1 files changed, 61 insertions(+), 44 deletions(-)

M src/deputies.rs

@@ 10,10 10,17 @@ use crate::{log_error, opt::Opt};
 use reqwest::Url;
 
 #[derive(Serialize, Deserialize, Debug)]
-struct CongressMember {
-    name: String,
-    state: String,
-    district: u8
+enum CongressMember {
+    Uninominal {
+        name: String,
+        state: String,
+        #[serde(skip_serializing_if = "Option::is_none")]
+        district: Option<u8>,
+    },
+    Plurinominal {
+        name: String,
+        circumscription: String
+    }
 }
 
 pub fn run(opts: Opt) {


@@ 59,7 66,10 @@ async fn process(_opts: &Opt) -> std::result::Result<(), Error> {
             };
 
             if let Some(l) = link {
-                extract(l)?
+                let deputies = extract(l)?;
+                let j = serde_json::to_string(&deputies)?;
+
+                println!("{}", j);
             }
         }
     }


@@ 67,7 77,7 @@ async fn process(_opts: &Opt) -> std::result::Result<(), Error> {
     Ok(())
 }
 
-fn extract(link: String) -> std::result::Result<(), Error> {
+fn extract(link: String) -> std::result::Result<Vec<CongressMember>, Error> {
     println!("{:?}", link);
     let url = Url::parse(&link)?;
 


@@ 77,38 87,12 @@ fn extract(link: String) -> std::result::Result<(), Error> {
     res.copy_to(&mut buf)?;
     let s: String = String::from_utf8(buf)?;
 
-    let document = kuchiki::parse_html().one(s);
-    // let h3_list = document.select("h3").unwrap();
-
-    // for h3 in h3_list.filter_map(|h| with_title(h, "Diputados por distrito".to_string())) {
-    //     let table = h3.as_node().next_sibling().unwrap().next_sibling().unwrap();
-    //     println!("  {:?}", h3.as_node().text_contents());
-    //     for tr in table.select("tr:not(:first-child)").unwrap() {
-    //         let cols: Vec<kuchiki::NodeDataRef<ElementData>> = tr.as_node()
-    //                     .select("td")
-    //                     .unwrap()
-    //                     .collect();
-
-    //         let count = cols.len();
-    //         let (first, second) = cols.split_at(count / 2);
-
-    //         for col in &[first, second] {
-    //             let person = CongressMember {
-    //                 district: col[1].text_contents().parse::<u8>().unwrap_or_default(),
-    //                 state: col[0].text_contents(),
-    //                 name: col[2]
-    //                     .as_node()
-    //                     .select_first("a")
-    //                     .map_or_else(|_| "".to_owned(), |v| v.text_contents())
-    //             };
-    
-    //             println!("{:?}", person);                
-    //         }
-    //     }
-    // }
+    let mut deputies: Vec<CongressMember> = vec![];
 
+    let document = kuchiki::parse_html().one(s);
     let h3_list = document.select("h3").unwrap();
-    for h3 in h3_list.filter_map(|h| with_title(h, "Diputados por representaci".to_string())) {
+
+    for h3 in h3_list.filter_map(|h| with_title(h, "Diputados por distrito".to_string())) {
         let table = h3.as_node().next_sibling().unwrap().next_sibling().unwrap();
         println!("  {:?}", h3.as_node().text_contents());
         for tr in table.select("tr:not(:first-child)").unwrap() {


@@ 119,24 103,57 @@ fn extract(link: String) -> std::result::Result<(), Error> {
 
             let count = cols.len();
             let (first, second) = cols.split_at(count / 2);
-            let sub_cols = &[first, second];
 
-            for col in sub_cols {
-                let person = CongressMember {
-                    district: 0,
-                    state: "".to_owned(),
-                    name: col[1]
+            for col in &[first, second] {
+                let person = CongressMember::Uninominal {
+                    district: col[1].text_contents().parse::<u8>().ok(),
+                    state: col[0].text_contents(),
+                    name: col[2]
                         .as_node()
                         .select_first("a")
                         .map_or_else(|_| "".to_owned(), |v| v.text_contents())
                 };
     
-                println!("{:?}", person);                
+                deputies.push(person);
+                //println!("{:?}", person);                
             }
         }
     }
 
-    Ok(())
+    let h3_list = document.select("h3").unwrap();
+    for h3 in h3_list.filter_map(|h| with_title(h, "Diputados por representaci".to_string())) {
+        let table = h3.as_node().next_sibling().unwrap().next_sibling().unwrap();
+        println!("  {:?}", h3.as_node().text_contents());
+        for tr in table.select("tr:not(:first-child)").unwrap() {
+            let td: Vec<kuchiki::NodeDataRef<ElementData>> = tr.as_node()
+                        .select("td")
+                        .unwrap()
+                        .collect();
+
+            let count = td.len();
+            let cols = if td.len() > 3 {
+                let (first, second) = td.split_at(count / 2);
+                [first, second].to_vec()
+            } else {
+                [td.as_slice()].to_vec()
+            };
+
+            for col in cols {
+                let person = CongressMember::Plurinominal {
+                    name: col[1]
+                        .as_node()
+                        .select_first("a")
+                        .map_or_else(|_| "".to_owned(), |v| v.text_contents()),
+                    circumscription: col[0].text_contents(),
+                };
+    
+                deputies.push(person);
+                //println!("{:?}", person);                
+            }
+        }
+    }
+
+    Ok(deputies)
 }
 
 fn with_title(