Implement i8 and detect types correctly in MySQL (#454)

* Update dependencies

* Add support for i8 subtype

* Add and fix MySQL tests

* Revert local testing, add note

* Cleanup

* Handle mediumtext in MySQL gracefully

* blobs

* Use correct dependencies from upstream

* Undo temporary local testing change

* Lint

core/src/graph/number.rs

@@ -51,6 +51,7 @@ macro_rules! any_range_int_impl {
     }
 }
 
+any_range_int_impl! { u8 }
 any_range_int_impl! { u16 }
 any_range_int_impl! { u32 }
 any_range_int_impl! { u64 }
@@ -219,6 +220,7 @@ macro_rules! standard_int_range_step_impl {
     }
 }
 
+standard_int_range_step_impl! { i8, i16, u8}
 standard_int_range_step_impl! { i16, i32, u16}
 standard_int_range_step_impl! { i32, i64, u32 }
 standard_int_range_step_impl! { u32, u32, u32 }
@@ -480,6 +482,12 @@ number_node!(
         I16Categorical as categorical,
         Incrementing as incrementing,
     ) for i16,
+    RandomI8 (
+        I8Range<StandardIntRangeStep<u8, i16>> as range,
+        I8Constant as constant,
+        I8Categorical as categorical,
+        Incrementing as incrementing,
+    ) for i8,
 );
 
 #[cfg(test)]

core/src/schema/content/array.rs

@@ -112,6 +112,15 @@ impl<'de> Deserialize<'de> for ArrayContent {
                             r.low .unwrap_or_default() .into(),
                         ).map_err(A::Error::custom)?
                     },
+                    Content::Number(NumberContent::I8(number_content::I8::Range(r))) => {
+                        if r.high.is_none() {
+                            return Err(A::Error::custom("missing high value for array length range"));
+                        }
+
+                        is_positive(
+                            r.low .unwrap_or_default() .into(),
+                        ).map_err(A::Error::custom)?
+                    },
                     Content::SameAs(_) => {},
                     Content::Null(_) => return Err(de::Error::custom("array length is missing. Try adding '\"length\": [number]' to the array type where '[number]' is a positive integer")),
                     Content::Empty(_) => return Err(de::Error::custom("array length is not a constant or number type. Try replacing the '\"length\": {}' with '\"length\": [number]' where '[number]' is a positive integer")),

core/src/schema/content/number.rs

@@ -4,7 +4,7 @@ use std::hash::{Hash, Hasher};
 
 use super::Categorical;
 
-use crate::graph::number::{RandomF32, RandomI16, RandomI32, RandomU32};
+use crate::graph::number::{RandomF32, RandomI16, RandomI32, RandomI8, RandomU32};
 use serde::{ser::Serializer, Serialize};
 
 #[derive(Clone, Copy)]
@@ -205,6 +205,7 @@ impl NumberContent {
         match self {
             NumberContent::U32(_) => Ok(Self::u32_default_id()),
             NumberContent::U64(_) => Ok(Self::u64_default_id()),
+            NumberContent::I8(_) => Ok(Self::i8_default_id()),
             NumberContent::I16(_) => Ok(Self::i16_default_id()),
             NumberContent::I32(_) => Ok(Self::i32_default_id()),
             NumberContent::I64(_) => Ok(Self::i64_default_id()),
@@ -221,6 +222,10 @@ impl NumberContent {
         NumberContent::U64(number_content::U64::Id(Id::default()))
     }
 
+    pub fn i8_default_id() -> Self {
+        NumberContent::I8(number_content::I8::Id(Id::default()))
+    }
+
     pub fn i16_default_id() -> Self {
         NumberContent::I16(number_content::I16::Id(Id::default()))
     }
@@ -372,7 +377,7 @@ macro_rules! derive_hash {
     };
 }
 
-derive_hash!(i16, i32, u32, i64, u64, f32, f64);
+derive_hash!(i8, i16, i32, u32, i64, u64, f32, f64);
 
 number_content!(
     #[derive(PartialEq, Hash)]
@@ -390,6 +395,13 @@ number_content!(
         Id(crate::schema::Id<u64>),
     },
     #[derive(PartialEq, Hash)]
+    i8[is_i8, default_i8_range] as I8 {
+        Range(RangeStep<i8>),
+        Categorical(Categorical<i8>),
+        Constant(i8),
+        Id(crate::schema::Id<i8>),
+    },
+    #[derive(PartialEq, Hash)]
     i16[is_i16, default_i16_range] as I16 {
         Range(RangeStep<i16>),
         Categorical(Categorical<i16>),
@@ -490,6 +502,19 @@ impl Compile for NumberContent {
                 };
                 random_f32.into()
             }
+            Self::I8(i8_content) => {
+                let random_i8 = match i8_content {
+                    number_content::I8::Range(range) => RandomI8::range(*range)?,
+                    number_content::I8::Categorical(categorical_content) => {
+                        RandomI8::categorical(categorical_content.clone())
+                    }
+                    number_content::I8::Constant(val) => RandomI8::constant(*val),
+                    number_content::I8::Id(id) => {
+                        RandomI8::incrementing(Incrementing::new_at(id.start_at.unwrap_or(1)))
+                    }
+                };
+                random_i8.into()
+            }
             Self::I16(i16_content) => {
                 let random_i16 = match i16_content {
                     number_content::I16::Range(range) => RandomI16::range(*range)?,

core/src/schema/inference/mod.rs

@@ -15,7 +15,7 @@ use super::{
     Content, DateTimeContent, Id, NumberContent, NumberKindExt, ObjectContent, OneOfContent,
     RangeStep, StringContent, ValueKindExt,
 };
-use crate::graph::prelude::content::number_content::{I16, I32, I64};
+use crate::graph::prelude::content::number_content::{I16, I32, I64, I8};
 use crate::schema::UniqueContent;
 use num::Zero;
 
@@ -335,6 +335,17 @@ impl MergeStrategy<number_content::I16, i16> for OptionalMergeStrategy {
     }
 }
 
+impl MergeStrategy<number_content::I8, i8> for OptionalMergeStrategy {
+    fn try_merge(self, master: &mut number_content::I8, candidate: &i8) -> Result<()> {
+        match master {
+            number_content::I8::Range(range) => self.try_merge(range, candidate),
+            number_content::I8::Categorical(cat) => self.try_merge(cat, candidate),
+            number_content::I8::Constant(cst) => self.try_merge(cst, candidate),
+            I8::Id(id) => self.try_merge(id, candidate),
+        }
+    }
+}
+
 impl MergeStrategy<NumberContent, Number> for OptionalMergeStrategy {
     fn try_merge(self, master: &mut NumberContent, value: &Number) -> Result<()> {
         match master {
@@ -383,6 +394,13 @@ impl MergeStrategy<NumberContent, Number> for OptionalMergeStrategy {
                     todo!()
                 }
             }
+            NumberContent::I8(i8_content) => {
+                if let Some(n) = value.as_i64() {
+                    self.try_merge(i8_content, &(n as i8))
+                } else {
+                    todo!()
+                }
+            }
             NumberContent::I16(i16_content) => {
                 if let Some(n) = value.as_i64() {
                     self.try_merge(i16_content, &(n as i16))

synth/src/datasource/mysql_datasource.rs

@@ -10,7 +10,7 @@ use rust_decimal::Decimal;
 use sqlx::mysql::{MySqlColumn, MySqlPoolOptions, MySqlRow};
 use sqlx::{Column, MySql, Pool, Row, TypeInfo};
 use std::collections::BTreeMap;
-use synth_core::schema::number_content::{F64, I64, U64};
+use synth_core::schema::number_content::{F64, I16, I32, I64, I8, U64};
 use synth_core::schema::{
     ChronoValueType, DateTimeContent, NumberContent, RangeStep, RegexContent, StringContent,
 };
@@ -88,7 +88,8 @@ impl SqlxDataSource for MySqlDataSource {
 
     fn decode_to_content(&self, column_info: &ColumnInfo) -> Result<Content> {
         let content = match column_info.data_type.to_lowercase().as_str() {
-            "char" | "varchar" | "text" | "binary" | "varbinary" | "enum" | "set" => {
+            "char" | "varchar" | "text" | "binary" | "varbinary" | "enum" | "set"
+            | "mediumtext" | "blob" => {
                 let pattern = "[a-zA-Z0-9]{0, {}}".replace(
                     "{}",
                     &format!("{}", column_info.character_maximum_length.unwrap_or(1)),
@@ -97,9 +98,12 @@ impl SqlxDataSource for MySqlDataSource {
                     RegexContent::pattern(pattern).context("pattern will always compile")?,
                 ))
             }
-            "int" | "integer" | "tinyint" | "smallint" | "mediumint" | "bigint" => {
-                Content::Number(NumberContent::I64(I64::Range(RangeStep::default())))
+            "int" | "integer" | "mediumint" => {
+                Content::Number(NumberContent::I32(I32::Range(RangeStep::default())))
             }
+            "tinyint" => Content::Number(NumberContent::I8(I8::Range(RangeStep::default()))),
+            "smallint" => Content::Number(NumberContent::I16(I16::Range(RangeStep::default()))),
+            "bigint" => Content::Number(NumberContent::I64(I64::Range(RangeStep::default()))),
             "serial" => Content::Number(NumberContent::U64(U64::Range(RangeStep::default()))),
             "float" | "double" | "numeric" | "decimal" => {
                 Content::Number(NumberContent::F64(F64::Range(RangeStep::default())))

synth/testing_harness/mysql/0_hospital_schema.sql

@@ -27,5 +27,6 @@ create table patients
     date_joined date,
     address     varchar(255),
     phone       varchar(20),
-    ssn         varchar(12)
-);
+    ssn         varchar(12),
+    freckles    tinyint
+);