feat(cli): add --dump markdown output mode (#54) (#78)

Surfaces the existing LP.getMarkdown CDP machinery as a CLI flag so text-pipeline users (LLM agents, scrapers feeding markdown to other tools) don't need a separate HTML-to-markdown converter step. obscura fetch --dump markdown 'https://example.com' Layout: - crates/obscura-js/src/markdown.rs holds HTML_TO_MARKDOWN_JS as a pub const so every layer can share the same DOM-walker. obscura-js is at the bottom of the dep DAG (obscura-browser -> obscura-js, obscura-cdp -> obscura-js, obscura-cli -> obscura-js) so a single source of truth works without a circular dependency. - crates/obscura-browser/src/lib.rs re-exports it for ergonomics. - crates/obscura-cdp/src/domains/lp.rs drops its inline duplicate and calls the shared const. - crates/obscura-cli/src/main.rs adds the `Markdown` DumpFormat variant + dump_markdown(&mut Page) which page.evaluate's the shared script and prints the result. Net diff: a new 75-line markdown.rs, a 70+ line deletion in lp.rs that's reborn as 3 lines, and ~10 lines added across the other four files. No behavior change for `--dump html|text|links`, and LP.getMarkdown over CDP keeps the exact same JSON shape. Four new unit tests in obscura-js cover headings, inline emphasis + links, ordered/unordered lists, and that <script> / <style> contents do not leak into the output. Generated by Ora Studio Vibe coded by ousamabenyounes Co-authored-by: Ora Studio <[email protected]>
2026-05-13 22:36:17 +02:00
1 parent 9fbd0c73a3
commit d8b7c5fad1
@@ -5,3 +5,6 @@ pub mod module_loader;
 pub mod runtime;
 pub mod ops;
 pub mod v8_lock;
+pub mod markdown;
+
+pub use markdown::HTML_TO_MARKDOWN_JS;
@@ -0,0 +1,71 @@
+//! Shared markdown extraction script used by the LP.getMarkdown CDP method
+//! and the CLI `--dump markdown` mode. Lives in obscura-browser so both
+//! obscura-cdp and obscura-cli can call it without depending on each other.
+
+/// JS expression that walks `document.body` and returns a markdown string.
+/// Must be evaluated against a Page that has a fully-bootstrapped JS runtime.
+pub const HTML_TO_MARKDOWN_JS: &str = r#"
+(function() {
+    function toMd(el, depth) {
+        if (!el) return '';
+        var out = '';
+        if (el.nodeType === 3) return el.textContent || '';
+        if (el.nodeType !== 1) return '';
+        var tag = (el.tagName || '').toLowerCase();
+        var children = '';
+        var cn = el.childNodes || [];
+        for (var i = 0; i < cn.length; i++) children += toMd(cn[i], depth);
+        children = children.replace(/\n{3,}/g, '\n\n');
+        switch(tag) {
+            case 'h1': return '\n# ' + children.trim() + '\n\n';
+            case 'h2': return '\n## ' + children.trim() + '\n\n';
+            case 'h3': return '\n### ' + children.trim() + '\n\n';
+            case 'h4': return '\n#### ' + children.trim() + '\n\n';
+            case 'h5': return '\n##### ' + children.trim() + '\n\n';
+            case 'h6': return '\n###### ' + children.trim() + '\n\n';
+            case 'p': return '\n' + children.trim() + '\n\n';
+            case 'br': return '\n';
+            case 'hr': return '\n---\n\n';
+            case 'strong': case 'b': return '**' + children + '**';
+            case 'em': case 'i': return '*' + children + '*';
+            case 'code': return '`' + children + '`';
+            case 'pre': return '\n```\n' + children + '\n```\n\n';
+            case 'blockquote': return '\n> ' + children.trim().replace(/\n/g, '\n> ') + '\n\n';
+            case 'a':
+                var href = el.getAttribute('href') || '';
+                if (href && children.trim()) return '[' + children.trim() + '](' + href + ')';
+                return children;
+            case 'img':
+                var src = el.getAttribute('src') || '';
+                var alt = el.getAttribute('alt') || '';
+                return '![' + alt + '](' + src + ')';
+            case 'ul': case 'ol':
+                return '\n' + children + '\n';
+            case 'li':
+                var parent = el.parentNode;
+                var isOrdered = parent && parent.tagName && parent.tagName.toLowerCase() === 'ol';
+                var bullet = isOrdered ? '1. ' : '- ';
+                return bullet + children.trim() + '\n';
+            case 'table': return '\n' + children + '\n';
+            case 'thead': case 'tbody': case 'tfoot': return children;
+            case 'tr':
+                var cells = [];
+                var tds = el.childNodes || [];
+                for (var j = 0; j < tds.length; j++) {
+                    if (tds[j].nodeType === 1) cells.push(toMd(tds[j], depth).trim());
+                }
+                return '| ' + cells.join(' | ') + ' |\n';
+            case 'th': case 'td': return children;
+            case 'script': case 'style': case 'noscript': case 'link': case 'meta': return '';
+            case 'div': case 'section': case 'article': case 'main': case 'aside': case 'nav': case 'header': case 'footer':
+                return '\n' + children;
+            case 'span': return children;
+            default: return children;
+        }
+    }
+    var body = document.body || document.documentElement;
+    var md = toMd(body, 0);
+    md = md.replace(/\n{3,}/g, '\n\n').trim();
+    return md;
+})()
+"#;
@@ -1466,6 +1466,72 @@ mod tests {
        assert_eq!(kind, serde_json::json!(true));
    }

+    #[test]
+    fn test_html_to_markdown_headings() {
+        let mut rt = setup_runtime("<html><body><h1>Title</h1><h2>Sub</h2><p>Body</p></body></html>");
+        let md = rt
+            .evaluate(crate::HTML_TO_MARKDOWN_JS)
+            .unwrap()
+            .as_str()
+            .unwrap()
+            .to_string();
+        assert!(md.contains("# Title"), "missing H1: {}", md);
+        assert!(md.contains("## Sub"), "missing H2: {}", md);
+        assert!(md.contains("Body"), "missing paragraph text: {}", md);
+    }
+
+    #[test]
+    fn test_html_to_markdown_links_and_inline() {
+        let mut rt = setup_runtime(
+            r#"<html><body><p>Hello <strong>world</strong> <a href="https://x.test/">link</a> <em>em</em></p></body></html>"#,
+        );
+        let md = rt
+            .evaluate(crate::HTML_TO_MARKDOWN_JS)
+            .unwrap()
+            .as_str()
+            .unwrap()
+            .to_string();
+        assert!(md.contains("**world**"), "missing strong: {}", md);
+        assert!(md.contains("*em*"), "missing em: {}", md);
+        assert!(
+            md.contains("[link](https://x.test/)"),
+            "missing link: {}",
+            md
+        );
+    }
+
+    #[test]
+    fn test_html_to_markdown_lists() {
+        let mut rt = setup_runtime(
+            "<html><body><ul><li>A</li><li>B</li></ul><ol><li>X</li><li>Y</li></ol></body></html>",
+        );
+        let md = rt
+            .evaluate(crate::HTML_TO_MARKDOWN_JS)
+            .unwrap()
+            .as_str()
+            .unwrap()
+            .to_string();
+        assert!(md.contains("- A"), "missing unordered A: {}", md);
+        assert!(md.contains("- B"), "missing unordered B: {}", md);
+        assert!(md.contains("1. X"), "missing ordered X: {}", md);
+    }
+
+    #[test]
+    fn test_html_to_markdown_skips_script_and_style() {
+        let mut rt = setup_runtime(
+            "<html><body><p>Text</p><script>alert(1)</script><style>body{color:red}</style></body></html>",
+        );
+        let md = rt
+            .evaluate(crate::HTML_TO_MARKDOWN_JS)
+            .unwrap()
+            .as_str()
+            .unwrap()
+            .to_string();
+        assert!(md.contains("Text"), "missing visible text: {}", md);
+        assert!(!md.contains("alert"), "leaked script content: {}", md);
+        assert!(!md.contains("color:red"), "leaked style content: {}", md);
+    }
+
    #[test]
    fn test_page_content_puppeteer_pattern() {
        let mut rt = setup_runtime("<!DOCTYPE html><html><head></head><body><p>Test</p></body></html>");