chore: improve json fixing

yiwen-ai · Nov 27, 2023 · 25e396e · 25e396e
1 parent 2a06607
commit 25e396e
Show file tree

Hide file tree

Showing 3 changed files with 62 additions and 3 deletions.
diff --git a/Cargo.lock b/Cargo.lock
diff --git a/Cargo.toml b/Cargo.toml
@@ -1,6 +1,6 @@
 [package]
 name = "jarvis"
-version = "1.1.5"
+version = "1.1.6"
 edition = "2021"
 rust-version = "1.64"
 description = ""

diff --git a/src/json_util.rs b/src/json_util.rs
@@ -58,6 +58,18 @@ impl RawJSONArray {
         }
     }
 
+    fn skip_space_v(&self) -> usize {
+        let mut offset = self.offset;
+        while offset < self.chars.len() {
+            if self.chars[offset].is_whitespace() {
+                offset += 1;
+            } else {
+                break;
+            }
+        }
+        return offset;
+    }
+
     // return error message if failed
     fn array(&mut self) -> Option<String> {
         self.result.push('[');
@@ -110,9 +122,15 @@ impl RawJSONArray {
 
             match self.chars[self.offset] {
                 ',' => {
-                    self.result.push(',');
                     self.offset += 1;
                     self.skip_space();
+                    if self.offset < self.chars.len() && self.chars[self.offset] == ']' {
+                        self.result.push(']');
+                        self.offset += 1;
+                        return None;
+                    } else {
+                        self.result.push(',');
+                    }
                 }
                 ']' => {
                     self.result.push(']');
@@ -201,6 +219,20 @@ impl RawJSONArray {
                         }
                     }
                 }
+                ']' => {
+                    self.offset += 1;
+                    let offset = self.skip_space_v();
+                    if offset >= self.chars.len()
+                        || (offset <= self.chars.len() - 1
+                            && (self.chars[offset] == ',' || self.chars[offset] == ']'))
+                    {
+                        self.result.push('"');
+                        self.offset -= 1;
+                        return None;
+                    }
+
+                    self.result.push(']');
+                }
                 _ => {
                     self.result.push(self.chars[self.offset]);
                     self.offset += 1;
@@ -329,6 +361,33 @@ mod tests {
                 output: r#"[[],["] Stream: ["],["Internet Engineering Task Force \\(IETF)"]]"#.to_string(),
                 err: None,
             },
+            Case {
+                input: r#"[
+                    [],
+                    [
+                        ""] Stream: ["
+                    ],
+                    [
+                        "Internet Engineering Task Force \(IETF)"
+                    ],
+                ]"#
+                .to_string(),
+                output: r#"[[],["] Stream: ["],["Internet Engineering Task Force \\(IETF)"]]"#.to_string(),
+                err: None,
+            },
+            Case {
+                input: r#"[
+                    [],
+                    [
+                        ""] Stream: ["
+                    ],
+                    [
+                        "Internet Engineering Task Force \(IETF)1]
+                ]"#
+                .to_string(),
+                output: r#"[[],["] Stream: ["],["Internet Engineering Task Force \\(IETF)1"]]"#.to_string(),
+                err: None,
+            },
             Case {
                 input: r#"[["作为UTF-8 [","RFC3629",""]编码的文本字符串（","第2节", "）。字符串中的字节数等于参数。包含无效UTF-8序列的字符串是格式良好但无效的（","第1.2节", "）。此类型适用于需要解释或显示人类可读文本的系统，并允许区分结构化字节和具有指定曲目（Unicode）和编码（UTF-8）的文本。与JSON等格式不同，此类型中的Unicode字符永远不会被转义。因此，换行符（U+000A）始终表示为字符串中的字节0x0a，而不是字符0x5c6e（字符“\”和“n”）或0x5c7530303061（字符“\”，“u”，“0”，“0”，“0”和“a”）。","¶"]]"#
                 .to_string(),