refactor: make crawler importable and add export pipeline

Move monkeyd and pdfout out of internal/ so other modules can import them,
and add export.Export, which holds the crawl-to-PDF sequence the CLI used to
inline. The CLI now parses flags and delegates, so an embedding program gets
the same defaults and validation.

monkeyd-crawler/monkeyd/css_words.go

@@ -0,0 +1,50 @@
+package monkeyd
+
+import (
+	"regexp"
+	"strconv"
+	"strings"
+)
+
+// The site hides part of every chapter behind CSS rather than putting it in the
+// markup. Chapter HTML carries empty elements such as
+//
+//	Nghe <span class="t-3e625e..."></span> trưởng tử
+//
+// and the page stylesheet supplies the missing word:
+//
+//	.t-3e625e...:before { content: "vị"; }
+//
+// Reading DOM text alone therefore drops hundreds of words per chapter without
+// any visible error. wordRule finds those rules so the words can be put back.
+var wordRule = regexp.MustCompile(`\.([A-Za-z0-9_-]+)\s*::?before\s*\{[^}]*?content\s*:\s*"((?:[^"\\]|\\.)*)"`)
+
+// cssEscape matches a CSS character escape: a hex code point, optionally
+// followed by one whitespace terminator, or an escaped literal character.
+var cssEscape = regexp.MustCompile(`\\([0-9A-Fa-f]{1,6})\s?|\\(.)`)
+
+// ParseWordClasses maps CSS class name to the word its :before rule injects.
+func ParseWordClasses(page []byte) map[string]string {
+	words := make(map[string]string)
+	for _, m := range wordRule.FindAllSubmatch(page, -1) {
+		words[string(m[1])] = decodeCSSString(string(m[2]))
+	}
+	return words
+}
+
+// decodeCSSString resolves the escape sequences allowed inside a CSS string.
+func decodeCSSString(s string) string {
+	if !strings.Contains(s, `\`) {
+		return s
+	}
+	return cssEscape.ReplaceAllStringFunc(s, func(esc string) string {
+		m := cssEscape.FindStringSubmatch(esc)
+		if m[1] != "" {
+			if cp, err := strconv.ParseInt(m[1], 16, 32); err == nil && cp > 0 {
+				return string(rune(cp))
+			}
+			return ""
+		}
+		return m[2]
+	})
+}