Merge pull request #10 from jblondin/fix/remove-undetected-metadata

Removed undetected metadata dialect options
jblondin · Mar 20, 2022 · 01eb1f8 · 01eb1f8
2 parents b7bbae5 + d46cbda
commit 01eb1f8
Show file tree

Hide file tree

Showing 4 changed files with 5 additions and 75 deletions.
diff --git a/examples/worldbank.rs b/examples/worldbank.rs
@@ -3,7 +3,6 @@ extern crate csv_sniffer;
 
 use std::path::Path;
 
-use csv::Terminator;
 use csv_sniffer::metadata::*;
 
 fn main() {
@@ -18,10 +17,6 @@ fn main() {
             num_preamble_rows: 4,
         },
         quote: Quote::Some(b'"'),
-        doublequote_escapes: true,
-        comment: Comment::Disabled,
-        escape: Escape::Disabled,
-        terminator: Terminator::CRLF,
         flexible: false,
     };
     let mut reader = dialect.open_path(data_filepath).unwrap();

diff --git a/src/metadata.rs b/src/metadata.rs
@@ -6,7 +6,7 @@ use std::fs::File;
 use std::io::{Read, Seek};
 use std::path::Path;
 
-use csv::{Reader, ReaderBuilder, Terminator};
+use csv::{Reader, ReaderBuilder};
 
 use error::*;
 use field_type::Type;
@@ -41,45 +41,22 @@ impl fmt::Display for Metadata {
 
 /// Dialect-level metadata. This type encapsulates the details to be used to derive a
 /// `ReaderBuilder` object (in the [`csv`](https://docs.rs/csv) crate).
-///
-/// Not all components of this type are currently detected by the sniffer, and may be detected in
-/// the future.
 #[derive(Clone)]
 pub struct Dialect {
-    /// CSV delimiter (field separator). Detected by sniffer.
+    /// CSV delimiter (field separator).
     pub delimiter: u8,
     /// [`Header`](struct.Header.html) subtype (header row boolean and number of preamble rows).
-    /// Detected by sniffer.
     pub header: Header,
-    /// Record terminator. Currently not detected by sniffer; defaults to `Terminator::CRLF`.
-    pub terminator: Terminator,
-    /// Record quoting details. Detected by sniffer.
+    /// Record quoting details.
     pub quote: Quote,
-    /// Whether or not doubled quotes are interpreted as escapes. Currently not detected by sniffer;
-    /// defaults to `true`.
-    pub doublequote_escapes: bool,
-    /// Character used as escape, if any. Currently not detected by sniffer; defaults to
-    /// `Escape::Disabled` (to escape a quote, use double quotes).
-    pub escape: Escape,
-    /// Character used as comment, if any. Currently not detected by sniffer; defaults to
-    /// `Comment::Disabled`.
-    pub comment: Comment,
-    /// Whether or not the number of fields in a record is allowed to change. Detected by sniffer.
+    /// Whether or not the number of fields in a record is allowed to change.
     pub flexible: bool,
 }
 impl PartialEq for Dialect {
     fn eq(&self, other: &Dialect) -> bool {
         self.delimiter == other.delimiter
             && self.header == other.header
-            && match (self.terminator, other.terminator) {
-                (Terminator::CRLF, Terminator::CRLF) => true,
-                (Terminator::Any(left), Terminator::Any(right)) => left == right,
-                _ => false,
-            }
             && self.quote == other.quote
-            && self.doublequote_escapes == other.doublequote_escapes
-            && self.escape == other.escape
-            && self.comment == other.comment
             && self.flexible == other.flexible
     }
 }
@@ -88,11 +65,7 @@ impl fmt::Debug for Dialect {
         f.debug_struct("Dialect")
             .field("delimiter", &char::from(self.delimiter))
             .field("header", &self.header)
-            .field("terminator", &self.terminator)
             .field("quote", &self.quote)
-            .field("doublequote_escapes", &self.doublequote_escapes)
-            .field("escape", &self.escape)
-            .field("comment", &self.comment)
             .field("flexible", &self.flexible)
             .finish()
     }
@@ -115,23 +88,6 @@ impl fmt::Display for Dialect {
                 Quote::None => "none".into(),
             }
         )?;
-        writeln!(f, "\tDouble-quote escapes?: {}", self.doublequote_escapes)?;
-        writeln!(
-            f,
-            "\tEscape character: {}",
-            match self.escape {
-                Escape::Enabled(chr) => format!("{}", char::from(chr)),
-                Escape::Disabled => "none".into(),
-            }
-        )?;
-        writeln!(
-            f,
-            "\tComment character: {}",
-            match self.comment {
-                Comment::Enabled(chr) => format!("{}", char::from(chr)),
-                Comment::Disabled => "none".into(),
-            }
-        )?;
         writeln!(f, "\tFlexible: {}", self.flexible)
     }
 }
@@ -155,10 +111,6 @@ impl From<Dialect> for ReaderBuilder {
         let mut bldr = ReaderBuilder::new();
         bldr.delimiter(dialect.delimiter)
             .has_headers(dialect.header.has_header_row)
-            .terminator(dialect.terminator)
-            .escape(dialect.escape.into())
-            .double_quote(dialect.doublequote_escapes)
-            .comment(dialect.comment.into())
             .flexible(dialect.flexible);
 
         match dialect.quote {

diff --git a/src/sniffer.rs b/src/sniffer.rs
@@ -3,7 +3,7 @@ use std::fs::File;
 use std::io::{Read, Seek, SeekFrom};
 use std::path::Path;
 
-use csv::{self, Reader, StringRecord, Terminator};
+use csv::{self, Reader, StringRecord};
 use csv_core as csvc;
 use regex::Regex;
 
@@ -128,11 +128,7 @@ impl Sniffer {
                     num_preamble_rows: self.num_preamble_rows.unwrap(),
                     has_header_row: self.has_header_row.unwrap(),
                 },
-                terminator: Terminator::CRLF,
                 quote: self.quote.clone().unwrap(),
-                doublequote_escapes: true,
-                escape: Escape::Disabled,
-                comment: Comment::Disabled,
                 flexible: self.flexible.unwrap(),
             },
             num_fields: self.delimiter_freq.unwrap() + 1,

diff --git a/tests/delimiter.rs b/tests/delimiter.rs
@@ -3,7 +3,6 @@ extern crate csv_sniffer;
 
 use std::path::Path;
 
-use csv::Terminator;
 use csv_sniffer::metadata::*;
 use csv_sniffer::{SampleSize, Sniffer, Type};
 
@@ -26,11 +25,7 @@ fn test_semicolon() {
                     has_header_row: true,
                     num_preamble_rows: 0,
                 },
-                terminator: Terminator::CRLF,
                 quote: Quote::None,
-                doublequote_escapes: true,
-                escape: Escape::Disabled,
-                comment: Comment::Disabled,
                 flexible: false
             },
             num_fields: 5,
@@ -64,11 +59,7 @@ fn test_comma() {
                     has_header_row: true,
                     num_preamble_rows: 0,
                 },
-                terminator: Terminator::CRLF,
                 quote: Quote::None,
-                doublequote_escapes: true,
-                escape: Escape::Disabled,
-                comment: Comment::Disabled,
                 flexible: false
             },
             num_fields: 5,
@@ -102,11 +93,7 @@ fn test_flexible() {
                     has_header_row: true,
                     num_preamble_rows: 0,
                 },
-                terminator: Terminator::CRLF,
                 quote: Quote::None,
-                doublequote_escapes: true,
-                escape: Escape::Disabled,
-                comment: Comment::Disabled,
                 flexible: true
             },
             num_fields: 7,