<Insert Picture Here>

Présentation au sujet: "<Insert Picture Here>"— Transcription de la présentation:

1 <Insert Picture Here>
3 Chargement et Indexation

2 Points couverts Formats des données géo-spatiales
Chargement des données Validation Création des index spatiaux Déplacement des données entre bases Export/import Tablespaces transportables

3 Sources de données géo-spatiales
Données applicatives Clients, magasins, entrepôts, agences, etc. Il faut les géocoder avant de pouvoir s'en servir Référentiel de géocodage et calcul Réseau routier, adresses Souvent fournies par Navteq ou Tele Atlas, … Autres données (vecteur et raster) Limites administratives, occupation du sol, points d'intérêt, … Multiples sources (IGN, etc)

4 Format des données géo-spatiales
Chaque outil SIG utilise son ou ses formats propres ESRI: shape files, coverages MapInfo: TAB files, MIF/MID Intergraph: DGN files Autodesk: DWG files Etc ...

5 Chargement des données Géo-spatiales
Au moyen des outils fournis avec les outils SIG Au moyen d'outils génériques FME (Safe Software) CITRA (CISS TDI) Pour les fichiers SHAPE: convertisseur SHP2SDO Oracle Mapbuilder Au moyen de SQL*Loader Les données doivent être formatées au préalable Charger les données avant de créer l’index spatial. !

6 Utilisation de SQL*Loader
Control file Data files Parameter file (optional) SQL*Loader Rejected Field processing Accepted Discarded Record selection Bad file Selected Oracle Server Rejected Discard file (optional) Inserted Log file Database files

7 SQL*Loader pour des Points
LOAD DATA INTO TABLE cities FIELDS TERMINATED BY '|' ( CITY, STATE_ABRV, POP90, RANK90, LOCATION COLUMN OBJECT ( SDO_GTYPE INTEGER EXTERNAL, SDO_POINT COLUMN OBJECT ( X FLOAT EXTERNAL, Y FLOAT EXTERNAL ) ) New York|NY| |1| 2001| | | Los Angeles|CA| |2| 2001| | | Chicago|IL| |3| 2001| | | Houston|TX| |4| 2001| | | Philadelphia|PA| |5| 2001| | | San Diego|CA| |6| 2001| | |

8 SQL*Loader pour des lignes et polygones
LOAD DATA CONTINUEIF NEXT(1:1) = '#' INTO TABLE counties FIELDS TERMINATED BY '|' ( COUNTY, STATE, POPPSQMI, GEOM COLUMN OBJECT ( SDO_GTYPE INTEGER EXTERNAL, SDO_ELEM_INFO VARRAY TERMINATED BY '/' (E FLOAT EXTERNAL), SDO_ORDINATES VARRAY TERMINATED BY '/' (O FLOAT EXTERNAL) ) ) Autauga|Alabama| | #2003|1|1003|1/ # | | | | |....| # | | | / Baldwin|Alabama| | # | | | | |..../ SQL*Loader Control File for Lines and Polygons (Oracle9i) The above example is very similar to the previous example, but are some differences. The first difference is: CONTINUEIF NEXT(1:1) = '#' This syntax causes SQL*Loader to look at the first character of each physical input record, and if it is # it will continue the logical record from the previous line. One of the requirements when using this syntax is that there should never be data in the first position of a physical record. SQL*Loader will always look at the first character position, and if it is a # it will treat the record as a continuation of the previous record. If there is no # it will treat everything after the first character as start of a new record to be loaded. This example is loading table counties, and the fields in this table are terminated by the vertical bar (|); In this example, the VARRAY fields (SDO_ELEM_INFO and SDO_ORDINATES) within the SDO_GEOMETRY are being loaded (the last example loaded the SDO_POINT field and left the SDO_ELEM_INFO and SDO_ORDINATES as NULL). There is special syntax to end each of the varray types (each value within the varray is terminated with a vertical bar, but the varray itself needs to be terminated also). The syntax: SDO_ELEM_INFO VARRAY TERMINATED BY '|/' (X FLOAT EXTERNAL) is the way to describe to SQL*Loader that the SDO_ELEM_INFO varray ends with '|/'. The same syntax is shown for the SDO_ORDINATES varray.

9 Applications spécifiques
Verbes SQL INSERT standards Utiliser des variables (pas des constantes) Faire des commit réguliers Utiliser des insertions par lots (array insert) pour de meilleures performances Conserver l'état d'avancement du chargement Les insertions sont plus rapides sans index spatial Grouper les insertions en lots (finir par COMMIT) INSERT INTO lines (id, … type, shape) VALUES (:id, … :type, :my_geom);

10 Convertisseur Shapefile : SHP2SDO
Traite un fichier SHAPE à la fois Convertit les attributs (.DBF) aussi bien que les géométries (.SHP et .SHX) Génère des scripts SQL pour la création de la table, la mise à jour des meta-données spatiales, et la création de l'index spatial. Génère les fichiers pour SQL*Loader: control file et data file Shapefile Converter The shapefile converter is an unsupported tool available both as part of this course and as a download from OTN that converts ESRI shapefiles (geometries and attributes) into SQL*Loader control and data files. The shapefile converter automatically generates a SQL script that can be used to create the single table that holds both geometry and attribute data. Additionally, the script also inserts the metadata about the spatial layer into USER_SDO_GEOM_METADATA. The shapefile converter appropriately uses continuation marks to ensure no physical record is greater than 64K characters. It can convert data for loading into Oracle Spatial version and beyond. After the shapefile converter is invoked, the generated SQL script must be executed to create the table and populate metadata in the USER_SDO_GEOM_METADATA view. The script can be edited to add storage parameters to the create table statement. After that, SQL*Loader can be invoked to bulk load data into Oracle. When the load is complete, migrate the spatial data to the latest Oracle Spatial format with the SDO_MIGRATE.TO_CURRENT procedure. This procedure ensures all elements have four-digit SDO_GTYPEs, and four-digit SDO_ETYPEs where appropriate. It also takes care of ring ordering and ensures exterior polygon elements are followed by all of their interior polygon elements. The following slides describe in detail how to use the shapefile converter.

11 Utilisation du convertisseur
shp2sdo <shapefile> <table-name> -g <geometry-col> -i <id-col> -x (<Xmin>,<Xmax>) -y (<Ymin>,<Ymax>) -s <srid> -t <tolerance> <shapefile> = nom du shapefile à convertir (sans inclure le suffixe .SHP) <table-name> = nom de la table en sortie (aussi préfixe des fichiers produits) -g <geometry-col> = nom donné à la colonne géométrique -i <id-col> = nom donné à une colonne numérique servant d'identifiant unique Invoking the SHP2SDO Converter This example shows how to invoke the shapefile to Oracle Spatial converter. The first parameter is the name of the shapefile. The converter will use input from the .shp file as well as the .dbf and .shx files. The second parameter is the name of the table to create, as well as the prefix for the output files generated by the shapefile converter. If not specified, the table name is the same as the shapefile. The -g <geometry-col> parameter specifies the geometry column name in the CREATE TABLE statement generated by the shapefile converter. This column will be of type MDSYS.SDO_GEOMETRY. -i <id-col> is an optional parameter that creates a numeric column that can be used to store an id field. Oracle Spatial has no requirement for a numeric key column, but many tools do require a numeric primary key. -n [<start-id>] is a parameter used to populate the numeric key field created by the -i qualifier. If a start-id is provided, the numbers are populated starting at this value. Otherwise, the numbers start at 0. The -d parameter is used to specify that the data should be included directly into the control file. Otherwise, a separate data file is generated.

12 Utilisation du convertisseur
-x (Xmin, Xmax) = bornes de la dimension X (si omis, alors on utilise les bornes réelles du shape file) -y (Ymin, Ymax) = bornes de la dimension Y (si omis, alors on utilise les bornes réelles du shape file) -s <srid> = numéro du système de coordonnées -t <tolerance> = valeur de tolérance Invoking the SHP2SDO Converter (continued) The -x (Xmin, Xmax) is used to tell Oracle the bounds (or extent) of the first dimension of the coordinate system. If not specified, the bounds are extracted from the shapefile. Oracle recommends using this parameter to specify the largest bounds of the coordinate system. For data stored as Longitude/Latitude, the values of Xmin and Xmax should be set to the smallest and largest Longitude values, that is, -180, 180. The -y (Ymin, Ymax) is used to tell Oracle the bounds (or extent) of the second dimension of the coordinate system used. If not specified, the bounds are extracted from the shapefile. Oracle recommends using this parameter to specify the largest bounds of the coordinate system. For data stored as Longitude/Latitude, the values should be set to the smallest and largest Latitude values, that is, -90, 90. The -s <srid> species the coordinate system (spatial reference system) associated with the data. By default, no coordinate system information is associated (NULL for SDO_SRID in the SDO_GEOMETRY column and NULL for SRID in USER_SDO_GEOM_METADATA).

13 Exemple d'utilisation Entrées Sorties
shp2sdo cities -g location -x (-180,180) –y (-90,90) –t 0.5 –s 8307 Entrées Fichiers CITIES.SHP, CITIES.SHX, CITIES.DBF Sorties CITIES.SQL: crée la table “CITIES” et remplit les méta-données pour la table. CITIES_SX.SQL : crée l'index spatial CITIES.CTL: fichier de contrôle pour SQL*Loader CITIES.DAT: fichier de données Example SHP2SDO Converter Run On the command line, the first parameter (“cities”) is the name of the shapefile, so cities.shp, cities.shx, and cities.dbf will be processed from the current directory. Full path names to the shapefile can be specified. Because <table-name> is not specified as the second parameter, the name of the table to be created will also be “cities”. The name of the column in table cities of type MDSYS.SDO_GEOMETRY will be “location”, as specified by the -g qualifier. The output files of the command are CITIES.SQL, CITIES.CTL, and CITIES.DAT. CITIES.SQL includes the command to create the cities table and the insert statement to populate the USER_SDO_GEOM_METADATA view. The -x (-180,180) -y (-90,90) populates the DIMINFO field of USER_SDO_GEOM_METADATA with coordinate system bounds of -180,180 and -90,90. CITIES.CTL includes the information for SQL*Loader to load the cities table. CITIES.DAT contains the data to be loaded in the cities table. After invoking the converter, execute cities.sql from SQL*Plus to create the table, and populate the USER_SDO_GEOM_METADATA view. Next run SQL*Loader to load the table. Then migrate the data loaded to the latest Oracle Spatial format, and create a spatial index on the SDO_GEOMETRY column. The migration of the data will be covered later in this chapter, and information about spatial indexing will follow in a later chapter.

14 Exemple d'utilisation shp2sdo - Shapefile(r) To Oracle Spatial Converter Version Apr-2004 Copyright 2003, 2004 Oracle Corporation Processing shapefile cities into spatial table CITIES Oracle version is 9.2 Geometry column is LOCATION Points stored in SDO_POINT attributes SRID used is 8307 Data is in a separate file No loading mode specified. Will use SQL*Loader default of INSERT Bounds set to X=[-180,180] Y=[-90,90] Tolerance: Conversion complete : 195 points processed The following files have been created: cities.sql : SQL script to create the table cities_sx.sql : SQL script to create the spatial index cities.ctl : Control file for loading the table cities.dat : Data file Log Output from shp2sdo Above is the log output from running the shapefile to Oracle Spatial conversion utility. The output shows input is coming from the cities shapefile, the output table will be called CITIES, and data is created in the Oracle Spatial object-relational format. Additionally, the SDO_GEOMETRY column in table CITIES is called LOCATION, and point data is stored in the SDO_POINT field. The data will be stored in a separate file. The extent of the coordinate system will be set to -180,180 and -90,90 in the DIMINFO field of the USER_SDO_GEOM_METADATA entry associated with the LOCATION column in the CITIES table. The output also shows that 195 cities were converted, and the output files are cities.sql, cities.ctl, and cities.dat.

15 Mapbuilder

16 Validation des données
Les données peuvent contenir des erreurs Polygones « self-crossing » Points redondants Erreur d'orientation (polygones) Pas de vérification automatique ! Certains outils acceptent certaines erreurs ... Certains demandent des données sans erreurs. Utiliser les fonctions de validation fournies

17 Validation des données
Fonction SDO_GEOM.VALIDATE_GEOMETRY Détermine la validité d'un objet isolé Procédure SDO_GEOM.VALIDATE_LAYER Détermine la validité de tous les objets d'une table Chaque fonction retourne un diagnostic précis (code erreur ORA-13xxx) Les variantes VALIDATE_GEOMETRY_WITH_CONTEXT et VALIDATE_LAYER_WITH_CONTEXT indiquent en plus à quel endroit se trouve l'erreur dans l'objet. Validating Geometries Oracle Spatial provides validation routines that provide detailed information about invalid geometries. The routines are called: SDO_GEOM.VALIDATE_GEOMETRY_WITH_CONTEXT SDO_GEOM.VALIDATE_LAYER_WITH_CONTEXT These routines replace: SDO_GEOM.VALIDATE_GEOMETRY SDO_GEOM.VALIDATE_LAYER The validation routines now provide information regarding: The specific coordinates where the geometry is invalid The specific edge where the problem occurred The specific ring where the problem occurred This new information will make it easier to debug invalid geometries.

18 Exemple de validation automatique
create or replace trigger counties_geom before insert or update of geom on counties for each row declare status varchar2(10); begin if :new.geom is not null then status := sdo_geom.validate_geometry (:new.geom, 0.5); if status <> 'TRUE' then raise_application_error (-20000, sqlerrm (-status)); end if; end; /

19 Radius Check

20 Indexation Spatiale ! Indexation R-tree Indexation Quad-tree
Arborescence de rectangles (MBR) Indexation en 2 ou 3 dimensions Indexation Quad-tree Utilisation d'une grille régulière Plus documentée en 10g Ne plus utiliser Sauf exceptions …. Un index spatial doit exister avant de pouvoir poser des requêtes spatiales sur une table! root R S a b c d Spatial Indexing Oracle Spatial uses a spatial index to optimize spatial query performance. The same way an index is created on scalar data to improve query performance, a spatial index improves spatial query performance. Oracle Spatial uses R-tree indexes, which generate and store a rough approximation of each geometry when you create a spatial index. This approximation is generated for all types of geometries (for example, points, line strings, and polygons). !

21 Quad-tree ou R-tree ? R-Tree est le choix par défaut et préféré
Le seul choix pour les données en Longitude/Latitude Les Quad-tree peuvent être utiles dans des cas particuliers

22 Concept du R-tree A D R-tree root root R B R S C A B C D S
MBRs, pointers to geometries

23 Création d'un index R-tree
create index CUSTOMERS_SIDX on CUSTOMERS (LOCATION) indextype is MDSYS.SPATIAL_INDEX; Index Information R-tree Index Structures When an R-tree spatial index is built, an Oracle table is created. This is also known as the spatial index table. The spatial index table is visible to the end user, and is flagged in Oracle’s metadata tables (USER_TABLES, ALL_TABLES, DBA_TABLES) as SECONDARY = ‘Y’. The spatial index name will always appear to be in the SYSTEM tablespace (if it is looked up in USER_INDEXES). This is because the spatial index name is only a logical structure associated with another physical structure (the spatial index table). The tablespace and storage parameters can be specified for the spatial index table associated with the spatial index name. This will be demonstrated later when the CREATE INDEX command is discussed. Table MDRT_7B50$

24 Syntaxe pour la création d'index
CREATE INDEX <index-name> ON <table-name> (<column-name>) INDEXTYPE IS MDSYS.SPATIAL_INDEX [PARAMETERS ( 'SDO_RTR_PCTFREE = <param_value> <storage_parameters> = <param_value> ... ‘) ] [PARALLEL [<parallel_degree>]]; Créer un index spatial est plus long que créer un index "normal" Dépend du nombre et de la complexité des objets.

25 DROP INDEX La création d'un index spatial n'est PAS une opération "atomique" En cas d'échec lors de la création, il faut d'abord détruire l'index existant avant de relancer la création Si cette destruction est refusée, utilisez la FORCE! DROP INDEX <index_name>; DROP INDEX <index_name> FORCE; DROP INDEX Syntax Spatial indexes created with the CREATE INDEX command can be dropped with the DROP INDEX command. Dropping a spatial index drops the spatial index table. Never manually drop spatial index tables. Sometimes a CREATE INDEX on a column of type SDO_GEOMETRY does not successfully complete. If DROP INDEX index_name does not work, the FORCE option should be used. For example: DROP INDEX index_name FORCE; Complete syntax information for dropping Oracle Spatial indexes can be found in the Oracle Spatial User’s Guide and Reference.

26 Comment trouver de l'information sur les index spatiaux
USER_INDEXES INDEX_TYPE = 'DOMAIN' et ITYP_NAME = 'SPATIAL_INDEX' USER_SDO_INDEX_INFO La colonne SDO_INDEX_TABLE identifie la table physique contenant l'index proprement dit (table dont le nom est de la forme MDRT_xxxx$) USER_SDO_INDEX_METADATA Comme ci-dessus, mais avec plus de détails: nombre de nœuds, taille des nœud, hauteur de l'arbre d'index, etc

27 Export/Import Export Import
Permet le déchargement de tables ou de groupes de tables Pas de syntaxe particulière pour les tables spatiales Les méta-données (USER_SDO_GEOM_METADATA) sont seulement extraites si la table est indexée. Import Permet le rechargement de données exportées par EXPORT Si la table est indexée: restaure les méta-données et re-crée l'index spatial Export/Import Export and import utilities are used for bulk unloading (export) and loading (import) of Oracle tables. There is no Oracle Spatial-specific syntax required for exporting tables with spatial columns. Export creates a file (called a dump file) with the contents of whatever tables were specified on the export command. By default, export unloads all of the data associated with a user’s schema. By specifying the tables parameter, the contents of one or more specific tables can be unloaded. If a spatial column is not indexed, export will only unload the data for the column (not the metadata). If the column is spatially indexed, export will also unload the contents of the USER_SDO_GEOM_METADATA view associated with that table, as well as the command to rebuild the index on import. The Import utility loads data from an export file generated by the export command. If the export file contains tables with spatial columns that were indexed, an entry is added to the USER_SDO_GEOM_METADATA view and the index is rebuilt using the same command it was originally built with. If the spatial column was not indexed, the metadata information will have to be added manually after the import. For more information about export and import, see the Oracle9i Database Utilities documentation.

28 Tablespaces Transportables
A faire sur la base source avant de déplacer le tablespace A faire sur la base cible après l'intégration du tablespace Il n'est pas possible de transporter un tablespace contenant un index spatial entre des plate-formes incompatibles EXECUTE SDO_UTIL.PREPARE_FOR_TTS('my_tablespace'); EXECUTE SDO_UTIL.INITIALIZE_INDEXES_FOR_TTS; !

29 En résumé … Formats des données géo-spatiales Chargement des données
Validation Création des index spatiaux Déplacement des données entre bases Export/import Tablespaces transportables

30