UCSC TwoBit caching works in main

CMakeLists.txt

@@ -8,7 +8,7 @@ project(fastafs)
 # Do this once in a while - find different bugs
 #set(CMAKE_CXX_COMPILER             "clang++")
 
-set(PROJECT_VERSION "1.1.1")
+set(PROJECT_VERSION "1.2.0")
 set(PACKAGE_URL "https://github.com/yhoogstrate/fastafs")
 set(PACKAGE_BUGREPORT "${PACKAGE_URL}/issues")
 
@@ -103,6 +103,7 @@ endif()
 add_executable(fastafs
 					src/main.cpp
 					src/fasta_to_fastafs.cpp
+					src/ucsc2bit_to_fastafs.cpp
 					src/fastafs.cpp
 					src/twobit_byte.cpp
 					src/database.cpp

Changelog

@@ -1,3 +1,7 @@
+2019-05-07  Youri Hoogstrate
+
+    * Full support for UCSC TwoBit files in `fastafs cache` (adding to db)
+
 2019-04-30  Youri Hoogstrate
 
     * Performance update.

include/utils.hpp

@@ -14,3 +14,5 @@ void sha1_digest_to_hash(unsigned char *, char *);
 
 std::string std_string_nullbyte_safe(char *, size_t, size_t);
 std::string std_string_nullbyte_safe(char *, size_t);
+
+bool is_fasta_file(char *filename);

src/main.cpp

@@ -6,6 +6,7 @@
 
 #include "config.hpp"
 #include "fasta_to_fastafs.hpp"
+#include "ucsc2bit_to_fastafs.hpp"
 #include "database.hpp"
 #include "fuse.hpp"
 
@@ -61,6 +62,12 @@ void usage_check(void)
     std::cout << std::endl;
 }
 
+void usage_cache(void)
+{
+    std::cout << "usage: " << PACKAGE << " cache <fastafs-id> <fasta file | ucsc TwoBit file>\n\n";
+    std::cout << "\n";
+}
+
 int main(int argc, char *argv[])
 {
     if (argc > 1) {
@@ -80,14 +87,15 @@ int main(int argc, char *argv[])
                 database d = database();
                 std::string fname_out = d.add(argv[argc - 2]);
 
-                fasta_to_fastafs(argv[argc - 1], fname_out);
-                //fasta_to_fastafs f = fasta_to_fastafs(std::string(argv[argc - 2]), std::string(argv[argc - 1]));
-                //f.cache();
-                //f.write(fname_out);
-                // @ todo progress bar
+                if(is_fasta_file(argv[argc - 1])) {
+                    fasta_to_fastafs(argv[argc - 1], fname_out);
+                } else {
+                    ucsc2bit_to_fastafs(argv[argc - 1], fname_out);
+                }
+
             } else {
-                std::cout << "usage: " << PACKAGE << " cache <fastafs-id> <fasta file>\n\n";
-                std::cout << "\n";
+                usage_cache();
+                exit(0);
             }
         } else if (strcmp(argv[1], "view") == 0) {
             uint32_t padding = 60;

src/utils.cpp

@@ -152,3 +152,22 @@ std::string std_string_nullbyte_safe(char *ref, size_t len)
     return s;
 }
 
+
+bool is_fasta_file(char *filename)
+{
+    char buf[2];
+    FILE *fp;
+
+    if ((fp = fopen(filename, "rb")) == NULL) {
+        throw std::runtime_error("Could not read first byte of putative FASTA file.");
+
+        return false;
+    }
+
+    if (fread(buf, 1, 2, fp) == 2) {
+        return (buf[0] == '>');// return true if first byte equals >
+    }
+
+    fclose(fp);
+}
+