package docsite

import (
	"fmt"
	"os"
	"path"
	"regexp"
	"slices"
	"strings"
)

// forbiddenName matches the consuming-application spellings framework docs
// must never contain (D-11), the accented variant included.
var forbiddenName = regexp.MustCompile(`(?i)fonoteka|p[lł][yý]tarium`)

const forbiddenMessage = "consuming-application name in output"

// forbiddenLines returns the 1-based line numbers of text that name a
// consuming application. The match itself is never reported.
func forbiddenLines(text []byte) []int {
	var lines []int
	for i, line := range strings.Split(string(text), "\n") {
		if forbiddenName.MatchString(line) {
			lines = append(lines, i+1)
		}
	}
	return lines
}

// checkForbiddenSources scans the full source file of every page (the
// frontmatter and a README's summary line included).
func (s *site) checkForbiddenSources() ([]Problem, error) {
	var problems []Problem
	for _, p := range s.pages {
		raw, err := os.ReadFile(p.abs)
		if err != nil {
			return nil, fmt.Errorf("docsite: read %s: %w", p.Source, err)
		}
		for _, line := range forbiddenLines(raw) {
			problems = append(problems, Problem{File: p.Source, Line: line, Rule: "forbidden", Message: forbiddenMessage})
		}
	}
	return problems, nil
}

// forbiddenOutputExts are the text outputs scanned after rendering.
var forbiddenOutputExts = []string{".html", ".md", ".txt", ".json", ".js", ".css"}

// checkForbiddenOutputs scans every rendered text output, which covers
// text pulled in through src= snippet copies and templates.
func (s *site) checkForbiddenOutputs() []Problem {
	names := make([]string, 0, len(s.outputs))
	for name := range s.outputs {
		if slices.Contains(forbiddenOutputExts, path.Ext(name)) {
			names = append(names, name)
		}
	}
	slices.Sort(names)
	var problems []Problem
	for _, name := range names {
		for _, line := range forbiddenLines(s.outputs[name]) {
			problems = append(problems, Problem{File: name, Line: line, Rule: "forbidden", Message: forbiddenMessage})
		}
	}
	return problems
}