tommy/libpostal
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

[phrases] Changing prefix/suffix chars so both are control characters and neither is the NUL-byte. Modifying transliteration special characters accordingly

Browse Source
This commit is contained in:
Al
2015-08-10 16:01:22 -04:00
parent 11a9881988
commit 51572d6575
3 changed files with 30 additions and 17 deletions
Download Patch File Download Diff File Expand all files Collapse all files

12
src/transliterate.h
View File

@@ -100,11 +100,11 @@ typedef struct transliteration_table {
#define WORD_BOUNDARY_CHAR "\x01"
#define WORD_BOUNDARY_CODEPOINT 1
#define WORD_BOUNDARY_CHAR_LEN strlen(WORD_BOUNDARY_CHAR)
#define PRE_CONTEXT_CHAR "\x02"
#define PRE_CONTEXT_CODEPOINT 2
#define PRE_CONTEXT_CHAR "\x86"
#define PRE_CONTEXT_CODEPOINT 134
#define PRE_CONTEXT_CHAR_LEN strlen(PRE_CONTEXT_CHAR)
#define POST_CONTEXT_CHAR "\x03"
#define POST_CONTEXT_CODEPOINT 3
#define POST_CONTEXT_CHAR "\x87"
#define POST_CONTEXT_CODEPOINT 135
#define POST_CONTEXT_CHAR_LEN strlen(POST_CONTEXT_CHAR)
#define EMPTY_TRANSITION_CHAR "\x04"
#define EMPTY_TRANSITION_CODEPOINT 4
@@ -112,8 +112,8 @@ typedef struct transliteration_table {
#define REPEAT_CHAR "\x05"
#define REPEAT_CODEPOINT 5
#define REPEAT_CHAR_LEN strlen(REPEAT_CHAR)
#define GROUP_INDICATOR_CHAR "\x06"
#define GROUP_INDICATOR_CODEPOINT 6
#define GROUP_INDICATOR_CHAR "\x1d"
#define GROUP_INDICATOR_CODEPOINT 29
#define GROUP_INDICATOR_CHAR_LEN strlen(GROUP_INDICATOR_CHAR)
#define BEGIN_SET_CHAR "\x0f"
#define BEGIN_SET_CODEPOINT 15