2 files changed, 29 insertions, 0 deletions

diff --git a/src/main.rs b/src/main.rs
index 066a07a..b733e6f 100644
--- a/src/main.rs
+++ b/src/main.rs
@@ -10,11 +10,19 @@ use std::env;
 use std::path::{Path, PathBuf};
 
 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 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(),
+    });
+
     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");
         let mut buf = String::new();
@@ -24,5 +32,9 @@ fn main() {
         }
 
         let buf = pre_process::strip_irrelevant_content(&buf);
+
+        pre_process::split_sections(&buf, &mut sections);
+
+        println!("{}", sections[0].content);
     }
 }
diff --git a/src/pre_process.rs b/src/pre_process.rs
index aa55bd8..d69cfce 100644
--- a/src/pre_process.rs
+++ b/src/pre_process.rs
@@ -1,3 +1,5 @@
+use super::regex::Regex;
+
 pub fn strip_irrelevant_content(s: &str) -> String {
     let mut retn = "";
     match s.find(r#"<div class=profile id=profile>"#) {
@@ -10,4 +12,19 @@ pub fn strip_irrelevant_content(s: &str) -> String {
         None => (),
     };
     return retn.into();
+}
+
+pub struct Section {
+    pub name: String,
+    pub re: Regex,
+    pub content: String,
+}
+
+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());
+        }
+    }
 }
\ No newline at end of file