id3 improvements - extract year/bpm as numbers; fix id3 extract job name; exit early if file isn't audio

also add EntryValue::guess_from

Cargo.lock

@@ -2918,6 +2918,7 @@ dependencies = [
  "once_cell",
  "opener",
  "rayon",
+ "regex",
  "serde",
  "serde_json",
  "tempfile",

Cargo.toml

@@ -43,6 +43,7 @@ chrono = { version = "0.4", features = ["serde"] }
 serde = { version = "1.0", features = ["derive"] }
 serde_json = "1.0"
 lexpr = "0.2.6"
+regex = "1"
 
 bs58 = "^0.4"
 filebuffer = "0.4.0"

src/database/entry.rs

@@ -2,6 +2,7 @@ use crate::addressing::{Address, Addressable};
 use crate::database::inner::models;
 use crate::util::hash::{b58_decode, hash, Hash, Hashable};
 use anyhow::{anyhow, Result};
+use regex::Regex;
 use serde::{Deserialize, Serialize};
 use std::convert::TryFrom;
 use std::io::{Cursor, Write};
@@ -161,6 +162,23 @@ impl EntryValue {
 
         Ok(format!("{}{}", type_char, content))
     }
+
+    pub fn guess_from<S: AsRef<str>>(string: S) -> Self {
+        let string = string.as_ref();
+        match string.parse::<f64>() {
+            Ok(num) => EntryValue::Number(num),
+            Err(_) => {
+                lazy_static! {
+                    static ref URL_REGEX: Regex = Regex::new("^[a-zA-Z0-9_]://").unwrap();
+                }
+                if URL_REGEX.is_match(string) {
+                    EntryValue::Address(Address::Url(string.to_string()))
+                } else {
+                    EntryValue::String(string.to_string())
+                }
+            }
+        }
+    }
 }
 
 impl std::str::FromStr for EntryValue {

src/extractors/audio.rs

@@ -1,7 +1,11 @@
 use super::Extractor;
 use crate::{
     addressing::Address,
-    database::{entry::Entry, UpEndConnection},
+    database::{
+        entry::{Entry, EntryValue},
+        UpEndConnection,
+    },
+    filesystem::FILE_MIME_KEY,
     util::jobs::{Job, JobContainer, State},
 };
 use anyhow::{anyhow, Result};
@@ -17,6 +21,19 @@ impl Extractor for ID3Extractor {
         job_container: Arc<RwLock<JobContainer>>,
     ) -> Result<Vec<Entry>> {
         if let Address::Hash(hash) = address {
+            let is_audio = connection.retrieve_object(address)?.iter().any(|e| {
+                if e.attribute == FILE_MIME_KEY {
+                    if let EntryValue::String(mime) = &e.value {
+                        return mime.starts_with("audio") || mime == "application/x-riff";
+                    }
+                }
+                false
+            });
+
+            if !is_audio {
+                return Ok(vec![]);
+            }
+
             let files = connection.retrieve_file(hash)?;
 
             if let Some(file) = files.get(0) {
@@ -26,8 +43,13 @@ impl Extractor for ID3Extractor {
                     .add_job(Job::new(
                         None,
                         &format!(
-                            "Getting ID3 info from {:?}",
+                            "Getting ID3 info from \"{:}\"",
-                            file.path.components().last().unwrap()
+                            file.path
+                                .components()
+                                .last()
+                                .unwrap()
+                                .as_os_str()
+                                .to_string_lossy()
                         ),
                     ))
                     .unwrap();
@@ -40,7 +62,10 @@ impl Extractor for ID3Extractor {
                         id3::Content::Text(text) => Some(Entry {
                             entity: address.clone(),
                             attribute: format!("ID3_{}", frame.id()),
-                            value: text.clone().into(),
+                            value: match frame.id() {
+                                "TYER" | "TBPM" => EntryValue::guess_from(text),
+                                _ => text.clone().into(),
+                            },
                         }),
                         _ => None,
                     })

src/filesystem.rs

@@ -27,7 +27,7 @@ use walkdir::WalkDir;
 
 const BLOB_TYPE: &str = "BLOB";
 const ALIAS_KEY: &str = "ALIAS";
-const FILE_MIME_KEY: &str = "FILE_MIME";
+pub const FILE_MIME_KEY: &str = "FILE_MIME";
 const FILE_MTIME_KEY: &str = "FILE_MTIME";
 const FILE_SIZE_KEY: &str = "FILE_SIZE";