-
Notifications
You must be signed in to change notification settings - Fork 7
/
main.go
213 lines (190 loc) · 5.12 KB
/
main.go
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
// Given as single file with crossref works API message, create a potentially
// smaller file, which contains only the most recent version of each document.
//
// Works in a three stage, two pass fashion: (1) extract, (2) identify, (3) extract.
// Performance data point (30M compressed records, 11m33.871s):
//
// 2017/07/24 18:26:10 stage 1: 8m13.799431646s
// 2017/07/24 18:26:55 stage 2: 45.746997314s
// 2017/07/24 18:29:30 stage 3: 2m34.23537293s
package main
import (
"bufio"
"bytes"
"encoding/json"
"flag"
"fmt"
"io"
"io/ioutil"
"log"
"os"
"os/exec"
"runtime/pprof"
"strings"
gzip "github.com/klauspost/pgzip"
"github.com/miku/clam"
"github.com/miku/span"
"github.com/miku/span/formats/crossref"
"github.com/miku/span/parallel"
)
// fallback awk script is used, if the filterline executable is not found.
var fallback = `
#!/bin/bash
LIST="$1" LC_ALL=C awk '
function nextline() {
if ((getline n < list) <=0) exit
}
BEGIN{
list = ENVIRON["LIST"]
nextline()
}
NR == n {
print
nextline()
}' < "$2"
`
// WriteFields writes a variable number of fields as tab separated values into a writer.
func WriteFields(w io.Writer, values ...interface{}) (int, error) {
var s []string
for _, v := range values {
s = append(s, fmt.Sprintf("%v", v))
}
return io.WriteString(w, fmt.Sprintf("%s\n", strings.Join(s, "\t")))
}
func main() {
excludeFile := flag.String("x", "", "a list of DOI to further ignore")
outputFile := flag.String("o", "", "output file")
compressed := flag.Bool("z", false, "input is gzip compressed")
batchsize := flag.Int("b", 100000, "batch size")
cpuprofile := flag.String("cpuprofile", "", "write cpuprofile to file")
flag.Parse()
if *cpuprofile != "" {
f, err := os.Create(*cpuprofile)
if err != nil {
log.Fatal(err)
}
pprof.StartCPUProfile(f)
defer pprof.StopCPUProfile()
}
if flag.NArg() == 0 {
log.Fatal("input file required")
}
if *outputFile == "" {
log.Fatal("output filename required")
}
var reader io.Reader
f, err := os.Open(flag.Arg(0))
if err != nil {
log.Fatal(err)
}
defer f.Close()
reader = f
if *compressed {
g, err := gzip.NewReader(f)
if err != nil {
log.Fatal(err)
}
defer g.Close()
reader = g
}
excludes := make(map[string]struct{})
if *excludeFile != "" {
file, err := os.Open(*excludeFile)
if err != nil {
log.Fatal(err)
}
defer file.Close()
if err := span.LoadSet(file, excludes); err != nil {
log.Fatal(err)
}
log.Printf("excludes: %d", len(excludes))
}
// Stage 1: Extract minimum amount of information from the raw data, write to tempfile.
tf, err := ioutil.TempFile("", "span-crossref-snapshot-")
if err != nil {
log.Fatal(err)
}
br := bufio.NewReader(reader)
bw := bufio.NewWriter(tf)
p := parallel.NewProcessor(br, bw, func(lineno int64, b []byte) ([]byte, error) {
var doc crossref.Document
if err := json.Unmarshal(b, &doc); err != nil {
return nil, err
}
date, err := doc.Deposited.Date()
if err != nil {
return nil, err
}
if _, ok := excludes[doc.DOI]; ok {
return nil, nil
}
var buf bytes.Buffer
if _, err := WriteFields(&buf, lineno+1, date.Format("2006-01-02"), doc.DOI); err != nil {
return nil, err
}
return buf.Bytes(), nil
})
p.BatchSize = *batchsize
if err := p.Run(); err != nil {
log.Fatal(err)
}
if err := bw.Flush(); err != nil {
log.Fatal(err)
}
if err := tf.Close(); err != nil {
log.Fatal(err)
}
// Stage 2: Identify relevant records. Sort by DOI (3), then date reversed (2);
// then unique by DOI (3). Should keep the entry of the last update (filename,
// document date, DOI).
fastsort := `LC_ALL=C sort -S20%`
cmd := `{{ f }} -k3,3 -rk2,2 {{ input }} | {{ f }} -k3,3 -u | cut -f1 | {{ f }} -n > {{ output }}`
output, err := clam.RunOutput(cmd, clam.Map{"f": fastsort, "input": tf.Name()})
if err != nil {
log.Fatal(err)
}
// External tools and fallbacks for stage 3. comp, decomp, filterline.
comp, decomp := `gzip -c`, `gunzip -c`
if _, err := exec.LookPath("unpigz"); err == nil {
comp, decomp = `pigz -c`, `unpigz -c`
}
filterline := `filterline`
if _, err := exec.LookPath("filterline"); err != nil {
if _, err := exec.LookPath("awk"); err != nil {
log.Fatal("filterline (git.io/v7qak) or awk is required")
}
tf, err := ioutil.TempFile("", "span-crossref-snapshot-filterline-")
if err != nil {
log.Fatal(err)
}
if _, err := io.WriteString(tf, fallback); err != nil {
log.Fatal(err)
}
if err := tf.Close(); err != nil {
log.Fatal(err)
}
if err := os.Chmod(tf.Name(), 0755); err != nil {
log.Fatal(err)
}
defer os.Remove(tf.Name())
filterline = tf.Name()
}
// Stage 3: Extract relevant records. Compressed input will be recompressed again.
cmd = `{{ filterline }} {{ L }} {{ F }} > {{ output }}`
if *compressed {
cmd = `{{ filterline }} {{ L }} <({{ decomp }} {{ F }}) | {{ comp }} > {{ output }}`
}
output, err = clam.RunOutput(cmd, clam.Map{
"L": output,
"F": f.Name(),
"filterline": filterline,
"decomp": decomp,
"comp": comp,
})
if err != nil {
log.Fatal(err)
}
if err := os.Rename(output, *outputFile); err != nil {
log.Fatal(err)
}
}