checksrc

checksrc.c (6544B)

// SPDX-FileCopyrightText: 2017 Haelwenn (lanodan) Monnier <contact+checksrc@hacktivis.me>
// SPDX-License-Identifier: MPL-2.0
#define _DEFAULT_SOURCE
#include <ctype.h>
#include <dirent.h>
#include <errno.h>
#include <fcntl.h>
#include <fnmatch.h>
#include <limits.h> // PATH_MAX
#include <stdbool.h>
#include <stdint.h> // uint8_t
#include <stdio.h>
#include <string.h>
#include <unistd.h>
#if 0
static const char *generated[] = {
	"do not modify",
	"do not edit",
	"was generated",
	"is generated",
	"generated by",
	"generated from",
	"generated using",
	"automatically generated",
	"generated automatically",
	"machine generated",
	"autogenerated",
	"generated code"
	"produced by", // makeinfo
	"produced from",
	"produced using",
};
// Check what yacc (original), oyacc, heirloom yacc & lex produces
// Maybe this table should be passed to lex for efficiency
#endif
bool verbose = false;
#define EXCLUDE_MAX 400
#define EXCLUDE_MAX_STR "400"
static const char *exclude[400];
static int exclude_len;
static int
checkfile(const char *fname)
{
	int fd = open(fname, O_RDONLY | O_NOCTTY);
	if(fd < 0)
	{
		fprintf(stderr, "checksrc: error: Failed opening file '%s': %s\n", fname, strerror(errno));
		return -1;
	}
	size_t lines_len[10];
	size_t chars = 0, line = 0;
	size_t last_dump = 0;
	unsigned int dumps = 0;
	off_t pos = 0;
	for(;;)
	{
		static uint8_t buf[4096];
		unsigned int puncts = 0, numbers = 0, letters = 0;
		int nread = read(fd, &buf, 4096);
		if(nread == 0) break;
		if(nread < 0)
		{
			fprintf(
			    stderr, "checksrc: error: Failed reading from file '%s': %s\n", fname, strerror(errno));
			close(fd);
			return -1;
		}
		for(size_t i = 0; i < nread; i++)
		{
			if(buf[i] == '\r') continue;
			if(buf[i] == '\t')
			{
				chars++;
				lines_len[line % 10]++;
				continue;
			}
			if(buf[i] == '\n')
			{
				if(line - last_dump > 10)
				{
					size_t maxdiff = 0;
					for(size_t li = 1; li < 10; li++)
					{
						size_t diff = 0;
						if(lines_len[li - 1] > lines_len[li])
							diff = lines_len[li - 1] - lines_len[li];
						else
							diff = lines_len[li] - lines_len[li - 1];
						if(diff > maxdiff) maxdiff = diff;
					}
					if(maxdiff < 3)
					{
						last_dump = line;
						printf("%s:[%zu - %zu] line length varied only by a maximum of %zu characters\n",
						       fname,
						       line - 10,
						       line,
						       maxdiff);
						if(++dumps == 3)
						{
							printf("%s: Found 3 dumps, stopping there\n", fname);
							close(fd);
							return 1;
						}
					}
				}
				line++;
				lines_len[line % 10] = 0;
				continue;
			}
			/* ('\e', ESC) */
			if(buf[i] == 0x1B) continue;
			/* 0x0 and obscure control characters. With \r, \t, \n, \e taken care off earlier */
			if(buf[i] < ' ')
			{
				printf("%s: binary (byte 0x%X found at position 0x%zX)\n", fname, buf[i], i);
				close(fd);
				return 1;
			}
			// somewhat UTF-8 centric
			if(buf[i] < 0x7F)
			{
				chars++;
				lines_len[line % 10]++;
				if(buf[i] == ' ') continue;
				if(isdigit(buf[i]))
					numbers++;
				else if(isalpha(buf[i]))
					letters++;
				else
					puncts++;
			}
		}
		if(puncts + numbers > letters)
		{
			printf("%s (block 0x%lX to 0x%lX): Got more punctuation(%d) + numbers(%d) than letters(%d)\n",
			       fname,
			       pos,
			       pos + nread,
			       puncts,
			       numbers,
			       letters);
		}
		pos += nread;
	}
	// TODO: Check for a "generated" string
	close(fd);
	if(chars > 200)
	{
		size_t lineavg = chars;
		if(line > 0) lineavg /= line;
		// Consider that lines are on average shorter than 100 character
		// One false-positive being formats like SVG, fine for this tool
		if(lineavg > 100)
		{
			printf("%s: minified (%zd characters / %zd newlines = %zd)\n", fname, chars, line, lineavg);
			return 1;
		}
	}
	if(verbose) printf("%s: OK\n", fname);
	return 0;
}
static int
exclude_match(char *path)
{
	for(int ei = 0; ei < exclude_len; ei++)
		if(fnmatch(exclude[ei], path, 0) == 0) return 0;
	return 1;
}
static int
checkdir(const char *dirname)
{
	int ret = 0;
	int dirfd = open(dirname, O_RDONLY | O_DIRECTORY);
	if(dirfd < 0)
	{
		fprintf(
		    stderr, "checksrc: error: Failed opening directory '%s': %s\n", dirname, strerror(errno));
		return 1;
	}
	for(;;)
	{
#define DENTSSIZ 10240
		char buf[DENTSSIZ];
		// Sadly posix_getdents is still a bit too recent
		ssize_t nread = getdents(dirfd, buf, DENTSSIZ);
		if(nread == 0) break;
		if(nread < 0)
		{
			fprintf(
			    stderr, "checksrc: error: Failed reading directory '%s': %s\n", dirname, strerror(errno));
			close(dirfd);
			return 1;
		}
		for(ssize_t bpos = 0; bpos < nread;)
		{
			struct dirent *dent = (void *)&buf[bpos];
			bpos += dent->d_reclen;
			if(strcmp(dent->d_name, ".") == 0) continue;
			if(strcmp(dent->d_name, "..") == 0) continue;
			if(dent->d_type == DT_DIR || dent->d_type == DT_REG)
			{
				char path[PATH_MAX] = "";
				if(snprintf(path, PATH_MAX, "%s/%s", dirname, dent->d_name) < 0)
				{
					fprintf(stderr,
					        "checksrc: error: Failed joining path in directory '%s': %s\n",
					        dirname,
					        strerror(errno));
					return 1;
				}
				if(exclude_match(path) == 0) continue;
				int err = 0;
				switch(dent->d_type)
				{
				case DT_DIR:
					err = checkdir(path);
					break;
				case DT_REG:
					err = checkfile(path);
					break;
				}
				if(err < 0) return 1;
				if(err > 0) ret = 1 + err;
			}
		}
	}
	close(dirfd);
	return ret;
}
static void
usage(void)
{
	fputs("Usage: checksrc [-v] [-d workdir] [-e excluded_path ...]\n", stderr);
}
int
main(int argc, char *argv[])
{
	char *workdir = NULL;
	for(int c = -1; (c = getopt(argc, argv, "vd:e:")) != -1;)
	{
		switch(c)
		{
		case 'v':
			verbose = true;
			break;
		case 'd':
			if(workdir)
			{
				fputs("checksrc: error: Option -d can only be passed once\n", stderr);
				usage();
				return 1;
			}
			workdir = optarg;
			break;
		case 'e':
			if(exclude_len == EXCLUDE_MAX)
			{
				fputs("checksrc: error: Can only use '-e' option " EXCLUDE_MAX_STR " times\n", stderr);
				return 1;
			}
			exclude[exclude_len++] = optarg;
			break;
		}
	}
	argc -= optind;
	argv += optind;
	if(argc > 0)
	{
		fprintf(stderr, "checksrc: error: Expected 0 arguments, got %d\n", argc);
		return 1;
	}
	if(workdir)
	{
		if(chdir(workdir) != 0)
		{
			fprintf(stderr,
			        "checksrc: error: Failed changing into directory '%s': %s\n",
			        workdir,
			        strerror(errno));
			return 1;
		}
	}
	return checkdir(".");
}