// Package filter provides a safety filter pipeline that runs over extracted content.
//
// This is defense-in-depth, not the primary security boundary.
// Primary security is the typed extraction itself — raw content never reaches
// the agent. The filter layer catches patterns that slipped through the
// structured extraction (e.g., code injection in prose paragraphs).
//
// Filters are deterministic regex/string operations. No LLM involvement.

package filter

import (
	"regexp"
	"strings"

	"github.com/sscoble/specstream/pkg/types"
)

// FilterFunc transforms an ExtractedContent in place. Returns warnings.
type FilterFunc func(c *types.ExtractedContent) []string

// StandardFilters returns the default filter pipeline.
func StandardFilters() []FilterFunc {
	return []FilterFunc{
		StripNullBytes,
		StripControlChars,
		TruncateParagraphs(5000),
		StripCommonInjectionPatterns,
	}
}

// StripNullBytes removes null bytes from all text fields.
func StripNullBytes(c *types.ExtractedContent) []string {
	var warnings []string
	for i := range c.Sections {
		c.Sections[i].Heading = strings.ReplaceAll(c.Sections[i].Heading, "\x00", "")
		for j := range c.Sections[i].Paragraphs {
			if strings.Contains(c.Sections[i].Paragraphs[j], "\x00") {
				c.Sections[i].Paragraphs[j] = strings.ReplaceAll(c.Sections[i].Paragraphs[j], "\x00", "")
			}
		}
	}
	return warnings
}

// StripControlChars removes ASCII control characters (except \n, \t, \r).
func StripControlChars(c *types.ExtractedContent) []string {
	var warnings []string
	re := regexp.MustCompile(`[\x00-\x08\x0b\x0c\x0e-\x1f\x7f]`)
	for i := range c.Sections {
		c.Sections[i].Heading = re.ReplaceAllString(c.Sections[i].Heading, "")
		for j := range c.Sections[i].Paragraphs {
			if re.MatchString(c.Sections[i].Paragraphs[j]) {
				c.Sections[i].Paragraphs[j] = re.ReplaceAllString(c.Sections[i].Paragraphs[j], "")
			}
		}
	}
	return warnings
}

// TruncateParagraphs caps paragraph length.
func TruncateParagraphs(maxLen int) FilterFunc {
	return func(c *types.ExtractedContent) []string {
		var warnings []string
		for i := range c.Sections {
			for j := range c.Sections[i].Paragraphs {
				if len(c.Sections[i].Paragraphs[j]) > maxLen {
					c.Sections[i].Paragraphs[j] = c.Sections[i].Paragraphs[j][:maxLen]
				}
			}
		}
		return warnings
	}
}

// StripCommonInjectionPatterns removes known prompt injection markers
// from paragraph text. This is a best-effort blocklist — the primary
// defense is the structured extraction itself.
func StripCommonInjectionPatterns(c *types.ExtractedContent) []string {
	var warnings []string

	patterns := []struct {
		name    string
		pattern *regexp.Regexp
	}{
		{
			"ignore-previous",
			regexp.MustCompile(`(?i)(ignore|forget|disregard|override)\s+(all\s+)?(previous|prior|above)\s+(instructions|commands|directions|prompts?)`),
		},
		{
			"new-instructions",
			regexp.MustCompile(`(?i)(now|instead|from\s+now\s+on)\s+(you\s+are|you\'?re|act\s+as|behave\s+as|pretend)\s+`),
		},
		{
			"system-override",
			regexp.MustCompile(`(?i)(system\s+prompt|system\s+message|SYSTEM:|system_message)\s*[:=]`),
		},
		{
			"role-switch",
			regexp.MustCompile(`(?i)^\s*(user|assistant|system)\s*[:]\s*`),
		},
		{
			"dangerous-command",
			regexp.MustCompile(`(?i)(rm\s+-(rf|r)\s+/|format\s+(drive|disk|volume)|dd\s+if=)\s*`),
		},
	}

	for _, p := range patterns {
		for i := range c.Sections {
			orig := c.Sections[i].Heading
			c.Sections[i].Heading = p.pattern.ReplaceAllString(c.Sections[i].Heading, "[REDACTED]")
			if orig != c.Sections[i].Heading {
				c.Metadata.StrippedPatterns[p.name]++
				warnings = append(warnings, p.name)
			}
			for j := range c.Sections[i].Paragraphs {
				orig = c.Sections[i].Paragraphs[j]
				c.Sections[i].Paragraphs[j] = p.pattern.ReplaceAllString(c.Sections[i].Paragraphs[j], "[REDACTED]")
				if orig != c.Sections[i].Paragraphs[j] {
					c.Metadata.StrippedPatterns[p.name]++
					warnings = append(warnings, p.name)
				}
			}
		}
	}
	if len(warnings) > 0 {
		// Deduplicate
		seen := make(map[string]bool)
		unique := []string{}
		for _, w := range warnings {
			if !seen[w] {
				seen[w] = true
				unique = append(unique, w)
			}
		}
		return unique
	}
	return nil
}