parsen von traits und tags, wir koennen den configkram eigentlich wieder komplett entfernen. klappt so nicht wirklich.

7 files changed, 130 insertions, 16 deletions

diff --git a/config.yml b/config.yml
index 07b5d03..313d8e2 100644
--- a/config.yml
+++ b/config.yml
@@ -1,11 +1,25 @@
 sections:
   name:
-    pattern: (?is)Romaji Name.*?<TD>(.*?)\s?</TD>.*?Japanese Name.*?<TD>(.*?)\s?</TD>
+    pattern: (?is)Romaji Name.*?<TD>(.*?)\s?</TD>.*?Japanese Name.*?<TD>(.*?)\s?</TD>.*?Aliases.*?<TD>(.*?)\s?</TD>
     groups:
       - romaji
       - japanese
+      - aliases
+  misc:
+    pattern: (?is)Role</TH>.*?<TD>(.*?)\s?</TD>
+    groups:
+      - role
   image:
     pattern: (?is)<H3 id="section99">.*<img src="(.*?)" alt=.*?></a><p><a href="(.*?)">View Full Size Image
     groups:
       - thumb
       - full
+  tags:
+    pattern: (?is)tagged as</P>.*?<TH>(.*?)</TH>
+    groups:
+      - tags_raw
+  traits:
+    pattern: (?is)indexed traits</P>.*?<dl>(.*?)</dl>.*?official traits\s?</P>.*?<dl>(.*?)</dl>
+    groups:
+      - indexed_raw
+      - official_raw
diff --git a/src/character.rs b/src/character.rs
index 9c548c1..98c4850 100644
--- a/src/character.rs
+++ b/src/character.rs
@@ -1,3 +1,31 @@
+use super::tags::Tag;
+use super::traits::Trait;
+
+#[derive(Debug)]
+pub struct Names {
+    pub romaji: String,
+    pub japanese: String,
+    pub aliases: Vec<String>,
+}
+
+#[derive(Debug)]
+pub struct Images {
+    pub thumb: String,
+    pub full: String,
+}
+
+#[derive(Debug)]
+pub struct Traits {
+    pub official: Vec<Trait>,
+    pub indexed: Vec<Trait>,
+}
+
+#[derive(Debug)]
 pub struct Character {
-    
-}
\ No newline at end of file
+    pub name: Names,
+    pub image: Images,
+    pub tags: Vec<Tag>,
+    pub traits: Traits,
+
+    pub role: Option<String>
+}
diff --git a/src/config.rs b/src/config.rs
index fc8ee03..f491852 100644
--- a/src/config.rs
+++ b/src/config.rs
@@ -15,7 +15,7 @@ pub struct Config {
 }
 
 impl Config {
-    pub fn from_file(p: &str) -> Self {
+    pub fn from_file(p: &str, expected: Vec<&'static str>) -> Self {
         let mut f = File::open(p).unwrap();
         let mut buf = String::new();
         f.read_to_string(&mut buf).unwrap();
@@ -23,8 +23,6 @@ impl Config {
 
         let doc = &docs[0];
 
-        println!("{:?}", doc);
-
         let mut sections: HashMap<String, SectionConfig> = HashMap::new();
         for (name, entry) in doc["sections"].as_hash().unwrap() {
             sections.insert(name.as_str().unwrap().into(),
@@ -39,6 +37,27 @@ impl Config {
                             });
         }
 
+
+        for ex in &expected {
+            if !sections.contains_key(&ex.to_string()) {
+                panic!("config: section '{}' not found", ex);
+            }
+        }
+
+        {
+            let traits = &sections["traits"];
+            if !traits.groups.contains(&"indexed_raw".to_string()) {
+                panic!("config: no group 'indexed_raw' found in section 'traits'");
+            }
+            if !traits.groups.contains(&"official_raw".to_string()) {
+                panic!("config: no group 'official_raw' found in section 'traits'");
+            }
+            let tags = &sections["tags"];
+            if !tags.groups.contains(&"tags_raw".to_string()) {
+                panic!("config: no group 'tags_raw' found in section 'tags'");
+            }
+        }
+
         Config { sections: sections }
     }
 }
diff --git a/src/main.rs b/src/main.rs
index 7d06fe7..ee8c3eb 100644
--- a/src/main.rs
+++ b/src/main.rs
@@ -16,15 +16,21 @@ use pre_process::Section;
 mod config;
 use config::Config;
 
+mod character;
+use character::{Images, Names, Traits, Character};
+
+mod tags;
+mod traits;
+
 fn main() {
     let raw_files = env::var("RAW_FILES").unwrap_or("characters".into());
     let base_path = Path::new(&raw_files);
 
-    let cfg = Config::from_file("config.yml");
+    let cfg = Config::from_file("config.yml", vec!["name", "image", "misc", "tags", "traits"]);
 
-    let mut sections: Vec<Section> = vec![];
+    let mut sections: HashMap<String, Section> = HashMap::new();
     for (name, sec) in &cfg.sections {
-        sections.push(Section::new(&name, &sec.pattern, sec.groups.clone()));
+        sections.insert(name.clone(), Section::new(&name, &sec.pattern, sec.groups.clone()));
     }
 
     for entry in WalkDir::new(base_path).min_depth(1).into_iter().filter_map(|e| e.ok()) {
@@ -39,11 +45,30 @@ fn main() {
 
         pre_process::split_sections(&buf, &mut sections);
 
-        let mut char: HashMap<String, HashMap<String, String>> = HashMap::new();
-        for s in &sections {
-            char.insert(s.name.clone(), s.data.clone());
+        {
+            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: name.data["aliases".into()].split(", ").map(|s| s.to_string()).collect(),
+                },
+                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: traits::parse(&(&sections["traits"] as &Section).data["official_raw".into()]),
+                    indexed: traits::parse(&(&sections["traits"] as &Section).data["indexed_raw".into()]),
+                },
+                role: match misc.data["role".into()].len() > 0 {
+                    true => Some(misc.data["role".into()].clone()),
+                    false => None
+                }
+            });
         }
-
-        println!("{:?}", char);
     }
 }
diff --git a/src/pre_process.rs b/src/pre_process.rs
index a4d6c14..273562d 100644
--- a/src/pre_process.rs
+++ b/src/pre_process.rs
@@ -34,8 +34,8 @@ impl Section {
     }
 }
 
-pub fn split_sections(d: &str, s: &mut Vec<Section>) {
-    for section in s {
+pub fn split_sections(d: &str, s: &mut HashMap<String, Section>) {
+    for (_, section) in s {
         for m in section.re.captures_iter(d) {
             assert!(m.len() >= section.keys.len() + 1);
 
diff --git a/src/tags.rs b/src/tags.rs
new file mode 100644
index 0000000..0fdf815
--- /dev/null
+++ b/src/tags.rs
@@ -0,0 +1,15 @@
+use super::regex::Regex;
+
+use std::str::FromStr;
+
+#[derive(Debug)]
+pub struct Tag {
+    pub id: u32,
+    pub name: String,
+}
+
+pub fn parse(s: &str) -> Vec<Tag> {
+    let reg_tag = Regex::new(r#"(?is)<a href="tags\.php\?id=([0-9]+)">(.*?)</a>"#).unwrap();
+
+    reg_tag.captures_iter(s).map(|c| Tag { id: u32::from_str(c.at(1).unwrap()).unwrap(), name: c.at(2).unwrap().into() }).collect()
+}
diff --git a/src/traits.rs b/src/traits.rs
new file mode 100644
index 0000000..db1b2eb
--- /dev/null
+++ b/src/traits.rs
@@ -0,0 +1,13 @@
+use super::regex::Regex;
+
+#[derive(Debug)]
+pub struct Trait {
+    name: String,
+    value: String,
+}
+
+pub fn parse(s: &str) -> Vec<Trait> {
+    let reg_trait = Regex::new(r#"(?is)<dt.*?>(.*?)</dt>.*?<dd>(.*?)</dd>"#).unwrap();
+
+    reg_trait.captures_iter(s).map(|c| Trait { name: c.at(1).unwrap().into(), value: c.at(2).unwrap().into() }).collect()
+}