From b87422dbc3a5656f49c887c1eb606a6fb5f9fe1c Mon Sep 17 00:00:00 2001
From: Jakob Ledermann <jakob-ledermann@users.noreply.github.com>
Date: Tue, 13 Sep 2022 22:57:04 +0200
Subject: [PATCH] Add a commandline option '--delimiter' to `map`.

This commandline option allows to specify the delimiter between keys and the values
in the input file to the `map` subcommand.

The provided delimiter should be a single byte utf-8 character.
If the value consists of multiple bytes only the first one is used.
This parameter is passed to `csv::ReaderBuilder::delimiter` which only
allows for a single byte as delimiter.
---
 fst-bin/src/app.rs     |  5 ++++-
 fst-bin/src/cmd/map.rs | 24 +++++++++++++++++++++++-
 fst-bin/src/util.rs    |  6 ++++--
 3 files changed, 31 insertions(+), 4 deletions(-)

diff --git a/fst-bin/src/app.rs b/fst-bin/src/app.rs
index f8ecb4c8..c18dc790 100644
--- a/fst-bin/src/app.rs
+++ b/fst-bin/src/app.rs
@@ -288,7 +288,10 @@ pub fn app() -> clap::App<'static, 'static> {
         ))
         .arg(flag("keep-tmp-dir").help(
             "Does not delete the temporary directory. Useful for debugging.",
-        ));
+        ))
+        .arg(flag("delimiter").help(
+            "The delimiter used in the CSV file to separate key and value in each line. \
+             This defaults to ','.",));
 
     let node = cmd("node", ABOUT_NODE)
         .arg(pos("input").required(true).help("The FST to inspect."))
diff --git a/fst-bin/src/cmd/map.rs b/fst-bin/src/cmd/map.rs
index d6e356e2..4a49d422 100644
--- a/fst-bin/src/cmd/map.rs
+++ b/fst-bin/src/cmd/map.rs
@@ -26,6 +26,17 @@ struct Args {
     keep_tmp_dir: bool,
     max: bool,
     min: bool,
+    delimiter: Option<u8>,
+}
+
+#[derive(Debug, Clone, Copy, PartialEq, Eq)]
+struct DelimiterInvalidError;
+
+impl std::error::Error for DelimiterInvalidError {}
+impl std::fmt::Display for DelimiterInvalidError {
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+        write!(f, "The provided value is no valid delimiter")
+    }
 }
 
 impl Args {
@@ -47,6 +58,15 @@ impl Args {
             keep_tmp_dir: m.is_present("keep-tmp-dir"),
             max: m.is_present("max"),
             min: m.is_present("min"),
+            delimiter: m
+                .value_of_lossy("delimiter")
+                .map(|x| {
+                    x.as_bytes()
+                        .get(0)
+                        .map(|y| *y)
+                        .ok_or(DelimiterInvalidError)
+                })
+                .transpose()?,
         })
     }
 
@@ -66,6 +86,7 @@ impl Args {
         let mut map = MapBuilder::new(wtr)?;
         for input in &self.input {
             let mut rdr = csv::ReaderBuilder::new()
+                .delimiter(self.delimiter.unwrap_or(b','))
                 .has_headers(false)
                 .from_reader(util::get_reader(Some(input))?);
             for row in rdr.deserialize() {
@@ -82,7 +103,8 @@ impl Args {
             .iter()
             .map(|inp| Path::new(inp).to_path_buf())
             .collect();
-        let keys = util::ConcatCsv::new(inputs);
+        let keys =
+            util::ConcatCsv::new(inputs, self.delimiter.unwrap_or(b','));
 
         let mut merger = Merger::new(keys, &self.output);
         merger = merger.fd_limit(self.fd_limit);
diff --git a/fst-bin/src/util.rs b/fst-bin/src/util.rs
index dfe4970c..17028f03 100644
--- a/fst-bin/src/util.rs
+++ b/fst-bin/src/util.rs
@@ -135,15 +135,16 @@ impl Iterator for ConcatLines {
 pub struct ConcatCsv {
     inputs: Vec<PathBuf>,
     cur: Option<Rows>,
+    delimiter: u8,
 }
 
 type Reader = Box<dyn io::Read + Send + Sync + 'static>;
 type Rows = csv::DeserializeRecordsIntoIter<Reader, (BString, u64)>;
 
 impl ConcatCsv {
-    pub fn new(mut inputs: Vec<PathBuf>) -> ConcatCsv {
+    pub fn new(mut inputs: Vec<PathBuf>, delimiter: u8) -> ConcatCsv {
         inputs.reverse(); // treat it as a stack
-        ConcatCsv { inputs, cur: None }
+        ConcatCsv { inputs, cur: None, delimiter }
     }
 
     fn read_row(&mut self) -> Option<Result<(BString, u64), Error>> {
@@ -173,6 +174,7 @@ impl Iterator for ConcatCsv {
                             Ok(rdr) => rdr,
                         };
                         let csvrdr = csv::ReaderBuilder::new()
+                            .delimiter(self.delimiter)
                             .has_headers(false)
                             .from_reader(rdr);
                         self.cur = Some(csvrdr.into_deserialize());