From 49665f74ce762ea475824f0423f9716e109160e4 Mon Sep 17 00:00:00 2001
From: Joscha <joscha@plugh.de>
Date: Mon, 30 Dec 2024 13:12:14 +0100
Subject: [PATCH] List links to and from an article

---
 brood/src/commands.rs            |  1 +
 brood/src/commands/list_links.rs | 87 ++++++++++++++++++++++++++++++++
 brood/src/main.rs                |  6 +++
 3 files changed, 94 insertions(+)
 create mode 100644 brood/src/commands/list_links.rs

diff --git a/brood/src/commands.rs b/brood/src/commands.rs
index ffff9d3..6da3050 100644
--- a/brood/src/commands.rs
+++ b/brood/src/commands.rs
@@ -1,4 +1,5 @@
 pub mod ingest;
+pub mod list_links;
 pub mod list_pages;
 pub mod longest_shortest_path;
 pub mod path;
diff --git a/brood/src/commands/list_links.rs b/brood/src/commands/list_links.rs
new file mode 100644
index 0000000..37c9972
--- /dev/null
+++ b/brood/src/commands/list_links.rs
@@ -0,0 +1,87 @@
+use std::{
+    collections::HashSet,
+    fs::File,
+    io::{self, BufReader},
+    path::Path,
+};
+
+use crate::{
+    data::{
+        adjacency_list::AdjacencyList,
+        info::{LinkInfo, PageInfo},
+        store,
+    },
+    util,
+};
+
+fn links_from(data: &AdjacencyList<PageInfo, LinkInfo>, idx: u32) -> HashSet<u32> {
+    data.links(idx).map(|(_, ld)| ld.to).collect()
+}
+
+fn links_to(data: &AdjacencyList<PageInfo, LinkInfo>, idx: u32) -> HashSet<u32> {
+    let mut links = HashSet::<u32>::new();
+    for (pi, _) in data.pages() {
+        for (_, ld) in data.links(pi) {
+            if ld.to == idx {
+                links.insert(pi);
+                continue;
+            }
+        }
+    }
+    links
+}
+
+fn print_links(data: &AdjacencyList<PageInfo, LinkInfo>, name: &str, links: &HashSet<u32>) {
+    let mut links = links
+        .iter()
+        .map(|pi| {
+            let page = data.page(*pi);
+            (&page.data.title as &str, page.data.redirect)
+        })
+        .collect::<Vec<_>>();
+
+    links.sort();
+
+    println!(">> {name} ({}):", links.len());
+    for (title, redirect) in links {
+        if redirect {
+            println!("v {title}");
+        } else {
+            println!("- {title}");
+        }
+    }
+}
+
+pub fn run(datafile: &Path, page: &str) -> io::Result<()> {
+    println!(">> Import");
+    let mut databuf = BufReader::new(File::open(datafile)?);
+    let data = store::read_adjacency_list(&mut databuf)?;
+
+    println!(">> Locate page");
+    let idx = util::resolve_redirects(&data, util::find_index_of_title(&data.pages, page));
+    println!("Page: {:?}", data.page(idx).data.title);
+
+    println!(">> Find links");
+    let from = links_from(&data, idx);
+    let to = links_to(&data, idx);
+    let twins = from.intersection(&to).copied().collect::<HashSet<_>>();
+    let twinless_from = from.difference(&twins).copied().collect::<HashSet<_>>();
+    let twinless_to = to.difference(&twins).copied().collect::<HashSet<_>>();
+
+    println!();
+    print_links(&data, "From", &from);
+
+    println!();
+    print_links(&data, "To", &to);
+
+    println!();
+    print_links(&data, "Twins", &twins);
+
+    println!();
+    print_links(&data, "From without twins", &twinless_from);
+
+    println!();
+    print_links(&data, "To without twins", &twinless_to);
+
+    Ok(())
+}
diff --git a/brood/src/main.rs b/brood/src/main.rs
index e4b4074..57d1b81 100644
--- a/brood/src/main.rs
+++ b/brood/src/main.rs
@@ -44,6 +44,11 @@ enum Command {
     },
     /// Print all page titles.
     ListPages,
+    /// Print all links.
+    ListLinks {
+        /// The page to inspect.
+        page: String,
+    },
 }
 
 #[derive(Debug, Parser)]
@@ -76,5 +81,6 @@ fn main() -> io::Result<()> {
             commands::philosophy_game::run(&args.datafile, subcmd)
         }
         Command::ListPages => commands::list_pages::run(&args.datafile),
+        Command::ListLinks { page } => commands::list_links::run(&args.datafile, &page),
     }
 }