From 441a47d384f837d7cbeb7077314ae634e8ea5b7a Mon Sep 17 00:00:00 2001
From: Till Wegmueller <toasterson@gmail.com>
Date: Mon, 18 May 2020 14:57:55 +0200
Subject: [PATCH] Add default derive Add Fast path parsing which catches the
 case of unescaped values with spaces.

---
 src/actions/mod.rs | 48 +++++++++++++++++++++++++++++++++++++++++++++-
 src/lib.rs         | 34 ++++++++++++++++++++++++++++----
 2 files changed, 77 insertions(+), 5 deletions(-)

diff --git a/src/actions/mod.rs b/src/actions/mod.rs
index 16cd064..5f6d608 100644
--- a/src/actions/mod.rs
+++ b/src/actions/mod.rs
@@ -3,7 +3,7 @@
 //  MPL was not distributed with this file, You can
 //  obtain one at https://mozilla.org/MPL/2.0/.
 
-use regex::Regex;
+use regex::{Regex, Captures};
 use std::collections::HashSet;
 use std::error;
 use std::fmt;
@@ -11,6 +11,7 @@ use std::fs::File;
 use std::io::BufRead;
 use std::io::BufReader;
 
+#[derive(Debug, Default)]
 pub struct Dir {
     pub path: String,
     pub group: String,
@@ -18,6 +19,7 @@ pub struct Dir {
     pub mode: String, //TODO implement as bitmask
 }
 
+#[derive(Debug, Default)]
 pub struct Attr {
     pub key: String,
     pub values: Vec<String>,
@@ -30,6 +32,7 @@ pub struct Property {
     pub value: String,
 }
 
+#[derive(Debug, Default)]
 pub struct Manifest {
     pub attributes: Vec<Attr>,
 }
@@ -130,6 +133,49 @@ fn is_attr_action(line: &String) -> bool {
 }
 
 pub fn parse_attr_action(line: String) -> Result<Attr, ManifestError> {
+    // Do a full line match to see if we can fast path this.
+    // This also catches values with spaces, that have not been properly escaped.
+    // Note: values with spaces must be properly escaped or the rest here will fail. Strings with
+    //  unescaped spaces are never valid but sadly present in the wild.
+    // Fast path will fail if a value has multiple values or a '=' sign in the values
+    let full_line_regex = match Regex::new(r"^set name=([^ ]+) value=(.+)$") {
+        Ok(re) => re,
+        Err(e) => return Err(ManifestError::Regex(e)),
+    };
+
+    if full_line_regex.is_match(line.trim_start()) {
+        match full_line_regex.captures(line.trim_start()) {
+            Some(captures) => {
+                let mut fast_path_fail = false;
+                let mut val = String::from(&captures[2]);
+
+                if val.contains("=") {
+                    fast_path_fail = true;
+                }
+
+                if val.contains("value=") {
+                    fast_path_fail = true;
+                }
+
+                if val.contains("name=") {
+                    fast_path_fail = true;
+                }
+
+                val = val.replace(&['"', '\\'][..], "");
+
+                if !fast_path_fail{
+                    return Ok(Attr{
+                        key: String::from(&captures[1]),
+                        values: vec![val],
+                        ..Attr::default()
+                    });
+                }
+            }
+            None => (),
+        };
+    }
+
+
     //Todo move regex initialisation out of for loop into static area
     let name_regex = match Regex::new(r"name=([^ ]+) value=") {
         Ok(re) => re,
diff --git a/src/lib.rs b/src/lib.rs
index 345c6aa..aa15ec9 100644
--- a/src/lib.rs
+++ b/src/lib.rs
@@ -8,7 +8,7 @@ mod actions;
 #[cfg(test)]
 mod tests {
 
-    use crate::actions::Manifest;
+    use crate::actions::{Manifest, Property};
     use crate::actions::{parse_manifest_string, Attr};
     use std::collections::HashSet;
 
@@ -25,7 +25,15 @@ mod tests {
         set name=info.source-url value=http://nginx.org/download/nginx-1.18.0.tar.gz
         set name=org.opensolaris.consolidation value=userland
         set name=com.oracle.info.version value=1.18.0
-        set name=variant.arch value=i386");
+        set name=pkg.summary value=\\\"provided mouse accessibility enhancements\\\"
+        set name=info.upstream value=X.Org Foundation
+        set name=pkg.description value=Latvian language support's extra files
+        set name=variant.arch value=i386 optional=testing optionalWithString=\"test ing\"");
+
+        let mut optional_hash = HashSet::new();
+        optional_hash.insert(Property{key: String::from("optional"), value:String::from("testing")});
+        optional_hash.insert(Property{key: String::from("optionalWithString"), value:String::from("test ing")});
+
         let test_results = vec![
             Attr{
                 key: String::from("pkg.fmri"),
@@ -82,10 +90,25 @@ mod tests {
                 values: vec![String::from("1.18.0")],
                 properties: HashSet::new(),
             },
+            Attr{
+                key: String::from("pkg.summary"),
+                values: vec![String::from("provided mouse accessibility enhancements")],
+                properties: HashSet::new(),
+            },
+            Attr{
+                key: String::from("info.upstream"),
+                values: vec![String::from("X.Org Foundation")],
+                properties: HashSet::new(),
+            },
+            Attr{
+                key: String::from("pkg.description"),
+                values: vec![String::from("Latvian language support's extra files")],
+                properties: HashSet::new(),
+            },
             Attr{
                 key: String::from("variant.arch"),
                 values: vec![String::from("i386")],
-                properties: HashSet::new(),
+                properties: optional_hash,
             }
         ];
 
@@ -94,9 +117,12 @@ mod tests {
             Ok(m) => manifest = m,
             Err(_) => assert!(false, "caught error"),
         };
-        assert_eq!(manifest.attributes.len(), 12);
+
+        assert_eq!(manifest.attributes.len(), 15);
+
         for (pos, attr) in manifest.attributes.iter().enumerate() {
             assert_eq!(attr.key, test_results[pos].key);
+
             for (vpos, val) in attr.values.iter().enumerate() {
                 assert_eq!(val, &test_results[pos].values[vpos]);
             }