| 1 |
<?php
|
| 2 |
// $Id: versioncontrol_hg.log.inc,v 1.9 2008/02/03 03:24:35 ezyang Exp $
|
| 3 |
|
| 4 |
/**
|
| 5 |
* @file
|
| 6 |
* Mercurial backend for Version Control API - Provides Mercurial commit
|
| 7 |
* information and account management as a pluggable backend.
|
| 8 |
*
|
| 9 |
* This file defines internal functions for processing logs.
|
| 10 |
*
|
| 11 |
* Copyright 2008 by Edward Z. Yang (ezyang, http://drupal.org/user/211688)
|
| 12 |
*/
|
| 13 |
|
| 14 |
/**
|
| 15 |
* Update repository's database tables by interfacing with Mercurial.
|
| 16 |
*
|
| 17 |
* @warning
|
| 18 |
* This code should not directly interface with the Mercurial
|
| 19 |
* executable; the hg_* functions defined in hg/hg.inc should be used
|
| 20 |
* instead. If you need functionality not yet defined, submit a patch
|
| 21 |
* for hg/hg.inc implementing it.
|
| 22 |
*
|
| 23 |
* @warning
|
| 24 |
* This function does not account for memory exhaustion.
|
| 25 |
* An alternative implementation could read each log entry one by
|
| 26 |
* one from the file (as CVS's implementation does), however, the
|
| 27 |
* resulting data-structure would still cause large amounts of memory
|
| 28 |
* usage (this is a primary deficiency of CVS) and would require
|
| 29 |
* multiple database queries. After determining which heuristics
|
| 30 |
* need to be able to access previous and later log entries, we will
|
| 31 |
* refactor accordingly; an alternate implementation would be to
|
| 32 |
* limit rev_range so that only batches of the logs are parsed at
|
| 33 |
* a time. Nevertheless, real world usage data is necessary before
|
| 34 |
* we can make a decision in this respect.
|
| 35 |
*
|
| 36 |
* @param $repository
|
| 37 |
* Repository variable as returned by versioncontrol_get_repository()
|
| 38 |
*/
|
| 39 |
function _versioncontrol_hg_log_update_repository(&$repository) {
|
| 40 |
|
| 41 |
// Because we only support local repositories, `hg pull` is NOT called
|
| 42 |
// before we perform these changes. It is the responsibility of all
|
| 43 |
// branch maintainers to push their changes to this repository so that
|
| 44 |
// their changes are registered.
|
| 45 |
|
| 46 |
$latest_rev = $repository['hg_specific']['latest_rev'];
|
| 47 |
if ($latest_rev === false) {
|
| 48 |
// This is necessary to ensure that the revisions are ordered
|
| 49 |
// properly, from earliest to latest. We need logs in this order
|
| 50 |
// because some algorithms reference earlier information.
|
| 51 |
$rev_range = '0:';
|
| 52 |
}
|
| 53 |
else {
|
| 54 |
$rev_range = ($latest_rev + 1) .':';
|
| 55 |
}
|
| 56 |
$raw_logs = hg_log($repository['root'], $rev_range);
|
| 57 |
// watchdog('special', hg_last_cmd());
|
| 58 |
|
| 59 |
if (empty($raw_logs)) {
|
| 60 |
// No new logs, abort.
|
| 61 |
return;
|
| 62 |
}
|
| 63 |
|
| 64 |
// In the future, this could be replaced with a one-way iterator
|
| 65 |
// which would let us process log entries piecemeal.
|
| 66 |
foreach ($raw_logs as $log) {
|
| 67 |
|
| 68 |
_versioncontrol_hg_log_preprocess_log($log);
|
| 69 |
|
| 70 |
// For branches and tags, the second parameter is invariably array();
|
| 71 |
// we might want to move it out to simplify the if-then logic.
|
| 72 |
|
| 73 |
// Branch handling must occur before commits
|
| 74 |
$raw_branch = _versioncontrol_hg_log_parse_branch_operation($log, $repository);
|
| 75 |
if ($raw_branch !== false) {
|
| 76 |
list($branch, $branched_items) = $raw_branch;
|
| 77 |
versioncontrol_insert_branch_operation($branch, $branched_items);
|
| 78 |
}
|
| 79 |
|
| 80 |
// Multiple tags operations can occur in one commit
|
| 81 |
$tags = _versioncontrol_hg_log_parse_tag_operation($log, $repository);
|
| 82 |
if ($tags !== false) {
|
| 83 |
foreach ($tags as $tag) {
|
| 84 |
versioncontrol_insert_tag_operation($tag, array());
|
| 85 |
}
|
| 86 |
}
|
| 87 |
|
| 88 |
// Commit handling
|
| 89 |
list($commit, $commit_actions) = _versioncontrol_hg_log_parse_commit($log, $repository);
|
| 90 |
versioncontrol_insert_commit($commit, $commit_actions);
|
| 91 |
|
| 92 |
}
|
| 93 |
|
| 94 |
// This depends on the previous foreach loop
|
| 95 |
$latest_rev = $log['rev'];
|
| 96 |
|
| 97 |
// Update latest revision
|
| 98 |
$repository['hg_specific']['latest_rev'] = $latest_rev;
|
| 99 |
db_query('UPDATE {versioncontrol_hg_repositories}
|
| 100 |
SET latest_rev = %d WHERE repo_id = %d',
|
| 101 |
$repository['hg_specific']['latest_rev'], $repository['repo_id']);
|
| 102 |
|
| 103 |
}
|
| 104 |
|
| 105 |
/**
|
| 106 |
* Perform common pre-processing on a log entry from hg_logs(), although no
|
| 107 |
* major changes to the array's structure.
|
| 108 |
*
|
| 109 |
* @warning
|
| 110 |
* This function MUST NOT be lossy or fundamentally change the log
|
| 111 |
* structure; any results of this pre-processing should be stored in
|
| 112 |
* unique keys prefixed with an underscore, so as to prevent collisions
|
| 113 |
* with the keys specified in hg/templates/changeset.tmpl.
|
| 114 |
*
|
| 115 |
* @param
|
| 116 |
* Raw logs variable to pre-process. The following additional indexes
|
| 117 |
* will be added:
|
| 118 |
*
|
| 119 |
* '_date' => String UTC Unix timestamp of changeset
|
| 120 |
* '_offset' => Integer time zone offset of commiter
|
| 121 |
* '_manifest' => Array parent nodeid
|
| 122 |
* '_parents' => array(0 => First parent nodeid, 1 => Second parent nodeid)
|
| 123 |
* '_author' => String author name w/o email
|
| 124 |
* '_email' => String author's email
|
| 125 |
* '_branch' => Either the contents of branches, or 'default'
|
| 126 |
*
|
| 127 |
* Nodeids take the form of array(string revision number, string SHA-1 hash)
|
| 128 |
*/
|
| 129 |
function _versioncontrol_hg_log_preprocess_log(&$log) {
|
| 130 |
|
| 131 |
// Parse Mercurial's log format into Unix timestamp and offset.
|
| 132 |
// We do NOT convert the timestamp into an integer to prevent overflow
|
| 133 |
list($log['_date'], $offset) = explode('.', $log['date']);
|
| 134 |
$log['_offset'] = (int) $offset;
|
| 135 |
|
| 136 |
// Split manifest nodeid (note that we are not converting the revision
|
| 137 |
// number to an integer either.)
|
| 138 |
$log['_manifest'] = explode(':', $log['manifest']);
|
| 139 |
|
| 140 |
// Split parents, and then split their nodeids
|
| 141 |
$parents = explode(' ', $log['parents']);
|
| 142 |
foreach ($parents as $k => $v) {
|
| 143 |
$log['_parents'][$k] = explode(':', $v);
|
| 144 |
}
|
| 145 |
|
| 146 |
// Parse author into author name and email
|
| 147 |
// This is somewhat fragile, but should work as long as a user does
|
| 148 |
// not have '>' in their name! :-)
|
| 149 |
$bits = explode('<', $log['author']);
|
| 150 |
$log['_author'] = rtrim($bits[0]);
|
| 151 |
if (!empty($bits[1])) {
|
| 152 |
$log['_email'] = rtrim($bits[1], '>');
|
| 153 |
}
|
| 154 |
else {
|
| 155 |
$log['_email'] = null;
|
| 156 |
}
|
| 157 |
|
| 158 |
$log['_branch'] = ($log['branches'] !== '') ? $log['branches'] : 'default';
|
| 159 |
|
| 160 |
}
|
| 161 |
|
| 162 |
/**
|
| 163 |
* Parses an array item of hg_log() into form ready for
|
| 164 |
* versioncontrol_insert_commit().
|
| 165 |
*
|
| 166 |
* @param $log
|
| 167 |
* Item of array from hg_log() to be parsed.
|
| 168 |
* @param $repository
|
| 169 |
* Repository array as per versioncontrol_get_repository().
|
| 170 |
* @return
|
| 171 |
* Array in form array(0 => $commit, 1 => $commit_actions)
|
| 172 |
* which are ready for insertion using versioncontrol_insert_commit().
|
| 173 |
*/
|
| 174 |
function _versioncontrol_hg_log_parse_commit($log, $repository) {
|
| 175 |
|
| 176 |
// Build $commit:
|
| 177 |
$commit = _versioncontrol_hg_log_parse_operation($log, $repository);
|
| 178 |
$commit['message'] = $log['desc'];
|
| 179 |
$commit['revision'] = $log['node'];
|
| 180 |
$commit['hg_specific'] = array(
|
| 181 |
'branch_id' => versioncontrol_ensure_branch($log['_branch'], $repository['repo_id']),
|
| 182 |
'rev' => $log['rev'],
|
| 183 |
// Using key [1] to retrieve only SHA-1 hash; the revision number
|
| 184 |
// should be the same as 'rev'
|
| 185 |
'manifest' => $log['_manifest'][1],
|
| 186 |
// We'll need this in order to look up the vc_op_id. In
|
| 187 |
// this case, we use revision because it's easier to handle and
|
| 188 |
// this is all internal processing.
|
| 189 |
'parent1' => $log['_parents'][0][0],
|
| 190 |
'parent2' => $log['_parents'][1][0],
|
| 191 |
);
|
| 192 |
|
| 193 |
// Build $commit_actions:
|
| 194 |
|
| 195 |
// Under very special conditions (basically, when a merge adds
|
| 196 |
// a file, but no edits were made to it), file_adds can contain
|
| 197 |
// files that are not in files. So, we need to merge these two
|
| 198 |
// arrays.
|
| 199 |
$files = array_unique(array_merge($log['files'], $log['file_adds']));
|
| 200 |
|
| 201 |
// These lookups must be tested with isset(), NOT !empty()
|
| 202 |
$lookup_adds = array_flip($log['file_adds']);
|
| 203 |
$lookup_dels = array_flip($log['file_dels']);
|
| 204 |
$lookup_copies = array_flip($log['file_copies']);
|
| 205 |
|
| 206 |
$commit_actions = array();
|
| 207 |
foreach ($files as $raw_file) {
|
| 208 |
$commit_action = array();
|
| 209 |
|
| 210 |
// Normalize Mercurial's file declaration "CVS" style to accomodate
|
| 211 |
// commitlog.
|
| 212 |
$file = "/$raw_file";
|
| 213 |
|
| 214 |
// CVS's implementation involves building a large array with
|
| 215 |
// everything you could possibly need, and then excising things
|
| 216 |
// based on options. While that approach may take less code,
|
| 217 |
// it is less robust to change.
|
| 218 |
|
| 219 |
// We need to use $raw_file to reference items in the logs.
|
| 220 |
if (isset($lookup_adds[$raw_file])) {
|
| 221 |
$commit_action['action'] = VERSIONCONTROL_ACTION_ADDED;
|
| 222 |
}
|
| 223 |
elseif (isset($lookup_dels[$raw_file])) {
|
| 224 |
$commit_action['action'] = VERSIONCONTROL_ACTION_DELETED;
|
| 225 |
}
|
| 226 |
elseif (isset($lookup_copies[$raw_file])) {
|
| 227 |
// This doesn't ever seem to happen.
|
| 228 |
$commit_action['action'] = VERSIONCONTROL_ACTION_COPIED;
|
| 229 |
}
|
| 230 |
else {
|
| 231 |
// File was modified
|
| 232 |
$commit_action['action'] = VERSIONCONTROL_ACTION_MODIFIED;
|
| 233 |
}
|
| 234 |
|
| 235 |
// Logs do not have information for 'modified', because
|
| 236 |
// file_copies is never filled with anything.
|
| 237 |
$commit_action['modified'] = FALSE;
|
| 238 |
|
| 239 |
$commit_action['current item'] = array();
|
| 240 |
|
| 241 |
// This isn't actually used, but we set it for completeness-sake.
|
| 242 |
$commit_action['current item']['path'] = $file;
|
| 243 |
|
| 244 |
if ($commit_action['action'] != VERSIONCONTROL_ACTION_DELETED) {
|
| 245 |
$commit_action['current item']['type'] = VERSIONCONTROL_ITEM_FILE;
|
| 246 |
}
|
| 247 |
else {
|
| 248 |
$commit_action['current item']['type'] = VERSIONCONTROL_ITEM_FILE_DELETED;
|
| 249 |
}
|
| 250 |
|
| 251 |
// This will be set later in versioncontrol_hg_commit(), because we
|
| 252 |
// need to ask the database for some information (the log doesn't tell
|
| 253 |
// us very much about this.)
|
| 254 |
$commit_action['source items'] = array();
|
| 255 |
|
| 256 |
$commit_actions[$file] = $commit_action;
|
| 257 |
|
| 258 |
}
|
| 259 |
|
| 260 |
return array($commit, $commit_actions);
|
| 261 |
}
|
| 262 |
|
| 263 |
/**
|
| 264 |
* Parse an array item of hg_log() into form ready for
|
| 265 |
* versioncontrol_insert_branch_operation().
|
| 266 |
*
|
| 267 |
* @return
|
| 268 |
* Array in form of array($branch, $branched_items = array()), with
|
| 269 |
* variables corresponding to parameters of versioncontrol_insert_branch_operation()
|
| 270 |
*/
|
| 271 |
function _versioncontrol_hg_log_parse_branch_operation($log, $repository) {
|
| 272 |
$branch_id = versioncontrol_get_branch_id($log['_branch'], $repository['repo_id'], TRUE);
|
| 273 |
if ($branch_id !== null) {
|
| 274 |
return false;
|
| 275 |
}
|
| 276 |
$commit = _versioncontrol_hg_log_parse_operation($log, $repository);
|
| 277 |
$commit['branch_name'] = $log['_branch'];
|
| 278 |
$commit['action'] = VERSIONCONTROL_ACTION_ADDED;
|
| 279 |
// All items are affected, so there's no need to check.
|
| 280 |
versioncontrol_ensure_branch($log['_branch'], $repository['repo_id']);
|
| 281 |
return array($commit, array());
|
| 282 |
}
|
| 283 |
|
| 284 |
/**
|
| 285 |
* Parses an array item of hg_log() into form ready for
|
| 286 |
* versioncontrol_insert_tag_operation().
|
| 287 |
*/
|
| 288 |
function _versioncontrol_hg_log_parse_tag_operation($log, $repository) {
|
| 289 |
if (!in_array('.hgtags', $log['files']) && !in_array('.hgtags', $log['file_adds'])) {
|
| 290 |
// No changes to .hgtags, so no tag operations
|
| 291 |
return false;
|
| 292 |
}
|
| 293 |
// We need to grab the contents of .hgtags at that revision
|
| 294 |
$hgtags = hg_cat($repository['root'], '.hgtags', $log['rev']);
|
| 295 |
$tags = _hg_parse_hgtags($hgtags);
|
| 296 |
|
| 297 |
// Now, lets grab the old set of tags
|
| 298 |
$result = db_query(
|
| 299 |
"SELECT t.name, c.revision FROM {versioncontrol_hg_tags} t
|
| 300 |
JOIN {versioncontrol_commits} c ON t.vc_op_id = c.vc_op_id
|
| 301 |
WHERE t.repo_id = %d
|
| 302 |
",
|
| 303 |
$repository['repo_id']
|
| 304 |
);
|
| 305 |
$old_tags = array();
|
| 306 |
while ($row = db_fetch_object($result)) {
|
| 307 |
$old_tags[$row->name] = $row->revision;
|
| 308 |
}
|
| 309 |
|
| 310 |
// Now, let's determine the changes
|
| 311 |
// We purposely never use VERSIONCONTROL_ACTION_MOVED because
|
| 312 |
// Mercurial doesn't think of tags this way; it'd be represented
|
| 313 |
// as an add and a delete.
|
| 314 |
|
| 315 |
$ret = array();
|
| 316 |
|
| 317 |
// This is to be copied for each tag operation we make.
|
| 318 |
$default = _versioncontrol_hg_log_parse_operation($log, $repository);
|
| 319 |
$default['message'] = $commit['des'];
|
| 320 |
|
| 321 |
foreach ($old_tags as $name => $nodeid) {
|
| 322 |
$op = $default;
|
| 323 |
$op['tag_name'] = $name;
|
| 324 |
if (!isset($tags[$name])) {
|
| 325 |
$op['action'] = VERSIONCONTROL_ACTION_DELETED;
|
| 326 |
}
|
| 327 |
elseif ($nodeid != $tags[$name]) {
|
| 328 |
$op['action'] = VERSIONCONTROL_ACTION_DELETED;
|
| 329 |
}
|
| 330 |
else {
|
| 331 |
// Everything's good!
|
| 332 |
unset($tags[$name]);
|
| 333 |
continue;
|
| 334 |
}
|
| 335 |
$ret[] = $op;
|
| 336 |
}
|
| 337 |
|
| 338 |
foreach ($tags as $name => $nodeid) {
|
| 339 |
$op = $default;
|
| 340 |
$op['tag_name'] = $name;
|
| 341 |
$op['action'] = VERSIONCONTROL_ACTION_ADDED;
|
| 342 |
$ret[] = $op;
|
| 343 |
}
|
| 344 |
|
| 345 |
return $ret;
|
| 346 |
|
| 347 |
}
|
| 348 |
|
| 349 |
/**
|
| 350 |
* Returns an operation array with the common values for all types
|
| 351 |
* of operations (commit, branch, tag) instantiated.
|
| 352 |
*/
|
| 353 |
function _versioncontrol_hg_log_parse_operation($log, $repository) {
|
| 354 |
return array(
|
| 355 |
'repository' => $repository,
|
| 356 |
'date' => $log['_date'],
|
| 357 |
'username' => $log['_email'],
|
| 358 |
);
|
| 359 |
}
|
Parent Directory
| 
Revision Log
| 
Revision Graph