feat: add --id flag and model not found validation (#32)

Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "untitled"
-version = "0.4.0"
+version = "0.4.1"
 edition = "2021"
 
 [dependencies]

src/cli.rs

@@ -15,9 +15,6 @@ pub struct Cli {
     #[arg(short, long, value_name = "FILE", default_value = "untitled.toml")]
     pub config: PathBuf,
 
-    #[arg(long, default_value = "false")]
-    pub full_size_image: bool,
-
     #[command(subcommand)]
     pub command: Option<Commands>,
 }
@@ -36,8 +33,12 @@ pub enum Commands {
         #[arg(short, long, value_name = "URL")]
         url: Option<String>,
 
+        /// Use just the girl's gallery ID
+        #[arg(short, long, value_name = "ID", conflicts_with = "url")]
+        id: Option<String>,
+
         /// Download full-size images
-        #[arg(long)]
+        #[arg(long, default_value = "false")]
         full_size_image: bool,
     },
     #[command(about = "Updates girl's galleries")]

src/main.rs

@@ -4,6 +4,7 @@ mod scraper;
 mod utilities;
 
 use crate::scraper::collector::scrape;
+use crate::utilities::validate_id;
 use clap::Parser;
 use cli::{Cli, Commands};
 use config::{print_config, read_or_create_config, MyConfig};
@@ -19,10 +20,24 @@ fn main() {
             }
             Some(Commands::Scrape {
                 url,
+                id,
                 full_size_image,
-            }) => match url {
-                Some(url) => scrape(&config, Some(&url), full_size_image),
-                None => scrape(&config, None, false),
+            }) => match (url, id) {
+                (Some(url), None) => scrape(&config, Some(&url), full_size_image),
+                (None, Some(id)) => {
+                    if !validate_id(&id) {
+                        eprintln!(
+                            "Girl's page ID's are only numbers! Double check the id and try again."
+                        );
+                        eprintln!("Note: id's between 0 and 5 are invalid.");
+                        return;
+                    }
+                    let constructed_url =
+                        format!("https://www.kindgirls.com/old/girls.php?id={}", id);
+                    scrape(&config, Some(&constructed_url), full_size_image)
+                }
+                (None, None) => eprintln!("You need to specify either a girl's page URL or ID!"),
+                (Some(_), Some(_)) => eprintln!("You can't use both URL and ID at the same time!"),
             },
             Some(Commands::Update) => {
                 scraper::updater::Updater::update(&config).unwrap();

src/scraper/collector.rs

@@ -1,7 +1,7 @@
 use super::structs::{Bio, Gallery, Girl, Selectors, Stats, Video, Visuals};
 use crate::config::Config;
 use crate::scraper::downloader::{Downloader, DownloaderImpl};
-use crate::utilities::{build_video_src_url, parse_video_duration, splitter, todays_date};
+use crate::utilities::{build_video_src_url, parse_video_duration, splitter, today_date};
 use reqwest::blocking::Client;
 use reqwest::header::{HeaderMap, HeaderValue, USER_AGENT};
 use reqwest::Error;
@@ -24,6 +24,12 @@ pub fn scrape<T: Config>(config: &T, url: Option<&str>, full_size_image: bool) {
             match body {
                 Ok(content) => {
                     let document = Html::parse_document(&content);
+
+                    if is_model_not_found(&document) {
+                        println!("Model not found! Please check the URl or ID and try again.");
+                        return;
+                    }
+
                     let girl = collect_girl(url, &document, full_size_image);
                     // Debug only; TODO: remove when logging is implemented
                     // println!("{:?}", girl);
@@ -171,9 +177,17 @@ fn collect_stats(visuals: &Visuals) -> Stats {
     }
 }
 
+fn is_model_not_found(document: &Html) -> bool {
+    let selector = Selector::parse(Selectors::MODEL_NOT_FOUND).unwrap();
+    if let Some(element) = document.select(&selector).next() {
+        return element.text().any(|s| s.contains("Model not found"));
+    }
+    false
+}
+
 pub(crate) fn collect_girl(url: &str, document: &Html, full_size_image: bool) -> Girl {
     let is_single_gallery = Girl::is_single_gallery(url);
-    let last_update: Option<String> = Some(todays_date());
+    let last_update: Option<String> = Some(today_date());
     let bio = collect_bio(document, url);
     let galleries = collect_gallery(document, full_size_image);
     let videos = collect_videos(document);

src/scraper/structs.rs

@@ -11,6 +11,8 @@ impl Selectors {
     pub const GALLERY_IMAGE_SRC: &'static str = r#".gal_list a img"#;
     pub const GALLERY_IMAGE_FULL_SIZE_SRC: &'static str = r#".gal_full a img"#;
     pub const MODEL_VIDEOS: &'static str = r#".video_list a"#;
+    // Use # for id's and . for classes.
+    pub const MODEL_NOT_FOUND: &'static str = "#cuerpo";
 }
 
 #[derive(Debug, Serialize, Deserialize)]

src/scraper/updater.rs

@@ -5,7 +5,7 @@ use crate::config::Config;
 use crate::scraper::collector::{collect_girl, fetch};
 use crate::scraper::downloader::Downloader;
 use crate::scraper::structs::Girl;
-use crate::utilities::todays_date;
+use crate::utilities::today_date;
 
 pub(crate) struct Updater;
 
@@ -21,7 +21,6 @@ impl Updater {
             let dir = dir?;
             let path = dir.path();
             if path.is_dir() {
-                println!("Updating: {:?}", path);
                 Self::update_girl_content(config, &dir.path())?;
             }
         }
@@ -42,7 +41,10 @@ impl Updater {
             let mut existing_girl: Girl = serde_json::from_str(&content)?;
 
             if let Some(link) = &existing_girl.bio.link {
-                println!("Checking updates for {}", existing_girl.bio.get_name());
+                println!(
+                    "Checking for new content for {}",
+                    existing_girl.bio.get_name()
+                );
 
                 let body = fetch(link)?;
                 let document = scraper::Html::parse_document(&body);
@@ -51,14 +53,14 @@ impl Updater {
                 let has_new_content = Self::compare_content(&existing_girl, &new_girl);
 
                 if has_new_content {
-                    println!("New content found for {}", existing_girl.bio.get_name());
+                    println!("New content found!");
                     Self::prompt_and_download(config, &new_girl)?;
 
                     // Update the existing girl with new content and timestamp
                     existing_girl.is_single_gallery = false;
                     existing_girl.content = new_girl.content;
                     existing_girl.stats = new_girl.stats;
-                    existing_girl.last_update = Some(todays_date());
+                    existing_girl.last_update = Some(today_date());
 
                     // Save updated JSON
                     let json = serde_json::to_string_pretty(&existing_girl)?;
@@ -80,8 +82,10 @@ impl Updater {
             .as_ref()
             .map(|v| v.len())
             .unwrap_or(0);
+
         let new_videos = new.content.videos.as_ref().map(|v| v.len()).unwrap_or(0);
 
+        // This is a simple quantity comparison, which should be enough for now
         new_galleries > existing_galleries || new_videos > existing_videos
     }
 

src/utilities.rs

@@ -61,6 +61,10 @@ pub fn format_date(date_str: &str) -> Option<String> {
         .map(|date| date.format("%d-%m-%Y").to_string())
 }
 
-pub fn todays_date() -> String {
+pub fn today_date() -> String {
     chrono::Local::now().format("%d-%m-%Y").to_string()
 }
+
+pub fn validate_id(id: &str) -> bool {
+    id.chars().all(char::is_numeric)
+}