Fix some tests.

[lightning.git] / taglib.go

diff --git a/taglib.go b/taglib.go
index 3dc45b80c08d67772cd4d6930fecf052bc44fce9..e4b25c17de2461beac6321f767e4dbb8331ff4cf 100644 (file)
--- a/taglib.go
+++ b/taglib.go
@@ -1,8 +1,11 @@
-package main
+// Copyright (C) The Lightning Authors. All rights reserved.
+//
+// SPDX-License-Identifier: AGPL-3.0
+
+package lightning
 
 import (
        "bufio"
-       "bytes"
        "fmt"
        "io"
 )
@@ -40,35 +43,72 @@ func (taglib *tagLibrary) Load(rdr io.Reader) error {
        return taglib.setTags(seqs)
 }
 
-func (taglib *tagLibrary) FindAll(buf []byte, fn func(id tagID, pos, taglen int)) {
+func (taglib *tagLibrary) FindAll(in *bufio.Reader, passthrough io.Writer, fn func(id tagID, pos, taglen int)) error {
+       var window = make([]byte, 0, taglib.keylen*1000)
        var key tagmapKey
-       valid := 0 // if valid < taglib.keylen, key has "no data" zeroes that are otherwise indistinguishable from "A"
-       for i, base := range buf {
+       for offset := 0; ; {
+               base, err := in.ReadByte()
+               if err == io.EOF {
+                       return nil
+               } else if err != nil {
+                       return err
+               } else if base == '\r' || base == '\n' {
+                       if buf, err := in.Peek(1); err == nil && len(buf) > 0 && buf[0] == '>' {
+                               return nil
+                       } else if err == io.EOF {
+                               return nil
+                       }
+                       continue
+               } else if base == '>' || base == ' ' {
+                       return fmt.Errorf("unexpected char %q at offset %d in fasta data", base, offset)
+               }
+
+               if passthrough != nil {
+                       if base >= 'A' && base <= 'Z' {
+                               // lowercase for passthrough
+                               base += 'a' - 'A'
+                       }
+                       _, err = passthrough.Write([]byte{base})
+                       if err != nil {
+                               return err
+                       }
+               }
+               offset++
                if !isbase[int(base)] {
-                       valid = 0
+                       // 'N' or various other chars meaning exact
+                       // base not known
+                       window = window[:0]
                        continue
                }
+               window = append(window, base)
+               if len(window) == cap(window) {
+                       copy(window, window[len(window)-taglib.keylen:])
+                       window = window[:taglib.keylen]
+               }
                key = ((key << 2) | twobit[int(base)]) & taglib.keymask
-               valid++
 
-               if valid < taglib.keylen {
+               if len(window) < taglib.keylen {
                        continue
                } else if taginfo, ok := taglib.tagmap[key]; !ok {
                        continue
-               } else if tagstart := i - taglib.keylen + 1; len(taginfo.tagseq) > taglib.keylen && (len(buf) < i+len(taginfo.tagseq) || !bytes.Equal(taginfo.tagseq, buf[i:i+len(taginfo.tagseq)])) {
-                       // key portion matches, but not the entire tag
-                       continue
+               } else if len(taginfo.tagseq) != taglib.keylen {
+                       return fmt.Errorf("assertion failed: len(%q) != keylen %d", taginfo.tagseq, taglib.keylen)
                } else {
-                       fn(taginfo.id, tagstart, len(taginfo.tagseq))
-                       valid = 0 // don't try to match overlapping tags
+                       fn(taginfo.id, offset-taglib.keylen, len(taginfo.tagseq))
+                       window = window[:0] // don't try to match overlapping tags
                }
        }
+       return nil
 }
 
 func (taglib *tagLibrary) Len() int {
        return len(taglib.tagmap)
 }
 
+func (taglib *tagLibrary) TagLen() int {
+       return taglib.keylen
+}
+
 var (
        twobit = func() []tagmapKey {
                r := make([]tagmapKey, 256)
@@ -117,3 +157,17 @@ func (taglib *tagLibrary) setTags(tags [][]byte) error {
        }
        return nil
 }
+
+func (taglib *tagLibrary) Tags() [][]byte {
+       out := make([][]byte, len(taglib.tagmap))
+       untwobit := []byte{'a', 'c', 'g', 't'}
+       for key, info := range taglib.tagmap {
+               seq := make([]byte, taglib.keylen)
+               for i := len(seq) - 1; i >= 0; i-- {
+                       seq[i] = untwobit[int(key)&3]
+                       key = key >> 2
+               }
+               out[int(info.id)] = seq
+       }
+       return out
+}