Local file paths in meta are recorded using relative paths.

Author: EthanYuan

devtools/text-optimizer/src/error.rs

@@ -1,7 +1,9 @@
 #[derive(Debug)]
 pub enum MyError {
+    CommitId,
     Io(std::io::Error),
     Serde(serde_yaml::Error),
+    PathError,
 }
 
 impl From<std::io::Error> for MyError {

devtools/text-optimizer/src/extractors/mod.rs

@@ -4,7 +4,7 @@ pub mod std_output_extractor;
 pub mod thiserror_extractor;
 
 use crate::{
-    types::TextInfo, yaml_processor::save_yaml, CLAP_TEXT_FILE, LOG_TEXT_FILE,
+    error::MyError, types::TextInfo, yaml_processor::save_yaml, CLAP_TEXT_FILE, LOG_TEXT_FILE,
     STD_OUTPUT_TEXT_FILE, THISERROR_TEXT_FILE,
 };
 
@@ -27,14 +27,11 @@ pub trait Extractor {
     fn text_list(&self) -> Vec<TextInfo>;
     fn scanning_file_path(&self) -> &PathBuf;
     fn save_file_path(&self) -> &PathBuf;
-    fn save_as_file(&self) {
-        save_yaml(self.save_file_path(), &self.text_list()).expect("save yaml");
-    }
     fn visit_file(&mut self, node: &File);
 }
 
-pub fn extract(project_root: PathBuf, output_dir: &PathBuf) {
-    // extractors
+pub fn extract(root_cargo_config_path: PathBuf, commit_id: &str, output_dir: &PathBuf) {
+    // init all extractors
     let mut clap_extractor = ClapExtractor::new(output_dir.join(CLAP_TEXT_FILE));
     let mut log_extractor = LogExtractor::new(output_dir.join(LOG_TEXT_FILE));
     let mut std_output_extractor = StdOutputExtractor::new(output_dir.join(STD_OUTPUT_TEXT_FILE));
@@ -48,21 +45,24 @@ pub fn extract(project_root: PathBuf, output_dir: &PathBuf) {
     ];
 
     let project_metadata = MetadataCommand::new()
-        .manifest_path(project_root)
+        .manifest_path(&root_cargo_config_path)
         .exec()
         .expect("Failed to get current directory");
 
     for package in project_metadata.workspace_packages() {
         log::info!("Scanning Crate: {}", package.name);
 
-        let crate_src_path = Path::new(&package.manifest_path)
+        let crate_src_path = PathBuf::from(&package.manifest_path)
             .parent()
             .expect("workspace member crate path")
             .join("src");
+        let crate_src_path =
+            to_relative_path(&crate_src_path, &root_cargo_config_path).expect("to_relative_path");
+
         process_rs_files_in_src(&crate_src_path, &mut extractors);
     }
 
-    save_extractors(output_dir, &extractors);
+    save_text_info_files(output_dir, &extractors, commit_id);
 }
 
 pub fn process_rs_files_in_src(dir_path: &Path, extractors: &mut [&mut dyn Extractor]) {
@@ -103,14 +103,34 @@ pub fn extract_contents_in_brackets(lit: String) -> Option<String> {
     None
 }
 
-fn save_extractors(output_dir: &PathBuf, extractors: &[&mut dyn Extractor]) {
+fn save_text_info_files(output_dir: &PathBuf, extractors: &[&mut dyn Extractor], commit_id: &str) {
     fs::create_dir_all(output_dir).expect("create dir all");
     println!();
 
     for extractor in extractors {
-        extractor.save_as_file();
+        save_yaml(
+            extractor.save_file_path(),
+            &extractor.text_list(),
+            commit_id,
+        )
+        .expect("save yaml");
+
         let file_name = extractor.save_file_path().file_name().unwrap();
         let text_len = extractor.text_list().len();
         println!("{:?}: {:?}", file_name, text_len);
     }
 }
+
+fn to_relative_path(
+    absolut_path: &Path,
+    root_cargo_config_path: &Path,
+) -> Result<PathBuf, MyError> {
+    let cargo_config_path_abs = root_cargo_config_path.canonicalize()?;
+    let project_root_path = cargo_config_path_abs.parent().ok_or(MyError::PathError)?;
+    let relative_path = absolut_path
+        .strip_prefix(project_root_path)
+        .map_err(|_| MyError::PathError)?
+        .to_str()
+        .ok_or(MyError::PathError)?;
+    Ok(Path::new("../..").join(relative_path))
+}

devtools/text-optimizer/src/main.rs

@@ -6,10 +6,14 @@ mod yaml_processor;
 
 use backfill::backfill;
 use clap::{Parser, Subcommand};
+use error::MyError;
 use extractors::extract;
 use std::path::PathBuf;
+use std::process::Command;
+use std::result::Result;
 
-pub const PROJECT_ROOT: &str = "../../Cargo.toml";
+pub const GITHUB_REPO: &str = "https://github.com/nervosnetwork/ckb/tree";
+pub const PROJECT_ROOT_CARGO_CONFIG_PATH: &str = "../../Cargo.toml";
 pub const LOG_TEXT_FILE: &str = "log_text_list.yml";
 pub const CLAP_TEXT_FILE: &str = "clap_text_list.yml";
 pub const STD_OUTPUT_TEXT_FILE: &str = "std_output_text_list.yml";
@@ -26,6 +30,10 @@ struct Cli {
 enum Commands {
     /// extracting text
     Extract {
+        /// specify a github commit id as the rev for generating the source link
+        #[arg(short, long)]
+        commit_id: String,
+
         /// specifies a directory path for .yml text files output
         #[arg(short, long, default_value = PathBuf::from("./").into_os_string())]
         output_dir: PathBuf,
@@ -46,12 +54,51 @@ fn main() {
     );
 
     match &cli.command {
-        Some(Commands::Extract { output_dir }) => {
-            extract(PathBuf::from(PROJECT_ROOT), output_dir);
+        Some(Commands::Extract {
+            commit_id,
+            output_dir,
+        }) => {
+            check_commit_id(commit_id).expect("check commit id");
+            extract(
+                PathBuf::from(PROJECT_ROOT_CARGO_CONFIG_PATH),
+                commit_id,
+                output_dir,
+            );
         }
         Some(Commands::Backfill { input_dir }) => {
             backfill(input_dir);
         }
         None => {}
     }
 }
+
+fn check_commit_id(commit_id: &str) -> Result<(), MyError> {
+    let output = Command::new("git")
+        .arg("rev-parse")
+        .arg("HEAD")
+        .output()
+        .expect("Failed to execute git command");
+
+    if output.status.success() {
+        let current_commit_id = String::from_utf8_lossy(&output.stdout);
+        let current_commit_id: String = current_commit_id.trim().chars().take(7).collect();
+        let commit_id: String = commit_id.trim().chars().take(7).collect();
+        if current_commit_id == commit_id {
+            log::info!(
+                "Current commit ID matches the expected commit ID: {}",
+                current_commit_id
+            );
+            Ok(())
+        } else {
+            log::warn!(
+                "Current commit ID {} does not match the expected commit ID {}.",
+                current_commit_id,
+                commit_id
+            );
+            Err(MyError::CommitId)
+        }
+    } else {
+        log::error!("Failed to retrieve the current commit ID");
+        Err(MyError::CommitId)
+    }
+}

devtools/text-optimizer/src/yaml_processor.rs

@@ -1,15 +1,76 @@
-use super::{error::MyError, types::TextInfo};
+use super::{
+    error::MyError,
+    types::{Category, Meta, TextInfo},
+    GITHUB_REPO,
+};
+use serde::{Deserialize, Serialize};
 use std::io::Read;
 use std::io::Write;
 use std::{fs::File, path::PathBuf};
 
-pub fn save_yaml(file: &PathBuf, data: &[TextInfo]) -> Result<(), MyError> {
+#[derive(Debug, Serialize, Deserialize, PartialEq, Eq, Hash, Clone)]
+pub struct TextInfoSave {
+    original: String,
+    editable: String,
+    metadata: Metadata,
+}
+
+impl TextInfoSave {
+    pub fn from_text_info(text_info: TextInfo, git_repo: &str, commit_id: &str) -> Self {
+        // 使用 Metadata 的 from_meta 方法进行 Meta 到 Metadata 的转换
+        let metadata = Metadata::from_meta(text_info.metadata(), git_repo, commit_id);
+
+        // 创建 TextInfoSave 结构体并返回
+        TextInfoSave {
+            original: text_info.original().to_owned(),
+            editable: text_info.editable().to_owned(),
+            metadata,
+        }
+    }
+}
+
+#[derive(Debug, Serialize, Deserialize, PartialEq, Eq, Hash, Clone)]
+pub struct Metadata {
+    category: Category,
+    file: PathBuf,
+    code_line_link: Vec<String>,
+}
+
+impl Metadata {
+    // 定义从 Meta 到 Metadata 的转换方法
+    pub fn from_meta(meta: &Meta, github_repo: &str, commit_id: &str) -> Self {
+        // 创建 GitHub 代码行链接的前缀
+        let github_link_prefix = format!("{}/{}/", github_repo, commit_id);
+
+        // 为每个代码行生成 GitHub 链接
+        let code_line_link: Vec<String> = meta
+            .start_lines()
+            .iter()
+            .map(|line| format!("{}#L{}", github_link_prefix, line))
+            .collect();
+
+        Metadata {
+            category: meta.category().to_owned(),
+            file: meta.file().to_owned(),
+            code_line_link,
+        }
+    }
+}
+
+pub fn save_yaml(file: &PathBuf, data: &[TextInfo], commit_id: &str) -> Result<(), MyError> {
     let mut file = File::create(file)?;
+
+    // Convert TextInfo to TextInfoSave
+    let data_save: Vec<TextInfoSave> = data
+        .iter()
+        .map(|text_info| TextInfoSave::from_text_info(text_info.clone(), GITHUB_REPO, commit_id))
+        .collect();
+
     file.write_fmt(format_args!(
         "# Number of TextInfo items: {}\n\n",
         data.len()
     ))?;
-    serde_yaml::to_writer(file, data)?;
+    serde_yaml::to_writer(file, &data_save)?;
     Ok(())
 }