add header detector

loadsig/loadsig/src/hspice.rs

@@ -1,5 +1,5 @@
 use super::shared::{self, Endian};
-use byteorder::{BigEndian, LittleEndian, NativeEndian, ReadBytesExt};
+use byteorder::ReadBytesExt;
 use std::io::{Read, Seek, SeekFrom};
 use thiserror::Error as TeError;
 
@@ -14,14 +14,20 @@ use thiserror::Error as TeError;
 pub enum Error {
     #[error("IO error {0}")]
     Io(#[from] std::io::Error),
-    #[error("fail to cast value into integral value")]
+    #[error(
+        "fail to cast value into integral value. this usually caused by some value out of range"
+    )]
     CastToInt(#[from] std::num::TryFromIntError),
+    #[error("something was overflow")]
+    Overflow,
     #[error("wrong endian integer {0:#X} in block head. it must be one of 0x4 or 0x4000000")]
     WrongEndianInt(u32),
     #[error("wrong block head value {0:#X}. it must be 0x4")]
     WrongBlockHead(u32),
     #[error("wrong block tail value {0:#X}. it must be equal to head given value {1:#X}")]
     WrongBlockTail(u32, u32),
+    #[error("expected magic words \"$&%#\" is not found at block tail")]
+    NoMagicWords,
 }
 type Result<T> = std::result::Result<T, Error>;
 
@@ -50,14 +56,217 @@ pub struct HSpiceData {}
 #[derive(Debug)]
 struct HspiceBlock {
     block_offset: u64,
-    block_len: usize,
+    block_len: u32,
+}
+impl HspiceBlock {
+    pub fn new(block_offset: u64, block_len: u32) -> Self {
+        Self {
+            block_offset,
+            block_len,
+        }
+    }
+    pub fn compute_offset(&self, relative_offset: u32) -> u64 {
+        if relative_offset >= self.block_len {
+            panic!("relative offset is out of range")
+        } else {
+            self.block_offset + u64::from(relative_offset)
+        }
+    }
 }
 #[derive(Debug)]
 struct HspiceBlocks {
     sim_source: SimSource,
     blocks: Vec<HspiceBlock>,
     header_blk_cnt: usize,
-    data_blk_cnt: usize,
+}
+impl HspiceBlocks {
+    pub fn new(sim_source: SimSource, blocks: Vec<HspiceBlock>, header_blk_cnt: usize) -> Self {
+        Self {
+            sim_source,
+            blocks,
+            header_blk_cnt,
+        }
+    }
+}
+impl<'a> HspiceBlocks {
+    pub fn to_reader<T>(&'a self, reader: T) -> HspiceBlocksReader<'a, T>
+    where
+        T: Read + Seek,
+    {
+        HspiceBlocksReader::new(reader, &self.blocks)
+    }
+    pub fn to_header_reader<T>(&'a self, reader: T) -> HspiceBlocksReader<'a, T>
+    where
+        T: Read + Seek,
+    {
+        HspiceBlocksReader::new(reader, &self.blocks[..self.header_blk_cnt])
+    }
+    pub fn to_data_reader<T>(&'a self, reader: T) -> HspiceBlocksReader<'a, T>
+    where
+        T: Read + Seek,
+    {
+        HspiceBlocksReader::new(reader, &self.blocks[self.header_blk_cnt..])
+    }
+}
+struct HspiceBlocksReader<'a, T>
+where
+    T: Read + Seek,
+{
+    reader: T,
+    blocks: &'a [HspiceBlock],
+
+    /// The full size of this reader
+    total_len: u64,
+    /// The table stored the offset relative to this reader.
+    /// It is convenient for Seek implement.
+    offset_table: Vec<u64>,
+
+    /// The cursor of block index.
+    block_cursor: usize,
+    /// The cursor inside block.
+    cursor: u32,
+}
+impl<'a, T> HspiceBlocksReader<'a, T>
+where
+    T: Read + Seek,
+{
+    pub fn new(reader: T, blocks: &'a [HspiceBlock]) -> Self {
+        // Build total length and offset table.
+        let mut total_len: u64 = 0;
+        let mut offset_table: Vec<u64> = Vec::with_capacity(blocks.len());
+        for block in blocks {
+            offset_table.push(total_len);
+            total_len += u64::from(block.block_len);
+        }
+        // NOTE: This tail is essential for Seek impl.
+        // otherwise Seek impl will panic when SeekFrom::Current.
+        offset_table.push(total_len);
+
+        HspiceBlocksReader {
+            reader,
+            blocks,
+            total_len,
+            offset_table,
+            block_cursor: 0,
+            cursor: 0,
+        }
+    }
+    pub fn into_inner(self) -> T {
+        self.reader
+    }
+}
+impl<'a, T> Read for HspiceBlocksReader<'a, T>
+where
+    T: Read + Seek,
+{
+    fn read(&mut self, mut buf: &mut [u8]) -> std::io::Result<usize> {
+        use std::io::{Error, ErrorKind, Result};
+
+        let old_block_cursor = self.block_cursor;
+        let old_cursor = self.cursor;
+        let mut total_read_count: usize = 0;
+
+        let rv: Result<()> = loop {
+            // Check whether there is space to be filled in buffer.
+            if buf.is_empty() {
+                break Ok(());
+            }
+            // Check whether we reach the tail of block chain.
+            if self.block_cursor >= self.blocks.len() {
+                break Err(Error::new(
+                    ErrorKind::UnexpectedEof,
+                    "reach the tail of blocks",
+                ));
+            }
+            // Check whether we need to switch to next block.
+            if self.cursor >= self.blocks[self.block_cursor].block_len {
+                self.block_cursor += 1;
+                self.cursor = 0;
+                continue;
+            }
+
+            // Decide the read count
+            let remain = self.blocks[self.block_cursor].block_len - self.cursor;
+            let read_count = std::cmp::min(usize::try_from(remain).unwrap(), buf.len());
+            // Read data
+            let offset = self.blocks[self.block_cursor].compute_offset(self.cursor);
+            if let Err(e) = self.reader.seek(SeekFrom::Start(offset)) {
+                break Err(e);
+            }
+            if let Err(e) = self.read_exact(&mut buf[..read_count]) {
+                break Err(e);
+            }
+            // Shift buffer and cursor.
+            buf = &mut buf[read_count..];
+            self.cursor += u32::try_from(read_count).unwrap();
+            // Add count into total counter
+            total_read_count += read_count;
+        };
+
+        // If fail to read, we restore its original states.
+        if let Err(e) = rv {
+            self.block_cursor = old_block_cursor;
+            self.cursor = old_cursor;
+            Err(e)
+        } else {
+            Ok(total_read_count)
+        }
+    }
+}
+impl<'a, T> Seek for HspiceBlocksReader<'a, T>
+where
+    T: Read + Seek,
+{
+    fn seek(&mut self, pos: SeekFrom) -> std::io::Result<u64> {
+        use std::io::{Error, ErrorKind};
+
+        let err_instance = Error::new(ErrorKind::UnexpectedEof, "offset is out of range");
+        let reader_cursor = match pos {
+            SeekFrom::Start(v) => {
+                if v > self.total_len {
+                    return Err(err_instance);
+                } else {
+                    v
+                }
+            }
+            SeekFrom::End(v) => {
+                if v > 0 {
+                    return Err(err_instance);
+                } else {
+                    match self.total_len.checked_add_signed(v) {
+                        Some(v) => v,
+                        None => return Err(err_instance),
+                    }
+                }
+            }
+            SeekFrom::Current(v) => {
+                let current = self.offset_table[self.block_cursor] + u64::from(self.cursor);
+                match current.checked_add_signed(v) {
+                    Some(v) => {
+                        if v > self.total_len {
+                            return Err(err_instance);
+                        } else {
+                            v
+                        }
+                    }
+                    None => return Err(err_instance),
+                }
+            }
+        };
+
+        let block_cursor = match self.offset_table.binary_search(&reader_cursor) {
+            Ok(v) => v,
+            // This unwrap should be safe.
+            // Because there is no possible that every items are greater than given value,
+            // due to that the first item of table must be zero.
+            Err(v) => v.checked_sub(1).unwrap(),
+        };
+        let cursor = reader_cursor - self.offset_table[block_cursor];
+
+        self.block_cursor = block_cursor;
+        self.cursor = u32::try_from(cursor).unwrap();
+        Ok(reader_cursor)
+    }
 }
 
 /// Guess whether given reader is a legal HSPICE signal file.
@@ -70,12 +279,12 @@ where
     T: ReadBytesExt + Seek,
 {
     // Back to the head of file.
-    if let Err(_) = reader.seek(SeekFrom::Start(0)) {
+    if let Err(_) = reader.rewind() {
         return false;
     }
 
     // Check whether first u32 is 4 (in little or big endian)
-    match reader.read_u32::<NativeEndian>() {
+    match shared::read_int_value(Endian::NativeEndian, reader) {
         Ok(val) => match val {
             LITTLE_ENDIAN_INT | BIG_ENDIAN_INT => true,
             _ => false,
@@ -101,10 +310,10 @@ where
     T: ReadBytesExt + Seek,
 {
     // Back to the file head
-    reader.seek(SeekFrom::Start(0))?;
+    reader.rewind()?;
 
     // Fetch the first u32 as little endian to decide the sim source
-    let endian_u32 = reader.read_u32::<LittleEndian>()?;
+    let endian_u32 = shared::read_int_value(Endian::LittleEndian, reader)?;
     let sim_source = match endian_u32 {
         LITTLE_ENDIAN_INT => SimSource::HspiceWin,
         BIG_ENDIAN_INT => SimSource::Hspice,
@@ -114,7 +323,7 @@ where
 
     // Iterate each block
     let mut blocks: Vec<HspiceBlock> = Vec::new();
-    reader.seek(SeekFrom::Start(0))?;
+    reader.rewind()?;
     loop {
         // Read block header.
         // If there is no more data, break this loop.
@@ -135,35 +344,52 @@ where
         }
 
         // Get the offset and length of block body
-        let raw_block_len = block_header[3];
+        let block_offset = reader.stream_position()?;
-        let block_offset = reader.seek(SeekFrom::Current(0))?;
+        let block_len = block_header[3];
-        let block_len = usize::try_from(raw_block_len).unwrap();
 
         // Skip block body to check block tail
-        reader.seek(SeekFrom::Current(i64::try_from(raw_block_len).unwrap()))?;
+        reader.seek(SeekFrom::Current(i64::try_from(block_len).unwrap()))?;
         let block_tail = shared::read_int_value(endian, reader)?;
-        if block_tail != raw_block_len {
+        if block_tail != block_len {
-            return Err(Error::WrongBlockTail(block_tail, raw_block_len));
+            return Err(Error::WrongBlockTail(block_tail, block_len));
         }
 
         // Push result
-        blocks.push(HspiceBlock {
+        blocks.push(HspiceBlock::new(block_offset, block_len));
-            block_offset,
-            block_len,
-        });
 
         // Cursor now at the beginning of next block.
         // We can continue analyze next blocks safely.
     }
 
-    // TODO:
     // Analyze header block count
+    let mut header_blk_cnt: Option<usize> = None;
+    for (i, block) in blocks.iter().enumerate() {
+        let signal_name_len = u32::try_from(SIGNAL_NAME_LEN).unwrap();
+        let offset = if block.block_len < signal_name_len {
+            continue;
+        } else {
+            block.compute_offset(block.block_len - signal_name_len)
+        };
+
+        reader.seek(SeekFrom::Start(offset))?;
+        let mut magic_word = [0u8; MAGIC_WORDS.len()];
+        reader.read_exact(&mut magic_word)?;
+        if &magic_word == MAGIC_WORDS {
+            match i.checked_add(1) {
+                Some(v) => header_blk_cnt = Some(v),
+                None => return Err(Error::Overflow),
+            }
+            break;
+        }
+    }
+    let header_blk_cnt = match header_blk_cnt {
+        Some(v) => v,
+        None => return Err(Error::NoMagicWords),
+    };
 
     // Build result and return.
-    Ok(HspiceBlocks {
+    Ok(HspiceBlocks::new(sim_source, blocks, header_blk_cnt))
-        sim_source,
-        blocks,
-        header_blk_cnt: 0,
-        data_blk_cnt: 0,
-    })
 }
+
+const SIGNAL_NAME_LEN: usize = 8;
+const MAGIC_WORDS: &[u8; 4] = b"$&%#";