aboutsummaryrefslogtreecommitdiffstats
path: root/notes/tasks/2021-12-06_regrobid.md
blob: d879277db25995794d6abf845c8de4717fb31ae1 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378

Want to test recent updates of GROBID (to fix regex issue), and also re-process
a number of PDFs which failed to process with GROBID initially.


## HTTP 503

These are attempts which failed because GROBID was too busy or not running.

    # IMPROVED BELOW
    COPY (
        SELECT row_to_json(cdx)
        FROM grobid
        LEFT JOIN cdx ON grobid.sha1hex = cdx.sha1hex
        WHERE
            grobid.status_code = 503
            AND cdx.sha1hex IS NOT NULL
        -- LIMIT 5;
    )
    TO '/srv/sandcrawler/tasks/ungrobided_fatcat.2021-12-06.grobid503.json'
    WITH NULL '';
    # COPY 4749

Not actually that many, which seems good. Confirm that these are uniq by sha1hex:

    cat ungrobided_fatcat.2021-12-06.grobid503.json | jq .sha1hex -r | sort | uniq -d | wc -l
    # 302

Nope! Need to add "distinct on":

    COPY (
        SELECT DISTINCT ON (cdx.sha1hex) row_to_json(cdx)
        FROM grobid
        LEFT JOIN cdx ON grobid.sha1hex = cdx.sha1hex
        WHERE
            grobid.status_code = 503
            AND cdx.sha1hex IS NOT NULL
        -- LIMIT 5;
    )
    TO '/srv/sandcrawler/tasks/ungrobided_fatcat.2021-12-06.grobid503.json'
    WITH NULL '';
    # COPY 4297

    cat /srv/sandcrawler/tasks/ungrobided_fatcat.2021-12-06.grobid503.json | rg -v "\\\\" | jq . -c | pv -l | kafkacat -P -b wbgrp-svc263.us.archive.org -t sandcrawler-prod.ungrobided-pg -p -1

## Never Processed CDX

PDFs in fatcat which have never been processed with GROBID.

    COPY (
        SELECT DISTINCT ON (cdx.sha1hex) row_to_json(cdx)
        FROM fatcat_file
        LEFT JOIN cdx ON fatcat_file.sha1hex = cdx.sha1hex
        LEFT JOIN grobid ON grobid.sha1hex = fatcat_file.sha1hex
        LEFT JOIN file_meta ON file_meta.sha1hex = fatcat_file.sha1hex
        WHERE
            grobid.sha1hex IS NULL
            AND cdx.sha1hex IS NOT NULL
            AND (file_meta.mimetype = 'application/pdf' OR file_meta.mimetype IS NULL)
        -- LIMIT 5;
    )
    TO '/srv/sandcrawler/tasks/ungrobided_fatcat.2021-12-06.cdx.json'
    WITH NULL '';
    # COPY 15488

    cat /srv/sandcrawler/tasks/ungrobided_fatcat.2021-12-06.cdx.json | rg -v "\\\\" | jq . -c | pv -l | kafkacat -P -b wbgrp-svc263.us.archive.org -t sandcrawler-prod.ungrobided-pg -p -1


PDFs in fatcat which have never been processed with pdfextract.

    # TODO
    COPY (
        SELECT DISTINCT ON (cdx.sha1hex) row_to_json(cdx)
        FROM fatcat_file
        LEFT JOIN cdx ON fatcat_file.sha1hex = cdx.sha1hex
        LEFT JOIN pdf_meta ON pdf_meta.sha1hex = fatcat_file.sha1hex
        LEFT JOIN file_meta ON file_meta.sha1hex = fatcat_file.sha1hex
        WHERE
            pdf_meta.sha1hex IS NULL
            AND cdx.sha1hex IS NOT NULL
            AND cdx.mimetype = 'application/pdf'
            AND (file_meta.mimetype = 'application/pdf' OR file_meta.mimetype IS NULL)
        -- LIMIT 5;
    )
    TO '/srv/sandcrawler/tasks/unextracted_fatcat.2021-12-08.cdx.json'
    WITH NULL '';
    # COPY 45535

    cat /srv/sandcrawler/tasks/unextracted_fatcat.2021-12-08.cdx.json \
        | rg -v "\\\\" \
        | jq . -c \
        | pv -l \
        | kafkacat -P -b wbgrp-svc263.us.archive.org -t sandcrawler-prod.unextracted -p -1
    # 45.5k 0:00:01 [30.2k/s]

## Timeout or Failure

    COPY (
        SELECT DISTINCT ON (cdx.sha1hex) row_to_json(cdx)
        FROM grobid
        LEFT JOIN cdx ON grobid.sha1hex = cdx.sha1hex
        LEFT JOIN file_meta ON grobid.sha1hex = file_meta.sha1hex
        WHERE
            (grobid.status_code = 500 OR grobid.status_code = -4)
            AND cdx.sha1hex IS NOT NULL
            AND file_meta.mimetype  = 'application/pdf'
        -- LIMIT 5;
    )
    TO '/srv/sandcrawler/tasks/ungrobided_fatcat.2021-12-06.grobid_failed.json'
    WITH NULL '';
    # COPY 8,084,296

    cat /srv/sandcrawler/tasks/ungrobided_fatcat.2021-12-06.grobid_failed.json | rg -v "\\\\" | jq . -c | pv -l | kafkacat -P -b wbgrp-svc263.us.archive.org -t sandcrawler-prod.ungrobided-pg -p -1

This seems to not be working very well, mostly errors, empty docs, etc. Will
roll-forward the kafka consumer group after attempting a couple hundred
thousand of these.

Let's try limiting to files actually in fatcat:

    COPY (
        SELECT DISTINCT ON (cdx.sha1hex) row_to_json(cdx)
        FROM grobid
        LEFT JOIN cdx ON grobid.sha1hex = cdx.sha1hex
        LEFT JOIN file_meta ON grobid.sha1hex = file_meta.sha1hex
        LEFT JOIN fatcat_file ON grobid.sha1hex = fatcat_file.sha1hex
        WHERE
            (grobid.status_code = 500 OR grobid.status_code = -4)
            AND cdx.sha1hex IS NOT NULL
            AND fatcat_file.sha1hex IS NOT NULL
            AND file_meta.mimetype  = 'application/pdf'
            -- sort of arbitary "not recently" date filter
            AND (grobid.updated IS NULL OR grobid.updated < '2021-11-15')
        -- LIMIT 5;
    )
    TO '/srv/sandcrawler/tasks/ungrobided_fatcat.2021-12-08.grobid_failed.json'
    WITH NULL '';
    # COPY 529265

That is a much more managable batch to retry.

    cat /srv/sandcrawler/tasks/ungrobided_fatcat.2021-12-08.grobid_failed.json \
        | rg -v "\\\\" \
        | jq . -c \
        | pv -l \
        | kafkacat -P -b wbgrp-svc263.us.archive.org -t sandcrawler-prod.ungrobided-pg -p -1
    # 529k 0:00:17 [31.0k/s]


## Missing Fatcat Files

There were around a half million fatcat file entities which didn't have `cdx`
rows in sandcrawler. Did some specific pdfextract processing; now we should do
GROBID ingest as well.

Enque the `CDX` objects for GROBID and pdfextract processing:

    zcat /schnell/fatcat_cleanups/file_meta/files_missing_sha256.cdx_rows.json.gz \
        | rg -v "\\\\" \
        | jq . -c \
        | pv -l \
        | kafkacat -P -b wbgrp-svc263.us.archive.org -t sandcrawler-prod.ungrobided-pg -p -1
    # 354k 0:00:11 [30.6k/s]

    zcat /schnell/fatcat_cleanups/file_meta/files_missing_sha256.cdx_rows.json.gz \
        | rg -v "\\\\" \
        | jq . -c \
        | pv -l \
        | kafkacat -P -b wbgrp-svc263.us.archive.org -t sandcrawler-prod.unextracted -p -1

And some earlier files of interest on `aitio`:

    cat files_missing_sha256.ingest_results.json \
        | rg '"application/pdf"' \
        | rg -v "\\\\" \
        | jq .cdx -c \
        | sort -u -S 4G \
        | pv -l \
        > files_missing_sha256.cdx.uniq.json
    # 100k 0:00:47 [2.09k/s]

    cat files_missing_sha256.cdx.uniq.json \
        | rg -v "\\\\" \
        | jq . -c \
        | pv -l \
        | kafkacat -P -b wbgrp-svc263.us.archive.org -t sandcrawler-prod.ungrobided-pg -p -1

    cat files_missing_sha256.cdx.uniq.json \
        | rg -v "\\\\" \
        | jq . -c \
        | pv -l \
        | kafkacat -P -b wbgrp-svc263.us.archive.org -t sandcrawler-prod.unextracted -p -1


## Ancient Fatcat Files

Files from an era where we didn't record GROBID version or status, even for
success.

    COPY (
        SELECT DISTINCT ON (cdx.sha1hex) row_to_json(cdx)
        FROM grobid
        LEFT JOIN cdx ON grobid.sha1hex = cdx.sha1hex
        LEFT JOIN fatcat_file ON grobid.sha1hex = fatcat_file.sha1hex
        WHERE
            grobid.status_code = 200
            AND grobid.status IS NULL
            AND cdx.sha1hex IS NOT NULL
            AND fatcat_file.sha1hex IS NOT NULL
            -- sort of arbitary "not recently" date filter
            AND (grobid.updated IS NULL OR grobid.updated < '2021-11-15')
        -- LIMIT 5;
    )
    TO '/srv/sandcrawler/tasks/ungrobided_fatcat.2021-12-11.grobid_status_null.json'
    WITH NULL '';

    cat /srv/sandcrawler/tasks/ungrobided_fatcat.2021-12-11.grobid_status_null.json \
        | rg -v "\\\\" \
        | jq . -c \
        | pv -l \
        | kafkacat -P -b wbgrp-svc263.us.archive.org -t sandcrawler-prod.ungrobided-pg -p -1
    # 107k 0:00:03 [29.9k/s]


## Start Re-Processing Old GROBID Versions

    COPY (
        SELECT DISTINCT ON (cdx.sha1hex) row_to_json(cdx)
        FROM grobid
        LEFT JOIN cdx ON grobid.sha1hex = cdx.sha1hex
        LEFT JOIN fatcat_file ON grobid.sha1hex = fatcat_file.sha1hex
        WHERE
            grobid.status = 'success'
            AND grobid.grobid_version NOT LIKE '0.7.%'
            AND cdx.sha1hex IS NOT NULL
            AND fatcat_file.sha1hex IS NOT NULL
            -- sort of arbitary "not recently" date filter
            AND (grobid.updated IS NULL OR grobid.updated < '2021-11-15')
        -- LIMIT 5;
    )
    TO '/srv/sandcrawler/tasks/ungrobided_fatcat.2021-12-11.grobid_old.json'
    WITH NULL '';

This one is huge, and want to process in batches/chunks of ~8 million at a time.

    cd /srv/sandcrawler/tasks/
    cat /srv/sandcrawler/tasks/ungrobided_fatcat.2021-12-11.grobid_old.json \
        | split --lines 5000000 - ungrobided_fatcat.2021-12-11.grobid_old.split_ -d --additional-suffix .json

Submit individual batches like:

    cat /srv/sandcrawler/tasks/ungrobided_fatcat.2021-12-11.grobid_old.split_01.json \
        | rg -v "\\\\" \
        | jq . -c \
        | pv -l \
        | kafkacat -P -b wbgrp-svc263.us.archive.org -t sandcrawler-prod.ungrobided-pg -p -1

Overall progress:

    x  ungrobided_fatcat.2021-12-11.grobid_old.split_00.json
    .  ungrobided_fatcat.2021-12-11.grobid_old.split_01.json
    => ungrobided_fatcat.2021-12-11.grobid_old.split_02.json
    => ungrobided_fatcat.2021-12-11.grobid_old.split_03.json
    => ungrobided_fatcat.2021-12-11.grobid_old.split_04.json
    => ungrobided_fatcat.2021-12-11.grobid_old.split_05.json
    => ungrobided_fatcat.2021-12-11.grobid_old.split_06.json
    => ungrobided_fatcat.2021-12-11.grobid_old.split_07.json
    => ungrobided_fatcat.2021-12-11.grobid_old.split_08.json (small)

## General Counts

How many fatcat files of what mimetype (reported in sandcrawler-db)?

    SELECT file_meta.mimetype, COUNT(*)
    FROM fatcat_file
    LEFT JOIN file_meta ON fatcat_file.sha1hex = file_meta.sha1hex
    WHERE
        fatcat_file.first_release_ident IS NOT NULL
        AND fatcat_file.any_url = true
        AND content_scope IS NULL
    GROUP BY file_meta.mimetype
    ORDER BY COUNT(*) DESC
    LIMIT 25;

                                     mimetype                                  |  count
    ---------------------------------------------------------------------------+----------
     application/pdf                                                           | 45227033
                                                                               |   433068
     application/octet-stream                                                  |    30634
     application/jats+xml                                                      |     6874
     text/html                                                                 |      876
     application/postscript                                                    |      199
     application/gzip                                                          |      173
     text/plain                                                                |       84
     application/xml                                                           |       48
     application/vnd.ms-powerpoint                                             |       38
     application/msword                                                        |       16
     application/vnd.openxmlformats-officedocument.wordprocessingml.document   |        8
     image/jpeg                                                                |        6
     application/vnd.openxmlformats-officedocument.presentationml.presentation |        4
     message/rfc822                                                            |        4
     application/zip                                                           |        4
     text/x-tex                                                                |        3
     application/x-dosexec                                                     |        3
     application/x-tar                                                         |        2
     application/vnd.ms-tnef                                                   |        2
     image/svg+xml                                                             |        1
     image/tiff                                                                |        1
     image/png                                                                 |        1
     image/gif                                                                 |        1
     application/vnd.ms-office                                                 |        1
    (25 rows)


PDF extract status?

    SELECT pdf_meta.status, COUNT(*)
    FROM fatcat_file
    LEFT JOIN pdf_meta ON fatcat_file.sha1hex = pdf_meta.sha1hex
    WHERE
        fatcat_file.first_release_ident IS NOT NULL
        AND fatcat_file.any_url = true
        AND content_scope IS NULL
    GROUP BY pdf_meta.status
    ORDER BY COUNT(*) DESC
    LIMIT 25;

         status     |  count
    ----------------+----------
     success        | 43415920
                    |  2018522
     text-too-large |   122730
     parse-error    |    94876
     not-pdf        |    32156
     error-wayback  |    14504
     bad-unicode    |      279
     bad-pdf        |       98
     empty-blob     |        2
    (9 rows)


What are the GROBID status codes for fatcat files? Narrowed down:

    SELECT grobid.status, grobid.status_code, COUNT(*)
    FROM fatcat_file
    LEFT JOIN grobid ON fatcat_file.sha1hex = grobid.sha1hex
    WHERE
        fatcat_file.first_release_ident IS NOT NULL
        AND fatcat_file.any_url = true
        AND content_scope IS NULL
    GROUP BY grobid.status, grobid.status_code
    ORDER BY COUNT(*) DESC
    LIMIT 25;

         status     | status_code |  count
    ----------------+-------------+----------
     success        |         200 | 44409069
     error          |         500 |   580402
                    |             |   468836
                    |         200 |   240660
     error-timeout  |          -4 |       79
     bad-grobid-xml |         200 |       38
     error          |         200 |        3
    (7 rows)

Ran the same query again on 2021-12-15:

         status     | status_code |  count   
    ----------------+-------------+----------
     success        |         200 | 45092915
     error          |         500 |   302373
                    |             |   250335
                    |         200 |    53352
     bad-grobid-xml |         200 |       39
     error-timeout  |          -4 |       37
     error          |         200 |       34
     error          |         503 |        2
    (8 rows)