Merge pull request #6 from steineggerlab/cfgmodule

Add config module

path.cfg

@@ -1,7 +1,8 @@
 mmseqs=mmseqs
 foldseek=foldseek
+foldmason=foldmason
 mafft=mafft
 mafft-linsi=mafft-linsi
-foldmason=foldmason
 iqtree=iqtree
-#fasttree=fasttree
+#fasttree=
+#raxml=

src/envs/error_handler.rs

@@ -4,22 +4,30 @@ use crate::util::message;
 
 pub const WRN_GENERAL: i32 = 0x00;
 pub const ERR_GENERAL: i32 = 0x01;
-pub const ERR_FILE_NOT_FOUND: i32 = 0x02;
-pub const ERR_MODULE_NOT_IMPLEMENTED: i32 = 0x03;
-pub const ERR_ARGPARSE: i32 = 0x04;
-pub const ERR_BINARY_NOT_FOUND: i32 = 0x05;
-pub const ERR_OUTPUT_EXISTS: i32 = 0x06;
+pub const ERR_FILE_NOT_FOUND: i32 = 0x10;
+pub const ERR_FILE_INVALID: i32 = 0x11;
+pub const ERR_BINARY_NOT_FOUND: i32 = 0x20;
+pub const ERR_BINARY_NOT_EXECUTABLE: i32 = 0x21;
+pub const ERR_BINARY_INVALID: i32 = 0x22;
+pub const ERR_MODULE_NOT_IMPLEMENTED: i32 = 0x30;
+pub const ERR_ARGPARSE: i32 = 0x40;
+pub const ERR_OUTPUT_EXISTS: i32 = 0x50;
+
 
 fn build_message(code: i32, passed_object: Option<String>) -> String {
     let object = passed_object.unwrap_or_else(|| "".to_string());
     match code {
         WRN_GENERAL => format!("Warning: {}", object),
         ERR_GENERAL => format!("Error: {}", object),
         ERR_FILE_NOT_FOUND => format!("File not found: {}", object),
+        ERR_FILE_INVALID => format!("Invalid file given: {}", object),
+        ERR_BINARY_NOT_FOUND => format!("Binary not found: {}", object),
+        ERR_BINARY_NOT_EXECUTABLE => format!("Binary not executable: {}", object),
+        ERR_BINARY_INVALID => format!("Invalid binary given: {}", object),
         ERR_MODULE_NOT_IMPLEMENTED => format!("Module not implemented: {}", object),
         ERR_ARGPARSE => format!("Argument parsing error: {}", object),
-        ERR_BINARY_NOT_FOUND => format!("Binary not found: {}", object),
         ERR_OUTPUT_EXISTS => format!("Output file already exists: {}; use -o to overwrite", object),
+
         _ => "Unknown error".to_string(),
     }
 }

src/envs/variables.rs

@@ -85,18 +85,20 @@ pub fn locate_encoder_py() -> String {
 }
 
 // binary paths
-const VALID_BINARY: [&str; 7] = [
-    "mmseqs", "foldseek", "mafft", "mafft-linsi", "foldmason", "iqtree", "fasttree",
+pub const VALID_BINARY: [&str; 8] = [
+    "mmseqs", "foldseek", "mafft", "mafft-linsi", "foldmason", "iqtree", "fasttree", "raxml"
 ];
 pub struct Binary {
     name: String,
     pub path: String,
+    pub set: bool,
 }
 impl Binary {
     fn new(name: &str, path: &str) -> Self {
         Binary {
             name: name.to_string(),
             path: path.to_string(),
+            set: false,
         }
     }
     fn test(&self, args: Vec<&str>) -> bool {
@@ -129,8 +131,10 @@ impl BinaryPaths {
             let mut split = line.split('=');
             let name = split.next().unwrap_or("");
             let path = split.next().unwrap_or("");
+            if path.len() == 0 { continue; }
             if let Some(&i) = self.map.get(name) {
                 self.bin[i].path = path.to_string();
+                self.bin[i].set = true;
             }
         }
         Ok(())

src/main.rs

@@ -44,6 +44,9 @@ fn run(args: &parser::Args, bin: &var::BinaryPaths, test: bool) -> Result<(), Bo
         Some(parser::Commands::EasySearch { .. }) => {
             workflow::easy_search::run(args, bin).unwrap_or_else(|e| err::error(err::ERR_GENERAL, Some(e.to_string())));
         }
+        Some(parser::Commands::Config { .. }) => {
+            modules::config::run(args, bin).unwrap_or_else(|e| err::error(err::ERR_GENERAL, Some(e.to_string())));
+        },
         /* Some(_) => {
             err::error(err::ERR_MODULE_NOT_IMPLEMENTED, std::env::args().nth(1));
         } */

src/modules/config.rs

@@ -0,0 +1,93 @@
+use std::os::unix::fs::MetadataExt;
+use std::io::Write;
+use crate::envs::error_handler as err;
+use crate::envs::variables as var;
+use crate::envs::variables::BinaryPaths;
+use crate::util::command as cmd;
+use crate::util::message as msg;
+use crate::util::arg_parser::Args;
+use color_print::cstr;
+
+fn task_check(bin: &BinaryPaths) -> Result<(), Box<dyn std::error::Error>> {
+    msg::println_message(&format!("{}", cstr!(r#"<bold><underline>System:</underline></bold>"#)), 3);
+    msg::println_message(&format!("Unicore version: {}", var::VERSION), 3);
+    msg::println_message(&format!("OS: {}", std::env::consts::OS), 3);
+    msg::println_message(&format!("Threads: {}", var::threads()), 3);
+    println!();
+    msg::println_message(&format!("{}", cstr!(r#"<bold><underline>Dependencies:</underline></bold>"#)), 3);
+    msg::println_message(&format!("MMseqs2: {} .. {}",
+                                  if let Some(&ref bin) = &bin.get("mmseqs") { if bin.set { bin.path.clone() } else { "Unset".to_string() } } else { "Undefined".to_string() },
+                                  if let Some(&ref bin) = &bin.get("mmseqs") { if bin.set { if binary_run_test(&bin.path, "mmseqs") { cstr!(r#"<green>ok</green>"#) } else { cstr!(r#"<red>no</red>"#) } } else { cstr!(r#"<dim>n/a</dim>"#) } } else { cstr!(r#"<dim>n/a</dim>"#) },
+    ), 3);
+    msg::println_message(&format!("Foldseek: {} .. {}",
+                                  if let Some(&ref bin) = &bin.get("foldseek") { if bin.set { bin.path.clone() } else { "Unset".to_string() } } else { "Undefined".to_string() },
+                                  if let Some(&ref bin) = &bin.get("foldseek") { if bin.set { if binary_run_test(&bin.path, "foldseek") { cstr!(r#"<green>ok</green>"#) } else { cstr!(r#"<red>no</red>"#) } } else { cstr!(r#"<dim>n/a</dim>"#) } } else { cstr!(r#"<dim>n/a</dim>"#) },
+    ), 3);
+    msg::println_message(&format!("FoldMason: {} .. {}",
+                                  if let Some(&ref bin) = &bin.get("foldmason") { if bin.set { bin.path.clone() } else { "Unset".to_string() } } else { "Undefined".to_string() },
+                                  if let Some(&ref bin) = &bin.get("foldmason") { if bin.set { if binary_run_test(&bin.path, "foldmason") { cstr!(r#"<green>ok</green>"#) } else { cstr!(r#"<red>no</red>"#) } } else { cstr!(r#"<dim>n/a</dim>"#) } } else { cstr!(r#"<dim>n/a</dim>"#) },
+    ), 3);
+    msg::println_message(&format!("MAFFT: {} .. {}",
+                                  if let Some(&ref bin) = &bin.get("mafft") { if bin.set { bin.path.clone() } else { "Unset".to_string() } } else { "Undefined".to_string() },
+                                  if let Some(&ref bin) = &bin.get("mafft") { if bin.set { if binary_run_test(&bin.path, "mafft") { cstr!(r#"<green>ok</green>"#) } else { cstr!(r#"<red>no</red>"#) } } else { cstr!(r#"<dim>n/a</dim>"#) } } else { cstr!(r#"<dim>n/a</dim>"#) },
+    ), 3);
+    msg::println_message(&format!("IQ-TREE: {} .. {}",
+                                  if let Some(&ref bin) = &bin.get("iqtree") { if bin.set { bin.path.clone() } else { "Unset".to_string() } } else { "Undefined".to_string() },
+                                  if let Some(&ref bin) = &bin.get("iqtree") { if bin.set { if binary_run_test(&bin.path, "iqtree") { cstr!(r#"<green>ok</green>"#) } else { cstr!(r#"<red>no</red>"#) } } else { cstr!(r#"<dim>n/a</dim>"#) } } else { cstr!(r#"<dim>n/a</dim>"#) },
+    ), 3);
+    msg::println_message(&format!("FastTree: {} .. {}",
+                                  if let Some(&ref bin) = &bin.get("fasttree") { if bin.set { bin.path.clone() } else { "Unset".to_string() } } else { "Undefined".to_string() },
+                                  if let Some(&ref bin) = &bin.get("fasttree") { if bin.set { if binary_run_test(&bin.path, "fasttree") { cstr!(r#"<green>ok</green>"#) } else { cstr!(r#"<red>no</red>"#) } } else { cstr!(r#"<dim>n/a</dim>"#) } } else { cstr!(r#"<dim>n/a</dim>"#) },
+    ), 3);
+    msg::println_message(&format!("RAxML: {} .. {}",
+                                  if let Some(&ref bin) = &bin.get("raxml") { if bin.set { bin.path.clone() } else { "Unset".to_string() } } else { "Undefined".to_string() },
+                                  if let Some(&ref bin) = &bin.get("raxml") { if bin.set { if binary_run_test(&bin.path, "raxml") { cstr!(r#"<green>ok</green>"#) } else { cstr!(r#"<red>no</red>"#) } } else { cstr!(r#"<dim>n/a</dim>"#) } } else { cstr!(r#"<dim>n/a</dim>"#) },
+    ), 3);
+    Ok(())
+}
+
+fn binary_run_test(path: &str, sw: &str) -> bool {
+    if !var::VALID_BINARY.contains(&sw) { return false; }
+    let mut test_command = std::process::Command::new(path);
+    let test_command = match sw {
+        "mmseqs" | "foldseek" | "foldmason" => test_command.arg("version"),
+        "mafft" | "mafft-linsi" | "iqtree" => test_command.arg("--version"),
+        "fasttree" => &mut test_command,
+        "raxml" => test_command.arg("-v"),
+        _ => return false,
+    };
+    cmd::run_code(test_command) == 0
+}
+
+fn set_binary(bin: &BinaryPaths, path: &str, sw: &str) -> Result<(), Box<dyn std::error::Error>> {
+    if !std::fs::File::open(path).is_ok() { err::error(err::ERR_BINARY_NOT_FOUND, Some(format!("{}", path))); }
+    if std::fs::metadata(path)?.is_dir() { err::error(err::ERR_FILE_INVALID, Some(format!("{}", path))); }
+    if std::fs::metadata(path)?.mode() & 0o111 == 0 { err::error(err::ERR_BINARY_NOT_EXECUTABLE, Some(format!("{}", path))); }
+    if !binary_run_test(path, sw) { err::error(err::ERR_BINARY_INVALID, Some(format!("{}", path))); }
+
+    let path = std::fs::canonicalize(path)?.to_str().unwrap().to_string();
+    msg::println_message(&format!("Setting dependency {} to {}...", sw, path), 3);
+    let mut cfg = std::fs::File::create(var::locate_path_cfg())?;
+    for &prog in var::VALID_BINARY.iter() {
+        if prog == sw { cfg.write_all(format!("{}={}\n", prog, path).as_bytes())?; }
+        else if bin.get(prog).is_none() || !bin.get(prog).unwrap().set { cfg.write_all(format!("#{}=\n", prog).as_bytes())?; }
+        else { cfg.write_all(format!("{}={}\n", prog, bin.get(prog).unwrap().path).as_bytes())?; }
+    }
+    msg::println_message(&"Done. Please run \"unicore config -c\" to check".to_string(), 3);
+
+    Ok(())
+}
+
+pub fn run(args: &Args, bin: &BinaryPaths) -> Result<(), Box<dyn std::error::Error>> {
+    if args.config_check.is_some() && args.config_check.unwrap() { task_check(bin)?; }
+    else if args.config_set_mmseqs.is_some() { set_binary(bin, args.config_set_mmseqs.clone().unwrap().as_str(), "mmseqs")?; }
+    else if args.config_set_foldseek.is_some() { set_binary(bin, args.config_set_foldseek.clone().unwrap().as_str(), "foldseek")?; }
+    else if args.config_set_foldmason.is_some() { set_binary(bin, args.config_set_foldmason.clone().unwrap().as_str(), "foldmason")?; }
+    else if args.config_set_mafft.is_some() { set_binary(bin, args.config_set_mafft.clone().unwrap().as_str(), "mafft")?; }
+    else if args.config_set_mafft_linsi.is_some() { set_binary(bin, args.config_set_mafft_linsi.clone().unwrap().as_str(), "mafft-linsi")?; }
+    else if args.config_set_iqtree.is_some() { set_binary(bin, args.config_set_iqtree.clone().unwrap().as_str(), "iqtree")?; }
+    else if args.config_set_fasttree.is_some() { set_binary(bin, args.config_set_fasttree.clone().unwrap().as_str(), "fasttree")?; }
+    else if args.config_set_raxml.is_some() { set_binary(bin, args.config_set_raxml.clone().unwrap().as_str(), "raxml")?; }
+    else { err::error(err::ERR_ARGPARSE, Some("No task specified".to_string())) };
+    Ok(())
+}

src/modules/mod.rs

@@ -4,4 +4,5 @@ pub mod cluster;
 pub mod search;
 pub mod profile;
 pub mod tree;
-pub mod genetree;
+pub mod genetree;
+pub mod config;

src/util/arg_parser.rs

@@ -362,6 +362,40 @@ pub enum Commands {
         #[arg(short='v', long, default_value="3")]
         verbosity: u8,
     },
+    /// Runtime environment configuration
+    #[clap(arg_required_else_help = true, allow_hyphen_values = true)]
+    Config {
+        /// Check current environment configuration
+        #[arg(short='c', long)]
+        check: bool,
+        /// Set mmseqs binary path
+        #[arg(long)]
+        set_mmseqs: Option<PathBuf>,
+        /// Set foldseek binary path
+        #[arg(long)]
+        set_foldseek: Option<PathBuf>,
+        /// Set foldmason binary path
+        #[arg(long)]
+        set_foldmason: Option<PathBuf>,
+        /// Set mafft binary path
+        #[arg(long)]
+        set_mafft: Option<PathBuf>,
+        /// Set mafft-linsi binary path
+        #[arg(long)]
+        set_mafft_linsi: Option<PathBuf>,
+        /// Set iqtree binary path
+        #[arg(long)]
+        set_iqtree: Option<PathBuf>,
+        /// Set fasttree binary path
+        #[arg(long)]
+        set_fasttree: Option<PathBuf>,
+        /// Set raxml binary path
+        #[arg(long)]
+        set_raxml: Option<PathBuf>,
+        /// Verbosity (0: quiet, 1: +errors, 2: +warnings, 3: +info, 4: +debug)
+        #[arg(short='v', long, default_value="3")]
+        verbosity: u8,
+    },
 }
 
 #[derive(Default)]
@@ -419,6 +453,16 @@ pub struct Args {
     pub genetree_realign: Option<bool>,
     pub genetree_aligner: Option<String>,
     pub genetree_aligner_options: Option<Option<String>>,
+
+    pub config_check: Option<bool>,
+    pub config_set_mmseqs: Option<String>,
+    pub config_set_foldseek: Option<String>,
+    pub config_set_foldmason: Option<String>,
+    pub config_set_mafft: Option<String>,
+    pub config_set_mafft_linsi: Option<String>,
+    pub config_set_iqtree: Option<String>,
+    pub config_set_fasttree: Option<String>,
+    pub config_set_raxml: Option<String>,
 }
 fn own(path: &PathBuf) -> String { path.clone().to_string_lossy().into_owned() }
 impl Args {
@@ -433,6 +477,7 @@ impl Args {
             Some(GeneTree { verbosity, .. }) => *verbosity,
             Some(EasyCore { verbosity, .. }) => *verbosity,
             Some(EasySearch { verbosity, .. }) => *verbosity,
+            Some(Config { verbosity, .. }) => *verbosity,
             _ => 3,
         };
         let threads = match &args.command {
@@ -641,6 +686,34 @@ impl Args {
             Some(GeneTree { threshold, .. }) => Some(*threshold), _ => None,
         };
 
+        let config_check = match &args.command {
+            Some(Config { check, .. }) => Some(*check), _ => None,
+        };
+        let config_set_mmseqs = match &args.command {
+            Some(Config { set_mmseqs, .. }) => match set_mmseqs { Some(p) => Some(own(p)), _ => None }, _ => None,
+        };
+        let config_set_foldseek = match &args.command {
+            Some(Config { set_foldseek, .. }) => match set_foldseek { Some(p) => Some(own(p)), _ => None }, _ => None,
+        };
+        let config_set_foldmason = match &args.command {
+            Some(Config { set_foldmason, .. }) => match set_foldmason { Some(p) => Some(own(p)), _ => None }, _ => None,
+        };
+        let config_set_mafft = match &args.command {
+            Some(Config { set_mafft, .. }) => match set_mafft { Some(p) => Some(own(p)), _ => None }, _ => None,
+        };
+        let config_set_mafft_linsi = match &args.command {
+            Some(Config { set_mafft_linsi, .. }) => match set_mafft_linsi { Some(p) => Some(own(p)), _ => None }, _ => None,
+        };
+        let config_set_iqtree = match &args.command {
+            Some(Config { set_iqtree, .. }) => match set_iqtree { Some(p) => Some(own(p)), _ => None }, _ => None,
+        };
+        let config_set_fasttree = match &args.command {
+            Some(Config { set_fasttree, .. }) => match set_fasttree { Some(p) => Some(own(p)), _ => None }, _ => None,
+        };
+        let config_set_raxml = match &args.command {
+            Some(Config { set_raxml, .. }) => match set_raxml { Some(p) => Some(own(p)), _ => None }, _ => None,
+        };
+
         Args {
             command: args.command, version: args.version, threads, verbosity,
             createdb_input, createdb_output, createdb_model, createdb_keep, createdb_overwrite, createdb_max_len, createdb_gpu, createdb_use_python, createdb_use_foldseek, createdb_afdb_lookup, createdb_afdb_local,
@@ -649,6 +722,7 @@ impl Args {
             cluster_input, cluster_output, cluster_tmp, cluster_keep_cluster_db, cluster_cluster_options,
             tree_db, tree_input, tree_output, tree_aligner, tree_tree_builder, tree_aligner_options, tree_tree_options, tree_threshold,
             genetree_input, genetree_names, genetree_tree_builder, genetree_tree_options, genetree_realign, genetree_aligner, genetree_aligner_options, genetree_threshold,
+            config_check, config_set_mmseqs, config_set_foldseek, config_set_foldmason, config_set_mafft, config_set_mafft_linsi, config_set_iqtree, config_set_fasttree, config_set_raxml,
         }
     }
 }

src/util/command.rs

@@ -23,6 +23,26 @@ pub fn run(cmd: &mut std::process::Command) {
     }
 }
 
+pub fn run_code(cmd: &mut std::process::Command) -> i32 {
+    let cmd = cmd.stdout(std::process::Stdio::null()).stderr(std::process::Stdio::null());
+    let cmdstr = format!("{:?}", cmd).replace("\"", "");
+    msg::println_message(&format!("Running command: {}", cmdstr), 4);
+    if let Ok(mut child) = cmd.spawn() {
+        let wait = child.wait();
+        if let Ok(status) = wait {
+            if let Some(code) = status.code() {
+                code
+            } else {
+                1
+            }
+        } else {
+            1
+        }
+    } else {
+        1
+    }
+}
+
 pub fn _run_at(cmd: &mut std::process::Command, path: &std::path::Path) {
     let cmdstr = format!("{:?}", cmd);
     if let Ok(mut child) = cmd.current_dir(path).spawn() {