From 9e59cd7e6a21751420ccbb853ac883154c6e578e Mon Sep 17 00:00:00 2001
From: jan <jan@ruken.pw>
Date: Thu, 29 Sep 2016 13:10:20 +0200
Subject: dynamischere sektionssuche, jetzt auch mit thumb images. koennten wir
 eigentlich in irgendeine config-datei auslagern.


diff --git a/src/main.rs b/src/main.rs
index b733e6f..2123c6b 100644
--- a/src/main.rs
+++ b/src/main.rs
@@ -13,15 +13,11 @@ mod pre_process;
 use pre_process::Section;
 
 fn main() {
-    let raw_files = env::var("RAW_FILES").unwrap_or("S:\\grilist\\acd\\acd_character_parser\\characters\\".into());
+    let raw_files = env::var("RAW_FILES").unwrap_or("characters".into());
     let base_path = Path::new(&raw_files);
 
     let mut sections: Vec<Section> = vec![];
-    sections.push(Section {
-        name: "image".into(),
-        re: Regex::new(r#"(?is)<H3 id="section99">.*<p><a href="(.*?)">View Full Size Image"#).unwrap(),
-        content: String::new(),
-    });
+    sections.push(Section::new("image", r#"(?is)<H3 id="section99">.*<img src="(.*?)" alt=.*?></a><p><a href="(.*?)">View Full Size Image"#, vec!["thumb".into(), "full".into()]));
 
     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");
@@ -35,6 +31,6 @@ fn main() {
 
         pre_process::split_sections(&buf, &mut sections);
 
-        println!("{}", sections[0].content);
+        println!("{:?}", sections[0].data);
     }
 }
diff --git a/src/pre_process.rs b/src/pre_process.rs
index d69cfce..0d8c6be 100644
--- a/src/pre_process.rs
+++ b/src/pre_process.rs
@@ -1,5 +1,7 @@
 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>"#) {
@@ -17,14 +19,31 @@ pub fn strip_irrelevant_content(s: &str) -> String {
 pub struct Section {
     pub name: String,
     pub re: Regex,
-    pub content: String,
+    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 Vec<Section>) {
     for section in s {
         for m in section.re.captures_iter(d) {
-            assert!(m.len() > 1);
-            section.content = format!("{}", m.at(1).unwrap());
+            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;
+			}
         }
     }
-}
\ No newline at end of file
+}
-- 
cgit v0.10.1