aboutsummaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
authorGravatar Steven Schubiger <stsc@refcnt.org> 2015-09-30 16:14:11 +0200
committerGravatar Steven Schubiger <stsc@refcnt.org> 2015-09-30 16:14:11 +0200
commit7f313b1c22e61541aea2bcbe666639e05e69c9ad (patch)
tree18c7f55a23e8c53cbbcbc3270b5973b1c483472c
parent3582f0672895201b3c460c4aeec9fa7b8ee7480f (diff)
parentf77dfa416cafc1a9db80349b31e6cb82e7f0491d (diff)
downloadcolorize-7f313b1c22e61541aea2bcbe666639e05e69c9ad.tar.gz
colorize-7f313b1c22e61541aea2bcbe666639e05e69c9ad.tar.bz2
Merge branch 'merge_part_line'
When cleaning lines with --clean[-all], reduce allocating space for the partial line in memory. Previously, string concatenation may have lead to excessive use of memory.
-rw-r--r--colorize.c172
-rwxr-xr-xt/merge.t153
-rwxr-xr-xtest.pl10
3 files changed, 306 insertions, 29 deletions
diff --git a/colorize.c b/colorize.c
index 498e566..06ef8fb 100644
--- a/colorize.c
+++ b/colorize.c
@@ -104,6 +104,8 @@
&& (streq (color_names[color2]->name, "none") \
|| streq (color_names[color2]->name, "default")) \
+#define ALLOC_COMPLETE_PART_LINE 8
+
#define COLOR_SEP_CHAR '/'
#define DEBUG_FILE "debug.txt"
@@ -210,6 +212,10 @@ static void free_color_names (struct color_name **);
static void process_args (unsigned int, char **, bool *, const struct color **, const char **, FILE **);
static void process_file_arg (const char *, const char **, FILE **);
static void read_print_stream (bool, const struct color **, const char *, FILE *);
+static void merge_print_line (bool, const struct color **, const char *, const char *, FILE *);
+static void complete_part_line (const char *, char **, FILE *);
+static bool get_next_char (char *, const char **, FILE *, bool *);
+static void save_char (char, char **, unsigned long *, size_t *);
static void find_color_entries (struct color_name **, const struct color **);
static void find_color_entry (const struct color_name *, unsigned int, const struct color **);
static void print_line (bool, const struct color **, const char * const, unsigned int);
@@ -709,23 +715,10 @@ process_file_arg (const char *file_string, const char **file, FILE **stream)
assert (*file);
}
-#define MERGE_PRINT_LINE(part_line, line, flags, check_eof) do { \
- char *current_line, *merged_line = NULL; \
- if (part_line) \
- { \
- merged_line = str_concat (part_line, line); \
- free_null (part_line); \
- } \
- current_line = merged_line ? merged_line : (char *)line; \
- if (!check_eof || *current_line != '\0') \
- print_line (bold, colors, current_line, flags); \
- free (merged_line); \
-} while (false)
-
static void
read_print_stream (bool bold, const struct color **colors, const char *file, FILE *stream)
{
- char buf[BUF_SIZE + 1], *part_line = NULL;
+ char buf[BUF_SIZE + 1];
unsigned int flags = 0;
while (!feof (stream))
@@ -754,26 +747,151 @@ read_print_stream (bool bold, const struct color **colors, const char *file, FIL
vfprintf_fail (formats[FMT_FILE], file, "unrecognized line ending");
p = eol + SKIP_LINE_ENDINGS (flags);
*eol = '\0';
- MERGE_PRINT_LINE (part_line, line, flags, false);
+ print_line (bold, colors, line, flags);
line = p;
}
- if (feof (stream)) {
- MERGE_PRINT_LINE (part_line, line, 0, true);
- }
- else if (*line != '\0')
+ if (feof (stream))
{
- if (!clean && !clean_all) /* efficiency */
+ if (*line != '\0')
print_line (bold, colors, line, 0);
- else if (!part_line)
- part_line = xstrdup (line);
+ }
+ else if (*line != '\0')
+ {
+ char *p;
+ if ((clean || clean_all) && (p = strrchr (line, '\033')))
+ merge_print_line (bold, colors, line, p, stream);
else
- {
- char *merged_line = str_concat (part_line, line);
- free (part_line);
- part_line = merged_line;
- }
+ print_line (bold, colors, line, 0);
+ }
+ }
+}
+
+static void
+merge_print_line (bool bold, const struct color **colors, const char *line, const char *p, FILE *stream)
+{
+ char *buf = NULL;
+ char *merged_part_line = NULL;
+ const char *part_line;
+
+ complete_part_line (p + 1, &buf, stream);
+
+ if (buf)
+ part_line = merged_part_line = str_concat (line, buf);
+ else
+ part_line = line;
+ free (buf);
+
+#ifdef TEST_MERGE_PART_LINE
+ printf ("%s", part_line);
+ free (merged_part_line);
+ exit (EXIT_SUCCESS);
+#else
+ print_line (bold, colors, part_line, 0);
+ free (merged_part_line);
+#endif
+}
+
+static void
+complete_part_line (const char *p, char **buf, FILE *stream)
+{
+ bool got_next_char = false, read_from_stream;
+ char ch;
+ unsigned long i = 0;
+ size_t size;
+
+ if (get_next_char (&ch, &p, stream, &read_from_stream))
+ {
+ if (ch == '[')
+ {
+ if (read_from_stream)
+ save_char (ch, buf, &i, &size);
+ }
+ else
+ {
+ if (read_from_stream)
+ ungetc ((int)ch, stream);
+ return; /* cancel */
+ }
+ }
+ else
+ return; /* cancel */
+
+ while (get_next_char (&ch, &p, stream, &read_from_stream))
+ {
+ if (isdigit (ch) || ch == ';')
+ {
+ if (read_from_stream)
+ save_char (ch, buf, &i, &size);
+ }
+ else /* read next character */
+ {
+ got_next_char = true;
+ break;
+ }
+ }
+
+ if (got_next_char)
+ {
+ if (ch == 'm')
+ {
+ if (read_from_stream)
+ save_char (ch, buf, &i, &size);
+ }
+ else
+ {
+ if (read_from_stream)
+ ungetc ((int)ch, stream);
+ return; /* cancel */
+ }
+ }
+ else
+ return; /* cancel */
+}
+
+static bool
+get_next_char (char *ch, const char **p, FILE *stream, bool *read_from_stream)
+{
+ if (**p == '\0')
+ {
+ int c;
+ if ((c = fgetc (stream)) != EOF)
+ {
+ *ch = (char)c;
+ *read_from_stream = true;
+ return true;
+ }
+ else
+ {
+ *read_from_stream = false;
+ return false;
}
}
+ else
+ {
+ *ch = **p;
+ (*p)++;
+ *read_from_stream = false;
+ return true;
+ }
+}
+
+static void
+save_char (char ch, char **buf, unsigned long *i, size_t *size)
+{
+ if (!*buf)
+ {
+ *size = ALLOC_COMPLETE_PART_LINE;
+ *buf = xmalloc (*size);
+ }
+ /* +1: effective occupied size of buffer */
+ else if ((*i + 1) == *size)
+ {
+ *size *= 2;
+ *buf = xrealloc (*buf, *size);
+ }
+ (*buf)[*i] = ch;
+ (*buf)[*i + 1] = '\0';
+ (*i)++;
}
static void
diff --git a/t/merge.t b/t/merge.t
new file mode 100755
index 0000000..b5c6dc7
--- /dev/null
+++ b/t/merge.t
@@ -0,0 +1,153 @@
+#!/usr/bin/perl
+
+use strict;
+use warnings;
+use constant true => 1;
+use constant false => 0;
+
+use File::Temp qw(tmpnam);
+use Test::More;
+
+# sequence, buffer sizes
+my @merge_success = (
+ [ "\e[30m", [ 1..4 ] ],
+ [ "\e[31m", [ 1..4 ] ],
+ [ "\e[32m", [ 1..4 ] ],
+ [ "\e[33m", [ 1..4 ] ],
+ [ "\e[34m", [ 1..4 ] ],
+ [ "\e[35m", [ 1..4 ] ],
+ [ "\e[36m", [ 1..4 ] ],
+ [ "\e[37m", [ 1..4 ] ],
+ [ "\e[39m", [ 1..4 ] ],
+ [ "\e[1;30m", [ 1..6 ] ],
+ [ "\e[1;31m", [ 1..6 ] ],
+ [ "\e[1;32m", [ 1..6 ] ],
+ [ "\e[1;33m", [ 1..6 ] ],
+ [ "\e[1;34m", [ 1..6 ] ],
+ [ "\e[1;35m", [ 1..6 ] ],
+ [ "\e[1;36m", [ 1..6 ] ],
+ [ "\e[1;37m", [ 1..6 ] ],
+ [ "\e[1;39m", [ 1..6 ] ],
+ [ "\e[40m", [ 1..4 ] ],
+ [ "\e[41m", [ 1..4 ] ],
+ [ "\e[42m", [ 1..4 ] ],
+ [ "\e[43m", [ 1..4 ] ],
+ [ "\e[44m", [ 1..4 ] ],
+ [ "\e[45m", [ 1..4 ] ],
+ [ "\e[46m", [ 1..4 ] ],
+ [ "\e[47m", [ 1..4 ] ],
+ [ "\e[49m", [ 1..4 ] ],
+ [ "\e[0m", [ 1..3 ] ],
+ [ "\e[m", [ 1..2 ] ],
+ [ "\e[;;m", [ 1..4 ] ],
+ [ "\e[123456m", [ 1 ] ], # tightly coupled to ALLOC_COMPLETE_PART_LINE
+);
+# sequence, buffer size
+my @merge_fail = (
+ [ "\e30m", 1 ], # missing bracket
+ [ "\e[am", 2 ], # not a digit nor ; nor m
+);
+# sequence
+my @buffer = (
+ "\e[30mz",
+ "\e[31mz",
+ "\e[32mz",
+ "\e[33mz",
+ "\e[34mz",
+ "\e[35mz",
+ "\e[36mz",
+ "\e[37mz",
+ "\e[39mz",
+ "\e[1;30mz",
+ "\e[1;31mz",
+ "\e[1;32mz",
+ "\e[1;33mz",
+ "\e[1;34mz",
+ "\e[1;35mz",
+ "\e[1;36mz",
+ "\e[1;37mz",
+ "\e[1;39mz",
+ "\e[40mz",
+ "\e[41mz",
+ "\e[42mz",
+ "\e[43mz",
+ "\e[44mz",
+ "\e[45mz",
+ "\e[46mz",
+ "\e[47mz",
+ "\e[49mz",
+ "\e[0mz",
+ "\e[mz",
+ "\e[;;mz",
+);
+# sequence, buffer size
+my @pushback = (
+ [ "\ezm", 1 ],
+ [ "\e[z", 2 ],
+);
+
+my $tests = 0;
+foreach (@merge_success) {
+ $tests += @{$_->[1]};
+}
+$tests += @merge_fail;
+$tests += @buffer;
+$tests += @pushback;
+
+my $source = 'colorize.c';
+my %programs;
+
+my $compile = sub
+{
+ my ($buf_size) = @_;
+ return true if exists $programs{$buf_size};
+ my $program = tmpnam();
+ return false unless system("gcc -DTEST_MERGE_PART_LINE -DBUF_SIZE=$buf_size -o $program $source") == 0;
+ $programs{$buf_size} = $program;
+ return true; # compiling succeeded
+};
+
+my $test_name = sub
+{
+ my ($sequence, $buf_size) = @_;
+ my $substr = substr($sequence, 0, $buf_size);
+ $substr =~ s/^\e/ESC/;
+ $sequence =~ s/^\e/ESC/;
+ return "$sequence: $substr";
+};
+
+plan tests => $tests;
+
+foreach my $test (@merge_success) {
+ foreach my $buf_size (@{$test->[1]}) {
+ SKIP: {
+ skip 'compiling failed (merge part line)', 1 unless $compile->($buf_size);
+ ok(qx(echo -n "$test->[0]" | $programs{$buf_size} --clean) eq $test->[0], 'merge success: ' . $test_name->($test->[0], $buf_size));
+ }
+ }
+}
+foreach my $test (@merge_fail) {
+ my $buf_size = $test->[1];
+ SKIP: {
+ skip 'compiling failed (merge part line)', 1 unless $compile->($buf_size);
+ ok(qx(echo -n "$test->[0]" | $programs{$buf_size} --clean) eq substr($test->[0], 0, $buf_size), 'merge fail: ' . $test_name->($test->[0], $buf_size));
+ }
+}
+foreach my $test (@buffer) {
+ my $buf_size = length($test) - 1;
+ SKIP: {
+ skip 'compiling failed (merge part line)', 1 unless $compile->($buf_size);
+ ok(qx(echo -n "$test" | $programs{$buf_size} --clean) eq substr($test, 0, $buf_size), 'buffer: ' . $test_name->($test, $buf_size));
+ }
+}
+foreach my $test (@pushback) {
+ my $buf_size = $test->[1];
+ SKIP: {
+ my $program = tmpnam();
+ skip 'compiling failed (merge part line)', 1 unless system("gcc -DBUF_SIZE=$buf_size -o $program $source") == 0;
+ ok(qx(echo -n "$test->[0]" | $program --clean) eq $test->[0], 'pushback: ' . $test_name->($test->[0], $buf_size));
+ unlink $program;
+ }
+}
+
+unlink $programs{$_} foreach keys %programs;
diff --git a/test.pl b/test.pl
index 959d83f..30c1fd1 100755
--- a/test.pl
+++ b/test.pl
@@ -8,6 +8,7 @@ use constant false => 0;
use File::Temp qw(tempfile tempdir tmpnam);
use IPC::Open3 qw(open3);
use Symbol qw(gensym);
+use Test::Harness qw(runtests);
use Test::More;
my $tests = 25;
@@ -30,6 +31,11 @@ my $write_to_tmpfile = sub
return $tmpfile;
};
+{
+ my @test_files = glob('t/*.t');
+ runtests(@test_files);
+}
+
plan tests => $tests;
SKIP: {
@@ -139,9 +145,9 @@ SKIP: {
my $switch = "--$type";
- # Check that line chunks are merged when cleaning text
+ # Check that line chunks are printed when cleaning text without sequences
my $short_text = 'Linux dev 2.6.32-5-openvz-686 #1 SMP Sun Sep 23 11:40:07 UTC 2012 i686 GNU/Linux';
- is(qx(echo -n "$short_text" | $program_buf $switch), $short_text, "merge ${\length $short_text} bytes (BUF_SIZE=$BUF_SIZE{short}, $type)");
+ is(qx(echo -n "$short_text" | $program_buf $switch), $short_text, "print ${\length $short_text} bytes (BUF_SIZE=$BUF_SIZE{short}, $type)");
};
SKIP: {