diff --git a/src/indexing/mod.rs b/src/indexing/mod.rs
index ef7a6f108..e98ff1ce5 100644
--- a/src/indexing/mod.rs
+++ b/src/indexing/mod.rs
@@ -21,8 +21,9 @@ use self::merge_function::{
     docid_word_positions_merge, documents_merge,
 };
 
-mod store;
 mod merge_function;
+mod store;
+mod transform;
 
 #[derive(Debug, Clone, StructOpt)]
 pub struct IndexerOpt {
diff --git a/src/indexing/transform.rs b/src/indexing/transform.rs
new file mode 100644
index 000000000..40d34a1b7
--- /dev/null
+++ b/src/indexing/transform.rs
@@ -0,0 +1,154 @@
+use std::borrow::Cow;
+use std::convert::{TryFrom, TryInto};
+use std::fs::File;
+use std::io::{Read, Write, Seek, SeekFrom};
+
+use anyhow::{bail, Context};
+use crate::{FieldsIdsMap, AvailableDocumentsIds};
+use fst::{IntoStreamer, Streamer};
+use grenad::{Writer, Sorter, Reader, CompressionType};
+use roaring::RoaringBitmap;
+
+pub struct TransformOutput {
+    pub fields_ids_map: FieldsIdsMap,
+    pub users_ids_documents_ids: fst::Map<Vec<u8>>,
+    pub new_documents_ids: RoaringBitmap,
+    pub replaced_documents_ids: RoaringBitmap,
+    pub documents_count: usize,
+    pub documents_file: File,
+}
+
+pub struct Transform<A> {
+    pub fields_ids_map: FieldsIdsMap,
+    pub available_documents_ids: AvailableDocumentsIds,
+    pub users_ids_documents_ids: fst::Map<A>,
+    pub compression_type: CompressionType,
+    pub compression_level: u32,
+    pub enable_file_fuzing: bool,
+}
+
+impl<A: AsRef<[u8]>> Transform<A> {
+    /// Extract the users ids, deduplicate and compute the new internal documents ids
+    /// and fields ids, writing all the documents under their internal ids into a final file.
+    ///
+    /// Outputs the new `FieldsIdsMap`, the new `UsersIdsDocumentsIds` map, the new documents ids,
+    /// the replaced documents ids, the number of documents in this update and the file
+    /// containing all those documents.
+    pub fn from_csv<R: Read>(mut self, reader: R) -> anyhow::Result<TransformOutput> {
+        let mut csv = csv::Reader::from_reader(reader);
+        let headers = csv.headers()?.clone();
+        let user_id_pos = headers.iter().position(|h| h == "id").context(r#"missing "id" header"#)?;
+
+        // Generate the new fields ids based on the current fields ids and this CSV headers.
+        let mut fields_ids = Vec::new();
+        for header in headers.iter() {
+            let id = self.fields_ids_map.insert(header)
+                .context("impossible to generate a field id (limit reached)")?;
+            fields_ids.push(id);
+        }
+
+        /// The last value associated with an id is kept.
+        fn merge_last_win(_key: &[u8], vals: &[Cow<[u8]>]) -> anyhow::Result<Vec<u8>> {
+            Ok(vals.last().unwrap().clone().into_owned())
+        }
+
+        // We initialize the sorter with the user indexing settings.
+        let mut sorter_builder = Sorter::builder(merge_last_win);
+        sorter_builder.chunk_compression_type(self.compression_type);
+        sorter_builder.chunk_compression_level(self.compression_level);
+        if self.enable_file_fuzing {
+            sorter_builder.enable_fusing();
+        }
+
+        // We write into the sorter to merge and deduplicate the documents
+        // based on the users ids.
+        let mut sorter = sorter_builder.build();
+        let mut buffer = Vec::new();
+        let mut record = csv::StringRecord::new();
+        while csv.read_record(&mut record)? {
+
+            buffer.clear();
+            let mut writer = obkv::KvWriter::new(&mut buffer);
+
+            // We retrieve the field id based on the CSV header position
+            // and zip it with the record value.
+            for (key, field) in fields_ids.iter().copied().zip(&record) {
+                // TODO we must serialize the values as JSON strings.
+                writer.insert(key, field)?;
+            }
+
+            // We extract the user id and use it as the key for this document.
+            let user_id = &record[user_id_pos];
+            sorter.insert(user_id, &buffer)?;
+        }
+
+        // Once we have sort and deduplicated the documents we write them into a final file.
+        let file = tempfile::tempfile()?;
+        let mut writer_builder = Writer::builder();
+        writer_builder.compression_type(self.compression_type);
+        writer_builder.compression_level(self.compression_level);
+
+        let mut writer = writer_builder.build(file)?;
+        let mut new_users_ids_documents_ids_builder = fst::MapBuilder::memory();
+        let mut replaced_documents_ids = RoaringBitmap::new();
+        let mut new_documents_ids = RoaringBitmap::new();
+
+        // While we write into final file we get or generate the internal documents ids.
+        let mut documents_count = 0;
+        let mut iter = sorter.into_iter()?;
+        while let Some((user_id, obkv)) = iter.next()? {
+
+            let docid = match self.users_ids_documents_ids.get(user_id) {
+                Some(docid) => {
+                    // If we find the user id in the current users ids documents ids map
+                    // we use it and insert it in the list of replaced documents.
+                    let docid = u32::try_from(docid).expect("valid document id");
+                    replaced_documents_ids.insert(docid);
+                    docid
+                },
+                None => {
+                    // If this user id is new we add it to the users ids documents ids map
+                    // for new ids and into the list of new documents.
+                    let new_docid = self.available_documents_ids.next()
+                        .context("no more available documents ids")?;
+                    new_users_ids_documents_ids_builder.insert(user_id, new_docid as u64)?;
+                    new_documents_ids.insert(new_docid);
+                    new_docid
+                },
+            };
+
+            // We insert the document under the documents ids map into the final file.
+            writer.insert(docid.to_be_bytes(), obkv)?;
+            documents_count += 1;
+        }
+
+        // Once we have written all the documents into the final file, we extract it
+        // from the writer and reset the seek to be able to read it again.
+        let mut documents_file = writer.into_inner()?;
+        documents_file.seek(SeekFrom::Start(0))?;
+
+        // We create the union between the existing users ids documents ids with the new ones.
+        let new_users_ids_documents_ids = new_users_ids_documents_ids_builder.into_map();
+        let mut union_ = fst::map::OpBuilder::new()
+            .add(&self.users_ids_documents_ids)
+            .add(&new_users_ids_documents_ids)
+            .r#union();
+
+        // We stream and merge the new users ids documents ids map with the existing one.
+        let mut users_ids_documents_ids_builder = fst::MapBuilder::memory();
+        let mut iter = union_.into_stream();
+        while let Some((user_id, vals)) = iter.next() {
+            assert_eq!(vals.len(), 1, "there must be exactly one document id");
+            users_ids_documents_ids_builder.insert(user_id, vals[0].value)?;
+        }
+
+        Ok(TransformOutput {
+            fields_ids_map: self.fields_ids_map,
+            users_ids_documents_ids: users_ids_documents_ids_builder.into_map(),
+            new_documents_ids,
+            replaced_documents_ids,
+            documents_count,
+            documents_file,
+        })
+    }
+}
diff --git a/src/lib.rs b/src/lib.rs
index 366bf5428..8b4246ad7 100644
--- a/src/lib.rs
+++ b/src/lib.rs
@@ -16,7 +16,9 @@ use std::collections::HashMap;
 use std::hash::BuildHasherDefault;
 use fxhash::{FxHasher32, FxHasher64};
 
+pub use self::available_documents_ids::AvailableDocumentsIds;
 pub use self::criterion::{Criterion, default_criteria};
+pub use self::fields_ids_map::FieldsIdsMap;
 pub use self::index::Index;
 pub use self::search::{Search, SearchResult};
 pub use self::update_store::UpdateStore;