summaryrefslogtreecommitdiff
path: root/klm/util/file_piece.hh
diff options
context:
space:
mode:
authorChris Dyer <redpony@gmail.com>2014-10-13 00:42:37 -0400
committerChris Dyer <redpony@gmail.com>2014-10-13 00:42:37 -0400
commitb1ed81ef3216b212295afa76c5d20a56fb647204 (patch)
tree9633cdc1b8a341dfa58b0b7fec0e2cae44d28835 /klm/util/file_piece.hh
parent1b17f61d359be6e1c3cea29f8c100db3bcdd73a0 (diff)
new kenlm
Diffstat (limited to 'klm/util/file_piece.hh')
-rw-r--r--klm/util/file_piece.hh29
1 files changed, 26 insertions, 3 deletions
diff --git a/klm/util/file_piece.hh b/klm/util/file_piece.hh
index ed3dc5ad..5495ddcc 100644
--- a/klm/util/file_piece.hh
+++ b/klm/util/file_piece.hh
@@ -1,5 +1,5 @@
-#ifndef UTIL_FILE_PIECE__
-#define UTIL_FILE_PIECE__
+#ifndef UTIL_FILE_PIECE_H
+#define UTIL_FILE_PIECE_H
#include "util/ersatz_progress.hh"
#include "util/exception.hh"
@@ -56,10 +56,33 @@ class FilePiece {
return Consume(FindDelimiterOrEOF(delim));
}
+ // Read word until the line or file ends.
+ bool ReadWordSameLine(StringPiece &to, const bool *delim = kSpaces) {
+ assert(delim[static_cast<unsigned char>('\n')]);
+ // Skip non-enter spaces.
+ for (; ; ++position_) {
+ if (position_ == position_end_) {
+ try {
+ Shift();
+ } catch (const util::EndOfFileException &e) { return false; }
+ // And break out at end of file.
+ if (position_ == position_end_) return false;
+ }
+ if (!delim[static_cast<unsigned char>(*position_)]) break;
+ if (*position_ == '\n') return false;
+ }
+ // We can't be at the end of file because there's at least one character open.
+ to = Consume(FindDelimiterOrEOF(delim));
+ return true;
+ }
+
// Unlike ReadDelimited, this includes leading spaces and consumes the delimiter.
// It is similar to getline in that way.
StringPiece ReadLine(char delim = '\n');
+ // Doesn't throw EndOfFileException, just returns false.
+ bool ReadLineOrEOF(StringPiece &to, char delim = '\n');
+
float ReadFloat();
double ReadDouble();
long int ReadLong();
@@ -132,4 +155,4 @@ class FilePiece {
} // namespace util
-#endif // UTIL_FILE_PIECE__
+#endif // UTIL_FILE_PIECE_H