mirror of
https://github.com/rhasspy/espeak-ng.git
synced 2026-10-31 13:38:47 +00:00
Add tests for the C-based API.
This commit is contained in:
@@ -16,6 +16,7 @@ data/ucd
|
||||
|
||||
src/libucd.la
|
||||
tests/printucddata
|
||||
tests/printucddata_cpp
|
||||
|
||||
# test output:
|
||||
|
||||
|
||||
+16
-5
@@ -6,6 +6,7 @@ mimedir = $(xdgdatadir)/mime
|
||||
xmldir = $(mimedir)/packages
|
||||
|
||||
AM_LDFLAGS = ${LTLIBINTL}
|
||||
AM_CFLAGS = -Isrc/include -Isrc
|
||||
AM_CXXFLAGS = -Isrc/include -Isrc
|
||||
|
||||
ACLOCAL_AMFLAGS = -I m4
|
||||
@@ -113,19 +114,29 @@ src_libucd_la_SOURCES = \
|
||||
############################# tests ###########################################
|
||||
|
||||
noinst_bin_PROGRAMS += tests/printucddata
|
||||
tests_printucddata_SOURCES = tests/printucddata.cpp
|
||||
tests_printucddata_SOURCES = tests/printucddata.c
|
||||
tests_printucddata_LDADD = src/libucd.la
|
||||
|
||||
noinst_bin_PROGRAMS += tests/printucddata_cpp
|
||||
tests_printucddata_cpp_SOURCES = tests/printucddata_cpp.cpp
|
||||
tests_printucddata_cpp_LDADD = src/libucd.la
|
||||
|
||||
tests/unicode-data.expected: tools/printdata.py tools/ucd.py \
|
||||
data/ucd/UnicodeData.txt \
|
||||
data/ucd/PropList.txt \
|
||||
data/ucd/Scripts.txt
|
||||
tools/printdata.py ${UCD_ROOTDIR} ${UCD_FLAGS} > $@
|
||||
|
||||
tests/unicode-data.actual: tests/printucddata
|
||||
tests/unicode-data-capi.actual: tests/printucddata
|
||||
tests/printucddata > $@
|
||||
|
||||
tests/unicode-data.diff: tests/unicode-data.expected tests/unicode-data.actual
|
||||
diff -U0 tests/unicode-data.expected tests/unicode-data.actual > tests/unicode-data.diff
|
||||
tests/unicode-data-cppapi.actual: tests/printucddata_cpp
|
||||
tests/printucddata_cpp > $@
|
||||
|
||||
check: tests/unicode-data.diff
|
||||
tests/unicode-data-capi.diff: tests/unicode-data.expected tests/unicode-data-capi.actual
|
||||
diff -U0 tests/unicode-data.expected tests/unicode-data-capi.actual > tests/unicode-data-capi.diff
|
||||
|
||||
tests/unicode-data-cppapi.diff: tests/unicode-data.expected tests/unicode-data-cppapi.actual
|
||||
diff -U0 tests/unicode-data.expected tests/unicode-data-cppapi.actual > tests/unicode-data-cppapi.diff
|
||||
|
||||
check: tests/unicode-data-capi.diff tests/unicode-data-cppapi.diff
|
||||
|
||||
@@ -0,0 +1,177 @@
|
||||
/*
|
||||
* Copyright (C) 2012-2016 Reece H. Dunn
|
||||
*
|
||||
* This file is part of ucd-tools.
|
||||
*
|
||||
* ucd-tools is free software: you can redistribute it and/or modify
|
||||
* it under the terms of the GNU General Public License as published by
|
||||
* the Free Software Foundation, either version 3 of the License, or
|
||||
* (at your option) any later version.
|
||||
*
|
||||
* ucd-tools is distributed in the hope that it will be useful,
|
||||
* but WITHOUT ANY WARRANTY; without even the implied warranty of
|
||||
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
||||
* GNU General Public License for more details.
|
||||
*
|
||||
* You should have received a copy of the GNU General Public License
|
||||
* along with ucd-tools. If not, see <http://www.gnu.org/licenses/>.
|
||||
*/
|
||||
|
||||
#include "ucd/ucd.h"
|
||||
|
||||
#include <string.h>
|
||||
#include <stdio.h>
|
||||
|
||||
void fput_utf8c(FILE *out, codepoint_t c)
|
||||
{
|
||||
if (c < 0x80)
|
||||
fputc((uint8_t)c, out);
|
||||
else if (c < 0x800)
|
||||
{
|
||||
fputc(0xC0 | (c >> 6), out);
|
||||
fputc(0x80 + (c & 0x3F), out);
|
||||
}
|
||||
else if (c < 0x10000)
|
||||
{
|
||||
fputc(0xE0 | (c >> 12), out);
|
||||
fputc(0x80 + ((c >> 6) & 0x3F), out);
|
||||
fputc(0x80 + (c & 0x3F), out);
|
||||
}
|
||||
else if (c < 0x200000)
|
||||
{
|
||||
fputc(0xF0 | (c >> 18), out);
|
||||
fputc(0x80 + ((c >> 12) & 0x3F), out);
|
||||
fputc(0x80 + ((c >> 6) & 0x3F), out);
|
||||
fputc(0x80 + (c & 0x3F), out);
|
||||
}
|
||||
}
|
||||
|
||||
int fget_utf8c(FILE *in, codepoint_t *c)
|
||||
{
|
||||
int ch = EOF;
|
||||
if ((ch = fgetc(in)) == EOF) return 0;
|
||||
if ((uint8_t)ch < 0x80)
|
||||
*c = (uint8_t)ch;
|
||||
else switch ((uint8_t)ch & 0xF0)
|
||||
{
|
||||
default:
|
||||
*c = (uint8_t)ch & 0x1F;
|
||||
if ((ch = fgetc(in)) == EOF) return 0;
|
||||
*c = (*c << 6) + ((uint8_t)ch & 0x3F);
|
||||
break;
|
||||
case 0xE0:
|
||||
*c = (uint8_t)ch & 0x0F;
|
||||
if ((ch = fgetc(in)) == EOF) return 0;
|
||||
*c = (*c << 6) + ((uint8_t)ch & 0x3F);
|
||||
if ((ch = fgetc(in)) == EOF) return 0;
|
||||
*c = (*c << 6) + ((uint8_t)ch & 0x3F);
|
||||
break;
|
||||
case 0xF0:
|
||||
*c = (uint8_t)ch & 0x07;
|
||||
if ((ch = fgetc(in)) == EOF) return 0;
|
||||
*c = (*c << 6) + ((uint8_t)ch & 0x3F);
|
||||
if ((ch = fgetc(in)) == EOF) return 0;
|
||||
*c = (*c << 6) + ((uint8_t)ch & 0x3F);
|
||||
if ((ch = fgetc(in)) == EOF) return 0;
|
||||
*c = (*c << 6) + ((uint8_t)ch & 0x3F);
|
||||
break;
|
||||
}
|
||||
return 1;
|
||||
}
|
||||
|
||||
void uprintf_codepoint(FILE *out, codepoint_t c, char mode)
|
||||
{
|
||||
switch (mode)
|
||||
{
|
||||
case 'c': // character
|
||||
switch (c)
|
||||
{
|
||||
case '\t': fputs("\\t", out); break;
|
||||
case '\r': fputs("\\r", out); break;
|
||||
case '\n': fputs("\\n", out); break;
|
||||
default: fput_utf8c(out, c); break;
|
||||
}
|
||||
break;
|
||||
case 'h': // hexadecimal (lower)
|
||||
fprintf(out, "%06x", c);
|
||||
break;
|
||||
case 'H': // hexadecimal (upper)
|
||||
fprintf(out, "%06X", c);
|
||||
break;
|
||||
}
|
||||
}
|
||||
|
||||
void uprintf(FILE *out, codepoint_t c, const char *format)
|
||||
{
|
||||
while (*format) switch (*format)
|
||||
{
|
||||
case '%':
|
||||
switch (*++format)
|
||||
{
|
||||
case 'c': // category
|
||||
fputs(ucd_get_category_string(ucd_lookup_category(c)), out);
|
||||
break;
|
||||
case 'C': // category group
|
||||
fputs(ucd_get_category_group_string(ucd_lookup_category_group(c)), out);
|
||||
break;
|
||||
case 'p': // codepoint
|
||||
uprintf_codepoint(out, c, *++format);
|
||||
break;
|
||||
case 'L': // lowercase
|
||||
uprintf_codepoint(out, ucd_tolower(c), *++format);
|
||||
break;
|
||||
case 's': // script
|
||||
fputs(ucd_get_script_string(ucd_lookup_script(c)), out);
|
||||
break;
|
||||
case 'T': // titlecase
|
||||
uprintf_codepoint(out, ucd_totitle(c), *++format);
|
||||
break;
|
||||
case 'U': // uppercase
|
||||
uprintf_codepoint(out, ucd_toupper(c), *++format);
|
||||
break;
|
||||
case 'W': // whitespace
|
||||
if (ucd_isspace(c))
|
||||
fputs("White_Space", out);
|
||||
break;
|
||||
}
|
||||
++format;
|
||||
break;
|
||||
default:
|
||||
fputc(*format, out);
|
||||
++format;
|
||||
break;
|
||||
}
|
||||
}
|
||||
|
||||
void print_file(FILE *in)
|
||||
{
|
||||
codepoint_t c = 0;
|
||||
while (fget_utf8c(in, &c))
|
||||
uprintf(stdout, c, "%pc\t%pH\t%s\t%c\t%Uc\t%Lc\t%Tc\t%W\n");
|
||||
}
|
||||
|
||||
int main(int argc, char **argv)
|
||||
{
|
||||
if (argc == 2)
|
||||
{
|
||||
if (!strcmp(argv[1], "--stdin") || !strcmp(argv[1], "-"))
|
||||
print_file(stdin);
|
||||
else
|
||||
{
|
||||
FILE *in = fopen(argv[1], "r");
|
||||
if (in)
|
||||
{
|
||||
print_file(in);
|
||||
fclose(in);
|
||||
}
|
||||
else
|
||||
fprintf(stdout, "cannot open `%s`\n", argv[1]);
|
||||
}
|
||||
}
|
||||
else
|
||||
{
|
||||
for (codepoint_t c = 0; c <= 0x10FFFF; ++c)
|
||||
uprintf(stdout, c, "%pH %s %C %c %UH %LH %TH %W\n");
|
||||
}
|
||||
return 0;
|
||||
}
|
||||
@@ -1,5 +1,5 @@
|
||||
/*
|
||||
* Copyright (C) 2012-2015 Reece H. Dunn
|
||||
* Copyright (C) 2012-2016 Reece H. Dunn
|
||||
*
|
||||
* This file is part of ucd-tools.
|
||||
*
|
||||
Reference in New Issue
Block a user