From 757b2e888007dcac15aef964e7193ecd2eee1b7e Mon Sep 17 00:00:00 2001
From: Dermot Haughey <hderms@gmail.com>
Date: Wed, 28 Apr 2021 09:30:22 -0500
Subject: [PATCH] add alternate delimiter (#13)

* add alternate delimiter

* fix test
---
 src/csv/csv_data.rs             |  6 ++++--
 src/main.rs                     |  8 +++++++-
 src/qsv.rs                      | 10 +++++++---
 testdata/slash_as_separator.csv |  3 +++
 tests/integration.rs            |  9 +++++++++
 5 files changed, 30 insertions(+), 6 deletions(-)
 create mode 100644 testdata/slash_as_separator.csv

diff --git a/src/csv/csv_data.rs b/src/csv/csv_data.rs
index c721e8c..ea31386 100644
--- a/src/csv/csv_data.rs
+++ b/src/csv/csv_data.rs
@@ -28,11 +28,12 @@ pub struct CsvData {
 }
 impl CsvData {
     ///Load CSVData from a filename
-    pub fn from_filename(filename: &str) -> Result<CsvData, Box<dyn Error>> {
+    pub fn from_filename(filename: &str, delimiter: char) -> Result<CsvData, Box<dyn Error>> {
         debug!("Trying to load CSV from filename {}", filename);
         let mut records = Vec::with_capacity(10000);
         let mut rdr = csv::ReaderBuilder::new()
             .buffer_capacity(16 * (1 << 10))
+            .delimiter(delimiter as u8)
             .from_path(filename)?;
 
         for result in rdr.records() {
@@ -52,9 +53,10 @@ impl CsvData {
 #[cfg(test)]
 mod tests {
     use super::*;
+    const delimiter: char = ',';
     #[test]
     fn it_can_load_file() {
-        let csv = CsvData::from_filename("testdata/test.csv").unwrap();
+        let csv = CsvData::from_filename("testdata/test.csv", delimiter).unwrap();
         assert_eq!(csv.records, vec!(StringRecord::from(vec!("bar", "13"))))
     }
 }
diff --git a/src/main.rs b/src/main.rs
index 74783b1..e959e77 100644
--- a/src/main.rs
+++ b/src/main.rs
@@ -7,18 +7,24 @@ use crate::qsv::{execute_query, write_to_stdout};
 use std::error::Error;
 use simple_logger::SimpleLogger;
 use clap::{AppSettings, Clap};
+use crate::qsv::Options;
+
 #[derive(Clap)]
 #[clap(version = "0.1", author = "Dermot H. <dermot.thomas.haughey@gmail.com>")]
 #[clap(setting = AppSettings::ColoredHelp)]
 struct Opts {
     query: String,
+    #[clap(short, long, default_value=",")]
+    delimiter: char
 }
 
 fn main() -> Result<(), Box<dyn Error>> {
     SimpleLogger::from_env().init()?;
     let opts: Opts = Opts::parse();
     let query = opts.query;
-    let results = execute_query(query.as_str())?;
+    let delimiter = opts.delimiter;
+    let options = Options{delimiter};
+    let results = execute_query(query.as_str(), &options)?;
     write_to_stdout(results)?;
     Ok(())
 }
diff --git a/src/qsv.rs b/src/qsv.rs
index 1fe8206..408b7da 100644
--- a/src/qsv.rs
+++ b/src/qsv.rs
@@ -14,9 +14,12 @@ use uuid::Uuid;
 use log::debug;
 
 type Rows = Vec<Vec<String>>;
+pub struct Options {
+    pub delimiter: char
+}
 
 ///Executes a query, possibly returning Rows
-pub fn execute_query(query: &str) -> Result<Rows, Box<dyn Error>> {
+pub fn execute_query(query: &str, options: &Options) -> Result<Rows, Box<dyn Error>> {
     let mut collector = Collector::new();
 
     let ast = Parser::parse_sql(query)?;
@@ -26,7 +29,7 @@ pub fn execute_query(query: &str) -> Result<Rows, Box<dyn Error>> {
     collector.collect(statement); //TODO: should we handle multiple SQL statements later?
     let mut files_to_tables = HashMap::new();
     for filename in collector.table_identifiers.iter() {
-        if let Ok(()) = maybe_load_file(&mut files_to_tables, filename, &mut db) {
+        if let Ok(()) = maybe_load_file(&mut files_to_tables, filename, &mut db, options) {
             debug!("Potential filename from SQL was able to be loaded: {}", filename);
         } else {
             debug!("Identifier in SQL could not be loaded as file: {}", filename);
@@ -42,8 +45,9 @@ fn maybe_load_file(
     files_to_tables: &mut HashMap<String, String>,
     filename: &str,
     db: &mut Db,
+    options: &Options
 ) -> Result<(), Box<dyn Error>> {
-    let csv = CsvData::from_filename(filename)?;
+    let csv = CsvData::from_filename(filename, options.delimiter)?;
     let path = Path::new(filename);
     debug!("Attempting to load identifier from SQL as file: {}", filename);
     let table_name = path.file_stem(); //TODO: should we canonicalize path?
diff --git a/testdata/slash_as_separator.csv b/testdata/slash_as_separator.csv
new file mode 100644
index 0000000..d088838
--- /dev/null
+++ b/testdata/slash_as_separator.csv
@@ -0,0 +1,3 @@
+occupation/minimum_age
+Bartender/32
+Construction Worker/25
diff --git a/tests/integration.rs b/tests/integration.rs
index bcf72ce..6aee10c 100644
--- a/tests/integration.rs
+++ b/tests/integration.rs
@@ -84,3 +84,12 @@ fn it_will_run_a_simple_query_with_stddev() -> Result<(), Box<dyn std::error::Er
     cmd.assert().success().stdout(predicates::str::contains("1.7078251276599"));
     Ok(())
 }
+
+#[test]
+fn it_will_run_with_an_alternate_delimiter() -> Result<(), Box<dyn std::error::Error>> {
+    let mut cmd = Command::cargo_bin("qsv")?;
+    cmd.arg("select min(minimum_age) from testdata/slash_as_separator.csv");
+    cmd.arg("--delimiter=/");
+    cmd.assert().success().stdout(predicates::str::contains("25"));
+    Ok(())
+}