Скачивание XLS документа по ссылке

.github/workflows/test.yml

@@ -22,4 +22,4 @@ jobs:
       - name: Build
         run: cargo build --verbose
       - name: Run tests
-        run: cargo test --verbose -p schedule_parser
+        run: cargo test --verbose -p schedule-parser-rusted -p schedule_parser

Cargo.toml

@@ -3,11 +3,14 @@ members = ["lib/schedule_parser"]
 
 [package]
 name = "schedule-parser-rusted"
-version = "0.2.0"
+version = "0.3.0"
 edition = "2024"
 publish = false
 
 [dependencies]
 serde = { version = "1.0.219", features = ["derive"] }
 serde_json = "1.0.140"
-schedule_parser = { path = "./lib/schedule_parser" }
+schedule_parser = { path = "./lib/schedule_parser" }
+chrono = "0.4.40"
+reqwest = "0.12.15"
+tokio = { version = "1.44.1", features = ["macros", "rt-multi-thread"] }

lib/schedule_parser/benches/parse.rs

@@ -1,11 +1,10 @@
 use criterion::{Criterion, criterion_group, criterion_main};
 use schedule_parser::parse_xls;
-use std::path::Path;
 
 pub fn bench_parse_xls(c: &mut Criterion) {
-    c.bench_function("parse_xls", |b| {
-        b.iter(|| parse_xls(Path::new("../../schedule.xls")))
-    });
+    let buffer: Vec<u8> = include_bytes!("../../../schedule.xls").to_vec();
+
+    c.bench_function("parse_xls", |b| b.iter(|| parse_xls(&buffer)));
 }
 
 criterion_group!(benches, bench_parse_xls);

lib/schedule_parser/src/lib/lib.rs

@@ -1,13 +1,13 @@
-use crate::LessonParseResult::{Lessons, Street};
 use crate::schema::LessonType::Break;
 use crate::schema::{Day, Lesson, LessonSubGroup, LessonTime, LessonType, ScheduleEntity};
-use calamine::{Reader, Xls, open_workbook};
+use crate::LessonParseResult::{Lessons, Street};
+use calamine::{open_workbook_from_rs, Reader, Xls};
 use chrono::{Duration, NaiveDateTime};
-use fuzzy_matcher::FuzzyMatcher;
 use fuzzy_matcher::skim::SkimMatcherV2;
+use fuzzy_matcher::FuzzyMatcher;
 use regex::Regex;
 use std::collections::HashMap;
-use std::path::Path;
+use std::io::Cursor;
 use std::sync::LazyLock;
 
 mod schema;
@@ -539,12 +539,13 @@ fn convert_groups_to_teachers(
 }
 
 pub fn parse_xls(
-    path: &Path,
+    buffer: &Vec<u8>,
 ) -> (
     HashMap<String, ScheduleEntity>,
     HashMap<String, ScheduleEntity>,
 ) {
-    let mut workbook: Xls<_> = open_workbook(path).expect("Can't open workbook");
+    let cursor = Cursor::new(&buffer);
+    let mut workbook: Xls<_> = open_workbook_from_rs(cursor).expect("Can't open workbook");
 
     let worksheet: WorkSheet = workbook
         .worksheets()
@@ -693,9 +694,10 @@ mod tests {
     use super::*;
 
     #[test]
-    fn it_works() {
-        let result = parse_xls(Path::new("../../schedule.xls"));
-        
+    fn read() {
+        let buffer: Vec<u8> = include_bytes!("../../../../schedule.xls").to_vec();
+        let result = parse_xls(&buffer);
+
         assert_ne!(result.0.len(), 0);
         assert_ne!(result.1.len(), 0);
     }

src/main.rs

@@ -1,9 +1,25 @@
-use std::fs;
-use std::path::Path;
+use crate::xls_downloader::basic_impl::BasicXlsDownloader;
+use crate::xls_downloader::interface::XLSDownloader;
 use schedule_parser::parse_xls;
+use std::{env, fs};
 
-fn main() {
-    let (teachers, groups) = parse_xls(Path::new("./schedule.xls"));
+mod xls_downloader;
+
+#[tokio::main]
+async fn main() {
+    let args: Vec<String> = env::args().collect();
+    assert_ne!(args.len(), 1);
+
+    let mut downloader = BasicXlsDownloader::new();
+
+    downloader
+        .set_url(args[1].to_string())
+        .await
+        .expect("Failed to set url");
+
+    let fetch_res = downloader.fetch(false).await.expect("Failed to fetch xls");
+
+    let (teachers, groups) = parse_xls(fetch_res.data.as_ref().unwrap());
 
     fs::write(
         "./schedule.json",
@@ -12,7 +28,7 @@ fn main() {
             .as_bytes(),
     )
     .expect("Failed to write schedule");
-    
+
     fs::write(
         "./teachers.json",
         serde_json::to_string_pretty(&teachers)

src/xls_downloader/basic_impl.rs

@@ -0,0 +1,219 @@
+use crate::xls_downloader::interface::{FetchError, FetchOk, FetchResult, XLSDownloader};
+use chrono::{DateTime, Utc};
+
+pub struct BasicXlsDownloader {
+    url: Option<String>,
+}
+
+async fn fetch_specified(url: &String, user_agent: String, head: bool) -> FetchResult {
+    let client = reqwest::Client::new();
+
+    let response = if head {
+        client.head(url)
+    } else {
+        client.get(url)
+    }
+    .header("User-Agent", user_agent)
+    .send()
+    .await;
+
+    match response {
+        Ok(r) => {
+            if r.status().as_u16() != 200 {
+                return Err(FetchError::BadStatusCode);
+            }
+
+            let headers = r.headers();
+
+            let content_type = headers.get("Content-Type");
+            let etag = headers.get("etag");
+            let last_modified = headers.get("last-modified");
+            let date = headers.get("date");
+
+            if content_type.is_none() || etag.is_none() || last_modified.is_none() || date.is_none()
+            {
+                Err(FetchError::BadHeaders)
+            } else if content_type.unwrap() != "application/vnd.ms-excel" {
+                Err(FetchError::BadContentType)
+            } else {
+                let etag = etag.unwrap().to_str().unwrap().to_string();
+                let last_modified =
+                    DateTime::parse_from_rfc2822(&last_modified.unwrap().to_str().unwrap())
+                        .unwrap()
+                        .with_timezone(&Utc);
+
+                Ok(if head {
+                    FetchOk::head(etag, last_modified)
+                } else {
+                    FetchOk::get(etag, last_modified, r.bytes().await.unwrap().to_vec())
+                })
+            }
+        }
+        Err(_) => Err(FetchError::Unknown),
+    }
+}
+
+impl BasicXlsDownloader {
+    pub fn new() -> Self {
+        BasicXlsDownloader { url: None }
+    }
+}
+
+impl XLSDownloader for BasicXlsDownloader {
+    async fn fetch(&self, head: bool) -> FetchResult {
+        if self.url.is_none() {
+            Err(FetchError::NoUrlProvided)
+        } else {
+            fetch_specified(
+                self.url.as_ref().unwrap(),
+                "t.me/polytechnic_next".to_string(),
+                head,
+            )
+            .await
+        }
+    }
+
+    async fn set_url(&mut self, url: String) -> Result<(), FetchError> {
+        let result = fetch_specified(&url, "t.me/polytechnic_next".to_string(), true).await;
+
+        if let Ok(_) = result {
+            Ok(self.url = Some(url))
+        } else {
+            Err(result.err().unwrap())
+        }
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use crate::xls_downloader::basic_impl::{BasicXlsDownloader, fetch_specified};
+    use crate::xls_downloader::interface::{FetchError, XLSDownloader};
+
+    #[tokio::test]
+    async fn bad_url() {
+        let url = "bad_url".to_string();
+        let user_agent = String::new();
+
+        let results = [
+            fetch_specified(&url, user_agent.clone(), true).await,
+            fetch_specified(&url, user_agent.clone(), false).await,
+        ];
+
+        assert!(results[0].is_err());
+        assert!(results[1].is_err());
+    }
+
+    #[tokio::test]
+    async fn bad_status_code() {
+        let url = "https://www.google.com/not-found".to_string();
+        let user_agent = String::new();
+
+        let results = [
+            fetch_specified(&url, user_agent.clone(), true).await,
+            fetch_specified(&url, user_agent.clone(), false).await,
+        ];
+
+        assert!(results[0].is_err());
+        assert!(results[1].is_err());
+
+        assert_eq!(
+            *results[0].as_ref().err().unwrap(),
+            FetchError::BadStatusCode
+        );
+        assert_eq!(
+            *results[1].as_ref().err().unwrap(),
+            FetchError::BadStatusCode
+        );
+    }
+
+    #[tokio::test]
+    async fn bad_headers() {
+        let url = "https://www.google.com/favicon.ico".to_string();
+        let user_agent = String::new();
+
+        let results = [
+            fetch_specified(&url, user_agent.clone(), true).await,
+            fetch_specified(&url, user_agent.clone(), false).await,
+        ];
+
+        assert!(results[0].is_err());
+        assert!(results[1].is_err());
+
+        assert_eq!(*results[0].as_ref().err().unwrap(), FetchError::BadHeaders);
+        assert_eq!(*results[1].as_ref().err().unwrap(), FetchError::BadHeaders);
+    }
+
+    #[tokio::test]
+    async fn bad_content_type() {
+        let url = "https://s3.aero-storage.ldragol.ru/679e5d1145a6ad00843ad3f1/67ddb59fd46303008396ac96%2Fexample.txt".to_string();
+        let user_agent = String::new();
+
+        let results = [
+            fetch_specified(&url, user_agent.clone(), true).await,
+            fetch_specified(&url, user_agent.clone(), false).await,
+        ];
+
+        assert!(results[0].is_err());
+        assert!(results[1].is_err());
+
+        assert_eq!(
+            *results[0].as_ref().err().unwrap(),
+            FetchError::BadContentType
+        );
+        assert_eq!(
+            *results[1].as_ref().err().unwrap(),
+            FetchError::BadContentType
+        );
+    }
+
+    #[tokio::test]
+    async fn ok() {
+        let url = "https://s3.aero-storage.ldragol.ru/679e5d1145a6ad00843ad3f1/67ddb5fad46303008396ac97%2Fschedule.xls".to_string();
+        let user_agent = String::new();
+
+        let results = [
+            fetch_specified(&url, user_agent.clone(), true).await,
+            fetch_specified(&url, user_agent.clone(), false).await,
+        ];
+
+        assert!(results[0].is_ok());
+        assert!(results[1].is_ok());
+    }
+
+    #[tokio::test]
+    async fn downloader_set_ok() {
+        let url = "https://s3.aero-storage.ldragol.ru/679e5d1145a6ad00843ad3f1/67ddb5fad46303008396ac97%2Fschedule.xls".to_string();
+
+        let mut downloader = BasicXlsDownloader::new();
+
+        assert!(downloader.set_url(url).await.is_ok());
+    }
+
+    #[tokio::test]
+    async fn downloader_set_err() {
+        let url = "bad_url".to_string();
+
+        let mut downloader = BasicXlsDownloader::new();
+
+        assert!(downloader.set_url(url).await.is_err());
+    }
+
+    #[tokio::test]
+    async fn downloader_ok() {
+        let url = "https://s3.aero-storage.ldragol.ru/679e5d1145a6ad00843ad3f1/67ddb5fad46303008396ac97%2Fschedule.xls".to_string();
+
+        let mut downloader = BasicXlsDownloader::new();
+
+        assert!(downloader.set_url(url).await.is_ok());
+        assert!(downloader.fetch(false).await.is_ok());
+    }
+
+    #[tokio::test]
+    async fn downloader_no_url_provided() {
+        let downloader = BasicXlsDownloader::new();
+        let result = downloader.fetch(false).await;
+
+        assert!(result.is_err());
+        assert_eq!(result.err().unwrap(), FetchError::NoUrlProvided);
+    }
+}

src/xls_downloader/interface.rs

@@ -0,0 +1,44 @@
+use chrono::{DateTime, Utc};
+
+#[derive(PartialEq, Debug)]
+pub enum FetchError {
+    NoUrlProvided,
+    Unknown,
+    BadStatusCode,
+    BadContentType,
+    BadHeaders,
+}
+
+pub struct FetchOk {
+    pub etag: String,
+    pub uploaded_at: DateTime<Utc>,
+    pub requested_at: DateTime<Utc>,
+    pub data: Option<Vec<u8>>,
+}
+
+impl FetchOk {
+    pub fn head(etag: String, uploaded_at: DateTime<Utc>) -> Self {
+        FetchOk {
+            etag,
+            uploaded_at,
+            requested_at: Utc::now(),
+            data: None,
+        }
+    }
+
+    pub fn get(etag: String, uploaded_at: DateTime<Utc>, data: Vec<u8>) -> Self {
+        FetchOk {
+            etag,
+            uploaded_at,
+            requested_at: Utc::now(),
+            data: Some(data),
+        }
+    }
+}
+
+pub type FetchResult = Result<FetchOk, FetchError>;
+
+pub trait XLSDownloader {
+    async fn fetch(&self, head: bool) -> FetchResult;
+    async fn set_url(&mut self, url: String) -> Result<(), FetchError>;
+}

src/xls_downloader/mod.rs

@@ -0,0 +1,2 @@
+pub mod basic_impl;
+pub mod interface;