refine did-url handling (percent encoding queries)

Signed-off-by: George Mulhearn <[email protected]>

aries/aries_vcx_ledger/Cargo.toml

@@ -9,7 +9,7 @@ edition.workspace = true
 # See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html
 [features]
 vdr_proxy_ledger = ["dep:indy-vdr-proxy-client"]
-cheqd = ["dep:did_cheqd", "dep:did_resolver"]
+cheqd = ["dep:did_cheqd", "dep:did_resolver", "dep:url"]
 
 [dependencies]
 aries_vcx_wallet = { path = "../aries_vcx_wallet" }
@@ -20,6 +20,7 @@ indy-vdr.workspace = true
 indy-vdr-proxy-client = { workspace = true, optional = true }
 did_cheqd = { path = "../../did_core/did_methods/did_cheqd", optional = true }
 did_resolver = { path = "../../did_core/did_resolver", optional = true }
+url = { version = "2.4.1", optional = true }
 serde_json = "1.0.95"
 public_key = { path = "../../did_core/public_key" }
 async-trait = "0.1.68"
@@ -40,4 +41,3 @@ tokio = { version = "1.38.0", default-features = false, features = [
 chrono = { version = "0.4", default-features = true }
 mockall = "0.13.1"
 uuid = { version = "1.4.1", default-features = false, features = ["v4"] }
-

aries/aries_vcx_ledger/src/ledger/cheqd/mod.rs

@@ -24,6 +24,7 @@ use models::{
     CheqdAnoncredsRevocationStatusList, CheqdAnoncredsSchema,
 };
 use serde::{Deserialize, Serialize};
+use url::Url;
 
 use super::base_ledger::{AnoncredsLedgerRead, AnoncredsLedgerSupport};
 use crate::errors::error::{VcxLedgerError, VcxLedgerResult};
@@ -185,12 +186,17 @@ impl AnoncredsLedgerRead for CheqdAnoncredsLedgerRead {
             DateTime::from_timestamp(timestamp as i64, 0).ok_or(VcxLedgerError::InvalidInput(
                 format!("input status list timestamp is not valid {timestamp}"),
             ))?;
+
+        // assemble query
         let xml_dt = resource_dt.to_rfc3339_opts(chrono::SecondsFormat::Millis, true);
-        let query = format!(
-            "{did}?resourceType={STATUS_LIST_RESOURCE_TYPE}&resourceName={name}&\
-             resourceVersionTime={xml_dt}"
-        );
-        let query_url = DidUrl::parse(query)?;
+        let mut query = Url::parse(did)
+            .map_err(|e| VcxLedgerError::InvalidInput(format!("cannot parse DID as URL: {e}")))?;
+        query
+            .query_pairs_mut()
+            .append_pair("resourceType", STATUS_LIST_RESOURCE_TYPE)
+            .append_pair("resourceName", name)
+            .append_pair("resourceVersionTime", &xml_dt);
+        let query_url = DidUrl::parse(query.to_string())?;
 
         let resource = self.resolver.resolve_resource(&query_url).await?;
         self.check_resource_type(&resource, STATUS_LIST_RESOURCE_TYPE)?;

did_core/did_parser_nom/Cargo.toml

@@ -7,6 +7,7 @@ edition = "2021"
 nom = "7.1.3"
 serde = "1.0.160"
 log = "0.4.16"
+percent-encoding = "2"
 
 [dev-dependencies]
 serde_test = "1.0.176"

did_core/did_parser_nom/src/did_url/mod.rs

@@ -59,8 +59,8 @@ impl DidUrl {
             .iter()
             .map(|(k, v)| {
                 (
-                    self.did_url[k.clone()].to_string(),
-                    self.did_url[v.clone()].to_string(),
+                    query_percent_decode(&self.did_url[k.clone()]),
+                    query_percent_decode(&self.did_url[v.clone()]),
                 )
             })
             .collect()
@@ -109,6 +109,14 @@ impl DidUrl {
     }
 }
 
+/// Decode percent-encoded URL query item (application/x-www-form-urlencoded encoded).
+/// Primary difference from general percent encoding is encoding of ' ' as '+'
+fn query_percent_decode(input: &str) -> String {
+    percent_encoding::percent_decode_str(&input.replace('+', " "))
+        .decode_utf8_lossy()
+        .into_owned()
+}
+
 impl TryFrom<String> for DidUrl {
     type Error = ParseError;
 

did_core/did_parser_nom/src/did_url/parsing.rs

@@ -3,11 +3,11 @@ use std::collections::HashMap;
 use nom::{
     branch::alt,
     bytes::complete::{tag, take_while1},
-    character::complete::{char, one_of},
+    character::complete::{char, one_of, satisfy},
     combinator::{all_consuming, cut, opt, recognize, success},
-    multi::{many0, separated_list0},
-    sequence::{preceded, separated_pair},
-    IResult,
+    multi::{many0, many1, separated_list0},
+    sequence::{preceded, separated_pair, tuple},
+    AsChar, IResult,
 };
 
 type UrlPart<'a> = (&'a str, Option<Vec<(&'a str, &'a str)>>, Option<&'a str>);
@@ -27,14 +27,29 @@ fn is_sub_delims(c: char) -> bool {
     "!$&'()*+,;=".contains(c)
 }
 
+// pct-encoded = "%" HEXDIG HEXDIG
+fn pct_encoded(input: &str) -> IResult<&str, &str> {
+    recognize(tuple((
+        tag("%"),
+        satisfy(|c| c.is_hex_digit()),
+        satisfy(|c| c.is_hex_digit()),
+    )))(input)
+}
+
 // pchar = unreserved / pct-encoded / sub-delims / ":" / "@"
-fn is_pchar(c: char) -> bool {
-    is_unreserved(c) || is_sub_delims(c) || ":@".contains(c)
+fn pchar(input: &str) -> IResult<&str, &str> {
+    alt((
+        recognize(satisfy(is_unreserved)),
+        pct_encoded,
+        recognize(satisfy(is_sub_delims)),
+        tag(":"),
+        tag("@"),
+    ))(input)
 }
 
 // segment = *pchar
 fn segment(input: &str) -> IResult<&str, &str> {
-    take_while1(is_pchar)(input)
+    recognize(many1(pchar))(input)
 }
 
 // path-abempty = *( "/" segment )
@@ -44,17 +59,17 @@ fn path_abempty(input: &str) -> IResult<&str, &str> {
 
 // fragment = *( pchar / "/" / "?" )
 pub(super) fn fragment_parser(input: &str) -> IResult<&str, &str> {
-    fn is_fragment_char(c: char) -> bool {
-        is_pchar(c) || "/?".contains(c)
+    fn fragment_element(input: &str) -> IResult<&str, &str> {
+        alt(((pchar), tag("/"), tag("?")))(input)
     }
 
-    take_while1(is_fragment_char)(input)
+    recognize(many1(fragment_element))(input)
 }
 
 // query = *( pchar / "/" / "?" )
 fn query_key_value_pair(input: &str) -> IResult<&str, (&str, &str)> {
-    fn is_query_char(c: char) -> bool {
-        is_pchar(c) || "/?".contains(c)
+    fn query_element(input: &str) -> IResult<&str, &str> {
+        alt(((pchar), tag("/"), tag("?")))(input)
     }
 
     let (remaining, (key, value)) = cut(separated_pair(
@@ -63,9 +78,9 @@ fn query_key_value_pair(input: &str) -> IResult<&str, (&str, &str)> {
         alt((take_while1(|c| !"&#?".contains(c)), success(""))),
     ))(input)?;
 
-    cut(all_consuming(take_while1(is_query_char)))(key)?;
+    cut(all_consuming(many1(query_element)))(key)?;
     if !value.is_empty() {
-        cut(all_consuming(take_while1(is_query_char)))(value)?;
+        cut(all_consuming(many1(query_element)))(value)?;
     }
 
     Ok((remaining, (key, value)))

did_core/did_parser_nom/tests/did_url/negative.rs

@@ -32,6 +32,10 @@ test_cases_negative! {
         "did:example:123456789abcdefghi&query1=value1"
     query_invalid_char:
         "did:example:123456789abcdefghi?query1=v^lue1"
+    query_unfinished_pct_encoding:
+        "did:example:123456789?query=a%3&query2=b"
+    query_invalid_space_char:
+        "did:example:123456789?query=a b"
     relative_empty_path: "/"
     relative_empty_path_and_query: "/?"
     relative_empty_path_and_fragment: "/#"

did_core/did_parser_nom/tests/did_url/positive.rs

@@ -384,4 +384,34 @@ test_cases_positive! {
                 ("resourceType".to_string(), "anonCredsCredDef".to_string()),
             ].into_iter().collect()
         }
+    test_case30:
+        "did:cheqd:testnet:36e695a3-f133-46ec-ac1e-79900a927f67?resourceType=anonCredsStatusList&resourceName=Example+schema-default-0&resourceVersionTime=2024-12-10T04%3A13%3A50.000Z",
+        Some("did:cheqd:testnet:36e695a3-f133-46ec-ac1e-79900a927f67"),
+        Some("cheqd"),
+        Some("testnet"),
+        Some("36e695a3-f133-46ec-ac1e-79900a927f67"),
+        None,
+        None,
+        {
+            vec![
+                ("resourceName".to_string(), "Example schema-default-0".to_string()),
+                ("resourceType".to_string(), "anonCredsStatusList".to_string()),
+                ("resourceVersionTime".to_string(), "2024-12-10T04:13:50.000Z".to_string()),
+            ].into_iter().collect()
+        }
+    test_case31:
+        "did:example:123?foo+bar=123&bar%20foo=123%20123&h3%21%210%20=w%40rld%3D%3D",
+        Some("did:example:123"),
+        Some("example"),
+        None,
+        Some("123"),
+        None,
+        None,
+        {
+            vec![
+                ("foo bar".to_string(), "123".to_string()),
+                ("bar foo".to_string(), "123 123".to_string()),
+                ("h3!!0 ".to_string(), "w@rld==".to_string())
+            ].into_iter().collect()
+        }
 }