Port path command

brood/src/commands.rs

@@ -1 +1,2 @@
 pub mod ingest;
+pub mod path;

brood/src/commands/path.rs

@@ -1,77 +1,61 @@
-use std::{
-    fs::File,
-    io::{self, BufReader},
-    path::Path,
-};
+use std::{io, path::Path};
 
 use crate::{
     algo::Dijkstra,
-    data::{info::PageInfo, store},
-    graph::{Graph, NodeIdx},
-    util,
+    data,
+    util::{self, TitleNormalizer},
 };
 
-pub fn find_index_of_title(pages: &[PageInfo], title: &str) -> NodeIdx {
-    let title = util::normalize_link(title);
-    pages
-        .iter()
-        .enumerate()
-        .find(|(_, p)| util::normalize_link(&p.title) == title)
-        .map(|(i, _)| NodeIdx::new(i))
-        .expect("invalid title")
+/// Find the shortest path between two articles.
+#[derive(Debug, clap::Parser)]
+pub struct Cmd {
+    start: String,
+    goal: String,
 }
 
-pub fn resolve_redirects(pages: &[PageInfo], graph: &Graph, mut page: NodeIdx) -> NodeIdx {
-    loop {
-        if pages[page.usize()].redirect {
-            if let Some(next) = graph.edges_for(page).first() {
-                page = *next;
-                continue;
+impl Cmd {
+    pub fn run(self, data: &Path) -> io::Result<()> {
+        let normalizer = TitleNormalizer::new();
+
+        println!(">> Import");
+        let (pages, _links, graph) = data::read_from_file(data)?;
+
+        println!(">> Resolve articles");
+        let start = util::resolve_title(&normalizer, &pages, &graph, &self.start);
+        let goal = util::resolve_title(&normalizer, &pages, &graph, &self.goal);
+        println!("Start: {}", pages[start.usize()].title);
+        println!("Goal:  {}", pages[goal.usize()].title);
+
+        println!(">> Find path");
+        println!("> Preparing dijkstra");
+        let mut dijkstra = Dijkstra::new(&graph);
+        println!("> Running dijkstra");
+        dijkstra.run(
+            start,
+            |node| node == goal,
+            |source, _edge, _target| !pages[source.usize()].redirect as u32,
+        );
+
+        if dijkstra.cost(goal) == u32::MAX {
+            println!("No path found");
+            return Ok(());
+        }
+
+        println!("> Collecting path");
+        let path = dijkstra.path(goal);
+        let cost = dijkstra.cost(goal);
+
+        println!();
+        println!("Path found (cost {cost}, length {}):", path.len());
+        for page in path {
+            let info = &pages[page.usize()];
+            if info.redirect {
+                println!("v {:?}", info.title);
+            } else {
+                println!("- {:?}", info.title);
             }
         }
 
-        return page;
+        Ok(())
     }
 }
-
-pub fn path(datafile: &Path, start: &str, goal: &str) -> io::Result<()> {
-    println!(">> Import");
-    let mut databuf = BufReader::new(File::open(datafile)?);
-    let (pages, _links, graph) = store::read_graph(&mut databuf)?;
-
-    println!(">> Locate from and to");
-    let start = resolve_redirects(&pages, &graph, find_index_of_title(&pages, start));
-    let goal = resolve_redirects(&pages, &graph, find_index_of_title(&pages, goal));
-    println!("Start: {:?}", pages[start.usize()].title);
-    println!("Goal:  {:?}", pages[goal.usize()].title);
-
-    println!(">> Find path");
-    println!("> Preparing dijkstra");
-    let mut dijkstra = Dijkstra::new(&graph);
-    println!("> Running dijkstra");
-    dijkstra.run(
-        start,
-        |node| node == goal,
-        |source, _edge, _target| !pages[source.usize()].redirect as u32,
-    );
-
-    if dijkstra.cost(goal) == u32::MAX {
-        println!("No path found");
-        return Ok(());
-    }
-
-    println!("> Collecting path");
-    let path = dijkstra.path(goal);
-    let cost = dijkstra.cost(goal);
-    println!("Path found (cost {cost}, length {}):", path.len());
-    for page in path {
-        let info = &pages[page.usize()];
-        if info.redirect {
-            println!(" v {:?}", info.title);
-        } else {
-            println!(" - {:?}", info.title);
-        }
-    }
-
-    Ok(())
-}

brood/src/main.rs

@@ -11,6 +11,7 @@ use clap::Parser;
 #[derive(Debug, Parser)]
 enum Command {
     Ingest(commands::ingest::Cmd),
+    Path(commands::path::Cmd),
 }
 
 #[derive(Debug, Parser)]
@@ -24,5 +25,6 @@ fn main() -> io::Result<()> {
     let args = Args::parse();
     match args.command {
         Command::Ingest(cmd) => cmd.run(&args.datafile),
+        Command::Path(cmd) => cmd.run(&args.datafile),
     }
 }

brood/src/util.rs

@@ -1,8 +1,13 @@
-use std::{fmt, iter, time::Instant};
+use std::{fmt, iter, thread::panicking, time::Instant};
 
 use regex::Regex;
 use thousands::Separable;
 
+use crate::{
+    data::Page,
+    graph::{Graph, NodeIdx},
+};
+
 pub struct Counter {
     n: usize,
     last_print: Instant,
@@ -145,27 +150,34 @@ impl TitleNormalizer {
     }
 }
 
-/*
-pub fn find_index_of_title(pages: &[Page<PageInfo>], title: &str) -> u32 {
-    let title = normalize_link(title);
+pub fn locate_title(normalizer: &TitleNormalizer, pages: &[Page], title: &str) -> NodeIdx {
+    let normalized = normalizer.normalize(title);
     pages
         .iter()
         .enumerate()
-        .find(|(_, p)| normalize_link(&p.data.title) == title)
-        .map(|(i, _)| i)
-        .expect("invalid title") as u32
+        .find(|(_, p)| normalizer.normalize(&p.title) == normalized)
+        .map(|(i, _)| NodeIdx::new(i))
+        .expect("invalid title")
 }
 
-pub fn resolve_redirects(data: &AdjacencyList<PageInfo, LinkInfo>, mut page_idx: u32) -> u32 {
+pub fn resolve_redirects(pages: &[Page], graph: &Graph, mut page: NodeIdx) -> NodeIdx {
     loop {
-        if data.page(page_idx).data.redirect {
-            if let Some(link_idx) = data.link_redirect(page_idx) {
-                page_idx = data.link(link_idx).to;
+        if pages[page.usize()].redirect {
+            if let Some(target) = graph.edges_for(page).first() {
+                page = *target;
                 continue;
             }
         }
 
-        return page_idx;
+        return page;
     }
 }
-*/
+
+pub fn resolve_title(
+    normalizer: &TitleNormalizer,
+    pages: &[Page],
+    graph: &Graph,
+    title: &str,
+) -> NodeIdx {
+    resolve_redirects(pages, graph, locate_title(normalizer, pages, title))
+}