Re-enabled Wuolah method

YM162 · Jan 31, 2024 · 81e7aa8 · 81e7aa8
1 parent 246f5e8
commit 81e7aa8
Show file tree

Hide file tree

Showing 8 changed files with 18 additions and 16 deletions.
diff --git a/gulagcleaner_python/Cargo.toml b/gulagcleaner_python/Cargo.toml
@@ -1,6 +1,6 @@
 [package]
 name = "gulagcleaner_python"
-version = "0.12.1"
+version = "0.12.2"
 edition = "2021"
 authors = ["YM162 <[email protected]>"]
 description = "Ad removal tool for PDFs."

diff --git a/gulagcleaner_python/pyproject.toml b/gulagcleaner_python/pyproject.toml
@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "gulagcleaner"
-version = "0.12.1"
+version = "0.12.2"
 description = "Ad removal tool for PDFs."
 authors = [
   {name = "YM162", email = "[email protected]"}]

diff --git a/gulagcleaner_python/python/gulagcleaner/command_line.py b/gulagcleaner_python/python/gulagcleaner/command_line.py
@@ -61,7 +61,7 @@ def main():
 
     # Check for the -v argument
     if arguments["version"]:
-        print("Current version: 0.12.1")
+        print("Current version: 0.12.2")
         return
 
     # Get the pdf_path argument

diff --git a/gulagcleaner_rs/Cargo.toml b/gulagcleaner_rs/Cargo.toml
@@ -1,6 +1,6 @@
 [package]
 name = "gulagcleaner_rs"
-version = "0.12.1"
+version = "0.12.2"
 edition = "2021"
 authors = ["YM162 <[email protected]>"]
 description = "Ad removal tool for PDFs."

diff --git a/gulagcleaner_rs/example_docs/wuolah-free-example.pdf b/gulagcleaner_rs/example_docs/wuolah-free-example.pdf
diff --git a/gulagcleaner_rs/src/clean.rs b/gulagcleaner_rs/src/clean.rs
@@ -71,13 +71,12 @@ fn match_method(doc: &Document, force_naive: bool) -> Method {
         .map(|x| doc.get_page_contents(*x.1))
         .filter(|x| x.len() > 1)
         .collect();
-    //let to_delete: Vec<u32> = pages
-    let _: Vec<u32> = pages
+    let to_delete: Vec<u32> = pages
         .iter()
         .filter(|x| {
             let contents = doc.get_page_contents(*x.1);
 
-            contents.len() == 1
+            contents.len() < 4
         })
         .map(|x| *x.0)
         .collect();
@@ -88,23 +87,26 @@ fn match_method(doc: &Document, force_naive: bool) -> Method {
         .filter(|x| *x == 3)
         .collect::<Vec<_>>()
         .len()
-        > 1
+        == pages.len()
     {
         return Method::StuDocu(content_list);
     }
+    let long_content_list: Vec<Vec<(u32, u16)>> = pages
+        .iter()
+        .map(|x| doc.get_page_contents(*x.1))
+        .filter(|x| x.len() > 3)
+        .collect();
 
-    if content_list.len() > 1
-        && content_list[0]
+    if long_content_list.len() > 1
+        && long_content_list[0]
             .iter()
             .collect::<HashSet<_>>()
-            .intersection(&content_list[1].iter().collect::<HashSet<_>>())
+            .intersection(&long_content_list[1].iter().collect::<HashSet<_>>())
             .collect::<Vec<_>>()
             .len()
             > 1
     {
-        //return Method::Wuolah(content_list, to_delete);
-        //SEE COMMENT AT THE TOP OF THE FUNCTION
-        return Method::Naive;
+        return Method::Wuolah(long_content_list, to_delete);
     }
     Method::Naive
 }
diff --git a/gulagcleaner_rs/src/models/method.rs b/gulagcleaner_rs/src/models/method.rs
@@ -46,7 +46,7 @@ impl Cleaner for Method {
 
                 let vector: Vec<(&u32, &(u32, u16))> = pages
                     .iter()
-                    .filter(|x| doc.get_page_contents(*x.1).len() > 1)
+                    .filter(|x| doc.get_page_contents(*x.1).len() > 3)
                     .collect();
                 for (i, page) in vector.iter().enumerate() {
                     let mutable_page = doc.get_object_mut(*page.1).unwrap().as_dict_mut().unwrap();

diff --git a/gulagcleaner_wasm/Cargo.toml b/gulagcleaner_wasm/Cargo.toml
@@ -1,6 +1,6 @@
 [package]
 name = "gulagcleaner_wasm"
-version = "0.12.1"
+version = "0.12.2"
 edition = "2021"
 authors = ["YM162 <[email protected]>"]
 description = "Ad removal tool for PDFs."