assignments & chars mit aehnlichen traits

3 files changed, 98 insertions, 48 deletions

diff --git a/src/character.rs b/src/character.rs
index b82ac18..b0322dd 100644
--- a/src/character.rs
+++ b/src/character.rs
@@ -1,6 +1,11 @@
+use super::tags;
 use super::tags::Tag;
+use super::dl_list;
 use super::dl_list::DLListItem;
+use super::section;
 use super::section::Section;
+use super::regex::Regex;
+use super::tiles;
 
 use std::collections::HashMap;
 
@@ -29,12 +34,86 @@ pub struct Character {
     pub image: Images,
     pub tags: Vec<Tag>,
     pub traits: Traits,
+    pub assignments: Vec<u32>,
+    pub chars_similar_traits: Vec<u32>,
     pub extra: Vec<DLListItem>,
 
-    pub role: Option<String>
+    pub role: Option<String>,
 }
 
-pub fn get_sections() -> HashMap<String, Section> {
+impl Names {
+    pub fn new() -> Self {
+        Names { romaji: String::new(), japanese: String::new(), aliases: vec![] }
+    }
+}
+
+impl Images {
+    pub fn new() -> Self {
+        Images { thumb: String::new(), full: String::new() }
+    }
+}
+
+impl Traits {
+    pub fn new() -> Self {
+        Traits { official: vec![], indexed: vec![] }
+    }
+}
+
+impl Character {
+    pub fn new() -> Self {
+        Character {
+            name: Names::new(),
+            image: Images::new(),
+            tags: vec![],
+            traits: Traits::new(),
+            assignments: vec![],
+            chars_similar_traits: vec![],
+            extra: vec![],
+            role: None
+        }
+    }
+
+    pub fn parse(&mut self, buf: &str) {
+        let re_extras = Regex::new(r#"(?is)Extra Details \| [0-9]+</H3>.*?<dl>(.*?)</dl>"#).unwrap();
+        let mut sections = get_sections();    
+        section::process(&buf, &mut sections);
+
+        let caps = re_extras.captures(&buf);
+        if caps.is_some() {
+            self.extra = dl_list::parse(caps.unwrap().at(1).unwrap());
+        }
+
+        {
+            let name: &Section = &sections["name".into()];
+            let image: &Section = &sections["image".into()];
+            let misc: &Section = &sections["misc".into()];
+            
+            self.name.romaji = name.data["romaji".into()].clone();
+            self.name.japanese = name.data["japanese".into()].clone();
+            
+            if name.data["aliases".into()].len() > 0 {
+                self.name.aliases = name.data["aliases".into()].split(", ").map(|s| s.to_string()).collect();
+            }
+            
+            self.image.thumb = image.data["thumb".into()].clone();
+            self.image.full = image.data["full".into()].clone();
+            
+            self.tags = tags::parse(&(&sections["tags".into()] as &Section).data["tags_raw".into()]);
+
+            self.traits.official = dl_list::parse(&(&sections["traits"] as &Section).data["official_raw".into()]);
+            self.traits.indexed = dl_list::parse(&(&sections["traits"] as &Section).data["indexed_raw".into()]);
+
+            self.assignments = tiles::parse_tile_link_ids(&(&sections["assignments"] as &Section).data["raw".into()], "series");
+            self.chars_similar_traits = tiles::parse_tile_link_ids(&(&sections["chars_similar_traits"] as &Section).data["raw".into()], "character");
+            
+            if misc.data["role".into()].len() > 0 {
+                self.role = Some(misc.data["role".into()].clone());
+            }
+        }
+   }
+}
+
+fn get_sections() -> HashMap<String, Section> {
     let mut s: HashMap<String, Section> = HashMap::new();
 
     s.insert("name".into(), Section::new("name", r#"(?is)Romaji Name.*?<TD>(.*?)\s?</TD>.*?Japanese Name.*?<TD>(.*?)\s?</TD>.*?Aliases.*?<TD>(.*?)\s?</TD>"#, vec!["romaji", "japanese", "aliases"]));
@@ -42,5 +121,8 @@ pub fn get_sections() -> HashMap<String, Section> {
     s.insert("image".into(), Section::new("image", r#"(?is)<H3 id="section99">.*<img src="(.*?)" alt=.*?></a><p><a href="(.*?)">View Full Size Image"#, vec!["thumb", "full"]));
     s.insert("tags".into(), Section::new("tags", r#"(?is)tagged as</P>.*?<TH>(.*?)</TH>"#, vec!["tags_raw"]));
     s.insert("traits".into(), Section::new("traits", r#"(?is)indexed traits</P>.*?<dl>(.*?)</dl>.*?official traits\s?</P>.*?<dl>(.*?)</dl>"#, vec!["indexed_raw", "official_raw"]));
+    s.insert("assignments".into(), Section::new("assignments", r#"(?is)appears in the following</P>(.*?)</UL>"#, vec!["raw"]));
+    s.insert("chars_similar_traits".into(), Section::new("assignments", r#"(?is)with Similar Traits</H3>(.*?)</UL>"#, vec!["raw"]));
+
     s
 }
diff --git a/src/main.rs b/src/main.rs
index e76da66..ea89b02 100644
--- a/src/main.rs
+++ b/src/main.rs
@@ -1,5 +1,4 @@
 extern crate regex;
-use regex::Regex;
 extern crate yaml_rust;
 
 extern crate walkdir;
@@ -9,26 +8,21 @@ use std::io::prelude::*;
 use std::fs::File;
 use std::env;
 use std::path::Path;
-use std::collections::HashMap;
 
 mod pre_process;
 mod section;
-use section::Section;
 
 mod character;
-use character::{Images, Names, Traits, Character};
+use character::Character;
 
 mod tags;
 mod dl_list;
-use dl_list::DLListItem;
+mod tiles;
 
 fn main() {
     let raw_files = env::var("RAW_FILES").unwrap_or("characters".into());
     let base_path = Path::new(&raw_files);
 
-    let re_extras = Regex::new(r#"(?is)Extra Details \| [0-9]+</H3>.*?<dl>(.*?)</dl>"#).unwrap();
-
-    let mut sections: HashMap<String, Section> = character::get_sections();
 
     for entry in WalkDir::new(base_path).min_depth(1).into_iter().filter_map(|e| e.ok()) {
         let mut f = File::open(entry.path()).expect("could not open file");
@@ -40,43 +34,8 @@ fn main() {
 
         let buf = pre_process::strip_irrelevant_content(&buf);
 
-        section::process(&buf, &mut sections);
-
-        // find optional extra details
-        let mut extra_details: Vec<DLListItem> = vec![];
-        let caps = re_extras.captures(&buf);
-        if caps.is_some() {
-            extra_details = dl_list::parse(caps.unwrap().at(1).unwrap());
-        }
-
-        {
-            let name: &Section = &sections["name".into()];
-            let image: &Section = &sections["image".into()];
-            let misc: &Section = &sections["misc".into()];
-            println!("{:?}", Character {
-                name: Names {
-                    romaji: name.data["romaji".into()].clone(),
-                    japanese: name.data["japanese".into()].clone(),
-                    aliases: match name.data["aliases".into()].len() > 0 {
-                        true => name.data["aliases".into()].split(", ").map(|s| s.to_string()).collect(),
-                        false => vec![]
-                    }
-                },
-                image: Images {
-                    thumb: image.data["thumb".into()].clone(),
-                    full: image.data["full".into()].clone(),
-                },
-                tags: tags::parse(&(&sections["tags".into()] as &Section).data["tags_raw".into()]),
-                traits: Traits {
-                    official: dl_list::parse(&(&sections["traits"] as &Section).data["official_raw".into()]),
-                    indexed: dl_list::parse(&(&sections["traits"] as &Section).data["indexed_raw".into()]),
-                },
-                extra: extra_details,
-                role: match misc.data["role".into()].len() > 0 {
-                    true => Some(misc.data["role".into()].clone()),
-                    false => None
-                }
-            });
-        }
+        let mut char = Character::new();
+        char.parse(&buf);
+        println!("{:?}", char);
     }
 }
diff --git a/src/tiles.rs b/src/tiles.rs
new file mode 100644
index 0000000..1ed7320
--- /dev/null
+++ b/src/tiles.rs
@@ -0,0 +1,9 @@
+use super::regex::Regex;
+
+use std::str::FromStr;
+
+pub fn parse_tile_link_ids(s: &str, php_file: &str) -> Vec<u32> {
+    let re = Regex::new(&format!(r#"(?is)<A href="{}\.php\?id=([0-9]+)"><IMG"#, php_file)).unwrap();
+
+    re.captures_iter(s).map(|cap| u32::from_str(cap.at(1).unwrap()).unwrap()).collect()
+}