From 7ba3cfccf27c5ef3c9ee54e0cf252083afc48989 Mon Sep 17 00:00:00 2001
From: Raphael Paul Laude <raphlaude@gmail.com>
Date: Fri, 3 Nov 2023 09:48:20 -0400
Subject: [PATCH] group by tag in csv export

---
 README.md   | 10 ++++++++--
 src/lib.rs  | 18 ++++++++++++++++++
 src/main.rs |  5 +++--
 3 files changed, 29 insertions(+), 4 deletions(-)

diff --git a/README.md b/README.md
index 795bc07..8f6ffed 100644
--- a/README.md
+++ b/README.md
@@ -4,14 +4,14 @@ US Addrs is a rust crate for parsing unstructured United States address strings
 It is a rust implementation of the awesome [usaddress](https://github.com/datamade/usaddress/tree/master) library.
 Thank you to the folks at [datamade](https://datamade.us/) for releasing such a cool tool.
 
-US Addrs is currently _79% (~5x) faster_ than usaddress, though additional optimizations should be possible. Accuracy stats TK.
+US Addrs is currently _79% (~5x) faster_ than usaddress, though additional optimizations should be possible. [Accuracy](#accuracy) is close to usaddress but not quite matching yet.
 The goal of this implementation is to faciliate use cases requiring better performance, such as geocoding large batches of addresses.
 
 :warning: This crate is under **active development** and may not match usaddress's accuracy. US Addrs will be better tested / documented shortly.
 
 ## Examples
 
-US Addrs can be run from the command line
+US Addrs can be run from the command line to parse an address
 
 ```bash
 cargo run -- parse --address '33 Nassau Avenue, Brooklyn, NY'
@@ -20,6 +20,12 @@ cargo run -- parse --address '33 Nassau Avenue, Brooklyn, NY'
 [("33", "AddressNumber"), ("Nassau", "StreetName"), ("Avenue", "StreetNamePostType"), ("Brooklyn", "PlaceName"), ("NY", "StateName")]
 ```
 
+or export a list of addresses to CSV
+
+```bash
+cargo run -- parse-file --file-path tests/test_data/test_addrs.txt test.csv
+```
+
 or by importing the crate and using the `parse` function
 
 ```rust
diff --git a/src/lib.rs b/src/lib.rs
index ed48626..325d569 100644
--- a/src/lib.rs
+++ b/src/lib.rs
@@ -324,3 +324,21 @@ pub fn read_xml_tagged_addresses(file_path: &str) -> (Vec<String>, Vec<Vec<Strin
 
     (addresses, tags)
 }
+
+pub fn group_by_tag(tokens: Vec<(String, String)>) -> Vec<(String, String)> {
+    let mut result = Vec::new();
+    let mut tokens = tokens.into_iter().peekable();
+
+    while let Some((mut token, tag)) = tokens.next() {
+        while tokens
+            .peek()
+            .map_or(false, |(_, ref next_tag)| &tag == next_tag)
+        {
+            let (next_token, _) = tokens.next().unwrap();
+            token = format!("{} {}", token, next_token);
+        }
+        result.push((token, tag));
+    }
+
+    result
+}
diff --git a/src/main.rs b/src/main.rs
index e49f361..5a63790 100644
--- a/src/main.rs
+++ b/src/main.rs
@@ -1,6 +1,6 @@
 use clap::Parser;
 use us_addrs::train::train_model;
-use us_addrs::{parse, parse_addresses_from_txt, TAGS};
+use us_addrs::{group_by_tag, parse, parse_addresses_from_txt, TAGS};
 
 // use std::path::PathBuf;
 
@@ -44,10 +44,11 @@ fn main() {
             wtr.write_record(TAGS.iter()).unwrap();
 
             for tagged_address in parsed_addresses {
+                let group_tagged_address = group_by_tag(tagged_address);
                 let mut record = Vec::new();
 
                 for tag in TAGS.iter() {
-                    if let Some((token, _)) = tagged_address
+                    if let Some((token, _)) = group_tagged_address
                         .iter()
                         .find(|&(_, token_tag)| *token_tag == *tag)
                     {