devops/postgres/schema.sql

   1 CREATE EXTENSION IF NOT EXISTS plpgsql WITH SCHEMA pg_catalog;
   2 COMMENT ON EXTENSION plpgsql IS 'PL/pgSQL procedural language';
   3
   4 CREATE EXTENSION IF NOT EXISTS tsm_system_rows;
   5 CREATE EXTENSION pgcrypto;
   6
   7 -----------------------------------------------------------------
   8 CREATE TABLE public.auth_user (
   9     id SERIAL,
  10     password     CHARACTER varying(128) NOT NULL,
  11     last_login   TIMESTAMP with time zone,
  12     is_superuser BOOLEAN NOT NULL,
  13     username     CHARACTER varying(150) NOT NULL,
  14     first_name   CHARACTER varying(30) NOT NULL,
  15     last_name    CHARACTER varying(30) NOT NULL,
  16     email        CHARACTER varying(254) NOT NULL,
  17     is_staff     BOOLEAN NOT NULL,
  18     is_active    BOOLEAN NOT NULL,
  19     date_joined  TIMESTAMP with time zone DEFAULT now() NOT NULL,
  20     PRIMARY KEY (id)
  21 );
  22 ALTER TABLE public.auth_user OWNER TO gargantua;
  23 -----------------------------------------------------------------
  24
  25 -- TODO add publication_date
  26 -- TODO typename -> type_id
  27 CREATE TABLE public.nodes (
  28     id        SERIAL,
  29     hash_id   CHARACTER varying(66) DEFAULT ''::character varying NOT NULL,
  30     typename  INTEGER NOT NULL,
  31     user_id   INTEGER NOT NULL,
  32     parent_id INTEGER REFERENCES public.nodes(id) ON DELETE CASCADE ,
  33     name      CHARACTER varying(255) DEFAULT ''::character varying NOT NULL,
  34     date      TIMESTAMP with time zone DEFAULT now() NOT NULL,
  35     hyperdata jsonb DEFAULT '{}'::jsonb NOT NULL,
  36     search tsvector,
  37     PRIMARY KEY (id),
  38     FOREIGN KEY (user_id)  REFERENCES public.auth_user(id) ON DELETE CASCADE
  39 );
  40 ALTER TABLE public.nodes OWNER TO gargantua;
  41 --------------------------------------------------------------
  42
  43 -- TODO add publication_date
  44 -- TODO typename -> type_id
  45 CREATE TABLE public.contexts (
  46     id        SERIAL,
  47     hash_id   CHARACTER varying(66) DEFAULT ''::character varying NOT NULL,
  48     typename  INTEGER NOT NULL,
  49     user_id   INTEGER NOT NULL,
  50     parent_id INTEGER REFERENCES public.contexts(id) ON DELETE CASCADE ,
  51     name      CHARACTER varying(255) DEFAULT ''::character varying NOT NULL,
  52     date      TIMESTAMP with time zone DEFAULT now() NOT NULL,
  53     hyperdata jsonb DEFAULT '{}'::jsonb NOT NULL,
  54     search tsvector,
  55     PRIMARY KEY (id),
  56     FOREIGN KEY (user_id)  REFERENCES public.auth_user(id) ON DELETE CASCADE
  57 );
  58 ALTER TABLE public.contexts OWNER TO gargantua;
  59
  60 --------------------------------------------------------------
  61 -- | Ngrams
  62 CREATE TABLE public.ngrams (
  63     id SERIAL,
  64     terms CHARACTER varying(255),
  65     n INTEGER,
  66     PRIMARY KEY (id)
  67 );
  68 ALTER TABLE public.ngrams OWNER TO gargantua;
  69
  70 -- | Ngrams PosTag
  71 CREATE TABLE public.ngrams_postag (
  72     id        SERIAL                  ,
  73     lang_id   INTEGER                 ,
  74     algo_id   INTEGER                 ,
  75     postag    CHARACTER varying(5)    ,
  76     ngrams_id INTEGER NOT NULL        ,
  77     lemm_id   INTEGER NOT NULL        ,
  78     score     INTEGER DEFAULT 1 ::integer NOT NULL                        ,
  79     FOREIGN KEY (ngrams_id) REFERENCES public.ngrams(id) ON DELETE CASCADE,
  80     FOREIGN KEY (lemm_id)   REFERENCES public.ngrams(id) ON DELETE CASCADE
  81 );
  82 ALTER TABLE public.ngrams_postag OWNER TO gargantua;
  83
  84 --------------------------------------------------------------
  85 -- Node here should have type NodeList
  86 CREATE TABLE public.node_ngrams (
  87     id SERIAL                          ,
  88     node_id        INTEGER NOT NULL    ,
  89     node_subtype   INTEGER             ,
  90     ngrams_id      INTEGER NOT NULL    ,
  91     ngrams_type    INTEGER             , -- change to ngrams_field? (no for pedagogic reason)
  92     ngrams_field   INTEGER             ,
  93     ngrams_tag     INTEGER             ,
  94     ngrams_class   INTEGER             ,
  95     weight double precision            ,
  96     PRIMARY KEY (id)                   ,
  97     FOREIGN KEY (node_id)   REFERENCES public.nodes(id)  ON DELETE CASCADE ,
  98     FOREIGN KEY (ngrams_id) REFERENCES public.ngrams(id) ON DELETE CASCADE
  99 );
 100 ALTER TABLE public.node_ngrams OWNER TO gargantua;
 101
 102 --CREATE TABLE public.context_nodengrams_nodengrams (
 103 --    context_id         INTEGER NOT NULL   ,
 104 --    node_ngrams1_id INTEGER NOT NULL   ,
 105 --    node_ngrams2_id INTEGER NOT NULL   ,
 106 --    weight double   precision          ,
 107 --    FOREIGN KEY (node_id) REFERENCES public.contexts(id) ON DELETE CASCADE              ,
 108 --    FOREIGN KEY (node_ngrams1_id) REFERENCES public.node_ngrams(id) ON DELETE CASCADE,
 109 --    FOREIGN KEY (node_ngrams2_id) REFERENCES public.node_ngrams(id) ON DELETE CASCADE,
 110 --    PRIMARY KEY (node_id, node_ngrams1_id, node_ngrams2_id)
 111 --);
 112 --ALTER TABLE public.context_nodengrams_nodengrams OWNER TO gargantua;
 113
 114 --------------------------------------------------------------
 115 --------------------------------------------------------------
 116 --
 117 --CREATE TABLE public.nodes_ngrams_ngrams (
 118 --    node_id   integer NOT NULL REFERENCES public.nodes(id)  ON DELETE CASCADE,
 119 --    ngram1_id integer NOT NULL REFERENCES public.ngrams(id) ON DELETE CASCADE,
 120 --    ngram2_id integer NOT NULL REFERENCES public.ngrams(id) ON DELETE CASCADE,
 121 --    weight double precision,
 122 --    PRIMARY KEY (node_id,ngram1_id,ngram2_id)
 123 --);
 124 --
 125 --ALTER TABLE public.nodes_ngrams_ngrams OWNER TO gargantua;
 126 ---------------------------------------------------------------
 127 -- TODO nodes_nodes(node1_id int, node2_id int, edge_type int , weight real)
 128 CREATE TABLE public.nodes_nodes (
 129     node1_id INTEGER NOT NULL REFERENCES public.nodes(id) ON DELETE CASCADE,
 130     node2_id INTEGER NOT NULL REFERENCES public.nodes(id) ON DELETE CASCADE,
 131     score    REAL    ,
 132     category INTEGER ,
 133     PRIMARY KEY (node1_id, node2_id)
 134 );
 135 ALTER TABLE public.nodes_nodes OWNER TO gargantua;
 136
 137
 138 -- To attach contexts to a Corpus
 139 CREATE TABLE public.nodes_contexts (
 140     node_id    INTEGER NOT NULL REFERENCES public.nodes(id)    ON DELETE CASCADE,
 141     context_id INTEGER NOT NULL REFERENCES public.contexts(id) ON DELETE CASCADE,
 142     score    REAL    ,
 143     category INTEGER ,
 144     PRIMARY KEY (node_id, context_id)
 145 );
 146 ALTER TABLE public.nodes_contexts OWNER TO gargantua;
 147
 148 ---------------------------------------------------------------
 149 CREATE TABLE public.context_node_ngrams (
 150     context_id    INTEGER NOT NULL REFERENCES public.contexts (id) ON DELETE CASCADE,
 151     node_id       INTEGER NOT NULL REFERENCES public.nodes    (id) ON DELETE CASCADE,
 152     ngrams_id     INTEGER NOT NULL REFERENCES public.ngrams   (id) ON DELETE CASCADE,
 153     ngrams_type   INTEGER  ,
 154     weight double precision,
 155     PRIMARY KEY (context_id, node_id, ngrams_id, ngrams_type)
 156   );
 157
 158
 159 ALTER TABLE public.context_node_ngrams OWNER TO gargantua;
 160
 161 CREATE TABLE public.context_node_ngrams2 (
 162     context_id      INTEGER NOT NULL REFERENCES public.contexts  (id)       ON DELETE CASCADE,
 163     nodengrams_id   INTEGER NOT NULL REFERENCES public.node_ngrams  (id) ON DELETE CASCADE,
 164     weight double   precision,
 165     PRIMARY KEY (context_id, nodengrams_id)
 166 );
 167 ALTER TABLE public.context_node_ngrams2 OWNER TO gargantua;
 168
 169
 170 --------------------------------------------------------------------
 171 CREATE TABLE public.node_node_ngrams (
 172 node1_id   INTEGER NOT NULL REFERENCES public.nodes  (id) ON DELETE CASCADE,
 173 node2_id   INTEGER NOT NULL REFERENCES public.nodes  (id) ON DELETE CASCADE,
 174 ngrams_id  INTEGER NOT NULL REFERENCES public.ngrams (id) ON DELETE CASCADE,
 175 ngrams_type INTEGER,
 176 weight double precision,
 177 PRIMARY KEY (node1_id, node2_id, ngrams_id, ngrams_type)
 178 );
 179 ALTER TABLE public.node_node_ngrams OWNER TO gargantua;
 180
 181 CREATE TABLE public.node_node_ngrams2 (
 182 node_id         INTEGER NOT NULL REFERENCES public.nodes  (id) ON DELETE CASCADE,
 183 nodengrams_id   INTEGER NOT NULL REFERENCES public.node_ngrams  (id) ON DELETE CASCADE,
 184 weight double precision,
 185 PRIMARY KEY (node_id, nodengrams_id)
 186 );
 187 ALTER TABLE public.node_node_ngrams2 OWNER TO gargantua;
 188
 189
 190
 191
 192 --------------------------------------------------------------
 193
 194 --CREATE TABLE public.nodes_ngrams_repo (
 195 --    version integer NOT NULL,
 196 --    patches jsonb DEFAULT '{}'::jsonb NOT NULL,
 197 --    PRIMARY KEY (version)
 198 --);
 199 --ALTER TABLE public.nodes_ngrams_repo OWNER TO gargantua;
 200
 201 ---------------------------------------------------------
 202
 203 -- If needed for rights management at row level
 204 -- CREATE EXTENSION IF NOT EXISTS acl WITH SCHEMA public;
 205 CREATE TABLE public.rights (
 206   user_id INTEGER NOT NULL REFERENCES public.auth_user(id) ON DELETE CASCADE,
 207   node_id INTEGER NOT NULL REFERENCES public.nodes(id)     ON DELETE CASCADE,
 208   rights  INTEGER NOT NULL,
 209   PRIMARY KEY (user_id, node_id)
 210 );
 211 ALTER TABLE public.rights OWNER TO gargantua;
 212
 213 ------------------------------------------------------------
 214 ------------------------------------------------------------
 215 -- INDEXES
 216 CREATE INDEX        ON public.auth_user USING btree (username varchar_pattern_ops);
 217 CREATE UNIQUE INDEX ON public.auth_user USING btree (username);
 218
 219 CREATE INDEX        ON public.rights USING btree (user_id,node_id);
 220
 221 CREATE INDEX        ON public.nodes USING gin (hyperdata);
 222 CREATE INDEX        ON public.nodes USING btree (user_id, typename, parent_id);
 223 CREATE INDEX        ON public.nodes USING btree (id, typename, date ASC);
 224 CREATE INDEX        ON public.nodes USING btree (id, typename, date DESC);
 225 CREATE INDEX        ON public.nodes USING btree (typename, id);
 226 CREATE UNIQUE INDEX ON public.nodes USING btree (hash_id);
 227
 228 CREATE INDEX        ON public.contexts USING gin (hyperdata);
 229 CREATE INDEX        ON public.contexts USING btree (user_id, typename, parent_id);
 230 CREATE INDEX        ON public.contexts USING btree (id, typename, date ASC);
 231 CREATE INDEX        ON public.contexts USING btree (id, typename, date DESC);
 232 CREATE INDEX        ON public.contexts USING btree (typename, id);
 233 CREATE UNIQUE INDEX ON public.contexts USING btree (hash_id);
 234
 235
 236 -- CREATE UNIQUE INDEX ON public.nodes USING btree (((hyperdata ->> 'uniqId'::text)));
 237 -- CREATE UNIQUE INDEX ON public.nodes USING btree (((hyperdata ->> 'uniqIdBdd'::text)));
 238 -- CREATE UNIQUE INDEX ON public.nodes USING btree (typename, parent_id, ((hyperdata ->> 'uniqId'::text)));
 239
 240 CREATE UNIQUE INDEX ON public.ngrams (terms); -- TEST GIN
 241 CREATE        INDEX ON public.ngrams USING btree (id, terms);
 242 CREATE UNIQUE INDEX ON public.ngrams_postag (lang_id,algo_id,postag,ngrams_id,lemm_id);
 243
 244 -- To save the Node Ngrams Repo
 245 CREATE        INDEX ON public.node_ngrams USING btree (node_id,node_subtype);
 246 CREATE UNIQUE INDEX ON public.node_ngrams USING btree (node_id,node_subtype, ngrams_id);
 247
 248
 249
 250 -- To make the links between Nodes in Tree/Forest
 251 CREATE UNIQUE INDEX ON public.nodes_nodes  USING btree (node1_id, node2_id);
 252 CREATE INDEX        ON public.nodes_nodes  USING btree (node1_id, node2_id, category);
 253
 254 -- To make the links between Corpus Node and its contexts
 255 CREATE UNIQUE INDEX ON public.nodes_contexts  USING btree (node_id, context_id);
 256 CREATE INDEX        ON public.nodes_contexts  USING btree (node_id, context_id, category);
 257
 258
 259 ------------------------------------------------------------------------
 260 CREATE UNIQUE INDEX ON public.context_node_ngrams USING btree (context_id, node_id, ngrams_id, ngrams_type);
 261 CREATE        INDEX ON public.context_node_ngrams USING btree (context_id,  node_id);
 262 CREATE        INDEX ON public.context_node_ngrams USING btree (ngrams_id, node_id);
 263 CREATE        INDEX ON public.context_node_ngrams USING btree (ngrams_type);
 264
 265 CREATE INDEX ON public.context_node_ngrams2 USING btree (context_id);
 266 CREATE INDEX ON public.context_node_ngrams2 USING btree (nodengrams_id);
 267 CREATE INDEX ON public.context_node_ngrams2 USING btree (context_id, nodengrams_id);
 268
 269
 270 CREATE UNIQUE INDEX ON public.node_node_ngrams USING btree (node1_id, node2_id, ngrams_id, ngrams_type);
 271 CREATE        INDEX ON public.node_node_ngrams USING btree (node1_id,  node2_id);
 272 CREATE        INDEX ON public.node_node_ngrams USING btree (ngrams_id, node2_id);
 273 CREATE        INDEX ON public.node_node_ngrams USING btree (ngrams_type);
 274 CREATE INDEX ON public.node_node_ngrams2 USING btree (node_id);
 275 CREATE INDEX ON public.node_node_ngrams2 USING btree (nodengrams_id);
 276 CREATE INDEX ON public.node_node_ngrams2 USING btree (node_id, nodengrams_id);
 277
 278 -- CREATE INDEX ON public.context_nodengrams_nodengrams USING btree (context_id, node_ngrams1_id, node_ngrams2_id);
 279 -- CREATE INDEX ON public.context_nodengrams_nodengrams USING btree (node_ngrams1_id);
 280 -- CREATE INDEX ON public.context_nodengrams_nodengrams USING btree (node_ngrams2_id);
 281
 282
 283 ------------------------------------------------------------------------
 284 -- Ngrams Full DB Extraction Optim
 285 -- TODO remove hard parameter and move elsewhere
 286 CREATE OR REPLACE function node_pos(int, int) returns bigint
 287    AS 'SELECT count(id) from nodes
 288       WHERE  id < $1
 289       AND typename = $2
 290       '
 291    LANGUAGE SQL immutable;
 292
 293 --drop index node_by_pos;
 294 --create index node_by_pos on nodes using btree(node_pos(id,typename));
 295