feat: Taglines from files (#7)

.gitignore

@@ -1 +1,3 @@
 /target
+
+/testfiles

Cargo.toml

@@ -8,3 +8,4 @@ edition = "2021"
 [dependencies]
 pest = "2.7.6"
 pest_derive = "2.7.6"
+walkdir = "2.4.0"

src/lib.rs

@@ -1,2 +1,5 @@
 /// parsers contains the relevant grammar parsers
 pub mod parsers;
+
+/// search contains functions for searching files and tags.
+pub mod search;

src/main.rs

@@ -1,3 +1,11 @@
-fn main() {
-    println!("Hello, world!");
+use tag::search::get_tags_from_files;
+
+fn main() -> Result<(), Box<dyn std::error::Error>> {
+    let tagged_files = get_tags_from_files("testfiles")?;
+
+    for file in tagged_files.iter() {
+        println!("File {} contains {:?}", file.path.display(), file.tags);
+    }
+
+    Ok(())
 }

src/parsers.rs

@@ -47,12 +47,6 @@ mod tests {
                 expected_tags: vec!["#1", "#asdf", "#something-idk"],
                 expected_error: false,
             },
-            TestCase {
-                name: "success_with_newline",
-                input: "tags:\n[\n\t#something_else\n]",
-                expected_tags: vec!["#something_else"],
-                expected_error: false,
-            },
             TestCase {
                 name: "fail_no_brackets",
                 input: "tags:#1#2#3",

src/search.rs

@@ -0,0 +1,60 @@
+use std::{
+    fs,
+    io::{BufRead, BufReader},
+    path::{Path, PathBuf},
+};
+
+use pest::Parser;
+use walkdir::WalkDir;
+
+use crate::parsers::tagline::{self, TaglineParser};
+
+/// TaggedFile is a file that contains tags.
+#[derive(Clone)]
+pub struct TaggedFile {
+    pub path: PathBuf,
+    pub tags: Vec<String>,
+}
+
+/// get_tags_from_file() returns a list of tags found in a file.
+/// It will return an error if a file has no parsable tags.
+fn get_tags_from_file(file: &Path) -> Result<Vec<String>, Box<dyn std::error::Error>> {
+    let file = fs::File::open(file)?;
+    let mut buffer = BufReader::new(file);
+    let mut tagline = String::new();
+    let _ = buffer.read_line(&mut tagline)?;
+
+    let parsed = TaglineParser::parse(tagline::Rule::tagline, tagline.trim())?;
+
+    let mut tags = Vec::new();
+
+    for tag in parsed {
+        if tag.as_rule() == tagline::Rule::tag {
+            tags.push(tag.as_str().to_string())
+        }
+    }
+
+    Ok(tags)
+}
+
+/// get_tags_from_files() recursively retrieves the tags of all files
+/// in a given directory.
+pub fn get_tags_from_files(directory: &str) -> Result<Vec<TaggedFile>, Box<dyn std::error::Error>> {
+    let mut tagged_files = Vec::new();
+
+    for entry in WalkDir::new(directory).follow_links(true) {
+        let entry = entry?;
+
+        if entry.file_type().is_dir() {
+            continue;
+        }
+
+        let tags = get_tags_from_file(entry.path())?;
+        tagged_files.push(TaggedFile {
+            path: entry.path().to_owned(),
+            tags,
+        })
+    }
+
+    Ok(tagged_files.clone())
+}

tagline.pest

@@ -2,5 +2,5 @@ tag = {"#" ~ (LETTER|NUMBER|CONNECTOR_PUNCTUATION|DASH_PUNCTUATION)+}
 taglist = _{"[" ~ tag* ~ "]"}
 tagline = _{SOI ~ "tags:" ~ taglist ~ EOI}
 
-WHITESPACE = _{" " | "\t" | NEWLINE}
+WHITESPACE = _{" " | "\t"}