Updated database schema

info579/scripts/bash/create_procedures.sh

@@ -8,8 +8,15 @@ DB_USER="your_username"
 DB_PASSWORD="your_password"
 DB_NAME="stixd_corpus"
 
+# Array of stored procedure scripts in the order they should be run
+procedures_scripts=(
+    "../../sql/procedures/sp_check_prolog_constraints.sql"
+    "../../sql/procedures/sp_check_lexicon_constraints.sql"
+)
+
 # Create procedures
 echo "Creating procedures..."
-mysql -u $DB_USER -p$DB_PASSWORD $DB_NAME < ../../sql/procedures/17_prc_check_prolog_constraints.sql
-
-echo "Procedures created successfully."
+for script in "${procedures_scripts[@]}"; do
+    echo "Running $script..."
+    mysql -u $DB_USER -p$DB_PASSWORD $DB_NAME < $script
+done

info579/scripts/bash/create_tables.sh

@@ -18,7 +18,6 @@ tables_scripts=(
     "../../sql/schema/tables/05_create_table_stix_objects.sql"
     "../../sql/schema/tables/06_create_table_documents.sql"
     "../../sql/schema/tables/07_create_table_sentences.sql"
-    "../../sql/schema/tables/08_create_table_validation_results.sql"
     "../../sql/schema/tables/09_create_table_lexicon.sql"
     "../../sql/schema/tables/20_create_table_obj_doc_jt.sql"
     "../../sql/schema/tables/21_create_table_doc_sent_jt.sql"

info579/scripts/bash/create_triggers.sh

@@ -10,13 +10,7 @@ DB_NAME="stixd_corpus"
 
 # Array of trigger creation scripts in the order they should be run
 triggers_scripts=(
-    "../../sql/triggers/11_trg_lexicon_preposition_gender.sql"
-    "../../sql/triggers/12_trg_lexicon_preposition_check.sql"
-    "../../sql/triggers/13_trg_lexicon_word_class_prep.sql"
-    "../../sql/triggers/14_trg_lexicon_word_class_gender.sql"
-    "../../sql/triggers/15_trg_lexicon_prohibited_words.sql"
-    "../../sql/triggers/16_trg_lexicon_prohibited_intersections.sql"
-    "../../sql/triggers/18_trg_lexicon_prolog_constraints.sql"
+    "../../sql/triggers/master_triggers_lexicon.sql"
 )
 
 # Create triggers

info579/scripts/bash/setup_database.sh

@@ -10,6 +10,6 @@ DB_NAME="stixd_corpus"
 
 # Create the database
 echo "Creating database..."
-mysql -u $DB_USER -p$DB_PASSWORD < ../../sql/setup/00_create_database.sql
+mysql -u $DB_USER -p$DB_PASSWORD < ../../sql/setup/create_database.sql
 
 echo "Database created successfully."

info579/scripts/python/create_tables.py

@@ -13,7 +13,6 @@ def main():
         "../../sql/schema/tables/05_create_table_stix_objects.sql",
         "../../sql/schema/tables/06_create_table_documents.sql",
         "../../sql/schema/tables/07_create_table_sentences.sql",
-        "../../sql/schema/tables/08_create_table_validation_results.sql",
         "../../sql/schema/tables/09_create_table_lexicon.sql",
         "../../sql/schema/tables/20_create_table_obj_doc_jt.sql",
         "../../sql/schema/tables/21_create_table_doc_sent_jt.sql",

info579/scripts/python/setup_database.py

@@ -4,7 +4,7 @@ def run_sql_script(script_path):
     subprocess.run(["mysql", "-u", "your_username", "-pyour_password", "<", script_path], shell=True)
 
 def main():
-    script_path = "../../sql/setup/00_create_database.sql"
+    script_path = "../../sql/setup/create_database.sql"
     print(f"Running {script_path}...")
     run_sql_script(script_path)
     print("Database created successfully.")

info579/sql/procedures/sp_check_lexicon_constraints.sql

@@ -0,0 +1,79 @@
+-- Create procedure to check for prolog constraints
+DELIMITER //
+CREATE PROCEDURE stixd_corpus.check_prolog_constraints(IN base_form VARCHAR(255))
+BEGIN
+    DECLARE done INT DEFAULT FALSE;
+    DECLARE constraint_violation VARCHAR(255);
+    DECLARE constraint_message VARCHAR(255);
+
+    DECLARE cur CURSOR FOR SELECT pattern, message FROM stixd_corpus.prolog_constraints;
+    DECLARE CONTINUE HANDLER FOR NOT FOUND SET done = TRUE;
+
+    OPEN cur;
+
+    read_loop: LOOP
+        FETCH cur INTO constraint_violation, constraint_message;
+        IF done THEN
+            LEAVE read_loop;
+        END IF;
+        IF NOT (base_form REGEXP constraint_violation) THEN
+            SIGNAL SQLSTATE '45000' SET MESSAGE_TEXT = constraint_message;
+        END IF;
+    END LOOP;
+
+    CLOSE cur;
+END;
+//
+DELIMITER ;
+
+-- Create main procedure to check lexicon constraints
+DELIMITER //
+CREATE PROCEDURE stixd_corpus.sp_check_lexicon_constraints(
+    IN new_base_form VARCHAR(255),
+    IN new_word_class VARCHAR(31),
+    IN new_preposition INT,
+    IN new_gender VARCHAR(7)
+)
+BEGIN
+    -- Step 1: Check prolog constraints
+    CALL stixd_corpus.check_prolog_constraints(new_base_form);
+
+    -- Step 2: Check prohibited words
+    DECLARE prohibited_word_found INT DEFAULT 0;
+    SELECT COUNT(*)
+    INTO prohibited_word_found
+    FROM stixd_corpus.prohibited_words
+    WHERE word = new_base_form;
+
+    IF prohibited_word_found > 0 THEN
+        SIGNAL SQLSTATE '45000'
+        SET MESSAGE_TEXT = 'base_form contains a prohibited word';
+    END IF;
+
+    -- Step 3: Check prohibited intersections
+    CALL stixd_corpus.check_prohib_intersections(new_base_form);
+
+    -- Step 4: Ensure preposition is not null if word_class starts with adj_tr or dv_
+    IF (new_word_class LIKE 'adj_tr%' OR new_word_class LIKE 'dv_%') THEN
+        IF (new_preposition IS NULL) THEN
+            SIGNAL SQLSTATE '45000'
+            SET MESSAGE_TEXT = 'preposition must be NOT NULL if word_class starts with "adj_tr" or "dv_"';
+        END IF;
+    END IF;
+
+    -- Step 5: Ensure gender is not null if word_class starts with noun_ or pn
+    IF (new_word_class LIKE 'noun_%' OR new_word_class LIKE 'pn%') THEN
+        IF (new_gender IS NULL) THEN
+            SIGNAL SQLSTATE '45000'
+            SET MESSAGE_TEXT = 'gender must be NOT NULL if word_class starts with "noun_" or "pn"';
+        END IF;
+    END IF;
+
+    -- Step 6: Enforce the mutual exclusivity of preposition and gender
+    IF (new_preposition IS NOT NULL AND new_gender IS NOT NULL) THEN
+        SIGNAL SQLSTATE '45000'
+        SET MESSAGE_TEXT = 'preposition and gender cannot both be NOT NULL';
+    END IF;
+END//
+//
+DELIMITER ;

info579/sql/procedures/17_pr_check_prolog_constraints.sql → info579/sql/procedures/sp_check_prolog_constraints.sql

@@ -1,4 +1,4 @@
--- Create procedure to check for prolog constraints (stix_corpus_18.sql)
+-- Create procedure to check for prolog constraints )
 DELIMITER //
 CREATE PROCEDURE stixd_corpus.check_prolog_constraints (IN base_form VARCHAR(255))
 BEGIN

info579/sql/procedures/sp_run_seed_scripts.sql

@@ -0,0 +1,42 @@
+DELIMITER //
+
+CREATE PROCEDURE stixd_corpus.run_seed_scripts()
+BEGIN
+    -- Seed Genders
+    SET @sql = 'SOURCE ../seeds/seed_genders.sql';
+    PREPARE stmt FROM @sql;
+    EXECUTE stmt;
+    DEALLOCATE PREPARE stmt;
+
+    -- Seed Prohibited Words
+    SET @sql = 'SOURCE ../seeds/seed_prohib_words.sql';
+    PREPARE stmt FROM @sql;
+    EXECUTE stmt;
+    DEALLOCATE PREPARE stmt;
+
+    -- Seed Prohibited Intersections
+    SET @sql = 'SOURCE ../seeds/seed_prohib_intersections.sql';
+    PREPARE stmt FROM @sql;
+    EXECUTE stmt;
+    DEALLOCATE PREPARE stmt;
+
+    -- Seed Prolog Constraints
+    SET @sql = 'SOURCE ../seeds/seed_prolog_constraints.sql';
+    PREPARE stmt FROM @sql;
+    EXECUTE stmt;
+    DEALLOCATE PREPARE stmt;
+
+    -- Seed Word Tags
+    SET @sql = 'SOURCE ../seeds/seed_word_tags.sql';
+    PREPARE stmt FROM @sql;
+    EXECUTE stmt;
+    DEALLOCATE PREPARE stmt;
+
+    -- Seed Special Characters
+    SET @sql = 'SOURCE ../seeds/seed_spec_chars.sql';
+    PREPARE stmt FROM @sql;
+    EXECUTE stmt;
+    DEALLOCATE PREPARE stmt;
+END //
+
+DELIMITER ;

info579/sql/schema/tables/creat_table_word_tags.sql

@@ -0,0 +1,14 @@
+-- Create the stixd_corpus.word_tags table in the specified database
+-- DROP TABLE IF EXISTS stixd_corpus.word_tags;
+CREATE TABLE IF NOT EXISTS stixd_corpus.word_tags (
+    id INT AUTO_INCREMENT PRIMARY KEY,
+    word_tag VARCHAR(12) NOT NULL,
+    first_arg VARCHAR(31) NOT NULL,
+    second_arg VARCHAR(31) NOT NULL,
+    third_arg VARCHAR(31) DEFAULT NULL,
+    word_class VARCHAR(31) NOT NULL
+);
+
+-- SHOW CREATE TABLE stixd_corpus.word_tags
+
+-- Populating the table done with SQL seed script.

info579/sql/schema/tables/22_create_table_doc_lex_jt.sql → info579/sql/schema/tables/create_table_doc_lex_jt.sql

@@ -1,6 +1,6 @@
--- Create the doc_lex_jt table in the specified database
-DROP TABLE IF EXISTS stixd_corpus.doc_lex_jt;
-CREATE TABLE stixd_corpus.doc_lex_jt (
+-- Create the stixd_corpus.doc_lex_jt table
+-- DROP TABLE IF EXISTS stixd_corpus.doc_lex_jt;
+CREATE TABLE IF NOT EXISTS stixd_corpus.doc_lex_jt (
     doc_id INT,
     lex_id INT,
     PRIMARY KEY (doc_id, lex_id),

info579/sql/schema/tables/21_create_table_doc_sent_jt.sql → info579/sql/schema/tables/create_table_doc_sent_jt.sql

@@ -1,6 +1,6 @@
--- Create the doc_sent_jt table in the specified database (stix_corpus_20.sql)
-DROP TABLE IF EXISTS stixd_corpus.doc_sent_jt;
-CREATE TABLE stixd_corpus.doc_sent_jt (
+-- Create the stixd_corpus.doc_sent_jt table
+-- DROP TABLE IF EXISTS stixd_corpus.doc_sent_jt;
+CREATE TABLE IF NOT EXISTS stixd_corpus.doc_sent_jt (
     doc_id INT,
     sent_id INT,
     PRIMARY KEY (doc_id, sent_id),

info579/sql/schema/tables/create_table_documents.sql

@@ -0,0 +1,12 @@
+-- Create the stixd_corpus.documents table
+-- DROP TABLE IF EXISTS stixd_corpus.documents;
+CREATE TABLE IF NOT EXISTS stixd_corpus.documents (
+    doc_id INT AUTO_INCREMENT PRIMARY KEY,
+    raw_text LONGTEXT,
+    raw_text_hash CHAR(64) UNIQUE,
+    proc_text JSON,
+    metadata JSON
+);
+-- SHOW CREATE TABLE stixd_corpus.documents;
+
+-- Populating table done from Python script.

info579/sql/schema/tables/create_table_genders.sql

@@ -0,0 +1,8 @@
+-- Create the stixd_corpus.genders table
+-- DROP TABLE IF EXISTS stixd_corpus.genders;
+CREATE TABLE IF NOT EXISTS stixd_corpus.genders (
+    gender VARCHAR(7) PRIMARY KEY
+);
+-- SHOW CREATE TABLE stixd_corpus.genders;
+
+-- Populating the table done with SQL seed script.

info579/sql/schema/tables/09_create_table_lexicon.sql → info579/sql/schema/tables/create_table_lexicon.sql

@@ -1,6 +1,6 @@
--- Create the lexicon table in the specified database
-DROP TABLE IF EXISTS stixd_corpus.lexicon;
-CREATE TABLE stixd_corpus.lexicon (
+-- Create the stixd_corpus.lexicon table
+-- DROP TABLE IF EXISTS stixd_corpus.lexicon;
+CREATE TABLE IF NOT EXISTS stixd_corpus.lexicon (
     lex_id INT AUTO_INCREMENT PRIMARY KEY,
     word_class VARCHAR(31),
     base_form VARCHAR(255),

info579/sql/schema/tables/20_create_table_obj_doc_jt.sql → info579/sql/schema/tables/create_table_obj_doc_jt.sql

@@ -1,6 +1,6 @@
--- Create the obj_doc_jt table in the specified database
-DROP TABLE IF EXISTS stixd_corpus.obj_doc_jt;
-CREATE TABLE stixd_corpus.obj_doc_jt (
+-- Create the stixd_corpus.obj_doc_jt table
+-- DROP TABLE IF EXISTS stixd_corpus.obj_doc_jt;
+CREATE TABLE IF NOT EXISTS stixd_corpus.obj_doc_jt (
     obj_id VARCHAR(292),
     doc_id INT,
     PRIMARY KEY (obj_id, doc_id),

info579/sql/schema/tables/24_create_table_obj_lex_jt.sql → info579/sql/schema/tables/create_table_obj_lex_jt.sql

@@ -1,6 +1,6 @@
--- Create the obj_lex_jt table in the specified database
-DROP TABLE IF EXISTS stixd_corpus.obj_lex_jt;
-CREATE TABLE stixd_corpus.obj_lex_jt (
+-- Create the stixd_corpus.obj_lex_jt table
+-- DROP TABLE IF EXISTS stixd_corpus.obj_lex_jt;
+CREATE TABLE IF NOT EXISTS stixd_corpus.obj_lex_jt (
     obj_id VARCHAR(292),
     lex_id INT,
     PRIMARY KEY (obj_id, lex_id),

info579/sql/schema/tables/create_table_prohib_intersections.sql

@@ -0,0 +1,10 @@
+-- Create the stixd_corpus.prohibited_intersections table
+-- DROP TABLE IF EXISTS stixd_corpus.prohibited_intersections;
+CREATE TABLE IF NOT EXISTS stixd_corpus.prohibited_intersections (
+    class1 VARCHAR(31),
+    class2 VARCHAR(31),
+    PRIMARY KEY (class1, class2)
+);
+-- SHOW CREATE TABLE stixd_corpus.prohibited_intersections;
+
+-- Populating the table done with SQL seed script.

info579/sql/schema/tables/create_table_prohib_words.sql

@@ -0,0 +1,8 @@
+-- Create the stixd_corpus.prohibited_words table
+-- DROP TABLE IF EXISTS stixd_corpus.prohibited_words;
+CREATE TABLE IF NOT EXISTS stixd_corpus.prohibited_words (
+    word VARCHAR(20) PRIMARY KEY
+);
+-- SHOW CREATE TABLE stixd_corpus.prohibited_words;
+
+-- Populating the table done with SQL seed script.

info579/sql/schema/tables/04_create_table_prolog_constraints.sql → info579/sql/schema/tables/create_table_prolog_constraints.sql

@@ -1,6 +1,6 @@
--- Create the prolog_constraints table in the specified database
-DROP TABLE IF EXISTS stixd_corpus.prolog_constraints;
-CREATE TABLE stixd_corpus.prolog_constraints (
+-- Create the stixd_corpus.prolog_constraints table
+-- DROP TABLE IF EXISTS stixd_corpus.prolog_constraints;
+CREATE TABLE IF NOT EXISTS stixd_corpus.prolog_constraints (
     constraint_id INT AUTO_INCREMENT PRIMARY KEY,
     description VARCHAR(511),
     pattern VARCHAR(255),

info579/sql/schema/tables/23_create_table_sent_lex_jt.sql → info579/sql/schema/tables/create_table_sent_lex_jt.sql

@@ -1,6 +1,6 @@
--- Create the sent_lex_jt table in the specified database
-DROP TABLE IF EXISTS stixd_corpus.sent_lex_jt;
-CREATE TABLE stixd_corpus.sent_lex_jt (
+-- Create the stixd_corpus.sent_lex_jt table
+-- DROP TABLE IF EXISTS stixd_corpus.sent_lex_jt;
+CREATE TABLE IF NOT EXISTS stixd_corpus.sent_lex_jt (
     sent_id INT,
     lex_id INT,
     PRIMARY KEY (sent_id, lex_id),