qsv 16.1.0

A Blazing-Fast Data-wrangling toolkit.
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
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
static USAGE: &str = r#"
Quickly sniff the first n rows and infer CSV metadata (delimiter, header row, number of
preamble rows, quote character, flexible, is_utf8, average record length, number of records,
content length and estimated number of records if sniffing a URL, file size, number of fields,
field names & data types).

`sniff` is also a mime type detector, returning the detected mime type, file size and
last modified date. If --no-infer is enabled, it doesn't even bother to infer the CSV's schema.
This makes it useful for accelerated CKAN harvesting and for checking stale/broken resource URLs.

When qsv is compiled with the optional `magika` feature, it uses Google's Magika
AI-powered content detection to identify file types with high accuracy. Magika detects over
200 content types including CSV, MS Office/Open Document files, JSON, XML, PDF, PNG, JPEG
and many more.
See https://opensource.googleblog.com/2025/11/announcing-magika-10-now-faster-smarter.html.

When the `magika` feature is not enabled in a build (e.g., MUSL builds, qsvlite, qsvdp), it falls back
to the file-format crate which provides basic MIME type detection.

NOTE: This command "sniffs" a CSV's schema by sampling the first n rows (default: 1000)
of a file. Its inferences are sometimes wrong if the the file is too small to infer a pattern
or if the CSV has unusual formatting - with atypical delimiters, quotes, etc.

In such cases, selectively use the --sample, --delimiter and --quote options to improve
the accuracy of the sniffed schema.

If you want more robust, guaranteed schemata, use the "schema" or "stats" commands
instead as they scan the entire file. However, they only work on local files and well-formed
CSVs, unlike `sniff` which can work with remote files, various CSV dialects and is very fast
regardless of file size.

Examples:

  # Sniff a local CSV file
  qsv sniff data.csv

  # Sniff a remote TSV file over HTTPS
  qsv sniff https://example.com/data.tsv

  # Get the mime type of a remote file without inferring the CSV schema
  qsv sniff --no-infer https://example.com/data.xlsx

  # Sniff the first 20 percent of a SSV file
  qsv sniff --sample 0.20 data.ssv

For examples, see https://github.com/dathere/qsv/blob/master/tests/test_sniff.rs.

Usage:
    qsv sniff [options] [<input>]
    qsv sniff --help

sniff arguments:
    <input>                  The file to sniff. This can be a local file, stdin or a
                             URL (http and https schemes supported).

                             Note that when input is a URL and is a CSV candidate,
                             sniff will automatically download a sample to a temporary
                             file and sniff it.

                             If the file sample is not a CSV, sniff will return as an error
                             - the detected mime type, file size & last modified date.

                             When --no-infer is enabled, sniff will not bother to infer the
                             CSV schema and just work as a general mime type detector -
                             returning the detected mime type, file size and last modified date.

                             When sniffing a local file, it will scan the file to detect the mime type.
                             When sniffing a URL, it will only download the first chunk of the file if 
                             the --quick option is enabled, otherwise it will download the entire file.
                             
sniff options:
    --sample <size>          First n rows to sample to sniff out the metadata.
                             When sample size is between 0 and 1 exclusive, 
                             it is treated as a percentage of the CSV to sample
                             (e.g. 0.20 is 20 percent).
                             When it is zero, the entire file will be sampled.
                             When the input is a URL, the sample size dictates
                             how many lines to sample without having to
                             download the entire file. Ignored when --no-infer is enabled.
                             [default: 1000]
    --prefer-dmy             Prefer to parse dates in dmy format. Otherwise, use mdy format.
                             Ignored when --no-infer is enabled.
    -d, --delimiter <arg>    The delimiter for reading CSV data.
                             Specify this when the delimiter is known beforehand,
                             as the delimiter inferencing algorithm can sometimes fail.
                             Must be a single ascii character.
    --quote <arg>            The quote character for reading CSV data.
                             Specify this when the quote character is known beforehand,
                             as the quote char inferencing algorithm can sometimes fail.
                             Must be a single ascii character - typically, double quote ("),
                             single quote ('), or backtick (`).
    --json                   Return results in JSON format.
    --pretty-json            Return results in pretty JSON format.
    --save-urlsample <file>  Save the URL sample to a file.
                             Valid only when input is a URL.
    --timeout <secs>         Timeout when sniffing URLs in seconds. If 0, no timeout is used.
                             [default: 30]
    --user-agent <agent>     Specify custom user agent to use when sniffing a CSV on a URL.
                             It supports the following variables - $QSV_VERSION, $QSV_TARGET,
                             $QSV_BIN_NAME, $QSV_KIND and $QSV_COMMAND. Try to follow the syntax here -
                             https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/User-Agent
    --stats-types            Use the same data type names as `stats`.
                             (Unsigned, Signed => Integer, Text => String, everything else the same)
    --no-infer               Do not infer the schema. Only return the file's mime type, size and
                             last modified date. Use this to use sniff as a general mime type detector.
                             Note that CSV and TSV files will only be detected as mime type plain/text
                             in this mode.
    --just-mime              Only return the file's mime type. Use this to use sniff as a general
                             mime type detector. Synonym for --no-infer.
    -Q, --quick              When sniffing a non-CSV remote file, only download the first chunk of the file
                             before attempting to detect the mime type. This is faster but less accurate as
                             some mime types cannot be detected with just the first downloaded chunk.
    --harvest-mode           This is a convenience flag when using sniff in CKAN harvesters. 
                             It is equivalent to --quick --timeout 10 --stats-types --json
                             and --user-agent "CKAN-harvest/$QSV_VERSION ($QSV_TARGET; $QSV_BIN_NAME)"

Common options:
    -h, --help               Display this message
    -p, --progressbar        Show progress bars. Only valid for URL input.
"#;

#[cfg(feature = "magika")]
use std::sync::{LazyLock, Mutex};
use std::{
    cmp::min,
    fmt, fs,
    io::{Seek, SeekFrom, Write, copy},
    path::PathBuf,
    time::Duration,
};

use bytes::Bytes;
use csv_nose::{DatePreference, SampleSize, Sniffer};
#[cfg(not(feature = "magika"))]
use file_format::FileFormat;
use futures::executor::block_on;
use futures_util::StreamExt;
use indicatif::HumanCount;
#[cfg(any(feature = "feature_capable", feature = "lite"))]
use indicatif::{HumanBytes, ProgressBar, ProgressDrawTarget, ProgressStyle};
#[cfg(feature = "magika")]
use magika::Session;
use qsv_tabwriter::TabWriter;
use serde::{Deserialize, Serialize};
use serde_json::json;
use tempfile::NamedTempFile;
use url::Url;

use crate::{
    CliResult,
    config::{Config, Delimiter},
    util,
    util::format_systemtime,
};

// Magika ML session - lazy initialized and reused for all detections
#[cfg(feature = "magika")]
static MAGIKA_SESSION: LazyLock<Result<Mutex<Session>, String>> =
    LazyLock::new(|| Session::new().map(Mutex::new).map_err(|e| e.to_string()));

/// Get label from file-format, with special handling for CSV/TSV
#[cfg(not(feature = "magika"))]
fn get_label(fmt: &file_format::FileFormat, mime_type: &str) -> String {
    // Special case for CSV/TSV which file-format doesn't specifically detect
    if mime_type == "text/csv" || mime_type == "application/csv" {
        return "csv".to_string();
    }
    if mime_type == "text/tab-separated-values" {
        return "tsv".to_string();
    }
    // Use short_name if available, otherwise fall back to name
    fmt.short_name()
        .unwrap_or_else(|| fmt.name())
        .to_lowercase()
        .replace(' ', "-")
}

/// Detect mime type from bytes using available backend
/// Returns (mime_type, label, inference_score)
#[cfg(feature = "magika")]
fn detect_mime_from_bytes(bytes: &[u8]) -> (String, String, f32) {
    // If Magika failed to initialize or the mutex is poisoned, fall back to a default mime type
    let default_mime = || {
        (
            "application/octet-stream".to_string(),
            "unknown".to_string(),
            1.0_f32, // default score for fallback
        )
    };

    match &*MAGIKA_SESSION {
        Ok(mutex) => match mutex.lock() {
            Ok(mut session) => match session.identify_content_sync(bytes) {
                Ok(result) => (
                    result.info().mime_type.to_string(),
                    result.info().label.to_string(),
                    result.score(),
                ),
                Err(_) => default_mime(),
            },
            Err(_) => default_mime(),
        },
        Err(_) => default_mime(),
    }
}

#[cfg(not(feature = "magika"))]
fn detect_mime_from_bytes(bytes: &[u8]) -> (String, String, f32) {
    let fmt = FileFormat::from_bytes(bytes);
    let mime = fmt.media_type().to_string();
    let label = get_label(&fmt, &mime);
    (mime, label, 1.0)
}

/// Detect mime type from file using available backend
/// Returns (mime_type, label, inference_score)
#[cfg(feature = "magika")]
fn detect_mime_from_file(path: &std::path::Path) -> Result<(String, String, f32), String> {
    match &*MAGIKA_SESSION {
        Ok(mutex) => mutex
            .lock()
            .map_err(|_| "Magika session mutex poisoned".to_string())?
            .identify_file_sync(path)
            .map(|r| {
                (
                    r.info().mime_type.to_string(),
                    r.info().label.to_string(),
                    r.score(),
                )
            })
            .map_err(|e| format!("Magika detection error: {e}")),
        Err(err) => Err(format!("Magika initialization error: {err}")),
    }
}

#[cfg(not(feature = "magika"))]
fn detect_mime_from_file(path: &std::path::Path) -> Result<(String, String, f32), String> {
    FileFormat::from_file(path)
        .map(|fmt| {
            let mime = fmt.media_type().to_string();
            let label = get_label(&fmt, &mime);
            (mime, label, 1.0)
        })
        .map_err(|e| format!("File format detection error: {e}"))
}

#[allow(dead_code)]
#[derive(Deserialize)]
struct Args {
    arg_input:           Option<String>,
    flag_sample:         f64,
    flag_prefer_dmy:     bool,
    flag_json:           bool,
    flag_save_urlsample: Option<String>,
    flag_pretty_json:    bool,
    flag_delimiter:      Option<Delimiter>,
    flag_quote:          Option<char>,
    flag_progressbar:    bool,
    flag_timeout:        u16,
    flag_user_agent:     Option<String>,
    flag_stats_types:    bool,
    flag_no_infer:       bool,
    flag_just_mime:      bool,
    flag_quick:          bool,
    flag_harvest_mode:   bool,
}

#[derive(Serialize, Deserialize, Default, Debug)]
struct SniffStruct {
    path:            String,
    sniff_timestamp: String,
    last_modified:   String,
    delimiter_char:  char,
    header_row:      bool,
    preamble_rows:   usize,
    quote_char:      String,
    flexible:        bool,
    is_utf8:         bool,
    detected_mime:   String,
    detected_label:  String,
    inference_score: Option<f32>,
    retrieved_size:  usize,
    file_size:       usize,
    sampled_records: usize,
    estimated:       bool,
    num_records:     usize,
    avg_record_len:  usize,
    num_fields:      usize,
    stats_types:     bool,
    fields:          Vec<String>,
    types:           Vec<String>,
}
impl fmt::Display for SniffStruct {
    fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
        writeln!(
            f,
            "Path: {}",
            // when sniffing a snappy compressed file, it is first decompressed
            // to a temporary file. The original file name is stored in the
            // temporary file name, so we extract the original file name
            if self.path.contains("qsv_temp_decompressed__") {
                // use a regular expression to extract the original file name
                // the original file name after "qsv_temp_decompressed__"
                let re = regex::Regex::new(r"qsv_temp_decompressed__(?P<filename>.*)$").unwrap();
                let caps = re.captures(&self.path).unwrap();
                let filename = caps.name("filename").unwrap().as_str();
                filename.to_string()
            } else {
                self.path.clone()
            }
        )?;
        writeln!(f, "Sniff Timestamp: {}", self.sniff_timestamp)?;
        writeln!(f, "Last Modified: {}", self.last_modified)?;
        writeln!(
            f,
            "Delimiter: {}",
            if self.delimiter_char == '\t' {
                "tab".to_string()
            } else {
                self.delimiter_char.to_string()
            }
        )?;
        writeln!(f, "Header Row: {}", self.header_row)?;
        writeln!(
            f,
            "Preamble Rows: {}",
            HumanCount(self.preamble_rows as u64)
        )?;
        writeln!(f, "Quote Char: {}", self.quote_char)?;
        writeln!(f, "Flexible: {}", self.flexible)?;
        writeln!(f, "Is UTF8: {}", self.is_utf8)?;
        writeln!(f, "Detected Mime Type: {}", self.detected_mime)?;
        writeln!(f, "Detected Label: {}", self.detected_label)?;
        if let Some(score) = self.inference_score {
            writeln!(f, "Inference Score: {}", util::round_num(score.into(), 4))?;
        }
        writeln!(
            f,
            "Retrieved Size (bytes): {}",
            HumanCount(self.retrieved_size as u64)
        )?;
        // show "Unknown" for usize::MAX (sentinel value when Content-Length is missing)
        if self.file_size == usize::MAX {
            writeln!(f, "File Size (bytes): Unknown")?;
        } else {
            writeln!(
                f,
                "File Size (bytes): {}",
                HumanCount(self.file_size as u64)
            )?;
        }
        writeln!(
            f,
            "Sampled Records: {}",
            HumanCount(self.sampled_records as u64)
        )?;
        writeln!(f, "Estimated: {}", self.estimated)?;
        // show "Unknown" when num_records was estimated from unknown file_size
        if self.file_size == usize::MAX {
            writeln!(f, "Num Records: Unknown")?;
        } else {
            writeln!(f, "Num Records: {}", HumanCount(self.num_records as u64))?;
        }
        writeln!(
            f,
            "Avg Record Len (bytes): {}",
            HumanCount(self.avg_record_len as u64)
        )?;
        writeln!(f, "Num Fields: {}", HumanCount(self.num_fields as u64))?;
        writeln!(f, "Stats Types: {}", self.stats_types)?;
        writeln!(f, "Fields:")?;

        let mut tabwtr = TabWriter::new(vec![]);

        for (i, ty) in self.types.iter().enumerate() {
            let data_type = if self.stats_types {
                match ty.as_str() {
                    "Unsigned" | "Signed" => "Integer",
                    "Text" => "String",
                    _ => ty,
                }
            } else {
                ty
            };

            writeln!(
                &mut tabwtr,
                "\t{}:\t{data_type}\t{}",
                i + 1,
                self.fields.get(i).unwrap_or(&String::new())
            )
            .unwrap_or_default();
        }
        tabwtr.flush().unwrap();

        let tabbed_field_list = String::from_utf8(tabwtr.into_inner().unwrap()).unwrap();
        writeln!(f, "{tabbed_field_list}")?;

        Ok(())
    }
}

#[derive(Debug, Clone)]
struct SniffFileStruct {
    display_path:       String,
    file_to_sniff:      String,
    detected_mime:      String,
    detected_label:     String,
    inference_score:    Option<f32>,
    tempfile_flag:      bool,
    retrieved_size:     usize,
    file_size:          usize,
    last_modified:      String,
    downloaded_records: usize,
}

const fn rowcount(
    metadata: &csv_nose::metadata::Metadata,
    sniff_file_info: &SniffFileStruct,
    count: usize,
) -> (usize, bool) {
    let mut estimated = false;
    let rowcount = if count == usize::MAX {
        // if the count == usize::MAX, it's a sentinel value for "Unknown" as the server
        // didn't provide a Content-Length header, so we estimate the rowcount by
        // dividing the file_size by avg_rec_len
        estimated = true;
        sniff_file_info.file_size / metadata.avg_record_len
    } else {
        count
    };

    let has_header_row = metadata.dialect.header.has_header_row;
    let num_preamble_rows = metadata.dialect.header.num_preamble_rows;
    let mut final_rowcount = rowcount;

    if !has_header_row {
        final_rowcount += 1;
    }

    final_rowcount -= num_preamble_rows;
    (final_rowcount, estimated)
}

async fn get_file_to_sniff(args: &Args, tmpdir: &tempfile::TempDir) -> CliResult<SniffFileStruct> {
    if let Some(uri) = args.arg_input.clone() {
        match uri {
            // its a URL, download sample to temp file
            url if Url::parse(&url).is_ok() && url.starts_with("http") => {
                // Transform GitHub blob URLs to raw URLs for direct file access
                let url = util::transform_github_url(&url);
                let snappy_flag = url.to_lowercase().ends_with(".sz");

                // setup the reqwest client
                let client = util::create_reqwest_async_client(
                    args.flag_user_agent.clone(),
                    util::timeout_secs(args.flag_timeout)
                        .unwrap_or(30)
                        .min(u16::MAX as u64) as u16,
                    Some(url.clone()),
                )?;

                let res = client
                    .get(url.clone())
                    .timeout(Duration::from_secs(
                        util::timeout_secs(args.flag_timeout).unwrap_or(30),
                    ))
                    .send()
                    .await?;

                // Check Content-Type header to detect HTML responses early
                // This helps catch cases where URLs serve HTML pages instead of raw files
                if let Some(content_type) = res.headers().get("content-type")
                    && let Ok(ct_str) = content_type.to_str()
                    && ct_str.contains("text/html")
                {
                    log::warn!(
                        "Response Content-Type is text/html. URL may not point to raw file \
                         content."
                    );
                }

                let last_modified = match res.headers().get("Last-Modified") {
                    Some(lm) => match lm.to_str() {
                        Ok(s) => {
                            // convert Last-Modified RFC2822 to RFC3339 format
                            let dt = chrono::DateTime::parse_from_rfc2822(s).unwrap();
                            dt.to_rfc3339_opts(chrono::SecondsFormat::Secs, false)
                        },
                        // server did not return Last-Modified header
                        Err(_) => String::from("Unknown"),
                    },
                    None => String::from("Unknown"),
                };

                let total_size = match res.content_length() {
                    Some(l) => l as usize,
                    None => {
                        // if we can't get the content length, just set it to a large value
                        // so we just end up downloading the entire file
                        usize::MAX
                    },
                };

                #[allow(clippy::cast_precision_loss)]
                let lines_sample_size = if snappy_flag {
                    // if it's a snappy compressed file, we need to download the entire file
                    // to uncompress and sniff it
                    usize::MAX
                } else if args.flag_sample > 1.0 {
                    args.flag_sample.round() as usize
                } else if args.flag_sample.abs() < f64::EPSILON {
                    // sample size is zero, so we want to download the entire file
                    usize::MAX
                } else {
                    // sample size is a percentage, download percentage number of lines
                    // from the file. Since we don't know how wide the lines are, we
                    // just download a percentage of the bytes, assuming the lines are
                    // 100 characters wide as a rough estimate.
                    ((total_size / 100_usize) as f64 * args.flag_sample) as usize
                };

                // prep progress bar
                #[cfg(any(feature = "feature_capable", feature = "lite"))]
                let show_progress =
                    args.flag_progressbar || util::get_envvar_flag("QSV_PROGRESSBAR");

                #[cfg(any(feature = "feature_capable", feature = "lite"))]
                let progress = ProgressBar::with_draw_target(
                    Some(total_size.try_into().unwrap_or(u64::MAX)),
                    ProgressDrawTarget::stderr_with_hz(5),
                );
                #[cfg(any(feature = "feature_capable", feature = "lite"))]
                if show_progress {
                    progress.set_style(
                        #[allow(clippy::to_string_in_format_args)]
                        #[allow(clippy::literal_string_with_formatting_args)]
                        ProgressStyle::default_bar()
                            .template(
                                "{msg}\n{spinner:.green} [{elapsed_precise}] \
                                 [{wide_bar:.white/blue}] {bytes}/{total_bytes} ({bytes_per_sec}, \
                                 {eta})",
                            )
                            .unwrap(),
                    );
                    if lines_sample_size == usize::MAX {
                        progress.set_message(format!(
                            "Downloading {}...",
                            HumanBytes(total_size as u64)
                        ));
                    } else {
                        progress.set_message(format!(
                            "Downloading {} samples...",
                            HumanCount(lines_sample_size as u64)
                        ));
                    }
                } else {
                    progress.set_draw_target(ProgressDrawTarget::hidden());
                }

                let mut file = NamedTempFile::new()?;
                let mut downloaded = 0_usize;
                let mut stream = res.bytes_stream();
                #[cfg(any(feature = "feature_capable", feature = "lite"))]
                let mut downloaded_lines = 0_usize;
                #[allow(unused_assignments)]
                let mut chunk = Bytes::new(); // amortize the allocation
                let mut firstchunk = Bytes::new();

                // download chunks until we have the desired sample size
                while let Some(item) = stream.next().await {
                    chunk = item?;

                    file.write_all(&chunk)?;
                    let chunk_len = chunk.len();

                    if downloaded == 0 && !snappy_flag && args.flag_quick {
                        let (mime, _kind, _score) = detect_mime_from_bytes(&chunk[..]);
                        log::debug!("scanned first {chunk_len} bytes - detected mime: {mime}");
                        if !mime.starts_with("text/") && mime != "application/csv" {
                            downloaded = chunk_len;
                            firstchunk = chunk.clone();
                            break;
                        }
                    }

                    downloaded = min(downloaded + chunk_len, total_size);

                    #[cfg(any(feature = "feature_capable", feature = "lite"))]
                    if show_progress {
                        progress.inc(chunk_len as u64);
                    }

                    // check if we're downloading the entire file
                    #[cfg(any(feature = "feature_capable", feature = "lite"))]
                    if lines_sample_size != usize::MAX {
                        // we're not downloading the entire file, so we need to
                        // scan chunk for newlines
                        let num_lines = chunk.into_iter().filter(|&x| x == b'\n').count();
                        // and keep track of the number of lines downloaded which is ~= sample_size
                        downloaded_lines += num_lines;
                        // we downloaded enough samples, stop downloading
                        if downloaded_lines > lines_sample_size {
                            downloaded_lines -= 1; // subtract 1 because we don't want to count the header row
                            break;
                        }
                    }
                }
                drop(client);
                file.flush()?;

                #[cfg(any(feature = "feature_capable", feature = "lite"))]
                if show_progress {
                    if snappy_flag {
                        progress.finish_with_message(format!(
                            "Downloaded {}.",
                            HumanBytes(downloaded as u64)
                        ));
                    } else {
                        progress.finish_with_message(format!(
                            "Downloaded {} samples.",
                            HumanCount(downloaded_lines as u64)
                        ));
                    }
                }

                let wtr_file_path;
                let mut csv_candidate = true;
                let mut detected_mime = String::new();
                let mut detected_label: String = String::new();

                let mut detected_score: Option<f32> = None;
                if !args.flag_quick {
                    let (mime, label, score) = detect_mime_from_file(file.path())?;
                    detected_mime = mime;
                    detected_label = label;
                    detected_score = Some(score);
                    csv_candidate =
                        detected_mime.starts_with("text/") || detected_mime == "application/csv";
                }

                // create a temporary file to write the download file to
                let wtr_file = NamedTempFile::new()?;

                // keep the temporary file around so we can sniff it later
                // we'll delete it when we're done
                let (mut tmp_file, path) =
                    wtr_file.keep().map_err(|_| "Cannot keep temporary file")?;

                let mut downloaded_records = 0_usize;

                if snappy_flag {
                    // we downloaded a snappy compressed file, we need to decompress it
                    // before we can sniff it
                    wtr_file_path =
                        util::decompress_snappy_file(&file.path().to_path_buf(), tmpdir)?;
                } else if args.flag_quick && !csv_candidate {
                    // when --quick is enabled, we short-circuit downloading by checking
                    // the file type from the first chunk. If the file is not a CSV,
                    // we just write the first chunk to a file and return
                    wtr_file_path = path.display().to_string();
                    tmp_file.write_all(&firstchunk)?;
                    tmp_file.flush()?;
                } else if csv_candidate {
                    // we downloaded a non-snappy file and it might be a CSV file.
                    // Rewrite it so we only have the exact sample size and truncate potentially
                    // incomplete lines. We do this coz we streamed the download and the downloaded
                    // file may be more than the sample size, and the final line may be incomplete.
                    wtr_file_path = path.display().to_string();
                    let mut wtr = Config::new(Some(wtr_file_path.clone()).as_ref())
                        .no_headers(false)
                        .flexible(true)
                        .quote_style(csv::QuoteStyle::NonNumeric)
                        .writer()?;

                    let retrieved_name = file.path().to_str().unwrap().to_string();
                    let config = Config::new(Some(retrieved_name).as_ref())
                        .delimiter(args.flag_delimiter)
                        // we say no_headers so we can just copy the downloaded file over
                        // including headers, to the exact sample size file
                        .no_headers(true)
                        .flexible(true)
                        // skip format check since temp files don't have recognizable extensions
                        .skip_format_check(true);

                    let mut rdr = config.reader()?;

                    // amortize allocation
                    #[allow(unused_assignments)]
                    let mut record = csv::ByteRecord::with_capacity(100, 20);

                    let header_row = rdr.byte_headers()?;
                    wtr.write_byte_record(header_row)?;
                    rdr.byte_records().next();

                    for rec in rdr.byte_records() {
                        record = rec?;
                        if downloaded_records >= lines_sample_size {
                            break;
                        }
                        downloaded_records += 1;
                        wtr.write_byte_record(&record)?;
                    }
                    wtr.flush()?;
                } else {
                    // we downloaded a non-snappy file and its not a CSV file just copy it over
                    wtr_file_path = path.display().to_string();
                    file.seek(SeekFrom::Start(0))?;
                    copy(&mut file, &mut tmp_file)?;
                    log::debug!("Not a CSV file, just copy it over");
                }

                Ok(SniffFileStruct {
                    display_path: url,
                    file_to_sniff: wtr_file_path,
                    detected_mime,
                    detected_label,
                    inference_score: detected_score,
                    tempfile_flag: true,
                    retrieved_size: downloaded,
                    file_size: if total_size == usize::MAX {
                        // the server didn't give us content length
                        // so send usize::MAX to indicate that we don't know the file size
                        usize::MAX
                    } else {
                        total_size
                    },
                    last_modified,
                    downloaded_records,
                })
            },
            // its a file. If its a snappy file, decompress it first
            // aftwerwards, check if its one of the supported file types
            // finally, check if its a utf8 file
            path => {
                let mut path = path;

                let mut pathbuf = PathBuf::from(path.clone());
                let file_ext = pathbuf.extension();
                match file_ext {
                    Some(ext) => {
                        let mut lower_ext =
                            ext.to_str().unwrap().to_lowercase().as_str().to_owned();
                        if lower_ext == "sz" {
                            path = util::decompress_snappy_file(&pathbuf, tmpdir)?;
                            pathbuf = PathBuf::from(path.clone());
                            lower_ext = pathbuf
                                .extension()
                                .unwrap()
                                .to_os_string()
                                .into_string()
                                .unwrap();
                            log::info!("Decompressed {lower_ext} file to {path}");
                        }
                    },
                    None => {
                        log::warn!("File extension not found");
                    },
                }

                let metadata = fs::metadata(&path)?;

                let file_size = metadata.len() as usize;
                let last_modified = match metadata.modified() {
                    Ok(time) => {
                        // format the datetime to RFC3339
                        format_systemtime(time, "%+")
                    },
                    Err(_) => "N/A".to_string(),
                };

                let canonical_path = fs::canonicalize(&path)?.to_str().unwrap().to_string();

                Ok(SniffFileStruct {
                    display_path: canonical_path,
                    file_to_sniff: path,
                    detected_mime: String::new(),
                    detected_label: String::new(),
                    inference_score: None,
                    tempfile_flag: false,
                    retrieved_size: file_size,
                    file_size,
                    last_modified,
                    downloaded_records: 0,
                })
            },
        }
    } else {
        // read from stdin and write to a temp file
        let mut stdin_file = NamedTempFile::new()?;
        let stdin = std::io::stdin();
        let mut stdin_handle = stdin.lock();
        std::io::copy(&mut stdin_handle, &mut stdin_file)?;
        drop(stdin_handle);
        let (file, path) = stdin_file
            .keep()
            .map_err(|_| "Cannot keep temporary file")?;

        if !util::isutf8_file(&path)? {
            return fail_clierror!("stdin input is not UTF8-encoded");
        }

        let metadata = file.metadata()?;

        let file_size = metadata.len() as usize;
        // set last_modified to now in RFC3339 format
        let last_modified = chrono::Utc::now().format("%+").to_string();
        let path_string = path
            .into_os_string()
            .into_string()
            .unwrap_or_else(|_| "???".to_string());

        Ok(SniffFileStruct {
            display_path: "stdin".to_string(),
            file_to_sniff: path_string,
            detected_mime: String::new(),
            detected_label: String::new(),
            inference_score: None,
            tempfile_flag: true,
            retrieved_size: file_size,
            file_size,
            last_modified,
            downloaded_records: 0,
        })
    }
}

fn cleanup_tempfile(
    tempfile_flag: bool,
    tempfile: String,
) -> Result<(), crate::clitypes::CliError> {
    if tempfile_flag {
        fs::remove_file(tempfile)?;
    }
    Ok(())
}

pub fn run(argv: &[&str]) -> CliResult<()> {
    let args: Args = util::get_args(USAGE, argv)?;

    let rt = tokio::runtime::Runtime::new().unwrap();
    rt.block_on(sniff_main(args))?;

    Ok(())
}

#[allow(clippy::unused_async)] // false positive lint
async fn sniff_main(mut args: Args) -> CliResult<()> {
    if args.flag_harvest_mode {
        args.flag_quick = true;
        args.flag_timeout = 10;
        args.flag_stats_types = true;
        args.flag_json = true;
        args.flag_user_agent =
            Some("CKAN-harvest/$QSV_VERSION ($QSV_TARGET; $QSV_BIN_NAME)".to_string());
    }

    if args.flag_just_mime {
        args.flag_no_infer = true;
    }

    let mut sample_size = args.flag_sample;
    let flag_json = args.flag_json || args.flag_pretty_json;
    if sample_size < 0.0 {
        if flag_json {
            let json_result = json!({
                "errors": [{
                    "title": "sniff error",
                    "detail": "Sample size must be greater than or equal to zero."
                }]
            });
            return fail_clierror!("{json_result}");
        }
        return fail_incorrectusage_clierror!("Sample size must be greater than or equal to zero.");
    }

    let sniffed_ts = chrono::Utc::now().to_rfc3339();
    let tmpdir = tempfile::tempdir()?;

    let future = get_file_to_sniff(&args, &tmpdir);
    let sfile_info = block_on(future)?;
    let tempfile_to_delete = sfile_info.file_to_sniff.clone();

    // if we don't have a mime type or its a snappy file and --no-infer is disabled,
    // let's try to infer the mime type
    let file_label: String;
    let file_score: Option<f32>;
    let file_type = if sfile_info.detected_mime.is_empty()
        || sfile_info.detected_mime == "application/x-snappy-framed" && !args.flag_no_infer
    {
        let (mime, label, score) =
            detect_mime_from_file(std::path::Path::new(&sfile_info.file_to_sniff))?;
        file_label = label;
        file_score = Some(score);
        mime
    } else {
        file_label = sfile_info.detected_label.clone();
        file_score = sfile_info.inference_score;
        sfile_info.detected_mime.clone()
    };

    // if the file is not a CSV candidate or --no-infer is specified,
    // we can just return the file type and exit
    if (file_type != "application/csv" && !file_type.starts_with("text/")) || args.flag_no_infer {
        cleanup_tempfile(sfile_info.tempfile_flag, tempfile_to_delete)?;

        let size = if sfile_info.file_size == usize::MAX {
            "Unknown".to_string()
        } else {
            sfile_info.file_size.to_string()
        };
        let last_modified = sfile_info.last_modified;

        if flag_json {
            if args.flag_no_infer {
                let json_result = json!({
                    "title": "sniff mime type",
                    "meta": {
                        "detected_mime_type": file_type,
                        "detected_label": file_label,
                        "size": size,
                        "last_modified": last_modified,
                    }
                });
                if args.flag_pretty_json {
                    woutinfo!("{}", simd_json::to_string_pretty(&json_result).unwrap());
                    return Ok(());
                }
                woutinfo!("{json_result}");
                return Ok(());
            }
            let json_result = json!({
                "errors": [{
                    "title": "sniff error",
                    "detail": format!("File is not a CSV file. Detected mime type: {file_type}"),
                    "meta": {
                        "detected_mime_type": file_type,
                        "detected_label": file_label,
                        "size": size,
                        "last_modified": last_modified,
                    }
                }]
            });
            if args.flag_pretty_json {
                return fail_clierror!("{}", simd_json::to_string_pretty(&json_result).unwrap());
            }
            return fail_clierror!("{json_result}");
        }
        if args.flag_no_infer {
            woutinfo!(
                "Detected mime type: {file_type}, size: {size}, last modified: {last_modified}"
            );
            return Ok(());
        }
        return fail_clierror!(
            "File is not a CSV file. Detected mime type: {file_type}, size: {size}, last \
             modified: {last_modified}"
        );
    }

    let conf = Config::new(Some(sfile_info.file_to_sniff.clone()).as_ref())
        .flexible(true)
        .delimiter(args.flag_delimiter);
    let n_rows = if sfile_info.downloaded_records == 0 {
        //if we have the whole file and not just a sample, we can count the number of rows
        match util::count_rows_regular(&conf) {
            Ok(n) => n as usize,
            Err(e) => {
                cleanup_tempfile(sfile_info.tempfile_flag, tempfile_to_delete)?;

                if flag_json {
                    let json_result = json!({
                        "errors": [{
                            "title": "count rows error",
                            "detail": e.to_string()
                        }]
                    });
                    return fail_clierror!("{json_result}");
                }
                return fail_clierror!("{}", e);
            },
        }
    } else {
        // usize::MAX is a sentinel value to let us
        // know that we need to estimate the number of records
        // since we only downloaded a sample, not the entire file
        usize::MAX
    };

    // its an empty file, exit with an error
    if n_rows == 0 {
        cleanup_tempfile(sfile_info.tempfile_flag, tempfile_to_delete)?;

        if flag_json {
            let json_result = json!({
                "errors": [{
                    "title": "sniff error",
                    "detail": "Empty file"
                }]
            });
            return fail_clierror!("{json_result}");
        }
        return fail_clierror!("Empty file");
    }

    let mut sample_all = false;
    // its a percentage, get the actual sample size
    #[allow(clippy::cast_precision_loss)]
    if sample_size < 1.0 {
        sample_size *= n_rows as f64;
    } else if (sample_size).abs() < f64::EPSILON {
        // its zero, the epsilon bit is because comparing a float
        // is really not precise - see https://floating-point-gui.de/errors/comparison/
        sample_all = true;
    }

    // for a local file and stdin, set sampled_records to the sample size
    // for a remote file, set sampled_records to the number of rows downloaded
    let sampled_records = if sfile_info.downloaded_records == 0 {
        sample_size as usize
    } else {
        sample_all = true;
        sfile_info.downloaded_records
    };

    let rdr = conf.clone().skip_format_check(true).reader_file()?;

    let dt_preference = if args.flag_prefer_dmy || conf.get_dmy_preference() {
        DatePreference::DmyFormat
    } else {
        DatePreference::MdyFormat
    };

    if let Some(save_urlsample) = args.flag_save_urlsample {
        fs::copy(sfile_info.file_to_sniff.clone(), save_urlsample)?;
    }

    let quote_char = match args.flag_quote {
        Some(quote_char) => {
            csv_nose::metadata::Quote::Some(*quote_char.to_string().as_bytes().first().unwrap())
        },
        _ => csv_nose::metadata::Quote::None,
    };

    // now that we have all the sniffing parameters, we can sniff the file
    let sniff_results = if sample_all {
        log::info!("Sniffing ALL rows...");
        if let Some(delimiter) = args.flag_delimiter {
            Sniffer::new()
                .sample_size(SampleSize::All)
                .date_preference(dt_preference)
                .delimiter(delimiter.as_byte())
                .quote(quote_char)
                .sniff_reader(rdr.into_inner())
        } else {
            Sniffer::new()
                .sample_size(SampleSize::All)
                .date_preference(dt_preference)
                .sniff_reader(rdr.into_inner())
        }
    } else {
        let mut sniff_size = sample_size as usize;
        // sample_size is at least 20
        if sniff_size < 20 {
            sniff_size = 20;
        }
        log::info!("Sniffing {sniff_size} rows...");
        if let Some(delimiter) = args.flag_delimiter {
            Sniffer::new()
                .sample_size(SampleSize::Records(sniff_size))
                .date_preference(dt_preference)
                .delimiter(delimiter.as_byte())
                .quote(quote_char)
                .sniff_reader(rdr.into_inner())
        } else {
            Sniffer::new()
                .sample_size(SampleSize::Records(sniff_size))
                .date_preference(dt_preference)
                .sniff_reader(rdr.into_inner())
        }
    };

    let mut processed_results = SniffStruct::default();
    let mut sniff_error: Option<String> = None;

    match sniff_results {
        Ok(metadata) => {
            let (num_records, estimated) = rowcount(&metadata, &sfile_info, n_rows);

            let sniffedfields = metadata
                .fields
                .iter()
                .map(std::string::ToString::to_string)
                .collect();
            let sniffedtypes = metadata
                .types
                .iter()
                .map(std::string::ToString::to_string)
                .collect();

            let delimiter_char = metadata.dialect.delimiter as char;
            processed_results = SniffStruct {
                path: sfile_info.display_path,
                sniff_timestamp: sniffed_ts,
                last_modified: sfile_info.last_modified.clone(),
                delimiter_char,
                header_row: metadata.dialect.header.has_header_row,
                preamble_rows: metadata.dialect.header.num_preamble_rows,
                quote_char: match metadata.dialect.quote {
                    csv_nose::metadata::Quote::Some(chr) => format!("{}", char::from(chr)),
                    csv_nose::metadata::Quote::None => "none".into(),
                },
                flexible: metadata.dialect.flexible,
                is_utf8: metadata.dialect.is_utf8,
                detected_mime: if delimiter_char == ',' {
                    "application/csv".to_string()
                } else {
                    file_type.clone()
                },
                detected_label: if delimiter_char == ',' {
                    "csv".to_string()
                } else if delimiter_char == '\t' {
                    "tsv".to_string()
                } else {
                    file_label.clone()
                },
                inference_score: file_score.map(|score| {
                    // round to 4 decimal places without string round-trip
                    (score * 10_000.0).round() / 10_000.0
                }),
                retrieved_size: sfile_info.retrieved_size,
                file_size: sfile_info.file_size,
                sampled_records: if sampled_records > num_records {
                    num_records
                } else {
                    sampled_records
                },
                estimated,
                num_records,
                avg_record_len: metadata.avg_record_len,
                num_fields: metadata.num_fields,
                fields: sniffedfields,
                types: sniffedtypes,
                stats_types: args.flag_stats_types,
            };
        },
        Err(e) => {
            sniff_error = Some(format!("{e}"));
        },
    }

    cleanup_tempfile(sfile_info.tempfile_flag, tempfile_to_delete)?;

    // safety: we just created all these json values above, so they are safe to unwrap
    if flag_json {
        if sniff_error.is_none() {
            if args.flag_pretty_json {
                println!(
                    "{}",
                    simd_json::to_string_pretty(&processed_results).unwrap()
                );
            } else {
                println!("{}", simd_json::to_string(&processed_results).unwrap());
            }
            return Ok(());
        }
        let sniff_error_json = json!({
            "title": "sniff error",
            "detail": format!("{}", sniff_error.unwrap()),
            "meta": {
                "detected_mime_type": file_type,
                "detected_label": file_label,
                "size": sfile_info.file_size,
                "last_modified": sfile_info.last_modified,
            }
        });
        let error_msg = if args.flag_pretty_json {
            simd_json::to_string_pretty(&sniff_error_json).unwrap()
        } else {
            simd_json::to_string(&sniff_error_json).unwrap()
        };
        return fail_clierror!("{error_msg}");
    }

    if let Some(error) = sniff_error {
        fail_clierror!(
            "{error}; size: {size}, last_modified: {last_modified}",
            size = sfile_info.file_size,
            last_modified = sfile_info.last_modified
        )
    } else {
        println!("{processed_results}");
        Ok(())
    }
}