Splits multi-row inserts with semicolons

src/sql-converter.ts

@@ -203,7 +203,16 @@ function convertStatement(
         return convertCreateTable(trimmed, context)
     }
 
-    return convertGeneralStatement(trimmed)
+    const converted = convertGeneralStatement(trimmed)
+
+    if (
+        /^INSERT\b/i.test(converted) &&
+        hasSemicolonInsideSingleQuotedString(converted)
+    ) {
+        return splitMultiRowInsert(converted)
+    }
+
+    return converted
 }
 
 function convertCreateTable(
@@ -476,6 +485,93 @@ function convertGeneralStatement(statement: string): string {
         .trim()
 }
 
+function hasSemicolonInsideSingleQuotedString(statement: string): boolean {
+    let inSingleQuote = false
+
+    for (let index = 0; index < statement.length; index += 1) {
+        const char = statement[index]
+        const next = statement[index + 1]
+
+        if (!inSingleQuote) {
+            if (char === "'") {
+                inSingleQuote = true
+            }
+            continue
+        }
+
+        if (char === "'" && next === "'") {
+            index += 1
+            continue
+        }
+
+        if (char === "'") {
+            inSingleQuote = false
+            continue
+        }
+
+        if (char === ';') {
+            return true
+        }
+    }
+
+    return false
+}
+
+function splitMultiRowInsert(statement: string): string {
+    const valuesKeywordIndex = findValuesKeywordOutsideQuotes(statement)
+
+    if (valuesKeywordIndex === -1) {
+        return statement
+    }
+
+    const prefix = statement
+        .slice(0, valuesKeywordIndex + 'VALUES'.length)
+        .trimEnd()
+    const values = statement.slice(valuesKeywordIndex + 'VALUES'.length).trim()
+    const valuesWithoutSemicolon = values.endsWith(';')
+        ? values.slice(0, -1)
+        : values
+    const rows = splitCommaSeparated(valuesWithoutSemicolon)
+
+    if (rows.length <= 1 || !rows.every(row => row.startsWith('('))) {
+        return statement
+    }
+
+    return rows.map(row => `${prefix} ${row}`).join(';\n')
+}
+
+function findValuesKeywordOutsideQuotes(statement: string): number {
+    let quote: "'" | '"' | '`' | null = null
+
+    for (let index = 0; index < statement.length; index += 1) {
+        const char = statement[index]
+        const next = statement[index + 1]
+
+        if (quote) {
+            if (char === "'" && quote === "'" && next === "'") {
+                index += 1
+                continue
+            }
+
+            if (char === quote) {
+                quote = null
+            }
+            continue
+        }
+
+        if (char === "'" || char === '"' || char === '`') {
+            quote = char
+            continue
+        }
+
+        if (/^VALUES\b/i.test(statement.slice(index))) {
+            return index
+        }
+    }
+
+    return -1
+}
+
 function extractIndexes(
     definitions: string[],
     tableName: string,

tests/sql-converter.test.ts

@@ -40,14 +40,36 @@ describe('convertMariaDbToSqlite', () => {
         expect(result.warnings).toHaveLength(1)
     })
 
-    test('preserves inserts with semicolons inside strings', () => {
+    test('splits multi-row inserts with semicolons inside strings', () => {
         const input =
             "INSERT INTO `notes` (`body`) VALUES ('one; two'), ('three');"
 
         const result = convertMariaDbToSqlite(input)
 
         expect(result.sql.trim()).toBe(
-            'INSERT INTO "notes" ("body") VALUES (\'one; two\'), (\'three\');',
+            'INSERT INTO "notes" ("body") VALUES (\'one; two\');\nINSERT INTO "notes" ("body") VALUES (\'three\');',
+        )
+    })
+
+    test('preserves single-row inserts with html entity semicolons', () => {
+        const input =
+            "INSERT INTO `expenses` (`description`) VALUES ('PENGANTARAN SOLAR EXCA SANY 01 & EXCA XCMG');"
+
+        const result = convertMariaDbToSqlite(input)
+
+        expect(result.sql.trim()).toBe(
+            'INSERT INTO "expenses" ("description") VALUES (\'PENGANTARAN SOLAR EXCA SANY 01 & EXCA XCMG\');',
+        )
+    })
+
+    test('splits risky multi-row inserts with apostrophes before semicolon values', () => {
+        const input =
+            "INSERT INTO `rent_item_rents` VALUES ('BIKIN JALAN DI KEBUN PAK MU\\'MIN'),('PENGANTARAN SOLAR EXCA SANY 01 & EXCA XCMG');"
+
+        const result = convertMariaDbToSqlite(input)
+
+        expect(result.sql.trim()).toBe(
+            "INSERT INTO \"rent_item_rents\" VALUES ('BIKIN JALAN DI KEBUN PAK MU''MIN');\nINSERT INTO \"rent_item_rents\" VALUES ('PENGANTARAN SOLAR EXCA SANY 01 & EXCA XCMG');",
         )
     })
 
@@ -190,6 +212,17 @@ describe('convertMariaDbToSqlite', () => {
         expect(result.sql).not.toContain("\\'")
     })
 
+    test('uses standard sqlite quote escaping for apostrophes in insert values', () => {
+        const input =
+            "INSERT INTO `rent_item_rents` VALUES ('BIKIN JALAN DI KEBUN PAK MU\\'MIN','PENGANTARAN SOLAR EXCA SANY 01 & EXCA XCMG');"
+
+        const result = convertMariaDbToSqlite(input)
+
+        expect(result.sql.trim()).toBe(
+            "INSERT INTO \"rent_item_rents\" VALUES ('BIKIN JALAN DI KEBUN PAK MU''MIN','PENGANTARAN SOLAR EXCA SANY 01 & EXCA XCMG');",
+        )
+    })
+
     test('preserves literal backticks inside JSON string values', () => {
         const input = `INSERT INTO \`activity_logs\` (\`model_value_changed\`) VALUES ('{\\"note\\":\\"\`\\",\\"uuid\\":\\"019cab37\\"}');`