revert: 移除回填抽奖姓名的管理命令

服务器环境是 Python 2.7 + Django 1.x,命令在输出链路上反复踩编码坑,
暂停这次回填,先不做。

FFIB 4 hours ago
parent
commit
5906e9dca4

+ 0 - 0
sales/management/__init__.py


+ 0 - 0
sales/management/commands/__init__.py


+ 0 - 294
sales/management/commands/backfill_lottery_names.py

@@ -1,294 +0,0 @@
1
-# -*- coding: utf-8 -*-
2
-
3
-"""
4
-把抽奖名额里丢失的姓名 / 手机号,从服务器日志里回填。
5
-
6
-背景
7
-----
8
-服务端曾只读 POST 的 'name',而小程序传的字段叫 'title',导致 2026-11-22
9
-修复上线之前产生的名额 name 为空(phone 正常,字段名一致)。djlogit 把每次
10
-请求的 POST 参数都记在日志里,可以按 (campaign_id, serial_no) 把姓名找回来。
11
-
12
-用法
13
-----
14
-    # 只看能补多少,不动数据库
15
-    python manage.py backfill_lottery_names --dry-run
16
-
17
-    # 只处理某个活动
18
-    python manage.py backfill_lottery_names --dry-run --campaign-id 4SRByFKXeSqH7VfUp5D9Hn
19
-
20
-    # 确认数字没问题后落库(只填空值,不覆盖已有数据)
21
-    python manage.py backfill_lottery_names --apply
22
-
23
-    # 只解析日志不查库(排查用)
24
-    python manage.py backfill_lottery_names --parse-only /tmp/kodotamron.logit.log
25
-"""
26
-
27
-from __future__ import print_function, unicode_literals
28
-
29
-import ast
30
-import glob
31
-import io
32
-import os
33
-import re
34
-import sys
35
-
36
-from django.core.management.base import BaseCommand, CommandError
37
-from django.db import transaction
38
-
39
-from sales.models import LotteryEntryInfo
40
-
41
-JOIN_MARKER = 'func=lottery_join_api'
42
-POST_MARKER = 'flag=post'
43
-TIME_RE = re.compile(r'^(?:DEBUG|INFO|WARNING|ERROR)\s+(\d{4}-\d{2}-\d{2} \d{2}:\d{2}:\d{2})')
44
-
45
-DEFAULT_LOG_GLOB = '/tmp/kodotamron.logit.log*'
46
-
47
-
48
-def parse_content(raw):
49
-    """把 '<QueryDict: {u'a': [u'b']}>' 解析成普通 dict,解析失败返回 None"""
50
-    text = raw.strip()
51
-    start = text.find('<QueryDict:')
52
-
53
-    if start < 0:
54
-        return None
55
-
56
-    text = text[start + len('<QueryDict:'):].strip()
57
-
58
-    if text.endswith('>'):
59
-        text = text[:-1].strip()
60
-
61
-    try:
62
-        value = ast.literal_eval(text)
63
-    except (ValueError, SyntaxError):
64
-        return None
65
-
66
-    if not isinstance(value, dict):
67
-        return None
68
-
69
-    # 每个 key 取第一个值
70
-    return {k: (v[0] if isinstance(v, list) and v else v) for k, v in value.items()}
71
-
72
-
73
-def parse_line(line):
74
-    """解析一行日志;不是 lottery_join_api 的 post 记录返回 None"""
75
-    if JOIN_MARKER not in line or POST_MARKER not in line:
76
-        return None
77
-
78
-    index = line.find('content=')
79
-
80
-    if index < 0:
81
-        return None
82
-
83
-    params = parse_content(line[index + len('content='):])
84
-
85
-    if not params:
86
-        return None
87
-
88
-    matched = TIME_RE.match(line)
89
-
90
-    return {
91
-        'campaign_id': (params.get('campaign_id') or '').strip(),
92
-        'serial_no': (params.get('serial_no') or '').strip(),
93
-        'name': (params.get('name') or params.get('title') or '').strip(),
94
-        'phone': (params.get('phone') or '').strip(),
95
-        'timestamp': matched.group(1) if matched else '',
96
-    }
97
-
98
-
99
-def iter_lines(path):
100
-    """
101
-    逐行读日志。
102
-
103
-    djlogit 的记录里,中文是 \u4e54 这种转义形式(纯 ASCII),
104
-    文件名和内容都按 utf-8 读即可;errors="replace" 兜住个别脏字节,
105
-    避免某个归档文件里混进坏数据就整批中断。
106
-    """
107
-    with io.open(path, 'r', encoding='utf-8', errors='replace') as handle:
108
-        for line in handle:
109
-            yield line
110
-
111
-
112
-def collect(log_glob):
113
-    """
114
-    扫描日志,返回 ({(campaign_id, serial_no): {name, phone, timestamp}}, 文件数, 扫描到的 join 条数)
115
-
116
-    同一条名额出现多次时,优先保留带姓名的那次。
117
-    """
118
-    found = {}
119
-    files = sorted(glob.glob(log_glob))
120
-    scanned = 0
121
-
122
-    for path in files:
123
-        for line in iter_lines(path):
124
-            record = parse_line(line)
125
-
126
-            if not record:
127
-                continue
128
-
129
-            scanned += 1
130
-            key = (record['campaign_id'], record['serial_no'])
131
-            old = found.get(key)
132
-
133
-            if old is None or (record['name'] and not old['name']):
134
-                found[key] = record
135
-
136
-    return found, len(files), scanned
137
-
138
-
139
-PY2 = sys.version_info[0] == 2
140
-
141
-
142
-def emit(stream, text):
143
-    """
144
-    统一输出,绕过 Django 的 OutputWrapper。
145
-
146
-    py2 + Django 1.x 下 OutputWrapper.write 会做 msg.endswith(unicode '\n'):
147
-    - 传 utf-8 字节 → bytes.endswith(unicode) 触发 ascii 解码 → UnicodeDecodeError
148
-    - 传 unicode   → sys.stdout 编码是 ascii 时 → UnicodeEncodeError
149
-    所以直接写底层流:py2 写 utf-8 字节,py3 写 str。
150
-    """
151
-    if PY2:
152
-        target = getattr(stream, '_out', None) or stream
153
-        target.write(text.encode('utf-8'))
154
-        target.flush()
155
-    else:
156
-        stream.write(text)
157
-
158
-
159
-def dump_ascii_traceback(stream):
160
-    """
161
-    把当前异常栈转成纯 ASCII 再输出。
162
-
163
-    py2 环境下 stderr 常见编码是 ascii,Django 打印含中文的堆栈时会二次抛错,
164
-    结果只剩一行 UnicodeDecodeError,看不到真正出错的位置。
165
-    """
166
-    import traceback
167
-
168
-    text = traceback.format_exc()
169
-
170
-    if not isinstance(text, type(u'')):
171
-        text = text.decode('utf-8', 'replace')
172
-
173
-    emit(stream, text.encode('ascii', 'backslashreplace').decode('ascii'))
174
-
175
-
176
-def setup_stdio_encoding():
177
-    """服务器 locale 可能是 POSIX,stdout 编码为 ascii,输出中文会直接抛异常"""
178
-    for stream in ('stdout', 'stderr'):
179
-        handle = getattr(sys, stream, None)
180
-        reconfigure = getattr(handle, 'reconfigure', None)
181
-
182
-        if reconfigure:
183
-            try:
184
-                reconfigure(encoding='utf-8', errors='replace')
185
-            except (ValueError, OSError):
186
-                pass
187
-
188
-
189
-class Command(BaseCommand):
190
-    help = '从 djlogit 日志回填抽奖名额的姓名/手机号(只填空值)'
191
-
192
-    def add_arguments(self, parser):
193
-        parser.add_argument('--log-glob', default=DEFAULT_LOG_GLOB, help='日志路径通配符')
194
-        parser.add_argument('--campaign-id', default='', help='只处理指定活动')
195
-        parser.add_argument('--dry-run', action='store_true', help='只统计,不写库')
196
-        parser.add_argument('--apply', action='store_true', help='真正写库')
197
-        parser.add_argument('--parse-only', default='', help='只解析日志并打印,不查数据库')
198
-
199
-    def handle(self, *args, **options):
200
-        setup_stdio_encoding()
201
-
202
-        try:
203
-            self.run_command(*args, **options)
204
-        except Exception:
205
-            dump_ascii_traceback(self.stderr)
206
-            raise CommandError('backfill failed, see ASCII traceback above')
207
-
208
-    def run_command(self, *args, **options):
209
-        log_glob = options['log_glob']
210
-
211
-        if options['parse_only']:
212
-            return self.handle_parse_only(options['parse_only'])
213
-
214
-        if not options['dry_run'] and not options['apply']:
215
-            raise CommandError('specify either --dry-run (count only) or --apply (write to db)')
216
-
217
-        found, file_count, scanned = collect(log_glob)
218
-
219
-        emit(self.stdout, '日志文件: %d 个,解析到 join 请求: %d 条,可回填组合: %d 个' % (
220
-            file_count, scanned, len(found)))
221
-
222
-        if not file_count:
223
-            emit(self.stdout, self.style.WARNING('没有匹配到日志文件:%s' % log_glob))
224
-
225
-        entries = LotteryEntryInfo.objects.filter(status=True)
226
-
227
-        if options['campaign_id']:
228
-            entries = entries.filter(campaign_id=options['campaign_id'])
229
-
230
-        missing = [entry for entry in entries.only('id', 'campaign_id', 'serial_no', 'name', 'phone')
231
-                   if not (entry.name or '').strip()]
232
-
233
-        emit(self.stdout, '待检查名额: %d 条,其中姓名为空: %d 条' % (entries.count(), len(missing)))
234
-
235
-        todo = []
236
-
237
-        for entry in missing:
238
-            record = found.get((entry.campaign_id, entry.serial_no))
239
-
240
-            if not record or not record['name']:
241
-                continue
242
-
243
-            todo.append((entry, record))
244
-
245
-        emit(self.stdout, '日志里能匹配到且带姓名的: %d 条' % len(todo))
246
-        emit(self.stdout, '补不上的: %d 条(对应日志已被清理)' % (len(missing) - len(todo)))
247
-
248
-        for entry, record in todo[:10]:
249
-            will_fill_phone = record['phone'] and not (entry.phone or '').strip()
250
-            emit(self.stdout, '  %s  %s → %s%s' % (
251
-                entry.campaign_id[:8],
252
-                entry.serial_no,
253
-                record['name'],
254
-                '(含手机号)' if will_fill_phone else '',
255
-            ))
256
-
257
-        if len(todo) > 10:
258
-            emit(self.stdout, '  ... 其余 %d 条省略' % (len(todo) - 10))
259
-
260
-        if options['dry_run']:
261
-            emit(self.stdout, self.style.SUCCESS('\n[dry-run] 未写库。确认后加 --apply 执行。'))
262
-            return
263
-
264
-        updated = 0
265
-
266
-        with transaction.atomic():
267
-            for entry, record in todo:
268
-                update_fields = []
269
-
270
-                if not (entry.name or '').strip():
271
-                    entry.name = record['name']
272
-                    update_fields.append('name')
273
-
274
-                if not (entry.phone or '').strip() and record['phone']:
275
-                    entry.phone = record['phone']
276
-                    update_fields.append('phone')
277
-
278
-                if update_fields:
279
-                    entry.save(update_fields=update_fields)
280
-                    updated += 1
281
-
282
-        emit(self.stdout, self.style.SUCCESS('\n已更新 %d 条名额' % updated))
283
-
284
-    def handle_parse_only(self, path):
285
-        if not os.path.exists(path):
286
-            raise CommandError('log file not found: %s' % path)
287
-
288
-        found, file_count, scanned = collect(path)
289
-
290
-        emit(self.stdout, '文件: %d 个,join 请求: %d 条,可回填组合: %d 个' % (file_count, scanned, len(found)))
291
-
292
-        for (campaign_id, serial_no), record in list(found.items())[:10]:
293
-            emit(self.stdout, '  %s / %s → %s %s' % (
294
-                campaign_id[:8], serial_no, record['name'] or '(空)', record['phone']))