X-Git-Url: https://www.chiark.greenend.org.uk/ucgi/~mdw/git/disorder/blobdiff_plain/8818b7fca12456e62410ef914a7bef250a0633c9..refs/heads/master:/lib/unicode.c?ds=sidebyside

diff --git a/lib/unicode.c b/lib/unicode.c
index b5b520c..27f82b7 100644
--- a/lib/unicode.c
+++ b/lib/unicode.c
@@ -1,21 +1,19 @@
 /*
  * This file is part of DisOrder
- * Copyright (C) 2007 Richard Kettlewell
+ * Copyright (C) 2007, 2009, 2013 Richard Kettlewell
  *
- * This program is free software; you can redistribute it and/or modify
+ * This program 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 2 of the License, or
+ * the Free Software Foundation, either version 3 of the License, or
  * (at your option) any later version.
- *
- * This program 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.
- *
+ * 
+ * This program 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 this program; if not, write to the Free Software
- * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307
- * USA
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
  */
 /** @file lib/unicode.c
  * @brief Unicode support functions
@@ -39,11 +37,7 @@
  * - @ref utf8
  */
 
-#include <config.h>
-#include "types.h"
-
-#include <string.h>
-#include <stdio.h>		/* TODO */
+#include "common.h"
 
 #include "mem.h"
 #include "vector.h"
@@ -109,6 +103,16 @@ static inline int utf32__combining_class(uint32_t c) {
   return utf32__unidata(c)->ccc;
 }
 
+/** @brief Return the combining class of @p c
+ * @param c Code point
+ * @return Combining class of @p c
+ *
+ * @p c can be any 32-bit value, a sensible value will be returned regardless.
+ */
+int utf32_combining_class(uint32_t c) {
+  return utf32__combining_class(c);
+}
+
 /** @brief Return the General_Category value for @p c
  * @param c Code point
  * @return General_Category property value
@@ -301,6 +305,133 @@ error:
   return 0;
 }
 
+/** @brief Convert UTF-16 to UTF-8
+ * @param s Source string
+ * @param ns Length of source string in code points
+ * @param ndp Where to store length of destination string (or NULL)
+ * @return Newly allocated destination string or NULL on error
+ *
+ * If the UTF-16 is not valid then NULL is returned.  A UTF-16 sequence t is
+ * invalid if it contains an incomplete surrogate.
+ *
+ * The return value is always 0-terminated.  The value returned via @p *ndp
+ * does not include the terminator.
+ */
+char *utf16_to_utf8(const uint16_t *s, size_t ns, size_t *ndp) {
+  struct dynstr d;
+  uint32_t c;
+
+  dynstr_init(&d);
+  while(ns > 0) {
+    c = *s++;
+    --ns;
+    if(c >= 0xD800 && c <= 0xDBFF) {
+      if(ns && *s >= 0xDC00 && c <= 0xDFFF)
+        c = ((c - 0xD800) << 10) + (*s++ - 0xDC00) + 0x10000;
+      else
+        goto error;
+    } else if(c >= 0xDC00 && c <= 0xDFFF)
+      goto error;
+    if(c < 0x80)
+      dynstr_append(&d, c);
+    else if(c < 0x0800) {
+      dynstr_append(&d, 0xC0 | (c >> 6));
+      dynstr_append(&d, 0x80 | (c & 0x3F));
+    } else if(c < 0x10000) {
+      if(c >= 0xD800 && c <= 0xDFFF)
+	goto error;
+      dynstr_append(&d, 0xE0 | (c >> 12));
+      dynstr_append(&d, 0x80 | ((c >> 6) & 0x3F));
+      dynstr_append(&d, 0x80 | (c & 0x3F));
+    } else if(c < 0x110000) {
+      dynstr_append(&d, 0xF0 | (c >> 18));
+      dynstr_append(&d, 0x80 | ((c >> 12) & 0x3F));
+      dynstr_append(&d, 0x80 | ((c >> 6) & 0x3F));
+      dynstr_append(&d, 0x80 | (c & 0x3F));
+    } else
+      goto error;
+  }
+  dynstr_terminate(&d);
+  if(ndp)
+    *ndp = d.nvec;
+  return d.vec;
+error:
+  xfree(d.vec);
+  return 0;
+}
+
+/** @brief Convert UTF-8 to UTF-16
+ * @param s Source string
+ * @param ns Length of source string in code points
+ * @param ndp Where to store length of destination string (or NULL)
+ * @return Newly allocated destination string or NULL on error
+ *
+ * The return value is always 0-terminated.  The value returned via @p *ndp
+ * does not include the terminator.
+ *
+ * If the UTF-8 is not valid then NULL is returned.  A UTF-8 sequence
+ * for a code point is invalid if:
+ * - it is not the shortest possible sequence for the code point
+ * - it codes for a UTF-16 surrogate
+ * - it codes for a value outside the unicode code space
+ */
+uint16_t *utf8_to_utf16(const char *s, size_t ns, size_t *ndp) {
+  struct dynstr_utf16 d;
+  uint32_t c32;
+  const uint8_t *ss = (const uint8_t *)s;
+  int n;
+
+  dynstr_utf16_init(&d);
+  while(ns > 0) {
+    const struct unicode_utf8_row *const r = &unicode_utf8_valid[*ss];
+    if(r->count <= ns) {
+      switch(r->count) {
+      case 1:
+        c32 = *ss;
+        break;
+      case 2:
+        if(ss[1] < r->min2 || ss[1] > r->max2)
+          goto error;
+        c32 = *ss & 0x1F;
+        break;
+      case 3:
+        if(ss[1] < r->min2 || ss[1] > r->max2)
+          goto error;
+        c32 = *ss & 0x0F;
+        break;
+      case 4:
+        if(ss[1] < r->min2 || ss[1] > r->max2)
+          goto error;
+        c32 = *ss & 0x07;
+        break;
+      default:
+        goto error;
+      }
+    } else
+      goto error;
+    for(n = 1; n < r->count; ++n) {
+      if(ss[n] < 0x80 || ss[n] > 0xBF)
+        goto error;
+      c32 = (c32 << 6) | (ss[n] & 0x3F);
+    }
+    if(c32 >= 0x10000) {
+      c32 -= 0x10000;
+      dynstr_utf16_append(&d, 0xD800 + (c32 >> 10));
+      dynstr_utf16_append(&d, 0xDC00 + (c32 & 0x03FF));
+    } else
+      dynstr_utf16_append(&d, c32);
+    ss += r->count;
+    ns -= r->count;
+  }
+  dynstr_utf16_terminate(&d);
+  if(ndp)
+    *ndp = d.nvec;
+  return d.vec;
+error:
+  xfree(d.vec);
+  return 0;
+}
+
 /** @brief Test whether [s,s+ns) is valid UTF-8
  * @param s Start of string
  * @param ns Length of string
@@ -372,21 +503,10 @@ struct utf32_iterator_data {
    * the value is (uint32_t)-1.
    */
   uint32_t last[2];
-};
 
-/** @brief Create a new iterator pointing at the start of a string
- * @param s Start of string
- * @param ns Length of string
- * @return New iterator
- */
-utf32_iterator utf32_iterator_new(const uint32_t *s, size_t ns) {
-  utf32_iterator it = xmalloc(sizeof *it);
-  it->s = s;
-  it->ns = ns;
-  it->n = 0;
-  it->last[0] = it->last[1] = -1;
-  return it;
-}
+  /** @brief Tailoring for Word_Break */
+  unicode_property_tailor *word_break;
+};
 
 /** @brief Initialize an internal private iterator
  * @param it Iterator
@@ -400,9 +520,54 @@ static void utf32__iterator_init(utf32_iterator it,
   it->ns = ns;
   it->n = 0;
   it->last[0] = it->last[1] = -1;
+  it->word_break = 0;
   utf32_iterator_set(it, n);
 }
 
+/** @brief Create a new iterator pointing at the start of a string
+ * @param s Start of string
+ * @param ns Length of string
+ * @return New iterator
+ */
+utf32_iterator utf32_iterator_new(const uint32_t *s, size_t ns) {
+  utf32_iterator it = xmalloc(sizeof *it);
+  utf32__iterator_init(it, s, ns, 0);
+  return it;
+}
+
+/** @brief Tailor this iterator's interpretation of the Word_Break property.
+ * @param it Iterator
+ * @param pt Property tailor function or NULL
+ *
+ * After calling this the iterator will call @p pt to determine the Word_Break
+ * property of each code point.  If it returns -1 the default value will be
+ * used otherwise the returned value will be used.
+ *
+ * @p pt can be NULL to revert to the default value of the property.
+ *
+ * It is safe to call this function at any time; the iterator's internal state
+ * will be reset to suit the new tailoring.
+ */
+void utf32_iterator_tailor_word_break(utf32_iterator it,
+                                      unicode_property_tailor *pt) {
+  it->word_break = pt;
+  utf32_iterator_set(it, it->n);
+}
+
+static inline enum unicode_Word_Break utf32__iterator_word_break(utf32_iterator it,
+                                                                 uint32_t c) {
+  if(!it->word_break)
+    return utf32__word_break(c);
+  else {
+    const int t = it->word_break(c);
+
+    if(t < 0)
+      return utf32__word_break(c);
+    else
+      return t;
+  }
+}
+
 /** @brief Destroy an iterator
  * @param it Iterator
  */
@@ -444,14 +609,18 @@ int utf32_iterator_set(utf32_iterator it, size_t n) {
     return -1;
   /* Walk backwards skipping ignorable code points */
   m = n;
-  while(m > 0 && (utf32__boundary_ignorable(utf32__word_break(it->s[m-1]))))
+  while(m > 0
+        && (utf32__boundary_ignorable(utf32__iterator_word_break(it,
+                                                                 it->s[m-1]))))
     --m;
   /* Either m=0 or s[m-1] is not ignorable */
   if(m > 0) {
     --m;
     /* s[m] is our first non-ignorable code; look for a second in the same
        way **/
-    while(m > 0 && (utf32__boundary_ignorable(utf32__word_break(it->s[m-1]))))
+    while(m > 0
+          && (utf32__boundary_ignorable(utf32__iterator_word_break(it,
+                                                                   it->s[m-1]))))
       --m;
     /* Either m=0 or s[m-1] is not ignorable */
     if(m > 0)
@@ -478,7 +647,7 @@ int utf32_iterator_advance(utf32_iterator it, size_t count) {
   if(count <= it->ns - it->n) {
     while(count > 0) {
       const uint32_t c = it->s[it->n];
-      const enum unicode_Word_Break wb = utf32__word_break(c);
+      const enum unicode_Word_Break wb = utf32__iterator_word_break(it, c);
       if(it->last[1] == (uint32_t)-1
          || !utf32__boundary_ignorable(wb)) {
         it->last[0] = it->last[1];
@@ -561,6 +730,12 @@ int utf32_iterator_grapheme_boundary(utf32_iterator it) {
   /* GB9 */
   if(gbafter == unicode_Grapheme_Break_Extend)
     return 0;
+  /* GB9a */
+  if(gbafter == unicode_Grapheme_Break_SpacingMark)
+    return 0;
+  /* GB9b */
+  if(gbbefore == unicode_Grapheme_Break_Prepend)
+    return 0;
   /* GB10 */
   return 1;
 
@@ -576,94 +751,112 @@ int utf32_iterator_grapheme_boundary(utf32_iterator it) {
  * string) and 0 otherwise.
  */
 int utf32_iterator_word_boundary(utf32_iterator it) {
-  enum unicode_Word_Break twobefore, before, after, twoafter;
+  uint32_t before, after;
+  enum unicode_Word_Break wbtwobefore, wbbefore, wbafter, wbtwoafter;
   size_t nn;
 
   /* WB1 and WB2 */
   if(it->n == 0 || it->n == it->ns)
     return 1;
+  before = it->s[it->n-1];
+  after = it->s[it->n];
   /* WB3 */
-  if(it->s[it->n-1] == 0x000D && it->s[it->n] == 0x000A)
+  if(before == 0x000D && after == 0x000A)
     return 0;
+  /* WB3a */
+  if(utf32__iterator_word_break(it, before) == unicode_Word_Break_Newline
+     || before == 0x000D
+     || before == 0x000A)
+    return 1;
+  /* WB3b */
+  if(utf32__iterator_word_break(it, after) == unicode_Word_Break_Newline
+     || after == 0x000D
+     || after == 0x000A)
+    return 1;
   /* WB4 */
   /* (!Sep) x (Extend|Format) as in UAX #29 s6.2 */
-  if(utf32__sentence_break(it->s[it->n-1]) != unicode_Sentence_Break_Sep
-     && utf32__boundary_ignorable(utf32__word_break(it->s[it->n])))
+  if(utf32__sentence_break(before) != unicode_Sentence_Break_Sep
+     && utf32__boundary_ignorable(utf32__iterator_word_break(it, after)))
     return 0;
   /* Gather the property values we'll need for the rest of the test taking the
    * s6.2 changes into account */
   /* First we look at the code points after the proposed boundary */
   nn = it->n;                           /* <it->ns */
-  after = utf32__word_break(it->s[nn++]);
-  if(!utf32__boundary_ignorable(after)) {
+  wbafter = utf32__iterator_word_break(it, it->s[nn++]);
+  if(!utf32__boundary_ignorable(wbafter)) {
     /* X (Extend|Format)* -> X */
     while(nn < it->ns
-          && utf32__boundary_ignorable(utf32__word_break(it->s[nn])))
+          && utf32__boundary_ignorable(utf32__iterator_word_break(it,
+                                                                  it->s[nn])))
       ++nn;
   }
   /* It's possible now that nn=ns */
   if(nn < it->ns)
-    twoafter = utf32__word_break(it->s[nn]);
+    wbtwoafter = utf32__iterator_word_break(it, it->s[nn]);
   else
-    twoafter = unicode_Word_Break_Other;
+    wbtwoafter = unicode_Word_Break_Other;
 
   /* We've already recorded the non-ignorable code points before the proposed
    * boundary */
-  before = utf32__word_break(it->last[1]);
-  twobefore = utf32__word_break(it->last[0]);
+  wbbefore = utf32__iterator_word_break(it, it->last[1]);
+  wbtwobefore = utf32__iterator_word_break(it, it->last[0]);
 
   /* WB5 */
-  if(before == unicode_Word_Break_ALetter
-     && after == unicode_Word_Break_ALetter)
+  if(wbbefore == unicode_Word_Break_ALetter
+     && wbafter == unicode_Word_Break_ALetter)
     return 0;
   /* WB6 */
-  if(before == unicode_Word_Break_ALetter
-     && after == unicode_Word_Break_MidLetter
-     && twoafter == unicode_Word_Break_ALetter)
+  if(wbbefore == unicode_Word_Break_ALetter
+     && (wbafter == unicode_Word_Break_MidLetter
+         || wbafter == unicode_Word_Break_MidNumLet)
+     && wbtwoafter == unicode_Word_Break_ALetter)
     return 0;
   /* WB7 */
-  if(twobefore == unicode_Word_Break_ALetter
-     && before == unicode_Word_Break_MidLetter
-     && after == unicode_Word_Break_ALetter)
+  if(wbtwobefore == unicode_Word_Break_ALetter
+     && (wbbefore == unicode_Word_Break_MidLetter
+         || wbbefore == unicode_Word_Break_MidNumLet)
+     && wbafter == unicode_Word_Break_ALetter)
     return 0;
-  /* WB8 */  
-  if(before == unicode_Word_Break_Numeric
-     && after == unicode_Word_Break_Numeric)
+  /* WB8 */
+  if(wbbefore == unicode_Word_Break_Numeric
+     && wbafter == unicode_Word_Break_Numeric)
     return 0;
   /* WB9 */
-  if(before == unicode_Word_Break_ALetter
-     && after == unicode_Word_Break_Numeric)
+  if(wbbefore == unicode_Word_Break_ALetter
+     && wbafter == unicode_Word_Break_Numeric)
     return 0;
   /* WB10 */
-  if(before == unicode_Word_Break_Numeric
-     && after == unicode_Word_Break_ALetter)
+  if(wbbefore == unicode_Word_Break_Numeric
+     && wbafter == unicode_Word_Break_ALetter)
     return 0;
    /* WB11 */
-  if(twobefore == unicode_Word_Break_Numeric
-     && before == unicode_Word_Break_MidNum
-     && after == unicode_Word_Break_Numeric)
+  if(wbtwobefore == unicode_Word_Break_Numeric
+     && (wbbefore == unicode_Word_Break_MidNum
+         || wbbefore == unicode_Word_Break_MidNumLet)
+     && wbafter == unicode_Word_Break_Numeric)
     return 0;
   /* WB12 */
-  if(before == unicode_Word_Break_Numeric
-     && after == unicode_Word_Break_MidNum
-     && twoafter == unicode_Word_Break_Numeric)
+  if(wbbefore == unicode_Word_Break_Numeric
+     && (wbafter == unicode_Word_Break_MidNum
+         || wbafter == unicode_Word_Break_MidNumLet)
+     && wbtwoafter == unicode_Word_Break_Numeric)
     return 0;
   /* WB13 */
-  if(before == unicode_Word_Break_Katakana
-     && after == unicode_Word_Break_Katakana)
+  if(wbbefore == unicode_Word_Break_Katakana
+     && wbafter == unicode_Word_Break_Katakana)
     return 0;
   /* WB13a */
-  if((before == unicode_Word_Break_ALetter
-      || before == unicode_Word_Break_Numeric
-      || before == unicode_Word_Break_Katakana
-      || before == unicode_Word_Break_ExtendNumLet)
-     && after == unicode_Word_Break_ExtendNumLet)
+  if((wbbefore == unicode_Word_Break_ALetter
+      || wbbefore == unicode_Word_Break_Numeric
+      || wbbefore == unicode_Word_Break_Katakana
+      || wbbefore == unicode_Word_Break_ExtendNumLet)
+     && wbafter == unicode_Word_Break_ExtendNumLet)
     return 0;
   /* WB13b */
-  if(before == unicode_Word_Break_ExtendNumLet
-     && (after == unicode_Word_Break_ALetter
-         || after == unicode_Word_Break_Numeric
-         || after == unicode_Word_Break_Katakana))
+  if(wbbefore == unicode_Word_Break_ExtendNumLet
+     && (wbafter == unicode_Word_Break_ALetter
+         || wbafter == unicode_Word_Break_Numeric
+         || wbafter == unicode_Word_Break_Katakana))
     return 0;
   /* WB14 */
   return 1;
@@ -1275,12 +1468,14 @@ int utf32_is_word_boundary(const uint32_t *s, size_t ns, size_t n) {
  * @param s Pointer to start of string
  * @param ns Length of string
  * @param nwp Where to store word count, or NULL
+ * @param wbreak Word_Break property tailor, or NULL
  * @return Pointer to array of pointers to words
  *
  * The returned array is terminated by a NULL pointer and individual
  * strings are 0-terminated.
  */
-uint32_t **utf32_word_split(const uint32_t *s, size_t ns, size_t *nwp) {
+uint32_t **utf32_word_split(const uint32_t *s, size_t ns, size_t *nwp,
+                            unicode_property_tailor *wbreak) {
   struct utf32_iterator_data it[1];
   size_t b1 = 0, b2 = 0 ,i;
   int isword;
@@ -1289,6 +1484,7 @@ uint32_t **utf32_word_split(const uint32_t *s, size_t ns, size_t *nwp) {
 
   vector32_init(v32);
   utf32__iterator_init(it, s, ns, 0);
+  it->word_break = wbreak;
   /* Work our way through the string stopping at each word break. */
   do {
     if(utf32_iterator_word_boundary(it)) {
@@ -1300,7 +1496,7 @@ uint32_t **utf32_word_split(const uint32_t *s, size_t ns, size_t *nwp) {
        * whether they are a word or not */
       isword = 0;
       for(i = b1; i < b2; ++i) {
-        switch(utf32__word_break(it->s[i])) {
+        switch(utf32__iterator_word_break(it, it->s[i])) {
         case unicode_Word_Break_ALetter:
         case unicode_Word_Break_Numeric:
         case unicode_Word_Break_Katakana:
@@ -1312,8 +1508,10 @@ uint32_t **utf32_word_split(const uint32_t *s, size_t ns, size_t *nwp) {
       }
       /* If it's a word add it to the list of results */
       if(isword) {
-        w = xcalloc(b2 - b1 + 1, sizeof(uint32_t));
-        memcpy(w, it->s + b1, (b2 - b1) * sizeof (uint32_t));
+        const size_t len = b2 - b1;
+        w = xcalloc_noptr(len + 1, sizeof(uint32_t));
+        memcpy(w, it->s + b1, len * sizeof (uint32_t));
+        w[len] = 0;
         vector32_append(v32, w);
       }
     }
@@ -1468,18 +1666,20 @@ char *utf8_casefold_compat(const char *s, size_t ns, size_t *ndp) {
  * @param s Pointer to start of string
  * @param ns Length of string
  * @param nwp Where to store word count, or NULL
+ * @param wbreak Word_Break property tailor, or NULL
  * @return Pointer to array of pointers to words
  *
  * The returned array is terminated by a NULL pointer and individual
  * strings are 0-terminated.
  */
-char **utf8_word_split(const char *s, size_t ns, size_t *nwp) {
+char **utf8_word_split(const char *s, size_t ns, size_t *nwp,
+                       unicode_property_tailor *wbreak) {
   uint32_t *to32 = 0, **v32 = 0;
   size_t nto32, nv, n;
   char **v8 = 0, **ret = 0;
-                                                                
+
   if(!(to32 = utf8_to_utf32(s, ns, &nto32))) goto error;
-  if(!(v32 = utf32_word_split(to32, nto32, &nv))) goto error;
+  if(!(v32 = utf32_word_split(to32, nto32, &nv, wbreak))) goto error;
   v8 = xcalloc(sizeof (char *), nv + 1);
   for(n = 0; n < nv; ++n)
     if(!(v8[n] = utf32_to_utf8(v32[n], utf32_len(v32[n]), 0)))
@@ -1487,7 +1687,7 @@ char **utf8_word_split(const char *s, size_t ns, size_t *nwp) {
   ret = v8;
   *nwp = nv;
   v8 = 0;                               /* don't free */
-error:                                                          
+error:
   if(v8) {
     for(n = 0; n < nv; ++n)
       xfree(v8[n]);
@@ -1505,6 +1705,20 @@ error:
 
 /*@}*/
 
+/** @brief Return the length of a 0-terminated UTF-16 string
+ * @param s Pointer to 0-terminated string
+ * @return Length of string in code points (excluding terminator)
+ *
+ * Unlike the conversion functions no validity checking is done on the string.
+ */
+size_t utf16_len(const uint16_t *s) {
+  const uint16_t *t = s;
+
+  while(*t)
+    ++t;
+  return (size_t)(t - s);
+}
+
 /*
 Local Variables:
 c-basic-offset:2