From 69421fee852248446dc4c3103bfb416753896d04 Mon Sep 17 00:00:00 2001
From: Thomas Krause <thomas.krause@hu-berlin.de>
Date: Mon, 7 Oct 2024 15:25:39 +0200
Subject: [PATCH] Add empty table importer

---
 src/importer/mod.rs   |  1 +
 src/importer/table.rs | 60 +++++++++++++++++++++++++++++++++++++++++++
 src/lib.rs            | 12 ++++++---
 3 files changed, 70 insertions(+), 3 deletions(-)
 create mode 100644 src/importer/table.rs

diff --git a/src/importer/mod.rs b/src/importer/mod.rs
index 2b70060b..35efb61e 100644
--- a/src/importer/mod.rs
+++ b/src/importer/mod.rs
@@ -9,6 +9,7 @@ pub mod opus;
 pub mod ptb;
 pub mod relannis;
 pub mod saltxml;
+pub mod table;
 pub mod textgrid;
 pub mod toolbox;
 pub mod treetagger;
diff --git a/src/importer/table.rs b/src/importer/table.rs
new file mode 100644
index 00000000..284ac2c9
--- /dev/null
+++ b/src/importer/table.rs
@@ -0,0 +1,60 @@
+use documented::{Documented, DocumentedFields};
+use graphannis::update::GraphUpdate;
+use serde::Deserialize;
+use struct_field_names_as_array::FieldNamesAsSlice;
+
+use super::Importer;
+const FILE_EXTENSIONS: [&str; 3] = ["csv", "tsv", "conll"];
+
+/// Imports table-like CSV files.
+#[derive(Deserialize, Documented, DocumentedFields, FieldNamesAsSlice)]
+#[serde(default, deny_unknown_fields)]
+pub struct ImportTable {
+    /// The provided character defines the column delimiter. The default value is tab.
+    ///
+    /// Example:
+    /// ```toml
+    /// [export.config]
+    /// delimiter = ";"
+    /// ```
+    #[serde(default = "default_delimiter")]
+    delimiter: char,
+    /// The provided character will be used for quoting values. If nothing is provided, all columns will contain bare values. If a character is provided,
+    /// all values will be quoted.
+    ///
+    /// Example:
+    /// ```toml
+    /// [export.config]
+    /// quote_char = "\""
+    /// ```
+    #[serde(default)]
+    quote_char: Option<char>,
+}
+
+impl Default for ImportTable {
+    fn default() -> Self {
+        Self {
+            delimiter: default_delimiter(),
+            quote_char: None,
+        }
+    }
+}
+
+fn default_delimiter() -> char {
+    '\t'
+}
+
+impl Importer for ImportTable {
+    fn import_corpus(
+        &self,
+        input_path: &std::path::Path,
+        step_id: crate::StepID,
+        tx: Option<crate::workflow::StatusSender>,
+    ) -> Result<GraphUpdate, Box<dyn std::error::Error>> {
+        todo!()
+    }
+
+    fn file_extensions(&self) -> &[&str] {
+        todo!()
+    }
+}
diff --git a/src/lib.rs b/src/lib.rs
index bb1a1e47..2ec27830 100644
--- a/src/lib.rs
+++ b/src/lib.rs
@@ -28,9 +28,9 @@ use graphannis::AnnotationGraph;
 use importer::{
     conllu::ImportCoNLLU, exmaralda::ImportEXMARaLDA, file_nodes::CreateFileNodes,
     graphml::GraphMLImporter, meta::AnnotateCorpus, none::CreateEmptyCorpus, opus::ImportOpusLinks,
-    ptb::ImportPTB, relannis::ImportRelAnnis, saltxml::ImportSaltXml, textgrid::ImportTextgrid,
-    toolbox::ImportToolBox, treetagger::ImportTreeTagger, xlsx::ImportSpreadsheet, xml::ImportXML,
-    Importer,
+    ptb::ImportPTB, relannis::ImportRelAnnis, saltxml::ImportSaltXml, table::ImportTable,
+    textgrid::ImportTextgrid, toolbox::ImportToolBox, treetagger::ImportTreeTagger,
+    xlsx::ImportSpreadsheet, xml::ImportXML, Importer,
 };
 use manipulator::{
     check::Check, chunker::Chunk, collapse::Collapse, enumerate::EnumerateMatches,
@@ -171,6 +171,7 @@ pub enum ReadFrom {
     PTB(#[serde(default)] ImportPTB),
     RelAnnis(#[serde(default)] ImportRelAnnis),
     SaltXml(#[serde(default)] ImportSaltXml),
+    Table(#[serde(default)] ImportTable),
     TextGrid(#[serde(default)] ImportTextgrid),
     Toolbox(#[serde(default)] ImportToolBox),
     TreeTagger(#[serde(default)] ImportTreeTagger),
@@ -198,6 +199,7 @@ impl ReadFrom {
             ReadFrom::PTB(m) => m,
             ReadFrom::RelAnnis(m) => m,
             ReadFrom::SaltXml(m) => m,
+            ReadFrom::Table(m) => m,
             ReadFrom::TextGrid(m) => m,
             ReadFrom::Toolbox(m) => m,
             ReadFrom::TreeTagger(m) => m,
@@ -220,6 +222,7 @@ impl ReadFromDiscriminants {
             ReadFromDiscriminants::PTB => ImportPTB::DOCS,
             ReadFromDiscriminants::RelAnnis => ImportRelAnnis::DOCS,
             ReadFromDiscriminants::SaltXml => ImportSaltXml::DOCS,
+            ReadFromDiscriminants::Table => ImportTable::DOCS,
             ReadFromDiscriminants::TextGrid => ImportTextgrid::DOCS,
             ReadFromDiscriminants::Toolbox => ImportToolBox::DOCS,
             ReadFromDiscriminants::TreeTagger => ImportTreeTagger::DOCS,
@@ -263,6 +266,9 @@ impl ReadFromDiscriminants {
                 ImportTextgrid::FIELD_NAMES_AS_SLICE,
                 ImportTextgrid::FIELD_DOCS,
             ),
+            ReadFromDiscriminants::Table => {
+                (ImportTable::FIELD_NAMES_AS_SLICE, ImportTable::FIELD_DOCS)
+            }
             ReadFromDiscriminants::TreeTagger => (
                 ImportTreeTagger::FIELD_NAMES_AS_SLICE,
                 ImportTreeTagger::FIELD_DOCS,