Add file_ignore setting

Author: zurawiki

e2e/test_githook_file_ignore.sh

@@ -0,0 +1,22 @@
+#!/bin/sh
+set -eu
+
+DIFF_CONTENT_PATH="$(pwd)/tests/data/example_2.diff"
+
+export TEMPDIR=$(mktemp -d)
+(
+    cd "${TEMPDIR}"
+    git init
+
+    export TEMPFILE=$(mktemp)
+    echo "foo" > $TEMPFILE
+
+    GPTCOMMIT__MODEL_PROVIDER="tester-foobar" \
+    gptcommit prepare-commit-msg \
+      --git-diff-content "${DIFF_CONTENT_PATH}" \
+      --commit-msg-file "${TEMPFILE}" \
+      --commit-source ""
+
+    cat $TEMPFILE
+)
+rm -rf "${TEMPDIR}"

src/actions/prepare_commit_msg.rs

@@ -92,8 +92,7 @@ pub(crate) async fn main(settings: Settings, args: PrepareCommitMsgArgs) -> Resu
     };
 
     let client = get_llm_client(&settings);
-    let summarization_client =
-        SummarizationClient::new(settings.prompt.unwrap(), settings.output.unwrap(), client)?;
+    let summarization_client = SummarizationClient::new(settings.to_owned(), client)?;
 
     println!("{}", "🤖 Asking GPT-3 to summarize diffs...".green().bold());
 

src/settings.rs

@@ -19,6 +19,13 @@ use crate::{
     },
 };
 
+static DEFAULT_FILES_TO_IGNORE: &[&str; 4] = &[
+    "package-lock.json",
+    "yarn.lock",
+    "pnpm-lock.yaml",
+    "Cargo.lock",
+];
+
 #[derive(Debug, Clone, Display, Serialize, Default, EnumString)]
 pub(crate) enum ModelProvider {
     #[default]
@@ -143,7 +150,10 @@ pub(crate) struct Settings {
     pub openai: Option<OpenAISettings>,
     pub prompt: Option<PromptSettings>,
     pub output: Option<OutputSettings>,
+    /// Whether to run githook when amending the commit
     pub allow_amend: Option<bool>,
+    /// Files to ignore, format similar to gitignore
+    pub file_ignore: Option<Vec<String>>,
 }
 
 impl Settings {
@@ -170,6 +180,14 @@ impl Settings {
     fn get_config_builder() -> Result<ConfigBuilder<DefaultState>, ConfigError> {
         let mut settings = Config::builder()
             .set_default("allow_amend", false)?
+            .set_default(
+                "file_ignore",
+                DEFAULT_FILES_TO_IGNORE
+                    .to_vec()
+                    .iter()
+                    .map(|s| s.to_string())
+                    .collect::<Vec<_>>(),
+            )?
             .set_default("model_provider", ModelProvider::OpenAI)?
             .set_default(
                 "openai",

src/summarize.rs

@@ -3,19 +3,17 @@ use std::str::FromStr;
 use std::sync::Arc;
 
 use crate::llms::llm_client::LlmClient;
+use crate::settings::Settings;
 use crate::util;
-use crate::{
-    prompt::format_prompt,
-    settings::PromptSettings,
-    settings::{Language, OutputSettings},
-};
+use crate::{prompt::format_prompt, settings::Language};
 use anyhow::Result;
 use tokio::task::JoinSet;
 use tokio::try_join;
 #[derive(Debug, Clone)]
 pub(crate) struct SummarizationClient {
     client: Arc<dyn LlmClient>,
 
+    file_ignore: Vec<String>,
     prompt_file_diff: String,
     prompt_commit_summary: String,
     prompt_commit_title: String,
@@ -24,21 +22,20 @@ pub(crate) struct SummarizationClient {
 }
 
 impl SummarizationClient {
-    pub(crate) fn new(
-        prompt_settings: PromptSettings,
-        output_settings: OutputSettings,
-        client: Box<dyn LlmClient>,
-    ) -> Result<Self> {
-        let prompt_file_diff = prompt_settings.file_diff.unwrap_or_default();
-        let prompt_commit_summary = prompt_settings.commit_summary.unwrap_or_default();
-        let prompt_commit_title = prompt_settings.commit_title.unwrap_or_default();
-        let prompt_translation = prompt_settings.translation.unwrap_or_default();
-        let prompt_lang = Language::from_str(&output_settings.lang.unwrap_or_default())
+    pub(crate) fn new(settings: Settings, client: Box<dyn LlmClient>) -> Result<Self> {
+        let prompt = settings.prompt.unwrap();
+
+        let prompt_file_diff = prompt.file_diff.unwrap_or_default();
+        let prompt_commit_summary = prompt.commit_summary.unwrap_or_default();
+        let prompt_commit_title = prompt.commit_title.unwrap_or_default();
+        let prompt_translation = prompt.translation.unwrap_or_default();
+        let prompt_lang = Language::from_str(&settings.output.unwrap().lang.unwrap_or_default())
             .unwrap()
             .to_string();
-
+        let file_ignore = settings.file_ignore.unwrap_or_default();
         Ok(Self {
             client: client.into(),
+            file_ignore,
             prompt_file_diff,
             prompt_commit_summary,
             prompt_commit_title,
@@ -103,6 +100,15 @@ impl SummarizationClient {
     /// https://git-scm.com/docs/git-diff
     async fn process_file_diff(&self, file_diff: &str) -> Option<(String, String)> {
         if let Some(file_name) = util::get_file_name_from_diff(file_diff) {
+            if self
+                .file_ignore
+                .iter()
+                .any(|ignore| file_name.contains(ignore))
+            {
+                warn!("skipping {file_name} due to file_ignore setting");
+
+                return None;
+            }
             let completion = self.diff_summary(file_name, file_diff).await;
             Some((
                 file_name.to_string(),

src/toml.rs

@@ -12,6 +12,7 @@ impl DeepKeysCollector<'_> {
         let mut visitor = DeepKeysCollector::default();
         visitor.visit_document(&document);
 
+        visitor.keys.dedup();
         visitor.keys.sort();
         visitor.keys
     }
@@ -76,11 +77,13 @@ the-force = { value = "surrounds-you" }
         visitor.visit_document(&document);
 
         assert_eq!(visitor.current_path, Vec::<&str>::new());
+        visitor.keys.dedup();
         visitor.keys.sort();
         assert_eq!(
             visitor.keys,
             vec![
                 "allow_amend",
+                "file_ignore",
                 "model_provider",
                 "openai.api_key",
                 "openai.model",
@@ -101,6 +104,7 @@ the-force = { value = "surrounds-you" }
             DeepKeysCollector::get_keys(input),
             vec![
                 "allow_amend",
+                "file_ignore",
                 "model_provider",
                 "openai.api_key",
                 "openai.model",

tests/data/example_2.diff

@@ -0,0 +1,72 @@
+diff --git a/.vscode/launch.json b/.vscode/launch.json
+new file mode 100644
+index 0000000..f26eafc
+--- /dev/null
++++ b/.vscode/launch.json
+@@ -0,0 +1,10 @@
++{
++    // Use IntelliSense to learn about possible attributes.
++    // Hover to view descriptions of existing attributes.
++    // For more information, visit: https://go.microsoft.com/fwlink/?linkid=830387
++    "version": "0.2.0",
++    "configurations": [
++      rust
++
++    ]
++}
+diff --git a/yarn.lock b/yarn.lock
+new file mode 100644
+index 0000000..a51b2a6
+--- /dev/null
++++ b/yarn.lock
+@@ -0,0 +1 @@
++sadasdas
+diff --git a/src/main.rs b/src/main.rs
+index 46d98b0..b1dd2af 100644
+--- a/src/main.rs
++++ b/src/main.rs
+@@ -1,10 +1,7 @@
+ use std::process::Command;
+
+ fn run_command(cmd: &str, args: &[&str]) -> Result<String, String> {
+-    let output = match Command::new(cmd)
+-        .args(args)
+-        .output()
+-    {
++    let output = match Command::new(cmd).args(args).output() {
+         Ok(o) => o,
+         Err(e) => return Err(format!("Failed to execute process: {}", e)),
+     };
+@@ -19,9 +16,28 @@ fn run_command(cmd: &str, args: &[&str]) -> Result<String, String> {
+ }
+
+ fn main() {
+-    let result = run_command("git", &["diff", "--staged"]);
+-    match result {
+-        Ok(output) => println!("{}", output),
+-        Err(error) => eprintln!("{}", error),
++    let output = run_command("git", &["diff", "--staged"]).unwrap();
++    let file_diffs = split_string_with_pattern(&output, "diff --git ");
++    println!("{:?}", file_diffs);
++}
++
++fn split_string_with_pattern<'a>(string: &'a str, pattern: &'a str) -> Vec<&'a str> {
++    let mut start = dbg!(pattern.len());
++    let mut substrings = Vec::new();
++
++    while let Some(i) = string[start..].find(pattern) {
++        substrings.push(&string[start..start + i]);
++        start += i;
+     }
++    substrings.push(&string[start..]);
++
++    substrings
++}
++
++#[test]
++fn test_split_string_with_pattern() {
++    let string = include_str!("../test_data/diff.txt");
++    let pattern = "diff --git ";
++    println!("testing");
++    assert!(split_string_with_pattern(string, pattern).len() == 2);
+ }