bug fix

src/main.rs

@@ -20,8 +20,8 @@
 //! オプションの各意味は以下のとおりです。
 //!
 //! - `--output`：解析で生成した情報を出力するフォルダ
-//! - `--start`：日時範囲の始端（オプション）
-//! - `--end`：日時範囲の終端（オプション）
+//! - `--start`：年範囲の始端（オプション）
+//! - `--end`：年範囲の終端（オプション）
 //! - `--rows`：一度の処理の重さ（オプション）
 //! - `--sleep-time`：一度の処理ごとに挟まるスリープ時間（オプション）
 //!
@@ -31,9 +31,8 @@
 //! (c) 2024 Naoki Kaneko (a.k.a. "puripuri2100")
 //!
 
-use anyhow::{anyhow, Context, Result};
+use anyhow::Result;
 use clap::Parser;
-use regex::Regex;
 use serde::{Deserialize, Serialize};
 use tokio::fs::File;
 use tokio::io::AsyncWriteExt;
@@ -44,10 +43,10 @@ use tracing::*;
 struct AppArgs {
   /// 検索する範囲の始端を<year>-<month>-<date>形式で与える
   #[clap(short, long)]
-  start: Option<String>,
+  start: Option<usize>,
   /// 検索する範囲の終端を<year>-<month>-<date>形式で与える
   #[clap(short, long)]
-  end: Option<String>,
+  end: Option<usize>,
   /// 出力先のフォルダ
   #[clap(short, long)]
   output: String,
@@ -59,25 +58,20 @@ struct AppArgs {
   sleep_time: u64,
 }
 
-fn gen_list_url(
-  start: Option<(usize, usize, usize)>,
-  end: Option<(usize, usize, usize)>,
-  n: usize,
-  rows: usize,
-) -> String {
-  let start_s = if let Some((y, m, d)) = start {
-    format!("{y}-{m}-{d}")
+fn gen_list_url(start: Option<usize>, end: Option<usize>, n: usize, rows: usize) -> String {
+  let start_s = if let Some(y) = start {
+    format!("{y:0>4}")
   } else {
-    String::from("1883-01-01")
+    String::from("*")
   };
-  let end_s = if let Some((y, m, d)) = end {
-    format!("{y:0>4}-{m:0>2}-{d:0>2}")
+  let end_s = if let Some(y) = end {
+    format!("{y:0>4}")
   } else {
-    String::from("NOW")
+    String::from("*")
   };
   let start_n = rows * n;
   format!(
-    r"https://jorei.slis.doshisha.ac.jp/api/reiki/select?f.municipality_id.facet.limit=1788&facet.mincount=1&facet.range=announcement_date&facet.range.gap=%2B1YEAR&facet.range.start={start_s}T00%3A00%3A00Z&facet.range.end={end_s}&hl=true&hl.fl=content&hl.usePhraseHighlighter=true&q=collection%3Alatest&start={start_n}&rows={rows}&fq=&facet=true&facet.field=municipality_type&facet.field=city&facet.field=type&facet.field=h_type&facet.field=municipality_id"
+    r"https://jorei.slis.doshisha.ac.jp/api/reiki/select?f.municipality_id.facet.limit=1788&facet.mincount=1&facet.range=announcement_date&facet.range.gap=%2B1YEAR&facet.range.start=1883-01-01T00%3A00%3A00Z&facet.range.end=NOW&q=collection%3Alatest%20AND%20announcement_date%3A%5B{start_s}%20TO%20{end_s}%5D&start={start_n}&rows={rows}&fq=&facet=true&facet.field=municipality_type&facet.field=city&facet.field=type&facet.field=h_type&facet.field=municipality_id"
   )
 }
 
@@ -91,10 +85,10 @@ struct JoreiInfoListResponseDocs {
   collected_date: Vec<String>,
   updated_date: Vec<String>,
   municipality_id: String,
-  prefecture: String,
-  city: String,
-  prefecture_kana: String,
-  city_kana: String,
+  prefecture: Option<String>,
+  city: Option<String>,
+  prefecture_kana: Option<String>,
+  city_kana: Option<String>,
   municipality_type: String,
   area: String,
   id: String,
@@ -129,10 +123,14 @@ struct JoreiInfoDocs {
   collected_date: Vec<String>,
   updated_date: Vec<String>,
   municipality_id: String,
-  prefecture: String,
-  city: String,
-  prefecture_kana: String,
-  city_kana: String,
+  #[serde(skip_serializing_if = "Option::is_none")]
+  prefecture: Option<String>,
+  #[serde(skip_serializing_if = "Option::is_none")]
+  city: Option<String>,
+  #[serde(skip_serializing_if = "Option::is_none")]
+  prefecture_kana: Option<String>,
+  #[serde(skip_serializing_if = "Option::is_none")]
+  city_kana: Option<String>,
   municipality_type: String,
   area: String,
   id: String,
@@ -201,32 +199,12 @@ async fn main() -> Result<()> {
 
   init_logger().await?;
 
-  let re_day = Regex::new(r"(?<y>\d{4})-(?<m>\d{2})-(?<d>\d{2})").unwrap();
-  let start = if let Some(s) = &args.start {
-    let caps = re_day.captures(s).with_context(|| anyhow!("err"))?;
-    let y = &caps["y"].parse::<usize>()?;
-    let m = &caps["m"].parse::<usize>()?;
-    let d = &caps["d"].parse::<usize>()?;
-    Some((*y, *m, *d))
-  } else {
-    None
-  };
-  let end = if let Some(s) = &args.end {
-    let caps = re_day.captures(s).with_context(|| anyhow!("err"))?;
-    let y = &caps["y"].parse::<usize>()?;
-    let m = &caps["m"].parse::<usize>()?;
-    let d = &caps["d"].parse::<usize>()?;
-    Some((*y, *m, *d))
-  } else {
-    None
-  };
-
   // jorei.slis.doshisa.ac.jpの証明書が壊れているので検証しない設定にする
   let client = reqwest::Client::builder()
     .danger_accept_invalid_certs(true)
     .build()?;
 
-  let first_api_url = gen_list_url(start, end, 0, args.rows);
+  let first_api_url = gen_list_url(args.start, args.end, 0, args.rows);
 
   let first_resp: JoreiListResponse = client.get(&first_api_url).send().await?.json().await?;
   let first_resp = first_resp.response;
@@ -237,7 +215,7 @@ async fn main() -> Result<()> {
 
   let mut stream = tokio_stream::iter(0..=(all_size / args.rows));
   while let Some(n) = stream.next().await {
-    let list_api_url = gen_list_url(start, end, n, args.rows);
+    let list_api_url = gen_list_url(args.start, args.end, n, args.rows);
 
     let list_resp: JoreiListResponse = client.get(&list_api_url).send().await?.json().await?;
     let id_lst = list_resp.response.docs.iter().map(|d| &d.id);