Moved to a locally patched docx-rs and made it so the excluded chars are read from a file

.gitignore

@@ -1,4 +1,4 @@
 /target
-
+excluded
 *.docx
 *.txt

Cargo.lock

@@ -22,9 +22,9 @@ checksum = "ace50bade8e6234aa140d9a2f552bbee1db4d353f69b8217bc503490fc1a9f26"
 
 [[package]]
 name = "base64"
-version = "0.13.1"
+version = "0.22.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "9e1b586273c5702936fe7b7d6896644d8be71e6314cfe09d3167c95f712589e8"
+checksum = "72b3254f16251a8381aa12e40e3c4d2f0199f8c6508fbecb9d91f575e0fbb8c6"
 
 [[package]]
 name = "bitflags"
@@ -82,9 +82,7 @@ dependencies = [
 
 [[package]]
 name = "docx-rs"
-version = "0.4.17"
+version = "0.4.18-rc19"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "e593b51d4fe95d69d70fd40da4b314b029736302c986c3c760826e842fd27dc3"
 dependencies = [
  "base64",
  "image",

Cargo.toml

@@ -5,5 +5,6 @@ edition = "2021"
 
 [dependencies]
 anyhow = "1.0.98"
-docx-rs = "0.4.17"
+docx-rs = { path = "../../Tests/docx-rs/docx-core" }
+# docx-rs = "0.4.17"
 serde_json = "1.0.140"

src/main.rs

@@ -1,5 +1,5 @@
 use std::{
-    env,
+    env, fs,
     io::{stdout, Read, Write},
 };
 
@@ -12,50 +12,30 @@ fn main() -> anyhow::Result<()> {
 
     let mut lines: Vec<String> = vec![];
 
-    // for docx
     parse_docx(file_name, &mut lines)?;
 
-    // for txt
+    let ascii_upper_limit = u32::from_str_radix("7F", 16).unwrap();
-    // for line in fs::read_to_string(file_name).unwrap().lines() {
+    let mut excluded = vec![];
-    //     if !line.is_empty() {
+    for line in fs::read_to_string("excluded")?.lines() {
-    //         lines.push(line.to_string())
+        if let Some(c) = line.chars().next() {
-    //     }
+            let c = format!("{:x}", c as u32);
-    // }
+            let u = u32::from_str_radix(&c, 16)?;
+            excluded.push(u);
+        }
+    }
 
     let mut lock = stdout().lock();
 
-    let ascii_upper_limit = u32::from_str_radix("7F", 16).unwrap();
-    let excluded = [
-        "0000021B", // ț
-        "00000219", // ș
-        "00000103", // ă
-        "000000E2", // â
-        "000000EE", // î
-        "0000021A", // Ț
-        "00000218", // Ș
-        "00000102", // Ă
-        "000000C2", // Â
-        "000000CE", // Î
-        "0000201E", // „
-        "0000201D", // ”
-        "0000201A", // ‚
-        "00002019", // ’
-    ];
-    let excluded: Vec<u32> = excluded
-        .into_iter()
-        .map(|c| u32::from_str_radix(c, 16).unwrap())
-        .collect();
-
     for line in lines {
         for c in line.chars() {
             let uc = c as u32;
             if uc > ascii_upper_limit && !excluded.contains(&uc) {
-                writeln!(lock, "{c} - {line}").unwrap();
+                writeln!(lock, "{c} - {line}")?;
             }
         }
     }
 
-    lock.flush().unwrap();
+    lock.flush()?;
 
     Ok(())
 }