Merge 'Implement json_error_position' from Peter Sooley

Furthering work for JSON compatibility as described in #127, this change adds support for the `json_error_position` function used to find the error in invalid JSON values. ![image](https://github.com/user-attachments/assets/d61403ef- bccf-41b0-9128-199066608fac) Closes #564
tursodatabase · Jan 13, 2025 · 9d42a48 · 9d42a48
2 parents 0dceb02 + 1e94dbf
commit 9d42a48
Show file tree

Hide file tree

Showing 6 changed files with 155 additions and 1 deletion.
diff --git a/COMPAT.md b/COMPAT.md
@@ -263,7 +263,7 @@ Feature support of [sqlite expr syntax](https://www.sqlite.org/lang_expr.html).
 | jsonb_array(value1,value2,...)     |         |                                                                                                                                              |
 | json_array_length(json)            | Yes     |                                                                                                                                              |
 | json_array_length(json,path)       | Yes     |                                                                                                                                              |
-| json_error_position(json)          |         |                                                                                                                                              |
+| json_error_position(json)          | Yes     |                                                                                                                                              |
 | json_extract(json,path,...)        | Partial | Does not fully support unicode literal syntax and does not allow numbers > 2^127 - 1 (which SQLite truncates to i32), does not support BLOBs |
 | jsonb_extract(json,path,...)       |         |                                                                                                                                              |
 | json -> path                       | Yes     |                                                                                                                                              |

diff --git a/core/function.rs b/core/function.rs
@@ -30,6 +30,7 @@ pub enum JsonFunc {
     JsonArrowShiftExtract,
     JsonExtract,
     JsonType,
+    JsonErrorPosition,
 }
 
 #[cfg(feature = "json")]
@@ -46,6 +47,7 @@ impl Display for JsonFunc {
                 Self::JsonArrowExtract => "->".to_string(),
                 Self::JsonArrowShiftExtract => "->>".to_string(),
                 Self::JsonType => "json_type".to_string(),
+                Self::JsonErrorPosition => "json_error_position".to_string(),
             }
         )
     }
@@ -379,6 +381,8 @@ impl Func {
             "json_extract" => Ok(Func::Json(JsonFunc::JsonExtract)),
             #[cfg(feature = "json")]
             "json_type" => Ok(Func::Json(JsonFunc::JsonType)),
+            #[cfg(feature = "json")]
+            "json_error_position" => Ok(Self::Json(JsonFunc::JsonErrorPosition)),
             "unixepoch" => Ok(Self::Scalar(ScalarFunc::UnixEpoch)),
             "julianday" => Ok(Self::Scalar(ScalarFunc::JulianDay)),
             "hex" => Ok(Self::Scalar(ScalarFunc::Hex)),

diff --git a/core/json/mod.rs b/core/json/mod.rs
@@ -6,10 +6,12 @@ mod ser;
 use std::rc::Rc;
 
 pub use crate::json::de::from_str;
+use crate::json::error::Error as JsonError;
 use crate::json::json_path::{json_path, JsonPath, PathElement};
 pub use crate::json::ser::to_string;
 use crate::types::{LimboText, OwnedValue, TextSubtype};
 use indexmap::IndexMap;
+use jsonb::Error as JsonbError;
 use serde::{Deserialize, Serialize};
 
 #[derive(Serialize, Deserialize, Debug, PartialEq, Clone)]
@@ -370,6 +372,32 @@ fn json_extract_single<'a>(
     Ok(Some(&current_element))
 }
 
+pub fn json_error_position(json: &OwnedValue) -> crate::Result<OwnedValue> {
+    match json {
+        OwnedValue::Text(t) => match crate::json::from_str::<Val>(&t.value) {
+            Ok(_) => Ok(OwnedValue::Integer(0)),
+            Err(JsonError::Message { location, .. }) => {
+                if let Some(loc) = location {
+                    Ok(OwnedValue::Integer(loc.column as i64))
+                } else {
+                    Err(crate::error::LimboError::InternalError(
+                        "failed to determine json error position".into(),
+                    ))
+                }
+            }
+        },
+        OwnedValue::Blob(b) => match jsonb::from_slice(b) {
+            Ok(_) => Ok(OwnedValue::Integer(0)),
+            Err(JsonbError::Syntax(_, pos)) => Ok(OwnedValue::Integer(pos as i64)),
+            _ => Err(crate::error::LimboError::InternalError(
+                "failed to determine json error position".into(),
+            )),
+        },
+        OwnedValue::Null => Ok(OwnedValue::Null),
+        _ => Ok(OwnedValue::Integer(0)),
+    }
+}
+
 #[cfg(test)]
 mod tests {
     use super::*;
@@ -698,4 +726,60 @@ mod tests {
             Err(e) => assert!(e.to_string().contains("JSON path error")),
         }
     }
+
+    #[test]
+    fn test_json_error_position_no_error() {
+        let input = OwnedValue::build_text(Rc::new("[1,2,3]".to_string()));
+        let result = json_error_position(&input).unwrap();
+        assert_eq!(result, OwnedValue::Integer(0));
+    }
+
+    #[test]
+    fn test_json_error_position_no_error_more() {
+        let input = OwnedValue::build_text(Rc::new(r#"{"a":55,"b":72 , }"#.to_string()));
+        let result = json_error_position(&input).unwrap();
+        assert_eq!(result, OwnedValue::Integer(0));
+    }
+
+    #[test]
+    fn test_json_error_position_object() {
+        let input = OwnedValue::build_text(Rc::new(r#"{"a":55,"b":72,,}"#.to_string()));
+        let result = json_error_position(&input).unwrap();
+        assert_eq!(result, OwnedValue::Integer(16));
+    }
+
+    #[test]
+    fn test_json_error_position_array() {
+        let input = OwnedValue::build_text(Rc::new(r#"["a",55,"b",72,,]"#.to_string()));
+        let result = json_error_position(&input).unwrap();
+        assert_eq!(result, OwnedValue::Integer(16));
+    }
+
+    #[test]
+    fn test_json_error_position_null() {
+        let input = OwnedValue::Null;
+        let result = json_error_position(&input).unwrap();
+        assert_eq!(result, OwnedValue::Null);
+    }
+
+    #[test]
+    fn test_json_error_position_integer() {
+        let input = OwnedValue::Integer(5);
+        let result = json_error_position(&input).unwrap();
+        assert_eq!(result, OwnedValue::Integer(0));
+    }
+
+    #[test]
+    fn test_json_error_position_float() {
+        let input = OwnedValue::Float(-5.5);
+        let result = json_error_position(&input).unwrap();
+        assert_eq!(result, OwnedValue::Integer(0));
+    }
+
+    #[test]
+    fn test_json_error_position_blob() {
+        let input = OwnedValue::Blob(Rc::new(r#"["a",55,"b",72,,]"#.as_bytes().to_owned()));
+        let result = json_error_position(&input).unwrap();
+        assert_eq!(result, OwnedValue::Integer(16));
+    }
 }
diff --git a/core/translate/expr.rs b/core/translate/expr.rs
@@ -812,6 +812,31 @@ pub fn translate_expr(
                             func_ctx,
                         )
                     }
+                    JsonFunc::JsonErrorPosition => {
+                        let args = if let Some(args) = args {
+                            if args.len() != 1 {
+                                crate::bail_parse_error!(
+                                    "{} function with not exactly 1 argument",
+                                    j.to_string()
+                                );
+                            }
+                            args
+                        } else {
+                            crate::bail_parse_error!(
+                                "{} function with no arguments",
+                                j.to_string()
+                            );
+                        };
+                        let json_reg = program.alloc_register();
+                        translate_expr(program, referenced_tables, &args[0], json_reg, resolver)?;
+                        program.emit_insn(Insn::Function {
+                            constant_mask: 0,
+                            start_reg: json_reg,
+                            dest: target_register,
+                            func: func_ctx,
+                        });
+                        Ok(target_register)
+                    }
                 },
                 Func::Scalar(srf) => {
                     match srf {

diff --git a/core/vdbe/mod.rs b/core/vdbe/mod.rs
@@ -42,6 +42,7 @@ use crate::vdbe::insn::Insn;
 use crate::{
     function::JsonFunc, json::get_json, json::json_array, json::json_array_length,
     json::json_arrow_extract, json::json_arrow_shift_extract, json::json_extract, json::json_type,
+    json::json_error_position,
 };
 use crate::{Connection, Result, Rows, TransactionState, DATABASE_VERSION};
 use datetime::{exec_date, exec_datetime_full, exec_julianday, exec_time, exec_unixepoch};
@@ -1421,6 +1422,14 @@ impl Program {
                                 Err(e) => return Err(e),
                             }
                         }
+                        #[cfg(feature = "json")]
+                        crate::function::Func::Json(JsonFunc::JsonErrorPosition) => {
+                            let json_value = &state.registers[*start_reg];
+                            match json_error_position(json_value) {
+                                Ok(pos) => state.registers[*dest] = pos,
+                                Err(e) => return Err(e),
+                            }
+                        }
                         crate::function::Func::Scalar(scalar_func) => match scalar_func {
                             ScalarFunc::Cast => {
                                 assert!(arg_count == 2);

diff --git a/testing/json.test b/testing/json.test
@@ -474,3 +474,35 @@ do_execsql_test json_type_cast {
 do_execsql_test json_type_null_arg {
   select json_type(null)
 } {{}}
+
+do_execsql_test json_error_position_valid {
+  SELECT json_error_position('{"a":55,"b":72,}');
+} {{0}}
+
+do_execsql_test json_error_position_valid_ws {
+  SELECT json_error_position('{"a":55,"b":72 , }');
+} {{0}}
+
+do_execsql_test json_error_position_object {
+  SELECT json_error_position('{"a":55,"b":72,,}');
+} {{16}}
+
+do_execsql_test json_error_position_array_valid {
+  SELECT json_error_position('["a",55,"b",72,]');
+} {{0}}
+
+do_execsql_test json_error_position_array_valid_ws {
+  SELECT json_error_position('["a",55,"b",72 , ]');
+} {{0}}
+
+do_execsql_test json_error_position_array {
+  SELECT json_error_position('["a",55,"b",72,,]');
+} {{16}}
+
+do_execsql_test json_error_position_null {
+  SELECT json_error_position(NULL);
+} {{}}
+
+do_execsql_test json_error_position_complex {
+  SELECT json_error_position('{a:null,{"h":[1,[1,2,3]],"j":"abc"}:true}');
+} {{9}}