Change downloading of images to be parallel

Signed-off-by: trivernis <trivernis@protonmail.com>

Cargo.lock

@@ -1167,7 +1167,7 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
 
 [[package]]
 name = "snekdown"
-version = "0.26.3"
+version = "0.26.4"
 dependencies = [
  "asciimath-rs 0.5.7 (registry+https://github.com/rust-lang/crates.io-index)",
  "base64 0.12.3 (registry+https://github.com/rust-lang/crates.io-index)",

Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "snekdown"
-version = "0.26.3"
+version = "0.26.4"
 authors = ["trivernis <trivernis@protonmail.com>"]
 edition = "2018"
 license-file = "LICENSE"

src/elements/mod.rs

@@ -4,14 +4,14 @@ use crate::format::PlaceholderTemplate;
 use crate::references::configuration::{ConfigRefEntry, Configuration, Value};
 use crate::references::placeholders::ProcessPlaceholders;
 use crate::references::templates::{Template, TemplateVariable};
+use crate::utils::downloads::{DownloadManager, PendingDownload};
 use asciimath_rs::elements::special::Expression;
 use bibliographix::bib_manager::BibManager;
 use bibliographix::bibliography::bibliography_entry::BibliographyEntryReference;
 use bibliographix::references::bib_reference::BibRefAnchor;
+use crossbeam_utils::sync::WaitGroup;
 use std::collections::HashMap;
-use std::fs::read;
 use std::iter::FromIterator;
-use std::path::PathBuf;
 use std::sync::atomic::{AtomicBool, Ordering};
 use std::sync::{Arc, Mutex, RwLock};
 
@@ -72,6 +72,7 @@ pub struct Document {
     pub(crate) placeholders: Vec<Arc<RwLock<Placeholder>>>,
     pub config: Configuration,
     pub bibliography: BibManager,
+    pub downloads: Arc<Mutex<DownloadManager>>,
     pub stylesheets: Vec<String>,
 }
 
@@ -232,6 +233,7 @@ pub struct Url {
 pub struct Image {
     pub(crate) url: Url,
     pub(crate) metadata: Option<InlineMetadata>,
+    pub(crate) download: Arc<Mutex<PendingDownload>>,
 }
 
 #[derive(Clone, Debug)]
@@ -297,10 +299,15 @@ impl Document {
             config: Configuration::default(),
             bibliography: BibManager::new(),
             stylesheets: Vec::new(),
+            downloads: Arc::new(Mutex::new(DownloadManager::new())),
         }
     }
 
-    pub fn new_with_manager(is_root: bool, bibliography: BibManager) -> Self {
+    pub fn new_with_manager(
+        is_root: bool,
+        bibliography: BibManager,
+        downloads: Arc<Mutex<DownloadManager>>,
+    ) -> Self {
         Self {
             elements: Vec::new(),
             is_root,
@@ -309,6 +316,7 @@ impl Document {
             config: Configuration::default(),
             bibliography,
             stylesheets: Vec::new(),
+            downloads,
         }
     }
 
@@ -680,28 +688,16 @@ impl Into<HashMap<String, Value>> for InlineMetadata {
 
 impl Image {
     pub fn get_content(&self) -> Option<Vec<u8>> {
-        let path = PathBuf::from(&self.url.url);
-        if path.exists() {
-            if let Ok(content) = read(path) {
-                Some(content)
-            } else {
-                None
-            }
-        } else {
-            self.download_content()
-        }
+        let mut download = self.download.lock().unwrap();
+        if let Some(wg) = &download.wg {
+            let wg = WaitGroup::clone(wg);
+            log::debug!("Waiting for content of {}", self.url.url.clone());
+            wg.wait();
         }
+        let mut data = None;
+        std::mem::swap(&mut data, &mut download.data);
 
-    fn download_content(&self) -> Option<Vec<u8>> {
-        if let Ok(content) = reqwest::blocking::get(&self.url.url) {
-            if let Ok(bytes) = content.bytes() {
-                Some(bytes.to_vec())
-            } else {
-                None
-            }
-        } else {
-            None
-        }
+        data
     }
 }
 

src/format/html/to_html.rs

@@ -7,6 +7,8 @@ use asciimath_rs::format::mathml::ToMathML;
 use htmlescape::encode_attribute;
 use minify::html::minify;
 use std::io;
+use std::sync::Arc;
+use std::thread;
 use syntect::highlighting::ThemeSet;
 use syntect::html::highlighted_html_for_string;
 use syntect::parsing::SyntaxSet;
@@ -95,6 +97,10 @@ impl ToHtml for MetadataValue {
 
 impl ToHtml for Document {
     fn to_html(&self, writer: &mut HTMLWriter) -> io::Result<()> {
+        let downloads = Arc::clone(&self.downloads);
+        thread::spawn(move || {
+            downloads.lock().unwrap().download_all();
+        });
         let path = if let Some(path) = &self.path {
             format!("path=\"{}\"", encode_attribute(path.as_str()))
         } else {

src/parser/inline.rs

@@ -132,7 +132,12 @@ impl ParseInline for Parser {
             } else {
                 None
             };
-            Ok(Image { url, metadata })
+            let path = url.url.clone();
+            Ok(Image {
+                url,
+                metadata,
+                download: self.document.downloads.lock().unwrap().add_download(path),
+            })
         } else {
             Err(self.ctm.rewind_with_error(start_index))
         }

src/parser/mod.rs

@@ -9,6 +9,7 @@ use crate::references::configuration::keys::{
     IMP_BIBLIOGRAPHY, IMP_CONFIGS, IMP_IGNORE, IMP_STYLESHEETS,
 };
 use crate::references::configuration::{Configuration, Value};
+use crate::utils::downloads::DownloadManager;
 use bibliographix::bib_manager::BibManager;
 use charred::tapemachine::{CharTapeMachine, TapeError, TapeResult};
 use crossbeam_utils::sync::WaitGroup;
@@ -57,6 +58,7 @@ impl Parser {
             false,
             Box::new(BufReader::new(f)),
             BibManager::new(),
+            Arc::new(Mutex::new(DownloadManager::new())),
         ))
     }
 
@@ -74,6 +76,7 @@ impl Parser {
             false,
             Box::new(Cursor::new(text_bytes.to_vec())),
             BibManager::new(),
+            Arc::new(Mutex::new(DownloadManager::new())),
         )
     }
 
@@ -83,6 +86,7 @@ impl Parser {
         path: PathBuf,
         paths: Arc<Mutex<Vec<PathBuf>>>,
         bib_manager: BibManager,
+        download_manager: Arc<Mutex<DownloadManager>>,
     ) -> Self {
         let text_bytes = text.as_bytes();
         Self::create(
@@ -91,6 +95,7 @@ impl Parser {
             true,
             Box::new(Cursor::new(text_bytes.to_vec())),
             bib_manager,
+            download_manager,
         )
     }
 
@@ -99,6 +104,7 @@ impl Parser {
         path: PathBuf,
         paths: Arc<Mutex<Vec<PathBuf>>>,
         bib_manager: BibManager,
+        download_manager: Arc<Mutex<DownloadManager>>,
     ) -> Result<Self, io::Error> {
         let f = File::open(&path)?;
         Ok(Self::create(
@@ -107,6 +113,7 @@ impl Parser {
             true,
             Box::new(BufReader::new(f)),
             bib_manager,
+            download_manager,
         ))
     }
 
@@ -116,6 +123,7 @@ impl Parser {
         is_child: bool,
         mut reader: Box<dyn BufRead>,
         bib_manager: BibManager,
+        download_manager: Arc<Mutex<DownloadManager>>,
     ) -> Self {
         if let Some(path) = path.clone() {
             paths.lock().unwrap().push(path.clone())
@@ -128,7 +136,7 @@ impl Parser {
             text.push('\n');
         }
 
-        let document = Document::new_with_manager(!is_child, bib_manager);
+        let document = Document::new_with_manager(!is_child, bib_manager, download_manager);
         Self {
             sections: Vec::new(),
             section_nesting: 0,
@@ -216,9 +224,11 @@ impl Parser {
         let paths = Arc::clone(&self.paths);
         let config = self.document.config.clone();
         let bibliography = self.document.bibliography.create_child();
+        let download_manager = Arc::clone(&self.document.downloads);
 
         let _ = thread::spawn(move || {
-            let mut parser = Parser::child_from_file(path, paths, bibliography).unwrap();
+            let mut parser =
+                Parser::child_from_file(path, paths, bibliography, download_manager).unwrap();
             parser.set_config(config);
             let document = parser.parse();
             anchor_clone.write().unwrap().set_document(document);

src/utils/downloads.rs

@@ -0,0 +1,84 @@
+use crossbeam_utils::sync::WaitGroup;
+use rayon::prelude::*;
+use std::fs::read;
+use std::path::PathBuf;
+use std::sync::{Arc, Mutex};
+
+/// A manager for downloading urls in parallel
+#[derive(Clone, Debug)]
+pub struct DownloadManager {
+    downloads: Vec<Arc<Mutex<PendingDownload>>>,
+}
+
+impl DownloadManager {
+    /// Creates a new download manager
+    pub fn new() -> Self {
+        Self {
+            downloads: Vec::new(),
+        }
+    }
+
+    /// Adds a new pending download
+    pub fn add_download(&mut self, path: String) -> Arc<Mutex<PendingDownload>> {
+        let pending = Arc::new(Mutex::new(PendingDownload::new(path)));
+        self.downloads.push(Arc::clone(&pending));
+
+        pending
+    }
+
+    /// Downloads all download entries
+    pub fn download_all(&self) {
+        self.downloads
+            .par_iter()
+            .for_each(|d| d.lock().unwrap().download())
+    }
+}
+
+/// A pending download entry.
+/// Download does not necessarily mean that it's not a local file
+#[derive(Clone, Debug)]
+pub struct PendingDownload {
+    path: String,
+    pub(crate) data: Option<Vec<u8>>,
+    pub(crate) wg: Option<WaitGroup>,
+}
+
+impl PendingDownload {
+    pub fn new(path: String) -> Self {
+        Self {
+            path,
+            data: None,
+            wg: Some(WaitGroup::new()),
+        }
+    }
+
+    /// Downloads the file and writes the content to the content field
+    pub fn download(&mut self) {
+        let wg = std::mem::replace(&mut self.wg, None);
+        if let Some(wg) = wg {
+            log::debug!("Reading {}...", self.path);
+            self.data = self.read_content();
+            log::debug!("{} read!", self.path);
+            drop(wg);
+        }
+    }
+
+    /// Reads the fiels content or downloads it if it doesn't exist in the filesystem
+    fn read_content(&self) -> Option<Vec<u8>> {
+        let path = PathBuf::from(&self.path);
+        if path.exists() {
+            read(path).ok()
+        } else {
+            self.download_content()
+        }
+    }
+
+    /// Downloads the content from the given url
+    fn download_content(&self) -> Option<Vec<u8>> {
+        reqwest::blocking::get(&self.path)
+            .ok()
+            .map(|c| c.bytes())
+            .and_then(|b| b.ok())
+            .map(|b| b.to_vec())
+    }
+}

src/utils/mod.rs

@@ -1 +1,2 @@
+pub mod downloads;
 pub mod parsing;