From 9c386d7b97464204b5ec6067e9cab5c924ff8dbb Mon Sep 17 00:00:00 2001
From: Manuel Holtgrewe <manuel.holtgrewe@bih-charite.de>
Date: Tue, 8 Oct 2024 12:29:54 +0200
Subject: [PATCH] fix: percent encoding clinvar annotation (#553)

---
 Cargo.lock                  |  1 +
 Cargo.toml                  | 11 ++++++-----
 src/annotate/seqvars/mod.rs | 29 ++++++++++++++++++++++++++++-
 3 files changed, 35 insertions(+), 6 deletions(-)

diff --git a/Cargo.lock b/Cargo.lock
index 31909ed8..9a9d9259 100644
--- a/Cargo.lock
+++ b/Cargo.lock
@@ -2653,6 +2653,7 @@ dependencies = [
  "pbjson",
  "pbjson-build",
  "pbjson-types",
+ "percent-encoding",
  "pprof",
  "pretty_assertions",
  "procfs",
diff --git a/Cargo.toml b/Cargo.toml
index 58a8a388..0b06499f 100644
--- a/Cargo.toml
+++ b/Cargo.toml
@@ -38,10 +38,12 @@ byte-unit = "5.1"
 chrono = "0.4"
 clap-verbosity-flag = "2.2"
 clap = { version = "4.5", features = ["derive"] }
+coz = "0.1.3"
 csv = "1.3"
 derivative = "2.2"
 derive_builder = { version = "0.20", features = ["clippy"] }
 derive-new = "0.7.0"
+dhat = "0.3.3"
 enumflags2 = { version = "0.7.10", features = ["serde"] }
 env_logger = "0.11"
 flate2 = "1.0"
@@ -54,7 +56,11 @@ jsonl = "4.0"
 log = "0.4"
 nom = "7.1"
 nutype = { version = "0.5.0", features = ["serde"] }
+once_cell = "1.20.1"
 parse-display = "0.10"
+pbjson = "0.7"
+pbjson-types = "0.7"
+percent-encoding = "2.3"
 procfs = "0.16"
 prost = "0.13.3"
 quick_cache = "0.6.9"
@@ -76,11 +82,6 @@ tracing-subscriber = "0.3"
 tracing = { version = "0.1", features = ["log"] }
 uuid = { version = "1.9", features = ["fast-rng", "serde"] }
 zstd = "0.13"
-pbjson = "0.7"
-pbjson-types = "0.7"
-coz = "0.1.3"
-dhat = "0.3.3"
-once_cell = "1.20.1"
 
 [dependencies.noodles]
 version = "0.77.0"
diff --git a/src/annotate/seqvars/mod.rs b/src/annotate/seqvars/mod.rs
index 24aed22a..99bc9730 100644
--- a/src/annotate/seqvars/mod.rs
+++ b/src/annotate/seqvars/mod.rs
@@ -1615,6 +1615,28 @@ pub struct ClinvarAnnotator {
     db: DBWithThreadMode<rocksdb::MultiThreaded>,
 }
 
+/// Helper code for percent encoding of strings.
+///
+/// cf. https://github.com/varfish-org/varfish-server-worker/issues/485
+mod vcf_encoding {
+    use percent_encoding::{utf8_percent_encode, AsciiSet, PercentEncode, CONTROLS};
+
+    // § 1.2 "Character encoding, non-printable characters and characters with special meaning" (2023-08-23)
+    const PERCENT_ENCODE_SET: &AsciiSet = &CONTROLS
+        .add(b':')
+        .add(b';')
+        .add(b'=')
+        .add(b'%')
+        .add(b',')
+        .add(b'\r')
+        .add(b'\n')
+        .add(b'\t');
+
+    pub(super) fn percent_encode(s: &str) -> PercentEncode<'_> {
+        utf8_percent_encode(s, PERCENT_ENCODE_SET)
+    }
+}
+
 impl ClinvarAnnotator {
     pub fn new(db: DBWithThreadMode<rocksdb::MultiThreaded>) -> Self {
         Self { db }
@@ -1674,7 +1696,12 @@ impl ClinvarAnnotator {
                 Some(field::Value::Array(field::value::Array::String(
                     clinvar_germline_classifications
                         .into_iter()
-                        .map(Some)
+                        .map(|value| {
+                            // Manually encode until the following is fixed.
+                            //
+                            // https://github.com/varfish-org/varfish-server-worker/issues/485
+                            Some(vcf_encoding::percent_encode(&value).to_string())
+                        })
                         .collect::<Vec<_>>(),
                 ))),
             );