From cfc70207996e202edbb577b2ad97a61ba9eb0eaa Mon Sep 17 00:00:00 2001
From: Akshay <nerdy@peppe.rs>
Date: Tue, 2 Aug 2022 19:50:46 +0530
Subject: add textual comparison

structural comparison helps detect a vast majority of duplicates, but it
has a few false positives when files contain only trivia. textual
similarity can help detect and eliminate those false positives.
---
 src/lib.rs | 20 +++++++++++++++++---
 1 file changed, 17 insertions(+), 3 deletions(-)

(limited to 'src')

diff --git a/src/lib.rs b/src/lib.rs
index b708984..df67a9a 100644
--- a/src/lib.rs
+++ b/src/lib.rs
@@ -17,6 +17,8 @@ use tree_sitter::{Node, Parser, Tree};
 
 type Result<T> = std::result::Result<T, SimError>;
 
+const THRESHOLD: f64 = 0.6;
+
 // Check if two tree-sitter objects (trees or nodes etc.) are "similar"
 pub trait Similarity {
     // This function accepts a predicate to filter out nodes,
@@ -32,6 +34,7 @@ pub struct ProgramFile {
     pub path: PathBuf,
     pub src: String,
     pub tree: Tree,
+    pub simhash: u64,
 }
 
 impl Similarity for ProgramFile {
@@ -49,12 +52,15 @@ impl Similarity for ProgramFile {
         // - one tree's traversal is longer than the other
         // - the trees have identical traversal lengths but non-identical
         //   nodes along the traversal
-        mine.zip_longest(theirs)
+        let structurally_similar = mine
+            .zip_longest(theirs)
             .any(|result| match result {
                 EitherOrBoth::Both(mine, theirs) => mine.kind_id() != theirs.kind_id(),
                 EitherOrBoth::Left(_) | EitherOrBoth::Right(_) => true,
             })
-            .not()
+            .not();
+        let textually_similar = (simhash::hash_similarity(self.simhash, other.simhash)) > THRESHOLD;
+        structurally_similar && textually_similar
     }
 }
 
@@ -72,7 +78,13 @@ impl ProgramFile {
         let src = std::fs::read_to_string(&path)?;
         let path = path.as_ref().to_owned();
         let tree = parser.parse(&src, None).ok_or(SimError::LanguageNotSet)?;
-        Ok(Self { path, src, tree })
+        let simhash = simhash::simhash(&src);
+        Ok(Self {
+            path,
+            src,
+            tree,
+            simhash,
+        })
     }
 }
 
@@ -92,12 +104,14 @@ mod test {
             path: PathBuf::from("mine"),
             src: mine.to_owned(),
             tree: parser.parse(&mine, None).unwrap(),
+            simhash: simhash::simhash(&mine),
         };
 
         let their_program_file = ProgramFile {
             path: PathBuf::from("their"),
             src: their.to_owned(),
             tree: parser.parse(&their, None).unwrap(),
+            simhash: simhash::simhash(&their),
         };
 
         my_program_file.is_similar(&their_program_file, |n| !n.is_extra())
-- 
cgit v1.2.3