X-Git-Url: https://pintos-os.org/cgi-bin/gitweb.cgi?a=blobdiff_plain;f=src%2Flibpspp%2Fstr.c;h=811674225c23c9f0f8f757bab266ccae662b7c38;hb=32538f9f35aee7145a49971f9dae1394a1a201b5;hp=d7c71b11f5509f15678693555676c54cc4662e3f;hpb=6f3865480503c571963d8a2d1af858a4d72d4e88;p=pspp

diff --git a/src/libpspp/str.c b/src/libpspp/str.c
index d7c71b11f5..811674225c 100644
--- a/src/libpspp/str.c
+++ b/src/libpspp/str.c
@@ -1,5 +1,6 @@
 /* PSPP - a program for statistical analysis.
-   Copyright (C) 1997-9, 2000, 2006, 2009, 2010, 2011, 2012 Free Software Foundation, Inc.
+   Copyright (C) 1997-9, 2000, 2006, 2009, 2010, 2011, 2012, 2014,
+   2020 Free Software Foundation, Inc.
 
    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
@@ -25,6 +26,7 @@
 #include <unistr.h>
 
 #include "libpspp/cast.h"
+#include "libpspp/i18n.h"
 #include "libpspp/message.h"
 #include "libpspp/pool.h"
 
@@ -142,9 +144,9 @@ buf_copy_str_lpad (char *dst, size_t dst_size, const char *src, char pad)
     memcpy (dst, src, dst_size);
   else
     {
-      size_t pad_cnt = dst_size - src_len;
-      memset (&dst[0], pad, pad_cnt);
-      memcpy (dst + pad_cnt, src, src_len);
+      size_t n_pad = dst_size - src_len;
+      memset (&dst[0], pad, n_pad);
+      memcpy (dst + n_pad, src, src_len);
     }
 }
 
@@ -189,7 +191,7 @@ buf_copy_rpad (char *dst, size_t dst_size,
 void
 str_copy_rpad (char *dst, size_t dst_size, const char *src)
 {
-  if (dst_size > 0) 
+  if (dst_size > 0)
     {
       size_t src_len = strlen (src);
       if (src_len < dst_size - 1)
@@ -257,9 +259,10 @@ str_lowercase (char *s)
 }
 
 /* Converts NUMBER into a string in 26-adic notation in BUFFER,
-   which has room for SIZE bytes.  Returns true if successful,
-   false if NUMBER, plus a trailing null, is too large to fit in
-   the available space.
+   which has room for SIZE bytes.  Uses uppercase if UPPERCASE is
+   true, otherwise lowercase, Returns true if successful, false
+   if NUMBER, plus a trailing null, is too large to fit in the
+   available space.
 
    26-adic notation is "spreadsheet column numbering": 1 = A, 2 =
    B, 3 = C, ... 26 = Z, 27 = AA, 28 = AB, 29 = AC, ...
@@ -271,15 +274,18 @@ str_lowercase (char *s)
    For more information, see
    http://en.wikipedia.org/wiki/Bijective_numeration. */
 bool
-str_format_26adic (unsigned long int number, char buffer[], size_t size)
+str_format_26adic (unsigned long int number, bool uppercase,
+                   char buffer[], size_t size)
 {
+  const char *alphabet
+    = uppercase ? "ABCDEFGHIJKLMNOPQRSTUVWXYZ" : "abcdefghijklmnopqrstuvwxyz";
   size_t length = 0;
 
   while (number-- > 0)
     {
       if (length >= size)
         goto overflow;
-      buffer[length++] = "ABCDEFGHIJKLMNOPQRSTUVWXYZ"[number % 26];
+      buffer[length++] = alphabet[number % 26];
       number /= 26;
     }
 
@@ -296,6 +302,39 @@ overflow:
   return false;
 }
 
+/* Copies IN to buffer OUT with size OUT_SIZE, appending a null terminator.  If
+   IN is too long for OUT, or if IN contains a new-line, replaces the tail with
+   "...".
+
+   OUT_SIZE must be at least 16. */
+void
+str_ellipsize (struct substring in, char *out, size_t out_size)
+{
+  assert (out_size >= 16);
+
+  size_t out_maxlen = out_size - 1;
+  if (in.length > out_maxlen - 3)
+    out_maxlen -= 3;
+
+  size_t out_len = 0;
+  while (out_len < in.length
+         && in.string[out_len] != '\n'
+         && in.string[out_len] != '\0'
+         && (in.string[out_len] != '\r'
+             || out_len + 1 >= in.length
+             || in.string[out_len + 1] != '\n'))
+    {
+      int mblen = u8_mblen (CHAR_CAST (const uint8_t *, in.string + out_len),
+                            in.length - out_len);
+      if (mblen < 0 || out_len + mblen > out_maxlen)
+        break;
+      out_len += mblen;
+    }
+
+  memcpy (out, in.string, out_len);
+  strcpy (&out[out_len], out_len < in.length ? "..." : "");
+}
+
 /* Sets the SIZE bytes starting at BLOCK to C,
    and returns the byte following BLOCK. */
 void *
@@ -307,32 +346,32 @@ mempset (void *block, int c, size_t size)
 
 /* Substrings. */
 
-/* Returns a substring whose contents are the CNT bytes
+/* Returns a substring whose contents are the N bytes
    starting at the (0-based) position START in SS. */
 struct substring
-ss_substr (struct substring ss, size_t start, size_t cnt)
+ss_substr (struct substring ss, size_t start, size_t n)
 {
   if (start < ss.length)
-    return ss_buffer (ss.string + start, MIN (cnt, ss.length - start));
+    return ss_buffer (ss.string + start, MIN (n, ss.length - start));
   else
     return ss_buffer (ss.string + ss.length, 0);
 }
 
-/* Returns a substring whose contents are the first CNT
+/* Returns a substring whose contents are the first N
    bytes in SS. */
 struct substring
-ss_head (struct substring ss, size_t cnt)
+ss_head (struct substring ss, size_t n)
 {
-  return ss_buffer (ss.string, MIN (cnt, ss.length));
+  return ss_buffer (ss.string, MIN (n, ss.length));
 }
 
-/* Returns a substring whose contents are the last CNT bytes
+/* Returns a substring whose contents are the last N bytes
    in SS. */
 struct substring
-ss_tail (struct substring ss, size_t cnt)
+ss_tail (struct substring ss, size_t n)
 {
-  if (cnt < ss.length)
-    return ss_buffer (ss.string + (ss.length - cnt), cnt);
+  if (n < ss.length)
+    return ss_buffer (ss.string + (ss.length - n), n);
   else
     return ss;
 }
@@ -346,12 +385,12 @@ ss_alloc_substring (struct substring *new, struct substring old)
   new->length = old.length;
 }
 
-/* Allocates room for a CNT-byte string in NEW. */
+/* Allocates room for a N-byte string in NEW. */
 void
-ss_alloc_uninit (struct substring *new, size_t cnt)
+ss_alloc_uninit (struct substring *new, size_t n)
 {
-  new->string = xmalloc (cnt);
-  new->length = cnt;
+  new->string = xmalloc (n);
+  new->length = n;
 }
 
 void
@@ -372,12 +411,12 @@ ss_alloc_substring_pool (struct substring *new, struct substring old,
   new->string[old.length] = '\0';
 }
 
-/* Allocates room for a CNT-byte string in NEW in POOL. */
+/* Allocates room for a N-byte string in NEW in POOL. */
 void
-ss_alloc_uninit_pool (struct substring *new, size_t cnt, struct pool *pool)
+ss_alloc_uninit_pool (struct substring *new, size_t n, struct pool *pool)
 {
-  new->string = pool_alloc_unaligned (pool, cnt);
-  new->length = cnt;
+  new->string = pool_alloc_unaligned (pool, n);
+  new->length = n;
 }
 
 /* Frees the string that SS points to. */
@@ -387,12 +426,21 @@ ss_dealloc (struct substring *ss)
   free (ss->string);
 }
 
-/* Truncates SS to at most CNT bytes in length. */
+/* Exchanges the contents of A and B. */
+void
+ss_swap (struct substring *a, struct substring *b)
+{
+  struct substring tmp = *a;
+  *a = *b;
+  *b = tmp;
+}
+
+/* Truncates SS to at most N bytes in length. */
 void
-ss_truncate (struct substring *ss, size_t cnt)
+ss_truncate (struct substring *ss, size_t n)
 {
-  if (ss->length > cnt)
-    ss->length = cnt;
+  if (ss->length > n)
+    ss->length = n;
 }
 
 /* Removes trailing bytes in TRIM_SET from SS.
@@ -400,13 +448,13 @@ ss_truncate (struct substring *ss, size_t cnt)
 size_t
 ss_rtrim (struct substring *ss, struct substring trim_set)
 {
-  size_t cnt = 0;
-  while (cnt < ss->length
+  size_t n = 0;
+  while (n < ss->length
          && ss_find_byte (trim_set,
-                          ss->string[ss->length - cnt - 1]) != SIZE_MAX)
-    cnt++;
-  ss->length -= cnt;
-  return cnt;
+                          ss->string[ss->length - n - 1]) != SIZE_MAX)
+    n++;
+  ss->length -= n;
+  return n;
 }
 
 /* Removes leading bytes in TRIM_SET from SS.
@@ -414,9 +462,9 @@ ss_rtrim (struct substring *ss, struct substring trim_set)
 size_t
 ss_ltrim (struct substring *ss, struct substring trim_set)
 {
-  size_t cnt = ss_span (*ss, trim_set);
-  ss_advance (ss, cnt);
-  return cnt;
+  size_t n = ss_span (*ss, trim_set);
+  ss_advance (ss, n);
+  return n;
 }
 
 /* Trims leading and trailing bytes in TRIM_SET from SS. */
@@ -506,18 +554,18 @@ ss_tokenize (struct substring ss, struct substring delimiters,
   ss_get_bytes (&ss, ss_cspan (ss, delimiters), token);
 
   found_token = ss_length (*token) > 0;
-  *save_idx += ss_length (*token) + found_token;
+  *save_idx += ss_length (*token) + (found_token?1:0);
   return found_token;
 }
 
-/* Removes the first CNT bytes from SS. */
+/* Removes the first N bytes from SS. */
 void
-ss_advance (struct substring *ss, size_t cnt)
+ss_advance (struct substring *ss, size_t n)
 {
-  if (cnt > ss->length)
-    cnt = ss->length;
-  ss->string += cnt;
-  ss->length -= cnt;
+  if (n > ss->length)
+    n = ss->length;
+  ss->string += n;
+  ss->length -= n;
 }
 
 /* If the first byte in SS is C, removes it and returns true.
@@ -592,15 +640,15 @@ ss_get_until (struct substring *ss, char delimiter, struct substring *out)
   return ss_match_byte (ss, delimiter);
 }
 
-/* Stores the first CNT bytes in SS in OUT (or fewer, if SS
-   is shorter than CNT bytes).  Trims the same bytes
-   from the beginning of SS.  Returns CNT. */
+/* Stores the first N bytes in SS in OUT (or fewer, if SS
+   is shorter than N bytes).  Trims the same bytes
+   from the beginning of SS.  Returns N. */
 size_t
-ss_get_bytes (struct substring *ss, size_t cnt, struct substring *out)
+ss_get_bytes (struct substring *ss, size_t n, struct substring *out)
 {
-  *out = ss_head (*ss, cnt);
-  ss_advance (ss, cnt);
-  return cnt;
+  *out = ss_head (*ss, n);
+  ss_advance (ss, n);
+  return n;
 }
 
 /* Parses and removes an optionally signed decimal integer from
@@ -687,6 +735,14 @@ ss_last (struct substring ss)
   return ss.length > 0 ? (unsigned char) ss.string[ss.length - 1] : EOF;
 }
 
+/* Returns true if SS starts with PREFIX, false otherwise. */
+bool
+ss_starts_with (struct substring ss, struct substring prefix)
+{
+  return (ss.length >= prefix.length
+          && !memcmp (ss.string, prefix.string, prefix.length));
+}
+
 /* Returns true if SS ends with SUFFIX, false otherwise. */
 bool
 ss_ends_with (struct substring ss, struct substring suffix)
@@ -725,10 +781,20 @@ ss_cspan (struct substring ss, struct substring stop_set)
 size_t
 ss_find_byte (struct substring ss, char c)
 {
-  const char *p = memchr (ss.string, c, ss.length);
+  const char *p = memchr (ss.string, (int) c, ss.length);
   return p != NULL ? p - ss.string : SIZE_MAX;
 }
 
+/* Returns the offset in HAYSTACK of the first instance of NEEDLE,
+   or SIZE_MAX if NEEDLE does not occur in HAYSTACK. */
+size_t
+ss_find_substring (struct substring haystack, struct substring needle)
+{
+  const char *p = memmem (haystack.string, haystack.length,
+                          needle.string, needle.length);
+  return p != NULL ? p - haystack.string : SIZE_MAX;
+}
+
 /* Compares A and B and returns a strcmp()-type comparison
    result. */
 int
@@ -863,6 +929,22 @@ ss_at_mblen (struct substring s, size_t ofs)
   else
     return 0;
 }
+
+size_t
+ss_utf8_count_columns (struct substring s)
+{
+  return utf8_count_columns (s.string, s.length);
+}
+
+/* Returns a substring of S starting at 0-based display column START and
+   running for N display columns. */
+struct substring
+ss_utf8_columns (struct substring s, size_t start, size_t n)
+{
+  ss_advance (&s, utf8_columns_to_bytes (s.string, s.length, start));
+  s.length = utf8_columns_to_bytes (s.string, s.length, n);
+  return s;
+}
 
 /* Initializes ST as an empty string. */
 void
@@ -982,35 +1064,35 @@ ds_ss (const struct string *st)
   return st->ss;
 }
 
-/* Returns a substring that contains CNT bytes from ST
+/* Returns a substring that contains N bytes from ST
    starting at position START.
 
    If START is greater than or equal to the length of ST, then
-   the substring will be the empty string.  If START + CNT
+   the substring will be the empty string.  If START + N
    exceeds the length of ST, then the substring will only be
    ds_length(ST) - START bytes long. */
 struct substring
-ds_substr (const struct string *st, size_t start, size_t cnt)
+ds_substr (const struct string *st, size_t start, size_t n)
 {
-  return ss_substr (ds_ss (st), start, cnt);
+  return ss_substr (ds_ss (st), start, n);
 }
 
-/* Returns a substring that contains the first CNT bytes in
-   ST.  If CNT exceeds the length of ST, then the substring will
+/* Returns a substring that contains the first N bytes in
+   ST.  If N exceeds the length of ST, then the substring will
    contain all of ST. */
 struct substring
-ds_head (const struct string *st, size_t cnt)
+ds_head (const struct string *st, size_t n)
 {
-  return ss_head (ds_ss (st), cnt);
+  return ss_head (ds_ss (st), n);
 }
 
-/* Returns a substring that contains the last CNT bytes in
-   ST.  If CNT exceeds the length of ST, then the substring will
+/* Returns a substring that contains the last N bytes in
+   ST.  If N exceeds the length of ST, then the substring will
    contain all of ST. */
 struct substring
-ds_tail (const struct string *st, size_t cnt)
+ds_tail (const struct string *st, size_t n)
 {
-  return ss_tail (ds_ss (st), cnt);
+  return ss_tail (ds_ss (st), n);
 }
 
 /* Ensures that ST can hold at least MIN_CAPACITY bytes plus a null
@@ -1059,10 +1141,10 @@ ds_rtrim (struct string *st, struct substring trim_set)
 size_t
 ds_ltrim (struct string *st, struct substring trim_set)
 {
-  size_t cnt = ds_span (st, trim_set);
-  if (cnt > 0)
-    ds_assign_substring (st, ds_substr (st, cnt, SIZE_MAX));
-  return cnt;
+  size_t n = ds_span (st, trim_set);
+  if (n > 0)
+    ds_assign_substring (st, ds_substr (st, n, SIZE_MAX));
+  return n;
 }
 
 /* Trims leading and trailing bytes in TRIM_SET from ST.
@@ -1070,8 +1152,8 @@ ds_ltrim (struct string *st, struct substring trim_set)
 size_t
 ds_trim (struct string *st, struct substring trim_set)
 {
-  size_t cnt = ds_rtrim (st, trim_set);
-  return cnt + ds_ltrim (st, trim_set);
+  size_t n = ds_rtrim (st, trim_set);
+  return n + ds_ltrim (st, trim_set);
 }
 
 /* If the last byte in ST is C, removes it and returns true.
@@ -1420,15 +1502,15 @@ ds_read_config_line (struct string *st, int *line_number, FILE *stream)
   return true;
 }
 
-/* Attempts to read SIZE * CNT bytes from STREAM and append them
+/* Attempts to read SIZE * N bytes from STREAM and append them
    to ST.
    Returns true if all the requested data was read, false otherwise. */
 bool
-ds_read_stream (struct string *st, size_t size, size_t cnt, FILE *stream)
+ds_read_stream (struct string *st, size_t size, size_t n, FILE *stream)
 {
   if (size != 0)
     {
-      size_t try_bytes = xtimes (cnt, size);
+      size_t try_bytes = xtimes (n, size);
       if (size_in_bounds_p (xsum (ds_length (st), try_bytes)))
         {
           char *buffer = ds_put_uninit (st, try_bytes);
@@ -1458,7 +1540,8 @@ ds_put_cstr (struct string *st, const char *s)
 void
 ds_put_substring (struct string *st, struct substring ss)
 {
-  memcpy (ds_put_uninit (st, ss_length (ss)), ss_data (ss), ss_length (ss));
+  if (ss.length)
+    memcpy (ds_put_uninit (st, ss_length (ss)), ss_data (ss), ss_length (ss));
 }
 
 /* Returns ds_end(ST) and THEN increases the length by INCR. */
@@ -1488,6 +1571,9 @@ ds_splice_uninit (struct string *st,
     {
       if (new_len > old_len)
         ds_extend (st, ds_length (st) + (new_len - old_len));
+
+      assert (ds_length (st) >= ofs + old_len);
+
       memmove (ds_data (st) + (ofs + new_len),
                ds_data (st) + (ofs + old_len),
                ds_length (st) - (ofs + old_len));
@@ -1518,25 +1604,22 @@ ds_put_c_format (struct string *st, const char *format, ...)
   va_end (args);
 }
 
-
-/* Formats FORMAT as a printf string, using fmt_func (a snprintf like function) 
-   and appends the result to ST. */
-static void
-ds_put_vformat_int (struct string *st, const char *format, va_list args_,
-		    int (*fmt_func) (char *, size_t, const char *, va_list))
+/* Formats FORMAT as a printf string and appends the result to ST. */
+void
+ds_put_vformat (struct string *st, const char *format, va_list args_)
 {
   int avail, needed;
   va_list args;
 
   va_copy (args, args_);
   avail = st->ss.string != NULL ? st->capacity - st->ss.length + 1 : 0;
-  needed = fmt_func (st->ss.string + st->ss.length, avail, format, args);
+  needed = vsnprintf (st->ss.string + st->ss.length, avail, format, args);
   va_end (args);
 
   if (needed >= avail)
     {
       va_copy (args, args_);
-      fmt_func (ds_put_uninit (st, needed), needed + 1, format, args);
+      vsnprintf (ds_put_uninit (st, needed), needed + 1, format, args);
       va_end (args);
     }
   else
@@ -1549,34 +1632,27 @@ ds_put_vformat_int (struct string *st, const char *format, va_list args_,
           avail = st->capacity - st->ss.length + 1;
 
           va_copy (args, args_);
-          needed = fmt_func (ds_end (st), avail, format, args);
+          needed = vsnprintf (ds_end (st), avail, format, args);
           va_end (args);
         }
       st->ss.length += needed;
     }
 }
 
-
-static int
-vasnwrapper (char *str, size_t size,  const char *format, va_list ap)
-{
-  c_vasnprintf (str, &size, format, ap);
-  return size;
-}
-
-/* Formats FORMAT as a printf string and appends the result to ST. */
-void
-ds_put_vformat (struct string *st, const char *format, va_list args_)
-{
-  ds_put_vformat_int (st, format, args_, vsnprintf);
-}
-
-/* Formats FORMAT as a printf string, as if in the C locale, 
+/* Formats FORMAT as a printf string, as if in the C locale,
    and appends the result to ST. */
 void
-ds_put_c_vformat (struct string *st, const char *format, va_list args_)
+ds_put_c_vformat (struct string *st, const char *format, va_list args)
 {
-  ds_put_vformat_int (st, format, args_, vasnwrapper);
+  char buf[128];
+  size_t len = sizeof buf;
+  char *output = c_vasnprintf (buf, &len, format, args);
+  if (output)
+    {
+      ds_put_cstr (st, output);
+      if (output != buf)
+        free (output);
+    }
 }
 
 /* Appends byte CH to ST. */
@@ -1586,11 +1662,11 @@ ds_put_byte (struct string *st, int ch)
   ds_put_uninit (st, 1)[0] = ch;
 }
 
-/* Appends CNT copies of byte CH to ST. */
+/* Appends N copies of byte CH to ST. */
 void
-ds_put_byte_multiple (struct string *st, int ch, size_t cnt)
+ds_put_byte_multiple (struct string *st, int ch, size_t n)
 {
-  memset (ds_put_uninit (st, cnt), ch, cnt);
+  memset (ds_put_uninit (st, n), ch, n);
 }
 
 /* Appends Unicode code point UC to ST in UTF-8 encoding. */
@@ -1601,6 +1677,18 @@ ds_put_unichar (struct string *st, ucs4_t uc)
   st->ss.length += u8_uctomb (CHAR_CAST (uint8_t *, ds_end (st)), uc, 6);
 }
 
+/* Appends N copies of S to ST. */
+void
+ds_put_substring_multiple (struct string *dst, struct substring src, size_t n)
+{
+  char *p = ds_put_uninit (dst, n * src.length);
+  for (size_t i = 0; i < n; i++)
+    {
+      memcpy (p, src.string, src.length);
+      p += src.length;
+    }
+}
+
 /* If relocation has been enabled, replace ST,
    with its relocated version */
 void
@@ -1609,7 +1697,7 @@ ds_relocate (struct string *st)
   const char *orig = ds_cstr (st);
   const char *rel = relocate (orig);
 
-  if ( orig != rel)
+  if (orig != rel)
     {
       ds_clear (st);
       ds_put_cstr (st, rel);