commits/batch/git-parse.php - dash/org.eclipse.dash.dashboard

#!/usr/bin/php

<?php

/*******************************************************************************

* are made available under the terms of the Eclipse Public License v1.0

* which accompanies this distribution, and is available at

* http://www.eclipse.org/legal/epl-v10.html

*

* Contributors:

* Wayne Beaton (Eclipse Foundation)- initial API and implementation

*******************************************************************************/

/**

* This script parses the output of the git log command (with the --numstat

* option to include file names) and generates the tabular output format

* required by Dash.

*

* Usage:

*

* git --git-dir=/gitroot/woolsey/iplog/.git/ log --numstat | ./git-parse.php

*

* Examples:

* echo -e "technology.dash.woolsey\t/gitroot/woolsey/iplog/.git/" | ./git-extract.php | ./git-parse.php

* wget http://www.eclipse.org/projects/web-api/roots-generator.php?type=git -O - 2> /dev/null | ./git-extract.php | ./git-parse.php

* wget http://www.eclipse.org/projects/web-api/roots-generator.php?type=git -O - 2> /dev/null | grep virgo | ./git-extract.php | ./git-parse.php | ./chunk.pl ./insert.pl

*

* @author Wayne Beaton

*/

$project = 'unknown';

$top = 'unknown';

$commit = 'unknown';

$login = 'unknown';

$company = 'unknown';

$date = 'unknown';

$year = 'unknown';

$yearmonth = 'unknown';

$yearmonthday = 'unknown';

$headers = array(

"DATE", "YEAR", "YEARMONTH","YEARMONTHDAY",

"TOPPROJECT","PROJECT","FILENAME","FILETYPE","REVISION",

"CHANGE_SIZE","MESSAGE_SIZE","LOGIN","COMPANY");

$projectNameSegmentPattern = "[a-zA-Z0-9\\-]+";

$projectNamePattern = "(([\w\-]+)(\.[\w\-]+){0,2})";

echo "#" . implode("\t", $headers) . "\n";

/*

* The logic here is this: We encounter the information

* in a line-by-line manner. As we encounter the basic information

* (like the committer identity, commit Id, etc), we remember it.

* As we encounter file names, we use the information we've collected

* so far to output a record for that file name.

*

* We assume that all the appropriate header information is always

* included as part of each block.

*

* Data comes to us in this form:

* ---------------------------------------------------------

* Project: technology.dash.woolsey

* commit 4e918dade5701c6cf03ad3608489ce00738fc8b5

* Author: Wayne Beaton <wayne@eclipse.org>

* AuthorDate: Thu Dec 16 23:53:58 2010 -0500

* Commit: Wayne Beaton <wayne@eclipse.org>

* CommitDate: Thu Dec 16 23:53:58 2010 -0500

*

* Bug 332692 Added support to capture date as part of submit request. Also added some rudimentary testing of input data and error reporting.

*

* 2 1 org.eclipse.woolsey.iplog.submit/META-INF/MANIFEST.MF

* 4 1 org.eclipse.woolsey.iplog.submit/futz.jpage

* 13 8 org.eclipse.woolsey.iplog.submit/src/org/eclipse/woolsey/iplog/submit/IpzillaClient.java

* 40 5 org.eclipse.woolsey.iplog.submit/src/org/eclipse/woolsey/iplog/submit/wizards/SubmitInfoWizardPage.java

* 60 8 org.eclipse.woolsey.iplog.submit/src/org/eclipse/woolsey/iplog/submit/wizards/SubmitOperation.java

* ---------------------------------------------------------

*

* So, as we encounter the Project, commit, Author, and Date lines, we

* record that information. It's only when we hit the file name lines

* (at the bottom) that we actually write anything out.

*/

while (!feof(STDIN)) {

$line = fgets(STDIN);

// For example:

// Project: technology.dash.woolsey

if (preg_match("/^Project: (([\w\-]+)(\.[\w\-]+){0,2})$/i", $line, $matches)) {

$project = $matches[1];

$top = $matches[2];

// Just to be safe and make sure that information doesn't

// bleed from one project to the next, we reset everything.

$commit = 'unknown';

$login = 'unknown';

$company = 'unknown';

$date = 'unknown';

$year = 'unknown';

$yearmonth = 'unknown';

$yearmonthday = 'unknown';

}

// For example:

// commit 4e918dade5701c6cf03ad3608489ce00738fc8b5

if (preg_match('/^commit ([a-f0-9]+)$/i', $line, $matches)) {

$commit = $matches[1];

// Just to be safe and make sure that information doesn't

// bleed from one commit to the next, we reset everything.

$login = 'unknown';

$company = 'unknown';

$date = 'unknown';

$year = 'unknown';

$yearmonth = 'unknown';

$yearmonthday = 'unknown';

}

// Note that the committer could be an email address or

// a committer id (we expect that it mostly likely an email

// address.

//

// For example:

// Commit: Wayne Beaton <wayne@eclipse.org>

// Commit: wbeaton

// Commit: spingel <>

// Commit: Steffen Pingel <steffen.pingel@tasktop.com>

// Commit: steffen.pingel@tasktop.com

if (preg_match('/^Commit:/', $line)) {

// First, look for a name between <>

if (preg_match('/<([^>]+)>/i', $line, $matches)) {

$login = getCommitterId(trim($matches[1]));

$company = getCommitterCompany($login);

} else

// Then, try to grab the first word after "Commit:" and hope for the best.

if (preg_match('/^Commit:\s*(\w+)/', $line, $matches)) {

$login = getCommitterId($matches[1]);

$company = getCommitterCompany($login);

}

// Otherwise, fail miserably.

}

// For example:

// CommitDate: Thu Dec 16 23:53:58 2010 -0500

if (preg_match('/^CommitDate: (.*)$/', $line, $matches)) {

$date = strtotime(trim($matches[1]));

$year = date('Y', $date);

$yearmonth = date('Ym', $date);

$yearmonthday = date('Ymd', $date);

$date = date("Y/m/d", $date);

}

// For example:

// 2 1 org.eclipse.woolsey.iplog.submit/META-INF/MANIFEST.MF

if (preg_match('/^([0-9]+)\s+([0-9]+)\s+(\w.*)$/', $line, $matches)) {

if ((int)$year < 2000) continue; // Ignore weird data (see Bug 333620).

$added = $matches[1];

$removed = $matches[2];

$change_size = $added + $removed;

$message_size = 0;

$filename = $matches[3];

$filetype = getFileType($filename);

echo "$date\t$year\t$yearmonth\t$yearmonthday\t$top\t$project\t$filename\t$filetype\t$commit\t$change_size\t$message_size\t$login\t$company\n";

}

// TODO Move to a "common" import

function getFileType($filename) {

if (preg_match('/.*\.(\w+)$/', $filename, $matches)) {

$extension = $matches[1];

if ($extension == 'htm') return 'html';

if ($extension == 'jpg') return 'jpeg';

return $extension;

}

return 'unknown';

}

function getCommitterId($address) {

$address = strtolower($address);

// If it's an email address, translate to a committer id.

$map = getEmailToCommitterMap();

if (isset($map[$address])) {

$id = $map[$address];

return $id;

}

/*

* If it looks like a committer id followed by a UUID, then

* it's probably a commit that's been migrated from CVS. Grab

* the committer id.

*

* e.g. dschaefer@6a79697e-3843-0410-8446-a9668620458d

*/

if (preg_match('/^(\w+)@[\w\-]+$/', $address, $matches)) {

return $matches[1];

}

return $address;

}

function &getEmailToCommitterMap() {

global $_emailToCommitterMap;

if ($_emailToCommitterMap) return $_emailToCommitterMap;

$_emailToCommitterMap = array();

$file = fopen('https://www.eclipse.org/projects/web-api/email-id-map.php', 'r');

if (!$file) return;

while (!feof($file)) {

$line = fgets($file);

$parts = split("\t", $line);

$email = strtolower(trim($parts[0]));

$id = trim($parts[1]);

$_emailToCommitterMap[$email] = $id;

}

fclose($file);

return $_emailToCommitterMap;

}

function getCommitterCompany($id) {

$id = strtolower($id);

$map = getCommitterToCompanyMap();

if (isset($map[$id])) {

$company = $map[$id];

if ($company) return $company;

}

return 'unknown';

}

function &getCommitterToCompanyMap() {

global $_committerToCompanyMap;

if ($_committerToCompanyMap) return $_committerToCompanyMap;

$_committerToCompanyMap = array();

$file = fopen('https://www.eclipse.org/projects/web-api/commit-companies.php', 'r');

if (!$file) return;

while (!feof($file)) {

$line = fgets($file);

$parts = split("\t", $line);

$id = strtolower(trim($parts[0]));

$company = trim($parts[1]);

$_committerToCompanyMap[$id] = $company;

}

fclose($file);

return $_committerToCompanyMap;

}

?>