From 9899a747d29bdeb717f796624bcbd6ed2c7c02fa Mon Sep 17 00:00:00 2001
From: sudeeptarlekar <sudeeptarlekar@gmail.com>
Date: Wed, 18 Oct 2023 11:21:59 +0200
Subject: [PATCH] [DO NOT MERGE] Detect file type

Rustcore should have functionality to detect the file type. If
file type is binary or text, and if text what is the encoding of
file.

Resolves: #1841
---
 .../apps/indexer/indexer_cli/src/main.rs      | 57 +++++++++++++++++++
 1 file changed, 57 insertions(+)

diff --git a/application/apps/indexer/indexer_cli/src/main.rs b/application/apps/indexer/indexer_cli/src/main.rs
index 79ad342ab..800a9b7a9 100644
--- a/application/apps/indexer/indexer_cli/src/main.rs
+++ b/application/apps/indexer/indexer_cli/src/main.rs
@@ -1799,3 +1799,60 @@ async fn detect_messages_type(input: &Path) -> Result<bool, DltParseError> {
         )))
     }
 }
+
+// Idea is to get the file path from the user,
+// read the first chunks from the file and return String
+// 'binary' or 'text' as of now. Later we can add enum for file
+// type.
+async fn detect_file_type(file_path: &Path) -> String {
+    let chunks = fetch_starting_chunk(file_path);
+    let buffer = match chunks {
+        Ok(buffer) => buffer,
+        // TODO: add correct error handling
+        Err(error) => panic!("Error while reading chunks from file: {}", error),
+    };
+    // println!("{:?}", buffer);
+    let result: bool = is_file_binary(buffer);
+
+    return String::from("Still in progress...");
+}
+
+fn fetch_starting_chunk(file_path: &Path) -> Result<Vec<u8>>{
+    let file = File::open(file_path)?;
+    let mut file = file.take(1024);
+    let mut buffer = Vec::new();
+    file.read_to_end(&mut buffer)?;
+    Ok(buffer)
+}
+
+fn is_printable(char_encoding: &u8) -> bool {
+    return match char_encoding {
+        32..=126 => true,
+        b'\r' | b'\n' | b'\t' => true,
+        _ => false
+    };
+}
+
+fn is_high_ascii(char_encoding: &u8) -> bool {
+    match char_encoding {
+        127..=255 => true,
+        _ => false,
+    }
+}
+
+fn is_file_binary(buffer: Vec<u8>) -> bool {
+    if buffer.is_empty() {
+        return false;
+    }
+
+    let buffer_length: f32 = buffer.len() as f32;
+
+    let non_printable_chars_count = buffer.iter().filter(|x| !is_printable(x)).count() as f32;
+    let non_ascii_ratio = non_printable_chars_count / buffer_length;
+
+    let high_ascii_count = buffer.iter().filter(|x| !is_high_ascii(x)).count() as f32;
+    let high_ascii_ratio = high_ascii_count / buffer_length;
+
+    (non_ascii_ratio > 0.3 && high_ascii_ratio < 0.05) ||
+        (non_ascii_ratio > 0.8 && high_ascii_ratio > 0.8)
+}