From 5c258209dc3f0c175d6338a3c658d7a8e01aedc0 Mon Sep 17 00:00:00 2001
From: Harald Eilertsen <haraldei@anduin.net>
Date: Wed, 8 May 2019 11:53:14 +0200
Subject: Help deserializer turn numbers and bools into their proper types.

The PhishTank dataset encodes the numeric `phish_id` as a string,
and boolean values as "yes" or "no". The default json deserializer
from serde isn't able to deserialize these values into their proper
types, so we have to helo it a bit.
---
 src/lib.rs | 47 +++++++++++++++++++++++++++++++++++++++++------
 1 file changed, 41 insertions(+), 6 deletions(-)

(limited to 'src')

diff --git a/src/lib.rs b/src/lib.rs
index 45add28..26be932 100644
--- a/src/lib.rs
+++ b/src/lib.rs
@@ -15,10 +15,12 @@
 // along with this program.  If not, see <https://www.gnu.org/licenses/>.
 
 use chrono::{ DateTime, Utc };
-use serde::Deserialize;
+use serde::{ Deserialize, Deserializer };
 use serde_json;
-use std::io::{ BufRead, Result };
+use std::io::{ self, BufRead };
 use std::net::IpAddr;
+use std::result::Result;
+use std::str::FromStr;
 
 #[derive(Deserialize)]
 pub struct PhishTank {
@@ -27,13 +29,18 @@ pub struct PhishTank {
 
 #[derive(Deserialize)]
 pub struct Phish {
-    pub phish_id: String,
+    #[serde(deserialize_with = "deserialize_number")]
+    pub phish_id: u64,
     pub url: String,
     pub phish_detail_url: String,
     pub submission_time: DateTime<Utc>,
-    pub verified: String,
+
+    #[serde(deserialize_with = "deserialize_yesno")]
+    pub verified: bool,
     pub verified_time: Option<DateTime<Utc>>,
-    pub online: String,
+
+    #[serde(deserialize_with = "deserialize_yesno")]
+    pub online: bool,
     pub target: String,
     pub details: Vec<PhishDetails>,
 }
@@ -48,7 +55,35 @@ pub struct PhishDetails {
     pub detail_time: DateTime<Utc>,
 }
 
-pub fn load_phistank<R: BufRead>(input: R) -> Result<PhishTank> {
+pub fn load_phistank<R: BufRead>(input: R) -> Result<PhishTank, io::Error> {
     let phishes = serde_json::from_reader(input)?;
     Ok(PhishTank { phishes })
 }
+
+// Helper function to deserialize a number represented as a string.
+//
+// The PhishTank dataset represents numbers this way, so the default
+// json deserializer implementation will only deserialize to a string.
+//
+fn deserialize_number<'de, D>(d: D) -> Result<u64, D::Error>
+    where D: Deserializer<'de>
+{
+    let s = String::deserialize(d)?;
+    u64::from_str(&s).map_err(serde::de::Error::custom)
+}
+
+// Helper function to deserialize "yes/no" into a bool.
+//
+// The PhishTank dataset uses the strings "yes" and "no" to represent
+// boolean values. This function helps us deserialize back into a proper
+// bool.
+//
+fn deserialize_yesno<'de, D>(d: D) -> Result<bool, D::Error>
+    where D: Deserializer<'de>
+{
+    match String::deserialize(d)?.as_ref() {
+        "yes" => Ok(true),
+        "no" => Ok(false),
+        _ => Err(serde::de::Error::custom("Expected \"yes\" or \"no\"."))
+    }
+}
-- 
cgit v1.2.3