Update CI

Author: SamarthPyati

.github/workflows/rust.yml

@@ -1,4 +1,4 @@
-name: Rust
+name: DocSense CI
 
 on:
   push:
@@ -16,7 +16,6 @@ jobs:
     steps:
     - uses: actions/checkout@v4
 
-    # Caching deps before build 
     - name: Cache Cargo dependencies
       uses: actions/cache@v4
       with:
@@ -35,7 +34,21 @@ jobs:
         override: true
 
     - name: Install system dependencies
-      run: sudo apt-get update && sudo apt-get install -y libcairo2-dev libglib2.0-dev pkg-config
+      run: |
+        sudo apt-get update
+        sudo apt-get install -y libpoppler-glib-dev pkg-config
+        PKG_PATH=$(dpkg -L libpoppler-glib-dev | grep 'poppler-glib.pc' || true)
+        echo "Found poppler-glib.pc at: $PKG_PATH"
+        if [ -z "$PKG_PATH" ]; then
+          echo "ERROR: poppler-glib.pc not found!" && exit 1
+        fi
+        export PKG_CONFIG_PATH=$(dirname $PKG_PATH)
+        echo "PKG_CONFIG_PATH=$PKG_CONFIG_PATH" >> $GITHUB_ENV
+
+    - name: Verify pkg-config
+      run: |
+        echo "PKG_CONFIG_PATH=${PKG_CONFIG_PATH}"
+        pkg-config --modversion poppler-glib
 
     - name: Build (Debug)
       run: cargo build --verbose

src/lexer.rs

@@ -59,14 +59,6 @@ impl<'a> Lexer<'a> {
             return Some(stemmed_token);
         }
 
-        // // Ignore single-character unwanted punctuation
-        // let unwanted_symbols  = &[',', ';', '*', '/', '?', '{', '}', '(', ')', '.', '$', '_', '-'];
-        
-        // if unwanted_symbols.contains(&self.content[0]) {
-        //     self.chop(1);   // skip this token 
-        //     return self.next_token();     // recursively fetch next token 
-        // }
-
         let token = self.chop(1);
         return Some(token.iter().collect());
     }

src/main.rs

@@ -64,7 +64,6 @@ fn parse_pdf_files(file_path: &Path) -> Result<String, ()> {
         eprintln!("{}: Failed to make poppler document out of PDF file {path} as {err}", "ERROR".red().bold(), path = file_path.to_string_lossy().bright_blue(), err = err.to_string().red());
     })?;
 
-
     let mut content = String::new();
     if pdf_file.n_pages() > 0 {
         for i in 0..pdf_file.n_pages() { 

src/model.rs

@@ -1,5 +1,5 @@
 use std::{
-    fs, path::{Path, PathBuf}, time::SystemTime
+    path::{Path, PathBuf}, time::SystemTime
 };
 
 use std::collections::HashMap;
@@ -294,7 +294,6 @@ impl Model for InMemoryModel {
         }
         return Ok(true);
     }
-
 }
 
 // TODO: Implement a efficient sqlite Model with parellel processing support