add option to exclude web fonts

README.md

@@ -49,6 +49,7 @@ The guide can be found [here](docs/containers.md)
 ## Options
  - `-c`: Ignore styles
  - `-f`: Exclude frames and iframes
+ - `-F`: Omit web fonts
  - `-i`: Remove images
  - `-I`: Isolate the document
  - `-j`: Exclude JavaScript

src/args.rs

@@ -4,6 +4,7 @@ use clap::{App, Arg};
 pub struct AppArgs {
     pub target: String,
     pub no_css: bool,
+    pub no_fonts: bool,
     pub no_frames: bool,
     pub no_images: bool,
     pub no_js: bool,
@@ -35,6 +36,7 @@ impl AppArgs {
             // .args_from_usage("-a, --include-audio 'Removes audio sources'")
             .args_from_usage("-c, --no-css 'Removes CSS'")
             .args_from_usage("-f, --no-frames 'Removes frames and iframes'")
+            .args_from_usage("-F, --no-fonts 'Removes fonts'")
             .args_from_usage("-i, --no-images 'Removes images'")
             .args_from_usage("-I, --isolate 'Cuts off document from the Internet'")
             .args_from_usage("-j, --no-js 'Removes JavaScript'")
@@ -52,6 +54,7 @@ impl AppArgs {
             .expect("please set target")
             .to_string();
         app_args.no_css = app.is_present("no-css");
+        app_args.no_fonts = app.is_present("no-fonts");
         app_args.no_frames = app.is_present("no-frames");
         app_args.no_images = app.is_present("no-images");
         app_args.no_js = app.is_present("no-js");

src/css.rs

@@ -61,6 +61,7 @@ pub fn process_css<'a>(
     rule_name: &str,
     prop_name: &str,
     func_name: &str,
+    opt_no_fonts: bool,
     opt_no_images: bool,
     opt_silent: bool,
 ) -> Result<String, ParseError<'a, String>> {
@@ -89,6 +90,10 @@ pub fn process_css<'a>(
             Token::Colon => result.push_str(":"),
             Token::Comma => result.push_str(","),
             Token::ParenthesisBlock | Token::SquareBracketBlock | Token::CurlyBracketBlock => {
+                if opt_no_fonts && curr_rule == "font-face" {
+                    continue;
+                }
+
                 let closure: &str;
                 if token == &Token::ParenthesisBlock {
                     result.push_str("(");
@@ -111,6 +116,7 @@ pub fn process_css<'a>(
                             rule_name,
                             curr_prop.as_str(),
                             func_name,
+                            opt_no_fonts,
                             opt_no_images,
                             opt_silent,
                         )
@@ -133,12 +139,18 @@ pub fn process_css<'a>(
             Token::WhiteSpace(ref value) => {
                 result.push_str(value);
             }
+            // div...
             Token::Ident(ref value) => {
+                curr_rule = str!();
                 curr_prop = str!(value);
                 result.push_str(&escape(value));
             }
+            // @import, @font-face, @charset, @media...
             Token::AtKeyword(ref value) => {
                 curr_rule = str!(value);
+                if opt_no_fonts && curr_rule == "font-face" {
+                    continue;
+                }
                 result.push_str("@");
                 result.push_str(value);
             }
@@ -147,13 +159,10 @@ pub fn process_css<'a>(
                 result.push_str(value);
             }
             Token::QuotedString(ref value) => {
-                let is_import: bool = curr_rule == "import";
-                if is_import {
+                if curr_rule == "import" {
                     // Reset current at-rule value
                     curr_rule = str!();
-                }
 
-                if is_import {
                     // Skip empty import values
                     if value.len() < 1 {
                         result.push_str("''");
@@ -182,6 +191,7 @@ pub fn process_css<'a>(
                                     client,
                                     final_url.as_str(),
                                     &css,
+                                    opt_no_fonts,
                                     opt_no_images,
                                     opt_silent,
                                 )
@@ -254,12 +264,15 @@ pub fn process_css<'a>(
                 result.push_str(str!(value).as_str());
                 result.push_str(str!(unit).as_str());
             }
+            // #selector, #id...
             Token::IDHash(ref value) => {
+                curr_rule = str!();
                 result.push_str("#");
                 result.push_str(&escape(value));
             }
             Token::UnquotedUrl(ref value) => {
                 let is_import: bool = curr_rule == "import";
+
                 if is_import {
                     // Reset current at-rule value
                     curr_rule = str!();
@@ -300,6 +313,7 @@ pub fn process_css<'a>(
                                     client,
                                     final_url.as_str(),
                                     &css,
+                                    opt_no_fonts,
                                     opt_no_images,
                                     opt_silent,
                                 )
@@ -347,6 +361,7 @@ pub fn process_css<'a>(
                             curr_rule.as_str(),
                             curr_prop.as_str(),
                             function_name,
+                            opt_no_fonts,
                             opt_no_images,
                             opt_silent,
                         )
@@ -368,6 +383,7 @@ pub fn embed_css(
     client: &Client,
     parent_url: &str,
     css: &str,
+    opt_no_fonts: bool,
     opt_no_images: bool,
     opt_silent: bool,
 ) -> String {
@@ -382,6 +398,7 @@ pub fn embed_css(
         "",
         "",
         "",
+        opt_no_fonts,
         opt_no_images,
         opt_silent,
     )

src/html.rs

@@ -42,10 +42,11 @@ pub fn walk_and_embed_assets(
     url: &str,
     node: &Handle,
     opt_no_css: bool,
+    opt_no_fonts: bool,
+    opt_no_frames: bool,
     opt_no_js: bool,
     opt_no_images: bool,
     opt_silent: bool,
-    opt_no_frames: bool,
 ) {
     match node.data {
         NodeData::Document => {
@@ -57,10 +58,11 @@ pub fn walk_and_embed_assets(
                     &url,
                     child,
                     opt_no_css,
+                    opt_no_fonts,
+                    opt_no_frames,
                     opt_no_js,
                     opt_no_images,
                     opt_silent,
-                    opt_no_frames,
                 );
             }
         }
@@ -157,17 +159,18 @@ pub fn walk_and_embed_assets(
                                         ) {
                                             // On successful retrieval, traverse CSS
                                             Ok((css_data, final_url)) => {
-                                                let x: String = embed_css(
+                                                let css: String = embed_css(
                                                     cache,
                                                     client,
                                                     &final_url,
                                                     &css_data,
+                                                    opt_no_fonts,
                                                     opt_no_images,
                                                     opt_silent,
                                                 );
                                                 data_to_data_url(
                                                     "text/css",
-                                                    x.as_bytes(),
+                                                    css.as_bytes(),
                                                     &final_url,
                                                     "",
                                                 )
@@ -462,6 +465,7 @@ pub fn walk_and_embed_assets(
                                     client,
                                     &url,
                                     tendril.as_ref(),
+                                    opt_no_fonts,
                                     opt_no_images,
                                     opt_silent,
                                 );
@@ -519,10 +523,11 @@ pub fn walk_and_embed_assets(
                                 &frame_final_url,
                                 &dom.document,
                                 opt_no_css,
+                                opt_no_fonts,
+                                opt_no_frames,
                                 opt_no_js,
                                 opt_no_images,
                                 opt_silent,
-                                opt_no_frames,
                             );
                             let mut buf: Vec<u8> = Vec::new();
                             serialize(&mut buf, &dom.document, SerializeOpts::default()).unwrap();
@@ -590,6 +595,7 @@ pub fn walk_and_embed_assets(
                         client,
                         &url,
                         attribute.value.as_ref(),
+                        opt_no_fonts,
                         opt_no_images,
                         opt_silent,
                     );
@@ -621,10 +627,11 @@ pub fn walk_and_embed_assets(
                     &url,
                     child,
                     opt_no_css,
+                    opt_no_fonts,
+                    opt_no_frames,
                     opt_no_js,
                     opt_no_images,
                     opt_silent,
-                    opt_no_frames,
                 );
             }
         }

src/main.rs

@@ -139,10 +139,11 @@ fn main() {
         &base_url,
         &dom.document,
         app_args.no_css,
+        app_args.no_fonts,
+        app_args.no_frames,
         app_args.no_js,
         app_args.no_images,
         app_args.silent,
-        app_args.no_frames,
     );
 
     let html: String = stringify_document(

src/tests/css/embed_css.rs

@@ -15,7 +15,10 @@ fn passing_empty_input() {
     let cache = &mut HashMap::new();
     let client = Client::new();
 
-    assert_eq!(css::embed_css(cache, &client, "", "", false, false,), "");
+    assert_eq!(
+        css::embed_css(cache, &client, "", "", false, false, false,),
+        ""
+    );
 }
 
 #[test]
@@ -37,6 +40,7 @@ height: calc(100vh - 10pt)";
             &client,
             "https://doesntmatter.local/",
             &STYLE,
+            false,
             true,
             true,
         ),
@@ -67,7 +71,7 @@ line-height: -1; \
 height: calc(100vh - 10pt)";
 
     assert_eq!(
-        css::embed_css(cache, &client, "", &STYLE, true, true,),
+        css::embed_css(cache, &client, "", &STYLE, false, true, true,),
         format!(
             "/* border: none;*/\
 background-image: url('{empty_image}'); \
@@ -95,7 +99,7 @@ fn passing_style_block() {
 html > body {}";
 
     assert_eq!(
-        css::embed_css(cache, &client, "file:///", &CSS, false, true,),
+        css::embed_css(cache, &client, "file:///", &CSS, false, false, true,),
         CSS
     );
 }
@@ -135,7 +139,10 @@ fn passing_attribute_selectors() {
 }
 ";
 
-    assert_eq!(css::embed_css(cache, &client, "", &CSS, false, false,), CSS);
+    assert_eq!(
+        css::embed_css(cache, &client, "", &CSS, false, false, false,),
+        CSS
+    );
 }
 
 #[test]
@@ -158,6 +165,7 @@ fn passing_import_string() {
             "https://doesntmatter.local/",
             &CSS,
             false,
+            false,
             true,
         ),
         "\
@@ -192,6 +200,7 @@ body {\n    \
             "https://doesntmatter.local/",
             &CSS,
             false,
+            false,
             true,
         ),
         CSS
@@ -218,6 +227,7 @@ div {\n    \
             "https://doesntmatter.local/",
             &CSS,
             false,
+            false,
             true,
         ),
         CSS
@@ -246,8 +256,62 @@ fn passing_unusual_indents() {
             "https://doesntmatter.local/",
             &CSS,
             false,
+            false,
             true,
         ),
         CSS
     );
 }
+
+#[test]
+fn passing_exclude_fonts() {
+    let cache = &mut HashMap::new();
+    let client = Client::new();
+
+    const CSS: &str = "\
+@font-face {\n    \
+    font-family: 'My Font';\n    \
+    src: url(my_font.woff);\n\
+}\n\
+\n\
+#identifier {\n    \
+    font-family: 'My Font' Arial\n\
+}\n\
+\n\
+@font-face {\n    \
+    font-family: 'My Font';\n    \
+    src: url(my_font.woff);\n\
+}\n\
+\n\
+div {\n    \
+    font-family: 'My Font' Verdana\n\
+}\n\
+";
+
+    const CSS_OUT: &str = " \
+\n\
+\n\
+#identifier {\n    \
+    font-family: 'My Font' Arial\n\
+}\n\
+\n \
+\n\
+\n\
+div {\n    \
+    font-family: 'My Font' Verdana\n\
+}\n\
+";
+
+    assert_eq!(
+        css::embed_css(
+            cache,
+            &client,
+            "https://doesntmatter.local/",
+            &CSS,
+            true,
+            false,
+            true,
+        ),
+        CSS_OUT
+    );
+}

src/tests/html/walk_and_embed_assets.rs

@@ -19,6 +19,7 @@ fn passing_basic() {
     let url = "http://localhost";
 
     let opt_no_css: bool = false;
+    let opt_no_fonts: bool = false;
     let opt_no_frames: bool = false;
     let opt_no_js: bool = false;
     let opt_no_images: bool = false;
@@ -32,10 +33,11 @@ fn passing_basic() {
         &url,
         &dom.document,
         opt_no_css,
+        opt_no_fonts,
+        opt_no_frames,
         opt_no_js,
         opt_no_images,
         opt_silent,
-        opt_no_frames,
     );
 
     let mut buf: Vec<u8> = Vec::new();
@@ -55,6 +57,7 @@ fn passing_ensure_no_recursive_iframe() {
     let cache = &mut HashMap::new();
 
     let opt_no_css: bool = false;
+    let opt_no_fonts: bool = false;
     let opt_no_frames: bool = false;
     let opt_no_js: bool = false;
     let opt_no_images: bool = false;
@@ -68,10 +71,11 @@ fn passing_ensure_no_recursive_iframe() {
         &url,
         &dom.document,
         opt_no_css,
+        opt_no_fonts,
+        opt_no_frames,
         opt_no_js,
         opt_no_images,
         opt_silent,
-        opt_no_frames,
     );
 
     let mut buf: Vec<u8> = Vec::new();
@@ -91,6 +95,7 @@ fn passing_ensure_no_recursive_frame() {
     let cache = &mut HashMap::new();
 
     let opt_no_css: bool = false;
+    let opt_no_fonts: bool = false;
     let opt_no_frames: bool = false;
     let opt_no_js: bool = false;
     let opt_no_images: bool = false;
@@ -104,10 +109,11 @@ fn passing_ensure_no_recursive_frame() {
         &url,
         &dom.document,
         opt_no_css,
+        opt_no_fonts,
+        opt_no_frames,
         opt_no_js,
         opt_no_images,
         opt_silent,
-        opt_no_frames,
     );
 
     let mut buf: Vec<u8> = Vec::new();
@@ -129,6 +135,7 @@ fn passing_no_css() {
     let cache = &mut HashMap::new();
 
     let opt_no_css: bool = true;
+    let opt_no_fonts: bool = false;
     let opt_no_frames: bool = false;
     let opt_no_js: bool = false;
     let opt_no_images: bool = false;
@@ -141,10 +148,11 @@ fn passing_no_css() {
         &url,
         &dom.document,
         opt_no_css,
+        opt_no_fonts,
+        opt_no_frames,
         opt_no_js,
         opt_no_images,
         opt_silent,
-        opt_no_frames,
     );
 
     let mut buf: Vec<u8> = Vec::new();
@@ -173,6 +181,7 @@ fn passing_no_images() {
     let cache = &mut HashMap::new();
 
     let opt_no_css: bool = false;
+    let opt_no_fonts: bool = false;
     let opt_no_frames: bool = false;
     let opt_no_js: bool = false;
     let opt_no_images: bool = true;
@@ -186,10 +195,11 @@ fn passing_no_images() {
         &url,
         &dom.document,
         opt_no_css,
+        opt_no_fonts,
+        opt_no_frames,
         opt_no_js,
         opt_no_images,
         opt_silent,
-        opt_no_frames,
     );
 
     let mut buf: Vec<u8> = Vec::new();
@@ -221,6 +231,7 @@ fn passing_no_body_background_images() {
     let cache = &mut HashMap::new();
 
     let opt_no_css: bool = false;
+    let opt_no_fonts: bool = false;
     let opt_no_frames: bool = false;
     let opt_no_js: bool = false;
     let opt_no_images: bool = true;
@@ -234,10 +245,11 @@ fn passing_no_body_background_images() {
         &url,
         &dom.document,
         opt_no_css,
+        opt_no_fonts,
+        opt_no_frames,
         opt_no_js,
         opt_no_images,
         opt_silent,
-        opt_no_frames,
     );
 
     let mut buf: Vec<u8> = Vec::new();
@@ -257,6 +269,7 @@ fn passing_no_frames() {
     let cache = &mut HashMap::new();
 
     let opt_no_css: bool = false;
+    let opt_no_fonts: bool = false;
     let opt_no_frames: bool = true;
     let opt_no_js: bool = false;
     let opt_no_images: bool = false;
@@ -269,10 +282,11 @@ fn passing_no_frames() {
         &url,
         &dom.document,
         opt_no_css,
+        opt_no_fonts,
+        opt_no_frames,
         opt_no_js,
         opt_no_images,
         opt_silent,
-        opt_no_frames,
     );
 
     let mut buf: Vec<u8> = Vec::new();
@@ -292,6 +306,7 @@ fn passing_no_iframes() {
     let cache = &mut HashMap::new();
 
     let opt_no_css: bool = false;
+    let opt_no_fonts: bool = false;
     let opt_no_frames: bool = true;
     let opt_no_js: bool = false;
     let opt_no_images: bool = false;
@@ -304,10 +319,11 @@ fn passing_no_iframes() {
         &url,
         &dom.document,
         opt_no_css,
+        opt_no_fonts,
+        opt_no_frames,
         opt_no_js,
         opt_no_images,
         opt_silent,
-        opt_no_frames,
     );
 
     let mut buf: Vec<u8> = Vec::new();
@@ -330,6 +346,7 @@ fn passing_no_js() {
     let cache = &mut HashMap::new();
 
     let opt_no_css: bool = false;
+    let opt_no_fonts: bool = false;
     let opt_no_frames: bool = false;
     let opt_no_js: bool = true;
     let opt_no_images: bool = false;
@@ -343,10 +360,11 @@ fn passing_no_js() {
         &url,
         &dom.document,
         opt_no_css,
+        opt_no_fonts,
+        opt_no_frames,
         opt_no_js,
         opt_no_images,
         opt_silent,
-        opt_no_frames,
     );
 
     let mut buf: Vec<u8> = Vec::new();
@@ -369,6 +387,7 @@ fn passing_with_no_integrity() {
     let cache = &mut HashMap::new();
     let client = Client::new();
     let opt_no_css: bool = true;
+    let opt_no_fonts: bool = false;
     let opt_no_frames: bool = true;
     let opt_no_js: bool = true;
     let opt_no_images: bool = true;
@@ -380,10 +399,11 @@ fn passing_with_no_integrity() {
         &url,
         &dom.document,
         opt_no_css,
+        opt_no_fonts,
+        opt_no_frames,
         opt_no_js,
         opt_no_images,
         opt_silent,
-        opt_no_frames,
     );
 
     let mut buf: Vec<u8> = Vec::new();