@recaptime-dev's working patches + fork for Phorge, a community fork of Phabricator. (Upstream dev and stable branches are at upstream/main and upstream/stable respectively.)
hq.recaptime.dev/wiki/Phorge
phorge
phabricator
Move Differential commit message parsing to a separate, tested class
Summary: Ref T2222. We have a hunk of logic that purely does text parsing here; separate it and get coverage on it.
Test Plan:
- Ran new unit tests.
- Used `differential.parsecommitmessage`.
Reviewers: btrahan
Reviewed By: btrahan
CC: aran
Maniphest Tasks: T2222
Differential Revision: https://secure.phabricator.com/D8444
+376
-95
+3
src/__phutil_library_map__.php
+3
src/__phutil_library_map__.php
···
356
356
'DifferentialCommentPreviewController' => 'applications/differential/controller/DifferentialCommentPreviewController.php',
357
357
'DifferentialCommentQuery' => 'applications/differential/query/DifferentialCommentQuery.php',
358
358
'DifferentialCommentSaveController' => 'applications/differential/controller/DifferentialCommentSaveController.php',
359
+
'DifferentialCommitMessageParser' => 'applications/differential/parser/DifferentialCommitMessageParser.php',
360
+
'DifferentialCommitMessageParserTestCase' => 'applications/differential/parser/__tests__/DifferentialCommitMessageParserTestCase.php',
359
361
'DifferentialCommitsField' => 'applications/differential/customfield/DifferentialCommitsField.php',
360
362
'DifferentialCommitsFieldSpecification' => 'applications/differential/field/specification/DifferentialCommitsFieldSpecification.php',
361
363
'DifferentialConflictsFieldSpecification' => 'applications/differential/field/specification/DifferentialConflictsFieldSpecification.php',
···
2919
2921
'DifferentialCommentPreviewController' => 'DifferentialController',
2920
2922
'DifferentialCommentQuery' => 'PhabricatorOffsetPagedQuery',
2921
2923
'DifferentialCommentSaveController' => 'DifferentialController',
2924
+
'DifferentialCommitMessageParserTestCase' => 'PhabricatorTestCase',
2922
2925
'DifferentialCommitsField' => 'DifferentialCustomField',
2923
2926
'DifferentialCommitsFieldSpecification' => 'DifferentialFieldSpecification',
2924
2927
'DifferentialConflictsFieldSpecification' => 'DifferentialFieldSpecification',
+10
-95
src/applications/differential/conduit/ConduitAPI_differential_parsecommitmessage_Method.php
+10
-95
src/applications/differential/conduit/ConduitAPI_differential_parsecommitmessage_Method.php
···
89
89
foreach ($aux_fields as $key => $aux_field) {
90
90
$labels = $aux_field->getSupportedCommitMessageLabels();
91
91
foreach ($labels as $label) {
92
-
$normal_label = strtolower($label);
92
+
$normal_label = DifferentialCommitMessageParser::normalizeFieldLabel(
93
+
$label);
93
94
if (!empty($label_map[$normal_label])) {
94
95
$previous = $label_map[$normal_label];
95
96
throw new Exception(
···
102
103
return $label_map;
103
104
}
104
105
105
-
private function buildLabelRegexp(array $label_map) {
106
-
$field_labels = array_keys($label_map);
107
-
foreach ($field_labels as $key => $label) {
108
-
$field_labels[$key] = preg_quote($label, '/');
109
-
}
110
-
$field_labels = implode('|', $field_labels);
111
-
112
-
$field_pattern = '/^(?P<field>'.$field_labels.'):(?P<text>.*)$/i';
113
-
114
-
return $field_pattern;
115
-
}
116
106
117
107
private function parseCommitMessage($corpus, array $label_map) {
118
-
$label_regexp = $this->buildLabelRegexp($label_map);
108
+
$parser = id(new DifferentialCommitMessageParser())
109
+
->setLabelMap($label_map)
110
+
->setTitleKey('title')
111
+
->setSummaryKey('summary');
119
112
120
-
// Note, deliberately not populating $seen with 'title' because it is
121
-
// optional to include the 'Title:' label. We're doing a little special
122
-
// casing to consume the first line as the title regardless of whether you
123
-
// label it as such or not.
124
-
$field = 'title';
113
+
$result = $parser->parseCorpus($corpus);
125
114
126
-
$seen = array();
127
-
$lines = explode("\n", trim($corpus));
128
-
$field_map = array();
129
-
foreach ($lines as $key => $line) {
130
-
$match = null;
131
-
if (preg_match($label_regexp, $line, $match)) {
132
-
$lines[$key] = trim($match['text']);
133
-
$field = $label_map[strtolower($match['field'])];
134
-
if (!empty($seen[$field])) {
135
-
$this->errors[] = "Field '{$field}' occurs twice in commit message!";
136
-
}
137
-
$seen[$field] = true;
138
-
}
139
-
$field_map[$key] = $field;
115
+
foreach ($parser->getErrors() as $error) {
116
+
$this->errors[] = $error;
140
117
}
141
118
142
-
$fields = array();
143
-
foreach ($lines as $key => $line) {
144
-
$fields[$field_map[$key]][] = $line;
145
-
}
146
-
147
-
// This is a piece of special-cased magic which allows you to omit the
148
-
// field labels for "title" and "summary". If the user enters a large block
149
-
// of text at the beginning of the commit message with an empty line in it,
150
-
// treat everything before the blank line as "title" and everything after
151
-
// as "summary".
152
-
if (isset($fields['title']) && empty($fields['summary'])) {
153
-
$lines = $fields['title'];
154
-
for ($ii = 0; $ii < count($lines); $ii++) {
155
-
if (strlen(trim($lines[$ii])) == 0) {
156
-
break;
157
-
}
158
-
}
159
-
if ($ii != count($lines)) {
160
-
$fields['title'] = array_slice($lines, 0, $ii);
161
-
$fields['summary'] = array_slice($lines, $ii);
162
-
}
163
-
}
164
-
165
-
// Implode all the lines back into chunks of text.
166
-
foreach ($fields as $name => $lines) {
167
-
$data = rtrim(implode("\n", $lines));
168
-
$data = ltrim($data, "\n");
169
-
$fields[$name] = $data;
170
-
}
171
-
172
-
// This is another piece of special-cased magic which allows you to
173
-
// enter a ridiculously long title, or just type a big block of stream
174
-
// of consciousness text, and have some sort of reasonable result conjured
175
-
// from it.
176
-
if (isset($fields['title'])) {
177
-
$terminal = '...';
178
-
$title = $fields['title'];
179
-
$short = phutil_utf8_shorten($title, 250, $terminal);
180
-
if ($short != $title) {
181
-
182
-
// If we shortened the title, split the rest into the summary, so
183
-
// we end up with a title like:
184
-
//
185
-
// Title title tile title title...
186
-
//
187
-
// ...and a summary like:
188
-
//
189
-
// ...title title title.
190
-
//
191
-
// Summary summary summary summary.
192
-
193
-
$summary = idx($fields, 'summary', '');
194
-
$offset = strlen($short) - strlen($terminal);
195
-
$remainder = ltrim(substr($fields['title'], $offset));
196
-
$summary = '...'.$remainder."\n\n".$summary;
197
-
$summary = rtrim($summary, "\n");
198
-
199
-
$fields['title'] = $short;
200
-
$fields['summary'] = $summary;
201
-
}
202
-
}
203
-
204
-
return $fields;
119
+
return $result;
205
120
}
206
121
207
122
+210
src/applications/differential/parser/DifferentialCommitMessageParser.php
+210
src/applications/differential/parser/DifferentialCommitMessageParser.php
···
1
+
<?php
2
+
3
+
/**
4
+
* Parses commit messages (containing relaively freeform text with textual
5
+
* field labels) into a dictionary of fields.
6
+
*
7
+
* $parser = id(new DifferentialCommitMessageParser())
8
+
* ->setLabelMap($label_map)
9
+
* ->setTitleKey($key_title)
10
+
* ->setSummaryKey($key_summary);
11
+
*
12
+
* $fields = $parser->parseCorpus($corpus);
13
+
* $errors = $parser->getErrors();
14
+
*
15
+
* This is used by Differential to parse messages entered from the command line.
16
+
*
17
+
* @task config Configuring the Parser
18
+
* @task parse Parsing Messages
19
+
* @task support Support Methods
20
+
* @task internal Internals
21
+
*/
22
+
final class DifferentialCommitMessageParser {
23
+
24
+
private $labelMap;
25
+
private $titleKey;
26
+
private $summaryKey;
27
+
private $errors;
28
+
29
+
30
+
/* -( Configuring the Parser )--------------------------------------------- */
31
+
32
+
33
+
/**
34
+
* @task config
35
+
*/
36
+
public function setLabelMap(array $label_map) {
37
+
$this->labelMap = $label_map;
38
+
return $this;
39
+
}
40
+
41
+
42
+
/**
43
+
* @task config
44
+
*/
45
+
public function setTitleKey($title_key) {
46
+
$this->titleKey = $title_key;
47
+
return $this;
48
+
}
49
+
50
+
51
+
/**
52
+
* @task config
53
+
*/
54
+
public function setSummaryKey($summary_key) {
55
+
$this->summaryKey = $summary_key;
56
+
return $this;
57
+
}
58
+
59
+
60
+
/* -( Parsing Messages )--------------------------------------------------- */
61
+
62
+
63
+
/**
64
+
* @task parse
65
+
*/
66
+
public function parseCorpus($corpus) {
67
+
$this->errors = array();
68
+
69
+
$label_map = $this->labelMap;
70
+
$key_title = $this->titleKey;
71
+
$key_summary = $this->summaryKey;
72
+
73
+
if (!$key_title || !$key_summary || ($label_map === null)) {
74
+
throw new Exception(
75
+
pht(
76
+
'Expected labelMap, summaryKey and titleKey to be set before '.
77
+
'parsing a corpus.'));
78
+
}
79
+
80
+
$label_regexp = $this->buildLabelRegexp($label_map);
81
+
82
+
// NOTE: We're special casing things here to make the "Title:" label
83
+
// optional in the message.
84
+
$field = $key_title;
85
+
86
+
$seen = array();
87
+
$lines = explode("\n", trim($corpus));
88
+
$field_map = array();
89
+
foreach ($lines as $key => $line) {
90
+
$match = null;
91
+
if (preg_match($label_regexp, $line, $match)) {
92
+
$lines[$key] = trim($match['text']);
93
+
$field = $label_map[self::normalizeFieldLabel($match['field'])];
94
+
if (!empty($seen[$field])) {
95
+
$this->errors[] = pht(
96
+
'Field "%s" occurs twice in commit message!',
97
+
$field);
98
+
}
99
+
$seen[$field] = true;
100
+
}
101
+
$field_map[$key] = $field;
102
+
}
103
+
104
+
$fields = array();
105
+
foreach ($lines as $key => $line) {
106
+
$fields[$field_map[$key]][] = $line;
107
+
}
108
+
109
+
// This is a piece of special-cased magic which allows you to omit the
110
+
// field labels for "title" and "summary". If the user enters a large block
111
+
// of text at the beginning of the commit message with an empty line in it,
112
+
// treat everything before the blank line as "title" and everything after
113
+
// as "summary".
114
+
if (isset($fields[$key_title]) && empty($fields[$key_summary])) {
115
+
$lines = $fields[$key_title];
116
+
for ($ii = 0; $ii < count($lines); $ii++) {
117
+
if (strlen(trim($lines[$ii])) == 0) {
118
+
break;
119
+
}
120
+
}
121
+
if ($ii != count($lines)) {
122
+
$fields[$key_title] = array_slice($lines, 0, $ii);
123
+
$summary = array_slice($lines, $ii);
124
+
if (strlen(trim(implode("\n", $summary)))) {
125
+
$fields[$key_summary] = $summary;
126
+
}
127
+
}
128
+
}
129
+
130
+
// Implode all the lines back into chunks of text.
131
+
foreach ($fields as $name => $lines) {
132
+
$data = rtrim(implode("\n", $lines));
133
+
$data = ltrim($data, "\n");
134
+
$fields[$name] = $data;
135
+
}
136
+
137
+
// This is another piece of special-cased magic which allows you to
138
+
// enter a ridiculously long title, or just type a big block of stream
139
+
// of consciousness text, and have some sort of reasonable result conjured
140
+
// from it.
141
+
if (isset($fields[$key_title])) {
142
+
$terminal = '...';
143
+
$title = $fields[$key_title];
144
+
$short = phutil_utf8_shorten($title, 250, $terminal);
145
+
if ($short != $title) {
146
+
147
+
// If we shortened the title, split the rest into the summary, so
148
+
// we end up with a title like:
149
+
//
150
+
// Title title tile title title...
151
+
//
152
+
// ...and a summary like:
153
+
//
154
+
// ...title title title.
155
+
//
156
+
// Summary summary summary summary.
157
+
158
+
$summary = idx($fields, $key_summary, '');
159
+
$offset = strlen($short) - strlen($terminal);
160
+
$remainder = ltrim(substr($fields[$key_title], $offset));
161
+
$summary = '...'.$remainder."\n\n".$summary;
162
+
$summary = rtrim($summary, "\n");
163
+
164
+
$fields[$key_title] = $short;
165
+
$fields[$key_summary] = $summary;
166
+
}
167
+
}
168
+
169
+
return $fields;
170
+
}
171
+
172
+
173
+
/**
174
+
* @task parse
175
+
*/
176
+
public function getErrors() {
177
+
return $this->errors;
178
+
}
179
+
180
+
181
+
/* -( Support Methods )---------------------------------------------------- */
182
+
183
+
184
+
/**
185
+
* @task support
186
+
*/
187
+
public static function normalizeFieldLabel($label) {
188
+
return phutil_utf8_strtolower($label);
189
+
}
190
+
191
+
192
+
/* -( Internals )---------------------------------------------------------- */
193
+
194
+
195
+
/**
196
+
* @task internal
197
+
*/
198
+
private function buildLabelRegexp(array $label_map) {
199
+
$field_labels = array_keys($label_map);
200
+
foreach ($field_labels as $key => $label) {
201
+
$field_labels[$key] = preg_quote($label, '/');
202
+
}
203
+
$field_labels = implode('|', $field_labels);
204
+
205
+
$field_pattern = '/^(?P<field>'.$field_labels.'):(?P<text>.*)$/i';
206
+
207
+
return $field_pattern;
208
+
}
209
+
210
+
}
+65
src/applications/differential/parser/__tests__/DifferentialCommitMessageParserTestCase.php
+65
src/applications/differential/parser/__tests__/DifferentialCommitMessageParserTestCase.php
···
1
+
<?php
2
+
3
+
final class DifferentialCommitMessageParserTestCase
4
+
extends PhabricatorTestCase {
5
+
6
+
public function testDifferentialCommitMessageParser() {
7
+
$dir = dirname(__FILE__).'/messages/';
8
+
$list = Filesystem::listDirectory($dir, $include_hidden = false);
9
+
foreach ($list as $file) {
10
+
if (!preg_match('/.txt$/', $file)) {
11
+
continue;
12
+
}
13
+
14
+
$data = Filesystem::readFile($dir.$file);
15
+
$divider = "~~~~~~~~~~\n";
16
+
$parts = explode($divider, $data);
17
+
if (count($parts) !== 4) {
18
+
throw new Exception(
19
+
pht(
20
+
'Expected test file "%s" to contain four parts (message, fields, '.
21
+
'output, errors) divided by "~~~~~~~~~~".',
22
+
$file));
23
+
}
24
+
25
+
list($message, $fields, $output, $errors) = $parts;
26
+
$fields = phutil_json_decode($fields, null);
27
+
$output = phutil_json_decode($output, null);
28
+
$errors = phutil_json_decode($errors, null);
29
+
30
+
if ($fields === null || $output === null || $errors === null) {
31
+
throw new Exception(
32
+
pht(
33
+
'Expected test file "%s" to contain valid JSON in its sections.',
34
+
$file));
35
+
}
36
+
37
+
$parser = id(new DifferentialCommitMessageParser())
38
+
->setLabelMap($fields)
39
+
->setTitleKey('title')
40
+
->setSummaryKey('summary');
41
+
42
+
$result_output = $parser->parseCorpus($message);
43
+
$result_errors = $parser->getErrors();
44
+
45
+
$this->assertEqual($output, $result_output);
46
+
$this->assertEqual($errors, $result_errors);
47
+
}
48
+
}
49
+
50
+
public function testDifferentialCommitMessageParserNormalization() {
51
+
$map = array(
52
+
'Test Plan' => 'test plan',
53
+
'REVIEWERS' => 'reviewers',
54
+
'sUmmArY' => 'summary',
55
+
);
56
+
57
+
foreach ($map as $input => $expect) {
58
+
$this->assertEqual(
59
+
$expect,
60
+
DifferentialCommitMessageParser::normalizeFieldLabel($input),
61
+
pht('Field normalization of label "%s".', $input));
62
+
}
63
+
}
64
+
65
+
}
+17
src/applications/differential/parser/__tests__/messages/double-field.txt
+17
src/applications/differential/parser/__tests__/messages/double-field.txt
+11
src/applications/differential/parser/__tests__/messages/long-title.txt
+11
src/applications/differential/parser/__tests__/messages/long-title.txt
···
1
+
123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890
2
+
3
+
~~~~~~~~~~
4
+
{}
5
+
~~~~~~~~~~
6
+
{
7
+
"title": "1234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567...",
8
+
"summary": "...89012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890"
9
+
}
10
+
~~~~~~~~~~
11
+
[]
+15
src/applications/differential/parser/__tests__/messages/multi-label.txt
+15
src/applications/differential/parser/__tests__/messages/multi-label.txt
+24
src/applications/differential/parser/__tests__/messages/normal.txt
+24
src/applications/differential/parser/__tests__/messages/normal.txt
···
1
+
Title
2
+
3
+
Summary: This is the summary.
4
+
5
+
Test Plan: Tested things.
6
+
7
+
Reviewers: alincoln
8
+
9
+
~~~~~~~~~~
10
+
{
11
+
"title": "title",
12
+
"summary": "summary",
13
+
"test plan": "test plan",
14
+
"reviewers": "reviewers"
15
+
}
16
+
~~~~~~~~~~
17
+
{
18
+
"title": "Title",
19
+
"summary": "This is the summary.",
20
+
"test plan": "Tested things.",
21
+
"reviewers": "alincoln"
22
+
}
23
+
~~~~~~~~~~
24
+
[]
+12
src/applications/differential/parser/__tests__/messages/simple.txt
+12
src/applications/differential/parser/__tests__/messages/simple.txt