19502a77ea0a16c395407f31af0bbb0afd0680adbElliott Hughes/* file.c - describe file type
29502a77ea0a16c395407f31af0bbb0afd0680adbElliott Hughes *
39502a77ea0a16c395407f31af0bbb0afd0680adbElliott Hughes * Copyright 2016 The Android Open Source Project
49502a77ea0a16c395407f31af0bbb0afd0680adbElliott Hughes *
59502a77ea0a16c395407f31af0bbb0afd0680adbElliott Hughes * See http://pubs.opengroup.org/onlinepubs/9699919799/utilities/file.html
69502a77ea0a16c395407f31af0bbb0afd0680adbElliott Hughes
7493dc573825a862ee533c8aaf91ab23112b591c2Elliott HughesUSE_FILE(NEWTOY(file, "<1hL[!hL]", TOYFLAG_USR|TOYFLAG_BIN))
89502a77ea0a16c395407f31af0bbb0afd0680adbElliott Hughes
99502a77ea0a16c395407f31af0bbb0afd0680adbElliott Hughesconfig FILE
109502a77ea0a16c395407f31af0bbb0afd0680adbElliott Hughes  bool "file"
114e21ddd440f5d1381c96ae6465a12888dcb23446Rob Landley  default y
129502a77ea0a16c395407f31af0bbb0afd0680adbElliott Hughes  help
13493dc573825a862ee533c8aaf91ab23112b591c2Elliott Hughes    usage: file [-hL] [file...]
149502a77ea0a16c395407f31af0bbb0afd0680adbElliott Hughes
159502a77ea0a16c395407f31af0bbb0afd0680adbElliott Hughes    Examine the given files and describe their content types.
16493dc573825a862ee533c8aaf91ab23112b591c2Elliott Hughes
17493dc573825a862ee533c8aaf91ab23112b591c2Elliott Hughes    -h	don't follow symlinks (default)
18493dc573825a862ee533c8aaf91ab23112b591c2Elliott Hughes    -L	follow symlinks
199502a77ea0a16c395407f31af0bbb0afd0680adbElliott Hughes*/
209502a77ea0a16c395407f31af0bbb0afd0680adbElliott Hughes
219502a77ea0a16c395407f31af0bbb0afd0680adbElliott Hughes#define FOR_file
229502a77ea0a16c395407f31af0bbb0afd0680adbElliott Hughes#include "toys.h"
239502a77ea0a16c395407f31af0bbb0afd0680adbElliott Hughes
249502a77ea0a16c395407f31af0bbb0afd0680adbElliott HughesGLOBALS(
259502a77ea0a16c395407f31af0bbb0afd0680adbElliott Hughes  int max_name_len;
269502a77ea0a16c395407f31af0bbb0afd0680adbElliott Hughes)
279502a77ea0a16c395407f31af0bbb0afd0680adbElliott Hughes
28cc4a1f4d520f090bd563647df2614d09b66bd54aRob Landley// We don't trust elf.h to be there, and two codepaths for 32/64 is awkward
29cc4a1f4d520f090bd563647df2614d09b66bd54aRob Landley// anyway, so calculate struct offsets manually. (It's a fixed ABI.)
30c5e7ee27f606ed0bdb20d54e9de1a4adb56fc5d3Elliott Hughesstatic void do_elf_file(int fd, struct stat *sb)
319502a77ea0a16c395407f31af0bbb0afd0680adbElliott Hughes{
32cc4a1f4d520f090bd563647df2614d09b66bd54aRob Landley  int endian = toybuf[5], bits = toybuf[4], i, j;
330332b60f62d71da61634cc69416129dfc64dced8Rob Landley  int64_t (*elf_int)(void *ptr, unsigned size) = peek_le;
340332b60f62d71da61634cc69416129dfc64dced8Rob Landley  // Values from include/linux/elf-em.h (plus arch/*/include/asm/elf.h)
3544c7958214e1690e820aaddf808267e0bf78df46Rob Landley  // Names are linux/arch/ directory (sometimes before 32/64 bit merges)
3644c7958214e1690e820aaddf808267e0bf78df46Rob Landley  struct {int val; char *name;} type[] = {{0x9026, "alpha"}, {93, "arc"},
3744c7958214e1690e820aaddf808267e0bf78df46Rob Landley    {195, "arcv2"}, {40, "arm"}, {183, "arm64"}, {0x18ad, "avr32"},
3844c7958214e1690e820aaddf808267e0bf78df46Rob Landley    {106, "blackfin"}, {140, "c6x"}, {23, "cell"}, {76, "cris"},
3944c7958214e1690e820aaddf808267e0bf78df46Rob Landley    {0x5441, "frv"}, {46, "h8300"}, {164, "hexagon"}, {50, "ia64"},
4044c7958214e1690e820aaddf808267e0bf78df46Rob Landley    {88, "m32r"}, {0x9041, "m32r"}, {4, "m68k"}, {174, "metag"},
4144c7958214e1690e820aaddf808267e0bf78df46Rob Landley    {0xbaab, "microblaze"}, {8, "mips"}, {10, "mips-old"}, {89, "mn10300"},
424b141e67eefc396687f22f8c136267c0f8081e05Rob Landley    {0xbeef, "mn10300-old"}, {113, "nios2"}, {92, "openrisc"},
4344c7958214e1690e820aaddf808267e0bf78df46Rob Landley    {0x8472, "openrisc-old"}, {15, "parisc"}, {20, "ppc"}, {21, "ppc64"},
4444c7958214e1690e820aaddf808267e0bf78df46Rob Landley    {22, "s390"}, {0xa390, "s390-old"}, {135, "score"}, {42, "sh"},
4544c7958214e1690e820aaddf808267e0bf78df46Rob Landley    {2, "sparc"}, {18, "sparc8+"}, {43, "sparc9"}, {188, "tile"},
4644c7958214e1690e820aaddf808267e0bf78df46Rob Landley    {191, "tilegx"}, {3, "386"}, {6, "486"}, {62, "x86-64"}, {94, "xtensa"},
4744c7958214e1690e820aaddf808267e0bf78df46Rob Landley    {0xabc7, "xtensa-old"}
4844c7958214e1690e820aaddf808267e0bf78df46Rob Landley  };
49c5e7ee27f606ed0bdb20d54e9de1a4adb56fc5d3Elliott Hughes  int dynamic = 0;
50c5e7ee27f606ed0bdb20d54e9de1a4adb56fc5d3Elliott Hughes  int stripped = 1;
51c5e7ee27f606ed0bdb20d54e9de1a4adb56fc5d3Elliott Hughes  char *map;
52c5e7ee27f606ed0bdb20d54e9de1a4adb56fc5d3Elliott Hughes  off_t phoff, shoff;
53fa84155632bf1c65f455b9872ee35a53cd0d2a19Elliott Hughes  int phentsize, phnum, shsize, shnum;
540332b60f62d71da61634cc69416129dfc64dced8Rob Landley
552e4aff2fd31fb9d949d6c47da1e46b97ecc23d24Rob Landley  printf("ELF ");
562e4aff2fd31fb9d949d6c47da1e46b97ecc23d24Rob Landley
572e4aff2fd31fb9d949d6c47da1e46b97ecc23d24Rob Landley  // executable (ELF says this is short but reality says byte, not MSB swapped)
582e4aff2fd31fb9d949d6c47da1e46b97ecc23d24Rob Landley  i = toybuf[16];
592e4aff2fd31fb9d949d6c47da1e46b97ecc23d24Rob Landley  if (i == 1) printf("relocatable");
602e4aff2fd31fb9d949d6c47da1e46b97ecc23d24Rob Landley  else if (i == 2) printf("executable");
612e4aff2fd31fb9d949d6c47da1e46b97ecc23d24Rob Landley  else if (i == 3) printf("shared object");
622e4aff2fd31fb9d949d6c47da1e46b97ecc23d24Rob Landley  else if (i == 4) printf("core dump");
632e4aff2fd31fb9d949d6c47da1e46b97ecc23d24Rob Landley  else printf("(bad type %d)", i);
642e4aff2fd31fb9d949d6c47da1e46b97ecc23d24Rob Landley  printf(", ");
659502a77ea0a16c395407f31af0bbb0afd0680adbElliott Hughes
669502a77ea0a16c395407f31af0bbb0afd0680adbElliott Hughes  // "64-bit"
672e4aff2fd31fb9d949d6c47da1e46b97ecc23d24Rob Landley  if (bits == 1) printf("32-bit ");
682e4aff2fd31fb9d949d6c47da1e46b97ecc23d24Rob Landley  else if (bits == 2) printf("64-bit ");
69cc4a1f4d520f090bd563647df2614d09b66bd54aRob Landley  else {
702e4aff2fd31fb9d949d6c47da1e46b97ecc23d24Rob Landley    printf("(bad class %d) ", bits);
71cc4a1f4d520f090bd563647df2614d09b66bd54aRob Landley    bits = 0;
72cc4a1f4d520f090bd563647df2614d09b66bd54aRob Landley  }
73cc4a1f4d520f090bd563647df2614d09b66bd54aRob Landley
749502a77ea0a16c395407f31af0bbb0afd0680adbElliott Hughes  // "LSB"
752e4aff2fd31fb9d949d6c47da1e46b97ecc23d24Rob Landley  if (endian == 1) printf("LSB ");
76cc4a1f4d520f090bd563647df2614d09b66bd54aRob Landley  else if (endian == 2) {
772e4aff2fd31fb9d949d6c47da1e46b97ecc23d24Rob Landley    printf("MSB ");
780332b60f62d71da61634cc69416129dfc64dced8Rob Landley    elf_int = peek_be;
790332b60f62d71da61634cc69416129dfc64dced8Rob Landley  } else {
802e4aff2fd31fb9d949d6c47da1e46b97ecc23d24Rob Landley    printf("(bad endian %d) \n", endian);
81cc4a1f4d520f090bd563647df2614d09b66bd54aRob Landley    endian = 0;
829502a77ea0a16c395407f31af0bbb0afd0680adbElliott Hughes  }
839502a77ea0a16c395407f31af0bbb0afd0680adbElliott Hughes
842e4aff2fd31fb9d949d6c47da1e46b97ecc23d24Rob Landley  // e_machine, ala "x86", from big table above
852e4aff2fd31fb9d949d6c47da1e46b97ecc23d24Rob Landley  j = elf_int(toybuf+18, 2);
862e4aff2fd31fb9d949d6c47da1e46b97ecc23d24Rob Landley  for (i = 0; i<ARRAY_LEN(type); i++) if (j==type[i].val) break;
872e4aff2fd31fb9d949d6c47da1e46b97ecc23d24Rob Landley  if (i<ARRAY_LEN(type)) printf("%s", type[i].name);
882e4aff2fd31fb9d949d6c47da1e46b97ecc23d24Rob Landley  else printf("(unknown arch %d)", j);
89cc4a1f4d520f090bd563647df2614d09b66bd54aRob Landley
90cc4a1f4d520f090bd563647df2614d09b66bd54aRob Landley  bits--;
91c5e7ee27f606ed0bdb20d54e9de1a4adb56fc5d3Elliott Hughes  // If what we've seen so far doesn't seem consistent, bail.
92fa84155632bf1c65f455b9872ee35a53cd0d2a19Elliott Hughes  if (!((bits&1)==bits && endian)) {
93c5e7ee27f606ed0bdb20d54e9de1a4adb56fc5d3Elliott Hughes    printf(", corrupt?\n");
94c5e7ee27f606ed0bdb20d54e9de1a4adb56fc5d3Elliott Hughes    return;
95c5e7ee27f606ed0bdb20d54e9de1a4adb56fc5d3Elliott Hughes  }
96c5e7ee27f606ed0bdb20d54e9de1a4adb56fc5d3Elliott Hughes
97c5e7ee27f606ed0bdb20d54e9de1a4adb56fc5d3Elliott Hughes  // Stash what we need from the header; it's okay to reuse toybuf after this.
98fa84155632bf1c65f455b9872ee35a53cd0d2a19Elliott Hughes  phentsize = elf_int(toybuf+42+12*bits, 2);
99c5e7ee27f606ed0bdb20d54e9de1a4adb56fc5d3Elliott Hughes  phnum = elf_int(toybuf+44+12*bits, 2);
100c5e7ee27f606ed0bdb20d54e9de1a4adb56fc5d3Elliott Hughes  phoff = elf_int(toybuf+28+4*bits, 4+4*bits);
101c5e7ee27f606ed0bdb20d54e9de1a4adb56fc5d3Elliott Hughes  shsize = elf_int(toybuf+46+12*bits, 2);
102c5e7ee27f606ed0bdb20d54e9de1a4adb56fc5d3Elliott Hughes  shnum = elf_int(toybuf+48+12*bits, 2);
103c5e7ee27f606ed0bdb20d54e9de1a4adb56fc5d3Elliott Hughes  shoff = elf_int(toybuf+32+8*bits, 4+4*bits);
104c5e7ee27f606ed0bdb20d54e9de1a4adb56fc5d3Elliott Hughes
105fa84155632bf1c65f455b9872ee35a53cd0d2a19Elliott Hughes  // With binutils, phentsize seems to only be non-zero if phnum is non-zero.
106fa84155632bf1c65f455b9872ee35a53cd0d2a19Elliott Hughes  // Such ELF files are rare, but do exist. (Android's crtbegin files, say.)
107fa84155632bf1c65f455b9872ee35a53cd0d2a19Elliott Hughes  if (phnum && (phentsize != 32+24*bits)) {
108fa84155632bf1c65f455b9872ee35a53cd0d2a19Elliott Hughes    printf(", corrupt phentsize %d?\n", phentsize);
109fa84155632bf1c65f455b9872ee35a53cd0d2a19Elliott Hughes    return;
110fa84155632bf1c65f455b9872ee35a53cd0d2a19Elliott Hughes  }
111fa84155632bf1c65f455b9872ee35a53cd0d2a19Elliott Hughes
11212f0744f340746b3a1c1accfdf993d6548f33e56Elliott Hughes  map = xmmap(0, sb->st_size, PROT_READ, MAP_SHARED, fd, 0);
113c5e7ee27f606ed0bdb20d54e9de1a4adb56fc5d3Elliott Hughes
114fa84155632bf1c65f455b9872ee35a53cd0d2a19Elliott Hughes  // We need to read the phdrs for dynamic vs static.
115c5e7ee27f606ed0bdb20d54e9de1a4adb56fc5d3Elliott Hughes  // (Note: fields got reordered for 64 bit)
116c5e7ee27f606ed0bdb20d54e9de1a4adb56fc5d3Elliott Hughes  for (i = 0; i<phnum; i++) {
117fa84155632bf1c65f455b9872ee35a53cd0d2a19Elliott Hughes    char *phdr = map+phoff+i*phentsize;
118c5e7ee27f606ed0bdb20d54e9de1a4adb56fc5d3Elliott Hughes    int p_type = elf_int(phdr, 4);
119c5e7ee27f606ed0bdb20d54e9de1a4adb56fc5d3Elliott Hughes    long long p_offset, p_filesz;
120c5e7ee27f606ed0bdb20d54e9de1a4adb56fc5d3Elliott Hughes
121c5e7ee27f606ed0bdb20d54e9de1a4adb56fc5d3Elliott Hughes    if (p_type==2 /*PT_DYNAMIC*/) dynamic = 1;
122c5e7ee27f606ed0bdb20d54e9de1a4adb56fc5d3Elliott Hughes    if (p_type!=3 /*PT_INTERP*/ && p_type!=4 /*PT_NOTE*/) continue;
123c5e7ee27f606ed0bdb20d54e9de1a4adb56fc5d3Elliott Hughes
124c5e7ee27f606ed0bdb20d54e9de1a4adb56fc5d3Elliott Hughes    j = bits+1;
125c5e7ee27f606ed0bdb20d54e9de1a4adb56fc5d3Elliott Hughes    p_offset = elf_int(phdr+4*j, 4*j);
126c5e7ee27f606ed0bdb20d54e9de1a4adb56fc5d3Elliott Hughes    p_filesz = elf_int(phdr+16*j, 4*j);
127c5e7ee27f606ed0bdb20d54e9de1a4adb56fc5d3Elliott Hughes
128c5e7ee27f606ed0bdb20d54e9de1a4adb56fc5d3Elliott Hughes    if (p_type==3 /*PT_INTERP*/)
129c5e7ee27f606ed0bdb20d54e9de1a4adb56fc5d3Elliott Hughes      printf(", dynamic (%.*s)", (int)p_filesz, map+p_offset);
130fa84155632bf1c65f455b9872ee35a53cd0d2a19Elliott Hughes  }
131fa84155632bf1c65f455b9872ee35a53cd0d2a19Elliott Hughes  if (!dynamic) printf(", static");
132fa84155632bf1c65f455b9872ee35a53cd0d2a19Elliott Hughes
133fa84155632bf1c65f455b9872ee35a53cd0d2a19Elliott Hughes  // We need to read the shdrs for stripped/unstripped and any notes.
134fa84155632bf1c65f455b9872ee35a53cd0d2a19Elliott Hughes  // Notes are in program headers *and* section headers, but some files don't
135fa84155632bf1c65f455b9872ee35a53cd0d2a19Elliott Hughes  // contain program headers, so we prefer to check here.
136fa84155632bf1c65f455b9872ee35a53cd0d2a19Elliott Hughes  // (Note: fields got reordered for 64 bit)
137fa84155632bf1c65f455b9872ee35a53cd0d2a19Elliott Hughes  for (i = 0; i<shnum; i++) {
138fa84155632bf1c65f455b9872ee35a53cd0d2a19Elliott Hughes    char *shdr = map+shoff+i*shsize;
139fa84155632bf1c65f455b9872ee35a53cd0d2a19Elliott Hughes    int sh_type = elf_int(shdr+4, 4);
140fa84155632bf1c65f455b9872ee35a53cd0d2a19Elliott Hughes    long sh_offset = elf_int(shdr+8+8*(bits+1), 4*(bits+1));
141fa84155632bf1c65f455b9872ee35a53cd0d2a19Elliott Hughes    int sh_size = elf_int(shdr+8+12*(bits+1), 4);
142fa84155632bf1c65f455b9872ee35a53cd0d2a19Elliott Hughes
143fa84155632bf1c65f455b9872ee35a53cd0d2a19Elliott Hughes    if (sh_type == 2 /*SHT_SYMTAB*/) {
144fa84155632bf1c65f455b9872ee35a53cd0d2a19Elliott Hughes      stripped = 0;
145fa84155632bf1c65f455b9872ee35a53cd0d2a19Elliott Hughes      break;
146fa84155632bf1c65f455b9872ee35a53cd0d2a19Elliott Hughes    } else if (sh_type == 7 /*SHT_NOTE*/) {
147fa84155632bf1c65f455b9872ee35a53cd0d2a19Elliott Hughes      char *note = map+sh_offset;
148c5e7ee27f606ed0bdb20d54e9de1a4adb56fc5d3Elliott Hughes
149fa84155632bf1c65f455b9872ee35a53cd0d2a19Elliott Hughes      // An ELF note is a sequence of entries, each consisting of an
150c5e7ee27f606ed0bdb20d54e9de1a4adb56fc5d3Elliott Hughes      // ndhr followed by n_namesz+n_descsz bytes of data (each of those
151c5e7ee27f606ed0bdb20d54e9de1a4adb56fc5d3Elliott Hughes      // rounded up to the next 4 bytes, without this being reflected in
152c5e7ee27f606ed0bdb20d54e9de1a4adb56fc5d3Elliott Hughes      // the header byte counts themselves).
153fa84155632bf1c65f455b9872ee35a53cd0d2a19Elliott Hughes      while (sh_size >= 3*4) { // Don't try to read a truncated entry.
154c5e7ee27f606ed0bdb20d54e9de1a4adb56fc5d3Elliott Hughes        int n_namesz = elf_int(note, 4);
155c5e7ee27f606ed0bdb20d54e9de1a4adb56fc5d3Elliott Hughes        int n_descsz = elf_int(note+4, 4);
156c5e7ee27f606ed0bdb20d54e9de1a4adb56fc5d3Elliott Hughes        int n_type = elf_int(note+8, 4);
157c5e7ee27f606ed0bdb20d54e9de1a4adb56fc5d3Elliott Hughes        int notesz = 3*4 + ((n_namesz+3)&~3) + ((n_descsz+3)&~3);
158c5e7ee27f606ed0bdb20d54e9de1a4adb56fc5d3Elliott Hughes
159c5e7ee27f606ed0bdb20d54e9de1a4adb56fc5d3Elliott Hughes        if (n_namesz==4 && !memcmp(note+12, "GNU", 4)) {
160fa84155632bf1c65f455b9872ee35a53cd0d2a19Elliott Hughes          if (n_type==3 /*NT_GNU_BUILD_ID*/) {
1618a34cbb5666a80153ea06ba8d59f7406f311bb7aElliott Hughes            printf(", BuildID=");
162c5e7ee27f606ed0bdb20d54e9de1a4adb56fc5d3Elliott Hughes            for (j = 0; j < n_descsz; ++j) printf("%02x", note[16 + j]);
163c5e7ee27f606ed0bdb20d54e9de1a4adb56fc5d3Elliott Hughes          }
164c5e7ee27f606ed0bdb20d54e9de1a4adb56fc5d3Elliott Hughes        } else if (n_namesz==8 && !memcmp(note+12, "Android", 8)) {
165fa84155632bf1c65f455b9872ee35a53cd0d2a19Elliott Hughes          if (n_type==1 /*.android.note.ident*/) {
166fa84155632bf1c65f455b9872ee35a53cd0d2a19Elliott Hughes            printf(", for Android %d", (int)elf_int(note+20, 4));
167fa84155632bf1c65f455b9872ee35a53cd0d2a19Elliott Hughes            if (n_descsz > 24)
168fa84155632bf1c65f455b9872ee35a53cd0d2a19Elliott Hughes              printf(", built by NDK %.64s (%.64s)", note+24, note+24+64);
169fa84155632bf1c65f455b9872ee35a53cd0d2a19Elliott Hughes          }
170c5e7ee27f606ed0bdb20d54e9de1a4adb56fc5d3Elliott Hughes        }
171c5e7ee27f606ed0bdb20d54e9de1a4adb56fc5d3Elliott Hughes
172c5e7ee27f606ed0bdb20d54e9de1a4adb56fc5d3Elliott Hughes        note += notesz;
173fa84155632bf1c65f455b9872ee35a53cd0d2a19Elliott Hughes        sh_size -= notesz;
174cc4a1f4d520f090bd563647df2614d09b66bd54aRob Landley      }
175cc4a1f4d520f090bd563647df2614d09b66bd54aRob Landley    }
176cc4a1f4d520f090bd563647df2614d09b66bd54aRob Landley  }
177c5e7ee27f606ed0bdb20d54e9de1a4adb56fc5d3Elliott Hughes  printf(", %sstripped", stripped ? "" : "not ");
1780332b60f62d71da61634cc69416129dfc64dced8Rob Landley  xputc('\n');
179c5e7ee27f606ed0bdb20d54e9de1a4adb56fc5d3Elliott Hughes
180c5e7ee27f606ed0bdb20d54e9de1a4adb56fc5d3Elliott Hughes  munmap(map, sb->st_size);
1819502a77ea0a16c395407f31af0bbb0afd0680adbElliott Hughes}
1829502a77ea0a16c395407f31af0bbb0afd0680adbElliott Hughes
183c5e7ee27f606ed0bdb20d54e9de1a4adb56fc5d3Elliott Hughesstatic void do_regular_file(int fd, char *name, struct stat *sb)
1849502a77ea0a16c395407f31af0bbb0afd0680adbElliott Hughes{
1850332b60f62d71da61634cc69416129dfc64dced8Rob Landley  char *s;
1860332b60f62d71da61634cc69416129dfc64dced8Rob Landley  int len = read(fd, s = toybuf, sizeof(toybuf)-256);
187830f2561757083ec6d34dc92c675447e91c9ed81Isaac Dunham  int magic;
1889502a77ea0a16c395407f31af0bbb0afd0680adbElliott Hughes
1890332b60f62d71da61634cc69416129dfc64dced8Rob Landley  if (len<0) perror_msg("%s", name);
1909502a77ea0a16c395407f31af0bbb0afd0680adbElliott Hughes
191c5e7ee27f606ed0bdb20d54e9de1a4adb56fc5d3Elliott Hughes  if (len>40 && strstart(&s, "\177ELF")) do_elf_file(fd, sb);
1926783156813d6c1c8469a7535f3cb575a776ceaa6Elliott Hughes  else if (len>=8 && strstart(&s, "!<arch>\n")) xprintf("ar archive\n");
193cc4a1f4d520f090bd563647df2614d09b66bd54aRob Landley  else if (len>28 && strstart(&s, "\x89PNG\x0d\x0a\x1a\x0a")) {
1940332b60f62d71da61634cc69416129dfc64dced8Rob Landley    // PNG is big-endian: https://www.w3.org/TR/PNG/#7Integers-and-byte-order
1950332b60f62d71da61634cc69416129dfc64dced8Rob Landley    int chunk_length = peek_be(s, 4);
1969502a77ea0a16c395407f31af0bbb0afd0680adbElliott Hughes
1970332b60f62d71da61634cc69416129dfc64dced8Rob Landley    xprintf("PNG image data");
1989502a77ea0a16c395407f31af0bbb0afd0680adbElliott Hughes
1990332b60f62d71da61634cc69416129dfc64dced8Rob Landley    // The IHDR chunk comes first: https://www.w3.org/TR/PNG/#11IHDR
2000332b60f62d71da61634cc69416129dfc64dced8Rob Landley    s += 4;
2010332b60f62d71da61634cc69416129dfc64dced8Rob Landley    if (chunk_length == 13 && strstart(&s, "IHDR")) {
2020332b60f62d71da61634cc69416129dfc64dced8Rob Landley      // https://www.w3.org/TR/PNG/#6Colour-values
2030332b60f62d71da61634cc69416129dfc64dced8Rob Landley      char *c = 0, *colors[] = {"grayscale", 0, "color RGB", "indexed color",
2040332b60f62d71da61634cc69416129dfc64dced8Rob Landley                                "grayscale with alpha", 0, "color RGBA"};
2059502a77ea0a16c395407f31af0bbb0afd0680adbElliott Hughes
2060332b60f62d71da61634cc69416129dfc64dced8Rob Landley      if (s[9]<ARRAY_LEN(colors)) c = colors[s[9]];
2070332b60f62d71da61634cc69416129dfc64dced8Rob Landley      if (!c) c = "unknown";
2089502a77ea0a16c395407f31af0bbb0afd0680adbElliott Hughes
2090332b60f62d71da61634cc69416129dfc64dced8Rob Landley      xprintf(", %d x %d, %d-bit/%s, %sinterlaced", (int)peek_be(s, 4),
2100332b60f62d71da61634cc69416129dfc64dced8Rob Landley        (int)peek_be(s+4, 4), s[8], c, s[12] ? "" : "non-");
2110332b60f62d71da61634cc69416129dfc64dced8Rob Landley    }
2129502a77ea0a16c395407f31af0bbb0afd0680adbElliott Hughes
2130332b60f62d71da61634cc69416129dfc64dced8Rob Landley    xputc('\n');
2149502a77ea0a16c395407f31af0bbb0afd0680adbElliott Hughes
2150332b60f62d71da61634cc69416129dfc64dced8Rob Landley  // https://www.w3.org/Graphics/GIF/spec-gif89a.txt
2160332b60f62d71da61634cc69416129dfc64dced8Rob Landley  } else if (len>16 && (strstart(&s, "GIF87a") || strstart(&s, "GIF89a")))
2170332b60f62d71da61634cc69416129dfc64dced8Rob Landley    xprintf("GIF image data, %d x %d\n",
2180332b60f62d71da61634cc69416129dfc64dced8Rob Landley      (int)peek_le(s, 2), (int)peek_le(s+8, 2));
2199502a77ea0a16c395407f31af0bbb0afd0680adbElliott Hughes
2200332b60f62d71da61634cc69416129dfc64dced8Rob Landley  // TODO: parsing JPEG for width/height is harder than GIF or PNG.
2216232dc68c4169332950333b259943e175ddcfaeaRob Landley  else if (len>32 && !memcmp(toybuf, "\xff\xd8", 2)) xputs("JPEG image data");
2229502a77ea0a16c395407f31af0bbb0afd0680adbElliott Hughes
2230332b60f62d71da61634cc69416129dfc64dced8Rob Landley  // https://docs.oracle.com/javase/specs/jvms/se7/html/jvms-4.html
2240332b60f62d71da61634cc69416129dfc64dced8Rob Landley  else if (len>8 && strstart(&s, "\xca\xfe\xba\xbe"))
2250332b60f62d71da61634cc69416129dfc64dced8Rob Landley    xprintf("Java class file, version %d.%d\n",
226ffc6fbbde3eeca29d1eb3470610eb7ae5b9025f1Elliott Hughes      (int)peek_be(s+2, 2), (int)peek_be(s, 2));
2279502a77ea0a16c395407f31af0bbb0afd0680adbElliott Hughes
228830f2561757083ec6d34dc92c675447e91c9ed81Isaac Dunham  // https://people.freebsd.org/~kientzle/libarchive/man/cpio.5.txt
229830f2561757083ec6d34dc92c675447e91c9ed81Isaac Dunham  // the lengths for cpio are size of header + 9 bytes, since any valid
230830f2561757083ec6d34dc92c675447e91c9ed81Isaac Dunham  // cpio archive ends with a record for "TARGET!!!"
231830f2561757083ec6d34dc92c675447e91c9ed81Isaac Dunham  else if (len>85 && strstart(&s, "07070")) {
232830f2561757083ec6d34dc92c675447e91c9ed81Isaac Dunham    char *cpioformat = "unknown type";
2336232dc68c4169332950333b259943e175ddcfaeaRob Landley
234830f2561757083ec6d34dc92c675447e91c9ed81Isaac Dunham    if (toybuf[5] == '7') cpioformat = "pre-SVR4 or odc";
235830f2561757083ec6d34dc92c675447e91c9ed81Isaac Dunham    else if (toybuf[5] == '1') cpioformat = "SVR4 with no CRC";
236830f2561757083ec6d34dc92c675447e91c9ed81Isaac Dunham    else if (toybuf[5] == '2') cpioformat = "SVR4 with CRC";
237830f2561757083ec6d34dc92c675447e91c9ed81Isaac Dunham    xprintf("ASCII cpio archive (%s)\n", cpioformat);
2389f42e832a186bb3f5320581a3a713d0a8a9c8384Rob Landley  } else if (len>33 && (magic=peek(&s,2), magic==0143561 || magic==070707)) {
239830f2561757083ec6d34dc92c675447e91c9ed81Isaac Dunham    if (magic == 0143561) printf("byte-swapped ");
240830f2561757083ec6d34dc92c675447e91c9ed81Isaac Dunham    xprintf("cpio archive\n");
241830f2561757083ec6d34dc92c675447e91c9ed81Isaac Dunham  // tar archive (ustar/pax or gnu)
2426232dc68c4169332950333b259943e175ddcfaeaRob Landley  } else if (len>500 && !strncmp(s+257, "ustar", 5))
243830f2561757083ec6d34dc92c675447e91c9ed81Isaac Dunham    xprintf("POSIX tar archive%s\n", strncmp(s+262,"  ",2)?"":" (GNU)");
244830f2561757083ec6d34dc92c675447e91c9ed81Isaac Dunham  // zip/jar/apk archive, ODF/OOXML document, or such
2456232dc68c4169332950333b259943e175ddcfaeaRob Landley  else if (len>5 && strstart(&s, "PK\03\04")) {
2466232dc68c4169332950333b259943e175ddcfaeaRob Landley    int ver = toybuf[4];
2476232dc68c4169332950333b259943e175ddcfaeaRob Landley
248830f2561757083ec6d34dc92c675447e91c9ed81Isaac Dunham    xprintf("Zip archive data");
2496232dc68c4169332950333b259943e175ddcfaeaRob Landley    if (ver) xprintf(", requires at least v%d.%d to extract", ver/10, ver%10);
250830f2561757083ec6d34dc92c675447e91c9ed81Isaac Dunham    xputc('\n');
2516232dc68c4169332950333b259943e175ddcfaeaRob Landley  } else if (len>4 && strstart(&s, "BZh") && isdigit(*s))
252af92dfdad8b8f2e8d35a1959832a3b6939261d9fElliott Hughes    xprintf("bzip2 compressed data, block size = %c00k\n", *s);
2536232dc68c4169332950333b259943e175ddcfaeaRob Landley  else if (len>10 && strstart(&s, "\x1f\x8b")) xputs("gzip compressed data");
254d1c664a31b656c1541c79044aa44467420f78f14Rob Landley  else {
2550332b60f62d71da61634cc69416129dfc64dced8Rob Landley    char *what = 0;
2560332b60f62d71da61634cc69416129dfc64dced8Rob Landley    int i, bytes;
2579502a77ea0a16c395407f31af0bbb0afd0680adbElliott Hughes
2580332b60f62d71da61634cc69416129dfc64dced8Rob Landley    // If shell script, report which interpreter
2590332b60f62d71da61634cc69416129dfc64dced8Rob Landley    if (len>3 && strstart(&s, "#!")) {
260ffc6fbbde3eeca29d1eb3470610eb7ae5b9025f1Elliott Hughes      // Whitespace is allowed between the #! and the interpreter
261ffc6fbbde3eeca29d1eb3470610eb7ae5b9025f1Elliott Hughes      while (isspace(*s)) s++;
262ffc6fbbde3eeca29d1eb3470610eb7ae5b9025f1Elliott Hughes      if (strstart(&s, "/usr/bin/env")) while (isspace(*s)) s++;
2630332b60f62d71da61634cc69416129dfc64dced8Rob Landley      for (what = s; (s-toybuf)<len && !isspace(*s); s++);
2640332b60f62d71da61634cc69416129dfc64dced8Rob Landley      strcpy(s, " script");
2659502a77ea0a16c395407f31af0bbb0afd0680adbElliott Hughes
2660332b60f62d71da61634cc69416129dfc64dced8Rob Landley    // Distinguish ASCII text, UTF-8 text, or data
2670332b60f62d71da61634cc69416129dfc64dced8Rob Landley    } else for (i = 0; i<len; ++i) {
2689502a77ea0a16c395407f31af0bbb0afd0680adbElliott Hughes      if (!(isprint(toybuf[i]) || isspace(toybuf[i]))) {
2690332b60f62d71da61634cc69416129dfc64dced8Rob Landley        wchar_t wc;
2700332b60f62d71da61634cc69416129dfc64dced8Rob Landley        if ((bytes = mbrtowc(&wc, s+i, len-i, 0))>0 && wcwidth(wc)>=0) {
2710332b60f62d71da61634cc69416129dfc64dced8Rob Landley          i += bytes-1;
2720332b60f62d71da61634cc69416129dfc64dced8Rob Landley          if (!what) what = "UTF-8 text";
2730332b60f62d71da61634cc69416129dfc64dced8Rob Landley        } else {
2740332b60f62d71da61634cc69416129dfc64dced8Rob Landley          what = "data";
2750332b60f62d71da61634cc69416129dfc64dced8Rob Landley          break;
2760332b60f62d71da61634cc69416129dfc64dced8Rob Landley        }
2779502a77ea0a16c395407f31af0bbb0afd0680adbElliott Hughes      }
2789502a77ea0a16c395407f31af0bbb0afd0680adbElliott Hughes    }
2790332b60f62d71da61634cc69416129dfc64dced8Rob Landley    xputs(what ? what : "ASCII text");
2809502a77ea0a16c395407f31af0bbb0afd0680adbElliott Hughes  }
2819502a77ea0a16c395407f31af0bbb0afd0680adbElliott Hughes}
2829502a77ea0a16c395407f31af0bbb0afd0680adbElliott Hughes
2830332b60f62d71da61634cc69416129dfc64dced8Rob Landleyvoid file_main(void)
2849502a77ea0a16c395407f31af0bbb0afd0680adbElliott Hughes{
2854b141e67eefc396687f22f8c136267c0f8081e05Rob Landley  char **arg;
2869502a77ea0a16c395407f31af0bbb0afd0680adbElliott Hughes
2874b141e67eefc396687f22f8c136267c0f8081e05Rob Landley  for (arg = toys.optargs; *arg; ++arg) {
2884b141e67eefc396687f22f8c136267c0f8081e05Rob Landley    int name_len = strlen(*arg);
2890332b60f62d71da61634cc69416129dfc64dced8Rob Landley
2909502a77ea0a16c395407f31af0bbb0afd0680adbElliott Hughes    if (name_len > TT.max_name_len) TT.max_name_len = name_len;
2919502a77ea0a16c395407f31af0bbb0afd0680adbElliott Hughes  }
2929502a77ea0a16c395407f31af0bbb0afd0680adbElliott Hughes
2934b141e67eefc396687f22f8c136267c0f8081e05Rob Landley  // Can't use loopfiles here because it doesn't call function when can't open
2944b141e67eefc396687f22f8c136267c0f8081e05Rob Landley  for (arg = toys.optargs; *arg; arg++) {
2954b141e67eefc396687f22f8c136267c0f8081e05Rob Landley    char *name = *arg, *what = "cannot open";
2969f42e832a186bb3f5320581a3a713d0a8a9c8384Rob Landley    struct stat sb;
2979f42e832a186bb3f5320581a3a713d0a8a9c8384Rob Landley    int fd = !strcmp(name, "-");
2984b141e67eefc396687f22f8c136267c0f8081e05Rob Landley
2994b141e67eefc396687f22f8c136267c0f8081e05Rob Landley    xprintf("%s: %*s", name, (int)(TT.max_name_len - strlen(name)), "");
3004b141e67eefc396687f22f8c136267c0f8081e05Rob Landley
301493dc573825a862ee533c8aaf91ab23112b591c2Elliott Hughes    if (fd || !((toys.optflags & FLAG_L) ? stat : lstat)(name, &sb)) {
3029f42e832a186bb3f5320581a3a713d0a8a9c8384Rob Landley      if (fd || S_ISREG(sb.st_mode)) {
3039f42e832a186bb3f5320581a3a713d0a8a9c8384Rob Landley        if (!sb.st_size) what = "empty";
3049f42e832a186bb3f5320581a3a713d0a8a9c8384Rob Landley        else if ((fd = openro(name, O_RDONLY)) != -1) {
3059f42e832a186bb3f5320581a3a713d0a8a9c8384Rob Landley          do_regular_file(fd, name, &sb);
3062e4aff2fd31fb9d949d6c47da1e46b97ecc23d24Rob Landley          if (fd) close(fd);
3079f42e832a186bb3f5320581a3a713d0a8a9c8384Rob Landley          continue;
3084b141e67eefc396687f22f8c136267c0f8081e05Rob Landley        }
3099f42e832a186bb3f5320581a3a713d0a8a9c8384Rob Landley      } else if (S_ISFIFO(sb.st_mode)) what = "fifo";
3109f42e832a186bb3f5320581a3a713d0a8a9c8384Rob Landley      else if (S_ISBLK(sb.st_mode)) what = "block special";
3114b141e67eefc396687f22f8c136267c0f8081e05Rob Landley      else if (S_ISCHR(sb.st_mode)) what = "character special";
3124b141e67eefc396687f22f8c136267c0f8081e05Rob Landley      else if (S_ISDIR(sb.st_mode)) what = "directory";
3134b141e67eefc396687f22f8c136267c0f8081e05Rob Landley      else if (S_ISSOCK(sb.st_mode)) what = "socket";
3144b141e67eefc396687f22f8c136267c0f8081e05Rob Landley      else if (S_ISLNK(sb.st_mode)) what = "symbolic link";
3154b141e67eefc396687f22f8c136267c0f8081e05Rob Landley      else what = "unknown";
3164b141e67eefc396687f22f8c136267c0f8081e05Rob Landley    }
3174b141e67eefc396687f22f8c136267c0f8081e05Rob Landley
3184b141e67eefc396687f22f8c136267c0f8081e05Rob Landley    xputs(what);
3194b141e67eefc396687f22f8c136267c0f8081e05Rob Landley  }
3209502a77ea0a16c395407f31af0bbb0afd0680adbElliott Hughes}
321