Updated SQL code

info579/sql/procedures/sp_check_lexicon_constraints.sql

@@ -1,45 +1,19 @@
--- Create procedure to check for prolog constraints
-DELIMITER //
-CREATE PROCEDURE stixd_corpus.check_prolog_constraints(IN word_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 (word_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
+-- Create main procedure to check lexicon constraints (sp_check_lexicon_constraints.sql)
 DELIMITER //
 CREATE PROCEDURE stixd_corpus.sp_check_lexicon_constraints(
-    IN new_word_tag VARCHAR(12),
-    IN new_word_form VARCHAR(99),
-    IN new_logical_symbol VARCHAR(99),
-    IN new_third_arg VARCHAR(15)
+    IN new_word_tag VARCHAR(12) CHARACTER SET utf8mb4 COLLATE utf8mb4_unicode_ci,
+    IN new_word_form VARCHAR(99) CHARACTER SET utf8mb4 COLLATE utf8mb4_unicode_ci,
+    IN new_logical_symbol VARCHAR(99) CHARACTER SET utf8mb4 COLLATE utf8mb4_unicode_ci,
+    IN new_third_arg VARCHAR(15) CHARACTER SET utf8mb4 COLLATE utf8mb4_unicode_ci
 )
 BEGIN
     DECLARE prohibited_word_found INT DEFAULT 0;
-    DECLARE tag_form_hash VARCHAR(64);
+    DECLARE tag_form_hash VARCHAR(64) CHARACTER SET utf8mb4 COLLATE utf8mb4_unicode_ci;
+    DECLARE new_word_class VARCHAR(31) CHARACTER SET utf8mb4 COLLATE utf8mb4_unicode_ci;
+    DECLARE new_preposition INT DEFAULT NULL;
 
     -- Step 1: Check prolog constraints for word_form
-    CALL stixd_corpus.check_prolog_constraints(new_word_form);
+    CALL stixd_corpus.sp_check_prolog_constraints(new_word_form);
 
     -- Step 2: Check prohibited words
     SELECT COUNT(*)
@@ -53,10 +27,10 @@ BEGIN
     END IF;
 
     -- Step 3: Check prohibited intersections for word_form
-    CALL stixd_corpus.check_prohib_intersections(new_word_form);
+    CALL stixd_corpus.sp_check_prohib_intersections(new_word_tag, new_word_form);
 
-    -- Step 4: Generate the SHA256 hash for word_tag and word_form combination
-    SET tag_form_hash = SHA2(CONCAT(new_word_tag, new_word_form), 256);
+    -- Step 4: Generate and validate hashes
+    CALL stixd_corpus.sp_gen_hashes(new_word_tag, new_word_form, tag_form_hash);
 
     -- Step 5: Check if the tag_form_hash already exists in the lexicon table
     IF EXISTS (SELECT 1 FROM stixd_corpus.lexicon WHERE tag_form_hash = tag_form_hash) THEN
@@ -85,6 +59,6 @@ BEGIN
         SIGNAL SQLSTATE '45000'
         SET MESSAGE_TEXT = 'logical_symbol and third_arg cannot both be NOT NULL';
     END IF;
-END//
+END //
 //
 DELIMITER ;

info579/sql/procedures/sp_check_prohib_intersections.sql

@@ -0,0 +1,46 @@
+-- Create procedure to check for prohibited word_word_tag intersections  (sp_check_prohib_intersections.sql)
+DELIMITER //
+
+CREATE PROCEDURE stixd_corpus.sp_check_prohib_intersections(
+    IN new_word_tag VARCHAR(12) CHARACTER SET utf8mb4 COLLATE utf8mb4_unicode_ci,
+    IN new_word_form VARCHAR(99) CHARACTER SET utf8mb4 COLLATE utf8mb4_unicode_ci
+)
+BEGIN
+    DECLARE tag_check INT DEFAULT 0;
+
+    -- Check if the new_word_form exists in the lexicon table
+    IF EXISTS (SELECT 1 FROM stixd_corpus.lexicon WHERE word_form = new_word_form) THEN
+        -- Check first condition
+        SELECT COUNT(*)
+        INTO tag_check
+        FROM stixd_corpus.prohibited_intersections
+        WHERE word_tag1 = new_word_tag
+          AND word_tag2 IN (SELECT word_tag
+                         FROM stixd_corpus.lexicon
+                         WHERE word_form = new_word_form);
+
+        IF tag_check > 0 THEN
+            SIGNAL SQLSTATE '45000'
+            SET MESSAGE_TEXT = 'This combination of word_tags is prohibited';
+        END IF;
+
+        -- Reset the variable
+        SET tag_check = 0;
+
+        -- Check second condition
+        SELECT COUNT(*)
+        INTO tag_check
+        FROM stixd_corpus.prohibited_intersections
+        WHERE word_tag2 = new_word_tag
+          AND word_tag1 IN (SELECT word_tag
+                         FROM stixd_corpus.lexicon
+                         WHERE word_form = new_word_form);
+
+        IF tag_check > 0 THEN
+            SIGNAL SQLSTATE '45000'
+            SET MESSAGE_TEXT = 'This combination of word_tags is prohibited';
+        END IF;
+    END IF;
+END;
+//
+DELIMITER ;

info579/sql/procedures/sp_check_prolog_constraints.sql

@@ -1,27 +1,50 @@
--- Create procedure to check for prolog constraints )
+-- Create procedure to check for prohibited word_class intersections of word_class and preposition (sp_check_prohib_intersections.sql)
 DELIMITER //
-CREATE PROCEDURE stixd_corpus.check_prolog_constraints (IN word_form VARCHAR(255))
+
+CREATE PROCEDURE stixd_corpus.sp_check_prohib_intersections(
+    IN new_word_tag VARCHAR(31) CHARACTER SET utf8mb4 COLLATE utf8mb4_unicode_ci,
+    IN new_word_form VARCHAR(255) CHARACTER SET utf8mb4 COLLATE utf8mb4_unicode_ci
+)
 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 (word_form REGEXP constraint_violation) THEN
-            SIGNAL SQLSTATE '45000' SET MESSAGE_TEXT = constraint_message;
-        END IF;
-    END LOOP;
-
-    CLOSE cur;
+    DECLARE class_check INT DEFAULT 0;
+
+    -- Get the word_class and preposition from the lexicon table
+    DECLARE new_word_class VARCHAR(31) CHARACTER SET utf8mb4 COLLATE utf8mb4_unicode_ci;
+    DECLARE new_preposition INT;
+
+    SELECT word_class INTO new_word_class FROM stixd_corpus.word_tags WHERE word_tag = new_word_tag;
+    SELECT preposition INTO new_preposition FROM stixd_corpus.lexicon WHERE word_form = new_word_form;
+
+    -- Check first condition
+    SELECT COUNT(*)
+    INTO class_check
+    FROM stixd_corpus.prohibited_intersections
+    WHERE class1 = new_word_class
+      AND class2 IN (SELECT word_class
+                     FROM stixd_corpus.lexicon
+                     WHERE lex_id = new_preposition);
+
+    IF class_check > 0 THEN
+        SIGNAL SQLSTATE '45000'
+        SET MESSAGE_TEXT = 'This combination of word_class and preposition is prohibited';
+    END IF;
+
+    -- Reset the variable
+    SET class_check = 0;
+
+    -- Check second condition
+    SELECT COUNT(*)
+    INTO class_check
+    FROM stixd_corpus.prohibited_intersections
+    WHERE class2 = new_word_class
+      AND class1 IN (SELECT word_class
+                     FROM stixd_corpus.lexicon
+                     WHERE lex_id = new_preposition);
+
+    IF class_check > 0 THEN
+        SIGNAL SQLSTATE '45000'
+        SET MESSAGE_TEXT = 'This combination of word_class and preposition is prohibited';
+    END IF;
 END;
 //
 DELIMITER ;

info579/sql/procedures/sp_gen_hases.sql

@@ -0,0 +1,14 @@
+-- Create procedure to generate and return hash (sp_gen_hashes.sql)
+DELIMITER //
+
+CREATE PROCEDURE stixd_corpus.sp_gen_hashes(
+    IN new_word_tag VARCHAR(12) CHARACTER SET utf8mb4 COLLATE utf8mb4_unicode_ci,
+    IN new_word_form VARCHAR(99) CHARACTER SET utf8mb4 COLLATE utf8mb4_unicode_ci,
+    OUT tag_form_hash VARCHAR(64) CHARACTER SET utf8mb4 COLLATE utf8mb4_unicode_ci
+)
+BEGIN
+    -- Generate the SHA256 hash for word_tag and word_form combination
+    SET tag_form_hash = SHA2(CONCAT(new_word_tag, new_word_form), 256);
+END;
+//
+DELIMITER ;

info579/sql/procedures/sp_run_seed_scripts.sql

@@ -1,3 +1,4 @@
+-- Seed all the ACE 6.7 tables in the database (sp_run_seed_scripts.sql)
 DELIMITER //
 
 CREATE PROCEDURE stixd_corpus.run_seed_scripts()

info579/sql/schema/tables/create_table_doc_lex_jt.sql

@@ -1,4 +1,4 @@
--- Create the stixd_corpus.doc_lex_jt table
+-- Create the stixd_corpus.doc_lex_jt table (create_table_doc_lex_jt.sql)
 -- DROP TABLE IF EXISTS stixd_corpus.doc_lex_jt;
 CREATE TABLE IF NOT EXISTS stixd_corpus.doc_lex_jt (
     doc_id INT,
@@ -7,4 +7,5 @@ CREATE TABLE IF NOT EXISTS stixd_corpus.doc_lex_jt (
     FOREIGN KEY (doc_id) REFERENCES stixd_corpus.documents(doc_id),
     FOREIGN KEY (lex_id) REFERENCES stixd_corpus.lexicon(lex_id)
 );
--- SHOW CREATE TABLE stixd_corpus.doc_lex_jt;
+-- SHOW CREATE TABLE stixd_corpus.doc_lex_jt;
+-- Populating table done from primary tables.

info579/sql/schema/tables/create_table_doc_sent_jt.sql

@@ -1,4 +1,4 @@
--- Create the stixd_corpus.doc_sent_jt table
+-- Create the stixd_corpus.doc_sent_jt table (create_table_doc_sent_jt.sql)
 -- DROP TABLE IF EXISTS stixd_corpus.doc_sent_jt;
 CREATE TABLE IF NOT EXISTS stixd_corpus.doc_sent_jt (
     doc_id INT,
@@ -7,4 +7,5 @@ CREATE TABLE IF NOT EXISTS stixd_corpus.doc_sent_jt (
     FOREIGN KEY (doc_id) REFERENCES stixd_corpus.documents(doc_id),
     FOREIGN KEY (sent_id) REFERENCES stixd_corpus.sentences(sent_id)
 );
--- SHOW CREATE TABLE stixd_corpus.doc_sent_jt;
+-- SHOW CREATE TABLE stixd_corpus.doc_sent_jt;
+-- Populating table done from primary tables.

info579/sql/schema/tables/create_table_documents.sql

@@ -1,12 +1,11 @@
--- Create the stixd_corpus.documents table
+-- Create the stixd_corpus.documents table (create_table_documents.sql)
 -- 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,
+    raw_text LONGTEXT CHARACTER SET utf8mb4 COLLATE utf8mb4_unicode_ci,
+    raw_text_hash CHAR(64) CHARACTER SET utf8mb4 COLLATE utf8mb4_unicode_ci 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

@@ -1,8 +1,7 @@
--- Create the stixd_corpus.genders table
+-- Create the stixd_corpus.genders table (create_table_genders.sql)
 -- DROP TABLE IF EXISTS stixd_corpus.genders;
 CREATE TABLE IF NOT EXISTS stixd_corpus.genders (
-    gender VARCHAR(7) PRIMARY KEY
+    gender VARCHAR(7) CHARACTER SET utf8mb4 COLLATE utf8mb4_unicode_ci PRIMARY KEY
 );
 -- SHOW CREATE TABLE stixd_corpus.genders;
-
--- Populating the table done with SQL seed script.
+-- Populating the table done with SQL seed script.

info579/sql/schema/tables/create_table_lexicon.sql

@@ -1,16 +1,15 @@
--- Create the stixd_corpus.lexicon table
+-- Create the stixd_corpus.lexicon table (create_table_lexicon.sql)
 DROP TABLE IF EXISTS stixd_corpus.lexicon;
 CREATE TABLE IF NOT EXISTS stixd_corpus.lexicon (
     lex_id INT AUTO_INCREMENT PRIMARY KEY,
-    word_tag VARCHAR(12),
-    word_form VARCHAR(99),
-    logical_symbol VARCHAR(99),
-    third_arg VARCHAR(15),
-    tag_form_hash VARCHAR(64) UNIQUE,
-    word_def TEXT,
+    word_tag VARCHAR(12) CHARACTER SET utf8mb4 COLLATE utf8mb4_unicode_ci,
+    word_form VARCHAR(99) CHARACTER SET utf8mb4 COLLATE utf8mb4_unicode_ci,
+    logical_symbol VARCHAR(99) CHARACTER SET utf8mb4 COLLATE utf8mb4_unicode_ci,
+    third_arg VARCHAR(15) CHARACTER SET utf8mb4 COLLATE utf8mb4_unicode_ci,
+    tag_form_hash VARCHAR(64) CHARACTER SET utf8mb4 COLLATE utf8mb4_unicode_ci UNIQUE,
+    word_def TEXT CHARACTER SET utf8mb4 COLLATE utf8mb4_unicode_ci,
     synsets JSON,
     tagsets JSON
 );
 -- SHOW CREATE TABLE stixd_corpus.lexicon;
-
--- Populating table done from Python script.
+-- Populating table done from Python script.

info579/sql/schema/tables/create_table_obj_doc_jt.sql

@@ -1,10 +1,11 @@
--- 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),
+-- Create the stixd_corpus.doc_sent_jt table (create_table_doc_sent_jt.sql)
+-- DROP TABLE IF EXISTS stixd_corpus.doc_sent_jt;
+CREATE TABLE IF NOT EXISTS stixd_corpus.doc_sent_jt (
     doc_id INT,
-    PRIMARY KEY (obj_id, doc_id),
-    FOREIGN KEY (obj_id) REFERENCES stixd_corpus.stix_objects(obj_id),
-    FOREIGN KEY (doc_id) REFERENCES stixd_corpus.documents(doc_id)
+    sent_id INT,
+    PRIMARY KEY (doc_id, sent_id),
+    FOREIGN KEY (doc_id) REFERENCES stixd_corpus.documents(doc_id),
+    FOREIGN KEY (sent_id) REFERENCES stixd_corpus.sentences(sent_id)
 );
--- SHOW CREATE TABLE stixd_corpus.obj_doc_jt;
+-- SHOW CREATE TABLE stixd_corpus.doc_sent_jt;
+-- Populating table done from primary tables.

info579/sql/schema/tables/create_table_obj_lex_jt.sql

@@ -1,4 +1,4 @@
--- Create the stixd_corpus.obj_lex_jt table
+-- Create the stixd_corpus.obj_lex_jt table (create_table_obj_lex_jt.sql)
 -- DROP TABLE IF EXISTS stixd_corpus.obj_lex_jt;
 CREATE TABLE IF NOT EXISTS stixd_corpus.obj_lex_jt (
     obj_id VARCHAR(292),
@@ -7,4 +7,5 @@ CREATE TABLE IF NOT EXISTS stixd_corpus.obj_lex_jt (
     FOREIGN KEY (obj_id) REFERENCES stixd_corpus.stix_objects(obj_id),
     FOREIGN KEY (lex_id) REFERENCES stixd_corpus.lexicon(lex_id)
 );
--- SHOW CREATE TABLE stixd_corpus.obj_lex_jt;
+-- SHOW CREATE TABLE stixd_corpus.obj_lex_jt;
+-- Populating table done from primary tables.

info579/sql/schema/tables/create_table_prohib_intersections.sql

@@ -1,10 +1,9 @@
--- Create the stixd_corpus.prohibited_intersections table
+-- Create the stixd_corpus.prohibited_intersections table (create_table_prohib_intersections.sql)
 -- 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)
+    word_tag1 VARCHAR(12) CHARACTER SET utf8mb4 COLLATE utf8mb4_unicode_ci,
+    word_tag2 VARCHAR(12) CHARACTER SET utf8mb4 COLLATE utf8mb4_unicode_ci,
+    PRIMARY KEY (word_tag1, word_tag2)
 );
 -- SHOW CREATE TABLE stixd_corpus.prohibited_intersections;
-
 -- Populating the table done with SQL seed script.

info579/sql/schema/tables/create_table_prohib_words.sql

@@ -1,8 +1,7 @@
--- Create the stixd_corpus.prohibited_words table
+-- Create the stixd_corpus.prohibited_words table (create_table_prohib_words.sql)
 -- DROP TABLE IF EXISTS stixd_corpus.prohibited_words;
 CREATE TABLE IF NOT EXISTS stixd_corpus.prohibited_words (
-    word VARCHAR(20) PRIMARY KEY
+    word VARCHAR(20) CHARACTER SET utf8mb4 COLLATE utf8mb4_unicode_ci PRIMARY KEY
 );
 -- SHOW CREATE TABLE stixd_corpus.prohibited_words;
-
--- Populating the table done with SQL seed script.
+-- Populating the table done with SQL seed script.

info579/sql/schema/tables/create_table_prolog_constraints.sql

@@ -1,10 +1,11 @@
--- Create the stixd_corpus.prolog_constraints table
+-- Create the stixd_corpus.prolog_constraints table (create_table_prolog_constraints.sql)
 -- 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),
-    message VARCHAR(255),
+    description VARCHAR(511) CHARACTER SET utf8mb4 COLLATE utf8mb4_unicode_ci,
+    pattern VARCHAR(255) CHARACTER SET utf8mb4 COLLATE utf8mb4_unicode_ci,
+    message VARCHAR(255) CHARACTER SET utf8mb4 COLLATE utf8mb4_unicode_ci,
     auto_correct BOOLEAN
 );
--- SHOW CREATE TABLE stixd_corpus.prolog_constraints;
+-- SHOW CREATE TABLE stixd_corpus.prolog_constraints;
+-- Populating the table done with SQL seed script.