Murrough-Foley · SupperTomato · Jul 21, 2026 · Jul 22, 2026 · Jul 22, 2026 · Jul 22, 2026
diff --git a/Cargo.toml b/Cargo.toml
@@ -1,6 +1,6 @@
 [package]
 name = "rs-trafilatura"
-version = "0.2.2"
+version = "0.3.0"
 edition = "2021"
 rust-version = "1.85"
 authors = ["Murrough Foley"]
@@ -21,7 +21,7 @@ dom_query = "0.24"         # Primary DOM manipulation
 tendril = "0.4"            # Zero-copy text operations via StrTendril
 thiserror = "2.0"
 regex = "1.11"
-chrono = "0.4"
+chrono = { version = "0.4", features = ["serde"] }
 encoding_rs = "0.8"
 serde = { version = "1.0", features = ["derive"] }
 serde_json = "1.0"

diff --git a/README.md b/README.md
@@ -4,6 +4,12 @@ Fast and accurate web content extraction in Rust.
 
 A high-performance Rust port of [trafilatura](https://github.com/adbar/trafilatura) / [go-trafilatura](https://github.com/markusmobius/go-trafilatura), extracting clean, readable content from web pages while removing boilerplate, navigation, and advertisements.
 
+## Project Status
+
+The crate is suitable for local services, batch extraction, and pipelines that can tolerate occasional extraction misses and use `extraction_quality` to route low-confidence pages to a fallback. For production-facing workloads, validate against your own URL corpus, cap output sizes, and keep browser/LLM fallbacks for JavaScript-heavy or low-confidence pages.
+
+The page-type classifier and extraction-quality predictor are Rust-specific additions. They are not part of upstream Python Trafilatura; they route pages to tuned extraction profiles and expose confidence signals for fallback decisions.
+
 ## Features
 
 - **Fast**: 71 files/s for articles, 46 files/s overall on a 1,497-page benchmark (pure Rust, compile-time regex)
@@ -13,7 +19,7 @@ A high-performance Rust port of [trafilatura](https://github.com/adbar/trafilatu
 - **Extraction Quality Predictor**: ML-based confidence scoring (0.0-1.0) using a 27-feature XGBoost model that predicts extraction F1 — pages below 0.80 are candidates for LLM fallback
 - **Markdown Output**: GitHub Flavored Markdown preserving headings, lists, tables, bold/italic, code blocks
 - **Rich Metadata**: Title, author, date, description, categories, tags, license, images from JSON-LD, Open Graph, Dublin Core, and HTML meta tags
-- **Configurable**: 28 options to tune precision/recall tradeoff, content selection, and output format
+- **Configurable**: 30 options to tune precision/recall tradeoff, content selection, and output format
 - **Robust**: Handles malformed HTML gracefully with automatic character encoding detection (UTF-8, ISO-8859-1, Windows-1252)
 
 ## Quick Start

diff --git a/benches/benchmark.rs b/benches/benchmark.rs
@@ -7,7 +7,7 @@
 //! - Real-world HTML files from benchmark dataset for realistic performance
 
 use criterion::{black_box, criterion_group, criterion_main, BenchmarkId, Criterion, Throughput};
-use rs_trafilatura::{extract, extract_with_options, Options};
+use rs_trafilatura::{baseline, extract, extract_with_options, html2txt, Options};
 use std::fs;
 
 const SAMPLE_HTML: &str = r#"
@@ -48,6 +48,66 @@ const SAMPLE_HTML: &str = r#"
 </html>
 "#;
 
+const JSON_LD_HTML: &str = r#"
+<!DOCTYPE html>
+<html>
+<head>
+    <title>Structured Data Article</title>
+    <script type="application/ld+json">
+    {
+        "@context": "https://schema.org",
+        "@type": "FAQPage",
+        "mainEntity": {
+            "@type": "Question",
+            "name": "How does extraction work?",
+            "acceptedAnswer": {
+                "@type": "Answer",
+                "text": "Rust extraction parses the DOM, scores content candidates, and falls back to structured JSON-LD text when visible content is not sufficient."
+            }
+        },
+        "recipeInstructions": [
+            {"text": "Parse the document once."},
+            {"itemListElement": [{"text": "Collect structured body text."}]}
+        ]
+    }
+    </script>
+    <script type="application/ld+json">
+    {
+        "@context": "https://schema.org",
+        "@type": "Product",
+        "description": "This product teaser is available only as structured metadata and should be kept as fallback text."
+    }
+    </script>
+</head>
+<body><nav>Home Products Cart</nav></body>
+</html>
+"#;
+
+const METADATA_ONLY_JSON_LD_HTML: &str = r#"
+<!DOCTYPE html>
+<html>
+<head>
+    <title>Metadata Only</title>
+    <script type="application/ld+json">
+    {"@context":"https://schema.org","@type":"Organization","name":"Example","url":"https://example.com"}
+    </script>
+    <script type="application/ld+json">
+    {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"name":"Home"},{"name":"Section"}]}
+    </script>
+    <script type="application/ld+json">
+    {"@context":"https://schema.org","@type":"WebSite","name":"Example","potentialAction":{"@type":"SearchAction"}}
+    </script>
+</head>
+<body>
+    <article>
+        <p>Fallback should eventually use visible article text when metadata-only JSON-LD contains no content body.</p>
+        <p>This benchmark measures the cost of scanning irrelevant structured data scripts.</p>
+        <p>The optimized path should skip JSON parsing when no content-bearing hook is present.</p>
+    </article>
+</body>
+</html>
+"#;
+
 fn bench_extract_default(c: &mut Criterion) {
     c.bench_function("extract_default", |b| {
         b.iter(|| extract(black_box(SAMPLE_HTML)));
@@ -66,6 +126,31 @@ fn bench_extract_with_options(c: &mut Criterion) {
     });
 }
 
+fn bench_official_api_helpers(c: &mut Criterion) {
+    let mut group = c.benchmark_group("official_api");
+    group.throughput(Throughput::Bytes(SAMPLE_HTML.len() as u64));
+
+    group.bench_function("baseline_sample", |b| {
+        b.iter(|| baseline(black_box(SAMPLE_HTML)));
+    });
+
+    group.bench_function("html2txt_clean_sample", |b| {
+        b.iter(|| html2txt(black_box(SAMPLE_HTML), black_box(true)));
+    });
+
+    group.throughput(Throughput::Bytes(JSON_LD_HTML.len() as u64));
+    group.bench_function("baseline_json_ld", |b| {
+        b.iter(|| baseline(black_box(JSON_LD_HTML)));
+    });
+
+    group.throughput(Throughput::Bytes(METADATA_ONLY_JSON_LD_HTML.len() as u64));
+    group.bench_function("baseline_metadata_only_json_ld", |b| {
+        b.iter(|| baseline(black_box(METADATA_ONLY_JSON_LD_HTML)));
+    });
+
+    group.finish();
+}
+
 /// Benchmark with real-world HTML files of varying sizes
 fn bench_real_world_html(c: &mut Criterion) {
     let html_dir = "../data/html_files";
@@ -97,6 +182,7 @@ criterion_group!(
     benches,
     bench_extract_default,
     bench_extract_with_options,
+    bench_official_api_helpers,
     bench_real_world_html
 );
 criterion_main!(benches);
diff --git a/rust-toolchain.toml b/rust-toolchain.toml
@@ -1,3 +1,3 @@
 [toolchain]
-channel = "1.85"
+channel = "1.86"
 components = ["rustfmt", "clippy"]
diff --git a/src/etree.rs b/src/etree.rs
@@ -20,6 +20,7 @@
 //! This module re-exports functions from the `html-cleaning` crate.
 
 // Re-export all tree functions from html-cleaning for backward compatibility
+#[allow(unused_imports)]
 pub use html_cleaning::tree::{
     append, element, extend, is_void_element, iter, iter_descendants, iter_text, remove,
     set_tail, set_text, strip, strip_elements, strip_tags, sub_element, tail, tail_nodes, text,