Merge pull request #37 from keszybz/pyc-parsing-2

Rewrite the pyc handler to use references for all repeated objects

Cargo.toml

@@ -170,6 +170,8 @@ itertools = "0.13.0"
 log = { version = "0.4", features = ["std"] }
 nix = { version = "0.28.0", features = ["fs", "socket"] }
 num-bigint-dig = "0.8.4"
+num-integer = "0.1.46"
+num-traits = "0.2.19"
 regex = { version = "1.10.0", default-features = false, features = ["std", "perf", "unicode-case"] }
 serde = { version = "1.0.203", features = ["derive"] }
 serde_cbor = "0.11"

src/config.rs

@@ -37,6 +37,19 @@ struct Options {
     #[arg(long)]
     pub check: bool,
 
+    /// Ignore file name extensions, always run handlers.
+    #[arg(long)]
+    pub ignore_extension: bool,
+
+    /// Print contents of files
+    #[arg(short, long,
+          conflicts_with = "brp",
+          conflicts_with = "check",
+          conflicts_with = "job_socket",
+          conflicts_with = "result_socket",
+          conflicts_with = "jobs")]
+    pub print: bool,
+
     /// Read paths to process from this socket.
     /// When used, an explicit list of handlers must be given.
     #[arg(long,
@@ -65,9 +78,11 @@ pub struct Config {
     pub inputs: Vec<PathBuf>,
     pub brp: bool,
     pub verbose: bool,
+    pub print: bool,
     pub job_socket: Option<RawFd>,
     pub result_socket: Option<RawFd>,
     pub check: bool,
+    pub ignore_extension: bool,
     pub jobs: Option<u32>,
     pub source_date_epoch: Option<i64>,
     pub handler_names: Vec<&'static str>,
@@ -193,9 +208,11 @@ impl Config {
             inputs: options.inputs,
             brp: options.brp,
             verbose: options.verbose,
+            print: options.print,
             job_socket: options.job_socket,
             result_socket: options.result_socket,
             check: options.check,
+            ignore_extension: options.ignore_extension,
             jobs,
             source_date_epoch,
             handler_names,
@@ -210,9 +227,11 @@ impl Config {
             inputs: vec![],
             brp: false,
             verbose: false,
+            print: false,
             job_socket: None,
             result_socket: None,
             check,
+            ignore_extension: false,
             jobs: None,
             source_date_epoch: Some(source_date_epoch),
             handler_names: vec![],

src/handlers/ar.rs

@@ -54,7 +54,7 @@ impl super::Processor for Ar {
     }
 
     fn filter(&self, path: &Path) -> Result<bool> {
-        Ok(path.extension().is_some_and(|x| x == "a"))
+        Ok(self.config.ignore_extension || path.extension().is_some_and(|x| x == "a"))
     }
 
     fn process(&self, input_path: &Path) -> Result<super::ProcessResult> {

src/handlers/jar.rs

@@ -29,7 +29,7 @@ impl super::Processor for Jar {
     }
 
     fn filter(&self, path: &Path) -> Result<bool> {
-        Ok(path.extension().is_some_and(|x| x == "jar"))
+        Ok(self.config.ignore_extension || path.extension().is_some_and(|x| x == "jar"))
     }
 
     fn process(&self, input_path: &Path) -> Result<super::ProcessResult> {

src/handlers/javadoc.rs

@@ -76,8 +76,10 @@ impl super::Processor for Javadoc {
     }
 
     fn filter(&self, path: &Path) -> Result<bool> {
-        Ok(path.extension().is_some_and(|x| x == "html")
-           // && path.to_str().is_some_and(|x| x.contains("/usr/share/javadoc/"))
+        Ok(
+            self.config.ignore_extension ||
+                path.extension().is_some_and(|x| x == "html")
+            // && path.to_str().is_some_and(|x| x.contains("/usr/share/javadoc/"))
         )
     }
 

src/handlers/mod.rs

@@ -10,6 +10,7 @@ use log::{log, debug, info, warn, Level};
 use serde::{Serialize, Deserialize};
 use std::ascii::escape_default;
 use std::collections::HashMap;
+use std::fmt::Write;
 use std::ffi::OsStr;
 use std::fs;
 use std::fs::{File, Metadata};
@@ -218,6 +219,22 @@ pub fn inodes_seen() -> HashMap<u64, u8> {
     HashMap::new()
 }
 
+pub fn do_print(config: &Rc<config::Config>) -> Result<()> {
+    let handler = pyc::Pyc::new(config);
+    let mut w = String::new();
+
+    for (n, input_path) in config.inputs.iter().enumerate() {
+        if n > 0 {
+            writeln!(w)?;  // separate outputs by empty line
+        }
+        handler.pretty_print(&mut w, input_path)?;
+    }
+
+    print!("{}", w);
+
+    Ok(())
+}
+
 pub fn do_normal_work(config: &Rc<config::Config>) -> Result<Stats> {
     let handlers = make_handlers(config)?;
     let mut inodes_seen = inodes_seen();