diff options
author | rtz12 <koenig@fagott.pw> | 2016-10-02 02:59:08 (UTC) |
---|---|---|
committer | rtz12 <koenig@fagott.pw> | 2016-10-02 02:59:08 (UTC) |
commit | 5997487fc5cc2126fb5afa12b75b67f3985e11d2 (patch) | |
tree | d0631885961a35daf7ec302161ea888ae02183aa /tools | |
parent | b799217c66df7c1c4ef285699273a5222efccfb0 (diff) |
Importer etwas netter strukturiert und Support fuer die neuen ACD-Daten
hinzugefuegt
Diffstat (limited to 'tools')
-rw-r--r-- | tools/importer/.gitignore | 1 | ||||
-rw-r--r-- | tools/importer/ACDv2Reader.go | 66 | ||||
-rw-r--r-- | tools/importer/main.go | 183 |
3 files changed, 175 insertions, 75 deletions
diff --git a/tools/importer/.gitignore b/tools/importer/.gitignore new file mode 100644 index 0000000..a6c57f5 --- /dev/null +++ b/tools/importer/.gitignore | |||
@@ -0,0 +1 @@ | |||
*.json | |||
diff --git a/tools/importer/ACDv2Reader.go b/tools/importer/ACDv2Reader.go new file mode 100644 index 0000000..2ddca4d --- /dev/null +++ b/tools/importer/ACDv2Reader.go | |||
@@ -0,0 +1,66 @@ | |||
1 | package main | ||
2 | |||
3 | import ( | ||
4 | "encoding/json" | ||
5 | "io/ioutil" | ||
6 | "path/filepath" | ||
7 | "strings" | ||
8 | |||
9 | "fagott.pw/grilist/modules/grils" | ||
10 | ) | ||
11 | |||
12 | type ACDv2Reader struct{} | ||
13 | |||
14 | type ACDv2NameValue struct { | ||
15 | Name string `json:"name"` | ||
16 | Value string `json:"value"` | ||
17 | } | ||
18 | |||
19 | type ACDv2Gril struct { | ||
20 | Name struct { | ||
21 | Romaji string `json:"romaji"` | ||
22 | Japanese string `json:"japanese"` | ||
23 | Aliases []string `json:"aliases"` | ||
24 | } `json:"name"` | ||
25 | Image struct { | ||
26 | Thumbnail string `json:"thumb"` | ||
27 | Full string `json:"full"` | ||
28 | } `json:"image"` | ||
29 | Tags []struct { | ||
30 | ID int `json:"id"` | ||
31 | Name string `json:"name"` | ||
32 | } `json:"tags"` | ||
33 | Traits struct { | ||
34 | Official []ACDv2NameValue `json:"official"` | ||
35 | Indexed []ACDv2NameValue `json:"indexed"` | ||
36 | } `json:"traits"` | ||
37 | Assignments []int `json:"assignments"` | ||
38 | CharsWithSimilarTraits []int `json:"chars_similar_traits"` | ||
39 | Extra []ACDv2NameValue `json:"extra"` | ||
40 | Role string `json:"role"` | ||
41 | } | ||
42 | |||
43 | func (r *ACDv2Reader) Read(path string) WrappedGril { | ||
44 | g := WrappedGril{} | ||
45 | g.Gril = grils.Gril{} | ||
46 | data, err := ioutil.ReadFile(path) | ||
47 | ifErrExit(err) | ||
48 | var jObj ACDv2Gril | ||
49 | err = json.Unmarshal(data, &jObj) | ||
50 | ifErrExit(err) | ||
51 | g.Image = jObj.Image.Full | ||
52 | g.Thumb = jObj.Image.Thumbnail | ||
53 | g.Gril.ID = toInt(strings.TrimSuffix(filepath.Base(path), ".json")) | ||
54 | g.Gril.KanjiName = jObj.Name.Japanese | ||
55 | g.Gril.RomajiName = jObj.Name.Romaji | ||
56 | g.Gril.OtherNames = jObj.Name.Aliases | ||
57 | g.Gril.Tags = make([]string, 0) | ||
58 | for _, v := range jObj.Tags { | ||
59 | g.Gril.Tags = append(g.Gril.Tags, v.Name) | ||
60 | } | ||
61 | return g | ||
62 | } | ||
63 | |||
64 | func (r *ACDv2Reader) ID() int { | ||
65 | return int(grils.DataSourceACD) | ||
66 | } | ||
diff --git a/tools/importer/main.go b/tools/importer/main.go index 9caf287..ba98fb5 100644 --- a/tools/importer/main.go +++ b/tools/importer/main.go | |||
@@ -13,6 +13,13 @@ import ( | |||
13 | _ "github.com/lib/pq" | 13 | _ "github.com/lib/pq" |
14 | ) | 14 | ) |
15 | 15 | ||
16 | var path string | ||
17 | var sourceType string | ||
18 | var imgUrlExportFile string | ||
19 | var exportFile *os.File | ||
20 | var db *sql.DB | ||
21 | var r GrilReader | ||
22 | |||
16 | type WrappedGril struct { | 23 | type WrappedGril struct { |
17 | Gril grils.Gril | 24 | Gril grils.Gril |
18 | Image string | 25 | Image string |
@@ -28,6 +35,8 @@ func GetGrilReader(rType string) GrilReader { | |||
28 | switch rType { | 35 | switch rType { |
29 | case "ACD": | 36 | case "ACD": |
30 | return &ACDReader{} | 37 | return &ACDReader{} |
38 | case "ACDv2": | ||
39 | return &ACDv2Reader{} | ||
31 | } | 40 | } |
32 | return nil | 41 | return nil |
33 | } | 42 | } |
@@ -38,15 +47,100 @@ func LogErr(err error) { | |||
38 | } | 47 | } |
39 | } | 48 | } |
40 | 49 | ||
50 | func TryGetExistingGril(g WrappedGril, idmap map[int]int) int { | ||
51 | dbID, _ := idmap[g.Gril.ID] | ||
52 | return dbID | ||
53 | } | ||
54 | |||
55 | func InsertGril(g WrappedGril) int { | ||
56 | var dbID int | ||
57 | row := db.QueryRow("INSERT INTO grilist.grils (age) VALUES (NULL) RETURNING id;") | ||
58 | row.Scan(&dbID) | ||
59 | fmt.Printf( | ||
60 | "Assigned %d from %s to %d\n", | ||
61 | g.Gril.ID, | ||
62 | sourceType, | ||
63 | dbID) | ||
64 | db.Exec(`INSERT INTO grilist.grils_id_mappings (gril_id, | ||
65 | source, source_id) VALUES ($1, $2, $3);`, | ||
66 | dbID, r.ID(), g.Gril.ID) | ||
67 | if g.Gril.KanjiName != "" { | ||
68 | _, err := db.Exec(`INSERT INTO | ||
69 | grilist.gril_names (gril_id, name, name_type) | ||
70 | VALUES ($1, $2, $3);`, dbID, g.Gril.KanjiName, | ||
71 | 0) | ||
72 | LogErr(err) | ||
73 | } | ||
74 | if g.Gril.RomajiName != "" { | ||
75 | _, err := db.Exec(`INSERT INTO | ||
76 | grilist.gril_names (gril_id, name, name_type) | ||
77 | VALUES ($1, $2, $3);`, dbID, g.Gril.RomajiName, | ||
78 | 1) | ||
79 | LogErr(err) | ||
80 | } | ||
81 | for _, v := range g.Gril.OtherNames { | ||
82 | _, err := db.Exec(`INSERT INTO | ||
83 | grilist.gril_names (gril_id, name, name_type) | ||
84 | VALUES ($1, $2, $3);`, dbID, v, 2) | ||
85 | LogErr(err) | ||
86 | } | ||
87 | if exportFile != nil { | ||
88 | if g.Image != "" { | ||
89 | fmt.Fprintf( | ||
90 | exportFile, | ||
91 | "%d %d %s\n", | ||
92 | dbID, | ||
93 | 0, | ||
94 | g.Image) | ||
95 | } | ||
96 | if g.Thumb != "" { | ||
97 | fmt.Fprintf( | ||
98 | exportFile, | ||
99 | "%d %d %s\n", | ||
100 | dbID, | ||
101 | 1, | ||
102 | g.Thumb) | ||
103 | } | ||
104 | } | ||
105 | fmt.Printf("Inserted %s\n", g.Gril.RomajiName) | ||
106 | return dbID | ||
107 | } | ||
108 | |||
109 | func InsertTags(dbID int, g WrappedGril, taglist map[string]int) { | ||
110 | for _, v := range g.Gril.Tags { | ||
111 | if _, ok := taglist[v]; ok { | ||
112 | continue | ||
113 | } | ||
114 | var id int | ||
115 | row := db.QueryRow("INSERT INTO grilist.tags (name) VALUES ($1) RETURNING id;", v) | ||
116 | row.Scan(&id) | ||
117 | taglist[v] = id | ||
118 | fmt.Printf("Inserted tag %s as %d\n", v, id) | ||
119 | } | ||
120 | rows, err := db.Query("SELECT tag_id FROM grils_tags WHERE gril_id = $1;", dbID) | ||
121 | LogErr(err) | ||
122 | existingTags := make(map[int]bool, 0) | ||
123 | for rows.Next() { | ||
124 | var tagID int | ||
125 | rows.Scan(&tagID) | ||
126 | existingTags[tagID] = true | ||
127 | } | ||
128 | for _, v := range g.Gril.Tags { | ||
129 | tagID := taglist[v] | ||
130 | if _, ok := existingTags[tagID]; ok { | ||
131 | continue | ||
132 | } | ||
133 | _, err := db.Exec(`INSERT INTO grilist.grils_tags (gril_id, tag_id) VALUES ($1, $2);`, dbID, tagID) | ||
134 | fmt.Printf("Assigned tag %s to %s\n", v, g.Gril.RomajiName) | ||
135 | LogErr(err) | ||
136 | } | ||
137 | } | ||
138 | |||
41 | func main() { | 139 | func main() { |
42 | var path string | ||
43 | var sourceType string | ||
44 | var imgUrlExportFile string | ||
45 | flag.StringVar(&path, "path", "", "path of the source files") | 140 | flag.StringVar(&path, "path", "", "path of the source files") |
46 | flag.StringVar(&sourceType, "type", "", "type of the files (ACD, AniDB)") | 141 | flag.StringVar(&sourceType, "type", "", "type of the files (ACD, AniDB)") |
47 | flag.StringVar(&imgUrlExportFile, "imageexportfile", "", "image URLs will be exported to this file, THIS WILL OVERWRITE EVERYTHING IN THE FILE!") | 142 | flag.StringVar(&imgUrlExportFile, "imageexportfile", "", "image URLs will be exported to this file, THIS WILL OVERWRITE EVERYTHING IN THE FILE!") |
48 | flag.Parse() | 143 | flag.Parse() |
49 | var exportFile *os.File | ||
50 | if imgUrlExportFile != "" { | 144 | if imgUrlExportFile != "" { |
51 | var err error | 145 | var err error |
52 | if exportFile, err = os.Create(imgUrlExportFile); err != nil { | 146 | if exportFile, err = os.Create(imgUrlExportFile); err != nil { |
@@ -60,14 +154,15 @@ func main() { | |||
60 | flag.Usage() | 154 | flag.Usage() |
61 | os.Exit(1) | 155 | os.Exit(1) |
62 | } | 156 | } |
63 | r := GetGrilReader(sourceType) | 157 | r = GetGrilReader(sourceType) |
64 | if r == nil { | 158 | if r == nil { |
65 | fmt.Fprintf(os.Stderr, "%s\n", "\"type\" must be specified and a valid gril source!") | 159 | fmt.Fprintf(os.Stderr, "%s\n", "\"type\" must be specified and a valid gril source!") |
66 | flag.Usage() | 160 | flag.Usage() |
67 | os.Exit(1) | 161 | os.Exit(1) |
68 | } | 162 | } |
69 | config := grilist.LoadConfig() | 163 | config := grilist.LoadConfig() |
70 | db, err := sql.Open("postgres", config.DBConnectionString()) | 164 | var err error |
165 | db, err = sql.Open("postgres", config.DBConnectionString()) | ||
71 | if err == nil { | 166 | if err == nil { |
72 | err = db.Ping() | 167 | err = db.Ping() |
73 | } | 168 | } |
@@ -76,6 +171,7 @@ func main() { | |||
76 | fmt.Fprintf(os.Stderr, "%v\n", err) | 171 | fmt.Fprintf(os.Stderr, "%v\n", err) |
77 | os.Exit(1) | 172 | os.Exit(1) |
78 | } | 173 | } |
174 | |||
79 | taglist := make(map[string]int) | 175 | taglist := make(map[string]int) |
80 | rows, _ := db.Query("SELECT id, name FROM grilist.tags;") | 176 | rows, _ := db.Query("SELECT id, name FROM grilist.tags;") |
81 | for rows.Next() { | 177 | for rows.Next() { |
@@ -92,82 +188,19 @@ func main() { | |||
92 | rows.Scan(&sourceID, &grilID) | 188 | rows.Scan(&sourceID, &grilID) |
93 | idmap[sourceID] = grilID | 189 | idmap[sourceID] = grilID |
94 | } | 190 | } |
191 | |||
95 | filepath.Walk(path, func(path string, info os.FileInfo, err error) error { | 192 | filepath.Walk(path, func(path string, info os.FileInfo, err error) error { |
96 | if info.IsDir() { | 193 | if info.IsDir() { |
97 | return nil | 194 | return nil |
98 | } | 195 | } |
99 | g := r.Read(path) | 196 | g := r.Read(path) |
197 | |||
100 | var dbID int | 198 | var dbID int |
101 | var ok bool | 199 | if dbID = TryGetExistingGril(g, idmap); dbID == 0 { |
102 | if dbID, ok = idmap[g.Gril.ID]; !ok { | 200 | dbID = InsertGril(g) |
103 | row := db.QueryRow("INSERT INTO grilist.grils (age) VALUES (NULL) RETURNING id;") | ||
104 | row.Scan(&dbID) | ||
105 | fmt.Printf( | ||
106 | "Assigned %d from %s to %d\n", | ||
107 | g.Gril.ID, | ||
108 | sourceType, | ||
109 | dbID) | ||
110 | db.Exec(`INSERT INTO grilist.grils_id_mappings (gril_id, | ||
111 | source, source_id) VALUES ($1, $2, $3);`, | ||
112 | dbID, r.ID(), g.Gril.ID) | ||
113 | if g.Gril.KanjiName != "" { | ||
114 | _, err := db.Exec(`INSERT INTO | ||
115 | grilist.gril_names (gril_id, name, name_type) | ||
116 | VALUES ($1, $2, $3);`, dbID, g.Gril.KanjiName, | ||
117 | 0) | ||
118 | LogErr(err) | ||
119 | } | ||
120 | if g.Gril.RomajiName != "" { | ||
121 | _, err := db.Exec(`INSERT INTO | ||
122 | grilist.gril_names (gril_id, name, name_type) | ||
123 | VALUES ($1, $2, $3);`, dbID, g.Gril.RomajiName, | ||
124 | 1) | ||
125 | LogErr(err) | ||
126 | } | ||
127 | for _, v := range g.Gril.OtherNames { | ||
128 | _, err := db.Exec(`INSERT INTO | ||
129 | grilist.gril_names (gril_id, name, name_type) | ||
130 | VALUES ($1, $2, $3);`, dbID, v, 2) | ||
131 | LogErr(err) | ||
132 | } | ||
133 | if exportFile != nil { | ||
134 | if g.Image != "" { | ||
135 | fmt.Fprintf( | ||
136 | exportFile, | ||
137 | "%d %d %s\n", | ||
138 | dbID, | ||
139 | 0, | ||
140 | g.Image) | ||
141 | } | ||
142 | if g.Thumb != "" { | ||
143 | fmt.Fprintf( | ||
144 | exportFile, | ||
145 | "%d %d %s\n", | ||
146 | dbID, | ||
147 | 1, | ||
148 | g.Thumb) | ||
149 | } | ||
150 | } | ||
151 | fmt.Printf("Inserted %s\n", g.Gril.RomajiName) | ||
152 | } | ||
153 | for _, v := range g.Gril.Tags { | ||
154 | if _, ok := taglist[v]; ok { | ||
155 | continue | ||
156 | } | ||
157 | var id int | ||
158 | row := db.QueryRow("INSERT INTO grilist.tags (name) VALUES ($1) RETURNING id;", v) | ||
159 | row.Scan(&id) | ||
160 | taglist[v] = id | ||
161 | } | ||
162 | _, err = db.Exec(`DELETE FROM grilist.grils_tags WHERE gril_id | ||
163 | = $1`, dbID) | ||
164 | LogErr(err) | ||
165 | for _, v := range g.Gril.Tags { | ||
166 | tagID := taglist[v] | ||
167 | _, err := db.Exec(`INSERT INTO grilist.grils_tags | ||
168 | (gril_id, tag_id) VALUES ($1, $2);`, dbID, tagID) | ||
169 | LogErr(err) | ||
170 | } | 201 | } |
202 | InsertTags(dbID, g, taglist) | ||
203 | |||
171 | return nil | 204 | return nil |
172 | }) | 205 | }) |
173 | } | 206 | } |