-
Notifications
You must be signed in to change notification settings - Fork 7
/
decomposition.clj
264 lines (238 loc) · 10.6 KB
/
decomposition.clj
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
(ns com.verybigthings.penkala.decomposition
(:require [clojure.spec.alpha :as s]
[com.verybigthings.penkala.util :refer [as-vec path-prefix-join col->alias]]
[camel-snake-kebab.core :refer [->kebab-case-string ->kebab-case-keyword]]
[clojure.set :as set]
[clojure.string :as str]))
(defrecord DecompositionSchema [pk decompose-to namespace columns])
(s/def ::decompose-to #{:indexed-by-pk :coll :map :parent :omit})
(s/def ::schema-map
(s/map-of
keyword?
(s/or
:keyword keyword?
:schema ::schema)))
(s/def ::schema-vec
(s/coll-of
(s/or
:keyword keyword?
:schema-map ::schema-map)
:kind vector?))
(s/def ::schema
(s/or
:schema-map ::schema-map
:schema-vec ::schema-vec))
(s/def ::pk
(s/or
:single keyword?
:multiple (s/coll-of keyword?)))
(s/def ::schema
(s/keys
:req-un [::pk ::schema]
:opt-un [::decompose-to]))
(declare process-schema)
(defn process-schema-columns [schema]
(let [columns (:schema schema)
ns-name (when-let [ns (:namespace schema)] (name ns))
rename (if ns-name (fn [k] (keyword ns-name (name k))) identity)
process-map-columns (fn [schema columns]
(reduce-kv
(fn [acc k v]
(if (map? v)
(assoc-in acc [:schemas (rename k)] (process-schema v))
(assoc-in acc [:renames (rename k)] v)))
schema
columns))]
(if (map? columns)
(process-map-columns schema columns)
(reduce
(fn [acc v]
(if (map? v)
(process-map-columns acc v)
(assoc-in acc [:renames v] (rename v))))
schema
columns))))
(defn process-schema- [schema]
(-> schema
process-schema-columns
(update :pk as-vec)))
(def process-schema (memoize process-schema-))
(defn assoc-columns [acc renames row]
(reduce-kv
(fn [acc' k v]
(if (contains? row v)
(assoc acc' k (get row v))
acc'))
acc
renames))
(declare build)
(defn assoc-descendants [acc schemas idx row]
(reduce-kv
(fn [m k v]
(let [descendant (build (get acc k) v idx row)]
(if descendant
(assoc m k descendant)
m)))
acc
schemas))
(defn build [acc schema idx row]
(let [pk (:pk schema)
is-composite-pk (< 1 (count pk))
id (if is-composite-pk
(mapv #(get row %) pk)
(get row (first pk)))
{:keys [renames schemas]} schema]
(if (or (and is-composite-pk (every? nil? id))
(and (not is-composite-pk) (nil? id)))
acc
(let [current (-> (get acc id {})
(vary-meta update ::idx #(or % idx))
(assoc-columns renames row)
(assoc-descendants schemas idx row))]
(assoc acc id current)))))
(defn transform [schema mapping]
(let [decompose-to (get schema :decompose-to :coll)
processor (get schema :processor identity)
schemas (:schemas schema)
transformed (reduce-kv
(fn [acc k row]
(let [transformed
(reduce-kv
(fn [row' k k-schema]
(let [transformed-child (transform k-schema (get row k))]
(if (= :parent (:decompose-to k-schema))
(-> row'
(dissoc k)
(merge transformed-child))
(assoc row' k transformed-child))))
row
schemas)
transformed' (processor transformed)]
(cond
(= :coll decompose-to)
(conj acc transformed')
(= :indexed-by-pk decompose-to)
(assoc acc k transformed')
;; decompose to :map and :parent should just return the map
:else
transformed')))
(if (= :coll decompose-to) [] {})
mapping)]
(if (= :coll decompose-to)
(->> transformed
(sort-by #(-> % meta ::idx))
vec)
transformed)))
(defn decompose
"Decomposes the data based on the schema."
[schema data]
(when (and (seq schema) (seq data))
(let [schema' (process-schema schema)
mapping (reduce
(fn [acc [idx row]]
(build acc schema' idx row))
{}
(map-indexed (fn [idx v] [idx v]) (as-vec data)))]
(transform schema' mapping))))
(s/fdef decompose
:args (s/cat ::schema (s/or :map map? :coll sequential?))
:ret (s/or :map? map? :coll sequential?))
(defn get-prefixed-col-name [path-prefix col-name]
(->> (conj path-prefix (col->alias col-name))
(mapv name)
path-prefix-join
keyword))
(defn get-rel-pk [rel]
(let [pk-ids (:pk rel)
pk-aliases (mapv #(get-in rel [:ids->aliases %]) pk-ids)
projection (:projection rel)]
(if (and (seq pk-aliases) (set/subset? (set pk-aliases) projection))
pk-aliases
(vec (sort projection)))))
(defn infer-flat-schema
([relation overrides path-prefix]
(let [columns (infer-flat-schema relation overrides path-prefix {})
processor (get overrides :processor identity)]
(map->DecompositionSchema
{:pk (-> (vals columns) sort vec)
:namespace false
:schema columns
:decompose-to :coll
:processor processor})))
([relation overrides path-prefix schema]
(let [with-columns (reduce
(fn [acc col-name]
(let [col-ns (when (seq path-prefix) (str/join "." (map name path-prefix)))
col-name' (if col-ns (keyword col-ns (name col-name)) col-name)]
(assoc acc col-name' (get-prefixed-col-name path-prefix col-name))))
schema
(:projection relation))]
(reduce-kv
(fn [acc alias join]
(infer-flat-schema (:relation join) overrides (conj path-prefix alias) acc))
with-columns
(:joins relation)))))
(defn infer-schema
"Infers the decomposition schema from a relation."
([relation] (infer-schema relation nil []))
([relation overrides] (infer-schema relation overrides []))
([relation overrides path-prefix]
(cond
(false? overrides)
(infer-flat-schema relation overrides path-prefix)
(= DecompositionSchema (type overrides))
overrides
:else
(let [pk (->> (or (:pk overrides) (get-rel-pk relation))
as-vec
(mapv #(get-prefixed-col-name path-prefix %)))
default-namespace (:namespace overrides)
namespace (if (nil? default-namespace)
(->kebab-case-string (get-in relation [:spec :name]))
default-namespace)
decompose-to (get overrides :decompose-to :coll)
processor (get overrides :processor identity)
columns (reduce
(fn [acc col-name]
(assoc acc col-name (get-prefixed-col-name path-prefix col-name)))
{}
(:projection relation))
columns-with-joined (reduce-kv
(fn [acc alias join]
(let [join-projection (:projection join)
join-relation (update (:relation join) :projection #(or join-projection %))
join-overrides (get-in overrides [:schema alias])
join-path (conj path-prefix alias)
join-type (:type join)]
(if (-> join-relation :projection seq)
;; If we encounter a join that might have nils on the left side
;; we switch to the flat decomposition, where we just namespace all columns
;; to avoid duplicates. The namespacing behavior is different than the default
;; because it's using join alias as a namespace instead of the relation name
;; to ensure that joining a same relation multiple times is not overriding values
(let [join-schema (if (contains? #{:left :left-lateral :inner :inner-lateral} join-type)
(infer-schema join-relation join-overrides join-path)
(infer-flat-schema join-relation join-overrides join-path))]
;; First we build join schema and then check if this level should be omitted.
;; If it should, we still need to pick up any joins on the levels below it.
;; This is not super efficient as we'll iterate through some columns multiple
;; times, but works until a better way is figured out.
(if (= :omit (:decompose-to join-schema))
(let [join-relation-projection (:projection join-relation)]
(reduce-kv
(fn [acc' col col-schema]
(if (contains? join-relation-projection col)
acc'
(assoc acc col col-schema)))
acc
(:schema join-schema)))
(assoc acc alias join-schema)))
acc)))
columns
(:joins relation))]
(map->DecompositionSchema
{:pk pk
:decompose-to decompose-to
:namespace namespace
:schema columns-with-joined
:processor processor})))))