// SiteOne Crawler - HtmlProcessor
// (c) Jan Reges <jan.reges@siteone.cz>
//
// Extracts URLs from HTML content and applies offline conversion changes.
use once_cell::sync::Lazy;
use regex::Regex;
use crate::content_processor::base_processor::{ProcessorConfig, convert_url_to_relative, is_relevant};
use crate::content_processor::content_processor::ContentProcessor;
use crate::engine::found_url::UrlSource;
use crate::engine::found_urls::FoundUrls;
use crate::engine::parsed_url::ParsedUrl;
use crate::types::ContentTypeId;
use crate::utils;
pub const JS_VARIABLE_NAME_URL_DEPTH: &str = "_SiteOneUrlDepth";
pub const HTML_PAGES_EXTENSIONS: &[&str] = &[
"htm", "html", "shtml", "php", "phtml", "ashx", "xhtml", "asp", "aspx", "jsp", "jspx", "do", "cfm", "cgi", "pl",
];
static HTML_EXT_REGEX: Lazy<Regex> = Lazy::new(|| {
let pattern = format!(r"(?i)\.({})", HTML_PAGES_EXTENSIONS.join("|"));
Regex::new(&pattern).unwrap()
});
static RE_A_HREF: Lazy<Regex> =
Lazy::new(|| Regex::new(r#"(?is)<a[^>]*\shref=["']?([^#][^"'\s>]+)["'\s]?[^>]*>"#).unwrap());
static RE_ESCAPED_HREF: Lazy<Regex> =
Lazy::new(|| Regex::new(r#"(?i)href\\["'][:=]\\["'](https?://[^"'\\]+)\\["']"#).unwrap());
static RE_FONT_URL: Lazy<Regex> = Lazy::new(|| {
Regex::new(r#"(?is)url\s*\(\s*['"]?([^'"\s>]+\.(eot|ttf|woff2|woff|otf)[^'")\s]*)['"]?\s*\)"#).unwrap()
});
static RE_FONT_LINK: Lazy<Regex> = Lazy::new(|| {
Regex::new(r#"(?is)<link\s+[^>]*href=["']?([^"' ]+\.(eot|ttf|woff2|woff|otf)[^"' ]*)["']?[^>]*>"#).unwrap()
});
static RE_IMG_SRC: Lazy<Regex> =
Lazy::new(|| Regex::new(r#"(?is)<img\s+[^>]*?src=["']?([^"'> ]+)["']?[^>]*>"#).unwrap());
static RE_IMG_DATA_SRC: Lazy<Regex> =
Lazy::new(|| Regex::new(r#"(?is)<img\s+[^>]*?data-src=["']?([^"'> ]+)["']?[^>]*>"#).unwrap());
static RE_INPUT_SRC: Lazy<Regex> =
Lazy::new(|| Regex::new(r#"(?is)<input\s+[^>]*?src=["']?([^"'> ]+\.[a-z0-9]{1,10})["']?[^>]*>"#).unwrap());
static RE_LINK_IMAGE: Lazy<Regex> = Lazy::new(|| {
Regex::new(r#"(?is)<link\s+[^>]*?href=["']?([^"'> ]+\.(png|gif|jpg|jpeg|webp|avif|tif|bmp|svg|ico)(\?[^"' ]*|))["']?[^>]*>"#).unwrap()
});
static RE_SOURCE_SRC: Lazy<Regex> =
Lazy::new(|| Regex::new(r#"(?is)<source\s+[^>]*?src=["']([^"'>]+)["'][^>]*>"#).unwrap());
static RE_CSS_URL_IMAGE: Lazy<Regex> = Lazy::new(|| {
Regex::new(r#"(?is)url\s*\(\s*['"]?([^'")\s]+\.(jpg|jpeg|png|gif|bmp|tif|webp|avif)[^'")\s]*)['"]?\s*\)"#).unwrap()
});
static RE_SOURCE_SRCSET: Lazy<Regex> =
Lazy::new(|| Regex::new(r#"(?is)<source\s+[^>]*?srcset=["']([^"'>]+)["'][^>]*>"#).unwrap());
static RE_IMG_SRCSET: Lazy<Regex> = Lazy::new(|| Regex::new(r#"(?is)<img[^>]+srcset=["']([^"']+)["']"#).unwrap());
static RE_IMAGESRCSET: Lazy<Regex> =
Lazy::new(|| Regex::new(r#"(?is)<[a-z]+[^>]+imagesrcset=["']([^"']+)["']"#).unwrap());
static RE_AUDIO_SRC: Lazy<Regex> =
Lazy::new(|| Regex::new(r#"(?is)<audio\s+[^>]*?src=["']?([^"'> ]+)["']?[^>]*>"#).unwrap());
static RE_VIDEO_SRC: Lazy<Regex> =
Lazy::new(|| Regex::new(r#"(?is)<video\s+[^>]*?src=["']?([^"'> ]+)["']?[^>]*>"#).unwrap());
static RE_SCRIPT_SRC: Lazy<Regex> =
Lazy::new(|| Regex::new(r#"(?is)<script\s+[^>]*?src=["']?([^"' ]+)["']?[^>]*>"#).unwrap());
static RE_LINK_JS: Lazy<Regex> =
Lazy::new(|| Regex::new(r#"(?is)<link\s+[^>]*href=["']?([^"'> ]+\.(json|js)(\?[^"']*|))["']?[^>]*>"#).unwrap());
static RE_DOT_SRC: Lazy<Regex> = Lazy::new(|| Regex::new(r#"(?is)\.src\s*=\s*["']([^"']+)["']"#).unwrap());
static RE_NEXTJS_CHUNKS: Lazy<Regex> =
Lazy::new(|| Regex::new(r#"(?is):([a-z0-9/._\-\[\]]+chunks[a-z0-9/._\-\[\]]+\.js)"#).unwrap());
static RE_LINK_STYLESHEET: Lazy<Regex> =
Lazy::new(|| Regex::new(r#"(?is)<link\s+[^>]*?href=["']([^"']+)["'][^>]*>"#).unwrap());
static RE_FILE_EXTENSION: Lazy<Regex> = Lazy::new(|| Regex::new(r"(?i)\.[a-z0-9]{1,10}(\?.*)?$").unwrap());
// Offline version regexes
static RE_HREF_SRC: Lazy<Regex> = Lazy::new(|| {
Regex::new(r#"(?is)(\.|<[a-z0-9]{1,10}[^>]*\s+)(href|src|component-url)\s*(=)\s*(['"]?)([^'">]+)['"]?([^>]*)"#)
.unwrap()
});
static RE_SRCSET_ATTR: Lazy<Regex> = Lazy::new(|| {
Regex::new(
r#"(?is)(\.|<[a-z0-9]{1,10}[^>]*\s+)(imagesrcset|srcset|renderer-url)\s*(=)\s*(['"]?)([^'">]+)['"]?([^>]*)"#,
)
.unwrap()
});
static RE_META_URL: Lazy<Regex> =
Lazy::new(|| Regex::new(r#"(?im)(<meta[^>]*)(url)\s*(=)\s*(['"]?)([^'">]+)['"]?(")"#).unwrap());
static RE_ESCAPED_HREF_SRC: Lazy<Regex> =
Lazy::new(|| Regex::new(r#"(?is)(.)(href\\["']|src\\["'])([:=])(\\["'])([^"'\\]+)\\["'](.)"#).unwrap());
static RE_META_REFRESH: Lazy<Regex> = Lazy::new(|| Regex::new(r#"(?i)(<meta[^>]*url=)([^"']+)(["'][^>]*>)"#).unwrap());
static RE_PORT_NORMALIZE: Lazy<Regex> = Lazy::new(|| Regex::new(r"(?i)((https?:)?//[a-z0-9._-]+):[0-9]+").unwrap());
static RE_CLOSE_HEAD: Lazy<Regex> = Lazy::new(|| Regex::new(r"(?i)<\s*/\s*head\s*>").unwrap());
static RE_CLOSE_BODY: Lazy<Regex> = Lazy::new(|| Regex::new(r"(?i)<\s*/\s*body\s*>").unwrap());
static RE_NON_HTTP_SCHEME: Lazy<Regex> = Lazy::new(|| Regex::new(r"(?i)^[a-z]+:[a-z0-9+]").unwrap());
static RE_EXTERNAL_SCRIPT: Lazy<Regex> =
Lazy::new(|| Regex::new(r#"(?is)<script[^>]*src=["']?(.*?)["']?[^>]*>.*?</script>"#).unwrap());
static RE_EXTERNAL_URL: Lazy<Regex> = Lazy::new(|| Regex::new(r"(?i)^(https?:)?//").unwrap());
static RE_CROSSORIGIN_LINK: Lazy<Regex> =
Lazy::new(|| Regex::new(r#"(?i)(<link[^>]+)\s*crossorigin(\s*=\s*["']?.*?["']?)?(\s*[^>]*>)"#).unwrap());
static RE_CROSSORIGIN_SCRIPT: Lazy<Regex> =
Lazy::new(|| Regex::new(r#"(?i)(<script[^>]+)\s*crossorigin(\s*=\s*["']?.*?["']?)?(\s*[^>]*>)"#).unwrap());
static RE_SCRIPT_BLOCK: Lazy<Regex> = Lazy::new(|| Regex::new(r"(?is)<script[^>]*>(.*?)</script>").unwrap());
static RE_SOCNET_IFRAME: Lazy<Regex> =
Lazy::new(|| Regex::new(r"(?is)<iframe[^>]*(facebook\.com|twitter\.com|linkedin\.com)[^>]*>.*?</iframe>").unwrap());
pub struct HtmlProcessor {
config: ProcessorConfig,
debug_mode: bool,
relevant_content_types: Vec<ContentTypeId>,
}
impl HtmlProcessor {
pub fn new(config: ProcessorConfig) -> Self {
Self {
config,
debug_mode: false,
relevant_content_types: vec![ContentTypeId::Html, ContentTypeId::Redirect],
}
}
/// Find <a href> URLs
fn find_href_urls(&self, html: &str, source_url: &ParsedUrl, found_urls: &mut FoundUrls) {
let source_url_str = source_url.get_full_url(true, false);
// Standard <a href="..."> links
let mut urls: Vec<String> = Vec::new();
for caps in RE_A_HREF.captures_iter(html) {
if let Some(m) = caps.get(1) {
urls.push(m.as_str().to_string());
}
}
// Escaped href URLs (e.g., href\":\")
for caps in RE_ESCAPED_HREF.captures_iter(html) {
if let Some(m) = caps.get(1) {
urls.push(m.as_str().to_string());
}
}
// If single_foreign_page is set and source is on a different 2nd-level domain, skip
if self.config.single_foreign_page && source_url.domain_2nd_level != self.config.initial_url.domain_2nd_level {
return;
}
// Filter by max depth
if self.config.max_depth > 0 {
urls.retain(|url_str| {
let parsed = ParsedUrl::parse(url_str, Some(source_url));
parsed.get_depth() <= self.config.max_depth as usize
});
}
// Filter out files if files are disabled
if !self.config.files_enabled {
urls.retain(|url_str| !RE_FILE_EXTENSION.is_match(url_str) || HTML_EXT_REGEX.is_match(url_str));
}
let url_refs: Vec<&str> = urls.iter().map(|s| s.as_str()).collect();
found_urls.add_urls_from_text_array(&url_refs, &source_url_str, UrlSource::AHref);
}
/// Find font URLs in CSS and link tags
fn find_fonts(&self, html: &str, source_url: &ParsedUrl, found_urls: &mut FoundUrls) {
let source_url_str = source_url.get_full_url(true, false);
// CSS @font-face url()
let font_urls: Vec<&str> = RE_FONT_URL
.captures_iter(html)
.filter_map(|caps| caps.get(1).map(|m| m.as_str()))
.collect();
found_urls.add_urls_from_text_array(&font_urls, &source_url_str, UrlSource::CssUrl);
// <link href="...(font extensions)"
let link_fonts: Vec<&str> = RE_FONT_LINK
.captures_iter(html)
.filter_map(|caps| caps.get(1).map(|m| m.as_str()))
.collect();
found_urls.add_urls_from_text_array(&link_fonts, &source_url_str, UrlSource::LinkHref);
}
/// Find image URLs from various sources
fn find_images(&self, html: &str, source_url: &ParsedUrl, found_urls: &mut FoundUrls) {
let source_url_str = source_url.get_full_url(true, false);
// <img src="..."
let img_srcs: Vec<&str> = RE_IMG_SRC
.captures_iter(html)
.filter_map(|caps| caps.get(1).map(|m| m.as_str()))
.collect();
found_urls.add_urls_from_text_array(&img_srcs, &source_url_str, UrlSource::ImgSrc);
// <img data-src="..." (lazy loading)
let data_srcs: Vec<&str> = RE_IMG_DATA_SRC
.captures_iter(html)
.filter_map(|caps| caps.get(1).map(|m| m.as_str()))
.collect();
found_urls.add_urls_from_text_array(&data_srcs, &source_url_str, UrlSource::ImgSrc);
// <input src="..."
let input_srcs: Vec<&str> = RE_INPUT_SRC
.captures_iter(html)
.filter_map(|caps| caps.get(1).map(|m| m.as_str()))
.collect();
found_urls.add_urls_from_text_array(&input_srcs, &source_url_str, UrlSource::InputSrc);
// <link href="...(image extensions)"
let link_imgs: Vec<&str> = RE_LINK_IMAGE
.captures_iter(html)
.filter_map(|caps| caps.get(1).map(|m| m.as_str()))
.collect();
found_urls.add_urls_from_text_array(&link_imgs, &source_url_str, UrlSource::LinkHref);
// <source src="..."
let source_srcs: Vec<&str> = RE_SOURCE_SRC
.captures_iter(html)
.filter_map(|caps| caps.get(1).map(|m| m.as_str()))
.collect();
found_urls.add_urls_from_text_array(&source_srcs, &source_url_str, UrlSource::SourceSrc);
// CSS url() with image extensions
let css_imgs: Vec<&str> = RE_CSS_URL_IMAGE
.captures_iter(html)
.filter_map(|caps| caps.get(1).map(|m| m.as_str()))
.collect();
found_urls.add_urls_from_text_array(&css_imgs, &source_url_str, UrlSource::CssUrl);
// srcset from <source>, <img>, and imagesrcset
let mut srcset_urls: Vec<String> = Vec::new();
let mut srcset_values: Vec<&str> = Vec::new();
for caps in RE_SOURCE_SRCSET.captures_iter(html) {
if let Some(m) = caps.get(1) {
srcset_values.push(m.as_str());
}
}
for caps in RE_IMG_SRCSET.captures_iter(html) {
if let Some(m) = caps.get(1) {
srcset_values.push(m.as_str());
}
}
for caps in RE_IMAGESRCSET.captures_iter(html) {
if let Some(m) = caps.get(1) {
srcset_values.push(m.as_str());
}
}
for srcset in &srcset_values {
// srcset sources are separated by ", " (comma+space)
for source in srcset.split(", ") {
let trimmed = source.trim();
if trimmed.is_empty() {
continue;
}
// Split by whitespace to separate URL from size descriptor
let url_part = trimmed.split_whitespace().next().unwrap_or("");
let url_trimmed = url_part.trim().to_string();
if !url_trimmed.is_empty() && !srcset_urls.contains(&url_trimmed) {
srcset_urls.push(url_trimmed);
}
}
}
let srcset_refs: Vec<&str> = srcset_urls.iter().map(|s| s.as_str()).collect();
found_urls.add_urls_from_text_array(&srcset_refs, &source_url_str, UrlSource::ImgSrcset);
}
/// Find audio URLs
fn find_audio(&self, html: &str, source_url: &ParsedUrl, found_urls: &mut FoundUrls) {
let source_url_str = source_url.get_full_url(true, false);
let urls: Vec<&str> = RE_AUDIO_SRC
.captures_iter(html)
.filter_map(|caps| caps.get(1).map(|m| m.as_str()))
.collect();
found_urls.add_urls_from_text_array(&urls, &source_url_str, UrlSource::AudioSrc);
}
/// Find video URLs
fn find_video(&self, html: &str, source_url: &ParsedUrl, found_urls: &mut FoundUrls) {
let source_url_str = source_url.get_full_url(true, false);
let urls: Vec<&str> = RE_VIDEO_SRC
.captures_iter(html)
.filter_map(|caps| caps.get(1).map(|m| m.as_str()))
.collect();
found_urls.add_urls_from_text_array(&urls, &source_url_str, UrlSource::VideoSrc);
}
/// Find script URLs from <script src>, <link href=".js">, .src= assignments, and NextJS chunks
fn find_scripts(&self, html: &str, source_url: &ParsedUrl, found_urls: &mut FoundUrls) {
let source_url_str = source_url.get_full_url(true, false);
// <script src="..."
let script_srcs: Vec<&str> = RE_SCRIPT_SRC
.captures_iter(html)
.filter_map(|caps| caps.get(1).map(|m| m.as_str()))
.collect();
found_urls.add_urls_from_text_array(&script_srcs, &source_url_str, UrlSource::ScriptSrc);
// <link href="...(json|js)"
let link_js: Vec<&str> = RE_LINK_JS
.captures_iter(html)
.filter_map(|caps| caps.get(1).map(|m| m.as_str()))
.collect();
found_urls.add_urls_from_text_array(&link_js, &source_url_str, UrlSource::LinkHref);
// .src = "..." (lazy loading in JS)
let dot_srcs: Vec<&str> = RE_DOT_SRC
.captures_iter(html)
.filter_map(|caps| caps.get(1).map(|m| m.as_str()))
.collect();
found_urls.add_urls_from_text_array(&dot_srcs, &source_url_str, UrlSource::InlineScriptSrc);
// NextJS chunks
let mut next_js_chunks: Vec<String> = Vec::new();
for caps in RE_NEXTJS_CHUNKS.captures_iter(html) {
if let Some(m) = caps.get(1) {
let matched = m.as_str();
let chunk_url = if matched.starts_with("//") {
format!("{}:{}", source_url.scheme.as_deref().unwrap_or("https"), matched)
} else if matched.starts_with("http://") || matched.starts_with("https://") {
matched.to_string()
} else if matched.contains("/_next/") {
let mut url = matched.to_string();
if source_url.host.is_some() && source_url.host != self.config.initial_url.host {
url = format!("{}{}", source_url.get_full_homepage_url(), url);
}
url
} else {
format!("{}/_next/{}", source_url.get_full_homepage_url(), matched)
};
next_js_chunks.push(chunk_url);
}
}
let chunk_refs: Vec<&str> = next_js_chunks.iter().map(|s| s.as_str()).collect();
found_urls.add_urls_from_text_array(&chunk_refs, &source_url_str, UrlSource::InlineScriptSrc);
}
/// Find stylesheet URLs from <link> tags with rel="stylesheet"
fn find_stylesheets(&self, html: &str, source_url: &ParsedUrl, found_urls: &mut FoundUrls) {
let source_url_str = source_url.get_full_url(true, false);
let mut stylesheet_urls: Vec<String> = Vec::new();
for caps in RE_LINK_STYLESHEET.captures_iter(html) {
let full_match = caps.get(0).map(|m| m.as_str()).unwrap_or("");
if let Some(href) = caps.get(1) {
// Only include if no rel= attribute or rel="stylesheet"
let full_lower = full_match.to_lowercase();
if !full_lower.contains("rel=") || full_lower.contains("stylesheet") {
stylesheet_urls.push(href.as_str().to_string());
}
}
}
let url_refs: Vec<&str> = stylesheet_urls.iter().map(|s| s.as_str()).collect();
found_urls.add_urls_from_text_array(&url_refs, &source_url_str, UrlSource::LinkHref);
}
/// Remove all unwanted code from HTML with respect to --disable-* options
fn remove_unwanted_code_from_html(&self, html: &str) -> String {
let mut result = html.to_string();
if !self.config.scripts_enabled {
result = utils::strip_javascript(&result);
}
if !self.config.styles_enabled {
result = utils::strip_styles(&result);
}
if !self.config.fonts_enabled {
result = utils::strip_fonts(&result);
}
if !self.config.images_enabled && result.to_lowercase().contains("<img") {
result = utils::strip_images(&result, None);
result = self.set_custom_css_for_tile_images(&result);
result = utils::add_class_to_html_images(&result, "siteone-crawler-bg");
}
result
}
/// Add custom CSS for placeholder tile images
fn set_custom_css_for_tile_images(&self, html: &str) -> String {
let background_base64 = "iVBORw0KGgoAAAANSUhEUgAAAEAAAAAkCAMAAAAO0sygAAAAAXNSR0IB2cksfwAAAAlwSFlzAAALEwAACxMBAJqcGAAAAMlQTFRFFxcXwMDA////1NTU5+fnpaWl0tLSIyMj5ubmlJSUxcXF29vbz8/P9PT01tbWxMTE8fHxaWlp39/f9fX1yMjI3NzciYmJeXl5Gxsb2dnZNTU18/PzXFxc5eXlJycnysrKZGRk6enp3d3dW1tbsrKyWFhYIiIi19fXvLy8w8PDuLi47e3tzMzM0dHRx8fH09PTHR0dzs7OLy8vwcHB0NDQSEhIqamp4uLiHh4eOzs74ODg3t7ewsLCISEhJCQkaGhoy8vLzc3N2NjYEPdgjAAAAaRJREFUeJzdlWlTgzAQhiGlWlsCCHog9vAWpalavK3X//9RJptACoEwTp1x9MPOht19k+VJCIZhIvNXzVh1DmPVHowfe5cOsrpr5dh6D1kb/VJs0LWQ3cAAO7gyp0tjXinGavBmPQOf5gaVvgIaC5eet5jeceoZbNPcjhjvcu9GGHl7sjYGfbXP3HyaG/Dx/pD7UYDwOFT0ThuDSYwOahgcCn0rgyNWZykMQNtpYXBM9+wkhtreKZ8zZ8D52RoGEeT6E9EntTPmzxP5/mEIXscAX8SF/hL6TSEHc6VTRGbNDMyrYaElRPRxncr1bVpzEzczyHugNjeIGHO9ydbNMjomGgbUUq5PlDMzp3p5FpoYmLei77sERUJ//yBy0wy8lkFf8s/XV/rVMXjk9VahnYF/KvWrY/AMuZdFrvdQCHtPSnVt52BOfa/YP6LcBzoGfj73K1s/q70PtAzYt/DGx8P3Dx6r3Ad6Br6ce2GLmLwPknGAgigAPfNBFDffB6WzKRiMvGJfK/urMlgyycBV9FjDoLAlBl5V/9nwPXzb/tO/8e8y+AJh0S3ETlwQiAAAAABJRU5ErkJggg==";
RE_CLOSE_HEAD
.replace(
html,
&format!(
"<style>\n\
.siteone-crawler-bg {{\n\
background-image: url(\"data:image/png;base64,{}\");\n\
background-repeat: repeat;\n\
opacity: 0.15;\n\
}}\n\
</style></head>",
background_base64
),
)
.to_string()
}
/// Set JS variable _SiteOneUrlDepth with number of levels before </head>
fn set_js_variable_with_url_depth(&self, html: &str, base_url: &str) -> String {
let base_path = if let Ok(parsed) = url::Url::parse(base_url) {
parsed.path().to_string()
} else {
"/".to_string()
};
let trimmed = base_path.trim_start_matches('/');
let mut depth = trimmed.matches('/').count();
let needs_index_html = base_path != "/" && base_path.ends_with('/');
if needs_index_html {
depth += 1;
}
RE_CLOSE_HEAD
.replace(
html,
&format!(
"<script>var {} = {};</script></head>",
JS_VARIABLE_NAME_URL_DEPTH, depth
),
)
.to_string()
}
/// Set JS function to remove all anchor listeners before </body>
fn set_js_function_to_remove_all_anchor_listeners(&self, html: &str) -> String {
RE_CLOSE_BODY
.replace(
html,
concat!(
"<script>\n",
"function _SiteOneRemoveAllAnchorListeners(){\n",
" var anchors=document.getElementsByTagName('a');\n",
" for(var i=0;i<anchors.length;i++){\n",
" var anchor=anchors[i];\n",
" var newAnchor=anchor.cloneNode(true);\n",
" anchor.parentNode.replaceChild(newAnchor,anchor);\n",
" }\n",
"}\n",
"setTimeout(_SiteOneRemoveAllAnchorListeners, 200);\n",
"setTimeout(_SiteOneRemoveAllAnchorListeners, 1000);\n",
"setTimeout(_SiteOneRemoveAllAnchorListeners, 5000);\n",
"</script></body>",
),
)
.to_string()
}
/// Remove scheme and host from full origin URLs to simplify relative paths conversion
fn remove_schema_and_host_from_full_origin_urls(&self, url: &ParsedUrl, content: &str) -> String {
static RE_BASE_URL_ROOT: Lazy<Regex> = Lazy::new(|| Regex::new(r"(?i)((https?:)?//[^/]+/?).*").unwrap());
let full_url = url.get_full_url(true, false);
// Extract base URL root (scheme://host/)
let base_url_root = RE_BASE_URL_ROOT.replace(&full_url, "$1").to_string();
let mut result = content.to_string();
// Normalize port numbers
result = RE_PORT_NORMALIZE.replace_all(&result, "$1").to_string();
// Build patterns for href=, src=, url=, url( attributes
let escaped_root = regex::escape(&base_url_root);
let attr_patterns = [
format!(r#"(?i)(href=(["'])){esc}([^"']*)(["'])"#, esc = escaped_root),
format!(r#"(?i)(src=(["'])){esc}([^"']*)(["'])"#, esc = escaped_root),
format!(r#"(?i)(url=(["'])){esc}([^"']*)(["'])"#, esc = escaped_root),
format!(r#"(?i)(url\((["']?)){esc}([^"')]*)(["']\)|\))"#, esc = escaped_root),
];
for pattern in &attr_patterns {
if let Ok(re) = Regex::new(pattern) {
let compiled_ignore = &self.config.compiled_ignore_regex;
result = re
.replace_all(&result, |caps: ®ex::Captures| {
let full_match = caps.get(0).map_or("", |m| m.as_str());
// Check against pre-compiled ignore patterns
for ire in compiled_ignore {
if ire.is_match(full_match) {
return full_match.to_string();
}
}
let attr_start = caps.get(1).map_or("", |m| m.as_str());
let path = caps.get(3).map_or("", |m| m.as_str());
let attr_end = caps.get(4).map_or("", |m| m.as_str());
format!("{}/{}{}", attr_start, path, attr_end)
})
.to_string();
}
}
result
}
/// Update all HTML paths to relative for offline version
fn update_html_paths_to_relative(&self, html: &str, parsed_base_url: &ParsedUrl) -> String {
let initial_url = &self.config.initial_url;
let compiled_ignore = &self.config.compiled_ignore_regex;
let replace_callback = |caps: ®ex::Captures| -> String {
let full_match = caps.get(0).map_or("", |m| m.as_str());
let start = caps.get(1).map_or("", |m| m.as_str());
let attribute_raw = caps.get(2).map_or("", |m| m.as_str());
let attribute = attribute_raw.trim_matches(|c: char| c == ' ' || c == '\\' || c == '"' || c == '\'');
let assignment_char = caps.get(3).map_or("", |m| m.as_str());
let quote = caps.get(4).map_or("", |m| m.as_str());
// Decode HTML entities in URL values (fixes Astro image query params like & → &)
let value_raw = caps.get(5).map_or("", |m| m.as_str());
let value_decoded = html_entity_decode(value_raw);
let value = value_decoded.as_str();
let end = caps.get(6).map_or("", |m| m.as_str());
// When modifying x.src (JS) and there is no quote, do not convert
if start == "." && quote.is_empty() {
return full_match.to_string();
}
// Ignore data URI, anchor, or non-http scheme
if value.starts_with('#') || RE_NON_HTTP_SCHEME.is_match(value) {
return full_match.to_string();
}
// Check against pre-compiled ignore regex patterns
for ire in compiled_ignore {
if ire.is_match(value) {
return full_match.to_string();
}
}
let attr_lower = attribute.to_lowercase();
let new_value = if attr_lower == "srcset" || attr_lower == "imagesrcset" {
// Handle srcset: multiple sources separated by ", "
let sources: Vec<&str> = value.split(", ").collect();
let converted: Vec<String> = sources
.iter()
.map(|source| {
let trimmed = source.trim();
if !trimmed.contains(' ') {
// URL without size descriptor
convert_url_to_relative(parsed_base_url, trimmed, initial_url, Some(&attr_lower))
} else {
// URL with size descriptor (e.g., "url 2x")
let mut parts = trimmed.splitn(2, char::is_whitespace);
let url_part = parts.next().unwrap_or("");
let size_part = parts.next().unwrap_or("");
let relative_url =
convert_url_to_relative(parsed_base_url, url_part, initial_url, Some(&attr_lower));
format!("{} {}", relative_url, size_part)
}
})
.collect();
converted.join(", ")
} else {
let mut converted = convert_url_to_relative(parsed_base_url, value, initial_url, Some(attribute));
// Handle component-url and renderer-url (Astro)
if attribute == "component-url" || attribute == "renderer-url" {
converted = format!("./{}", converted);
}
converted
};
format!(
"{}{}{}{}{}{}{}",
start, attribute_raw, assignment_char, quote, new_value, quote, end
)
};
let mut result = html.to_string();
result = RE_HREF_SRC.replace_all(&result, replace_callback).to_string();
result = RE_SRCSET_ATTR.replace_all(&result, replace_callback).to_string();
result = RE_META_URL.replace_all(&result, replace_callback).to_string();
result = RE_ESCAPED_HREF_SRC.replace_all(&result, replace_callback).to_string();
result
}
/// Apply specific HTML changes for offline version
#[allow(clippy::too_many_arguments)]
fn apply_specific_html_changes(
&self,
html: &mut String,
parsed_base_url: &ParsedUrl,
remove_external_js: bool,
remove_cross_origins: bool,
remove_analytics: bool,
remove_socnets: bool,
remove_cookies_related: bool,
) {
if html.trim().is_empty() {
return;
}
let base_host = parsed_base_url.host.as_deref().unwrap_or("");
// Remove external JS
if remove_external_js {
let base_host_owned = base_host.to_string();
*html = RE_EXTERNAL_SCRIPT
.replace_all(html, |caps: ®ex::Captures| {
let full_match = caps.get(0).map_or("", |m| m.as_str());
let src = caps.get(1).map_or("", |m| m.as_str());
if RE_EXTERNAL_URL.is_match(src) {
// Parse host from the src URL
let parsed_src = if src.starts_with("//") {
format!("https:{}", src)
} else {
src.to_string()
};
if let Ok(parsed) = url::Url::parse(&parsed_src)
&& parsed.host_str().unwrap_or("") != base_host_owned
{
return String::new();
}
}
full_match.to_string()
})
.to_string();
}
// Remove crossorigin attributes
if remove_cross_origins {
*html = RE_CROSSORIGIN_LINK.replace_all(html, "$1$3").to_string();
*html = RE_CROSSORIGIN_SCRIPT.replace_all(html, "$1$3").to_string();
}
// Remove analytics and social network scripts
if remove_analytics || remove_socnets || remove_cookies_related {
let mut patterns: Vec<&str> = Vec::new();
if remove_analytics {
patterns.extend_from_slice(&[
"googletagmanager.com",
"google-analytics.com",
"ga.js",
"gtag.js",
"gtag(",
"analytics.",
"connect.facebook.net",
"fbq(",
]);
}
if remove_socnets {
patterns.extend_from_slice(&[
"connect.facebook.net",
"connect.facebook.com",
"twitter.com",
".x.com",
"linkedin.com",
"instagram.com",
"pinterest.com",
"tumblr.com",
"plus.google.com",
"curator.io",
]);
}
if remove_cookies_related {
patterns.extend_from_slice(&["cookies", "cookiebot"]);
}
// Deduplicate
patterns.sort();
patterns.dedup();
*html = RE_SCRIPT_BLOCK
.replace_all(html, |caps: ®ex::Captures| {
let full_match = caps.get(0).map_or("", |m| m.as_str());
let full_lower = full_match.to_lowercase();
for keyword in &patterns {
if full_lower.contains(&keyword.to_lowercase()) {
return String::new();
}
}
full_match.to_string()
})
.to_string();
// Remove social network iframes
if remove_socnets {
*html = RE_SOCNET_IFRAME.replace_all(html, "").to_string();
}
}
}
/// Check if anchor listener removal is forced (e.g., for NextJS sites)
fn is_forced_to_remove_anchor_listeners(&self, html: &str) -> bool {
html.contains("_next/")
}
}
impl ContentProcessor for HtmlProcessor {
fn find_urls(&self, content: &str, source_url: &ParsedUrl) -> Option<FoundUrls> {
let mut found_urls = FoundUrls::new();
if !self.config.single_page {
self.find_href_urls(content, source_url, &mut found_urls);
}
if self.config.fonts_enabled {
self.find_fonts(content, source_url, &mut found_urls);
}
if self.config.images_enabled {
self.find_images(content, source_url, &mut found_urls);
}
if self.config.files_enabled {
self.find_audio(content, source_url, &mut found_urls);
self.find_video(content, source_url, &mut found_urls);
}
if self.config.scripts_enabled {
self.find_scripts(content, source_url, &mut found_urls);
}
if self.config.styles_enabled {
self.find_stylesheets(content, source_url, &mut found_urls);
}
if found_urls.get_count() > 0 {
Some(found_urls)
} else {
None
}
}
fn apply_content_changes_before_url_parsing(
&self,
_content: &mut String,
_content_type: ContentTypeId,
_url: &ParsedUrl,
) {
// No changes needed before URL parsing in HtmlProcessor
}
fn apply_content_changes_for_offline_version(
&self,
content: &mut String,
_content_type: ContentTypeId,
url: &ParsedUrl,
remove_unwanted_code: bool,
) {
let base_url = url.get_full_url(true, false);
// Remove schema and host from full origin URLs
*content = self.remove_schema_and_host_from_full_origin_urls(url, content);
// Remove unwanted code from HTML
*content = self.remove_unwanted_code_from_html(content);
// Update all paths to relative
*content = self.update_html_paths_to_relative(content, url);
// Meta redirects (e.g., in Astro projects)
if let Some(caps) = RE_META_REFRESH.captures(content) {
let full_match = caps.get(0).map_or("", |m| m.as_str());
let prefix = caps.get(1).map_or("", |m| m.as_str());
let meta_url = caps.get(2).map_or("", |m| m.as_str());
let suffix = caps.get(3).map_or("", |m| m.as_str());
let relative = convert_url_to_relative(url, meta_url, &self.config.initial_url, None);
*content = content.replace(full_match, &format!("{}{}{}", prefix, relative, suffix));
}
// Apply specific HTML changes
self.apply_specific_html_changes(
content,
url,
self.config.disable_javascript,
remove_unwanted_code,
remove_unwanted_code,
remove_unwanted_code,
remove_unwanted_code,
);
// Set JS variable and remove anchor listeners
if self.config.scripts_enabled {
*content = self.set_js_variable_with_url_depth(content, &base_url);
if self.config.remove_all_anchor_listeners || self.is_forced_to_remove_anchor_listeners(content) {
*content = self.set_js_function_to_remove_all_anchor_listeners(content);
}
}
}
fn is_content_type_relevant(&self, content_type: ContentTypeId) -> bool {
is_relevant(content_type, &self.relevant_content_types)
}
fn get_name(&self) -> &str {
"HtmlProcessor"
}
fn set_debug_mode(&mut self, debug_mode: bool) {
self.debug_mode = debug_mode;
}
}
/// Decode HTML entities in URL attribute values.
/// Single-pass implementation to avoid double-decoding (e.g. `&amp;` → `&`, not `&`).
fn html_entity_decode(input: &str) -> String {
let mut result = String::with_capacity(input.len());
let bytes = input.as_bytes();
let len = bytes.len();
let mut i = 0;
while i < len {
if bytes[i] == b'&' {
// Try to match a named or numeric entity
if let Some((decoded, advance)) = try_decode_entity(&input[i..]) {
result.push_str(decoded);
i += advance;
continue;
}
}
result.push(input[i..].chars().next().unwrap());
i += input[i..].chars().next().unwrap().len_utf8();
}
result
}
/// Try to decode a single HTML entity at the start of `s`. Returns (decoded, bytes_consumed).
fn try_decode_entity(s: &str) -> Option<(&'static str, usize)> {
// Named entities
for (entity, decoded) in [
("&", "&"),
("<", "<"),
(">", ">"),
(""", "\""),
("'", "'"),
] {
if s.starts_with(entity) {
return Some((decoded, entity.len()));
}
}
// Numeric entities (decimal and hex)
for (entity, decoded) in [
("&", "&"),
("&", "&"),
("<", "<"),
("<", "<"),
("<", "<"),
(">", ">"),
(">", ">"),
(">", ">"),
(""", "\""),
(""", "\""),
("'", "'"),
("'", "'"),
("'", "'"),
] {
if s.starts_with(entity) {
return Some((decoded, entity.len()));
}
}
None
}
#[cfg(test)]
mod tests {
use super::*;
fn make_config() -> ProcessorConfig {
ProcessorConfig::new(ParsedUrl::parse("https://example.com/", None))
}
#[test]
fn test_find_href_urls() {
let processor = HtmlProcessor::new(make_config());
let html = r#"<html><body><a href="/about">About</a><a href="/contact">Contact</a></body></html>"#;
let source = ParsedUrl::parse("https://example.com/", None);
let result = processor.find_urls(html, &source);
assert!(result.is_some());
let found = result.unwrap();
assert!(found.get_count() >= 2);
}
#[test]
fn test_find_images() {
let processor = HtmlProcessor::new(make_config());
let html = r#"<html><body><img src="/img/logo.png"><img data-src="/img/lazy.jpg"></body></html>"#;
let source = ParsedUrl::parse("https://example.com/", None);
let result = processor.find_urls(html, &source);
assert!(result.is_some());
}
#[test]
fn test_find_scripts() {
let processor = HtmlProcessor::new(make_config());
let html = r#"<html><head><script src="/js/app.js"></script></head></html>"#;
let source = ParsedUrl::parse("https://example.com/", None);
let result = processor.find_urls(html, &source);
assert!(result.is_some());
}
#[test]
fn test_single_page_no_hrefs() {
let mut config = make_config();
config.single_page = true;
let processor = HtmlProcessor::new(config);
let html = r#"<html><body><a href="/about">About</a><script src="/app.js"></script></body></html>"#;
let source = ParsedUrl::parse("https://example.com/", None);
let result = processor.find_urls(html, &source);
assert!(result.is_some());
// Should only find script, not href
let found = result.unwrap();
for (_key, url) in found.get_urls() {
assert_ne!(url.url, "/about");
}
}
#[test]
fn test_find_srcset() {
let processor = HtmlProcessor::new(make_config());
let html = r#"<img srcset="/img/small.jpg 1x, /img/large.jpg 2x">"#;
let source = ParsedUrl::parse("https://example.com/", None);
let result = processor.find_urls(html, &source);
assert!(result.is_some());
}
}