diff options
author | Sergey Poznyakoff <gray@gnu.org.ua> | 2011-06-04 10:27:59 +0000 |
---|---|---|
committer | Sergey Poznyakoff <gray@gnu.org.ua> | 2011-06-04 10:27:59 +0000 |
commit | ceb837f01112d2cfde96ba9e6ddc9c9ccbd0d0a4 (patch) | |
tree | 316933c27051392c5cd48b873ae0697cd389d52a /src/ellinika | |
parent | 99076de629a6f5f2b654118cde3612f9ba05edf0 (diff) | |
download | ellinika-ceb837f01112d2cfde96ba9e6ddc9c9ccbd0d0a4.tar.gz ellinika-ceb837f01112d2cfde96ba9e6ddc9c9ccbd0d0a4.tar.bz2 |
Implement new morphological functions. Move elmorph to scm/ellinika
git-svn-id: file:///home/puszcza/svnroot/ellinika/trunk@554 941c8c0f-9102-463b-b60b-cd22ce0e6858
Diffstat (limited to 'src/ellinika')
-rw-r--r-- | src/ellinika/Makefile.am | 44 | ||||
-rw-r--r-- | src/ellinika/aorist.c | 73 | ||||
-rw-r--r-- | src/ellinika/elchr.c | 701 | ||||
-rw-r--r-- | src/ellinika/elmorph.c | 655 | ||||
-rw-r--r-- | src/ellinika/elmorph.h | 46 | ||||
-rw-r--r-- | src/ellinika/elmorph.scm4 | 5 | ||||
-rw-r--r-- | src/ellinika/utf8.c | 2149 | ||||
-rw-r--r-- | src/ellinika/utf8.h | 71 |
8 files changed, 3742 insertions, 2 deletions
diff --git a/src/ellinika/Makefile.am b/src/ellinika/Makefile.am index 136b44f..274eea8 100644 --- a/src/ellinika/Makefile.am +++ b/src/ellinika/Makefile.am @@ -15,7 +15,7 @@ # along with this program. If not, see <http://www.gnu.org/licenses/>. guiledir=$(GUILE_SITE)/$(PACKAGE) -guile_DATA=xlat.scm cgi.scm i18n.scm config.scm dico.scm +guile_DATA=xlat.scm cgi.scm i18n.scm config.scm dico.scm elmorph.scm cgi.m4: Makefile echo 'divert(-1)' > $@ @@ -31,13 +31,53 @@ cgi.m4: Makefile echo 'define([SYSCONFDIR],$(sysconfdir))' >> $@ echo 'define([LOCALEDIR],$(datadir)/locale)' >> $@ echo 'define([HTMLDIR],$(HTMLDIR))' >> $@ + echo 'define([VERSION],$(VERSION))' >> $@ + echo 'define([LIBDIR],$(pkglibdir))' >> $@ echo 'divert(0)dnl' >> $@ echo '@AUTOGENERATED@' >> $@ -SUFFIXES = .scm4 .scm +SUFFIXES = .scm4 .scm .x .scm4.scm: m4 cgi.m4 $< > $@ cgi.scm: cgi.scm4 cgi.m4 config.scm: config.scm4 cgi.m4 +elmorph.scm: elmorph.scm4 cgi.m4 + +pkglib_LTLIBRARIES=libelmorph.la + +libelmorph_la_SOURCES = \ + aorist.c\ + utf8.c\ + elchr.c\ + elmorph.c\ + elmorph.h + +DOT_X_FILES = elmorph.x + +BUILT_SOURCES = $(DOT_X_FILES) + +DISTCLEANFILES = $(DOT_X_FILES) + +snarfcppopts = $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(AM_CFLAGS) $(CFLAGS) + +.c.x: + AWK=$(AWK) \ + guile-snarf -o $@ $< $(snarfcppopts) + +pkglibnames=elmorph + +install-data-hook: + here=`pwd`; \ + cd $(DESTDIR)$(pkglibdir);\ + for name in $(pkglibnames); do \ + if test -f lib$$name.la; then \ + dlname=`sed -n 's/dlname='\''\(.*\)'\''/\1/p' lib$$name.la`; \ + test -z "$$dlname" && dlname='lib$$name.so'; \ + $(LN_S) -f "$$dlname" libguile-$$name-v-$(VERSION).so; \ + fi; \ + done; \ + cd $$here + + diff --git a/src/ellinika/aorist.c b/src/ellinika/aorist.c new file mode 100644 index 0000000..995fce8 --- /dev/null +++ b/src/ellinika/aorist.c @@ -0,0 +1,73 @@ +#ifdef HAVE_CONFIG_H +# include <config.h> +#endif +#include <errno.h> +#include <stdlib.h> +#include <libguile.h> +#include "utf8.h" +#include "elmorph.h" + +int +elmorph_thema_aoristoy(unsigned *word, size_t len, + unsigned **thema, size_t *tlen) +{ + unsigned ch, *pw; + + switch (word[len-1]) { + case 0x03B6: /* ζ */ + /* FIXME: This can produce ξ as well: αλλάζω => άλλαξα */ + case 0x03B8: /* θ */ + ch = 0x03C3; /* σ */ + break; + + case 0x03B3: /* γ */ + case 0x03C7: /* χ */ + ch = 0x03BE; /* ξ */ + break; + + case 0x03BA: /* κ */ + if (len > 1 && word[len-2] == 0x03C3 /* σκ */) + len--; + ch = 0x03BE; /* ξ */ + break; + + case 0x03BD: /* ν */ + if (len > 1 && word[len-2] == 0x03C7 /* χν */) { + len--; + ch = 0x03BE; /* ξ */ + } else + ch = 0x03C3; /* σ */ + break; + + case 0x03B2: /* β */ + case 0x03C0: /* π */ + case 0x03C6: /* φ */ + ch = 0x03C8; /* ψ */ + break; + + case 0x03CD: /* ύ */ + case 0x03C5: /* υ FIXME: This assumes the word has been deaccentized */ + if (len > 1 && (word[len-2] == 0x03B1 /* αύ */ || + word[len-2] == 0x03B5 /* εύ */)) { + ch = 0x03C8; /* ψ */ + break; + } + + default: + len++; + ch = 0x03C3; /* σ */ + } + + pw = calloc(len, sizeof(pw[0])); + if (!pw) + return -1; + memcpy(pw, word, sizeof(word[0]) * (len - 1)); + pw[len-1] = ch; + + *thema = pw; + *tlen = len; + return 0; +} + + + diff --git a/src/ellinika/elchr.c b/src/ellinika/elchr.c new file mode 100644 index 0000000..9b4e7ad --- /dev/null +++ b/src/ellinika/elchr.c @@ -0,0 +1,701 @@ +#ifdef HAVE_CONFIG_H +# include <config.h> +#endif +#include <errno.h> +#include <stdlib.h> +#include <libguile.h> +#include "utf8.h" +#include "elmorph.h" + +struct char_info_st { + unsigned ch; /* Characters */ + int flags; /* Flags (see above) */ + unsigned base; /* for vowels - a corresponding vowel with all diacritics + removed */ + unsigned trans; /* a counter-case equivalent, i.e. a corresponding uppercase + letter if flags & CHF_LOWER and a corresponding lowerrcase + letter if flags & CHF_UPPER */ + unsigned numval; /* Numeric value */ + unsigned accented[3]; /* For vowels - corresponding accented variant */ + unsigned deaccent; /* For accented vowels with diaeresis - corresponding + non-accented character */ +}; + +/* See http://www.unicode.org/charts/PDF/Unicode-5.1/U51-0370.pdf */ +struct char_info_st el_basic_ctype[] = { + { 0x0300, }, + { 0x0301, }, + { 0x0302, }, + { 0x0303, }, + { 0x0304, }, + { 0x0305, }, + { 0x0306, }, + { 0x0307, }, + { 0x0308, }, + { 0x0309, }, + { 0x030A, }, + { 0x030B, }, + { 0x030C, }, + { 0x030D, }, + { 0x030E, }, + { 0x030F, }, + { 0x0310, }, + { 0x0311, }, + { 0x0312, }, + { 0x0313, }, + { 0x0314, }, + { 0x0315, }, + { 0x0316, }, + { 0x0317, }, + { 0x0318, }, + { 0x0319, }, + { 0x031A, }, + { 0x031B, }, + { 0x031C, }, + { 0x031D, }, + { 0x031E, }, + { 0x031F, }, + { 0x0320, }, + { 0x0321, }, + { 0x0322, }, + { 0x0323, }, + { 0x0324, }, + { 0x0325, }, + { 0x0326, }, + { 0x0327, }, + { 0x0328, }, + { 0x0329, }, + { 0x032A, }, + { 0x032B, }, + { 0x032C, }, + { 0x032D, }, + { 0x032E, }, + { 0x032F, }, + { 0x0330, }, + { 0x0331, }, + { 0x0332, }, + { 0x0333, }, + { 0x0334, }, + { 0x0335, }, + { 0x0336, }, + { 0x0337, }, + { 0x0338, }, + { 0x0339, }, + { 0x033A, }, + { 0x033B, }, + { 0x033C, }, + { 0x033D, }, + { 0x033E, }, + { 0x033F, }, + { 0x0340, }, + { 0x0341, }, + { 0x0342, }, + { 0x0343, }, + { 0x0344, }, + { 0x0345, }, + { 0x0346, }, + { 0x0347, }, + { 0x0348, }, + { 0x0349, }, + { 0x034A, }, + { 0x034B, }, + { 0x034C, }, + { 0x034D, }, + { 0x034E, }, + { 0x034F, }, + { 0x0350, }, + { 0x0351, }, + { 0x0352, }, + { 0x0353, }, + { 0x0354, }, + { 0x0355, }, + { 0x0356, }, + { 0x0357, }, + { 0x0358, }, + { 0x0359, }, + { 0x035A, }, + { 0x035B, }, + { 0x035C, }, + { 0x035D, }, + { 0x035E, }, + { 0x035F, }, + { 0x0360, }, + { 0x0361, }, + { 0x0362, }, + { 0x0363, }, + { 0x0364, }, + { 0x0365, }, + { 0x0366, }, + { 0x0367, }, + { 0x0368, }, + { 0x0369, }, + { 0x036A, }, + { 0x036B, }, + { 0x036C, }, + { 0x036D, }, + { 0x036E, }, + { 0x036F, }, + { 0x0370, CHF_ARCHAIC|CHF_CONSONANT|CHF_UPPER, 0, 0x0371 }, /* CAPITAL HETTA */ + { 0x0371, CHF_ARCHAIC|CHF_CONSONANT|CHF_LOWER, 0, 0x0370 }, /* SMALL HETA */ + { 0x0372, CHF_ARCHAIC|CHF_CONSONANT|CHF_UPPER, 0, 0x0373 }, /* CAPITAL SAMPI */ + { 0x0373, CHF_ARCHAIC|CHF_CONSONANT|CHF_LOWER, 0, 0x0372 }, /* SMALL SAMPI */ + { 0x0374, CHF_MODIFIER|CHF_UPPER, 0, 0x0375 }, /* NUMERAL SIGN = dexia keraia */ + { 0x0375, CHF_MODIFIER|CHF_LOWER, 0, 0x0374 }, /* aristeri keraia */ + { 0x0376, CHF_ARCHAIC|CHF_SEMIVOWEL|CHF_UPPER, 0, 0x0377}, /* CAPITAL PAMPHYLIAN DIGAMMA */ + { 0x0377, CHF_ARCHAIC|CHF_SEMIVOWEL|CHF_LOWER, 0, 0x0376}, /* SMALL PAMPHYLIAN DIGAMMA */ + { 0x0378, }, + { 0x0379, }, + { 0x037A, CHF_ARCHAIC|CHF_MODIFIER }, /* YPOGEGRAMMENI */ + { 0x037B, CHF_SYMBOL, 0, 0x03FD }, /* SMALL REVERSED LUNATE SIGMA */ + { 0x037C, CHF_SYMBOL, 0, 0x03FE }, /* SMALL DOTTED LUNATE SIGMA */ + { 0x037D, CHF_SYMBOL, 0, 0x03FF }, /* SMALL REVERSED DOTTED LUNATE SIGMA */ + { 0x037E, CHF_PUNCT }, /* erotimatiko */ + { 0x037F, }, + { 0x0380, }, + { 0x0381, }, + { 0x0382, }, + { 0x0383, }, + { 0x0384, CHF_MODIFIER }, /* Oxeia */ + { 0x0385, CHF_MODIFIER }, /* dialytika */ + { 0x0386, CHF_VOWEL|CHF_UPPER|CHF_OXEIA, 0x0391, 0x03AC }, /* Ά */ + { 0x0387, CHF_PUNCT }, /* ano teleia */ + { 0x0388, CHF_VOWEL|CHF_UPPER|CHF_OXEIA, 0x0395, 0x03AD }, /* Έ */ + { 0x0389, CHF_VOWEL|CHF_UPPER|CHF_OXEIA|CHF_DIPH2, 0x0397, 0x03AE }, /* Ή */ + { 0x038A, CHF_VOWEL|CHF_UPPER|CHF_OXEIA|CHF_DIPH2, 0x0399, 0x03AF }, /* Ί */ + { 0x038B, }, + { 0x038C, CHF_VOWEL|CHF_UPPER|CHF_OXEIA, 0x039F, 0x03CC }, /* Ό */ + { 0x038D, }, + { 0x038E, CHF_VOWEL|CHF_UPPER|CHF_OXEIA|CHF_DIPH2, 0x03A5, 0x03CD }, /* Ύ */ + { 0x038F, CHF_VOWEL|CHF_UPPER|CHF_OXEIA, 0x03A9, 0x03CE }, /* Ώ */ + { 0x0390, CHF_VOWEL|CHF_LOWER|CHF_TREMA|CHF_OXEIA, 0x03B9, 0, 0, 0, 0, 0x03CA }, /* ΐ */ + { 0x0391, CHF_VOWEL|CHF_UPPER|CHF_NUMERIC|CHF_DIPH1, 0, 0x03B1, 1, 0x0386 }, /* Α */ + { 0x0392, CHF_CONSONANT|CHF_UPPER|CHF_NUMERIC, 0, 0x03B2, 2 }, /* Β */ + { 0x0393, CHF_CONSONANT|CHF_UPPER|CHF_NUMERIC, 0, 0x03B3, 3 }, /* Γ */ + { 0x0394, CHF_CONSONANT|CHF_UPPER|CHF_NUMERIC, 0, 0x03B4, 4 }, /* Δ */ + { 0x0395, CHF_VOWEL|CHF_UPPER|CHF_NUMERIC|CHF_DIPH1, 0, 0x03B5, 5, 0x0388 }, /* Ε */ + { 0x0396, CHF_CONSONANT|CHF_UPPER|CHF_NUMERIC, 0, 0x03B6, 7 }, /* Ζ */ + { 0x0397, CHF_VOWEL|CHF_UPPER|CHF_NUMERIC, 0, 0x03B7, 8, 0x0389 }, /* Η */ + { 0x0398, CHF_CONSONANT|CHF_UPPER|CHF_NUMERIC, 0, 0x03B8, 9 }, /* Θ */ + { 0x0399, CHF_VOWEL|CHF_UPPER|CHF_NUMERIC|CHF_DIPH2, 0, 0x03B9, 10, 0x038A }, /* Ι */ + { 0x039A, CHF_CONSONANT|CHF_UPPER|CHF_NUMERIC, 0, 0x03BA, 20 }, /* Κ */ + { 0x039B, CHF_CONSONANT|CHF_UPPER|CHF_NUMERIC, 0, 0x03BB, 30 }, /* Λ */ + { 0x039C, CHF_CONSONANT|CHF_UPPER|CHF_NUMERIC, 0, 0x03BC, 40 }, /* Μ */ + { 0x039D, CHF_CONSONANT|CHF_UPPER|CHF_NUMERIC, 0, 0x03BD, 50 }, /* Ν */ + { 0x039E, CHF_CONSONANT|CHF_UPPER|CHF_NUMERIC, 0, 0x03BE, 60 }, /* Ξ */ + { 0x039F, CHF_VOWEL|CHF_UPPER|CHF_NUMERIC, 0, 0x03BF, 70, 0x038C }, /* Ο */ + { 0x03A0, CHF_CONSONANT|CHF_UPPER|CHF_NUMERIC, 0, 0x03C0, 80 }, /* Π */ + { 0x03A1, CHF_CONSONANT|CHF_UPPER|CHF_NUMERIC, 0, 0x03C1, 100 }, /* Ρ */ + { 0x03A2, }, + { 0x03A3, CHF_CONSONANT|CHF_UPPER|CHF_NUMERIC, 0, 0x03C3, 200 }, /* Σ */ + { 0x03A4, CHF_CONSONANT|CHF_UPPER|CHF_NUMERIC, 0, 0x03C4, 300 }, /* Τ */ + { 0x03A5, CHF_VOWEL|CHF_UPPER|CHF_NUMERIC|CHF_DIPH1|CHF_DIPH2, 0, 0x03C5, 400, 0x038E }, /* Υ */ + { 0x03A6, CHF_CONSONANT|CHF_UPPER|CHF_NUMERIC, 0, 0x03C6, 500 }, /* Φ */ + { 0x03A7, CHF_CONSONANT|CHF_UPPER|CHF_NUMERIC, 0, 0x03C7, 600 }, /* Χ */ + { 0x03A8, CHF_CONSONANT|CHF_UPPER|CHF_NUMERIC, 0, 0x03C8, 700 }, /* Ψ */ + { 0x03A9, CHF_VOWEL|CHF_UPPER|CHF_NUMERIC, 0, 0x03C9, 800, 0x038F }, /* Ω */ + { 0x03AA, CHF_VOWEL|CHF_UPPER|CHF_TREMA|CHF_DIPH2, 0x0399, 0x03CA }, /* Ϊ */ + { 0x03AB, CHF_VOWEL|CHF_UPPER|CHF_TREMA, 0x03A5, 0x03CB }, /* Ϋ */ + { 0x03AC, CHF_VOWEL|CHF_LOWER|CHF_OXEIA, 0x03B1, 0x0386 }, /* ά */ + { 0x03AD, CHF_VOWEL|CHF_LOWER|CHF_OXEIA, 0x03B4, 0x0388 }, /* έ */ + { 0x03AE, CHF_VOWEL|CHF_LOWER|CHF_OXEIA|CHF_DIPH2, 0x03B7, 0x0389 }, /* ή */ + { 0x03AF, CHF_VOWEL|CHF_LOWER|CHF_OXEIA|CHF_DIPH2, 0x03B9, 0x038A }, /* ί */ + { 0x03B0, CHF_VOWEL|CHF_OXEIA|CHF_TREMA, 0x03C5, 0, 0, 0, 0, 0x03CB }, /* ΰ */ + { 0x03B1, CHF_VOWEL|CHF_LOWER|CHF_NUMERIC|CHF_DIPH1, 0, 0x0391, 1, 0x03AC }, /* α */ + { 0x03B2, CHF_CONSONANT|CHF_LOWER|CHF_NUMERIC, 0, 0x0392, 2 }, /* β */ + { 0x03B3, CHF_CONSONANT|CHF_LOWER|CHF_NUMERIC, 0, 0x0393, 3 }, /* γ */ + { 0x03B4, CHF_CONSONANT|CHF_LOWER|CHF_NUMERIC, 0, 0x0394, 4 }, /* δ */ + { 0x03B5, CHF_CONSONANT|CHF_VOWEL|CHF_LOWER|CHF_NUMERIC|CHF_DIPH1, 0, 0x0395, 5, 0x03AD }, /* ε */ + { 0x03B6, CHF_CONSONANT|CHF_LOWER|CHF_NUMERIC, 0, 0x0396, 7 }, /* ζ */ + { 0x03B7, CHF_VOWEL|CHF_LOWER|CHF_NUMERIC|CHF_DIPH1|CHF_DIPH2, 0, 0x0397, 8, 0x03AE }, /* η */ + { 0x03B8, CHF_CONSONANT|CHF_LOWER|CHF_NUMERIC, 0, 0x0398, 9 }, /* θ */ + { 0x03B9, CHF_VOWEL|CHF_LOWER|CHF_NUMERIC, 0, 0x0399, 10, 0x03AF }, /* ι */ + { 0x03BA, CHF_CONSONANT|CHF_LOWER|CHF_NUMERIC, 0, 0x039A, 20 }, /* κ */ + { 0x03BB, CHF_CONSONANT|CHF_LOWER|CHF_NUMERIC, 0, 0x039B, 30 }, /* λ */ + { 0x03BC, CHF_CONSONANT|CHF_LOWER|CHF_NUMERIC, 0, 0x039C, 40 }, /* μ */ + { 0x03BD, CHF_CONSONANT|CHF_LOWER|CHF_NUMERIC, 0, 0x039D, 50 }, /* ν */ + { 0x03BE, CHF_CONSONANT|CHF_LOWER|CHF_NUMERIC, 0, 0x039E, 60 }, /* ξ */ + + { 0x03BF, CHF_VOWEL|CHF_LOWER|CHF_NUMERIC, 0, 0x039F, 70, 0x03CC }, /* ο */ + { 0x03C0, CHF_CONSONANT|CHF_LOWER|CHF_NUMERIC, 0, 0x03A0, 80 }, /* π */ + { 0x03C1, CHF_CONSONANT|CHF_LOWER|CHF_NUMERIC, 0, 0x03A1, 100 }, /* ρ */ + { 0x03C2, CHF_CONSONANT|CHF_LOWER, 0, 0x03A3 }, /* ς */ + { 0x03C3, CHF_CONSONANT|CHF_LOWER|CHF_NUMERIC, 0, 0x03A3, 200 }, /* σ */ + { 0x03C4, CHF_CONSONANT|CHF_LOWER|CHF_NUMERIC, 0, 0x03A4, 300 }, /* τ */ + { 0x03C5, CHF_VOWEL|CHF_LOWER|CHF_NUMERIC|CHF_DIPH2, 0, 0x03A5, 400, 0x03CD }, /* υ */ + { 0x03C6, CHF_CONSONANT|CHF_LOWER|CHF_NUMERIC, 0, 0x03A6, 500 }, /* φ */ + { 0x03C7, CHF_CONSONANT|CHF_LOWER|CHF_NUMERIC, 0, 0x03A7, 600 }, /* χ */ + { 0x03C8, CHF_CONSONANT|CHF_LOWER|CHF_NUMERIC, 0, 0x03A8, 700 }, /* ψ */ + { 0x03C9, CHF_VOWEL|CHF_LOWER|CHF_NUMERIC, 0, 0x03A9, 800, 0x03CE }, /* ω */ + { 0x03CA, CHF_VOWEL|CHF_LOWER|CHF_TREMA|CHF_DIPH2, 0x03B9, 0x03AA, 0, 0x0390 }, /* ϊ */ + { 0x03CB, CHF_VOWEL|CHF_LOWER|CHF_TREMA, 0x03C5, 0x03AB, 0, 0x03B0 }, /* ϋ */ + { 0x03CC, CHF_VOWEL|CHF_LOWER|CHF_OXEIA, 0x03BF, 0x038C }, /* ό */ + { 0x03CD, CHF_VOWEL|CHF_LOWER|CHF_OXEIA|CHF_DIPH2, 0x03C5, 0x038E }, /* ύ */ + { 0x03CE, CHF_VOWEL|CHF_LOWER|CHF_OXEIA, 0x03CE, 0x038F }, /* ώ */ + { 0x03CF, CHF_SYMBOL|CHF_UPPER, 0x03D7 }, /* KAI */ + { 0x03D0, CHF_CONSONANT|CHF_LOWER, 0, 0x0392 }, /* curled beta */ + { 0x03D1, CHF_CONSONANT|CHF_LOWER, 0, 0x0398 }, /* script theta */ + { 0x03D2, CHF_VOWEL|CHF_UPPER, }, /* capital ypsilon with hook */ + { 0x03D3, CHF_VOWEL|CHF_OXEIA, 0x03D2 }, /* capital ypsilon with acute & hook */ + { 0x03D4, CHF_VOWEL|CHF_TREMA, 0x03D2 }, /* capital ypsilon with diaeresis & hook */ + { 0x03D5, CHF_CONSONANT|CHF_LOWER, 0, 0x03A6 }, /* phi */ + { 0x03D6, CHF_CONSONANT|CHF_LOWER, 0, 0x03A0 }, /* pi */ + { 0x03D7, CHF_SYMBOL|CHF_LOWER, 0, 0x03CF }, /* kai */ + { 0x03D8, CHF_ARCHAIC|CHF_CONSONANT|CHF_UPPER, 0, 0x03D9 }, /* QOPPA */ + { 0x03D9, CHF_ARCHAIC|CHF_CONSONANT|CHF_LOWER, 0, 0x03D8 }, /* qoppa */ + { 0x03DA, CHF_ARCHAIC|CHF_CONSONANT|CHF_UPPER|CHF_NUMERIC, 0, 0x03DB, 6 }, /* STIGMA */ + { 0x03DB, CHF_ARCHAIC|CHF_CONSONANT|CHF_LOWER|CHF_NUMERIC, 0, 0x03DA, 6 }, /* stigma */ + { 0x03DC, CHF_ARCHAIC|CHF_CONSONANT|CHF_UPPER|CHF_NUMERIC, 0, 0x03DD, 6 }, /* DIGAMMA */ + { 0x03DD, CHF_ARCHAIC|CHF_CONSONANT|CHF_LOWER|CHF_NUMERIC, 0, 0x03DC, 6 }, /* digamma */ + { 0x03DE, CHF_ARCHAIC|CHF_CONSONANT|CHF_UPPER|CHF_NUMERIC, 0, 0x03DF, 6 }, /* KOPPA */ + { 0x03DF, CHF_ARCHAIC|CHF_CONSONANT|CHF_LOWER|CHF_NUMERIC, 0, 0x03DE, 6 }, /* koppa */ + { 0x03E0, CHF_ARCHAIC|CHF_CONSONANT|CHF_UPPER|CHF_NUMERIC, 0, 0x03E1, 900 }, /* SAMPI */ + { 0x03E1, CHF_ARCHAIC|CHF_CONSONANT|CHF_LOWER|CHF_NUMERIC, 0, 0x03E0, 900 }, /* sampi */ + { 0x03E2, }, + { 0x03E3, }, + { 0x03E4, }, + { 0x03E5, }, + { 0x03E6, }, + { 0x03E7, }, + { 0x03E8, }, + { 0x03E9, }, + { 0x03EA, }, + { 0x03EB, }, + { 0x03EC, }, + { 0x03ED, }, + { 0x03EE, }, + { 0x03EF, }, + { 0x03F0, CHF_CONSONANT|CHF_LOWER, 0, 0x039A }, /* kappa */ + { 0x03F1, CHF_CONSONANT|CHF_LOWER, 0, 0x03A1 }, /* tailed rho */ + { 0x03F2, CHF_CONSONANT, 0, 0x03F9 }, /* lunate sigma */ + { 0x03F3, CHF_SEMIVOWEL|CHF_LOWER, }, /* yot */ + { 0x03F4, CHF_CONSONANT|CHF_UPPER, 0, 0x03B8 }, /* THETA */ + { 0x03F5, CHF_SYMBOL|CHF_LOWER, 0, 0x0395 }, /* lunate epsilon */ + { 0x03F6, CHF_SYMBOL|CHF_LOWER, }, /* reversed lunate epsilon */ + { 0x03F7, }, + { 0x03F8, }, + { 0x03F9, CHF_CONSONANT|CHF_UPPER, 0, 0x03F2 }, /* LUNATE SIGMA */ + { 0x03FA, CHF_ARCHAIC|CHF_CONSONANT|CHF_UPPER, 0, 0x03FB }, /* SAN */ + { 0x03FB, CHF_ARCHAIC|CHF_CONSONANT|CHF_LOWER, 0, 0x03FA }, /* san */ + { 0x03FC, CHF_SYMBOL|CHF_CONSONANT|CHF_LOWER, }, /* rho with stroke */ + { 0x03FD, CHF_SYMBOL|CHF_CONSONANT|CHF_UPPER, 0, 0x037B}, /* CAPITAL REV. LUNATE SIGMA + antisigma */ + { 0x03FE, CHF_SYMBOL|CHF_CONSONANT|CHF_UPPER, 0, 0x037C }, /* CAPITAL DOTTED LUNATE SIGMA + sigma periestigmenon */ + { 0x03FF, CHF_SYMBOL|CHF_CONSONANT|CHF_UPPER, 0, 0x037D }, /* antisigma periestigmenon */ +}; + +/* FIXME: Implement http://www.unicode.org/charts/PDF/U1F00.pdf */ +struct char_info_st el_extended_ctype[] = { + { 0x1F00, }, + { 0x1F01, }, + { 0x1F02, }, + { 0x1F03, }, + { 0x1F04, }, + { 0x1F05, }, + { 0x1F06, }, + { 0x1F07, }, + { 0x1F08, }, + { 0x1F09, }, + { 0x1F0A, }, + { 0x1F0B, }, + { 0x1F0C, }, + { 0x1F0D, }, + { 0x1F0E, }, + { 0x1F0F, }, + { 0x1F10, }, + { 0x1F11, }, + { 0x1F12, }, + { 0x1F13, }, + { 0x1F14, }, + { 0x1F15, }, + { 0x1F16, }, + { 0x1F17, }, + { 0x1F18, }, + { 0x1F19, }, + { 0x1F1A, }, + { 0x1F1B, }, + { 0x1F1C, }, + { 0x1F1D, }, + { 0x1F1E, }, + { 0x1F1F, }, + { 0x1F20, }, + { 0x1F21, }, + { 0x1F22, }, + { 0x1F23, }, + { 0x1F24, }, + { 0x1F25, }, + { 0x1F26, }, + { 0x1F27, }, + { 0x1F28, }, + { 0x1F29, }, + { 0x1F2A, }, + { 0x1F2B, }, + { 0x1F2C, }, + { 0x1F2D, }, + { 0x1F2E, }, + { 0x1F2F, }, + { 0x1F30, }, + { 0x1F31, }, + { 0x1F32, }, + { 0x1F33, }, + { 0x1F34, }, + { 0x1F35, }, + { 0x1F36, }, + { 0x1F37, }, + { 0x1F38, }, + { 0x1F39, }, + { 0x1F3A, }, + { 0x1F3B, }, + { 0x1F3C, }, + { 0x1F3D, }, + { 0x1F3E, }, + { 0x1F3F, }, + { 0x1F40, }, + { 0x1F41, }, + { 0x1F42, }, + { 0x1F43, }, + { 0x1F44, }, + { 0x1F45, }, + { 0x1F46, }, + { 0x1F47, }, + { 0x1F48, }, + { 0x1F49, }, + { 0x1F4A, }, + { 0x1F4B, }, + { 0x1F4C, }, + { 0x1F4D, }, + { 0x1F4E, }, + { 0x1F4F, }, + { 0x1F50, }, + { 0x1F51, }, + { 0x1F52, }, + { 0x1F53, }, + { 0x1F54, }, + { 0x1F55, }, + { 0x1F56, }, + { 0x1F57, }, + { 0x1F58, }, + { 0x1F59, }, + { 0x1F5A, }, + { 0x1F5B, }, + { 0x1F5C, }, + { 0x1F5D, }, + { 0x1F5E, }, + { 0x1F5F, }, + { 0x1F60, }, + { 0x1F61, }, + { 0x1F62, }, + { 0x1F63, }, + { 0x1F64, }, + { 0x1F65, }, + { 0x1F66, }, + { 0x1F67, }, + { 0x1F68, }, + { 0x1F69, }, + { 0x1F6A, }, + { 0x1F6B, }, + { 0x1F6C, }, + { 0x1F6D, }, + { 0x1F6E, }, + { 0x1F6F, }, + { 0x1F70, }, + { 0x1F71, }, + { 0x1F72, }, + { 0x1F73, }, + { 0x1F74, }, + { 0x1F75, }, + { 0x1F76, }, + { 0x1F77, }, + { 0x1F78, }, + { 0x1F79, }, + { 0x1F7A, }, + { 0x1F7B, }, + { 0x1F7C, }, + { 0x1F7D, }, + { 0x1F7E, }, + { 0x1F7F, }, + { 0x1F80, }, + { 0x1F81, }, + { 0x1F82, }, + { 0x1F83, }, + { 0x1F84, }, + { 0x1F85, }, + { 0x1F86, }, + { 0x1F87, }, + { 0x1F88, }, + { 0x1F89, }, + { 0x1F8A, }, + { 0x1F8B, }, + { 0x1F8C, }, + { 0x1F8D, }, + { 0x1F8E, }, + { 0x1F8F, }, + { 0x1F90, }, + { 0x1F91, }, + { 0x1F92, }, + { 0x1F93, }, + { 0x1F94, }, + { 0x1F95, }, + { 0x1F96, }, + { 0x1F97, }, + { 0x1F98, }, + { 0x1F99, }, + { 0x1F9A, }, + { 0x1F9B, }, + { 0x1F9C, }, + { 0x1F9D, }, + { 0x1F9E, }, + { 0x1F9F, }, + { 0x1FA0, }, + { 0x1FA1, }, + { 0x1FA2, }, + { 0x1FA3, }, + { 0x1FA4, }, + { 0x1FA5, }, + { 0x1FA6, }, + { 0x1FA7, }, + { 0x1FA8, }, + { 0x1FA9, }, + { 0x1FAA, }, + { 0x1FAB, }, + { 0x1FAC, }, + { 0x1FAD, }, + { 0x1FAE, }, + { 0x1FAF, }, + { 0x1FB0, }, + { 0x1FB1, }, + { 0x1FB2, }, + { 0x1FB3, }, + { 0x1FB4, }, + { 0x1FB5, }, + { 0x1FB6, }, + { 0x1FB7, }, + { 0x1FB8, }, + { 0x1FB9, }, + { 0x1FBA, }, + { 0x1FBB, }, + { 0x1FBC, }, + { 0x1FBD, }, + { 0x1FBE, }, + { 0x1FBF, }, + { 0x1FC0, }, + { 0x1FC1, }, + { 0x1FC2, }, + { 0x1FC3, }, + { 0x1FC4, }, + { 0x1FC5, }, + { 0x1FC6, }, + { 0x1FC7, }, + { 0x1FC8, }, + { 0x1FC9, }, + { 0x1FCA, }, + { 0x1FCB, }, + { 0x1FCC, }, + { 0x1FCD, }, + { 0x1FCE, }, + { 0x1FCF, }, + { 0x1FD0, }, + { 0x1FD1, }, + { 0x1FD2, }, + { 0x1FD3, }, + { 0x1FD4, }, + { 0x1FD5, }, + { 0x1FD6, }, + { 0x1FD7, }, + { 0x1FD8, }, + { 0x1FD9, }, + { 0x1FDA, }, + { 0x1FDB, }, + { 0x1FDC, }, + { 0x1FDD, }, + { 0x1FDE, }, + { 0x1FDF, }, + { 0x1FE0, }, + { 0x1FE1, }, + { 0x1FE2, }, + { 0x1FE3, }, + { 0x1FE4, }, + { 0x1FE5, }, + { 0x1FE6, }, + { 0x1FE7, }, + { 0x1FE8, }, + { 0x1FE9, }, + { 0x1FEA, }, + { 0x1FEB, }, + { 0x1FEC, }, + { 0x1FED, }, + { 0x1FEE, }, + { 0x1FEF, }, + { 0x1FF0, }, + { 0x1FF1, }, + { 0x1FF2, }, + { 0x1FF3, }, + { 0x1FF4, }, + { 0x1FF5, }, + { 0x1FF6, }, + { 0x1FF7, }, + { 0x1FF8, }, + { 0x1FF9, }, + { 0x1FFA, }, + { 0x1FFB, }, + { 0x1FFC, }, + { 0x1FFD, }, + { 0x1FFE, }, + { 0x1FFF, } +}; + +static struct char_info_st * +elchr_info(unsigned ch) +{ + if (ch >= 0x0300 && ch <= 0x03FF) + return el_basic_ctype + ch - 0x0300; + else if (ch >= 0x1F00 && ch <= 0x1FFF) + return el_extended_ctype + ch - 0x1F00; + return NULL; +} + +int +elchr_flags(unsigned ch) +{ + struct char_info_st *ci = elchr_info(ch); + return ci ? ci->flags : 0; +} + +int +elchr_isupper(unsigned ch) +{ + return elchr_flags(ch) & CHF_UPPER; +} + +int +elchr_islower(unsigned ch) +{ + return elchr_flags(ch) & CHF_LOWER; +} + +int +elchr_getaccent(unsigned ch) +{ + return elchr_flags(ch) & CHF_ACCENT_MASK; +} + +int +elchr_istrema(unsigned ch) +{ + return elchr_flags(ch) & CHF_TREMA; +} + + +int +elchr_isvowel(unsigned ch) +{ + return elchr_flags(ch) & CHF_VOWEL; +} + +int +elchr_isconsonant(unsigned ch) +{ + return elchr_flags(ch) & CHF_CONSONANT; +} + +int +elchr_issemivowel(unsigned ch) +{ + return elchr_flags(ch) & CHF_SEMIVOWEL; +} + +int +elchr_ispunct(unsigned ch) +{ + return elchr_flags(ch) & CHF_PUNCT; +} + +int +elchr_issymbol(unsigned ch) +{ + return elchr_flags(ch) & CHF_SYMBOL; +} + +int +elchr_ismodifier(unsigned ch) +{ + return elchr_flags(ch) & CHF_MODIFIER; +} + +int +elchr_isarchaic(unsigned ch) +{ + return elchr_flags(ch) & CHF_ARCHAIC; +} + +int +elchr_isnumeric(unsigned ch) +{ + return elchr_flags(ch) & CHF_NUMERIC; +} + +unsigned +elchr_numeric_value(unsigned ch) +{ + struct char_info_st *ci = elchr_info(ch); + return (ci && (ci->flags & CHF_NUMERIC)) ? ci->numval: 0; +} + +unsigned +elchr_toupper(unsigned ch) +{ + struct char_info_st *ci = elchr_info(ch); + return (ci && (ci->flags & CHF_LOWER)) ? ci->trans: ch; +} + +unsigned +elchr_tolower(unsigned ch) +{ + struct char_info_st *ci = elchr_info(ch); + return (ci && (ci->flags & CHF_UPPER)) ? ci->trans : ch; +} + +unsigned +elchr_base(unsigned ch) +{ + struct char_info_st *ci = elchr_info(ch); + return (ci && (ci->flags & CHF_ACCENT_MASK) && ci->base) ? ci->base : ch; +} + +unsigned +elchr_deaccent(unsigned ch) +{ + struct char_info_st *ci = elchr_info(ch); + if (ci && (ci->flags & CHF_ACCENT_MASK)) + return ci->deaccent ? ci->deaccent : ci->base ? ci->base : ch; + return ch; +} + +unsigned +elchr_accent(unsigned ch, int acc) +{ + struct char_info_st *ci = elchr_info(ch); + return (ci && ci->accented[acc-1]) ? ci->accented[acc-1] : ch; +} + +int +elchr_diphthong(unsigned ch, int state) +{ + struct char_info_st *ci = elchr_info(ch); + + if (!ci || !(ci->flags & CHF_VOWEL)) + return 0; + switch (state) { + case 0: + if (ci->flags & CHF_DIPH1) + state = 1; + break; + case 1: + if (ci->flags & CHF_DIPH2) + state = 2; + break; + default: + state = 0; + } + return state; +} diff --git a/src/ellinika/elmorph.c b/src/ellinika/elmorph.c new file mode 100644 index 0000000..5234eda --- /dev/null +++ b/src/ellinika/elmorph.c @@ -0,0 +1,655 @@ +#ifdef HAVE_CONFIG_H +# include <config.h> +#endif +#include <errno.h> +#include <stdlib.h> +#include <libguile.h> +#include "utf8.h" +#include "elmorph.h" + +struct elstr { + unsigned *str; /* UTF-8 string */ + size_t len; /* Its length */ + unsigned nsyl; /* Number of syllables. */ + unsigned *sylmap; /* Syllable map (nsyl elements) */ + unsigned acc_syl; /* Number of the accented syllable + (1-based, from the last syllable) */ + unsigned acc_pos; /* Number of the accented character + (0-based, from str[0]) */ +}; + +scm_t_bits _elstr_tag; + +static void +_elstr_syllabize(struct elstr *elstr) +{ + unsigned *sylmap; + unsigned i, nsyl = 0, accsyl = 0, accchr = 0; + int dstate = 0; + int acc = 0; + + sylmap = scm_gc_malloc(sizeof(sylmap[0])*elstr->len, "syllable map"); + + for (i = 0; i < elstr->len; i++) { + int nstate; + + if (elchr_getaccent(elstr->str[i])) { + accsyl = nsyl; + accchr = i; + } + nstate = elchr_diphthong(elstr->str[i], dstate); + if (nstate) + /* skip */; + else if (dstate) + sylmap[nsyl++] = i - 1; + else if (elchr_isvowel(elstr->str[i])) + sylmap[nsyl++] = i; + dstate = nstate; + } + if (dstate) + sylmap[nsyl++] = i - 1; + else + sylmap[nsyl-1] = i - 1; + elstr->sylmap = sylmap; + elstr->nsyl = nsyl; + elstr->acc_pos = accchr; + elstr->acc_syl = nsyl - accsyl; +} + +static SCM +_elstr_alloc(const char *instr) +{ + struct elstr *elstr; + unsigned *wptr; + size_t wlen; + + if (utf8_mbstr_to_wc(instr, &wptr, &wlen)) + return SCM_EOL; + + elstr = scm_gc_malloc(sizeof(*elstr), "Elstr"); + elstr->str = wptr; + elstr->len = wlen; + + _elstr_syllabize(elstr); + + SCM_RETURN_NEWSMOB(_elstr_tag, elstr); +} + +static SCM +_elstr_dup(struct elstr *elstr) +{ + struct elstr *elnew; + + elnew = scm_gc_malloc(sizeof(*elstr), "Elstr"); + elnew->str = calloc(elstr->len, sizeof(elnew->str[0])); + if (!elnew->str) + scm_memory_error("_elstr_dup"); + elnew->sylmap = calloc(elstr->nsyl, sizeof(elnew->sylmap[0])); + if (!elnew->sylmap) { + free(elnew->str); + scm_memory_error("_elstr_dup"); + } + memcpy(elnew->str, elstr->str, sizeof(elstr->str[0]) * elstr->len); + elnew->len = elstr->len; + elnew->nsyl = elstr->nsyl; + memcpy(elnew->sylmap, elstr->sylmap, + sizeof(elstr->sylmap[0]) * elstr->nsyl); + elnew->acc_syl = elstr->acc_syl; + elnew->acc_pos = elstr->acc_pos; + SCM_RETURN_NEWSMOB(_elstr_tag, elnew); +} + +static scm_sizet +_elstr_free(SCM smob) +{ + struct elstr *elstr = (struct elstr *) SCM_CDR(smob); + free(elstr->str); + free(elstr->sylmap); + free(elstr); + return 0; +} + +static int +_elstr_print(SCM smob, SCM port, scm_print_state *pstate) +{ + struct elstr *elstr = (struct elstr *) SCM_CDR(smob); + int i, j, an; + char *s; + + scm_puts("#<elstr ``", port); + an = elstr->nsyl - elstr->acc_syl; + if (an == 0) + scm_puts("[", port); + for (i = j = 0; i < elstr->len; i++) { + char r[6]; + int n; + + if (i == elstr->sylmap[j] + 1) { + if (j == an) + scm_puts("]", port); + scm_puts("-", port); + if (++j == an) + scm_puts("[", port); + } + n = utf8_wctomb(r, elstr->str[i]); + if (n == -1) + continue; + r[n] = 0; + scm_puts(r, port); + } + if (j == an) + scm_puts("]", port); + scm_puts("''>", port); + return 1; +} + +static void +_elstr_init() +{ + _elstr_tag = scm_make_smob_type("Elstr", sizeof(struct elstr)); + scm_set_smob_free(_elstr_tag, _elstr_free); + scm_set_smob_print(_elstr_tag, _elstr_print); +} + +SCM_DEFINE_PUBLIC(scm_string__elstr, "string->elstr", 1, 0, 0, + (SCM string), +"Create new ELSTR from STRING\n") +#define FUNC_NAME s_scm_string__elstr +{ + char *str; + SCM scm; + + SCM_ASSERT(scm_is_string(string), string, SCM_ARG1, FUNC_NAME); + str = scm_to_locale_string(string); + scm = _elstr_alloc(str); + free(str); + if (scm == SCM_EOL) + scm_misc_error(FUNC_NAME, + "Invalid input string: ~S", + scm_list_1(string)); + return scm; +} +#undef FUNC_NAME + +#define scm_is_elstr(s) (!SCM_IMP(s) && SCM_CELL_TYPE(s) == _elstr_tag) + +SCM_DEFINE_PUBLIC(scm_elstr__string, "elstr->string", 1, 0, 0, + (SCM el), +"Convert EL to a STRING\n") +#define FUNC_NAME s_scm_elstr__string +{ + struct elstr *elstr; + char *s; + SCM scm; + + SCM_ASSERT(scm_is_elstr(el), el, SCM_ARG1, FUNC_NAME); + elstr = (struct elstr*) SCM_CDR(el); + if (utf8_wc_to_mbstr(elstr->str, elstr->len, &s)) + scm_misc_error(FUNC_NAME, + "cannot convert elstr to Scheme", + SCM_EOL); + scm = scm_from_locale_string(s); + free(s); + return scm; +} +#undef FUNC_NAME + +SCM_DEFINE_PUBLIC(scm_elstr_length, "elstr-length", 1, 0, 0, + (SCM el), +"Returns the number of characters in EL\n") +#define FUNC_NAME s_scm_elstr_length +{ + struct elstr *elstr; + + SCM_ASSERT(scm_is_elstr(el), el, SCM_ARG1, FUNC_NAME); + elstr = (struct elstr*) SCM_CDR(el); + return scm_from_uint(elstr->len); +} +#undef FUNC_NAME + +SCM_DEFINE_PUBLIC(scm_elstr_number_of_syllables, "elstr-number-of-syllables", + 1, 0, 0, + (SCM el), +"Returns the number of characters in EL\n") +#define FUNC_NAME s_scm_elstr_number_of_syllables +{ + struct elstr *elstr; + + SCM_ASSERT(scm_is_elstr(el), el, SCM_ARG1, FUNC_NAME); |