1
|
|
|
// Copyright 2018 Fedir RYKHTIK. All rights reserved. |
2
|
|
|
// Use of this source code is governed by the GNU GPL 3.0 |
3
|
|
|
// license that can be found in the LICENSE file. |
4
|
|
|
|
5
|
|
|
// ghstat - statistical multi-criteria decision-making comparator for Github's projects. |
6
|
|
|
package main |
7
|
|
|
|
8
|
|
|
import ( |
9
|
|
|
"encoding/csv" |
10
|
|
|
"encoding/json" |
11
|
|
|
"flag" |
12
|
|
|
"fmt" |
13
|
|
|
"log" |
14
|
|
|
"os" |
15
|
|
|
"path/filepath" |
16
|
|
|
"strings" |
17
|
|
|
"time" |
18
|
|
|
|
19
|
|
|
"github.com/fedir/ghstat/github" |
20
|
|
|
"github.com/fedir/ghstat/httpcache" |
21
|
|
|
"github.com/fedir/ghstat/timing" |
22
|
|
|
) |
23
|
|
|
|
24
|
|
|
func main() { |
25
|
1 |
|
var ( |
26
|
|
|
clearHTTPCache = flag.Bool("cc", false, "Clear HTTP cache") |
27
|
|
|
clearHTTPCacheDryRun = flag.Bool("ccdr", false, "Clear HTTP cache (dry run)") |
28
|
|
|
debug = flag.Bool("d", false, "Debug mode") |
29
|
|
|
resultFileSavePath = flag.String("f", "", "File path where result CSV file will be saved") |
30
|
|
|
rateLimitCheck = flag.Bool("l", false, "Rate limit check") |
31
|
|
|
repositoriesKeysManual = flag.String("r", "", "Repositories keys") |
32
|
|
|
tmpFolder = flag.String("t", "test_data", "Clear HTTP cache (dry run)") |
33
|
|
|
repositoriesKeys = []string{} |
34
|
|
|
) |
35
|
1 |
|
flag.Parse() |
36
|
1 |
|
if *clearHTTPCache || *clearHTTPCacheDryRun { |
37
|
|
|
clearHTTPCacheFolder(*tmpFolder, *clearHTTPCacheDryRun) |
38
|
|
|
os.Exit(0) |
39
|
|
|
} |
40
|
1 |
|
if *rateLimitCheck { |
41
|
|
|
checkAndPrintRateLimit() |
42
|
|
|
os.Exit(0) |
43
|
|
|
} |
44
|
1 |
|
if *rateLimitCheck { |
45
|
|
|
checkAndPrintRateLimit() |
46
|
|
|
os.Exit(0) |
47
|
|
|
} |
48
|
1 |
|
if *repositoriesKeysManual != "" { |
49
|
|
|
repositoriesKeys = strings.Split(*repositoriesKeysManual, ",") |
50
|
|
|
} else { |
51
|
1 |
|
repositoriesKeys = []string{ |
52
|
|
|
"astaxie/beego", |
53
|
|
|
"gohugoio/hugo", |
54
|
|
|
"gin-gonic/gin", |
55
|
|
|
"labstack/echo", |
56
|
|
|
"revel/revel", |
57
|
|
|
"gobuffalo/buffalo", |
58
|
|
|
"go-chi/chi", |
59
|
|
|
"kataras/iris", |
60
|
|
|
"zenazn/goji", |
61
|
|
|
} |
62
|
|
|
} |
63
|
1 |
|
csvFilePath := "" |
64
|
1 |
|
if *resultFileSavePath != "" { |
65
|
|
|
csvFilePath = *resultFileSavePath |
66
|
|
|
} else { |
67
|
1 |
|
csvFilePath = "result.csv" |
68
|
|
|
} |
69
|
1 |
|
var ghData = [][]string{} |
70
|
1 |
|
headers := []string{ |
71
|
|
|
"Name", |
72
|
|
|
"URL", |
73
|
|
|
"Author", |
74
|
|
|
"License", |
75
|
|
|
"Author's followers", |
76
|
|
|
"Top 10 contributors followers", |
77
|
|
|
"Created at", |
78
|
|
|
"Age in days", |
79
|
|
|
"Total commits", |
80
|
|
|
"Total additions", |
81
|
|
|
"Total deletions", |
82
|
|
|
"Total code changes", |
83
|
|
|
"Last commit date", |
84
|
|
|
"Commits/day", |
85
|
|
|
"Medium commit size", |
86
|
|
|
"Total releases", |
87
|
|
|
"Stargazers", |
88
|
|
|
"Forks", |
89
|
|
|
"Contributors", |
90
|
|
|
"Active forkers, %", |
91
|
|
|
"Open issues", |
92
|
|
|
"Total issues", |
93
|
|
|
"Issue/day", |
94
|
|
|
"Closed issues, %", |
95
|
|
|
"Place", |
96
|
|
|
} |
97
|
1 |
|
ghDataColumnIndexes := map[string]int{ |
98
|
|
|
"nameColumn": 0, |
99
|
|
|
"urlColumn": 1, |
100
|
|
|
"authorColumn": 2, |
101
|
|
|
"licenseColumn": 3, |
102
|
|
|
"authorsFollowersColumn": 4, |
103
|
|
|
"top10ContributorsFollowersColumn": 5, |
104
|
|
|
"ageColumn": 7, |
105
|
|
|
"totalCommitsColumn": 8, |
106
|
|
|
"totalAdditionsColumn": 9, |
107
|
|
|
"totalDeletionsColumn": 10, |
108
|
|
|
"totalCodeChangesColumn": 11, |
109
|
|
|
"lastCommitDateColumn": 12, |
110
|
|
|
"commitsByDayColumn": 13, |
111
|
|
|
"mediCommitSizeColumn": 14, |
112
|
|
|
"totalTagsColumn": 15, |
113
|
|
|
"stargazersColumn": 16, |
114
|
|
|
"activeForkersColumn": 19, |
115
|
|
|
"issuesByDayColumn": 22, |
116
|
|
|
"closedIssuesPercentageColumn": 23, |
117
|
|
|
"totalPointsColumnIndex": 24, |
118
|
|
|
} |
119
|
1 |
|
dataChan := make(chan []string, len(repositoriesKeys)) |
120
|
1 |
|
for _, rKey := range repositoriesKeys { |
121
|
1 |
|
go fillRepositoryStatistics(rKey, *tmpFolder, *debug, dataChan) |
122
|
|
|
} |
123
|
1 |
|
for range repositoriesKeys { |
124
|
1 |
|
ghData = append(ghData, <-dataChan) |
125
|
|
|
} |
126
|
1 |
|
greetings := rateGhData(ghData, ghDataColumnIndexes) |
127
|
1 |
|
fmt.Println(greetings) |
128
|
1 |
|
writeCsv(csvFilePath, headers, ghData) |
129
|
|
|
} |
130
|
|
|
|
131
|
|
|
func fillRepositoryStatistics(rKey string, tmpFolder string, debug bool, dataChan chan []string) { |
132
|
1 |
|
repositoryData := github.GetRepositoryStatistics(rKey, tmpFolder, debug) |
133
|
1 |
|
repositoryAge := int(time.Since(repositoryData.CreatedAt).Seconds() / 86400) |
134
|
1 |
|
authorLogin, lastCommitDate := github.GetRepositoryCommitsData(rKey, tmpFolder, debug) |
135
|
1 |
|
authorFollowers := 0 |
136
|
1 |
|
if authorLogin != "" { |
137
|
1 |
|
authorFollowers = github.GetUserFollowers(authorLogin, tmpFolder, debug) |
138
|
|
|
} |
139
|
1 |
|
closedIssues := github.GetRepositoryClosedIssues(rKey, tmpFolder, debug) |
140
|
1 |
|
topContributorsFollowers, totalContributors := github.GetRepositoryContributors(rKey, tmpFolder, debug) |
141
|
1 |
|
totalTags := github.GetRepositoryTagsNumber(rKey, tmpFolder, debug) |
142
|
1 |
|
activeForkersPercentage := github.GetActiveForkersPercentage(totalContributors, repositoryData.Forks) |
143
|
1 |
|
issueByDay := github.GetIssueByDay(closedIssues+repositoryData.OpenIssues, repositoryAge) |
144
|
1 |
|
closedIssuesPercentage := github.GetClosedIssuesPercentage(repositoryData.OpenIssues, int(closedIssues)) |
145
|
1 |
|
contributionStatistics := github.GetContributionStatistics(rKey, tmpFolder, debug) |
146
|
1 |
|
commitsByDay := github.GetCommitsByDay(contributionStatistics.TotalCommits, repositoryAge) |
147
|
1 |
|
ghProjectData := []string{ |
148
|
|
|
repositoryData.FullName, |
149
|
|
|
fmt.Sprintf("https://github.com/%s", repositoryData.FullName), |
150
|
|
|
fmt.Sprintf("%s", func(a string) string { |
151
|
1 |
|
if a == "" { |
152
|
1 |
|
a = "[Account removed]" |
153
|
|
|
} |
154
|
1 |
|
return a |
155
|
|
|
}(authorLogin)), |
156
|
|
|
fmt.Sprintf("%s", func(l string) string { |
157
|
1 |
|
if l == "" { |
158
|
1 |
|
l = "[Unknown]" |
159
|
|
|
} |
160
|
1 |
|
return l |
161
|
|
|
}(repositoryData.License.SPDXID)), |
162
|
|
|
fmt.Sprintf("%d", authorFollowers), |
163
|
|
|
fmt.Sprintf("%d", topContributorsFollowers), |
164
|
|
|
fmt.Sprintf("%d/%02d", repositoryData.CreatedAt.Year(), repositoryData.CreatedAt.Month()), |
165
|
|
|
fmt.Sprintf("%d", repositoryAge), |
166
|
|
|
fmt.Sprintf("%d", contributionStatistics.TotalCommits), |
167
|
|
|
fmt.Sprintf("%d", contributionStatistics.TotalAdditions), |
168
|
|
|
fmt.Sprintf("%d", contributionStatistics.TotalDeletions), |
169
|
|
|
fmt.Sprintf("%d", contributionStatistics.TotalCodeChanges), |
170
|
|
|
fmt.Sprintf(lastCommitDate.Format("2006-01-02 15:04:05")), |
|
|
|
|
171
|
|
|
fmt.Sprintf("%.4f", commitsByDay), |
172
|
|
|
fmt.Sprintf("%d", contributionStatistics.MediumCommitSize), |
173
|
|
|
fmt.Sprintf("%d", totalTags), |
174
|
|
|
fmt.Sprintf("%d", repositoryData.Watchers), |
175
|
|
|
fmt.Sprintf("%d", repositoryData.Forks), |
176
|
|
|
fmt.Sprintf("%d", totalContributors), |
177
|
|
|
fmt.Sprintf("%.2f", activeForkersPercentage), |
178
|
|
|
fmt.Sprintf("%d", repositoryData.OpenIssues), |
179
|
|
|
fmt.Sprintf("%d", closedIssues+repositoryData.OpenIssues), |
180
|
|
|
fmt.Sprintf("%.4f", issueByDay), |
181
|
|
|
fmt.Sprintf("%.2f", closedIssuesPercentage), |
182
|
|
|
"0", |
183
|
|
|
} |
184
|
1 |
|
dataChan <- ghProjectData |
185
|
|
|
} |
186
|
|
|
|
187
|
|
|
func clearHTTPCacheFolder(tmpFolderPath string, dryRun bool) error { |
188
|
|
|
d, err := os.Open(tmpFolderPath) |
189
|
|
|
if err != nil { |
190
|
|
|
log.Fatalf("Could not open %s", tmpFolderPath) |
191
|
|
|
} |
192
|
|
|
defer d.Close() |
193
|
|
|
names, err := d.Readdirnames(-1) |
194
|
|
|
if err != nil { |
195
|
|
|
log.Fatalf("Could not read from %s", tmpFolderPath) |
196
|
|
|
} |
197
|
|
|
for _, name := range names { |
198
|
|
|
fp := filepath.Join(tmpFolderPath, name) |
199
|
|
|
if dryRun { |
200
|
|
|
fmt.Printf("Deleted %s\n", fp) |
201
|
|
|
} else { |
202
|
|
|
err = os.RemoveAll(fp) |
203
|
|
|
if err != nil { |
204
|
|
|
log.Fatalf("Could not remove %s", fp) |
205
|
|
|
} |
206
|
|
|
fmt.Printf("Deleted %s\n", fp) |
207
|
|
|
} |
208
|
|
|
} |
209
|
|
|
return nil |
210
|
|
|
} |
211
|
|
|
|
212
|
|
|
func checkAndPrintRateLimit() { |
213
|
|
|
type RateLimits struct { |
214
|
|
|
Resources struct { |
215
|
|
|
Core struct { |
216
|
|
|
Limit int `json:"limit"` |
217
|
|
|
Remaining int `json:"remaining"` |
218
|
|
|
Reset int `json:"reset"` |
219
|
|
|
} `json:"core"` |
220
|
|
|
Search struct { |
221
|
|
|
Limit int `json:"limit"` |
222
|
|
|
Remaining int `json:"remaining"` |
223
|
|
|
Reset int `json:"reset"` |
224
|
|
|
} `json:"search"` |
225
|
|
|
GraphQL struct { |
226
|
|
|
Limit int `json:"limit"` |
227
|
|
|
Remaining int `json:"remaining"` |
228
|
|
|
Reset int `json:"reset"` |
229
|
|
|
} `json:"graphql"` |
230
|
|
|
} `json:"resources"` |
231
|
|
|
Rate struct { |
232
|
|
|
Limit int `json:"limit"` |
233
|
|
|
Remaining int `json:"remaining"` |
234
|
|
|
Reset int `json:"reset"` |
235
|
|
|
} `json:"rate"` |
236
|
|
|
} |
237
|
|
|
url := "https://api.github.com/rate_limit" |
238
|
|
|
resp, statusCode, err := httpcache.MakeHTTPRequest(url) |
239
|
|
|
if err != nil { |
240
|
|
|
log.Fatalf("Error during checking rate limit : %d %v#", statusCode, err) |
241
|
|
|
} |
242
|
|
|
jsonResponse, _, _ := httpcache.ReadResp(resp) |
243
|
|
|
rateLimits := RateLimits{} |
244
|
|
|
json.Unmarshal(jsonResponse, &rateLimits) |
245
|
|
|
fmt.Printf("Core: %d/%d (reset in %d minutes)\n", rateLimits.Resources.Core.Remaining, rateLimits.Resources.Core.Limit, timing.GetRelativeTime(rateLimits.Resources.Core.Reset)) |
246
|
|
|
fmt.Printf("Search: %d/%d (reset in %d minutes)\n", rateLimits.Resources.Search.Remaining, rateLimits.Resources.Search.Limit, timing.GetRelativeTime(rateLimits.Resources.Search.Reset)) |
247
|
|
|
fmt.Printf("GraphQL: %d/%d (reset in %d minutes)\n", rateLimits.Resources.GraphQL.Remaining, rateLimits.Resources.GraphQL.Limit, timing.GetRelativeTime(rateLimits.Resources.GraphQL.Reset)) |
248
|
|
|
fmt.Printf("Rate: %d/%d (reset in %d minutes)\n", rateLimits.Rate.Remaining, rateLimits.Rate.Limit, timing.GetRelativeTime(rateLimits.Rate.Reset)) |
249
|
|
|
} |
250
|
|
|
|
251
|
|
|
func writeCsv(csvFilePath string, headers []string, ghData [][]string) { |
252
|
1 |
|
file, err := os.Create(csvFilePath) |
253
|
1 |
|
if err != nil { |
254
|
|
|
log.Fatal("Cannot create file", err) |
255
|
|
|
} |
256
|
1 |
|
defer file.Close() |
257
|
1 |
|
writer := csv.NewWriter(file) |
258
|
1 |
|
defer writer.Flush() |
259
|
1 |
|
err = writer.Write(headers) |
260
|
1 |
|
if err != nil { |
261
|
|
|
log.Fatal("Cannot write to file", err) |
262
|
|
|
} |
263
|
1 |
|
for _, value := range ghData { |
264
|
1 |
|
err := writer.Write(value) |
265
|
1 |
|
if err != nil { |
266
|
|
|
log.Fatal("Cannot write to file", err) |
267
|
|
|
} |
268
|
|
|
} |
269
|
|
|
} |
270
|
|
|
|