"""
SQL syntax highlighter for QPlainTextEdit.
Highlights keywords, types, functions, strings, comments, and numbers.
"""
import re
from PyQt6.QtCore import QRegularExpression, Qt
from PyQt6.QtGui import (
    QSyntaxHighlighter, QTextCharFormat, QColor, QFont
)


def _fmt(color: str, bold: bool = False, italic: bool = False) -> QTextCharFormat:
    f = QTextCharFormat()
    f.setForeground(QColor(color))
    if bold:   f.setFontWeight(QFont.Weight.Bold)
    if italic: f.setFontItalic(True)
    return f


# ── Token categories (Catppuccin Frappé palette) ─────────────────────────────
_KEYWORD_FMT   = _fmt("#8caaee", bold=True)   # blue   — DDL/DML/control
_TYPE_FMT      = _fmt("#ef9f76")               # peach  — data types
_FUNCTION_FMT  = _fmt("#a6d189")               # green  — functions
_STRING_FMT    = _fmt("#a6d189")               # green  — string literals
_NUMBER_FMT    = _fmt("#ef9f76")               # orange — numeric literals
_COMMENT_FMT   = _fmt("#737994", italic=True)  # grey   — comments
_OPERATOR_FMT  = _fmt("#ca9ee6")               # mauve  — operators/special
_STAR_FMT      = _fmt("#ca9ee6", bold=True)    # mauve  — * wildcard

# ── Keyword lists ─────────────────────────────────────────────────────────────
_KEYWORDS = [
    "SELECT", "FROM", "WHERE", "JOIN", "LEFT", "RIGHT", "INNER", "OUTER",
    "FULL", "CROSS", "ON", "AS", "AND", "OR", "NOT", "IN", "LIKE", "BETWEEN",
    "IS", "NULL", "EXISTS", "CASE", "WHEN", "THEN", "ELSE", "END",
    "INSERT", "INTO", "VALUES", "UPDATE", "SET", "DELETE", "MERGE",
    "CREATE", "TABLE", "VIEW", "INDEX", "DATABASE", "SCHEMA", "SEQUENCE",
    "DROP", "ALTER", "ADD", "COLUMN", "PRIMARY", "KEY", "FOREIGN",
    "REFERENCES", "UNIQUE", "DEFAULT", "CONSTRAINT", "CHECK",
    "IF", "EXISTS", "NOT", "TRUNCATE", "RENAME", "MODIFY", "CHANGE",
    "GRANT", "REVOKE", "COMMIT", "ROLLBACK", "SAVEPOINT", "BEGIN",
    "TRANSACTION", "START", "END", "LOCK", "UNLOCK", "EXPLAIN",
    "ANALYZE", "VACUUM", "SHOW", "DESCRIBE", "DESC", "ASC",
    "LIMIT", "OFFSET", "ORDER", "BY", "GROUP", "HAVING", "DISTINCT",
    "UNION", "ALL", "INTERSECT", "EXCEPT", "PARTITION", "OVER",
    "WITH", "RECURSIVE", "USE", "CALL", "EXEC", "EXECUTE",
    "PROCEDURE", "FUNCTION", "TRIGGER", "EVENT", "REPLACE",
    "TRUE", "FALSE", "RETURNING",
]

_TYPES = [
    "INT", "INTEGER", "BIGINT", "SMALLINT", "TINYINT", "MEDIUMINT",
    "FLOAT", "DOUBLE", "DECIMAL", "NUMERIC", "REAL", "MONEY",
    "VARCHAR", "CHAR", "TEXT", "TINYTEXT", "MEDIUMTEXT", "LONGTEXT",
    "BLOB", "TINYBLOB", "MEDIUMBLOB", "LONGBLOB", "BINARY", "VARBINARY",
    "DATE", "DATETIME", "TIMESTAMP", "TIME", "YEAR",
    "BOOLEAN", "BOOL", "BIT", "ENUM", "SET",
    "JSON", "UUID", "SERIAL", "BYTEA", "NVARCHAR", "NCHAR",
    "IMAGE", "XML", "CURSOR", "ROWVERSION", "UNIQUEIDENTIFIER",
]

_FUNCTIONS = [
    "COUNT", "SUM", "AVG", "MIN", "MAX", "COALESCE", "IFNULL", "NULLIF",
    "NOW", "CURDATE", "CURTIME", "DATE", "YEAR", "MONTH", "DAY",
    "CONCAT", "SUBSTRING", "LENGTH", "TRIM", "UPPER", "LOWER",
    "REPLACE", "ROUND", "FLOOR", "CEIL", "ABS", "MOD", "POWER",
    "CAST", "CONVERT", "ISNULL", "ISNUMERIC",
    "ROW_NUMBER", "RANK", "DENSE_RANK", "LAG", "LEAD", "NTILE",
    "FIRST_VALUE", "LAST_VALUE", "NTH_VALUE",
    "STRING_AGG", "GROUP_CONCAT", "ARRAY_AGG",
    "TO_CHAR", "TO_DATE", "TO_NUMBER", "EXTRACT",
    "DATEDIFF", "DATEADD", "DATE_FORMAT", "DATE_TRUNC",
    "IF", "IIF", "DECODE", "GREATEST", "LEAST",
]


class SQLHighlighter(QSyntaxHighlighter):
    """Applies syntax colouring to a SQL document."""

    def __init__(self, document):
        super().__init__(document)
        self._rules: list[tuple] = []

        def kw_pattern(words: list[str]) -> str:
            return r"\b(?:" + "|".join(words) + r")\b"

        self._rules = [
            # Keywords (case-insensitive handled via flag)
            (QRegularExpression(kw_pattern(_KEYWORDS),
                                QRegularExpression.PatternOption.CaseInsensitiveOption),
             _KEYWORD_FMT),
            # Data types
            (QRegularExpression(kw_pattern(_TYPES),
                                QRegularExpression.PatternOption.CaseInsensitiveOption),
             _TYPE_FMT),
            # Functions
            (QRegularExpression(kw_pattern(_FUNCTIONS),
                                QRegularExpression.PatternOption.CaseInsensitiveOption),
             _FUNCTION_FMT),
            # Numbers
            (QRegularExpression(r"\b\d+(\.\d+)?\b"), _NUMBER_FMT),
            # Single-quoted strings
            (QRegularExpression(r"'[^'\\]*(?:\\.[^'\\]*)*'"), _STRING_FMT),
            # Double-quoted identifiers
            (QRegularExpression(r'"[^"]*"'), _fmt("#99d1db")),
            # Backtick identifiers (MySQL)
            (QRegularExpression(r"`[^`]*`"), _fmt("#99d1db")),
            # * wildcard
            (QRegularExpression(r"\bSELECT\s+\*|\*(?=\s*FROM)",
                                QRegularExpression.PatternOption.CaseInsensitiveOption),
             _STAR_FMT),
            # Operators
            (QRegularExpression(r"[=<>!%&|^~+\-*/]"), _OPERATOR_FMT),
            # Single-line comment  --
            (QRegularExpression(r"--[^\n]*"), _COMMENT_FMT),
            # Single-line comment  #
            (QRegularExpression(r"#[^\n]*"), _COMMENT_FMT),
        ]

        # Multi-line block comment  /* ... */
        self._block_comment_start = QRegularExpression(r"/\*")
        self._block_comment_end   = QRegularExpression(r"\*/")

    def highlightBlock(self, text: str) -> None:
        # Single-line rules
        for pattern, fmt in self._rules:
            it = pattern.globalMatch(text)
            while it.hasNext():
                m = it.next()
                self.setFormat(m.capturedStart(), m.capturedLength(), fmt)

        # Multi-line block comments
        self.setCurrentBlockState(0)
        start_idx = 0
        if self.previousBlockState() != 1:
            m = self._block_comment_start.match(text)
            start_idx = m.capturedStart() if m.hasMatch() else -1

        while start_idx >= 0:
            end_m = self._block_comment_end.match(text, start_idx)
            if end_m.hasMatch():
                end_idx = end_m.capturedStart() + end_m.capturedLength()
                self.setFormat(start_idx, end_idx - start_idx, _COMMENT_FMT)
                m = self._block_comment_start.match(text, end_idx)
                start_idx = m.capturedStart() if m.hasMatch() else -1
            else:
                self.setCurrentBlockState(1)
                self.setFormat(start_idx, len(text) - start_idx, _COMMENT_FMT)
                break