function _versioncontrol_git_fill_source_item($item, $parents) { + $parent_count = count($parents); + $path_stripped = substr($item->path, 1); + $prev_revisions = _versioncontrol_git_get_prev_commits_touching_file($path_stripped, $item->revision, $parent_count); -function _versioncontrol_git_insert_commit($repository, $date, $username, $message, $revision, $branches, $op_items) { - //FIXME: get author and commiter - $op = new VersioncontrolGitOperation(VERSIONCONTROL_OPERATION_COMMIT, $username, $date, $revision, $message, $username, $repository); - $op->labels = $branches; - $op->insert($op_items); + for($i = 0; $i < $parent_count; ++$i) { + $revision = trim($prev_revisions[$i + 2]); + $item->source_items[] = new VersioncontrolGitItem( + VERSIONCONTROL_ITEM_FILE, + $item->path, + $revision, + null, + $item->repository + ); + } } -function _versioncontrol_git_parse_items($repository, &$logs, &$line, $revision, &$branches_per_commit, $parents, $merge) { +/** + * Takes parts of the output of git log and returns all affected OperationItems for a commit. + * @param VersioncontrolRepository $repository + * @param array $logs + * @param string $line + * @param string $revision + * @param array $parents The parent commit(s) + * @param bool $merge + * @return array All items affected by a commit. + */ +function _versioncontrol_git_parse_items($repository, &$logs, &$line, $revision, $parents, $merge) { $op_items = array(); - $read = FALSE; - - // Read file line revisions. + // Parse the diffstat for the changed files. do { - if (preg_match('/^(\S+)'."\t".'(\S+)'."\t".'(.+)$/', $line, $matches)) { // Begins with num lines added and matches expression. - $read = TRUE; - $path = '/'. $matches[3]; - $op_items[$path] = new VersioncontrolGitItem( - VERSIONCONTROL_ITEM_FILE, - $path, - $revision, - //'source_items' => array(),//filled later - ($merge ? VERSIONCONTROL_ACTION_MERGED : VERSIONCONTROL_ACTION_MODIFIED), - $repository - ); - if (is_numeric($matches[1]) && is_numeric($matches[2])) { - $op_items[$path]->line_changes = array( - 'added' => $matches[1], - 'removed' => $matches[2] - ); - } + if (!preg_match('/^(\S+)' . "\t" . '(\S+)' . "\t" . '(.+)$/', $line, $matches)) { + break; } - else { - break; + $path = '/'. $matches[3]; + $op_items[$path] = new VersioncontrolGitItem( + VERSIONCONTROL_ITEM_FILE, + $path, + $revision, + ($merge ? VERSIONCONTROL_ACTION_MERGED : VERSIONCONTROL_ACTION_MODIFIED), + $repository + ); + if (is_numeric($matches[1]) && is_numeric($matches[2])) { + $op_items[$path]->line_changes = array( + 'added' => $matches[1], + 'removed' => $matches[2] + ); } } while (($line = next($logs)) !== FALSE); - // Read file actions. + // Parse file actions. do { - if (preg_match('/^ (\S+) (\S+) (\S+) (.+)$/', $line, $matches)) { // Ensure that same file, they should be in same order. - $read = TRUE; - // We also can get 'mode' here if someone changes the file permissions. - if ($matches[1] == 'create') { - $op_items['/'. $matches[4]]->action = VERSIONCONTROL_ACTION_ADDED; - } - else if ($matches[1] == 'delete') { - $op_items['/'. $matches[4]]->action = VERSIONCONTROL_ACTION_DELETED; - } + if (!preg_match('/^ (\S+) (\S+) (\S+) (.+)$/', $line, $matches)) { + break; } - else { - break; + // We also can get 'mode' here if someone changes the file permissions. + if ($matches[1] == 'create') { + $op_items['/'. $matches[4]]->action = VERSIONCONTROL_ACTION_ADDED; + } + else if ($matches[1] == 'delete') { + $op_items['/'. $matches[4]]->action = VERSIONCONTROL_ACTION_DELETED; } } while (($line = next($logs)) !== FALSE); - //This is an inconsistency in git log output... - if ($read) { - $line = next($logs); - } + // Fill in the source_items for non-added items foreach ($op_items as $path => $item) { if ($item->action != VERSIONCONTROL_ACTION_ADDED) { - $op_items[$path]->source_items = _versioncontrol_git_get_source_item($repository, $path, $parents, $branches_per_commit[$revision], $revision); + _versioncontrol_git_fill_source_item($item, $parents); } } return $op_items; } -function _versioncontrol_git_check_already_parsed_commits($repository, $revision, &$branches_per_commit, &$line, &$logs) { - $constraints = array( - 'types' => array(VERSIONCONTROL_OPERATION_COMMIT), - 'revisions' => array($revision), - 'vcs' => array('git'), - 'repo_ids' => array($repository->repo_id) - ); - $same_rev_commit = VersioncontrolOperationCache::getInstance()->getOperations($constraints); - $adjusted_commit = FALSE; - foreach ($same_rev_commit as $vc_op_id => $rev_commit) { - // We already have a commit with this revision recorded, so use a faster parser then. - $adjusted_commit = TRUE; - $labels = array(); - foreach ( $rev_commit->labels as $label) { - if ($label['type'] == VERSIONCONTROL_OPERATION_TAG) { - $labels[] = $label; - } - } - $labels = array_merge($labels, $branches_per_commit[$revision]); - $rev_commit->updateLabels($labels); - $line = next($logs); // Get $parents - $line = next($logs); // Get Author - $line = next($logs); // Get Date as Timestamp - // Pretend message parsing - while (($line = next($logs)) !== FALSE) { - if (trim($line) == 'ENDOFOUTPUTGITMESSAGEHERE') { - break; - } - } - $line = next($logs); - // Skip everything --summary or --numstat related output - while (!(preg_match("/^([a-f0-9]{40})$/", trim($line))) && $line !== FALSE) { - $line = next($logs); - } - //$branches_per_commit[$revision] = TRUE; - } - return $adjusted_commit; -} - /** - * Parse the output of 'git log' and insert commits based on it's data. + * Parse the output of 'git log' and insert a commit based on it's data. * - * @param $repository - * The repository array, as given by the Version Control API. - * @param $logs The output of 'git log' to parse - * @param array $branches_per_commit An array which has all branches for all commits in it. - * It is used to construct $operation['labels']. + * @param VersioncontrolRepository $repository + * @param array $logs The output of 'git log' to parse + * @param array $commits_in_db + * @param array $branch_label_list An associative list of branchname => VersioncontrolBranch */ -function _versioncontrol_git_log_parse_commits($repository, &$logs, &$branches_per_commit, &$existing_revs) { - // If the log was retrieved by taking the return value of exec(), we've - // got an array and navigate it via next(). If we stored the log in a - // temporary file, $logs is a file handle that we need to fgets() instead. - $root_path = $repository['root']; - $line = next($logs); // Get Revision +function _versioncontrol_git_log_parse_and_insert_commit($repository, $logs, &$commits_in_db, $branch_label_list) { $merge = FALSE; - // $line already points to the revision - $revision = trim($line); - foreach ($branches_per_commit[$revision] as $label) { - $existing_revs[$label['name']][$revision] = TRUE; - } - - $adjusted_commit = _versioncontrol_git_check_already_parsed_commits($repository, - $revision, $branches_per_commit, $line, $logs); - if ($adjusted_commit) { - return; - } + // Get Revision + $revision = trim(next($logs)); - $line = next($logs); // Get $parents - $parents = explode(" ", trim($line)); + // Get $parents + $parents = explode(" ", trim(next($logs))); if ($parents[0] == '') { $parents = array(); } if (isset($parents[1])) { $merge = TRUE; } - $line = next($logs); // Get Author - $username = trim($line); - $line = next($logs); // Get Date as Timestamp - $date = trim($line); + + // Get Author + $username = trim(next($logs)); + + // Get Date as Timestamp + $date = trim(next($logs)); + // Get revision message. + // TODO: revisit! $message = ''; $i = 0; - while (($line = next($logs)) !== FALSE) { - $line = trim($line); + while (($line = trim(next($logs))) !== FALSE) { if ($line == 'ENDOFOUTPUTGITMESSAGEHERE') { if (substr($message, -2) === "\n\n") { $message = substr($message, 0, strlen($message) - 1); @@ -608,11 +370,252 @@ function _versioncontrol_git_log_parse_commits($repository, &$logs, &$branches_p $message .= $line ."\n"; $i++; } - $line = next($logs); // Points to either the next entry or the first items modified or to the file actions + + // This is either a (kind of) diffstat for each modified file or a list of + // file actions like moved, created, deleted, mode changed. + $line = next($logs); // Get the items $op_items = _versioncontrol_git_parse_items($repository, $logs, $line, $revision, - $branches_per_commit, $parents, $merge); + $parents, $merge); - _versioncontrol_git_insert_commit($repository, $date, $username, - $message, $revision, $branches_per_commit[$revision], $op_items); + $op = new VersioncontrolGitOperation(VERSIONCONTROL_OPERATION_COMMIT, $username, $date, $revision, $message, $username, $repository); + $op->labels = _versioncontrol_git_log_get_branches_of_commit($revision, $branch_label_list); + $op->insert($op_items); + + $commits_in_db[] = $revision; +} + + + +/// All branch related functions + +/** + * Returns a list of all existing branches in the database. + * @param VersioncontrolRepository $repository + * @return array An array of branch names. + */ +function _versioncontrol_git_log_get_branches_in_db($repository) { + return _versioncontrol_git_log_get_labels_in_db($repository, VERSIONCONTROL_LABEL_BRANCH); +} + +/** + * Get branches from Git using 'branch -l' command. + * @return array List of branches. + */ +function _versioncontrol_git_log_get_branches_in_repo() { + $logs = _versioncontrol_git_log_exec('git show-ref --heads'); // Query branches. + $branches = array(); + while (($line = next($logs)) !== FALSE) { + // the output of git show-ref --heads looks like this + // c5bbe46edb29eb6142942ba971a51dc9d5becbc0 refs/heads/master + // to get 'master', we have to skip 52 chars. + $branches[] = substr(trim($line), 52); + } + return $branches; +} + +/** + * Attaches the label $branch to every commit if it is not there yet. + * @param VersioncontrolBranch $branch + * @param array $commits + * @return none + */ +function _versioncontrol_git_log_attach_branch_to_commits($branch, $commits) { + $constraints = array( + 'vcs' => array('git'), + 'repo_ids' => array($branch->repository->repo_id), + 'types' => array(VERSIONCONTROL_OPERATION_COMMIT), + 'revisions' => $commits, + ); + $commits_as_op = VersioncontrolOperationCache::getInstance()->getOperations($constraints); + foreach($commits_as_op as $vc_op_id => $op) { + // We need this complicated logic to avoid adding a branch twice to a commit. + $already_there = false; + foreach($op->labels as $label) { + if($label->type == VERSIONCONTROL_LABEL_BRANCH && $label->name == $branch->name) { + $already_there = true; + } + } + if(!$already_there) { + $op->labels[] = $branch; + $op->updateLabels($op->labels); + } + } +} + +/** + * This function removes the given branch from all commits. + * @param VersioncontrolRepository $repository + * @param string $branch + * @return none + */ +function _versioncontrol_git_log_remove_branch_from_commits($repository, $branch) { + $constraints = array( + 'vcs' => array('git'), + 'repo_ids' => array($repository->repo_id), + 'types' => array(VERSIONCONTROL_OPERATION_COMMIT), + 'branches' => array($branch), + ); + $commits_as_op = VersioncontrolOperationCache::getInstance()->getOperations($constraints); + foreach($commits_as_op as $commit_op) { + $new_labels = array(); + foreach($commit_op->labels as $label) { + if($label->type == VERSIONCONTROL_LABEL_BRANCH && $label->name != $branch) { + $new_labels[] = $label; + } + } + $commit_op->updateLabels($new_labels); + } +} + + +/// Tag related functions + +/** + * Returns a list of all existing tags in the database. + * @param VersioncontrolRepository $repository + * @return array An array of tag names. + */ +function _versioncontrol_git_log_get_tags_in_db($repository) { + return _versioncontrol_git_log_get_labels_in_db($repository, VERSIONCONTROL_LABEL_TAG); +} + +/** + * Get all tags present in the repository. + * @return array + */ +function _versioncontrol_git_log_get_tags_in_repo() { + $log = _versioncontrol_git_log_exec('git show-ref --tags'); // Query tags. + $tags = array(); + while (($line = next($log)) !== FALSE) { + // the output of git show-ref --tags looks like this + // 94a5915923d5a9a6af935e4055c95582fd1a1136 refs/tags/DRUPAL-5--1-0 + // to get 'DRUPAL-5--1-0', we have to skip 51 chars. + $tags[] = substr(trim($line), 51); + } + return $tags; +} + + +/** + * Returns a string with fully qualified tag names from an array of tag names. + * @param array $tags + * @return string + */ +function _versioncontrol_git_get_tag_string($tags) { + $tag_string = ''; + // $tag_string is a list of fully qualified tag names + foreach ($tags as $tag) { + $tag_string .= escapeshellarg("refs/tags/$tag") . ' '; + } + return $tag_string; +} + +/** + * Returns a list of tag names with the tagged commits. + * Handles annotated tags. + * @param array $tags An array of tag names + * @return array A list of all tags with the respective tagged commit. + */ +function _versioncontrol_git_log_get_tag_commit_list($tags) { + if(empty($tags)) { + return array(); + } + $tag_string = _versioncontrol_git_get_tag_string($tags); + $exec = "git show-ref -d $tag_string"; + $tag_commit_list_raw = _versioncontrol_git_log_exec($exec); + $tag_commit_list = array(); + $tags_annotated = array(); + foreach($tag_commit_list_raw as $tag_commit_line) { + if($tag_commit_line == '') { + continue; + } + $tag_commit = substr($tag_commit_line, 0, 40); + // annotated tag mark + // 9c70f55549d3f4e70aaaf30c0697f704d02e9249 refs/tags/tag^{} + if (substr($tag_commit_line, -3, 3) == '^{}') { + $tag_name = substr($tag_commit_line, 51, -3); + $tags_annotated[$tag_name] = $tag_commit; + } + // Simple tags + // 9c70f55549d3f4e70aaaf30c0697f704d02e9249 refs/tags/tag + else { + $tag_name = substr($tag_commit_line, 51); + } + $tag_commit_list[$tag_name] = $tag_commit; + } + // Because annotated tags show up twice in the output of git show-ref, once + // with a 'tag' object and once with a commit-id we will go through them and + // adjust the array so we just keep the commits. + foreach($tags_annotated as $tag_name => $tag_commit) { + $tag_commit_list[$tag_name] = $tag_commit; + } + return $tag_commit_list; +} + +/** + * Does all processing to insert the tags in $tags_new in the database. + * @param VersioncontrolRepository $repository + * @param array $tags_new All new tags. + * @return none + */ +function _versioncontrol_git_log_process_tags($repository, $tags_new) { + if (empty($tags_new)) { + return array(); + } + $tag_ops = array(); + + // get a list of all tag names with the corresponding commit. + $tag_commit_list = _versioncontrol_git_log_get_tag_commit_list($tags_new); + $format = '%(objecttype)%0a%(objectname)%0a%(refname)%0a%(taggername) %(taggeremail)%0a%(taggerdate)%0a%(contents)ENDOFGITTAGOUTPUTMESAGEHERE'; + foreach($tag_commit_list as $tag_name => $tag_commit) { + $exec = "git for-each-ref --format=\"$format\" refs/tags/" . escapeshellarg($tag_name); + $logs_tag_msg = _versioncontrol_git_log_exec($exec); + $tagged_commit_op = _versioncontrol_git_log_get_commit($repository, $tag_commit); + // Get the specific tag data for annotated vs not annotated tags. + if($logs_tag_msg[1] == 'commit') { + // simple tag + // [2] is tagged commit [3] tagname [4] and [5] empty [6] commit log message + // We get the tagger, the tag_date and the tag_message from the tagged commit. + $tagger = $tagged_commit_op->author; + $tag_date = $tagged_commit_op->date + 1; + $message = $tagged_commit_op->message; + } + else if($logs_tag_msg[1] == 'tag') { + // annotated tag + // [2] is the tagged commit [3] tag name + $tagger = $logs_tag_msg[4]; + $tag_date = strtotime($logs_tag_msg[5]); + // Get the tag message + $message = ''; + $i = 0; + while (true) { + $line = $logs_tag_msg[$i + 6]; + if($logs_tag_msg[$i + 7] == 'ENDOFGITTAGOUTPUTMESAGEHERE') { + $message .= $line; + break; + } + $message .= $line ."\n"; + $i++; + } + } + else { + drupal_set_message(t('Serious problem in tag parsing, please check that you\'re using a supported version of git!', 'error')); + } + $tag_op = new VersioncontrolGitOperation ( + VERSIONCONTROL_OPERATION_TAG, + $tagger, + $tag_date, + $tag_commit, + $message, + $tagger, + $repository + ); + $tag_op->labels = array(new VersioncontrolTag($tag_name, VERSIONCONTROL_ACTION_ADDED, null, $repository)); + $empty_array = array(); + $tag_op->insert($empty_array); + // Update the tagged commit to include the label + $tagged_commit_op->labels[] = new VersioncontrolTag($tag_name, VERSIONCONTROL_ACTION_MODIFIED, null, $repository); + $tagged_commit_op->updateLabels($tagged_commit_op->labels); + } }