added uchime_src folder. added biom parameter to make.shared. added biom as a current...

[mothur.git] / uchime_src / sfasta.h

diff --git a/uchime_src/sfasta.h b/uchime_src/sfasta.h
new file mode 100644 (file)
index 0000000..ed2f2ff
--- /dev/null
+++ b/uchime_src/sfasta.h
@@ -0,0 +1,93 @@
+#ifndef sfasta_h\r
+#define sfasta_h\r
+\r
+#include "myutils.h"\r
+#include "seq.h"\r
+\r
+typedef void (*ON_START_XSEQ)(const SeqData &SD);\r
+typedef void (*ON_END_XSEQ)(const SeqData &SD);\r
+\r
+// Sequential reader for FASTA file format.\r
+// Serves sequences in file order to save memory.\r
+// Caches biggish chunks to compromise memory vs. speed.\r
+class SFasta\r
+       {\r
+public:\r
+       string m_FileName;\r
+       FILE *m_File;\r
+       bool m_AllowGaps;\r
+\r
+       off_t m_FileSize;\r
+\r
+// Position to start next read\r
+       off_t m_FilePos;\r
+\r
+// Cached data.\r
+       byte *m_Buffer;\r
+\r
+// Bytes allocated to m_Buffer\r
+       unsigned m_BufferSize;\r
+\r
+// Current position in buffer, normally points to '>'\r
+       unsigned m_BufferOffset;\r
+\r
+// File data in buffer <= m_BufferSize\r
+       unsigned m_BufferBytes;\r
+\r
+// Current label\r
+// Points into m_Buffer, not a separate buffer.\r
+       char *m_Label;\r
+\r
+// Current sequence length\r
+       unsigned m_SeqLength;\r
+\r
+// Current seq index\r
+       unsigned m_SeqIndex;\r
+\r
+       unsigned m_ShortestLength;\r
+       unsigned m_LongestLength;\r
+       unsigned m_TooShortCount;\r
+       unsigned m_TooLongCount;\r
+       unsigned m_TooPolyCount;\r
+\r
+private:\r
+       bool m_IsNucleoSet;\r
+       bool m_IsNucleo;\r
+\r
+public:\r
+       SFasta();\r
+       ~SFasta();\r
+\r
+       void Clear();\r
+       void Open(const string &FileName);\r
+       void Rewind();\r
+       bool SetIsNucleo();\r
+       bool GetIsNucleo() const { asserta(m_IsNucleoSet); return m_IsNucleo; };\r
+\r
+// Get next sequence.\r
+// Returns zero on end-of-file\r
+       const byte *GetNextSeq();\r
+\r
+// Get next sequence as SeqData object, return false on end-of-file.\r
+       bool GetNextSD(SeqData &SD);\r
+\r
+// Length of most recent sequence returned by GetNextSeq().\r
+       unsigned GetSeqLength() const { return m_SeqLength; }\r
+\r
+// Label of most recent sequence returned by GetNextSeq().\r
+       const char *GetLabel() const { return m_Label; }\r
+\r
+// Index of most recent sequence returned by GetNextSeq().\r
+       unsigned GetSeqIndex() const { return m_SeqIndex; }\r
+\r
+       unsigned GetPctDoneX10() const;\r
+       double GetPctDone() const;\r
+\r
+       void LogMe() const;\r
+\r
+private:\r
+       void FillCache();\r
+       const byte *GetNextSeqLo();\r
+       };\r
+\r
+#endif // sfasta_h\r