Test fastxpp implmentations

ishlib/formats/fastxpp_bench.mojo

@@ -0,0 +1,135 @@
+import sys
+from time.time import perf_counter
+from ishlib.vendor.kseq import FastxReader, BufferedReader, KRead
+from ExtraMojo.utils.ir import dump_ir
+
+
+# ── thin wrapper so FileHandle implements KRead ─────────────────────────
+struct FileReader(KRead):
+    var fh: FileHandle
+
+    fn __init__(out self, owned fh: FileHandle):
+        self.fh = fh^
+
+    fn __moveinit__(out self, owned other: Self):
+        self.fh = other.fh^
+
+    fn unbuffered_read[
+        o: MutableOrigin
+    ](mut self, buffer: Span[UInt8, o]) raises -> Int:
+        return Int(self.fh.read(buffer.unsafe_ptr(), len(buffer)))
+
+
+# ────────────────────────────────────────────────────────────────────────
+
+
+fn bench_original(path: String) raises -> (Int, Int, Float64):
+    var fh = open(path, "r")
+    var rdr = FastxReader[read_comment=False](BufferedReader(FileReader(fh^)))
+    var rec = 0
+    var seq = 0
+    var t0 = perf_counter()
+    while rdr.read() > 0:
+        rec += 1
+        seq += len(rdr.seq)
+    return (rec, seq, perf_counter() - t0)
+
+
+fn bench_fastxpp(path: String) raises -> (Int, Int, Float64):
+    var fh = open(path, "r")
+    var rdr = FastxReader[read_comment=False](BufferedReader(FileReader(fh^)))
+    var rec = 0
+    var seq = 0
+    var t0 = perf_counter()
+    while True:
+        var n = rdr.read_fastxpp()
+        if n < 0:
+            break
+        rec += 1
+        seq += n
+    return (rec, seq, perf_counter() - t0)
+
+
+fn bench_fastxpp_bpl(path: String) raises -> (Int, Int, Float64):
+    var fh = open(path, "r")
+    var rdr = FastxReader[read_comment=False](BufferedReader(FileReader(fh^)))
+    var rec = 0
+    var seq = 0
+    var t0 = perf_counter()
+    while True:
+        var n = rdr.read_fastxpp_bpl()
+        if n < 0:
+            break
+        rec += 1
+        seq += n
+    return (rec, seq, perf_counter() - t0)
+
+
+fn bench_fastxpp_swar(path: String) raises -> (Int, Int, Float64):
+    var fh = open(path, "r")
+    var rdr = FastxReader[read_comment=False](BufferedReader(FileReader(fh^)))
+    var rec = 0
+    var seq = 0
+    var t0 = perf_counter()
+    while True:
+        var n = rdr.read_fastxpp_swar()
+        if n < 0:
+            break
+        rec += 1
+        seq += n
+    return (rec, seq, perf_counter() - t0)
+
+
+fn bench_fastxpp_bpl2(path: String) raises -> (Int, Int, Float64):
+    var fh = open(path, "r")
+    var rdr = FastxReader[read_comment=False](BufferedReader(FileReader(fh^)))
+    var rec = 0
+    var seq = 0
+    var t0 = perf_counter()
+    while True:
+        var n = rdr.read_fastxpp_bpl()
+        if n < 0:
+            break
+        rec += 1
+        seq += n
+    return (rec, seq, perf_counter() - t0)
+
+
+fn main() raises:
+    var argv = sys.argv()
+    if len(argv) < 2 or len(argv) > 3:
+        print("Usage: mojo run fastxpp_bench.mojo <file> [orig|fastxpp|bpl]")
+        return
+
+    var path = String(argv[1])
+    var mode: String = "orig"  # default when no flag given
+    if len(argv) == 3:
+        mode = String(argv[2])
+
+    if mode == "orig":
+        r, s, t = bench_original(path)
+        print(
+            "mode=orig          records=", r, "  bases=", s, "  time=", t, "s"
+        )
+    elif mode == "bpl":
+        r, s, t = bench_fastxpp_bpl(path)
+        print(
+            "mode=fastxpp_bpl   records=", r, "  bases=", s, "  time=", t, "s"
+        )
+    elif mode == "fastxpp":
+        r, s, t = bench_fastxpp(path)
+        print(
+            "mode=fastxpp       records=", r, "  bases=", s, "  time=", t, "s"
+        )
+    elif mode == "swar":
+        r, s, t = bench_fastxpp_swar(path)
+        print(
+            "mode=fastxpp_swar   records=", r, "  bases=", s, "  time=", t, "s"
+        )
+    elif mode == "filler":
+        r, s, t = bench_fastxpp_bpl2(path)
+        print(
+            "mode=fastxpp_bpl   records=", r, "  bases=", s, "  time=", t, "s"
+        )
+    else:
+        print("Unknown mode:", mode)

ishlib/formats/generate_fastxpp.mojo

@@ -0,0 +1,183 @@
+import sys
+from collections import Optional
+from ExtraMojo.io.buffered import BufferedReader, BufferedWriter
+from collections import List                # dynamic grow-able buffer
+from memory import Span                     # view into the List for zero-copy writes
+
+# ---------- helpers -------------------------------------------------
+
+
+fn string_count(s: String) -> Int:
+    var n: Int = 0
+    for _ in s.codepoints():
+        n = n + 1
+    return n
+
+
+fn read_line(mut rdr: BufferedReader) raises -> String:
+    var buf = List[UInt8]()
+    var n = rdr.read_until(buf, ord("\n"))
+    if n == 0:
+        return ""
+    var s = String()
+    s.write_bytes(Span(buf))
+    return s
+
+
+# ---------- FASTX++ builder -----------------------------------------
+
+
+fn generate_fastxpp(
+    marker: String,
+    header: String,
+    seq_lines: List[String],
+    qualities: Optional[List[String]] = None,
+) -> String:
+    var bpl = string_count(seq_lines[0]) + 1  # bases + LF
+    var seq_len: Int = 0
+    for i in range(len(seq_lines)):
+        seq_len = seq_len + string_count(seq_lines[i])
+
+    var meta = String(string_count(header)) + ":" + String(
+        seq_len
+    ) + ":" + String(len(seq_lines))
+
+    var rec = marker + "`" + meta + "`" + header + "\n"
+
+    for i in range(len(seq_lines)):
+        rec.write(seq_lines[i], "\n")
+
+    if qualities:
+        var q = qualities.value()
+        rec += "+\n"
+        for i in range(len(q)):
+            rec.write(q[i], "\n")
+
+    return rec
+
+
+fn generate_fastxpp_bpl(
+    marker: String,
+    header: String,
+    seq_lines: List[String],
+    qualities: Optional[List[String]] = None,
+) -> String:
+    var bpl = string_count(seq_lines[0]) + 1          # bases + LF
+    var slen = (bpl - 1) * (len(seq_lines) - 1) +     # (bases per full line)
+               string_count(seq_lines[-1])            # + last (ragged) line
+    var meta = String(string_count(header)) + ":" +
+               String(slen) + ":" +
+               String(len(seq_lines)) + ":" +
+               String(bpl)
+    var rec = marker + "`" + meta + "`" + header + "\n"
+    for i in range(len(seq_lines)):
+        rec.write(seq_lines[i], "\n")
+    if qualities:
+        var q = qualities.value()
+        for i in range(len(q)):
+            rec.write(q[i], "\n")
+    return rec
+
+# Helper: encode an unsigned ≤9-digit value as zero-padded ASCII.
+fn to_ascii_padded(value: Int, width: Int) -> String:
+    # build the decimal text first …
+    var digits = String(value)                       # e.g. "123"
+    var pad     = width - string_count(digits)       # how many zeros needed
+
+    # … then emit into a single pre-sized String
+    var out = String(capacity=width)
+    for _ in range(pad):
+        out.write("0")
+    out.write(digits)                                # concat is zero-copy
+    return out                                       # length == width
+
+fn generate_fastxpp_bpl_fixed(
+        marker: String,
+        header: String,
+        seq_lines: List[String],
+        qualities: Optional[List[String]] = None,
+) -> String:
+
+    # --- numeric fields ------------------------------------------------
+    var bpl  = string_count(seq_lines[0]) + 1                       # incl. LF
+    var slen = (bpl - 1) * (len(seq_lines) - 1) +
+               string_count(seq_lines[-1])
+
+    # --- fixed-width metadata block ------------------------------------
+    var meta = "`" +
+        #to_ascii_padded(string_count(header), 6) +      # hlen
+        to_ascii_padded(slen,                 9) +      # slen
+        to_ascii_padded(len(seq_lines),        7) +      # nlin
+        to_ascii_padded(bpl,                   3) +      # bpl
+        "`"
+
+    # --- assemble record -----------------------------------------------
+    var rec = marker + meta + header + "\n"
+    for i in range(len(seq_lines)):
+        rec.write(seq_lines[i], "\n")
+    if qualities:
+        var q = qualities.value()
+        for i in range(len(q)):
+            rec.write(q[i], "\n")
+    return rec
+
+# ---------- main ----------------------------------------------------
+
+
+fn main() raises:
+    var argv = sys.argv()
+    if len(argv) != 3:
+        print(
+            "Usage: mojo run generate_fastxpp.mojo <input.fastx>"
+            " <output.fastxpp>"
+        )
+        return
+
+    var reader = BufferedReader(
+        open(String(argv[1]), "r"), buffer_capacity=128 * 1024
+    )
+    var writer = BufferedWriter(
+        open(String(argv[2]), "w"), buffer_capacity=128 * 1024
+    )
+
+    var pending_header = String()  # carries a header we already read
+
+    while True:
+        var header_line = pending_header
+        if header_line == "":
+            header_line = read_line(reader)
+        pending_header = String()
+
+        if header_line == "":
+            break
+
+        var marker = String(header_line[0:1])
+        var header = String(header_line[1:])
+
+        var seq = List[String]()
+        var line: String
+
+        while True:
+            line = read_line(reader)
+            if line == "":
+                break
+            if (
+                line.startswith(">")
+                or line.startswith("@")
+                or (marker == "@" and line.startswith("+"))
+            ):
+                pending_header = line  # save for the next record
+                break
+            seq.append(line)
+
+        var qual: Optional[List[String]] = None
+        if marker == "@" and line.startswith("+"):
+            var qlines = List[String]()
+            for _ in range(len(seq)):
+                qlines.append(read_line(reader))
+            qual = Optional[List[String]](qlines)
+
+        writer.write(generate_fastxpp_bpl_fixed(marker, header, seq, qual))
+
+    writer.flush()
+    writer.close()