Merge pull request #187 from vibinex/tr/phind/2.0

Diff Graph implementation

Dockerfile

@@ -4,7 +4,7 @@ FROM ubuntu:latest
 # # Install dependencies required by the application
 RUN \
   apt-get update && \
-  apt-get install ca-certificates git -y && \
+  apt-get install ca-certificates git ripgrep -y && \
   apt-get clean
 ADD http://archive.ubuntu.com/ubuntu/pool/main/o/openssl/libssl1.1_1.1.1f-1ubuntu2_amd64.deb /tmp
 RUN chmod a+x /tmp/libssl1.1_1.1.1f-1ubuntu2_amd64.deb && \
@@ -45,7 +45,7 @@ ENV PROVIDER=$PROVIDER
 COPY ./vibi-dpu/target/debug/vibi-dpu /app/vibi-dpu
 COPY ./pubsub-sa.json /app/pubsub-sa.json
 COPY ./repo-profiler.pem /app/repo-profiler.pem
-
+COPY ./prompts /app/prompts
 # Create directory for configuration
 RUN mkdir /app/config
 

vibi-dpu/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "vibi-dpu"
-version = "1.0.0"
+version = "2.0.0"
 edition = "2021"
 authors = ["Tapish Rathore <tapish@vibinex.com>"]
 license = "GPL-3.0-or-later"
@@ -21,7 +21,7 @@ serde = { version = "1.0", features = ["derive"] }
 serde_json = "1.0"
 base64ct = "1.5.3"
 sha256 = "1.1.1"
-reqwest = { version = "0.11", features = ["json", "blocking"] }
+reqwest = { version = "0.11", features = ["json", "blocking", "stream"] }
 google-cloud-pubsub = "0.15.0"
 google-cloud-default = { version = "0.3.0", features = ["pubsub"] }
 google-cloud-googleapis = "0.9.0"
@@ -37,5 +37,5 @@ once_cell = "1.18.0" # MIT
 jsonwebtoken = "8.3.0" # MIT
 fern = "0.6.2" # MIT
 log = "0.4.20" # MIT/Apache2
-
+walkdir = "2.5.0" # Unlicence/MIT
 # todo - check all lib licences

vibi-dpu/src/core/diff_graph.rs

@@ -0,0 +1,46 @@
+use crate::graph::mermaid_elements::generate_mermaid_flowchart;
+use crate::utils::user::ProviderEnum;
+use crate::utils::review::Review;
+use crate::core::github;
+use crate::utils::gitops::StatItem;
+
+pub async fn send_diff_graph(review: &Review, excluded_files: &Vec<StatItem>, small_files: &Vec<StatItem>, access_token: &str) {
+	let comment = diff_graph_comment_text(excluded_files, small_files, review).await;
+	// add comment for GitHub
+	if review.provider().to_string() == ProviderEnum::Github.to_string() {
+		log::info!("Inserting comment on repo {}...", review.repo_name());
+		github::comment::add_comment(&comment, review, &access_token).await;
+	}
+
+	// TODO: add comment for Bitbucket
+}
+
+async fn diff_graph_comment_text(excluded_files: &Vec<StatItem>, small_files: &Vec<StatItem>, review: &Review) -> String {
+    let mut comment = "Diff Graph:\n\n".to_string();  
+
+    let all_diff_files: Vec<StatItem> = excluded_files
+        .iter()
+        .chain(small_files.iter())
+        .cloned()  // Clone the StatItem instances since `iter` returns references
+        .collect(); // Collect into a new vector
+    if let Some(mermaid_text) = mermaid_comment(&all_diff_files, review).await {
+        comment += mermaid_text.as_str();
+    }
+    comment += "\nTo modify DiffGraph settings, go to [your Vibinex settings page.](https://vibinex.com/settings)\n";
+    return comment;
+}
+
+async fn mermaid_comment(diff_files: &Vec<StatItem>, review: &Review) -> Option<String> {
+    let flowchart_str_opt = generate_mermaid_flowchart(diff_files, review).await;
+    if flowchart_str_opt.is_none() {
+        log::error!("[mermaid_comment] Unable to generate flowchart for review: {}", review.id());
+        return None;
+    }
+    let flowchart_str = flowchart_str_opt.expect("Empty flowchart_str_opt");
+    let mermaid_comment = format!(
+        "### Call Stack Diff\n```mermaid\n{}\n```",
+        flowchart_str,
+    );
+    return Some(mermaid_comment);
+}
+

vibi-dpu/src/core/mod.rs

@@ -4,4 +4,5 @@ pub mod utils;
 pub mod approval;
 pub mod bitbucket;
 pub mod github;
-pub mod trigger;
+pub mod trigger;
+pub mod diff_graph;

vibi-dpu/src/core/relevance.rs

@@ -1,10 +1,10 @@
 use std::collections::{HashMap, HashSet};
 
-use crate::{bitbucket::{self, user::author_from_commit}, core::github, db::review::save_review_to_db, utils::{aliases::get_login_handles, relevance::Relevance, hunk::{HunkMap, PrHunkItem}, user::ProviderEnum}};
+use crate::{bitbucket::{self, user::author_from_commit}, core::github, db::review::save_review_to_db, utils::{aliases::get_login_handles, gitops::StatItem, hunk::{HunkMap, PrHunkItem}, relevance::Relevance, user::ProviderEnum}};
 use crate::utils::review::Review;
 use crate::utils::repo_config::RepoConfig;
 
-pub async fn process_relevance(hunkmap: &HunkMap, review: &Review,
+pub async fn process_relevance(hunkmap: &HunkMap, excluded_files: &Vec<StatItem>, review: &Review,
 	repo_config: &mut RepoConfig, access_token: &str, old_review_opt: &Option<Review>,
 ) {
 	log::info!("Processing relevance of code authors...");
@@ -22,7 +22,7 @@ pub async fn process_relevance(hunkmap: &HunkMap, review: &Review,
 		let relevance_vec = relevance_vec_opt.expect("Empty coverage_obj_opt");
 		if repo_config.comment() {
 			// create comment text
-			let comment = comment_text(&relevance_vec, repo_config.auto_assign());
+			let comment = relevant_reviewers_comment_text(&relevance_vec, repo_config.auto_assign(), excluded_files).await;
 			// add comment
 			if review.provider().to_string() == ProviderEnum::Bitbucket.to_string() {
 				// TODO - add feature flag check
@@ -184,7 +184,8 @@ async fn calculate_relevance(prhunk: &PrHunkItem, review: &mut Review) -> Option
     return Some(relevance_vec);
 }
 
-fn comment_text(relevance_vec: &Vec<Relevance>, auto_assign: bool) -> String {
+async fn relevant_reviewers_comment_text(relevance_vec: &Vec<Relevance>, auto_assign: bool,
+    excluded_files: &Vec<StatItem>) -> String {
     let mut comment = "Relevant users for this PR:\n\n".to_string();  // Added two newlines
     comment += "| Contributor Name/Alias  | Relevance |\n";  // Added a newline at the end
     comment += "| -------------- | --------------- |\n";  // Added a newline at the end
@@ -208,6 +209,14 @@ fn comment_text(relevance_vec: &Vec<Relevance>, auto_assign: bool) -> String {
         comment += &format!("Missing profile handles for {} aliases. [Go to your Vibinex settings page](https://vibinex.com/settings) to map aliases to profile handles.", unmapped_aliases.len());
     }
 
+    if !excluded_files.is_empty() {
+        comment += "\n\n";
+        comment += "Ignoring following files due to large size: ";
+        for file_item in excluded_files {
+            comment += &format!("- {}\n", file_item.filepath.as_str());
+        }
+    }
+
     if auto_assign {
         comment += "\n\n";
         comment += "Auto assigning to relevant reviewers.";
@@ -216,7 +225,6 @@ fn comment_text(relevance_vec: &Vec<Relevance>, auto_assign: bool) -> String {
     comment += "If you are a relevant reviewer, you can use the [Vibinex browser extension](https://chromewebstore.google.com/detail/vibinex-code-review/jafgelpkkkopeaefadkdjcmnicgpcncc) to see parts of the PR relevant to you\n";  // Added a newline at the end
     comment += "Relevance of the reviewer is calculated based on the git blame information of the PR. To know more, hit us up at contact@vibinex.com.\n\n";  // Added two newlines
     comment += "To change comment and auto-assign settings, go to [your Vibinex settings page.](https://vibinex.com/u)\n";  // Added a newline at the end
-
     return comment;
 }
 

vibi-dpu/src/core/review.rs

@@ -1,17 +1,17 @@
-use std::env;
+use std::{env, thread, time::Duration};
 
 use serde_json::Value;
 
 use crate::{
-    core::{relevance::process_relevance, utils::get_access_token},
+    core::{relevance::process_relevance, diff_graph::send_diff_graph, utils::get_access_token},
     db::{
         hunk::{get_hunk_from_db, store_hunkmap_to_db},
         repo::get_clone_url_clone_dir,
         repo_config::save_repo_config_to_db,
         review::{get_review_from_db, save_review_to_db},
     },
     utils::{
-        gitops::{commit_exists, generate_blame, generate_diff, get_excluded_files, git_pull, process_diffmap},
+        gitops::{commit_exists, generate_blame, generate_diff, get_excluded_files, git_pull, process_diffmap, StatItem},
         hunk::{HunkMap, PrHunkItem},
         repo_config::RepoConfig,
         reqwest_client::get_client,
@@ -41,11 +41,24 @@ pub async fn process_review(message_data: &Vec<u8>) {
 	}
 	let access_token = access_token_opt.expect("Empty access_token_opt");
 	commit_check(&review, &access_token).await;
-	let hunkmap_opt = process_review_changes(&review).await;
-	send_hunkmap(&hunkmap_opt, &review, &repo_config, &access_token, &old_review_opt).await;
+	process_review_changes(&review, &repo_config, &access_token, &old_review_opt).await;
 }
 
-pub async fn send_hunkmap(hunkmap_opt: &Option<HunkMap>, review: &Review,
+pub async fn process_review_changes(review: &Review, repo_config: &RepoConfig, access_token: &str, old_review_opt: &Option<Review>) {
+	log::info!("Processing changes in code...");
+	if let Some((excluded_files, smallfiles)) = get_included_and_excluded_files(review) {
+		let hunkmap_opt = calculate_hunkmap(review, &smallfiles).await;
+		send_hunkmap(&hunkmap_opt, &excluded_files, review, repo_config, access_token, old_review_opt).await;
+
+		if repo_config.diff_graph() {
+			send_diff_graph(review, &excluded_files, &smallfiles, access_token).await;
+		}
+	} else {
+		log::error!("Failed to get included and excluded files");
+	}
+}
+
+pub async fn send_hunkmap(hunkmap_opt: &Option<HunkMap>, excluded_files: &Vec<StatItem>, review: &Review,
 	repo_config: &RepoConfig, access_token: &str, old_review_opt: &Option<Review>) {
 	if hunkmap_opt.is_none() {
 		log::error!("[send_hunkmap] Empty hunkmap in send_hunkmap");
@@ -58,7 +71,7 @@ pub async fn send_hunkmap(hunkmap_opt: &Option<HunkMap>, review: &Review,
 	let hunkmap_async = hunkmap.clone();
 	let review_async = review.clone();
 	let mut repo_config_clone = repo_config.clone();
-	process_relevance(&hunkmap_async, &review_async,
+	process_relevance(&hunkmap_async, &excluded_files, &review_async,
 		&mut repo_config_clone, access_token, old_review_opt).await;
 }
 
@@ -73,16 +86,20 @@ fn hunk_already_exists(review: &Review) -> bool {
 	log::debug!("[hunk_already_exists] Hunk already in db!");
 	return true;
 }
-pub async fn process_review_changes(review: &Review) -> Option<HunkMap>{
-	log::info!("Processing changes in code...");
-	let mut prvec = Vec::<PrHunkItem>::new();
+
+fn get_included_and_excluded_files(review: &Review) -> Option<(Vec<StatItem>, Vec<StatItem>)> {
 	let fileopt = get_excluded_files(&review);
 	log::debug!("[process_review_changes] fileopt = {:?}", &fileopt);
 	if fileopt.is_none() {
 		log::error!("[process_review_changes] No files to review for PR {}", review.id());
 		return None;
 	}
-	let (_, smallfiles) = fileopt.expect("fileopt is empty");
+	let (excluded_files, smallfiles) = fileopt.expect("fileopt is empty");
+	return Some(( excluded_files, smallfiles));
+}
+
+async fn calculate_hunkmap(review: &Review, smallfiles: &Vec<StatItem>) -> Option<HunkMap> {
+	let mut prvec = Vec::<PrHunkItem>::new();
 	let diffmap = generate_diff(&review, &smallfiles);
 	log::debug!("[process_review_changes] diffmap = {:?}", &diffmap);
 	let linemap = process_diffmap(&diffmap);
@@ -109,6 +126,7 @@ pub async fn commit_check(review: &Review, access_token: &str) {
 	if !commit_exists(&review.base_head_commit(), &review.clone_dir()) 
 		|| !commit_exists(&review.pr_head_commit(), &review.clone_dir()) {
 		log::info!("Executing git pull on repo {}...", &review.repo_name());
+		thread::sleep(Duration::from_secs(1));
 		git_pull(review, access_token).await;
 	}
 }
@@ -213,7 +231,7 @@ fn create_and_save_github_review_object(deserialized_data: &Value) -> Option<Rev
 	let repo_provider = ProviderEnum::Github.to_string().to_lowercase();
 	let clone_opt = get_clone_url_clone_dir(&repo_provider, &repo_owner, &repo_name);
 	if clone_opt.is_none() {
-		log::error!("[create_and_save_github_review_object] Unable to get clone url and directory for bitbucket review");
+		log::error!("[create_and_save_github_review_object] Unable to get clone url and directory for github review");
 		return None;
 	}
 	let (clone_url, clone_dir) = clone_opt.expect("Empty clone_opt");

vibi-dpu/src/core/trigger.rs

@@ -48,9 +48,7 @@ pub async fn process_trigger(message_data: &Vec<u8>) {
 	// commit_check
 	commit_check(&review, &access_token).await;
 	// process_review_changes
-	let hunkmap_opt = process_review_changes(&review).await;
-	// send_hunkmap
-	send_hunkmap(&hunkmap_opt, &review, &repo_config, &access_token, &None).await;
+	process_review_changes(&review, &repo_config, &access_token, &None).await;
 }
 
 fn parse_message_fields(msg: &Value) -> Option<TriggerReview> {

vibi-dpu/src/db/review.rs

@@ -38,7 +38,7 @@ pub fn get_review_from_db(repo_name: &str, repo_owner: &str,
     let review_res = serde_json::from_slice(&ivec);
     if let Err(e) = review_res {
         log::error!(
-            "[get_handles_from_db] Failed to deserialize review from json: {:?}",
+            "[get_review_from_db] Failed to deserialize review from json: {:?}",
             e
         );
         return None;