3 files changed, 56 insertions(+), 15 deletions(-)

M src/congress/legislatures.rs
M src/congress/mod.rs
M src/congress/senators.rs

@@ 62,7 62,7 @@ async fn process(_opts: &Opt) -> std::result::Result<Vec<Legislature>, Error> {
             .to_string();
         let name = a.text_contents();
 
-        let link = if href.starts_with("/wiki") {
+        let _link = if href.starts_with("/wiki") {
             let link_str =
                 format!("{}://{}{}", url.scheme(), url.host_str().unwrap(), href).to_string();
             Some(link_str)

@@ 6,7 6,7 @@ fn with_title(
     h: kuchiki::NodeDataRef<kuchiki::ElementData>,
     title: String,
 ) -> Option<kuchiki::NodeDataRef<kuchiki::ElementData>> {
-    if h.text_contents().starts_with(&title) {
+    if h.text_contents().contains(&title) {
         Some(h)
     } else {
         None

@@ 10,10 10,19 @@ use crate::{log_error, opt::Opt};
 use reqwest::Url;
 
 #[derive(Serialize, Deserialize, Debug)]
-struct CongressMember {
-    name: String,
-    state: String,
-    district: u8
+enum CongressMember {
+    ByState {
+        name: String,
+        state: String,
+        #[serde(skip_serializing_if = "Option::is_none")]
+        party: Option<String>
+    },
+
+    ByList {
+        name: String,
+        #[serde(skip_serializing_if = "Option::is_none")]
+        party: Option<String>
+    }
 }
 
 pub fn run(opts: Opt) {


@@ 59,7 68,10 @@ async fn process(_opts: &Opt) -> std::result::Result<(), Error> {
             };
 
             if let Some(l) = link {
-                extract(l)?
+                let senators = extract(l)?;
+                let j = serde_json::to_string(&senators)?;
+
+                println!("{}", j);
             }
         }
     }


@@ 67,7 79,7 @@ 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)?;
 


@@ 80,7 92,9 @@ fn extract(link: String) -> std::result::Result<(), Error> {
     let document = kuchiki::parse_html().one(s);
     let h3_list = document.select("h3").unwrap();
 
-    for h3 in h3_list.filter_map(|h| super::with_title(h, "Diputados por distrito".to_string())) {
+    let mut senators = vec![];
+
+    for h3 in h3_list.filter_map(|h| super::with_title(h, "entidad federativa".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() {


@@ 93,19 107,46 @@ fn extract(link: String) -> std::result::Result<(), Error> {
             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(),
+                let person = CongressMember::ByState {
                     state: col[0].text_contents(),
-                    name: col[2]
+                    name: col[1]
                         .as_node()
                         .select_first("a")
-                        .map_or_else(|_| "".to_owned(), |v| v.text_contents())
+                        .map_or_else(|_| "".to_owned(), |v| v.text_contents()),
+                    party: None
+                };
+                
+                senators.push(person);
+            }
+        }
+    }
+
+    let h3_list = document.select("h3").unwrap();
+    for h3 in h3_list.filter_map(|h| super::with_title(h, "lista nacional".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 (first, rest) = cols.split_at(2);
+            let (second, third) = rest.split_at(2);
+
+            for col in &[first, second, third] {
+                let person = CongressMember::ByList {
+                    name: col[0]
+                        .as_node()
+                        .select_first("a")
+                        .map_or_else(|_| "".to_owned(), |v| v.text_contents()),
+                    party: None
                 };
     
-                println!("{:?}", person);                
+                senators.push(person);
             }
         }
     }
 
-    Ok(())
+    Ok(senators)
 }