Fixed Naive method for latest PDFs

YM162 · Sep 28, 2024 · 7836e7e · 7836e7e
1 parent 3b98ac6
commit 7836e7e
Show file tree

Hide file tree

Showing 9 changed files with 27 additions and 10 deletions.
diff --git a/gulagcleaner_python/Cargo.toml b/gulagcleaner_python/Cargo.toml
@@ -1,6 +1,6 @@
 [package]
 name = "gulagcleaner_python"
-version = "0.14.3"
+version = "0.14.4"
 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 = "0.14.3"
+version = "4"
 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: 0.14.3")
+        print("Current version: 0.14.4")
         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 = "0.14.3"
+version = "0.14.4"
 edition = "2021"
 authors = ["YM162 <[email protected]>"]
 description = "Ad removal tool for PDFs."

diff --git a/gulagcleaner_rs/example_docs/wuolah-280924-2-example.pdf b/gulagcleaner_rs/example_docs/wuolah-280924-2-example.pdf
diff --git a/gulagcleaner_rs/src/models/method.rs b/gulagcleaner_rs/src/models/method.rs
@@ -124,6 +124,7 @@ impl Cleaner for Method {
                 for page in &pages {
                     let page_type =
                         page_type::PageType::get_page_type(doc, page.1).unwrap_or_default();
+                        println!("{:?}", page_type);
                     let mutable_page = doc.get_object_mut(*page.1).unwrap().as_dict_mut().unwrap();
 
                     let mediabox = mutable_page.get(b"MediaBox").unwrap().as_array().unwrap();
@@ -245,7 +246,7 @@ pub fn find_iobj_pairs(first_page: &[(u32, u16)], second_page: &[(u32, u16)]) ->
     if len < 2 {
         return (0, 0);
     }
-    
+
     (indexes[len - 3], indexes[len - 2])
 }
 

diff --git a/gulagcleaner_rs/src/models/page_type.rs b/gulagcleaner_rs/src/models/page_type.rs
@@ -5,6 +5,7 @@ use lopdf::{Document, ObjectId};
 use super::method::{get_images, get_xobjs};
 
 #[derive(Default)]
+#[derive(Debug)]
 /// Represents the different methods used in the Gulag Cleaner application.
 pub enum PageType {
     BannerAds,
@@ -14,18 +15,20 @@ pub enum PageType {
     Idk,
 }
 
-pub const LOGO_DIMS: [(i64, i64); 3] = [(71, 390), (37, 203), (73, 390)];
+pub const LOGO_DIMS: [(i64, i64); 6] = [(71, 390), (37, 203), (73, 390),(23,130),(19,109),(72,391)];
 
-const HORIZONTAL_BANNER_DIMS: [(i64, i64); 7] = [
+const HORIZONTAL_BANNER_DIMS: [(i64, i64); 9] = [
     (247, 1414),
     (213, 1219),
     (215, 1219),
     (249, 1414),
     (217, 1240),
     (147, 1757),
     (221, 1240),
+    (136, 780),
+    (218,1241)
 ];
-const VERTICAL_BANNER_DIMS: [(i64, i64); 8] = [
+const VERTICAL_BANNER_DIMS: [(i64, i64); 10] = [
     (1753, 170),
     (1518, 248),
     (1520, 147),
@@ -34,15 +37,19 @@ const VERTICAL_BANNER_DIMS: [(i64, i64); 8] = [
     (1537, 147),
     (1093, 217),
     (1534, 150),
+    (970, 92),
+    (1538, 148)
 ];
-const FULL_PAGE_DIMS: [(i64, i64); 7] = [
+const FULL_PAGE_DIMS: [(i64, i64); 9] = [
     (842, 595),
     (1754, 1240),
     (2526, 1785),
     (1733, 1219),
     (3508, 2480),
     (2339, 1653),
     (1785, 2526),
+    (1109, 782),
+    (1759,1241)
 ];
 
 impl PageType {
@@ -64,6 +71,7 @@ impl PageType {
     pub fn get_page_type(doc: &Document, page: &ObjectId) -> Result<PageType, Box<dyn Error>> {
         let xobjs = get_xobjs(doc, page)?;
         let images = get_images(doc, xobjs)?;
+        println!("{:?}", images);
         let has_logo = !LOGO_DIMS
             .iter()
             .collect::<HashSet<_>>()

diff --git a/gulagcleaner_rs/src/tests.rs b/gulagcleaner_rs/src/tests.rs
@@ -113,6 +113,14 @@ fn test_280924_wuolah_pdf() {
     });
 }
 
+#[test]
+fn test_280924_2_wuolah_pdf() {
+    run_test_for_config(&TestConfig {
+        input_path: "example_docs/wuolah-280924-2-example.pdf",
+        output_filename: "wuolah-280924-2-example_clean.pdf",
+    });
+}
+
 #[test]
 fn test_studocu_pdf() {
     run_test_for_config(&TestConfig {

diff --git a/gulagcleaner_wasm/Cargo.toml b/gulagcleaner_wasm/Cargo.toml
@@ -1,6 +1,6 @@
 [package]
 name = "gulagcleaner_wasm"
-version = "0.14.3"
+version = "0.14.4"
 edition = "2021"
 authors = ["YM162 <[email protected]>"]
 description = "Ad removal tool for PDFs."