config entfernt, extra details parsen

7 files changed, 87 insertions, 127 deletions

diff --git a/src/character.rs b/src/character.rs
index 98c4850..b82ac18 100644
--- a/src/character.rs
+++ b/src/character.rs
@@ -1,5 +1,8 @@
 use super::tags::Tag;
-use super::traits::Trait;
+use super::dl_list::DLListItem;
+use super::section::Section;
+
+use std::collections::HashMap;
 
 #[derive(Debug)]
 pub struct Names {
@@ -16,8 +19,8 @@ pub struct Images {
 
 #[derive(Debug)]
 pub struct Traits {
-    pub official: Vec<Trait>,
-    pub indexed: Vec<Trait>,
+    pub official: Vec<DLListItem>,
+    pub indexed: Vec<DLListItem>,
 }
 
 #[derive(Debug)]
@@ -26,6 +29,18 @@ pub struct Character {
     pub image: Images,
     pub tags: Vec<Tag>,
     pub traits: Traits,
+    pub extra: Vec<DLListItem>,
 
     pub role: Option<String>
 }
+
+pub 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"]));
+    s.insert("misc".into(), Section::new("misc", r#"(?is)Role</TH>.*?<TD>(.*?)\s?</TD>"#, vec!["role"]));
+    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
+}
diff --git a/src/config.rs b/src/config.rs
deleted file mode 100644
index f491852..0000000
--- a/src/config.rs
+++ /dev/null
@@ -1,63 +0,0 @@
-extern crate yaml_rust;
-use self::yaml_rust::YamlLoader;
-
-use std::collections::HashMap;
-use std::fs::File;
-use std::io::prelude::*;
-
-pub struct SectionConfig {
-    pub pattern: String,
-    pub groups: Vec<String>,
-}
-
-pub struct Config {
-    pub sections: HashMap<String, SectionConfig>,
-}
-
-impl Config {
-    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();
-        let docs = YamlLoader::load_from_str(&buf).unwrap();
-
-        let doc = &docs[0];
-
-        let mut sections: HashMap<String, SectionConfig> = HashMap::new();
-        for (name, entry) in doc["sections"].as_hash().unwrap() {
-            sections.insert(name.as_str().unwrap().into(),
-                            SectionConfig {
-                                pattern: entry["pattern"].as_str().unwrap().into(),
-                                groups: entry["groups"]
-                                    .as_vec()
-                                    .unwrap()
-                                    .into_iter()
-                                    .map(|v| v.as_str().unwrap().into())
-                                    .collect(),
-                            });
-        }
-
-
-        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/dl_list.rs b/src/dl_list.rs
new file mode 100644
index 0000000..979c332
--- /dev/null
+++ b/src/dl_list.rs
@@ -0,0 +1,13 @@
+use super::regex::Regex;
+
+#[derive(Debug)]
+pub struct DLListItem {
+    name: String,
+    value: String,
+}
+
+pub fn parse(s: &str) -> Vec<DLListItem> {
+    let reg_list_item = Regex::new(r#"(?is)<dt.*?>(.*?)</dt>.*?<dd>(.*?)</dd>"#).unwrap();
+
+    reg_list_item.captures_iter(s).map(|c| DLListItem { name: c.at(1).unwrap().into(), value: c.at(2).unwrap().into() }).collect()
+}
diff --git a/src/main.rs b/src/main.rs
index ee8c3eb..e76da66 100644
--- a/src/main.rs
+++ b/src/main.rs
@@ -1,4 +1,5 @@
 extern crate regex;
+use regex::Regex;
 extern crate yaml_rust;
 
 extern crate walkdir;
@@ -11,27 +12,23 @@ use std::path::Path;
 use std::collections::HashMap;
 
 mod pre_process;
-use pre_process::Section;
-
-mod config;
-use config::Config;
+mod section;
+use section::Section;
 
 mod character;
 use character::{Images, Names, Traits, Character};
 
 mod tags;
-mod traits;
+mod dl_list;
+use dl_list::DLListItem;
 
 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", vec!["name", "image", "misc", "tags", "traits"]);
+    let re_extras = Regex::new(r#"(?is)Extra Details \| [0-9]+</H3>.*?<dl>(.*?)</dl>"#).unwrap();
 
-    let mut sections: HashMap<String, Section> = HashMap::new();
-    for (name, sec) in &cfg.sections {
-        sections.insert(name.clone(), Section::new(&name, &sec.pattern, sec.groups.clone()));
-    }
+    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");
@@ -43,7 +40,14 @@ fn main() {
 
         let buf = pre_process::strip_irrelevant_content(&buf);
 
-        pre_process::split_sections(&buf, &mut sections);
+        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()];
@@ -53,7 +57,10 @@ fn main() {
                 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(),
+                    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(),
@@ -61,9 +68,10 @@ fn main() {
                 },
                 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()]),
+                    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
diff --git a/src/pre_process.rs b/src/pre_process.rs
index 273562d..877ddc3 100644
--- a/src/pre_process.rs
+++ b/src/pre_process.rs
@@ -1,7 +1,3 @@
-use super::regex::Regex;
-
-use std::collections::HashMap;
-
 pub fn strip_irrelevant_content(s: &str) -> String {
     let mut retn = "";
     match s.find(r#"<div class=profile id=profile>"#) {
@@ -16,34 +12,4 @@ pub fn strip_irrelevant_content(s: &str) -> String {
     return retn.into();
 }
 
-pub struct Section {
-    pub name: String,
-    pub re: Regex,
-    pub keys: Vec<String>,
-    pub data: HashMap<String, String>,
-}
 
-impl Section {
-    pub fn new(name: &str, re: &str, groups: Vec<String>) -> Self {
-        Section {
-            name: name.into(),
-            re: Regex::new(re).unwrap(),
-            keys: groups,
-            data: HashMap::new(),
-        }
-    }
-}
-
-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);
-
-            let mut idx = 0;
-            for key in &section.keys {
-                section.data.insert(key.clone(), m.at(idx + 1).unwrap().into());
-                idx += 1;
-            }
-        }
-    }
-}
diff --git a/src/section.rs b/src/section.rs
new file mode 100644
index 0000000..7e492b1
--- /dev/null
+++ b/src/section.rs
@@ -0,0 +1,34 @@
+use super::regex::Regex;
+use std::collections::HashMap;
+
+pub struct Section {
+    pub name: String,
+    pub re: Regex,
+    pub keys: Vec<String>,
+    pub data: HashMap<String, String>,
+}
+
+impl Section {
+    pub fn new(name: &str, re: &str, groups: Vec<&'static str>) -> Self {
+        Section {
+            name: name.into(),
+            re: Regex::new(re).unwrap(),
+            keys: groups.into_iter().map(|s| s.into()).collect(),
+            data: HashMap::new(),
+        }
+    }
+}
+
+pub fn process(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);
+
+            let mut idx = 0;
+            for key in &section.keys {
+                section.data.insert(key.clone(), m.at(idx + 1).unwrap().into());
+                idx += 1;
+            }
+        }
+    }
+}
diff --git a/src/traits.rs b/src/traits.rs
deleted file mode 100644
index db1b2eb..0000000
--- a/src/traits.rs
+++ /dev/null
@@ -1,13 +0,0 @@
-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()
-}