Rewrite python script in rust

.gitignore

@@ -3,4 +3,7 @@
 /tmp
 .idea/
 transform*.yaml
-__pycache__
+__pycache__
+
+target/
+Cargo.lock

.vscode/settings.json

@@ -1,11 +1,8 @@
 {
-  "rust-analyzer.assist.importMergeBehavior": "last",
   "editor.formatOnSave": true,
   "rust-analyzer.cargo.allFeatures": false,
   "rust-analyzer.checkOnSave.allFeatures": false,
   "rust-analyzer.checkOnSave.allTargets": false,
-  "rust-analyzer.cargo.target": "thumbv7em-none-eabihf",
-  "rust-analyzer.checkOnSave.target": "thumbv7em-none-eabihf",
   "rust-analyzer.procMacro.enable": true,
   "rust-analyzer.cargo.loadOutDirsFromCheck": true,
   "files.watcherExclude": {

Cargo.toml

@@ -0,0 +1,27 @@
+[package]
+name = "stm32-data"
+version = "0.1.0"
+edition = "2021"
+
+# See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html
+
+[features]
+default = ["rayon"]
+rayon = ["dep:rayon"]
+
+[dependencies]
+anyhow = "1.0.66"
+glob = "0.3.0"
+num = "0.4.0"
+quick-xml = { version = "0.26.0", features = ["serialize"] }
+regex = "1.6.0"
+serde = { version = "1.0.147", features = ["derive"] }
+serde_yaml = "0.9.14"
+chiptool = { git = "https://github.com/embassy-rs/chiptool", rev = "28ffa8a19d84914089547f52900ffb5877a5dc23" }
+serde_json = "1.0.87"
+rayon = { version = "1.5.3", optional = true }
+stm32-data-serde = { version = "0.1.0", path = "stm32-data-serde" }
+ref_thread_local = "0.1.1"
+
+[profile.release]
+debug = true

header_map.yaml

@@ -52,7 +52,7 @@ STM32F407xx: STM32F407VG, STM32F407VE, STM32F407ZG, STM32F407ZE, STM32F407IG, ST
 STM32F417xx: STM32F417VG, STM32F417VE, STM32F417ZG, STM32F417ZE, STM32F417IG, STM32F417IE
 STM32F427xx: STM32F427VG, STM32F427VI, STM32F427ZG, STM32F427ZI, STM32F427IG, STM32F427II
 STM32F437xx: STM32F437VG, STM32F437VI, STM32F437ZG, STM32F437ZI, STM32F437IG, STM32F437II
-STM32F429xx: STM32F429VG, STM32F429VI, STM32F429ZG, STM32F429ZI, STM32F429BG, STM32F429BI, STM32F429NG, STM32F439NI, STM32F429IG, STM32F429II
+STM32F429xx: STM32F429VG, STM32F429VI, STM32F429ZG, STM32F429ZI, STM32F429BG, STM32F429BI, STM32F429NG, STM32F429IG, STM32F429II
 STM32F439xx: STM32F439VG, STM32F439VI, STM32F439ZG, STM32F439ZI, STM32F439BG, STM32F439BI, STM32F439NG, STM32F439NI, STM32F439IG, STM32F439II
 STM32F401xC: STM32F401CB, STM32F401CC, STM32F401RB, STM32F401RC, STM32F401VB, STM32F401VC
 STM32F401xE: STM32F401CD, STM32F401RD, STM32F401VD, STM32F401CE, STM32F401RE, STM32F401VE
@@ -69,7 +69,7 @@ STM32F412Vx: STM32F412VET, STM32F412VGT, STM32F412VEH, STM32F412VGH
 STM32F412Rx: STM32F412RET, STM32F412RGT, STM32F412REY, STM32F412RGY
 STM32F413xx: STM32F413CH, STM32F413MH, STM32F413RH, STM32F413VH, STM32F413ZH, STM32F413CG, STM32F413MG, STM32F413RG, STM32F413VG, STM32F413ZG
 STM32F423xx: STM32F423CH, STM32F423RH, STM32F423VH, STM32F423ZH
-STM32F756xx: STM32F756VG, STM32F756ZG, STM32F756ZG, STM32F756IG, STM32F756BG, STM32F756NG
+STM32F756xx: STM32F756VG, STM32F756ZG, STM32F756IG, STM32F756BG, STM32F756NG
 STM32F746xx: STM32F746VE, STM32F746VG, STM32F746ZE, STM32F746ZG, STM32F746IE, STM32F746IG, STM32F746BE, STM32F746BG, STM32F746NE, STM32F746NG
 STM32F745xx: STM32F745VE, STM32F745VG, STM32F745ZG, STM32F745ZE, STM32F745IE, STM32F745IG
 STM32F765xx: STM32F765BI, STM32F765BG, STM32F765NI, STM32F765NG, STM32F765II, STM32F765IG, STM32F765ZI, STM32F765ZG, STM32F765VI, STM32F765VG
@@ -87,7 +87,6 @@ STM32L010x4: STM32L010K4, STM32L010F4
 STM32L010x6: STM32L010C6
 STM32L010x8: STM32L010K8, STM32L010R8
 STM32L010xB: STM32L010RB
-STM32L011xx: STM32L031C6, STM32L031E6, STM32L031F6, STM32L031G6, STM32L031K6
 STM32L021xx: STM32L021D4, STM32L021F4, STM32L021G4, STM32L021K4
 STM32L031xx: STM32L031C6, STM32L031E6, STM32L031F6, STM32L031G6, STM32L031K6
 STM32L041xx: STM32L041C6, STM32L041K6, STM32L041G6, STM32L041F6, STM32L041E6
@@ -164,7 +163,7 @@ STM32H7A3xxQ: STM32H7A3QIY6Q, STM32H7A3IIK6Q, STM32H7A3IIT6Q, STM32H7A3LIH6Q, ST
 STM32H7B3xx: STM32H7B3IIK6, STM32H7B3IIT6, STM32H7B3NIH6, STM32H7B3RIT6, STM32H7B3VIH6, STM32H7B3VIT6, STM32H7B3ZIT6
 STM32H7B3xxQ: STM32H7B3QIY6Q, STM32H7B3IIK6Q, STM32H7B3IIT6Q, STM32H7B3LIH6Q, STM32H7B3VIH6Q, STM32H7B3VIT6Q, STM32H7B3AII6Q, STM32H7B3ZIT6Q
 STM32H735xx: STM32H735AGI6, STM32H735IGK6, STM32H735RGV6, STM32H735VGT6, STM32H735VGY6, STM32H735ZGT6
-STM32H733xx: STM32H733VGH6, STM32H733VGT6, STM32H733ZGI6, STM32H733ZGT6,
+STM32H733xx: STM32H733VGH6, STM32H733VGT6, STM32H733ZGI6, STM32H733ZGT6
 STM32H730xx: STM32H730VBH6, STM32H730VBT6, STM32H730ZBT6, STM32H730ZBI6
 STM32H730xxQ: STM32H730IBT6Q, STM32H730ABI6Q, STM32H730IBK6Q
 STM32H725xx: STM32H725AGI6, STM32H725IGK6, STM32H725IGT6, STM32H725RGV6, STM32H725VGT6, STM32H725VGY6, STM32H725ZGT6, STM32H725REV6, SM32H725VET6, STM32H725ZET6, STM32H725AEI6, STM32H725IET6, STM32H725IEK6

src/dma.rs

@@ -0,0 +1,337 @@
+use std::collections::HashMap;
+
+mod xml {
+    use serde::Deserialize;
+
+    #[derive(Debug, Deserialize, PartialEq)]
+    pub struct Ip {
+        #[serde(rename = "Name")]
+        pub name: String,
+        #[serde(rename = "Version")]
+        pub version: String,
+        #[serde(rename = "RefParameter")]
+        pub ref_parameters: Vec<RefParameter>,
+        #[serde(rename = "RefMode")]
+        pub ref_modes: Vec<RefMode>,
+        #[serde(rename = "ModeLogicOperator")]
+        pub mode_logic_operator: ModeLogicOperator,
+    }
+
+    #[derive(Debug, Deserialize, PartialEq)]
+    pub struct RefMode {
+        #[serde(rename = "Name")]
+        pub name: String,
+        #[serde(rename = "BaseMode")]
+        pub base_mode: Option<String>,
+        #[serde(rename = "Parameter")]
+        pub parameters: Vec<Parameter>,
+    }
+
+    #[derive(Debug, Deserialize, PartialEq)]
+    pub struct Parameter {
+        #[serde(rename = "Name")]
+        pub name: String,
+        #[serde(rename = "PossibleValue", default)]
+        pub possible_values: Vec<String>,
+    }
+
+    #[derive(Debug, Deserialize, PartialEq)]
+    pub struct RefParameter {
+        #[serde(rename = "Name")]
+        pub name: String,
+        #[serde(rename = "PossibleValue", default)]
+        pub possible_values: Vec<PossibleValue>,
+    }
+
+    #[derive(Debug, Deserialize, PartialEq)]
+    pub struct PossibleValue {
+        #[serde(rename = "Comment")]
+        pub comment: String,
+        #[serde(rename = "Value")]
+        pub value: String,
+    }
+
+    #[derive(Debug, Deserialize, PartialEq)]
+    pub struct ModeLogicOperator {
+        #[serde(rename = "Name")]
+        pub name: String,
+        #[serde(rename = "Mode")]
+        pub modes: Vec<Mode>,
+    }
+
+    #[derive(Debug, Deserialize, PartialEq)]
+    pub struct Mode {
+        #[serde(rename = "Name")]
+        pub name: String,
+        #[serde(rename = "ModeLogicOperator", default)]
+        pub mode_logic_operator: Option<ModeLogicOperator>,
+    }
+}
+
+#[derive(Debug, PartialEq)]
+pub struct ChipDma {
+    pub peripherals: HashMap<String, Vec<stm32_data_serde::chip::core::peripheral::DmaChannel>>,
+    pub channels: Vec<stm32_data_serde::chip::core::DmaChannels>,
+}
+
+#[derive(Debug)]
+pub struct DmaChannels(pub HashMap<String, ChipDma>);
+
+impl DmaChannels {
+    pub fn parse() -> anyhow::Result<Self> {
+        let mut dma_channels = HashMap::new();
+        for f in glob::glob("sources/cubedb/mcu/IP/DMA*Modes.xml")?
+            .chain(glob::glob("sources/cubedb/mcu/IP/BDMA*Modes.xml")?)
+        {
+            let parsed: xml::Ip = quick_xml::de::from_str(&std::fs::read_to_string(f?)?)?;
+
+            let ff = parsed.version.clone();
+            let is_explicitly_bdma = match parsed.name.as_str() {
+                "DMA" | "DMA2D" => false,
+                "BDMA" | "BDMA1" | "BDMA2" => true,
+                name => panic!("Unrecognized DMA name: {name}"),
+            };
+
+            let mut chip_dma = ChipDma {
+                peripherals: HashMap::new(),
+                channels: Vec::new(),
+            };
+
+            for dma in parsed.mode_logic_operator.modes {
+                let dma_peri_name = dma.name.clone();
+                if dma_peri_name.contains(" Context") {
+                    continue;
+                }
+                let channels = dma.mode_logic_operator.unwrap().modes;
+                if channels.len() == 1 {
+                    // ========== CHIP WITH DMAMUX
+
+                    let dmamux_file = {
+                        if ff.starts_with("STM32L4P") {
+                            "L4PQ"
+                        } else if ff.starts_with("STM32L4S") {
+                            "L4RS"
+                        } else {
+                            &ff[5..7]
+                        }
+                    };
+
+                    let dmamux = match is_explicitly_bdma {
+                        true => "DMAMUX2",
+                        false => "DMAMUX1",
+                    };
+
+                    let mut mfs: Vec<_> = glob::glob(&format!("data/dmamux/{dmamux_file}_*.yaml"))?
+                        .map(Result::unwrap)
+                        .collect();
+                    mfs.sort();
+                    for mf in mfs {
+                        let y: HashMap<String, u8> = serde_yaml::from_str(&std::fs::read_to_string(&mf)?)?;
+
+                        let mf = mf.file_name().unwrap().to_string_lossy();
+                        let (_, req_dmamux) = mf.strip_suffix(".yaml").unwrap().split_once('_').unwrap(); // DMAMUX1 or DMAMUX2
+
+                        if req_dmamux == dmamux {
+                            for (request_name, request_num) in y {
+                                let parts: Vec<_> = request_name.split('_').collect();
+                                let target_peri_name = parts[0];
+                                let request = {
+                                    if parts.len() < 2 {
+                                        target_peri_name
+                                    } else {
+                                        parts[1]
+                                    }
+                                };
+                                chip_dma
+                                    .peripherals
+                                    .entry(target_peri_name.to_string())
+                                    .or_default()
+                                    .push(stm32_data_serde::chip::core::peripheral::DmaChannel {
+                                        signal: request.to_string(),
+                                        channel: None,
+                                        dmamux: Some(req_dmamux.to_string()),
+                                        request: Some(request_num),
+                                        dma: None,
+                                    })
+                            }
+                        }
+                    }
+
+                    let mut dmamux_channel = 0;
+                    for n in dma_peri_name.split(',') {
+                        let n = n.trim();
+                        let re = regex::Regex::new(&format!(".*{n}{}", r"_(Channel|Stream)\[(\d+)-(\d+)\]")).unwrap();
+                        if let Some(result) = re.captures(&channels[0].name) {
+                            let low: u8 = result.get(2).unwrap().as_str().parse()?;
+                            let high: u8 = result.get(3).unwrap().as_str().parse()?;
+                            for i in low..=high {
+                                chip_dma.channels.push(stm32_data_serde::chip::core::DmaChannels {
+                                    name: format!("{n}_CH{i}"),
+                                    dma: n.to_string(),
+                                    // Make sure all channels numbers start at 0
+                                    channel: i - low,
+                                    dmamux: Some(dmamux.to_string()),
+                                    dmamux_channel: Some(dmamux_channel),
+                                    supports_2d: None,
+                                });
+                                dmamux_channel += 1;
+                            }
+                        }
+                    }
+                } else {
+                    // ========== CHIP WITHOUT DMAMUX
+
+                    // see if we can scrape out requests
+                    let mut requests = HashMap::<String, u8>::new();
+                    for block in parsed
+                        .ref_modes
+                        .iter()
+                        .filter(|x| x.base_mode == Some("DMA_Request".to_string()))
+                    {
+                        let name = block.name.clone();
+                        // Depending on the chip, the naming is "Channel" or "Request"...
+                        if let Some(request_num) = block
+                            .parameters
+                            .iter()
+                            .find(|x| x.name == "Channel" || x.name == "Request")
+                        {
+                            assert_eq!(request_num.possible_values.len(), 1);
+                            let request_num = request_num.possible_values[0].clone();
+                            if request_num.starts_with("BDMA1_REQUEST_") {
+                                continue;
+                            }
+                            let request_num = request_num
+                                .strip_prefix("DMA_CHANNEL_")
+                                .or_else(|| request_num.strip_prefix("DMA_REQUEST_"))
+                                .unwrap();
+                            requests.insert(name, request_num.parse().unwrap());
+                        }
+                    }
+
+                    let mut channel_names: Vec<u8> = Vec::new();
+
+                    for channel in channels {
+                        let channel_name = channel.name;
+                        let (_, channel_name) = channel_name.split_once('_').unwrap();
+                        let channel_name = channel_name
+                            .strip_prefix("Channel")
+                            .or_else(|| channel_name.strip_prefix("Stream"))
+                            .unwrap();
+
+                        channel_names.push(channel_name.parse().unwrap());
+                        chip_dma.channels.push(stm32_data_serde::chip::core::DmaChannels {
+                            name: format!("{dma_peri_name}_CH{channel_name}"),
+                            dma: dma_peri_name.clone(),
+                            channel: channel_name.parse().unwrap(),
+                            dmamux: None,
+                            dmamux_channel: None,
+                            supports_2d: None,
+                        });
+                        for target in channel.mode_logic_operator.unwrap().modes {
+                            let original_target_name = target.name;
+                            let parts: Vec<_> = original_target_name.split(':').collect();
+                            let target_name = parts[0];
+
+                            //  Chips with single DAC refer to channels by DAC1/DAC2
+                            let target_name = match target_name {
+                                "DAC1" => "DAC_CH1",
+                                "DAC2" => "DAC_CH2",
+                                x => x,
+                            };
+
+                            let parts: Vec<_> = target_name.split('_').collect();
+                            let target_peri_name = parts[0];
+                            let target_requests = {
+                                if parts.len() < 2 {
+                                    vec![target_peri_name]
+                                } else {
+                                    target_name.split('_').nth(1).unwrap().split('/').collect()
+                                }
+                            };
+                            if target_name != "MEMTOMEM" {
+                                let target_peri_name = match target_peri_name {
+                                    "LPUART" => "LPUART1",
+                                    x => x,
+                                };
+                                for request in target_requests {
+                                    assert!(!request.contains(':'));
+                                    let entry = stm32_data_serde::chip::core::peripheral::DmaChannel {
+                                        signal: request.to_string(),
+                                        channel: Some(format!("{dma_peri_name}_CH{channel_name}")),
+                                        dmamux: None,
+                                        request: requests.get(&original_target_name).copied(),
+                                        dma: None,
+                                    };
+                                    chip_dma
+                                        .peripherals
+                                        .entry(target_peri_name.to_string())
+                                        .or_default()
+                                        .push(entry);
+                                }
+                            }
+                        }
+                    }
+
+                    // Make sure all channels numbers start at 0
+                    if channel_names.iter().min().unwrap() != &0 {
+                        for ch in &mut chip_dma.channels {
+                            if ch.dma == dma_peri_name {
+                                ch.channel -= 1;
+                            }
+                        }
+                    }
+                }
+            }
+
+            dma_channels.insert(ff, chip_dma);
+        }
+
+        // STM32U5
+
+        let mut chip_dma = ChipDma {
+            peripherals: HashMap::new(),
+            channels: Vec::new(),
+        };
+
+        let parsed: HashMap<String, u8> =
+            serde_yaml::from_str(&std::fs::read_to_string("data/dmamux/U5_GPDMA1.yaml")?)?;
+
+        for (request_name, request_num) in parsed {
+            let parts: Vec<_> = request_name.split('_').collect();
+            let target_peri_name = parts[0];
+            let request = {
+                if parts.len() < 2 {
+                    target_peri_name
+                } else {
+                    parts[1]
+                }
+            };
+            chip_dma
+                .peripherals
+                .entry(target_peri_name.to_string())
+                .or_default()
+                .push(stm32_data_serde::chip::core::peripheral::DmaChannel {
+                    signal: request.to_string(),
+                    dma: Some("GPDMA1".to_string()),
+                    channel: None,
+                    dmamux: None,
+                    request: Some(request_num),
+                });
+        }
+
+        for i in 0..16 {
+            chip_dma.channels.push(stm32_data_serde::chip::core::DmaChannels {
+                name: format!("GPDMA1_CH{i}"),
+                dma: "GPDMA1".to_string(),
+                channel: i,
+                dmamux: None,
+                dmamux_channel: None,
+                supports_2d: Some(i >= 12),
+            });
+        }
+
+        dma_channels.insert("STM32U5_dma3_Cube".to_string(), chip_dma);
+
+        Ok(Self(dma_channels))
+    }
+}

src/docs.rs

@@ -0,0 +1,150 @@
+use std::collections::HashMap;
+
+mod mcufinder {
+    use serde::{Deserialize, Serialize};
+
+    #[derive(Debug, Serialize, Deserialize, PartialEq)]
+    pub struct Files {
+        #[serde(rename = "Files")]
+        pub files: Vec<File>,
+    }
+
+    #[derive(Debug, Serialize, Deserialize, PartialEq)]
+    pub struct File {
+        #[serde(rename = "URL")]
+        pub url: String,
+        // #[serde(rename = "displayName")]
+        // pub display_name: String,
+        pub id_file: String,
+        pub name: String,
+        // #[serde(rename = "related_MCU_count")]
+        // pub related_mcu_count: String,
+        pub title: String,
+        pub r#type: String,
+        // #[serde(rename = "versionNumber")]
+        // pub version_number: String,
+    }
+
+    #[derive(Debug, Serialize, Deserialize, PartialEq)]
+    pub struct Mcus {
+        #[serde(rename = "MCUs")]
+        pub mcus: Vec<Mcu>,
+    }
+
+    #[derive(Debug, Serialize, Deserialize, PartialEq)]
+    pub struct Mcu {
+        #[serde(rename = "RPN")]
+        pub rpn: String,
+        pub files: Vec<McuFile>,
+    }
+
+    #[derive(Debug, Serialize, Deserialize, PartialEq)]
+    pub struct McuFile {
+        pub file_id: String,
+    }
+}
+
+impl From<mcufinder::File> for stm32_data_serde::chip::Doc {
+    fn from(file: mcufinder::File) -> Self {
+        Self {
+            name: file.name,
+            title: file.title,
+            url: file.url,
+            r#type: parse_document_type(&file.r#type).to_string(),
+        }
+    }
+}
+
+#[derive(Debug, PartialEq)]
+pub struct AllMcuFiles(HashMap<String, stm32_data_serde::chip::Doc>);
+
+impl AllMcuFiles {
+    pub fn parse() -> anyhow::Result<Self> {
+        let j = std::fs::read_to_string("sources/mcufinder/files.json")?;
+        let parsed: mcufinder::Files = serde_json::from_str(&j)?;
+        let all_mcu_files = parsed
+            .files
+            .into_iter()
+            .map(|file| (file.id_file.clone(), file.into()))
+            .collect();
+        Ok(Self(all_mcu_files))
+    }
+}
+
+#[derive(Debug, PartialEq)]
+pub struct PerMcuFiles(HashMap<String, Vec<String>>);
+
+impl PerMcuFiles {
+    pub fn parse() -> anyhow::Result<Self> {
+        let j = std::fs::read_to_string("sources/mcufinder/mcus.json")?;
+        let parsed: mcufinder::Mcus = serde_json::from_str(&j)?;
+
+        let mut per_mcu_files = HashMap::<String, Vec<String>>::new();
+
+        for mcu in parsed.mcus {
+            let rpn = mcu.rpn;
+            let files = mcu.files.into_iter().map(|file| file.file_id);
+            per_mcu_files.entry(rpn.to_string()).or_default().extend(files);
+        }
+
+        Ok(Self(per_mcu_files))
+    }
+}
+
+pub struct Docs {
+    pub all_mcu_files: AllMcuFiles,
+    pub per_mcu_files: PerMcuFiles,
+}
+
+impl Docs {
+    pub fn parse() -> anyhow::Result<Self> {
+        Ok(Self {
+            all_mcu_files: AllMcuFiles::parse()?,
+            per_mcu_files: PerMcuFiles::parse()?,
+        })
+    }
+
+    pub fn documents_for(&self, chip_name: &str) -> Vec<stm32_data_serde::chip::Doc> {
+        let mut docs: Vec<_> = self
+            .per_mcu_files
+            .0
+            .get(chip_name)
+            .into_iter()
+            .flatten()
+            .flat_map(|id| {
+                if let Some(file) = self.all_mcu_files.0.get(id) {
+                    let order = order_doc_type(&file.r#type);
+                    Some((order, file))
+                } else {
+                    None
+                }
+            })
+            .collect();
+        docs.sort_by_key(|(order, file)| (*order, file.name.clone()));
+
+        docs.into_iter().map(|(_order, file)| file.clone()).collect()
+    }
+}
+
+fn parse_document_type(t: &str) -> &'static str {
+    match t {
+        "Reference manual" => "reference_manual",
+        "Programming manual" => "programming_manual",
+        "Datasheet" => "datahseet", // TODO: fix me
+        "Errata sheet" => "errata_sheet",
+        "Application note" => "application_note",
+        "User manual" => "user_manual",
+        _ => panic!("Unknown doc type {t}"),
+    }
+}
+
+fn order_doc_type(t: &str) -> u8 {
+    match t {
+        "reference_manual" => 0,
+        "programming_manual" => 1,
+        "datahseet" => 2, // TODO: fix me
+        "errata_sheet" => 3,
+        "application_note" => 4,
+        _ => panic!("Unknown doc type {t}"),
+    }
+}

src/gpio_af.rs

@@ -0,0 +1,125 @@
+use std::collections::HashMap;
+
+use crate::regex;
+
+mod xml {
+    use serde::Deserialize;
+
+    #[derive(Debug, Deserialize, PartialEq)]
+    pub struct Ip {
+        #[serde(rename = "Name")]
+        pub name: String,
+        #[serde(rename = "Version")]
+        pub version: String,
+        #[serde(rename = "GPIO_Pin")]
+        pub gpio_pins: Vec<GpioPin>,
+    }
+
+    #[derive(Debug, Deserialize, PartialEq)]
+    pub struct GpioPin {
+        #[serde(rename = "Name")]
+        pub name: String,
+        #[serde(rename = "PinSignal", default)]
+        pub pin_signals: Vec<PinSignal>,
+    }
+
+    #[derive(Debug, Deserialize, PartialEq)]
+    pub struct PinSignal {
+        #[serde(rename = "Name")]
+        pub name: String,
+        #[serde(rename = "SpecificParameter", default)]
+        pub specific_parameter: Option<SpecificParameter>,
+    }
+
+    #[derive(Debug, Deserialize, PartialEq)]
+    pub struct SpecificParameter {
+        #[serde(rename = "PossibleValue")]
+        pub possible_value: String,
+    }
+}
+
+pub fn clean_pin(pin_name: &str) -> Option<stm32_data_serde::chip::core::peripheral::pin::Pin> {
+    let pin_name = regex!(r"^P[A-Z]\d+").find(pin_name)?.as_str();
+
+    stm32_data_serde::chip::core::peripheral::pin::Pin::parse(pin_name)
+}
+
+#[derive(Debug, PartialEq, Eq)]
+pub struct Af(pub HashMap<String, HashMap<String, Vec<stm32_data_serde::chip::core::peripheral::Pin>>>);
+
+impl Af {
+    pub fn parse() -> anyhow::Result<Self> {
+        let mut af = HashMap::new();
+        for f in glob::glob("sources/cubedb/mcu/IP/GPIO-*_gpio_v1_0_Modes.xml")? {
+            let parsed: xml::Ip = quick_xml::de::from_str(&std::fs::read_to_string(f?)?)?;
+
+            let ff = parsed.version.strip_suffix("_gpio_v1_0").unwrap().to_string();
+
+            let mut peris = HashMap::<_, Vec<_>>::new();
+
+            for pin in parsed.gpio_pins {
+                // Cleanup pin name
+                let Some(pin_name) = clean_pin(&pin.name) else {continue};
+
+                // Extract AFs
+                for signal in pin.pin_signals {
+                    let Some((peri_name, signal_name)) = parse_signal_name(&signal.name) else {continue};
+                    let afn = if parsed.version.starts_with("STM32F1") {
+                        None
+                    } else {
+                        let afn = signal.specific_parameter.unwrap();
+                        let afn = afn
+                            .possible_value
+                            .split('_')
+                            .nth(1)
+                            .unwrap()
+                            .strip_prefix("AF")
+                            .unwrap()
+                            .parse()
+                            .unwrap();
+                        Some(afn)
+                    };
+                    peris.entry(peri_name.to_string()).or_default().push(
+                        stm32_data_serde::chip::core::peripheral::Pin {
+                            pin: pin_name,
+                            signal: signal_name.to_string(),
+                            af: afn,
+                        },
+                    );
+                }
+            }
+
+            for p in peris.values_mut() {
+                p.sort();
+                p.dedup();
+            }
+
+            af.insert(ff, peris);
+        }
+
+        Ok(Self(af))
+    }
+}
+
+fn parse_signal_name(signal_name: &str) -> Option<(&str, &str)> {
+    let (peri_name, signal_name) = {
+        if let Some(signal_name) = signal_name.strip_prefix("USB_OTG_FS_") {
+            ("USB_OTG_FS", signal_name)
+        } else if let Some(signal_name) = signal_name.strip_prefix("USB_OTG_HS_") {
+            ("USB_OTG_HS", signal_name)
+        } else {
+            signal_name.split_once('_')?
+        }
+    };
+
+    if signal_name.starts_with("EXTI") {
+        return None;
+    }
+    if peri_name.starts_with("DEBUG") && signal_name.starts_with("SUBGHZSPI") {
+        let (peri_name, signal_name) = signal_name.split_once('-').unwrap();
+
+        Some((peri_name, signal_name.strip_suffix("OUT").unwrap_or(signal_name)))
+    } else {
+        Some((peri_name, signal_name))
+    }
+}

src/header.rs

@@ -0,0 +1,286 @@
+use std::collections::HashMap;
+
+use crate::regex;
+
+pub struct Headers {
+    map: HeaderMap,
+    parsed: HeadersParsed,
+    regexes: Vec<(regex::Regex, String)>,
+}
+
+impl Headers {
+    pub fn parse() -> anyhow::Result<Self> {
+        let map = HeaderMap::parse()?;
+        let parsed = HeadersParsed::parse()?;
+        let regexes = parsed
+            .0
+            .keys()
+            .map(|h| {
+                let pattern = h.replace('x', ".");
+                let regex = regex::Regex::new(&format!("^{pattern}$")).unwrap();
+                (regex, h.clone())
+            })
+            .collect();
+        Ok(Self { map, parsed, regexes })
+    }
+
+    pub fn get_for_chip(&self, model: &str) -> Option<&ParsedHeader> {
+        let model = model.to_ascii_lowercase();
+        match self.map.0.get(&model) {
+            // if it's in the map, just go
+            Some(name) => Some(self.parsed.0.get(name).unwrap()),
+            // if not, find it by regex, taking `x` meaning `anything`
+            None => {
+                let mut results = self
+                    .regexes
+                    .iter()
+                    .filter_map(|(r, name)| if r.is_match(&model) { Some(name) } else { None });
+                let res = results.next();
+                assert_eq!(results.next(), None, "found more than one match");
+                res.map(|name| self.parsed.0.get(name).unwrap())
+            }
+        }
+    }
+}
+
+#[derive(Debug, PartialEq, Eq)]
+pub struct HeaderMap(pub HashMap<String, String>);
+
+impl HeaderMap {
+    pub fn parse() -> anyhow::Result<Self> {
+        let mut res = HashMap::new();
+        for (mut header, chips) in
+            serde_yaml::from_str::<HashMap<String, String>>(&std::fs::read_to_string("header_map.yaml")?)?
+        {
+            header.make_ascii_lowercase();
+            for chip in chips.split(',') {
+                let chip = chip.trim().to_ascii_lowercase();
+                if let Some(old) = res.insert(chip.clone(), header.clone()) {
+                    panic!("Duplicate {chip} found! Overwriting {old} with {header}");
+                }
+            }
+        }
+
+        Ok(Self(res))
+    }
+}
+
+#[derive(Debug, PartialEq, Eq)]
+pub struct HeadersParsed(pub HashMap<String, ParsedHeader>);
+
+impl HeadersParsed {
+    pub fn parse() -> anyhow::Result<Self> {
+        let files = glob::glob("sources/headers/*.h").unwrap().map(Result::unwrap);
+
+        let for_each_file = |f: std::path::PathBuf| {
+            let ff = f.file_name().unwrap().to_string_lossy();
+            let ff = ff.strip_suffix(".h").unwrap();
+            let parsed_header = ParsedHeader::parse(&f).unwrap();
+            (ff.to_string(), parsed_header)
+        };
+
+        #[cfg(feature = "rayon")]
+        {
+            use rayon::prelude::*;
+            Ok(Self(files.par_bridge().map(for_each_file).collect()))
+        }
+        #[cfg(not(feature = "rayon"))]
+        {
+            Ok(Self(files.map(for_each_file).collect()))
+        }
+    }
+}
+
+fn parens_ok(val: &str) -> bool {
+    let mut n: i32 = 0;
+    for c in val.chars() {
+        match c {
+            '(' => n += 1,
+            ')' => {
+                n -= 1;
+                if n < 0 {
+                    return false;
+                }
+            }
+            _ => {}
+        }
+    }
+    n == 0
+}
+
+#[derive(Clone, Debug, Default, PartialEq, Eq)]
+pub struct Defines(pub HashMap<String, i64>);
+
+impl Defines {
+    // warning: horrible abomination ahead
+    fn parse_value(&self, val: &str) -> i64 {
+        let val = val.trim();
+        if val.is_empty() {
+            0
+        } else if let Some(m) = regex!(r"^(0([1-9][0-9]*)(U))").captures(val) {
+            m.get(2).unwrap().as_str().parse().unwrap()
+        } else if let Some(m) = regex!(r"^((0x[0-9a-fA-F]+|\d+))(|u|ul|U|UL)$").captures(val) {
+            let x = m.get(1).unwrap().as_str();
+            match x.strip_prefix("0x") {
+                Some(x) => i64::from_str_radix(x, 16),
+                None => x.parse(),
+            }
+            .unwrap()
+        } else if let Some(m) = regex!(r"^([0-9A-Za-z_]+)$").captures(val) {
+            self.0.get(m.get(1).unwrap().as_str()).copied().unwrap_or(0)
+        } else if let Some(x) = regex!(r"^\((.*)\)$")
+            .captures(val)
+            .map(|m| m.get(1).unwrap().as_str())
+            .filter(|x| parens_ok(x))
+        {
+            self.parse_value(x)
+        } else if let Some(m) = regex!(r"^\*?\([0-9A-Za-z_]+ *\*?\)(.*)$").captures(val) {
+            self.parse_value(m.get(1).unwrap().as_str())
+        } else if let Some(m) = regex!(r"^(.*)/(.*)$").captures(val) {
+            self.parse_value(m.get(1).unwrap().as_str()) / self.parse_value(m.get(2).unwrap().as_str())
+        } else if let Some(m) = regex!(r"^(.*)<<(.*)$").captures(val) {
+            self.parse_value(m.get(1).unwrap().as_str()) << self.parse_value(m.get(2).unwrap().as_str()) & 0xFFFFFFFF
+        } else if let Some(m) = regex!(r"^(.*)>>(.*)$").captures(val) {
+            self.parse_value(m.get(1).unwrap().as_str()) >> self.parse_value(m.get(2).unwrap().as_str())
+        } else if let Some(m) = regex!(r"^(.*)\|(.*)$").captures(val) {
+            self.parse_value(m.get(1).unwrap().as_str()) | self.parse_value(m.get(2).unwrap().as_str())
+        } else if let Some(m) = regex!(r"^(.*)&(.*)$").captures(val) {
+            self.parse_value(m.get(1).unwrap().as_str()) & self.parse_value(m.get(2).unwrap().as_str())
+        } else if let Some(m) = regex!(r"^~(.*)$").captures(val) {
+            !self.parse_value(m.get(1).unwrap().as_str()) & 0xFFFFFFFF
+        } else if let Some(m) = regex!(r"^(.*)\+(.*)$").captures(val) {
+            self.parse_value(m.get(1).unwrap().as_str()) + self.parse_value(m.get(2).unwrap().as_str())
+        } else if let Some(m) = regex!(r"^(.*)-(.*)$").captures(val) {
+            self.parse_value(m.get(1).unwrap().as_str()) - self.parse_value(m.get(2).unwrap().as_str())
+        } else {
+            panic!("can't parse: {val:?}")
+        }
+    }
+
+    pub fn get_peri_addr(&self, pname: &str) -> Option<u32> {
+        const ALT_PERI_DEFINES: &[(&str, &[&str])] = &[
+            ("DBGMCU", &["DBGMCU_BASE", "DBG_BASE"]),
+            ("FLASH", &["FLASH_R_BASE", "FLASH_REG_BASE"]),
+            (
+                "ADC_COMMON",
+                &["ADC_COMMON", "ADC1_COMMON", "ADC12_COMMON", "ADC123_COMMON"],
+            ),
+            ("CAN", &["CAN_BASE", "CAN1_BASE"]),
+            ("FMC", &["FMC_BASE", "FMC_R_BASE"]),
+            ("FSMC", &["FSMC_R_BASE"]),
+        ];
+        let alt_peri_defines: HashMap<_, _> = ALT_PERI_DEFINES.iter().copied().collect();
+
+        let possible_defines: Vec<String> = alt_peri_defines
+            .get(pname)
+            .map(|x| x.iter().map(ToString::to_string).collect())
+            .unwrap_or_else(|| vec![format!("{pname}_BASE"), pname.to_string()]);
+        possible_defines
+            .into_iter()
+            .find_map(|d| self.0.get(&d).filter(|&&addr| addr != 0))
+            .map(|x| u32::try_from(*x).unwrap())
+    }
+}
+
+#[derive(Debug, PartialEq, Eq)]
+pub struct ParsedHeader {
+    pub cores: Vec<String>,
+    pub interrupts: HashMap<String, HashMap<String, u8>>,
+    pub defines: HashMap<String, Defines>,
+}
+
+impl ParsedHeader {
+    fn parse(f: impl AsRef<std::path::Path>) -> anyhow::Result<Self> {
+        let mut irqs = HashMap::<String, HashMap<String, u8>>::new();
+        let mut defines = HashMap::<String, Defines>::new();
+        let mut cores = Vec::<String>::new();
+        let mut cur_core = "all".to_string();
+
+        let mut accum = String::new();
+        let f = std::fs::read(f)?;
+        for l in f.split(|b| b == &b'\n') {
+            let l = String::from_utf8_lossy(l);
+            let l = l.trim();
+            let l = accum.clone() + l;
+            if l.ends_with('\\') {
+                accum = l.strip_suffix('\\').unwrap().to_string();
+                continue;
+            }
+            accum = String::new();
+
+            // Scoped by a single core
+            if let Some(m) = regex!(r".*if defined.*CORE_CM(\d+)(PLUS)?.*").captures(&l) {
+                cur_core = format!("cm{}", m.get(1).unwrap().as_str());
+                if m.get(2).is_some() {
+                    cur_core += "p";
+                }
+                if !cores.contains(&cur_core) {
+                    cores.push(cur_core.clone())
+                }
+            } else if regex!(r".*else.*").is_match(&l) {
+                cur_core = "all".to_string();
+                if let Some(m) = regex!(".*else.*CORE_CM(\\d+)(PLUS)?.*").captures(&l) {
+                    cur_core = format!("cm{}", m.get(1).unwrap().as_str());
+                    if m.get(2).is_some() {
+                        cur_core += "p";
+                    }
+                } else if cores.len() > 1 {
+                    // Pick the second core assuming we've already parsed one
+                    cur_core = cores[1].clone();
+                }
+
+                if !cores.contains(&cur_core) {
+                    cores.push(cur_core.clone());
+                }
+            } else if regex!(r".*endif.*").is_match(&l) {
+                cur_core = "all".to_string();
+            }
+
+            let irq_entry = irqs.entry(cur_core.clone()).or_default();
+            let defines_entry = defines.entry(cur_core.clone()).or_default();
+
+            if let Some(m) = regex!(r"^([a-zA-Z0-9_]+)_IRQn += (\d+),? +/\*!< (.*) \*/").captures(&l) {
+                irq_entry.insert(
+                    m.get(1).unwrap().as_str().to_string(),
+                    m.get(2).unwrap().as_str().parse().unwrap(),
+                );
+            }
+
+            if let Some(m) = regex!(r"^#define +([0-9A-Za-z_]+)\(").captures(&l) {
+                defines_entry.0.insert(m.get(1).unwrap().as_str().to_string(), -1);
+            }
+
+            if let Some(m) = regex!(r"^#define +([0-9A-Za-z_]+) +(.*)").captures(&l) {
+                let name = m.get(1).unwrap().as_str().trim();
+                if name == "FLASH_SIZE" {
+                    continue;
+                }
+                let val = m.get(2).unwrap().as_str();
+                let val = val.split("/*").next().unwrap().trim();
+                let val = defines_entry.parse_value(val);
+
+                defines_entry.0.insert(name.to_string(), val);
+            }
+        }
+
+        if cores.is_empty() {
+            cores = vec!["all".to_string()];
+        }
+
+        for core in &mut cores {
+            if core != "all" {
+                let all_irqs = irqs.get("all").unwrap().clone();
+                irqs.get_mut(core).unwrap().extend(all_irqs);
+
+                let all_defines = defines.get("all").unwrap().clone();
+                defines.get_mut(core).unwrap().0.extend(all_defines.0);
+            }
+        }
+
+        Ok(Self {
+            cores,
+            interrupts: irqs,
+            defines,
+        })
+    }
+}

src/interrupts.rs

@@ -0,0 +1,384 @@
+use std::collections::{HashMap, HashSet};
+
+use crate::regex;
+
+mod xml {
+    use serde::Deserialize;
+
+    #[derive(Debug, Deserialize, PartialEq)]
+    pub struct Ip {
+        #[serde(rename = "Name")]
+        pub name: String,
+        #[serde(rename = "Version")]
+        pub version: String,
+        #[serde(rename = "RefParameter")]
+        pub ref_parameters: Vec<RefParameter>,
+    }
+
+    #[derive(Debug, Deserialize, PartialEq)]
+    pub struct RefParameter {
+        #[serde(rename = "Name")]
+        pub name: String,
+        #[serde(rename = "PossibleValue", default)]
+        pub possible_values: Vec<PossibleValue>,
+    }
+
+    #[derive(Debug, Deserialize, PartialEq)]
+    pub struct PossibleValue {
+        #[serde(rename = "Comment")]
+        pub comment: String,
+        #[serde(rename = "Value")]
+        pub value: String,
+    }
+}
+
+#[derive(Debug)]
+pub struct ChipInterrupts(
+    pub HashMap<(String, String), HashMap<String, Vec<stm32_data_serde::chip::core::peripheral::Interrupt>>>,
+);
+
+impl ChipInterrupts {
+    pub fn parse() -> anyhow::Result<Self> {
+        let mut chip_interrupts = HashMap::new();
+
+        let mut files: Vec<_> = glob::glob("sources/cubedb/mcu/IP/NVIC*_Modes.xml")?
+            .map(Result::unwrap)
+            .filter(|file| !file.to_string_lossy().contains("STM32MP1"))
+            .collect();
+        files.sort();
+
+        for f in files {
+            let mut irqs = HashMap::<String, _>::new();
+            let file = std::fs::read_to_string(f)?;
+            let parsed: xml::Ip = quick_xml::de::from_str(&file)?;
+            for irq in parsed
+                .ref_parameters
+                .into_iter()
+                .filter(|param| param.name == "IRQn")
+                .flat_map(|param| param.possible_values)
+            {
+                let parts = {
+                    let mut iter = irq.value.split(':');
+                    let parts = [(); 5].map(|_| iter.next().unwrap());
+                    assert!(iter.next().is_none());
+                    parts
+                };
+
+                let name = {
+                    let name = parts[0].strip_suffix("_IRQn").unwrap();
+
+                    // Fix typo in STM32Lxx and L083 devices
+                    let contains_rng = || parts[2..].iter().flat_map(|x| x.split(',')).any(|x| x == "RNG");
+                    if name == "AES_RNG_LPUART1" && !contains_rng() {
+                        "AES_LPUART1"
+                    } else {
+                        name
+                    }
+                };
+
+                let entry = match irqs.entry(name.to_string()) {
+                    std::collections::hash_map::Entry::Occupied(_) => continue,
+                    std::collections::hash_map::Entry::Vacant(entry) => entry,
+                };
+
+                // Flags.
+                // Y
+                //   unknown, it's in all of them
+                // H3, nHS
+                //   ???
+                // 2V, 3V, nV, 2V1
+                //   unknown, it has to do with the fact the irq is shared among N peripehrals
+                // DMA, DMAL0, DMAF0, DMAL0_DMAMUX, DMAF0_DMAMUX
+                //   special format for DMA
+                // DFSDM
+                //   special format for DFSDM
+                // EXTI
+                //   special format for EXTI
+                let flags: Vec<_> = parts[1].split(',').collect();
+
+                // F100xE MISC_REMAP remaps some DMA IRQs, so ST decided to give two names
+                // to the same IRQ number.
+                if parsed.version == "STM32F100E" && name == "DMA2_Channel4_5" {
+                    continue;
+                }
+                // F3 can remap USB IRQs, ignore them.
+                if parsed.version.starts_with("STM32F3") && irq.comment.contains("remap") {
+                    continue;
+                }
+                let mut signals = HashSet::<(String, String)>::new();
+                if [
+                    "NonMaskableInt",
+                    "HardFault",
+                    "MemoryManagement",
+                    "BusFault",
+                    "UsageFault",
+                    "SVCall",
+                    "DebugMonitor",
+                    "PendSV",
+                    "SysTick",
+                ]
+                .contains(&name)
+                {
+                    // pass
+                } else if flags
+                    .iter()
+                    .map(|flag| ["DMA", "DMAL0", "DMAF0", "DMAL0_DMAMUX", "DMAF0_DMAMUX"].contains(flag))
+                    .any(std::convert::identity)
+                {
+                    let mut dmas_iter = parts[3].split(',');
+                    let mut chans_iter = parts[4].split(';');
+                    for (dma, chan) in std::iter::zip(&mut dmas_iter, &mut chans_iter) {
+                        let range = {
+                            let mut ch = chan.split(',');
+                            let ch_from: usize = ch.next().unwrap().parse().unwrap();
+                            let ch_to = match ch.next() {
+                                Some(ch_to) => ch_to.parse().unwrap(),
+                                None => ch_from,
+                            };
+                            assert!(ch.next().is_none());
+                            ch_from..=ch_to
+                        };
+                        for ch in range {
+                            signals.insert((dma.to_string(), format!("CH{ch}")));
+                        }
+                    }
+                    assert!(dmas_iter.next().is_none());
+                    assert!(chans_iter.next().is_none());
+                } else if name == "DMAMUX1" || name == "DMAMUX1_S" || name == "DMAMUX_OVR" || name == "DMAMUX1_OVR" {
+                    signals.insert(("DMAMUX1".to_string(), "OVR".to_string()));
+                } else if name == "DMAMUX2_OVR" {
+                    signals.insert(("DMAMUX2".to_string(), "OVR".to_string()));
+                } else if flags.contains(&"DMAMUX") {
+                    panic!("should've been handled above");
+                } else if flags.contains(&"EXTI") {
+                    for signal in parts[2].split(',') {
+                        signals.insert(("EXTI".to_string(), signal.to_string()));
+                    }
+                } else if name == "FLASH" {
+                    signals.insert(("FLASH".to_string(), "GLOBAL".to_string()));
+                } else if name == "CRS" {
+                    signals.insert(("RCC".to_string(), "CRS".to_string()));
+                } else if name == "RCC" {
+                    signals.insert(("RCC".to_string(), "GLOBAL".to_string()));
+                } else {
+                    if parts[2].is_empty() {
+                        continue;
+                    }
+
+                    let peri_names: Vec<_> = parts[2].split(',').map(ToString::to_string).collect();
+
+                    let name2 = {
+                        if name == "USBWakeUp" || name == "USBWakeUp_RMP" {
+                            "USB_WKUP"
+                        } else {
+                            name.strip_suffix("_S").unwrap_or(name)
+                        }
+                    };
+
+                    let mut peri_signals: HashMap<_, _> = peri_names
+                        .iter()
+                        .map(|name| (name.clone(), Vec::<String>::new()))
+                        .collect();
+
+                    let mut curr_peris = Vec::new();
+                    if peri_names.len() == 1 {
+                        curr_peris = peri_names.clone();
+                    }
+
+                    // Parse IRQ signals from the IRQ name.
+                    for part in tokenize_name(name2) {
+                        let part = {
+                            if part == "TAMPER" {
+                                "TAMP".to_string()
+                            } else {
+                                part
+                            }
+                        };
+
+                        if part == "LSECSS" {
+                            signals.insert(("RCC".to_string(), "LSECSS".to_string()));
+                        } else if part == "CSS" {
+                            signals.insert(("RCC".to_string(), "CSS".to_string()));
+                        } else if part == "LSE" {
+                            signals.insert(("RCC".to_string(), "LSE".to_string()));
+                        } else if part == "CRS" {
+                            signals.insert(("RCC".to_string(), "CRS".to_string()));
+                        } else {
+                            let pp = match_peris(&peri_names, &part);
+                            if !pp.is_empty() {
+                                curr_peris = pp;
+                            } else {
+                                assert!(!curr_peris.is_empty());
+                                for p in &curr_peris {
+                                    peri_signals.entry(p.clone()).or_default().push(part.clone());
+                                }
+                            }
+                        }
+                    }
+
+                    for (p, mut ss) in peri_signals.into_iter() {
+                        let known = valid_signals(&p);
+
+                        // If we have no signals for the peri, assume it's "global" so assign it all known ones
+                        if ss.is_empty() {
+                            if p.starts_with("COMP") {
+                                ss = vec!["WKUP".to_string()];
+                            } else {
+                                ss = known.clone();
+                            }
+                        }
+
+                        for s in ss {
+                            if !known.contains(&s.clone()) {
+                                panic!("Unknown signal {s} for peri {p}, known={known:?}");
+                            }
+                            signals.insert((p.clone(), s));
+                        }
+                    }
+                }
+
+                // for (peri, signal) in &signals {
+                //     println!("    {peri}:{signal}");
+                // }
+                entry.insert(signals);
+            }
+
+            let mut irqs2 = HashMap::<_, Vec<_>>::new();
+            for (name, signals) in irqs {
+                for (p, s) in signals {
+                    irqs2
+                        .entry(p)
+                        .or_default()
+                        .push(stm32_data_serde::chip::core::peripheral::Interrupt {
+                            signal: s,
+                            interrupt: name.clone(),
+                        });
+                }
+            }
+
+            for pirqs in irqs2.values_mut() {
+                let mut psirqs = HashMap::<_, Vec<_>>::new();
+                for irq in pirqs {
+                    psirqs
+                        .entry(irq.signal.clone())
+                        .or_default()
+                        .push(irq.interrupt.clone());
+                }
+                // for (s, irqs) in psirqs {
+                //     if irqs.len() != 1 {
+                //         println!("DUPE: {p} {s} {irqs:?}");
+                //     }
+                // }
+            }
+
+            chip_interrupts.insert((parsed.name, parsed.version), irqs2);
+        }
+
+        Ok(Self(chip_interrupts))
+    }
+}
+
+fn tokenize_name(name: &str) -> Vec<String> {
+    // Treat IRQ names are "tokens" separated by `_`, except some tokens
+    // contain `_` themselves, such as `C1_RX`.
+    let r = regex!(r"(SPDIF_RX|EP\d+_(IN|OUT)|OTG_FS|OTG_HS|USB_FS|C1_RX|C1_TX|C2_RX|C2_TX|[A-Z0-9]+(_\d+)*)_*");
+    let name = name.to_ascii_uppercase();
+
+    r.captures_iter(&name)
+        .map(|cap| cap.get(1).unwrap().as_str().to_string())
+        .collect()
+}
+
+fn match_peris(peris: &[String], name: &str) -> Vec<String> {
+    const PERI_OVERRIDE: &[(&str, &[&str])] = &[
+        ("USB_FS", &["USB"]),
+        ("OTG_HS", &["USB_OTG_HS"]),
+        ("OTG_FS", &["USB_OTG_FS"]),
+        ("USB", &["USB_DRD_FS"]),
+        ("UCPD1_2", &["UCPD1", "UCPD2"]),
+        ("ADC1", &["ADC"]),
+        ("CEC", &["HDMI_CEC"]),
+        ("SPDIF_RX", &["SPDIFRX1", "SPDIFRX"]),
+        ("CAN1", &["CAN"]),
+        ("TEMP", &["TEMPSENS"]),
+        ("DSI", &["DSIHOST"]),
+        ("HRTIM1", &["HRTIM"]),
+        ("GTZC", &["GTZC_S"]),
+        ("TZIC", &["GTZC_S"]),
+    ];
+
+    let peri_override: HashMap<_, _> = PERI_OVERRIDE.iter().copied().collect();
+
+    if let Some(over) = peri_override.get(name) {
+        let mut res = Vec::new();
+        for p in *over {
+            if peris.contains(&p.to_string()) {
+                res.push(p.to_string());
+            }
+        }
+        if !res.is_empty() {
+            return res;
+        }
+    }
+    let mut name = name;
+    let mut res = Vec::new();
+    if let Some(m) = regex!(r"^(I2C|[A-Z]+)(\d+(_\d+)*)$").captures(name) {
+        name = m.get(1).unwrap().as_str();
+        for n in m.get(2).unwrap().as_str().split('_') {
+            let p = format!("{name}{n}");
+            if !peris.contains(&p) {
+                return Vec::new();
+            }
+            res.push(p);
+        }
+    } else {
+        for p in peris {
+            if p == name || { p.starts_with(name) && regex!(r"^\d+$").is_match(p.strip_prefix(name).unwrap_or(p)) } {
+                res.push(p.to_string());
+            }
+        }
+    }
+
+    res
+}
+
+fn valid_signals(peri: &str) -> Vec<String> {
+    const IRQ_SIGNALS_MAP: &[(&str, &[&str])] = &[
+        ("CAN", &["TX", "RX0", "RX1", "SCE"]),
+        ("FDCAN", &["IT0", "IT1", "CAL"]),
+        ("I2C", &["ER", "EV"]),
+        ("FMPI2C", &["ER", "EV"]),
+        ("TIM", &["BRK", "UP", "TRG", "COM", "CC"]),
+        // ("HRTIM", &["Master", "TIMA", "TIMB", "TIMC", "TIMD", "TIME", "TIMF"]),
+        ("RTC", &["ALARM", "WKUP", "TAMP", "STAMP", "SSRU"]),
+        ("SUBGHZ", &["RADIO"]),
+        ("IPCC", &["C1_RX", "C1_TX", "C2_RX", "C2_TX"]),
+        (
+            "HRTIM",
+            &["MASTER", "TIMA", "TIMB", "TIMC", "TIMD", "TIME", "TIMF", "FLT"],
+        ),
+        ("COMP", &["WKUP", "ACQ"]),
+        ("RCC", &["RCC", "CRS"]),
+        ("MDIOS", &["GLOBAL", "WKUP"]),
+        ("ETH", &["GLOBAL", "WKUP"]),
+        ("LTDC", &["GLOBAL", "ER"]),
+        (
+            "DFSDM",
+            &["FLT0", "FLT1", "FLT2", "FLT3", "FLT4", "FLT5", "FLT6", "FLT7"],
+        ),
+        ("MDF", &["FLT0", "FLT1", "FLT2", "FLT3", "FLT4", "FLT5", "FLT6", "FLT7"]),
+        ("PWR", &["S3WU"]),
+        ("GTZC", &["GLOBAL", "ILA"]),
+        ("WWDG", &["GLOBAL", "RST"]),
+        ("USB_OTG_FS", &["GLOBAL", "EP1_OUT", "EP1_IN", "WKUP"]),
+        ("USB_OTG_HS", &["GLOBAL", "EP1_OUT", "EP1_IN", "WKUP"]),
+        ("USB", &["LP", "HP", "WKUP"]),
+    ];
+
+    for (prefix, signals) in IRQ_SIGNALS_MAP {
+        if peri.starts_with(prefix) {
+            return signals.iter().map(ToString::to_string).collect();
+        }
+    }
+    vec!["GLOBAL".to_string()]
+}

src/main.rs

@@ -0,0 +1,102 @@
+mod chips;
+mod dma;
+mod docs;
+mod gpio_af;
+mod header;
+mod interrupts;
+mod memory;
+mod rcc;
+
+#[macro_export]
+macro_rules! regex {
+    ($re:literal) => {{
+        ::ref_thread_local::ref_thread_local! {
+            static managed REGEX: ::regex::Regex = ::regex::Regex::new($re).unwrap();
+        }
+        <REGEX as ::ref_thread_local::RefThreadLocal<::regex::Regex>>::borrow(&REGEX)
+    }};
+}
+
+struct Stopwatch {
+    start: std::time::Instant,
+    section_start: Option<std::time::Instant>,
+}
+
+impl Stopwatch {
+    fn new() -> Self {
+        eprintln!("Starting timer");
+        let start = std::time::Instant::now();
+        Self {
+            start,
+            section_start: None,
+        }
+    }
+
+    fn section(&mut self, status: &str) {
+        let now = std::time::Instant::now();
+        self.print_done(now);
+        eprintln!("  {status}");
+        self.section_start = Some(now);
+    }
+
+    fn stop(self) {
+        let now = std::time::Instant::now();
+        self.print_done(now);
+        let total_elapsed = now - self.start;
+        eprintln!("Total time: {:.2} seconds", total_elapsed.as_secs_f32());
+    }
+
+    fn print_done(&self, now: std::time::Instant) {
+        if let Some(section_start) = self.section_start {
+            let elapsed = now - section_start;
+            eprintln!("    done in {:.2} seconds", elapsed.as_secs_f32());
+        }
+    }
+}
+
+fn main() -> anyhow::Result<()> {
+    let mut stopwatch = Stopwatch::new();
+
+    stopwatch.section("Parsing headers");
+    let headers = header::Headers::parse()?;
+
+    stopwatch.section("Parsing other stuff");
+
+    // stopwatch.section("Parsing memories");
+    let memories = memory::Memories::parse()?;
+
+    // stopwatch.section("Parsing interrupts");
+    let chip_interrupts = interrupts::ChipInterrupts::parse()?;
+
+    // stopwatch.section("Parsing RCC registers");
+    let peripheral_to_clock = rcc::PeripheralToClock::parse()?;
+
+    // stopwatch.section("Parsing docs");
+    let docs = docs::Docs::parse()?;
+
+    // stopwatch.section("Parsing DMA");
+    let dma_channels = dma::DmaChannels::parse()?;
+
+    // stopwatch.section("Parsing GPIO AF");
+    let af = gpio_af::Af::parse()?;
+
+    stopwatch.section("Parsing chip groups");
+    let (chips, chip_groups) = chips::parse_groups()?;
+
+    stopwatch.section("Processing chips");
+    chips::dump_all_chips(
+        chip_groups,
+        headers,
+        af,
+        chip_interrupts,
+        peripheral_to_clock,
+        dma_channels,
+        chips,
+        memories,
+        docs,
+    )?;
+
+    stopwatch.stop();
+
+    Ok(())
+}

src/memory.rs

@@ -0,0 +1,348 @@
+use std::fs;
+
+#[derive(Debug, PartialEq)]
+struct Memory {
+    pub device_id: u16,
+    pub names: Vec<String>,
+    pub ram: Ram,
+    pub flash: Option<Flash>,
+}
+
+#[derive(Clone, Copy, Debug, PartialEq)]
+struct Ram {
+    pub address: u32,
+    pub bytes: u32,
+}
+
+#[derive(Clone, Debug, PartialEq)]
+struct Flash {
+    pub address: u32,
+    pub bytes: u32,
+    pub settings: stm32_data_serde::chip::memory::Settings,
+}
+
+fn splat_names(base: &str, parts: Vec<&str>) -> Vec<String> {
+    let mut names = Vec::new();
+    for part in parts {
+        if part.starts_with("STM32") {
+            names.push(base.to_string());
+        } else if part.starts_with(&base[5..6]) {
+            names.push("STM32".to_string() + part);
+        } else {
+            let diff = base.len() - part.len();
+            names.push((base[..diff]).to_string() + part);
+        }
+    }
+
+    names
+}
+
+fn split_names(str: &str) -> Vec<String> {
+    let mut cleaned = Vec::new();
+    let mut current_base = None;
+    for name in str.split('/') {
+        let name = name.split(' ').next().unwrap().trim();
+        if name.contains('-') {
+            let parts: Vec<_> = name.split('-').collect();
+            current_base = parts.first().map(ToString::to_string);
+            let splatted = splat_names(&current_base.unwrap(), parts);
+            current_base = splatted.first().map(Clone::clone);
+            cleaned.extend(splatted);
+        } else if name.starts_with("STM32") {
+            current_base = Some(name.to_string());
+            cleaned.push(name.to_string())
+        } else if name.starts_with(&current_base.clone().unwrap()[5..6]) {
+            // names.append('STM32' + name)
+            cleaned.push("STM32".to_string() + name);
+        } else {
+            cleaned.push(
+                (current_base.clone().unwrap()[0..(current_base.clone().unwrap().len() - name.len())]).to_string()
+                    + name,
+            )
+        }
+    }
+
+    cleaned
+}
+
+mod xml {
+    use serde::Deserialize;
+
+    pub fn from_hex<'de, T, D>(deserializer: D) -> Result<T, D::Error>
+    where
+        D: serde::Deserializer<'de>,
+        T: num::Num,
+        T::FromStrRadixErr: std::fmt::Display,
+    {
+        use serde::de::Error;
+        let s: &str = Deserialize::deserialize(deserializer)?;
+        let s = s.trim();
+        let (prefix, num) = s.split_at(2);
+        if prefix != "0x" && prefix != "0X" {
+            panic!("no hex prefix");
+        }
+        T::from_str_radix(num, 16).map_err(D::Error::custom)
+    }
+
+    pub fn opt_from_hex<'de, T, D>(deserializer: D) -> Result<Option<T>, D::Error>
+    where
+        D: serde::Deserializer<'de>,
+        T: num::Num,
+        T::FromStrRadixErr: std::fmt::Display,
+    {
+        Ok(Some(from_hex(deserializer)?))
+    }
+
+    #[derive(Debug, Deserialize, PartialEq)]
+    pub struct Root {
+        #[serde(rename = "Device")]
+        pub device: root::Device,
+    }
+
+    mod root {
+        use serde::Deserialize;
+
+        use super::from_hex;
+
+        #[derive(Debug, Deserialize, PartialEq)]
+        pub struct Device {
+            #[serde(rename = "DeviceID", deserialize_with = "from_hex")]
+            pub device_id: u16,
+            #[serde(rename = "Name")]
+            pub name: String,
+            #[serde(rename = "Peripherals")]
+            pub peripherals: device::Peripherals,
+        }
+
+        mod device {
+            use serde::Deserialize;
+
+            #[derive(Debug, Deserialize, PartialEq)]
+            pub struct Peripherals {
+                #[serde(rename = "Peripheral")]
+                pub peripharal: Vec<peripherals::Peripheral>,
+            }
+
+            mod peripherals {
+                use serde::Deserialize;
+
+                use super::super::super::opt_from_hex;
+
+                #[derive(Debug, Deserialize, PartialEq)]
+                pub struct Peripheral {
+                    #[serde(rename = "Name")]
+                    // pub name: peripheral::Name,
+                    pub name: String,
+                    #[serde(rename = "ErasedValue", deserialize_with = "opt_from_hex", default)]
+                    pub erased_value: Option<u8>,
+                    #[serde(rename = "Configuration", default)]
+                    pub configuration: Vec<peripheral::Configuration>,
+                }
+
+                mod peripheral {
+                    use serde::Deserialize;
+
+                    use super::super::super::super::opt_from_hex;
+
+                    #[derive(Debug, Deserialize, PartialEq)]
+                    pub struct Configuration {
+                        #[serde(rename = "Parameters", default)]
+                        pub parameters: Option<configuration::Parameters>,
+                        #[serde(rename = "Allignement", deserialize_with = "opt_from_hex", default)]
+                        pub allignement: Option<u32>,
+                        #[serde(rename = "Bank")]
+                        pub bank: Vec<configuration::Bank>,
+                    }
+
+                    mod configuration {
+                        use serde::Deserialize;
+
+                        use super::super::super::super::super::from_hex;
+
+                        #[derive(Debug, Deserialize, PartialEq)]
+                        pub struct Parameters {
+                            #[serde(deserialize_with = "from_hex")]
+                            pub address: u32,
+                            #[serde(deserialize_with = "from_hex")]
+                            pub size: u32,
+                        }
+
+                        #[derive(Debug, Deserialize, PartialEq)]
+                        pub struct Bank {
+                            #[serde(rename = "Field", default)]
+                            pub field: Vec<bank::Field>,
+                        }
+
+                        mod bank {
+                            use serde::Deserialize;
+
+                            #[derive(Debug, Deserialize, PartialEq)]
+                            pub struct Field {
+                                #[serde(rename = "Parameters")]
+                                pub parameters: super::Parameters,
+                            }
+                        }
+                    }
+                }
+            }
+        }
+    }
+}
+pub struct Memories(Vec<Memory>);
+
+impl Memories {
+    pub fn parse() -> anyhow::Result<Self> {
+        let mut paths: Vec<_> = glob::glob("sources/cubeprogdb/db/*.xml")
+            .unwrap()
+            .map(Result::unwrap)
+            .collect();
+        paths.sort();
+
+        let mut memories = Vec::new();
+
+        for f in paths {
+            // println!("Parsing {f:?}");
+            let file = fs::read_to_string(f)?;
+            let parsed: xml::Root = quick_xml::de::from_str(&file)?;
+            // dbg!(&parsed);
+
+            let device_id = parsed.device.device_id;
+            let names = split_names(&parsed.device.name);
+
+            let mut ram = None;
+            let mut flash = None;
+
+            for peripheral in parsed.device.peripherals.peripharal {
+                if peripheral.name == "Embedded SRAM" && ram.is_none() {
+                    let config = peripheral.configuration.first().unwrap();
+                    let parameters = config.parameters.as_ref().unwrap();
+                    ram = Some(Ram {
+                        address: parameters.address,
+                        bytes: parameters.size,
+                    });
+                }
+
+                if peripheral.name == "Embedded Flash" && flash.is_none() {
+                    let config = peripheral.configuration.first().unwrap();
+                    let parameters = config.parameters.as_ref().unwrap();
+                    let bank = config.bank.first().unwrap();
+                    let erase_size = bank.field.iter().map(|field| field.parameters.size).max().unwrap();
+
+                    flash = Some(Flash {
+                        address: parameters.address,
+                        bytes: parameters.size,
+                        settings: stm32_data_serde::chip::memory::Settings {
+                            erase_value: peripheral.erased_value.unwrap(),
+                            write_size: config.allignement.unwrap(),
+                            erase_size,
+                        },
+                    });
+                }
+            }
+
+            memories.push(Memory {
+                device_id,
+                names,
+                ram: ram.unwrap(),
+                flash,
+            });
+        }
+
+        // The chips below are missing from cubeprogdb
+        memories.push(Memory {
+            device_id: 0,
+            names: vec!["STM32F302xD".to_string()],
+            ram: Ram {
+                address: 0x20000000,
+                bytes: 64 * 1024,
+            },
+            flash: Some(Flash {
+                address: 0x08000000,
+                bytes: 384 * 1024,
+                settings: stm32_data_serde::chip::memory::Settings {
+                    erase_value: 0xFF,
+                    write_size: 8,
+                    erase_size: 2048,
+                },
+            }),
+        });
+
+        memories.push(Memory {
+            device_id: 0,
+            names: vec!["STM32F303xD".to_string()],
+            ram: Ram {
+                address: 0x20000000,
+                bytes: 80 * 1024,
+            },
+            flash: Some(Flash {
+                address: 0x08000000,
+                bytes: 384 * 1024,
+                settings: stm32_data_serde::chip::memory::Settings {
+                    erase_value: 0xFF,
+                    write_size: 8,
+                    erase_size: 2048,
+                },
+            }),
+        });
+
+        memories.push(Memory {
+            device_id: 0,
+            names: vec!["STM32L100x6".to_string()],
+            ram: Ram {
+                address: 0x20000000,
+                bytes: 32 * 1024,
+            },
+            flash: Some(Flash {
+                address: 0x08000000,
+                bytes: 4 * 1024,
+                settings: stm32_data_serde::chip::memory::Settings {
+                    erase_value: 0xFF,
+                    write_size: 4,
+                    erase_size: 256,
+                },
+            }),
+        });
+
+        Ok(Self(memories))
+    }
+
+    fn lookup_chip(&self, chip_name: &str) -> &Memory {
+        for each in &self.0 {
+            for name in &each.names {
+                if is_chip_name_match(name, chip_name) {
+                    return each;
+                }
+            }
+        }
+        panic!("could not find memory information for {chip_name}");
+    }
+
+    pub fn determine_ram_size(&self, chip_name: &str) -> u32 {
+        self.lookup_chip(chip_name).ram.bytes
+    }
+
+    pub fn determine_flash_size(&self, chip_name: &str) -> u32 {
+        self.lookup_chip(chip_name).flash.as_ref().unwrap().bytes
+    }
+
+    pub fn determine_flash_settings(&self, chip_name: &str) -> stm32_data_serde::chip::memory::Settings {
+        self.lookup_chip(chip_name).flash.as_ref().unwrap().settings.clone()
+    }
+
+    pub fn determine_device_id(&self, chip_name: &str) -> u16 {
+        self.lookup_chip(chip_name).device_id
+    }
+}
+
+fn is_chip_name_match(pattern: &str, chip_name: &str) -> bool {
+    let mut chip_name = chip_name.replace("STM32F479", "STM32F469"); // F479 is missing, it's the same as F469.
+    chip_name = chip_name.replace("STM32G050", "STM32G051"); // same...
+    chip_name = chip_name.replace("STM32G060", "STM32G061"); // same...
+    chip_name = chip_name.replace("STM32G070", "STM32G071"); // same...
+    chip_name = chip_name.replace("STM32G0B0", "STM32G0B1"); // same...
+    chip_name = chip_name.replace("STM32G4A", "STM32G49"); // same...
+    chip_name = chip_name.replace("STM32L422", "STM32L412"); // same...
+    chip_name = chip_name.replace("STM32WB30", "STM32WB35"); // same...
+    let pattern = pattern.replace('x', ".");
+    regex::Regex::new(&pattern).unwrap().is_match(&chip_name)
+}

src/rcc.rs

@@ -0,0 +1,92 @@
+use std::collections::HashMap;
+
+use crate::regex;
+
+#[derive(Debug)]
+pub struct PeripheralToClock(
+    HashMap<(String, String, String), HashMap<String, stm32_data_serde::chip::core::peripheral::Rcc>>,
+);
+
+impl PeripheralToClock {
+    pub fn parse() -> anyhow::Result<Self> {
+        let mut peripheral_to_clock = HashMap::new();
+        for f in glob::glob("data/registers/rcc_*")? {
+            let f = f?;
+            let ff = f
+                .file_name()
+                .unwrap()
+                .to_string_lossy()
+                .strip_prefix("rcc_")
+                .unwrap()
+                .strip_suffix(".yaml")
+                .unwrap()
+                .to_string();
+            let mut family_clocks = HashMap::new();
+            let y: chiptool::ir::IR = serde_yaml::from_str(&std::fs::read_to_string(f)?)?;
+            for (reg, body) in &y.fieldsets {
+                let key = format!("fieldset/{reg}");
+                if let Some(m) = regex!(r"^fieldset/((A[PH]B\d?)|GPIO)[LH]?ENR\d?$").captures(&key) {
+                    let clock = m.get(1).unwrap().as_str();
+                    let clock = match clock {
+                        "AHB" => "AHB1",
+                        "APB" => "APB1",
+                        clock => clock,
+                    };
+                    for field in &body.fields {
+                        if let Some(peri) = field.name.strip_suffix("EN") {
+                            // Timers are a bit special, they may have a x2 freq
+                            let peri_clock = {
+                                if regex!(r"^TIM\d+$").is_match(peri) {
+                                    format!("{clock}_TIM")
+                                } else {
+                                    clock.to_string()
+                                }
+                            };
+
+                            let mut reset = None;
+                            if let Some(rstr) = y.fieldsets.get(&reg.replace("ENR", "RSTR")) {
+                                if let Some(_field) =
+                                    rstr.fields.iter().find(|field| field.name == format!("{peri}RST"))
+                                {
+                                    reset = Some(stm32_data_serde::chip::core::peripheral::rcc::Reset {
+                                        register: reg.replace("ENR", "RSTR"),
+                                        field: format!("{peri}RST"),
+                                    });
+                                }
+                            }
+
+                            let res = stm32_data_serde::chip::core::peripheral::Rcc {
+                                clock: peri_clock,
+                                enable: stm32_data_serde::chip::core::peripheral::rcc::Enable {
+                                    register: reg.clone(),
+                                    field: field.name.clone(),
+                                },
+                                reset,
+                            };
+
+                            family_clocks.insert(peri.to_string(), res);
+                        }
+                    }
+                }
+            }
+            peripheral_to_clock.insert(("rcc".to_string(), ff, "RCC".to_string()), family_clocks);
+        }
+
+        Ok(Self(peripheral_to_clock))
+    }
+
+    pub fn match_peri_clock(
+        &self,
+        rcc_block: (String, String, String),
+        peri_name: &str,
+    ) -> Option<&stm32_data_serde::chip::core::peripheral::Rcc> {
+        let clocks = self.0.get(&rcc_block)?;
+        if let Some(res) = clocks.get(peri_name) {
+            Some(res)
+        } else if let Some(peri_name) = peri_name.strip_suffix('1') {
+            self.match_peri_clock(rcc_block, peri_name)
+        } else {
+            None
+        }
+    }
+}

stm32-data-serde/Cargo.toml

@@ -0,0 +1,14 @@
+[package]
+name = "stm32-data-serde"
+version = "0.1.0"
+edition = "2021"
+
+# See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html
+
+[dependencies]
+serde = { version = "1.0.147", features = ["derive"] }
+
+[dev-dependencies]
+itertools = "0.10.5"
+rayon = "1.5.3"
+serde_json = "1.0.87"

stm32-data-serde/src/lib.rs

@@ -0,0 +1,287 @@
+use serde::{Deserialize, Serialize};
+
+#[derive(Clone, Debug, Eq, PartialEq, Hash, PartialOrd, Ord, Serialize, Deserialize)]
+pub struct Chip {
+    pub name: String,
+    pub family: String,
+    pub line: String,
+    pub die: String,
+    pub device_id: u16,
+    pub packages: Vec<chip::Package>,
+    pub memory: Vec<chip::Memory>,
+    pub docs: Vec<chip::Doc>,
+    pub cores: Vec<chip::Core>,
+}
+
+pub mod chip {
+    use serde::{Deserialize, Serialize};
+
+    #[derive(Clone, Debug, Eq, PartialEq, Hash, PartialOrd, Ord, Serialize, Deserialize)]
+    pub struct Package {
+        pub name: String,
+        pub package: String,
+    }
+
+    #[derive(Clone, Debug, Eq, PartialEq, Hash, PartialOrd, Ord, Serialize, Deserialize)]
+    pub struct Memory {
+        pub name: String,
+        pub kind: memory::Kind,
+        pub address: u32,
+        pub size: u32,
+        #[serde(skip_serializing_if = "Option::is_none")]
+        pub settings: Option<memory::Settings>,
+    }
+
+    pub mod memory {
+        use serde::{Deserialize, Serialize};
+
+        #[derive(Clone, Debug, Eq, PartialEq, Hash, PartialOrd, Ord, Serialize, Deserialize)]
+        #[serde(rename_all = "lowercase")]
+        pub enum Kind {
+            Flash,
+            Ram,
+        }
+
+        #[derive(Clone, Debug, Eq, PartialEq, Hash, PartialOrd, Ord, Serialize, Deserialize)]
+        pub struct Settings {
+            pub erase_size: u32,
+            pub write_size: u32,
+            pub erase_value: u8,
+        }
+    }
+
+    #[derive(Clone, Debug, Eq, PartialEq, Hash, PartialOrd, Ord, Serialize, Deserialize)]
+    pub struct Doc {
+        pub r#type: String,
+        pub title: String,
+        pub name: String,
+        pub url: String,
+    }
+
+    #[derive(Clone, Debug, Eq, PartialEq, Hash, PartialOrd, Ord, Serialize, Deserialize)]
+    pub struct Core {
+        pub name: String,
+        pub peripherals: Vec<core::Peripheral>,
+        pub interrupts: Vec<core::Interrupt>,
+        pub dma_channels: Vec<core::DmaChannels>,
+    }
+
+    pub mod core {
+        use serde::{Deserialize, Serialize};
+
+        #[derive(Clone, Debug, Eq, PartialEq, Hash, PartialOrd, Ord, Serialize, Deserialize)]
+        pub struct Peripheral {
+            pub name: String,
+            pub address: u32,
+            #[serde(skip_serializing_if = "Option::is_none")]
+            pub registers: Option<peripheral::Registers>,
+            #[serde(skip_serializing_if = "Option::is_none")]
+            pub rcc: Option<peripheral::Rcc>,
+            #[serde(default, skip_serializing_if = "Vec::is_empty")]
+            pub pins: Vec<peripheral::Pin>,
+            #[serde(skip_serializing_if = "Option::is_none")]
+            pub interrupts: Option<Vec<peripheral::Interrupt>>, // TODO: This should just be a Vec
+            #[serde(default, skip_serializing_if = "Vec::is_empty")]
+            pub dma_channels: Vec<peripheral::DmaChannel>,
+        }
+
+        pub mod peripheral {
+            use serde::{Deserialize, Serialize};
+
+            #[derive(Clone, Debug, Eq, PartialEq, Hash, PartialOrd, Ord, Serialize, Deserialize)]
+            pub struct Registers {
+                pub kind: String,
+                pub version: String,
+                pub block: String,
+            }
+
+            #[derive(Clone, Debug, Eq, PartialEq, Hash, PartialOrd, Ord, Serialize, Deserialize)]
+            pub struct Rcc {
+                pub clock: String,
+                pub enable: rcc::Enable,
+                #[serde(skip_serializing_if = "Option::is_none")]
+                pub reset: Option<rcc::Reset>,
+            }
+
+            pub mod rcc {
+                use serde::{Deserialize, Serialize};
+
+                #[derive(Clone, Debug, Eq, PartialEq, Hash, PartialOrd, Ord, Serialize, Deserialize)]
+                pub struct Enable {
+                    pub register: String,
+                    pub field: String,
+                }
+
+                #[derive(Clone, Debug, Eq, PartialEq, Hash, PartialOrd, Ord, Serialize, Deserialize)]
+                pub struct Reset {
+                    pub register: String,
+                    pub field: String,
+                }
+            }
+
+            #[derive(Clone, Debug, Eq, PartialEq, Hash, PartialOrd, Ord, Serialize, Deserialize)]
+            pub struct Pin {
+                pub pin: pin::Pin,
+                pub signal: String,
+                #[serde(skip_serializing_if = "Option::is_none")]
+                pub af: Option<u8>,
+            }
+
+            pub mod pin {
+                use serde::{Deserialize, Serialize};
+
+                #[derive(Clone, Copy, Debug, Eq, PartialEq, Hash, PartialOrd, Ord)]
+                pub struct Pin {
+                    pub port: char,
+                    pub num: u8,
+                }
+
+                impl Pin {
+                    pub fn parse(pin: &str) -> Option<Self> {
+                        let mut chars = pin.chars();
+                        let p = chars.next()?;
+                        if p != 'P' {
+                            return None;
+                        }
+                        let port = chars.next()?;
+                        let num = chars.as_str().parse().ok()?;
+
+                        Some(Self { port, num })
+                    }
+                }
+
+                impl std::fmt::Display for Pin {
+                    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+                        write!(f, "P{}{}", self.port, self.num)
+                    }
+                }
+
+                impl Serialize for Pin {
+                    fn serialize<S>(&self, serializer: S) -> Result<S::Ok, S::Error>
+                    where
+                        S: serde::Serializer,
+                    {
+                        serializer.serialize_str(&format!("{self}"))
+                    }
+                }
+
+                struct PinVisitor;
+
+                impl<'de> serde::de::Visitor<'de> for PinVisitor {
+                    type Value = Pin;
+
+                    fn expecting(&self, formatter: &mut std::fmt::Formatter) -> std::fmt::Result {
+                        formatter.write_str("pin")
+                    }
+
+                    fn visit_str<E>(self, v: &str) -> Result<Self::Value, E>
+                    where
+                        E: serde::de::Error,
+                    {
+                        Ok(Pin::parse(v).unwrap())
+                    }
+                }
+
+                impl<'de> Deserialize<'de> for Pin {
+                    fn deserialize<D>(deserializer: D) -> Result<Self, D::Error>
+                    where
+                        D: serde::Deserializer<'de>,
+                    {
+                        deserializer.deserialize_str(PinVisitor)
+                    }
+                }
+            }
+
+            #[derive(Clone, Debug, Eq, PartialEq, Hash, PartialOrd, Ord, Serialize, Deserialize)]
+            pub struct Interrupt {
+                pub signal: String,
+                pub interrupt: String,
+            }
+
+            #[derive(Clone, Debug, Eq, PartialEq, Hash, PartialOrd, Ord, Serialize, Deserialize)]
+            pub struct DmaChannel {
+                pub signal: String,
+                #[serde(skip_serializing_if = "Option::is_none")]
+                pub dma: Option<String>,
+                #[serde(skip_serializing_if = "Option::is_none")]
+                pub channel: Option<String>,
+                #[serde(skip_serializing_if = "Option::is_none")]
+                pub dmamux: Option<String>,
+                #[serde(skip_serializing_if = "Option::is_none")]
+                pub request: Option<u8>,
+            }
+        }
+
+        #[derive(Clone, Debug, Eq, PartialEq, Hash, PartialOrd, Ord, Serialize, Deserialize)]
+        pub struct Interrupt {
+            pub name: String,
+            pub number: u8,
+        }
+
+        #[derive(Clone, Debug, Eq, PartialEq, Hash, PartialOrd, Ord, Serialize, Deserialize)]
+        pub struct DmaChannels {
+            pub name: String,
+            pub dma: String,
+            pub channel: u8,
+            #[serde(skip_serializing_if = "Option::is_none")]
+            pub dmamux: Option<String>,
+            #[serde(skip_serializing_if = "Option::is_none")]
+            pub dmamux_channel: Option<u8>,
+            #[serde(skip_serializing_if = "Option::is_none")]
+            pub supports_2d: Option<bool>,
+        }
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use std::borrow::Cow;
+    use std::path::Path;
+    use std::{fs, str};
+
+    use super::*;
+
+    fn normalize_line(line: &str) -> Cow<'_, str> {
+        // The python script saves with 4 spaces instead of 2
+        let line = line.trim_start();
+
+        // The python script escapes unicode
+        let mut line = Cow::Borrowed(line);
+        for symbol in [("\\u00ae", "\u{00ae}"), ("\\u2122", "\u{2122}")] {
+            if line.contains(symbol.0) {
+                line = Cow::Owned(line.replace(symbol.0, symbol.1));
+            }
+        }
+
+        line
+    }
+
+    fn normalize(file: &[u8]) -> impl Iterator<Item = Cow<'_, str>> + '_ {
+        str::from_utf8(file).unwrap().lines().map(normalize_line)
+    }
+
+    fn check_file(path: impl AsRef<Path>) {
+        println!("Checking {:?}", path.as_ref());
+        let original = fs::read(path).unwrap();
+        let parsed: Chip = serde_json::from_slice(&original).unwrap();
+        let reencoded = serde_json::to_vec_pretty(&parsed).unwrap();
+        itertools::assert_equal(normalize(&original), normalize(&reencoded))
+    }
+
+    const CHIPS_DIR: &str = "../data/chips/";
+
+    #[test]
+    fn test_one() {
+        let path = Path::new(CHIPS_DIR).join("STM32F030C6.json");
+        check_file(path);
+    }
+
+    #[test]
+    fn test_all() {
+        use rayon::prelude::*;
+
+        Path::new(CHIPS_DIR).read_dir().unwrap().par_bridge().for_each(|chip| {
+            check_file(chip.unwrap().path());
+        });
+    }
+}

stm32data/header.py

@@ -1,209 +0,0 @@
-import re
-import os
-import json
-from glob import glob
-
-from stm32data import yaml
-from stm32data.util import removeprefix, removesuffix
-
-
-headers_parsed = {}
-header_map = {}
-with open('header_map.yaml', 'r') as f:
-    y = yaml.load(f)
-    for header, chips in y.items():
-        for chip in chips.split(','):
-            header_map[chip.strip().lower()] = header.lower()
-
-
-def get_for_chip(model):
-    if header := get_header_name_for_chip(model):
-        return headers_parsed[header]
-    return None
-
-
-def get_header_name_for_chip(model):
-    # for a, b in header_map:
-    #    model = re.sub(a, b, model, flags=re.IGNORECASE)
-    model = model.lower()
-
-    # if it's in the map, just go
-    if r := header_map.get(model):
-        return r
-
-    # if not, find it by regex, taking `x` meaning `anything`
-    res = []
-    for h in headers_parsed.keys():
-        if re.match('^' + h.replace('x', '.') + '$', model):
-            res.append(h)
-
-    if len(res) == 0:
-        return None
-    assert len(res) == 1
-    return res[0]
-
-
-def paren_ok(val):
-    n = 0
-    for c in val:
-        if c == '(':
-            n += 1
-        if c == ')':
-            n -= 1
-        if n < 0:
-            return False
-    return n == 0
-
-
-# warning: horrible abomination ahead
-
-
-def parse_value(val, defines):
-    val = val.strip()
-    if val == '':
-        return 0
-    if m := re.match('(0([1-9][0-9]*)(U))', val):
-        return int(m.group(2), 10)
-    if m := re.match('((0x[0-9a-fA-F]+|\\d+))(|u|ul|U|UL)$', val):
-        return int(m.group(1), 0)
-    if m := re.match('([0-9A-Za-z_]+)$', val):
-        return defines.get(m.group(1), 0)
-    if m := re.match('\\((.*)\\)$', val):
-        if paren_ok(m.group(1)):
-            return parse_value(m.group(1), defines)
-    if m := re.match('\\*?\\([0-9A-Za-z_]+ *\\*?\\)(.*)$', val):
-        return parse_value(m.group(1), defines)
-    # if m := re.match('\\*?\\(u?int(8|16|32|64)_t\\ *)(.*)$', val):
-    #    return parse_value(m.group(1), defines)
-    if m := re.match('(.*)/(.*)$', val):
-        return parse_value(m.group(1), defines) / parse_value(m.group(2), defines)
-    if m := re.match('(.*)<<(.*)$', val):
-        return (parse_value(m.group(1), defines) << parse_value(m.group(2), defines)) & 0xFFFFFFFF
-    if m := re.match('(.*)>>(.*)$', val):
-        return parse_value(m.group(1), defines) >> parse_value(m.group(2), defines)
-    if m := re.match('(.*)\\|(.*)$', val):
-        return parse_value(m.group(1), defines) | parse_value(m.group(2), defines)
-    if m := re.match('(.*)&(.*)$', val):
-        return parse_value(m.group(1), defines) | parse_value(m.group(2), defines)
-    if m := re.match('~(.*)$', val):
-        return (~parse_value(m.group(1), defines)) & 0xFFFFFFFF
-    if m := re.match('(.*)\\+(.*)$', val):
-        return parse_value(m.group(1), defines) + parse_value(m.group(2), defines)
-    if m := re.match('(.*)-(.*)$', val):
-        return parse_value(m.group(1), defines) - parse_value(m.group(2), defines)
-    raise Exception("can't parse: " + val)
-
-
-def parse_header(f):
-    irqs = {}
-    defines = {}
-    cores = []
-    cur_core = 'all'
-
-    accum = ''
-    for l in open(f, 'r', encoding='utf-8', errors='ignore'):
-        l = l.strip()
-        l = accum + l
-        if l.endswith('\\'):
-            accum = l[:-1]
-            continue
-        accum = ''
-
-        # Scoped by a single core
-        if m := re.match('.*if defined.*CORE_CM(\\d+)(PLUS)?.*', l):
-            cur_core = "cm" + str(m.group(1))
-            if m.group(2) != None:
-                cur_core += "p"
-            # print("Cur core is ", cur_core, "matched", l)
-            found = False
-            for core in cores:
-                if core == cur_core:
-                    found = True
-            if not found:
-                cores.append(cur_core)
-            # print("Switching to core", cur_core, "for", f)
-        elif m := re.match('.*else.*', l):
-            cur_core = "all"
-            if m := re.match('.*else.*CORE_CM(\\d+)(PLUS)?.*', l):
-                cur_core = "cm" + str(m.group(1))
-                if m.group(2) != None:
-                    cur_core += "p"
-                # print("Cur core is ", cur_core, "matched", l)
-            elif len(cores) > 1:
-                # Pick the second core assuming we've already parsed one
-                cur_core = cores[1]
-
-            found = False
-            for core in cores:
-                if core == cur_core:
-                    found = True
-            if not found:
-                cores.append(cur_core)
-            # print("Switching to core", cur_core, "for", f)
-        elif m := re.match('.*endif.*', l):
-            # print("Switching to common core for", f)
-            cur_core = "all"
-
-        if cur_core not in irqs:
-            # print("Registering new core", cur_core)
-            irqs[cur_core] = {}
-        if cur_core not in defines:
-            defines[cur_core] = {}
-
-        if m := re.match('([a-zA-Z0-9_]+)_IRQn += (\\d+),? +/\\*!< (.*) \\*/', l):
-            # print("Found irq for", cur_core)
-            irqs[cur_core][m.group(1)] = int(m.group(2))
-
-        if m := re.match('#define +([0-9A-Za-z_]+)\\(', l):
-            defines[cur_core][m.group(1)] = -1
-        if m := re.match('#define +([0-9A-Za-z_]+) +(.*)', l):
-            name = m.group(1)
-            val = m.group(2)
-            name = name.strip()
-            if name == 'FLASH_SIZE':
-                continue
-            val = val.split('/*')[0].strip()
-            val = parse_value(val, defines[cur_core])
-            # print("Found define for", cur_core)
-            defines[cur_core][name] = val
-
-    # print("Found", len(cores), "cores for", f)
-    # print("Found", len(irqs['all']), "shared interrupts for", f)
-
-    if len(cores) == 0:
-        cores.append("all")
-
-    for core in cores:
-        if core != "all":
-            irqs[core].update(irqs['all'])
-            defines[core].update(defines['all'])
-
-    return {
-        'cores': cores,
-        'interrupts': irqs,
-        'defines': defines,
-    }
-
-
-def parse_headers():
-    os.makedirs('sources/headers_parsed', exist_ok=True)
-    print('loading headers...')
-    for f in glob('sources/headers/*.h'):
-        f = f.replace(os.path.sep, '/')
-        # if 'stm32f4' not in f: continue
-        ff = removeprefix(f, 'sources/headers/')
-        ff = removesuffix(ff, '.h')
-
-        try:
-            with open('sources/headers_parsed/{}.json'.format(ff), 'r') as j:
-                res = json.load(j)
-        except:
-            print(f)
-            res = parse_header(f)
-            with open('sources/headers_parsed/{}.json'.format(ff), 'w') as j:
-                json.dump(res, j)
-
-        headers_parsed[ff] = res
-
-
-parse_headers()

stm32data/interrupts.py

@@ -1,292 +0,0 @@
-from stm32data.util import *
-from glob import glob
-import xmltodict
-import re
-import os
-
-
-chip_interrupts = {}
-
-
-def get(nvic_name, nvic_version, core):
-    return chip_interrupts[(nvic_name, nvic_version)]
-
-
-def parse():
-    print("parsing interrupts")
-    for f in sorted(glob('sources/cubedb/mcu/IP/NVIC*_Modes.xml')):
-        if 'STM32MP1' in f:
-            continue
-        f = f.replace(os.path.sep, '/')
-        ff = removeprefix(f, 'sources/cubedb/mcu/IP/')
-        ff = removesuffix(ff, '_Modes.xml')
-
-        [nvic_name, nvic_version] = ff.split('-')
-
-        irqs = {}
-        r = xmltodict.parse(open(f, 'rb'))
-
-        xml_irqs = next(filter(lambda x: x['@Name'] == 'IRQn', r['IP']['RefParameter']))
-        for irq in xml_irqs['PossibleValue']:
-            value = irq['@Value']
-            parts = value.split(':')
-
-            # Interrupt name
-            name = removesuffix(parts[0], "_IRQn")
-
-            # Fix typo in STM32Lxx and L083 devices
-            if name == "AES_RNG_LPUART1" and "RNG" not in str(parts[1:]):
-                name = "AES_LPUART1"
-
-            if name in irqs:
-                continue
-
-            print(f'{name:25} {nvic_version:12} {nvic_name:5}  {parts[1]:8} {parts[2]:45} {parts[3]:45} {parts[4]:15}')
-            # Flags.
-            # Y
-            #   unknown, it's in all of them
-            # H3, nHS
-            #   ???
-            # 2V, 3V, nV, 2V1
-            #   unknown, it has to do with the fact the irq is shared among N peripehrals
-            # DMA, DMAL0, DMAF0, DMAL0_DMAMUX, DMAF0_DMAMUX
-            #   special format for DMA
-            # DFSDM
-            #   special format for DFSDM
-            # EXTI
-            #   special format for EXTI
-            flags = parts[1].split(',')
-
-            # F100xE MISC_REMAP remaps some DMA IRQs, so ST decided to give two names
-            # to the same IRQ number.
-            if nvic_version == 'STM32F100E' and name == 'DMA2_Channel4_5':
-                continue
-
-            # F3 can remap USB IRQs, ignore them.
-            if nvic_version.startswith('STM32F3') and 'remap' in irq['@Comment']:
-                continue
-
-            signals = set()
-
-            if name in ['NonMaskableInt', 'HardFault', 'MemoryManagement', 'BusFault', 'UsageFault', 'SVCall', 'DebugMonitor', 'PendSV', 'SysTick']:
-                pass
-            elif any(f in flags for f in ['DMA', 'DMAL0', 'DMAF0', 'DMAL0_DMAMUX', 'DMAF0_DMAMUX']):
-                dmas = parts[3].split(',')
-                chans = parts[4].split(';')
-                assert len(dmas) == len(chans)
-                for i in range(len(dmas)):
-                    dma = dmas[i]
-                    if ',' in chans[i]:
-                        ch_from, ch_to = chans[i].split(',')
-                    else:
-                        ch_from = chans[i]
-                        ch_to = chans[i]
-                    ch_from = int(ch_from)
-                    ch_to = int(ch_to)
-                    for ch in range(ch_from, ch_to+1):
-                        signals.add((dma, f'CH{ch}'))
-            elif name == 'DMAMUX1':  # TODO does DMAMUX have more irq signals? seen in U5
-                signals.add(('DMAMUX1', 'OVR'))
-            elif name == 'DMAMUX1_S':  # TODO does DMAMUX have more irq signals? seen in U5
-                signals.add(('DMAMUX1', 'OVR'))
-            elif name == 'DMAMUX_OVR':
-                signals.add(('DMAMUX1', 'OVR'))
-            elif name == 'DMAMUX1_OVR':
-                signals.add(('DMAMUX1', 'OVR'))
-            elif name == 'DMAMUX2_OVR':
-                signals.add(('DMAMUX2', 'OVR'))
-            elif 'DMAMUX' in flags:
-                assert False  # should've been handled above
-            elif 'EXTI' in flags:
-                for signal in parts[2].split(','):
-                    signals.add(('EXTI', signal))
-            elif name == 'FLASH':
-                signals.add(('FLASH', 'GLOBAL'))
-            elif name == 'CRS':
-                signals.add(('RCC', 'CRS'))
-            elif name == 'RCC':
-                signals.add(('RCC', 'GLOBAL'))
-            else:
-                if parts[2] == '':
-                    continue
-
-                peri_names = parts[2].split(',')
-
-                name2 = name
-                if name2 == 'USBWakeUp':
-                    name2 = 'USB_WKUP'
-                if name2 == 'USBWakeUp_RMP':
-                    name2 = 'USB_WKUP'
-                if name2.endswith('_S'):
-                    name2 = removesuffix(name2, '_S')
-
-                peri_signals = {p: [] for p in peri_names}
-
-                curr_peris = None
-                if len(peri_names) == 1:
-                    curr_peris = peri_names
-
-                # Parse IRQ signals from the IRQ name.
-                for part in tokenize_name(name2):
-                    if part == 'TAMPER':
-                        part = 'TAMP'
-
-                    if part == 'LSECSS':
-                        signals.add(('RCC', 'LSECSS'))
-                    elif part == 'CSS':
-                        signals.add(('RCC', 'CSS'))
-                    elif part == 'LSE':
-                        signals.add(('RCC', 'LSE'))
-                    elif part == 'CRS':
-                        signals.add(('RCC', 'CRS'))
-                    elif pp := match_peris(peri_names, part):
-                        curr_peris = pp
-                    else:
-                        assert curr_peris is not None
-                        for p in curr_peris:
-                            peri_signals[p].append(part)
-
-                for p, ss in peri_signals.items():
-                    known = valid_signals(p)
-
-                    # If we have no signals for the peri, assume it's "global" so assign it all known ones
-                    if ss == []:
-                        if p.startswith('COMP'):
-                            ss = ['WKUP']
-                        else:
-                            ss = known
-
-                    for s in ss:
-                        if s not in known:
-                            raise Exception(f'Unknown signal {s} for peri {p}, known={known}')
-                        signals.add((p, s))
-
-            for (peri, signal) in signals:
-                print(f'    {peri}:{signal}')
-            irqs[name] = signals
-
-        irqs2 = {}
-        for name, signals in irqs.items():
-            for (p, s) in signals:
-                irqs2.setdefault(p, []).append({
-                    'signal': s,
-                    'interrupt': name,
-                })
-
-        for p, pirqs in irqs2.items():
-            psirqs = {}
-            for irq in pirqs:
-                psirqs.setdefault(irq['signal'], []).append(irq['interrupt'])
-            for s, irqs in psirqs.items():
-                if len(irqs) != 1:
-                    print(f'DUPE: {p} {s} {irqs}')
-
-        chip_interrupts[(nvic_name, nvic_version)] = irqs2
-
-
-def tokenize_name(name):
-    # Treat IRQ names are "tokens" separated by `_`, except some tokens
-    # contain `_` themselves, such as `C1_RX`.
-    r = re.compile('(SPDIF_RX|EP\d+_(IN|OUT)|OTG_FS|OTG_HS|USB_FS|C1_RX|C1_TX|C2_RX|C2_TX|[A-Z0-9]+(_\d+)*)_*')
-
-    name = name.upper()
-
-    res = []
-    i = 0
-    while i < len(name):
-        m = r.match(name, i)
-        assert m is not None
-        res.append(m.group(1))
-        i = m.end()
-    return res
-
-
-PERI_OVERRIDE = {
-    'USB_FS':   ['USB'],
-    'OTG_HS':   ['USB_OTG_HS'],
-    'OTG_FS':   ['USB_OTG_FS'],
-    'USB':      ['USB_DRD_FS'],
-    'UCPD1_2':  ['UCPD1', 'UCPD2'],
-    'ADC1':     ['ADC'],
-    'CEC':      ['HDMI_CEC'],
-    'SPDIF_RX': ['SPDIFRX1', 'SPDIFRX'],
-    'CAN1':     ['CAN'],
-    'TEMP':     ['TEMPSENS'],
-    'DSI':      ['DSIHOST'],
-    'HRTIM1':   ['HRTIM'],
-    'GTZC':     ['GTZC_S'],
-    'TZIC':     ['GTZC_S'],
-}
-
-
-def match_peris(peris, name):
-    if over := PERI_OVERRIDE.get(name):
-        res = []
-        for p in over:
-            if p in peris:
-                res.append(p)
-        if len(res) != 0:
-            return res
-
-    res = []
-    if m := re.fullmatch('(I2C|[A-Z]+)(\d+(_\d+)*)', name):
-        name = m.group(1)
-        for n in m.group(2).split('_'):
-            p = f'{name}{n}'
-            if p not in peris:
-                return []
-            res.append(p)
-    else:
-        for p in peris:
-            if p == name or (p.startswith(name) and re.fullmatch('\d+', removeprefix(p, name))):
-                res.append(p)
-
-    return res
-
-
-def merge_peri_irq_signals(peri_irqs, additional):
-    for key, value in additional.items():
-        if key not in peri_irqs:
-            peri_irqs[key] = []
-        peri_irqs[key].append(value)
-
-
-irq_signals_map = {
-    'CAN': ['TX', 'RX0', 'RX1', 'SCE'],
-    'FDCAN': ['IT0', 'IT1', 'CAL'],
-    'I2C': ['ER', 'EV'],
-    'FMPI2C': ['ER', 'EV'],
-    'TIM': ['BRK', 'UP', 'TRG', 'COM', 'CC'],
-    'HRTIM': ['Master', 'TIMA', 'TIMB', 'TIMC', 'TIMD', 'TIME', 'TIMF'],
-    'RTC': ['ALARM', 'WKUP', 'TAMP', 'STAMP', 'SSRU'],
-    'SUBGHZ': ['RADIO'],
-    'IPCC': ['C1_RX', 'C1_TX', 'C2_RX', 'C2_TX'],
-    'HRTIM': ['MASTER', 'TIMA', 'TIMB', 'TIMC', 'TIMD', 'TIME', 'TIMF', 'FLT'],
-    'COMP': ['WKUP', 'ACQ'],
-    'RCC': ['RCC', 'CRS'],
-    'MDIOS': ['GLOBAL', 'WKUP'],
-    'ETH': ['GLOBAL', 'WKUP'],
-    'LTDC': ['GLOBAL', 'ER'],
-    'DFSDM': ['FLT0', 'FLT1', 'FLT2', 'FLT3', 'FLT4', 'FLT5', 'FLT6', 'FLT7'],
-    'MDF': ['FLT0', 'FLT1', 'FLT2', 'FLT3', 'FLT4', 'FLT5', 'FLT6', 'FLT7'],
-    'PWR': ['S3WU'],
-    'GTZC': ['GLOBAL', 'ILA'],
-    'WWDG': ['GLOBAL', 'RST'],
-
-    'USB_OTG_FS': ['GLOBAL', 'EP1_OUT', 'EP1_IN', 'WKUP'],
-    'USB_OTG_HS': ['GLOBAL', 'EP1_OUT', 'EP1_IN', 'WKUP'],
-    'USB': ['LP', 'HP', 'WKUP'],
-}
-
-
-def valid_signals(peri):
-    for prefix, signals in irq_signals_map.items():
-        if peri.startswith(prefix):
-            return signals
-    return ['GLOBAL']
-
-
-def filter_interrupts(peri_irqs, all_irqs):
-    return [
-        i for i in peri_irqs if i['interrupt'] in all_irqs
-    ]

stm32data/memory.py

@@ -1,206 +0,0 @@
-import sys
-import re
-import xmltodict
-from glob import glob
-from stm32data.util import *
-
-
-def splat_names(base, parts):
-    names = []
-    for part in parts:
-        if part.startswith("STM32"):
-            names.append(base)
-        elif part.startswith(base[5]):
-            names.append('STM32' + part)
-        else:
-            names.append(base[0: len(base) - len(part)] + part)
-
-    return names
-
-
-def split_names(str):
-    cleaned = []
-    names = str.split("/")
-    current_base = None
-    for name in names:
-        name = name.split(' ')[0].strip()
-        if '-' in name:
-            parts = name.split('-')
-            current_base = parts[0]
-            splatted = splat_names(current_base, parts)
-            current_base = splatted[0]
-            cleaned = cleaned + splatted
-        elif name.startswith("STM32"):
-            current_base = name
-            cleaned.append(name)
-        elif name.startswith(current_base[5]):
-            names.append('STM32' + name)
-        else:
-            cleaned.append(current_base[0: len(current_base) - len(name)] + name)
-    return cleaned
-
-
-memories = []
-
-
-def parse():
-    for f in sorted(glob('sources/cubeprogdb/db/*.xml')):
-        # print("parsing ", f);
-        device = xmltodict.parse(open(f, 'rb'))['Root']['Device']
-        device_id = device['DeviceID']
-        name = device['Name']
-        names = split_names(name)
-        flash_size = None
-        flash_addr = None
-        write_size = None
-        erase_size = None
-        erase_value = None
-        ram_size = None
-        ram_addr = None
-
-        for peripheral in device['Peripherals']['Peripheral']:
-            if peripheral['Name'] == 'Embedded SRAM' and ram_size is None:
-                configs = peripheral['Configuration']
-                if type(configs) != list:
-                    configs = [configs]
-                ram_addr = int(configs[0]['Parameters']['@address'], 16)
-                ram_size = int(configs[0]['Parameters']['@size'], 16)
-                #print( f'ram {addr} {size}')
-            if peripheral['Name'] == 'Embedded Flash' and flash_size is None:
-                configs = peripheral['Configuration']
-                if type(configs) != list:
-                    configs = [configs]
-                flash_addr = int(configs[0]['Parameters']['@address'], 16)
-                flash_size = int(configs[0]['Parameters']['@size'], 16)
-                erase_value = int(peripheral['ErasedValue'], 16)
-                write_size = int(configs[0]['Allignement'], 16)
-                bank = configs[0]['Bank']
-                if type(bank) != list:
-                    bank = [bank]
-                fields = bank[0]['Field']
-                if type(fields) != list:
-                    fields = [fields]
-
-                erase_size = int(fields[0]['Parameters']['@size'], 16)
-                for field in fields:
-                    # print("Field", field)
-                    erase_size = max(erase_size, int(field['Parameters']['@size'], 16))
-                    #print( f'flash {addr} {size}')
-
-        chunk = {
-            'device-id': int(device_id, 16),
-            'names': names,
-        }
-
-        if ram_size is not None:
-            chunk['ram'] = {
-                'address': ram_addr,
-                'bytes': ram_size,
-            }
-
-        if flash_size is not None:
-            chunk['flash'] = {
-                'address': flash_addr,
-                'bytes': flash_size,
-                'erase_value': erase_value,
-                'write_size': write_size,
-                'erase_size': erase_size,
-            }
-
-        memories.append(chunk)
-
-    # The chips below are missing from cubeprogdb
-    memories.append({
-        'device-id': 0,
-        'names': ['STM32F302xD'],
-        'ram': {
-            'address': 0x20000000,
-            'bytes': 64*1024,
-        },
-        'flash': {
-            'address': 0x08000000,
-            'bytes': 384*1024,
-            'erase_value': 0xFF,
-            'write_size': 8,
-            'erase_size': 2048,
-        }
-    })
-    memories.append({
-        'device-id': 0,
-        'names': ['STM32F303xD'],
-        'ram': {
-            'address': 0x20000000,
-            'bytes': 80*1024,
-        },
-        'flash': {
-            'address': 0x08000000,
-            'bytes': 384*1024,
-            'erase_value': 0xFF,
-            'write_size': 8,
-            'erase_size': 2048,
-        }
-    })
-    memories.append({
-        'device-id': 0,
-        'names': ['STM32L100x6'],
-        'ram': {
-            'address': 0x20000000,
-            'bytes': 32*1024,
-        },
-        'flash': {
-            'address': 0x08000000,
-            'bytes': 4*1024,
-            'erase_value': 0xFF,
-            'write_size': 4,
-            'erase_size': 256,
-        }
-    })
-
-
-def determine_ram_size(chip_name):
-    for each in memories:
-        for name in each['names']:
-            if is_chip_name_match(name, chip_name):
-                return each['ram']['bytes']
-    raise Exception(f'could not find ram size for {chip_name}')
-
-
-def determine_flash_size(chip_name):
-    for each in memories:
-        for name in each['names']:
-            if is_chip_name_match(name, chip_name):
-                return each['flash']['bytes']
-    raise Exception(f'could not find flash size for {chip_name}')
-
-
-def determine_flash_settings(chip_name):
-    for each in memories:
-        for name in each['names']:
-            if is_chip_name_match(name, chip_name):
-                return {
-                    'erase_size': each['flash']['erase_size'],
-                    'write_size': each['flash']['write_size'],
-                    'erase_value': each['flash']['erase_value'],
-                }
-    raise Exception(f'could not find flash settings for {chip_name}')
-
-
-def determine_device_id(chip_name):
-    for each in memories:
-        for name in each['names']:
-            if is_chip_name_match(name, chip_name):
-                return each['device-id']
-    return None
-
-
-def is_chip_name_match(pattern, chip_name):
-    chip_name = chip_name.replace("STM32F479", "STM32F469")  # F479 is missing, it's the same as F469.
-    chip_name = chip_name.replace("STM32G050", "STM32G051")  # same...
-    chip_name = chip_name.replace("STM32G060", "STM32G061")  # same...
-    chip_name = chip_name.replace("STM32G070", "STM32G071")  # same...
-    chip_name = chip_name.replace("STM32G0B0", "STM32G0B1")  # same...
-    chip_name = chip_name.replace("STM32G4A", "STM32G49")  # same...
-    chip_name = chip_name.replace("STM32L422", "STM32L412")  # same...
-    chip_name = chip_name.replace("STM32WB30", "STM32WB35")  # same...
-    pattern = pattern.replace('x', '.')
-    return re.match(pattern + ".*", chip_name)

stm32data/util.py

@@ -1,13 +0,0 @@
-
-def removeprefix(value: str, prefix: str) -> str:
-    if value.startswith(prefix):
-        return value[len(prefix):]
-    else:
-        return value[:]
-
-
-def removesuffix(value: str, suffix: str, /) -> str:
-    if value.endswith(suffix):
-        return value[:-len(suffix)]
-    else:
-        return value[:]

stm32data/yaml.py

@@ -1,19 +0,0 @@
-import yaml
-
-try:
-    from yaml import CSafeLoader as SafeLoader
-except ImportError:
-    from yaml import SafeLoader
-
-try:
-    from yaml import CDumper as Dumper
-except ImportError:
-    from yaml import Dumper
-
-
-def load(*args, **kwargs):
-    return yaml.load(*args, Loader=SafeLoader, **kwargs)
-
-
-def dump(*args, **kwargs):
-    return yaml.dump(*args, Dumper=Dumper, sort_keys=False, **kwargs)